1 // NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --function-signature --include-generated-funcs --replace-value-regex "__omp_offloading_[0-9a-z]+_[0-9a-z]+" "reduction_size[.].+[.]" "pl_cond[.].+[.|,]" --prefix-filecheck-ir-name _ 2 // expected-no-diagnostics 3 #ifndef HEADER 4 #define HEADER 5 // Test host codegen. 6 // RUN: %clang_cc1 -no-opaque-pointers -DCK1 -verify -fopenmp -x c++ -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -emit-llvm %s -o - | FileCheck %s --check-prefix=CHECK1 7 // RUN: %clang_cc1 -no-opaque-pointers -DCK1 -fopenmp -x c++ -std=c++11 -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -emit-pch -o %t %s 8 // RUN: %clang_cc1 -no-opaque-pointers -DCK1 -fopenmp -x c++ -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck %s --check-prefix=CHECK1 9 // RUN: %clang_cc1 -no-opaque-pointers -DCK1 -verify -fopenmp -x c++ -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -emit-llvm %s -o - | FileCheck %s --check-prefix=CHECK3 10 // RUN: %clang_cc1 -no-opaque-pointers -DCK1 -fopenmp -x c++ -std=c++11 -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -emit-pch -o %t %s 11 // RUN: %clang_cc1 -no-opaque-pointers -DCK1 -fopenmp -x c++ -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck %s --check-prefix=CHECK3 12 13 // RUN: %clang_cc1 -no-opaque-pointers -DCK1 -verify -fopenmp-simd -x c++ -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -emit-llvm %s -o - | FileCheck %s --check-prefix=CHECK5 14 // RUN: %clang_cc1 -no-opaque-pointers -DCK1 -fopenmp-simd -x c++ -std=c++11 -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -emit-pch -o %t %s 15 // RUN: %clang_cc1 -no-opaque-pointers -DCK1 -fopenmp-simd -x c++ -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck %s --check-prefix=CHECK5 16 // RUN: %clang_cc1 -no-opaque-pointers -DCK1 -verify -fopenmp-simd -x c++ -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -emit-llvm %s -o - | FileCheck %s --check-prefix=CHECK7 17 // RUN: %clang_cc1 -no-opaque-pointers -DCK1 -fopenmp-simd -x c++ -std=c++11 -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -emit-pch -o %t %s 18 // RUN: %clang_cc1 -no-opaque-pointers -DCK1 -fopenmp-simd -x c++ -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck %s --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 -no-opaque-pointers -DCK2 -verify -fopenmp -x c++ -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -emit-llvm %s -o - | FileCheck %s --check-prefix=CHECK9 57 // RUN: %clang_cc1 -no-opaque-pointers -DCK2 -fopenmp -x c++ -std=c++11 -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -emit-pch -o %t %s 58 // RUN: %clang_cc1 -no-opaque-pointers -DCK2 -fopenmp -x c++ -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck %s --check-prefix=CHECK9 59 // RUN: %clang_cc1 -no-opaque-pointers -DCK2 -verify -fopenmp -x c++ -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -emit-llvm %s -o - | FileCheck %s --check-prefix=CHECK11 60 // RUN: %clang_cc1 -no-opaque-pointers -DCK2 -fopenmp -x c++ -std=c++11 -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -emit-pch -o %t %s 61 // RUN: %clang_cc1 -no-opaque-pointers -DCK2 -fopenmp -x c++ -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck %s --check-prefix=CHECK11 62 63 // RUN: %clang_cc1 -no-opaque-pointers -DCK2 -verify -fopenmp-simd -x c++ -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -emit-llvm %s -o - | FileCheck %s --check-prefix=CHECK13 64 // RUN: %clang_cc1 -no-opaque-pointers -DCK2 -fopenmp-simd -x c++ -std=c++11 -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -emit-pch -o %t %s 65 // RUN: %clang_cc1 -no-opaque-pointers -DCK2 -fopenmp-simd -x c++ -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck %s --check-prefix=CHECK13 66 // RUN: %clang_cc1 -no-opaque-pointers -DCK2 -verify -fopenmp-simd -x c++ -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -emit-llvm %s -o - | FileCheck %s --check-prefix=CHECK15 67 // RUN: %clang_cc1 -no-opaque-pointers -DCK2 -fopenmp-simd -x c++ -std=c++11 -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -emit-pch -o %t %s 68 // RUN: %clang_cc1 -no-opaque-pointers -DCK2 -fopenmp-simd -x c++ -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck %s --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 -no-opaque-pointers -DCK3 -verify -fopenmp -x c++ -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -emit-llvm %s -o - | FileCheck %s --check-prefix=CHECK17 90 // RUN: %clang_cc1 -no-opaque-pointers -DCK3 -fopenmp -x c++ -std=c++11 -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -emit-pch -o %t %s 91 // RUN: %clang_cc1 -no-opaque-pointers -DCK3 -fopenmp -x c++ -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck %s --check-prefix=CHECK17 92 // RUN: %clang_cc1 -no-opaque-pointers -DCK3 -verify -fopenmp -x c++ -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -emit-llvm %s -o - | FileCheck %s --check-prefix=CHECK19 93 // RUN: %clang_cc1 -no-opaque-pointers -DCK3 -fopenmp -x c++ -std=c++11 -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -emit-pch -o %t %s 94 // RUN: %clang_cc1 -no-opaque-pointers -DCK3 -fopenmp -x c++ -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck %s --check-prefix=CHECK19 95 // RUN: %clang_cc1 -no-opaque-pointers -DCK3 -verify -fopenmp -fopenmp-version=50 -DOMP5 -x c++ -triple powerpc64le-unknown-unknown -fopenmp -fopenmp-version=50 -fopenmp-targets=powerpc64le-ibm-linux-gnu -emit-llvm %s -o - | FileCheck %s --check-prefix=CHECK21 96 // RUN: %clang_cc1 -no-opaque-pointers -DCK3 -fopenmp -fopenmp-version=50 -DOMP5 -x c++ -std=c++11 -triple powerpc64le-unknown-unknown -fopenmp -fopenmp-version=50 -fopenmp-targets=powerpc64le-ibm-linux-gnu -emit-pch -o %t %s 97 // RUN: %clang_cc1 -no-opaque-pointers -DCK3 -fopenmp -fopenmp-version=50 -DOMP5 -x c++ -triple powerpc64le-unknown-unknown -fopenmp -fopenmp-version=50 -fopenmp-targets=powerpc64le-ibm-linux-gnu -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck %s --check-prefix=CHECK21 98 // RUN: %clang_cc1 -no-opaque-pointers -DCK3 -verify -fopenmp -fopenmp-version=50 -DOMP5 -x c++ -triple i386-unknown-unknown -fopenmp -fopenmp-version=50 -fopenmp-targets=i386-pc-linux-gnu -emit-llvm %s -o - | FileCheck %s --check-prefix=CHECK23 99 // RUN: %clang_cc1 -no-opaque-pointers -DCK3 -fopenmp -fopenmp-version=50 -DOMP5 -x c++ -std=c++11 -triple i386-unknown-unknown -fopenmp -fopenmp-version=50 -fopenmp-targets=i386-pc-linux-gnu -emit-pch -o %t %s 100 // RUN: %clang_cc1 -no-opaque-pointers -DCK3 -fopenmp -fopenmp-version=50 -DOMP5 -x c++ -triple i386-unknown-unknown -fopenmp -fopenmp-version=50 -fopenmp-targets=i386-pc-linux-gnu -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck %s --check-prefix=CHECK23 101 102 // RUN: %clang_cc1 -no-opaque-pointers -DCK3 -verify -fopenmp-simd -x c++ -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -emit-llvm %s -o - | FileCheck %s --check-prefix=CHECK25 103 // RUN: %clang_cc1 -no-opaque-pointers -DCK3 -fopenmp-simd -x c++ -std=c++11 -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -emit-pch -o %t %s 104 // RUN: %clang_cc1 -no-opaque-pointers -DCK3 -fopenmp-simd -x c++ -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck %s --check-prefix=CHECK25 105 // RUN: %clang_cc1 -no-opaque-pointers -DCK3 -verify -fopenmp-simd -x c++ -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -emit-llvm %s -o - | FileCheck %s --check-prefix=CHECK27 106 // RUN: %clang_cc1 -no-opaque-pointers -DCK3 -fopenmp-simd -x c++ -std=c++11 -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -emit-pch -o %t %s 107 // RUN: %clang_cc1 -no-opaque-pointers -DCK3 -fopenmp-simd -x c++ -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck %s --check-prefix=CHECK27 108 // RUN: %clang_cc1 -no-opaque-pointers -DCK3 -verify -fopenmp-simd -fopenmp-version=50 -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 -no-opaque-pointers -DCK3 -fopenmp-simd -fopenmp-version=50 -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 -no-opaque-pointers -DCK3 -fopenmp-simd -fopenmp-version=50 -DOMP5 -x c++ -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck %s --check-prefix=CHECK29 111 // RUN: %clang_cc1 -no-opaque-pointers -DCK3 -verify -fopenmp-simd -fopenmp-version=50 -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 -no-opaque-pointers -DCK3 -fopenmp-simd -fopenmp-version=50 -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 -no-opaque-pointers -DCK3 -fopenmp-simd -fopenmp-version=50 -DOMP5 -x c++ -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck %s --check-prefix=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 -no-opaque-pointers -DCK4 -verify -fopenmp -x c++ -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -emit-llvm %s -o - | FileCheck %s --check-prefix=CHECK33 149 // RUN: %clang_cc1 -no-opaque-pointers -DCK4 -fopenmp -x c++ -std=c++11 -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -emit-pch -o %t %s 150 // RUN: %clang_cc1 -no-opaque-pointers -DCK4 -fopenmp -x c++ -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck %s --check-prefix=CHECK33 151 // RUN: %clang_cc1 -no-opaque-pointers -DCK4 -verify -fopenmp -x c++ -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -emit-llvm %s -o - | FileCheck %s --check-prefix=CHECK35 152 // RUN: %clang_cc1 -no-opaque-pointers -DCK4 -fopenmp -x c++ -std=c++11 -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -emit-pch -o %t %s 153 // RUN: %clang_cc1 -no-opaque-pointers -DCK4 -fopenmp -x c++ -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck %s --check-prefix=CHECK35 154 // RUN: %clang_cc1 -no-opaque-pointers -DCK4 -verify -fopenmp -fopenmp-version=50 -DOMP5 -x c++ -triple powerpc64le-unknown-unknown -fopenmp -fopenmp-version=50 -fopenmp-targets=powerpc64le-ibm-linux-gnu -emit-llvm %s -o - | FileCheck %s --check-prefix=CHECK37 155 // RUN: %clang_cc1 -no-opaque-pointers -DCK4 -fopenmp -fopenmp-version=50 -DOMP5 -x c++ -std=c++11 -triple powerpc64le-unknown-unknown -fopenmp -fopenmp-version=50 -fopenmp-targets=powerpc64le-ibm-linux-gnu -emit-pch -o %t %s 156 // RUN: %clang_cc1 -no-opaque-pointers -DCK4 -fopenmp -fopenmp-version=50 -DOMP5 -x c++ -triple powerpc64le-unknown-unknown -fopenmp -fopenmp-version=50 -fopenmp-targets=powerpc64le-ibm-linux-gnu -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck %s --check-prefix=CHECK37 157 // RUN: %clang_cc1 -no-opaque-pointers -DCK4 -verify -fopenmp -fopenmp-version=50 -DOMP5 -x c++ -triple i386-unknown-unknown -fopenmp -fopenmp-version=50 -fopenmp-targets=i386-pc-linux-gnu -emit-llvm %s -o - | FileCheck %s --check-prefix=CHECK39 158 // RUN: %clang_cc1 -no-opaque-pointers -DCK4 -fopenmp -fopenmp-version=50 -DOMP5 -x c++ -std=c++11 -triple i386-unknown-unknown -fopenmp -fopenmp-version=50 -fopenmp-targets=i386-pc-linux-gnu -emit-pch -o %t %s 159 // RUN: %clang_cc1 -no-opaque-pointers -DCK4 -fopenmp -fopenmp-version=50 -DOMP5 -x c++ -triple i386-unknown-unknown -fopenmp -fopenmp-version=50 -fopenmp-targets=i386-pc-linux-gnu -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck %s --check-prefix=CHECK39 160 161 // RUN: %clang_cc1 -no-opaque-pointers -DCK4 -verify -fopenmp-simd -x c++ -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -emit-llvm %s -o - | FileCheck %s --check-prefix=CHECK41 162 // RUN: %clang_cc1 -no-opaque-pointers -DCK4 -fopenmp-simd -x c++ -std=c++11 -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -emit-pch -o %t %s 163 // RUN: %clang_cc1 -no-opaque-pointers -DCK4 -fopenmp-simd -x c++ -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck %s --check-prefix=CHECK41 164 // RUN: %clang_cc1 -no-opaque-pointers -DCK4 -verify -fopenmp-simd -x c++ -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -emit-llvm %s -o - | FileCheck %s --check-prefix=CHECK43 165 // RUN: %clang_cc1 -no-opaque-pointers -DCK4 -fopenmp-simd -x c++ -std=c++11 -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -emit-pch -o %t %s 166 // RUN: %clang_cc1 -no-opaque-pointers -DCK4 -fopenmp-simd -x c++ -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck %s --check-prefix=CHECK43 167 // RUN: %clang_cc1 -no-opaque-pointers -DCK4 -verify -fopenmp-simd -fopenmp-version=50 -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 -no-opaque-pointers -DCK4 -fopenmp-simd -fopenmp-version=50 -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 -no-opaque-pointers -DCK4 -fopenmp-simd -fopenmp-version=50 -DOMP5 -x c++ -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck %s --check-prefix=CHECK45 170 // RUN: %clang_cc1 -no-opaque-pointers -DCK4 -verify -fopenmp-simd -fopenmp-version=50 -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 -no-opaque-pointers -DCK4 -fopenmp-simd -fopenmp-version=50 -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 -no-opaque-pointers -DCK4 -fopenmp-simd -fopenmp-version=50 -DOMP5 -x c++ -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck %s --check-prefix=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 213 214 // CHECK1-LABEL: define {{[^@]+}}@_Z21teams_argument_globali 215 // CHECK1-SAME: (i32 noundef signext [[N:%.*]]) #[[ATTR0:[0-9]+]] { 216 // CHECK1-NEXT: entry: 217 // CHECK1-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4 218 // CHECK1-NEXT: [[I:%.*]] = alloca i32, align 4 219 // CHECK1-NEXT: [[TE:%.*]] = alloca i32, align 4 220 // CHECK1-NEXT: [[TH:%.*]] = alloca i32, align 4 221 // CHECK1-NEXT: [[TE_CASTED:%.*]] = alloca i64, align 8 222 // CHECK1-NEXT: [[TH_CASTED:%.*]] = alloca i64, align 8 223 // CHECK1-NEXT: [[I_CASTED:%.*]] = alloca i64, align 8 224 // CHECK1-NEXT: [[N_CASTED:%.*]] = alloca i64, align 8 225 // CHECK1-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [5 x i8*], align 8 226 // CHECK1-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [5 x i8*], align 8 227 // CHECK1-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [5 x i8*], align 8 228 // CHECK1-NEXT: [[TMP:%.*]] = alloca i32, align 4 229 // CHECK1-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 230 // CHECK1-NEXT: [[DOTCAPTURE_EXPR_4:%.*]] = alloca i32, align 4 231 // CHECK1-NEXT: [[N_CASTED7:%.*]] = alloca i64, align 8 232 // CHECK1-NEXT: [[DOTOFFLOAD_BASEPTRS9:%.*]] = alloca [2 x i8*], align 8 233 // CHECK1-NEXT: [[DOTOFFLOAD_PTRS10:%.*]] = alloca [2 x i8*], align 8 234 // CHECK1-NEXT: [[DOTOFFLOAD_MAPPERS11:%.*]] = alloca [2 x i8*], align 8 235 // CHECK1-NEXT: [[_TMP12:%.*]] = alloca i32, align 4 236 // CHECK1-NEXT: [[DOTCAPTURE_EXPR_13:%.*]] = alloca i32, align 4 237 // CHECK1-NEXT: [[DOTCAPTURE_EXPR_14:%.*]] = alloca i32, align 4 238 // CHECK1-NEXT: store i32 [[N]], i32* [[N_ADDR]], align 4 239 // CHECK1-NEXT: [[TMP0:%.*]] = load i32, i32* [[N_ADDR]], align 4 240 // CHECK1-NEXT: [[DIV:%.*]] = sdiv i32 [[TMP0]], 128 241 // CHECK1-NEXT: store i32 [[DIV]], i32* [[TE]], align 4 242 // CHECK1-NEXT: store i32 128, i32* [[TH]], align 4 243 // CHECK1-NEXT: [[TMP1:%.*]] = load i32, i32* [[TE]], align 4 244 // CHECK1-NEXT: [[CONV:%.*]] = bitcast i64* [[TE_CASTED]] to i32* 245 // CHECK1-NEXT: store i32 [[TMP1]], i32* [[CONV]], align 4 246 // CHECK1-NEXT: [[TMP2:%.*]] = load i64, i64* [[TE_CASTED]], align 8 247 // CHECK1-NEXT: [[TMP3:%.*]] = load i32, i32* [[TH]], align 4 248 // CHECK1-NEXT: [[CONV1:%.*]] = bitcast i64* [[TH_CASTED]] to i32* 249 // CHECK1-NEXT: store i32 [[TMP3]], i32* [[CONV1]], align 4 250 // CHECK1-NEXT: [[TMP4:%.*]] = load i64, i64* [[TH_CASTED]], align 8 251 // CHECK1-NEXT: [[TMP5:%.*]] = load i32, i32* [[I]], align 4 252 // CHECK1-NEXT: [[CONV2:%.*]] = bitcast i64* [[I_CASTED]] to i32* 253 // CHECK1-NEXT: store i32 [[TMP5]], i32* [[CONV2]], align 4 254 // CHECK1-NEXT: [[TMP6:%.*]] = load i64, i64* [[I_CASTED]], align 8 255 // CHECK1-NEXT: [[TMP7:%.*]] = load i32, i32* [[N_ADDR]], align 4 256 // CHECK1-NEXT: [[CONV3:%.*]] = bitcast i64* [[N_CASTED]] to i32* 257 // CHECK1-NEXT: store i32 [[TMP7]], i32* [[CONV3]], align 4 258 // CHECK1-NEXT: [[TMP8:%.*]] = load i64, i64* [[N_CASTED]], align 8 259 // CHECK1-NEXT: [[TMP9:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 260 // CHECK1-NEXT: [[TMP10:%.*]] = bitcast i8** [[TMP9]] to i64* 261 // CHECK1-NEXT: store i64 [[TMP2]], i64* [[TMP10]], align 8 262 // CHECK1-NEXT: [[TMP11:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 263 // CHECK1-NEXT: [[TMP12:%.*]] = bitcast i8** [[TMP11]] to i64* 264 // CHECK1-NEXT: store i64 [[TMP2]], i64* [[TMP12]], align 8 265 // CHECK1-NEXT: [[TMP13:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0 266 // CHECK1-NEXT: store i8* null, i8** [[TMP13]], align 8 267 // CHECK1-NEXT: [[TMP14:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1 268 // CHECK1-NEXT: [[TMP15:%.*]] = bitcast i8** [[TMP14]] to i64* 269 // CHECK1-NEXT: store i64 [[TMP4]], i64* [[TMP15]], align 8 270 // CHECK1-NEXT: [[TMP16:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1 271 // CHECK1-NEXT: [[TMP17:%.*]] = bitcast i8** [[TMP16]] to i64* 272 // CHECK1-NEXT: store i64 [[TMP4]], i64* [[TMP17]], align 8 273 // CHECK1-NEXT: [[TMP18:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 1 274 // CHECK1-NEXT: store i8* null, i8** [[TMP18]], align 8 275 // CHECK1-NEXT: [[TMP19:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2 276 // CHECK1-NEXT: [[TMP20:%.*]] = bitcast i8** [[TMP19]] to [100 x i32]** 277 // CHECK1-NEXT: store [100 x i32]* @a, [100 x i32]** [[TMP20]], align 8 278 // CHECK1-NEXT: [[TMP21:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2 279 // CHECK1-NEXT: [[TMP22:%.*]] = bitcast i8** [[TMP21]] to [100 x i32]** 280 // CHECK1-NEXT: store [100 x i32]* @a, [100 x i32]** [[TMP22]], align 8 281 // CHECK1-NEXT: [[TMP23:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2 282 // CHECK1-NEXT: store i8* null, i8** [[TMP23]], align 8 283 // CHECK1-NEXT: [[TMP24:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 3 284 // CHECK1-NEXT: [[TMP25:%.*]] = bitcast i8** [[TMP24]] to i64* 285 // CHECK1-NEXT: store i64 [[TMP6]], i64* [[TMP25]], align 8 286 // CHECK1-NEXT: [[TMP26:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 3 287 // CHECK1-NEXT: [[TMP27:%.*]] = bitcast i8** [[TMP26]] to i64* 288 // CHECK1-NEXT: store i64 [[TMP6]], i64* [[TMP27]], align 8 289 // CHECK1-NEXT: [[TMP28:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 3 290 // CHECK1-NEXT: store i8* null, i8** [[TMP28]], align 8 291 // CHECK1-NEXT: [[TMP29:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 4 292 // CHECK1-NEXT: [[TMP30:%.*]] = bitcast i8** [[TMP29]] to i64* 293 // CHECK1-NEXT: store i64 [[TMP8]], i64* [[TMP30]], align 8 294 // CHECK1-NEXT: [[TMP31:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 4 295 // CHECK1-NEXT: [[TMP32:%.*]] = bitcast i8** [[TMP31]] to i64* 296 // CHECK1-NEXT: store i64 [[TMP8]], i64* [[TMP32]], align 8 297 // CHECK1-NEXT: [[TMP33:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 4 298 // CHECK1-NEXT: store i8* null, i8** [[TMP33]], align 8 299 // CHECK1-NEXT: [[TMP34:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 300 // CHECK1-NEXT: [[TMP35:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 301 // CHECK1-NEXT: [[TMP36:%.*]] = load i32, i32* [[TE]], align 4 302 // CHECK1-NEXT: [[TMP37:%.*]] = load i32, i32* [[N_ADDR]], align 4 303 // CHECK1-NEXT: store i32 [[TMP37]], i32* [[DOTCAPTURE_EXPR_]], align 4 304 // CHECK1-NEXT: [[TMP38:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 305 // CHECK1-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP38]], 0 306 // CHECK1-NEXT: [[DIV5:%.*]] = sdiv i32 [[SUB]], 1 307 // CHECK1-NEXT: [[SUB6:%.*]] = sub nsw i32 [[DIV5]], 1 308 // CHECK1-NEXT: store i32 [[SUB6]], i32* [[DOTCAPTURE_EXPR_4]], align 4 309 // CHECK1-NEXT: [[TMP39:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_4]], align 4 310 // CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP39]], 1 311 // CHECK1-NEXT: [[TMP40:%.*]] = zext i32 [[ADD]] to i64 312 // CHECK1-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2:[0-9]+]], i64 -1, i64 [[TMP40]]) 313 // CHECK1-NEXT: [[TMP41:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z21teams_argument_globali_l30.region_id, i32 5, i8** [[TMP34]], i8** [[TMP35]], i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_sizes, i32 0, i32 0), i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 [[TMP36]], i32 1) 314 // CHECK1-NEXT: [[TMP42:%.*]] = icmp ne i32 [[TMP41]], 0 315 // CHECK1-NEXT: br i1 [[TMP42]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] 316 // CHECK1: omp_offload.failed: 317 // CHECK1-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z21teams_argument_globali_l30(i64 [[TMP2]], i64 [[TMP4]], [100 x i32]* @a, i64 [[TMP6]], i64 [[TMP8]]) #[[ATTR3:[0-9]+]] 318 // CHECK1-NEXT: br label [[OMP_OFFLOAD_CONT]] 319 // CHECK1: omp_offload.cont: 320 // CHECK1-NEXT: [[TMP43:%.*]] = load i32, i32* [[N_ADDR]], align 4 321 // CHECK1-NEXT: [[CONV8:%.*]] = bitcast i64* [[N_CASTED7]] to i32* 322 // CHECK1-NEXT: store i32 [[TMP43]], i32* [[CONV8]], align 4 323 // CHECK1-NEXT: [[TMP44:%.*]] = load i64, i64* [[N_CASTED7]], align 8 324 // CHECK1-NEXT: [[TMP45:%.*]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[DOTOFFLOAD_BASEPTRS9]], i32 0, i32 0 325 // CHECK1-NEXT: [[TMP46:%.*]] = bitcast i8** [[TMP45]] to i64* 326 // CHECK1-NEXT: store i64 [[TMP44]], i64* [[TMP46]], align 8 327 // CHECK1-NEXT: [[TMP47:%.*]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[DOTOFFLOAD_PTRS10]], i32 0, i32 0 328 // CHECK1-NEXT: [[TMP48:%.*]] = bitcast i8** [[TMP47]] to i64* 329 // CHECK1-NEXT: store i64 [[TMP44]], i64* [[TMP48]], align 8 330 // CHECK1-NEXT: [[TMP49:%.*]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[DOTOFFLOAD_MAPPERS11]], i64 0, i64 0 331 // CHECK1-NEXT: store i8* null, i8** [[TMP49]], align 8 332 // CHECK1-NEXT: [[TMP50:%.*]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[DOTOFFLOAD_BASEPTRS9]], i32 0, i32 1 333 // CHECK1-NEXT: [[TMP51:%.*]] = bitcast i8** [[TMP50]] to [100 x i32]** 334 // CHECK1-NEXT: store [100 x i32]* @a, [100 x i32]** [[TMP51]], align 8 335 // CHECK1-NEXT: [[TMP52:%.*]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[DOTOFFLOAD_PTRS10]], i32 0, i32 1 336 // CHECK1-NEXT: [[TMP53:%.*]] = bitcast i8** [[TMP52]] to [100 x i32]** 337 // CHECK1-NEXT: store [100 x i32]* @a, [100 x i32]** [[TMP53]], align 8 338 // CHECK1-NEXT: [[TMP54:%.*]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[DOTOFFLOAD_MAPPERS11]], i64 0, i64 1 339 // CHECK1-NEXT: store i8* null, i8** [[TMP54]], align 8 340 // CHECK1-NEXT: [[TMP55:%.*]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[DOTOFFLOAD_BASEPTRS9]], i32 0, i32 0 341 // CHECK1-NEXT: [[TMP56:%.*]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[DOTOFFLOAD_PTRS10]], i32 0, i32 0 342 // CHECK1-NEXT: [[TMP57:%.*]] = load i32, i32* [[N_ADDR]], align 4 343 // CHECK1-NEXT: store i32 [[TMP57]], i32* [[DOTCAPTURE_EXPR_13]], align 4 344 // CHECK1-NEXT: [[TMP58:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_13]], align 4 345 // CHECK1-NEXT: [[SUB15:%.*]] = sub nsw i32 [[TMP58]], 0 346 // CHECK1-NEXT: [[DIV16:%.*]] = sdiv i32 [[SUB15]], 1 347 // CHECK1-NEXT: [[SUB17:%.*]] = sub nsw i32 [[DIV16]], 1 348 // CHECK1-NEXT: store i32 [[SUB17]], i32* [[DOTCAPTURE_EXPR_14]], align 4 349 // CHECK1-NEXT: [[TMP59:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_14]], align 4 350 // CHECK1-NEXT: [[ADD18:%.*]] = add nsw i32 [[TMP59]], 1 351 // CHECK1-NEXT: [[TMP60:%.*]] = zext i32 [[ADD18]] to i64 352 // CHECK1-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 [[TMP60]]) 353 // CHECK1-NEXT: [[TMP61:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z21teams_argument_globali_l36.region_id, i32 2, i8** [[TMP55]], i8** [[TMP56]], i64* getelementptr inbounds ([2 x i64], [2 x i64]* @.offload_sizes.2, i32 0, i32 0), i64* getelementptr inbounds ([2 x i64], [2 x i64]* @.offload_maptypes.3, i32 0, i32 0), i8** null, i8** null, i32 0, i32 1) 354 // CHECK1-NEXT: [[TMP62:%.*]] = icmp ne i32 [[TMP61]], 0 355 // CHECK1-NEXT: br i1 [[TMP62]], label [[OMP_OFFLOAD_FAILED19:%.*]], label [[OMP_OFFLOAD_CONT20:%.*]] 356 // CHECK1: omp_offload.failed19: 357 // CHECK1-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z21teams_argument_globali_l36(i64 [[TMP44]], [100 x i32]* @a) #[[ATTR3]] 358 // CHECK1-NEXT: br label [[OMP_OFFLOAD_CONT20]] 359 // CHECK1: omp_offload.cont20: 360 // CHECK1-NEXT: [[TMP63:%.*]] = load i32, i32* getelementptr inbounds ([100 x i32], [100 x i32]* @a, i64 0, i64 0), align 4 361 // CHECK1-NEXT: ret i32 [[TMP63]] 362 // 363 // 364 // CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z21teams_argument_globali_l30 365 // CHECK1-SAME: (i64 noundef [[TE:%.*]], i64 noundef [[TH:%.*]], [100 x i32]* noundef nonnull align 4 dereferenceable(400) [[A:%.*]], i64 noundef [[I:%.*]], i64 noundef [[N:%.*]]) #[[ATTR1:[0-9]+]] { 366 // CHECK1-NEXT: entry: 367 // CHECK1-NEXT: [[TE_ADDR:%.*]] = alloca i64, align 8 368 // CHECK1-NEXT: [[TH_ADDR:%.*]] = alloca i64, align 8 369 // CHECK1-NEXT: [[A_ADDR:%.*]] = alloca [100 x i32]*, align 8 370 // CHECK1-NEXT: [[I_ADDR:%.*]] = alloca i64, align 8 371 // CHECK1-NEXT: [[N_ADDR:%.*]] = alloca i64, align 8 372 // CHECK1-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB2]]) 373 // CHECK1-NEXT: store i64 [[TE]], i64* [[TE_ADDR]], align 8 374 // CHECK1-NEXT: store i64 [[TH]], i64* [[TH_ADDR]], align 8 375 // CHECK1-NEXT: store [100 x i32]* [[A]], [100 x i32]** [[A_ADDR]], align 8 376 // CHECK1-NEXT: store i64 [[I]], i64* [[I_ADDR]], align 8 377 // CHECK1-NEXT: store i64 [[N]], i64* [[N_ADDR]], align 8 378 // CHECK1-NEXT: [[CONV:%.*]] = bitcast i64* [[TE_ADDR]] to i32* 379 // CHECK1-NEXT: [[CONV1:%.*]] = bitcast i64* [[TH_ADDR]] to i32* 380 // CHECK1-NEXT: [[TMP1:%.*]] = load [100 x i32]*, [100 x i32]** [[A_ADDR]], align 8 381 // CHECK1-NEXT: [[CONV2:%.*]] = bitcast i64* [[I_ADDR]] to i32* 382 // CHECK1-NEXT: [[CONV3:%.*]] = bitcast i64* [[N_ADDR]] to i32* 383 // CHECK1-NEXT: [[TMP2:%.*]] = load i32, i32* [[CONV]], align 4 384 // CHECK1-NEXT: [[TMP3:%.*]] = load i32, i32* [[CONV1]], align 4 385 // CHECK1-NEXT: call void @__kmpc_push_num_teams(%struct.ident_t* @[[GLOB2]], i32 [[TMP0]], i32 [[TMP2]], i32 [[TMP3]]) 386 // CHECK1-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2]], i32 3, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32*, i32*, [100 x i32]*)* @.omp_outlined. to void (i32*, i32*, ...)*), i32* [[CONV2]], i32* [[CONV3]], [100 x i32]* [[TMP1]]) 387 // CHECK1-NEXT: ret void 388 // 389 // 390 // CHECK1-LABEL: define {{[^@]+}}@.omp_outlined. 391 // CHECK1-SAME: (i32* noalias noundef [[DOTGLOBAL_TID_:%.*]], i32* noalias noundef [[DOTBOUND_TID_:%.*]], i32* noundef nonnull align 4 dereferenceable(4) [[I:%.*]], i32* noundef nonnull align 4 dereferenceable(4) [[N:%.*]], [100 x i32]* noundef nonnull align 4 dereferenceable(400) [[A:%.*]]) #[[ATTR1]] { 392 // CHECK1-NEXT: entry: 393 // CHECK1-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 394 // CHECK1-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 395 // CHECK1-NEXT: [[I_ADDR:%.*]] = alloca i32*, align 8 396 // CHECK1-NEXT: [[N_ADDR:%.*]] = alloca i32*, align 8 397 // CHECK1-NEXT: [[A_ADDR:%.*]] = alloca [100 x i32]*, align 8 398 // CHECK1-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 399 // CHECK1-NEXT: [[TMP:%.*]] = alloca i32, align 4 400 // CHECK1-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 401 // CHECK1-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 402 // CHECK1-NEXT: [[I3:%.*]] = alloca i32, align 4 403 // CHECK1-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 404 // CHECK1-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4 405 // CHECK1-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 406 // CHECK1-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 407 // CHECK1-NEXT: [[I4:%.*]] = alloca i32, align 4 408 // CHECK1-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8 409 // CHECK1-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8 410 // CHECK1-NEXT: store i32* [[I]], i32** [[I_ADDR]], align 8 411 // CHECK1-NEXT: store i32* [[N]], i32** [[N_ADDR]], align 8 412 // CHECK1-NEXT: store [100 x i32]* [[A]], [100 x i32]** [[A_ADDR]], align 8 413 // CHECK1-NEXT: [[TMP0:%.*]] = load i32*, i32** [[I_ADDR]], align 8 414 // CHECK1-NEXT: [[TMP1:%.*]] = load i32*, i32** [[N_ADDR]], align 8 415 // CHECK1-NEXT: [[TMP2:%.*]] = load [100 x i32]*, [100 x i32]** [[A_ADDR]], align 8 416 // CHECK1-NEXT: [[TMP3:%.*]] = load i32, i32* [[TMP1]], align 4 417 // CHECK1-NEXT: store i32 [[TMP3]], i32* [[DOTCAPTURE_EXPR_]], align 4 418 // CHECK1-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 419 // CHECK1-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP4]], 0 420 // CHECK1-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 421 // CHECK1-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1 422 // CHECK1-NEXT: store i32 [[SUB2]], i32* [[DOTCAPTURE_EXPR_1]], align 4 423 // CHECK1-NEXT: store i32 0, i32* [[I3]], align 4 424 // CHECK1-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 425 // CHECK1-NEXT: [[CMP:%.*]] = icmp slt i32 0, [[TMP5]] 426 // CHECK1-NEXT: br i1 [[CMP]], label [[OMP_PRECOND_THEN:%.*]], label [[OMP_PRECOND_END:%.*]] 427 // CHECK1: omp.precond.then: 428 // CHECK1-NEXT: [[ARRAYDECAY:%.*]] = getelementptr inbounds [100 x i32], [100 x i32]* [[TMP2]], i64 0, i64 0 429 // CHECK1-NEXT: call void @llvm.assume(i1 true) [ "align"(i32* [[ARRAYDECAY]], i64 16) ] 430 // CHECK1-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4 431 // CHECK1-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 432 // CHECK1-NEXT: store i32 [[TMP6]], i32* [[DOTOMP_UB]], align 4 433 // CHECK1-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4 434 // CHECK1-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4 435 // CHECK1-NEXT: [[TMP7:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 436 // CHECK1-NEXT: [[TMP8:%.*]] = load i32, i32* [[TMP7]], align 4 437 // CHECK1-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1:[0-9]+]], i32 [[TMP8]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1) 438 // CHECK1-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 439 // CHECK1-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 440 // CHECK1-NEXT: [[CMP5:%.*]] = icmp sgt i32 [[TMP9]], [[TMP10]] 441 // CHECK1-NEXT: br i1 [[CMP5]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] 442 // CHECK1: cond.true: 443 // CHECK1-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 444 // CHECK1-NEXT: br label [[COND_END:%.*]] 445 // CHECK1: cond.false: 446 // CHECK1-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 447 // CHECK1-NEXT: br label [[COND_END]] 448 // CHECK1: cond.end: 449 // CHECK1-NEXT: [[COND:%.*]] = phi i32 [ [[TMP11]], [[COND_TRUE]] ], [ [[TMP12]], [[COND_FALSE]] ] 450 // CHECK1-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4 451 // CHECK1-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4 452 // CHECK1-NEXT: store i32 [[TMP13]], i32* [[DOTOMP_IV]], align 4 453 // CHECK1-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] 454 // CHECK1: omp.inner.for.cond: 455 // CHECK1-NEXT: [[TMP14:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !5 456 // CHECK1-NEXT: [[TMP15:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4, !llvm.access.group !5 457 // CHECK1-NEXT: [[CMP6:%.*]] = icmp sle i32 [[TMP14]], [[TMP15]] 458 // CHECK1-NEXT: br i1 [[CMP6]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] 459 // CHECK1: omp.inner.for.body: 460 // CHECK1-NEXT: [[TMP16:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !5 461 // CHECK1-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP16]], 1 462 // CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] 463 // CHECK1-NEXT: store i32 [[ADD]], i32* [[I4]], align 4, !llvm.access.group !5 464 // CHECK1-NEXT: [[TMP17:%.*]] = load i32, i32* [[I4]], align 4, !llvm.access.group !5 465 // CHECK1-NEXT: [[IDXPROM:%.*]] = sext i32 [[TMP17]] to i64 466 // CHECK1-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [100 x i32], [100 x i32]* [[TMP2]], i64 0, i64 [[IDXPROM]] 467 // CHECK1-NEXT: store i32 0, i32* [[ARRAYIDX]], align 4, !llvm.access.group !5 468 // CHECK1-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] 469 // CHECK1: omp.body.continue: 470 // CHECK1-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] 471 // CHECK1: omp.inner.for.inc: 472 // CHECK1-NEXT: [[TMP18:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !5 473 // CHECK1-NEXT: [[ADD7:%.*]] = add nsw i32 [[TMP18]], 1 474 // CHECK1-NEXT: store i32 [[ADD7]], i32* [[DOTOMP_IV]], align 4, !llvm.access.group !5 475 // CHECK1-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP6:![0-9]+]] 476 // CHECK1: omp.inner.for.end: 477 // CHECK1-NEXT: br label [[OMP_LOOP_EXIT:%.*]] 478 // CHECK1: omp.loop.exit: 479 // CHECK1-NEXT: [[TMP19:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 480 // CHECK1-NEXT: [[TMP20:%.*]] = load i32, i32* [[TMP19]], align 4 481 // CHECK1-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP20]]) 482 // CHECK1-NEXT: [[TMP21:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4 483 // CHECK1-NEXT: [[TMP22:%.*]] = icmp ne i32 [[TMP21]], 0 484 // CHECK1-NEXT: br i1 [[TMP22]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]] 485 // CHECK1: .omp.final.then: 486 // CHECK1-NEXT: [[TMP23:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 487 // CHECK1-NEXT: [[SUB8:%.*]] = sub nsw i32 [[TMP23]], 0 488 // CHECK1-NEXT: [[DIV9:%.*]] = sdiv i32 [[SUB8]], 1 489 // CHECK1-NEXT: [[MUL10:%.*]] = mul nsw i32 [[DIV9]], 1 490 // CHECK1-NEXT: [[ADD11:%.*]] = add nsw i32 0, [[MUL10]] 491 // CHECK1-NEXT: store i32 [[ADD11]], i32* [[TMP0]], align 4 492 // CHECK1-NEXT: br label [[DOTOMP_FINAL_DONE]] 493 // CHECK1: .omp.final.done: 494 // CHECK1-NEXT: br label [[OMP_PRECOND_END]] 495 // CHECK1: omp.precond.end: 496 // CHECK1-NEXT: ret void 497 // 498 // 499 // CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z21teams_argument_globali_l36 500 // CHECK1-SAME: (i64 noundef [[N:%.*]], [100 x i32]* noundef nonnull align 4 dereferenceable(400) [[A:%.*]]) #[[ATTR1]] { 501 // CHECK1-NEXT: entry: 502 // CHECK1-NEXT: [[N_ADDR:%.*]] = alloca i64, align 8 503 // CHECK1-NEXT: [[A_ADDR:%.*]] = alloca [100 x i32]*, align 8 504 // CHECK1-NEXT: store i64 [[N]], i64* [[N_ADDR]], align 8 505 // CHECK1-NEXT: store [100 x i32]* [[A]], [100 x i32]** [[A_ADDR]], align 8 506 // CHECK1-NEXT: [[CONV:%.*]] = bitcast i64* [[N_ADDR]] to i32* 507 // CHECK1-NEXT: [[TMP0:%.*]] = load [100 x i32]*, [100 x i32]** [[A_ADDR]], align 8 508 // CHECK1-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2]], i32 2, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32*, [100 x i32]*)* @.omp_outlined..1 to void (i32*, i32*, ...)*), i32* [[CONV]], [100 x i32]* [[TMP0]]) 509 // CHECK1-NEXT: ret void 510 // 511 // 512 // CHECK1-LABEL: define {{[^@]+}}@.omp_outlined..1 513 // CHECK1-SAME: (i32* noalias noundef [[DOTGLOBAL_TID_:%.*]], i32* noalias noundef [[DOTBOUND_TID_:%.*]], i32* noundef nonnull align 4 dereferenceable(4) [[N:%.*]], [100 x i32]* noundef nonnull align 4 dereferenceable(400) [[A:%.*]]) #[[ATTR1]] { 514 // CHECK1-NEXT: entry: 515 // CHECK1-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 516 // CHECK1-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 517 // CHECK1-NEXT: [[N_ADDR:%.*]] = alloca i32*, align 8 518 // CHECK1-NEXT: [[A_ADDR:%.*]] = alloca [100 x i32]*, align 8 519 // CHECK1-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 520 // CHECK1-NEXT: [[TMP:%.*]] = alloca i32, align 4 521 // CHECK1-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 522 // CHECK1-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 523 // CHECK1-NEXT: [[I:%.*]] = alloca i32, align 4 524 // CHECK1-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 525 // CHECK1-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4 526 // CHECK1-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 527 // CHECK1-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 528 // CHECK1-NEXT: [[I3:%.*]] = alloca i32, align 4 529 // CHECK1-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8 530 // CHECK1-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8 531 // CHECK1-NEXT: store i32* [[N]], i32** [[N_ADDR]], align 8 532 // CHECK1-NEXT: store [100 x i32]* [[A]], [100 x i32]** [[A_ADDR]], align 8 533 // CHECK1-NEXT: [[TMP0:%.*]] = load i32*, i32** [[N_ADDR]], align 8 534 // CHECK1-NEXT: [[TMP1:%.*]] = load [100 x i32]*, [100 x i32]** [[A_ADDR]], align 8 535 // CHECK1-NEXT: [[TMP2:%.*]] = load i32, i32* [[TMP0]], align 4 536 // CHECK1-NEXT: store i32 [[TMP2]], i32* [[DOTCAPTURE_EXPR_]], align 4 537 // CHECK1-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 538 // CHECK1-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP3]], 0 539 // CHECK1-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 540 // CHECK1-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1 541 // CHECK1-NEXT: store i32 [[SUB2]], i32* [[DOTCAPTURE_EXPR_1]], align 4 542 // CHECK1-NEXT: store i32 0, i32* [[I]], align 4 543 // CHECK1-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 544 // CHECK1-NEXT: [[CMP:%.*]] = icmp slt i32 0, [[TMP4]] 545 // CHECK1-NEXT: br i1 [[CMP]], label [[OMP_PRECOND_THEN:%.*]], label [[OMP_PRECOND_END:%.*]] 546 // CHECK1: omp.precond.then: 547 // CHECK1-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4 548 // CHECK1-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 549 // CHECK1-NEXT: store i32 [[TMP5]], i32* [[DOTOMP_UB]], align 4 550 // CHECK1-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4 551 // CHECK1-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4 552 // CHECK1-NEXT: [[TMP6:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 553 // CHECK1-NEXT: [[TMP7:%.*]] = load i32, i32* [[TMP6]], align 4 554 // CHECK1-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1]], i32 [[TMP7]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1) 555 // CHECK1-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 556 // CHECK1-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 557 // CHECK1-NEXT: [[CMP4:%.*]] = icmp sgt i32 [[TMP8]], [[TMP9]] 558 // CHECK1-NEXT: br i1 [[CMP4]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] 559 // CHECK1: cond.true: 560 // CHECK1-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 561 // CHECK1-NEXT: br label [[COND_END:%.*]] 562 // CHECK1: cond.false: 563 // CHECK1-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 564 // CHECK1-NEXT: br label [[COND_END]] 565 // CHECK1: cond.end: 566 // CHECK1-NEXT: [[COND:%.*]] = phi i32 [ [[TMP10]], [[COND_TRUE]] ], [ [[TMP11]], [[COND_FALSE]] ] 567 // CHECK1-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4 568 // CHECK1-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4 569 // CHECK1-NEXT: store i32 [[TMP12]], i32* [[DOTOMP_IV]], align 4 570 // CHECK1-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] 571 // CHECK1: omp.inner.for.cond: 572 // CHECK1-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 573 // CHECK1-NEXT: [[TMP14:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 574 // CHECK1-NEXT: [[CMP5:%.*]] = icmp sle i32 [[TMP13]], [[TMP14]] 575 // CHECK1-NEXT: br i1 [[CMP5]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] 576 // CHECK1: omp.inner.for.body: 577 // CHECK1-NEXT: [[TMP15:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 578 // CHECK1-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP15]], 1 579 // CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] 580 // CHECK1-NEXT: store i32 [[ADD]], i32* [[I3]], align 4 581 // CHECK1-NEXT: [[TMP16:%.*]] = load i32, i32* [[I3]], align 4 582 // CHECK1-NEXT: [[IDXPROM:%.*]] = sext i32 [[TMP16]] to i64 583 // CHECK1-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [100 x i32], [100 x i32]* [[TMP1]], i64 0, i64 [[IDXPROM]] 584 // CHECK1-NEXT: store i32 0, i32* [[ARRAYIDX]], align 4 585 // CHECK1-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] 586 // CHECK1: omp.body.continue: 587 // CHECK1-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] 588 // CHECK1: omp.inner.for.inc: 589 // CHECK1-NEXT: [[TMP17:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 590 // CHECK1-NEXT: [[ADD6:%.*]] = add nsw i32 [[TMP17]], 1 591 // CHECK1-NEXT: store i32 [[ADD6]], i32* [[DOTOMP_IV]], align 4 592 // CHECK1-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP12:![0-9]+]] 593 // CHECK1: omp.inner.for.end: 594 // CHECK1-NEXT: br label [[OMP_LOOP_EXIT:%.*]] 595 // CHECK1: omp.loop.exit: 596 // CHECK1-NEXT: [[TMP18:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 597 // CHECK1-NEXT: [[TMP19:%.*]] = load i32, i32* [[TMP18]], align 4 598 // CHECK1-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP19]]) 599 // CHECK1-NEXT: [[TMP20:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4 600 // CHECK1-NEXT: [[TMP21:%.*]] = icmp ne i32 [[TMP20]], 0 601 // CHECK1-NEXT: br i1 [[TMP21]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]] 602 // CHECK1: .omp.final.then: 603 // CHECK1-NEXT: [[TMP22:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 604 // CHECK1-NEXT: [[SUB7:%.*]] = sub nsw i32 [[TMP22]], 0 605 // CHECK1-NEXT: [[DIV8:%.*]] = sdiv i32 [[SUB7]], 1 606 // CHECK1-NEXT: [[MUL9:%.*]] = mul nsw i32 [[DIV8]], 1 607 // CHECK1-NEXT: [[ADD10:%.*]] = add nsw i32 0, [[MUL9]] 608 // CHECK1-NEXT: store i32 [[ADD10]], i32* [[I3]], align 4 609 // CHECK1-NEXT: br label [[DOTOMP_FINAL_DONE]] 610 // CHECK1: .omp.final.done: 611 // CHECK1-NEXT: br label [[OMP_PRECOND_END]] 612 // CHECK1: omp.precond.end: 613 // CHECK1-NEXT: ret void 614 // 615 // 616 // CHECK1-LABEL: define {{[^@]+}}@.omp_offloading.requires_reg 617 // CHECK1-SAME: () #[[ATTR4:[0-9]+]] { 618 // CHECK1-NEXT: entry: 619 // CHECK1-NEXT: call void @__tgt_register_requires(i64 1) 620 // CHECK1-NEXT: ret void 621 // 622 // 623 // CHECK3-LABEL: define {{[^@]+}}@_Z21teams_argument_globali 624 // CHECK3-SAME: (i32 noundef [[N:%.*]]) #[[ATTR0:[0-9]+]] { 625 // CHECK3-NEXT: entry: 626 // CHECK3-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4 627 // CHECK3-NEXT: [[I:%.*]] = alloca i32, align 4 628 // CHECK3-NEXT: [[TE:%.*]] = alloca i32, align 4 629 // CHECK3-NEXT: [[TH:%.*]] = alloca i32, align 4 630 // CHECK3-NEXT: [[TE_CASTED:%.*]] = alloca i32, align 4 631 // CHECK3-NEXT: [[TH_CASTED:%.*]] = alloca i32, align 4 632 // CHECK3-NEXT: [[I_CASTED:%.*]] = alloca i32, align 4 633 // CHECK3-NEXT: [[N_CASTED:%.*]] = alloca i32, align 4 634 // CHECK3-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [5 x i8*], align 4 635 // CHECK3-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [5 x i8*], align 4 636 // CHECK3-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [5 x i8*], align 4 637 // CHECK3-NEXT: [[TMP:%.*]] = alloca i32, align 4 638 // CHECK3-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 639 // CHECK3-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 640 // CHECK3-NEXT: [[N_CASTED4:%.*]] = alloca i32, align 4 641 // CHECK3-NEXT: [[DOTOFFLOAD_BASEPTRS5:%.*]] = alloca [2 x i8*], align 4 642 // CHECK3-NEXT: [[DOTOFFLOAD_PTRS6:%.*]] = alloca [2 x i8*], align 4 643 // CHECK3-NEXT: [[DOTOFFLOAD_MAPPERS7:%.*]] = alloca [2 x i8*], align 4 644 // CHECK3-NEXT: [[_TMP8:%.*]] = alloca i32, align 4 645 // CHECK3-NEXT: [[DOTCAPTURE_EXPR_9:%.*]] = alloca i32, align 4 646 // CHECK3-NEXT: [[DOTCAPTURE_EXPR_10:%.*]] = alloca i32, align 4 647 // CHECK3-NEXT: store i32 [[N]], i32* [[N_ADDR]], align 4 648 // CHECK3-NEXT: [[TMP0:%.*]] = load i32, i32* [[N_ADDR]], align 4 649 // CHECK3-NEXT: [[DIV:%.*]] = sdiv i32 [[TMP0]], 128 650 // CHECK3-NEXT: store i32 [[DIV]], i32* [[TE]], align 4 651 // CHECK3-NEXT: store i32 128, i32* [[TH]], align 4 652 // CHECK3-NEXT: [[TMP1:%.*]] = load i32, i32* [[TE]], align 4 653 // CHECK3-NEXT: store i32 [[TMP1]], i32* [[TE_CASTED]], align 4 654 // CHECK3-NEXT: [[TMP2:%.*]] = load i32, i32* [[TE_CASTED]], align 4 655 // CHECK3-NEXT: [[TMP3:%.*]] = load i32, i32* [[TH]], align 4 656 // CHECK3-NEXT: store i32 [[TMP3]], i32* [[TH_CASTED]], align 4 657 // CHECK3-NEXT: [[TMP4:%.*]] = load i32, i32* [[TH_CASTED]], align 4 658 // CHECK3-NEXT: [[TMP5:%.*]] = load i32, i32* [[I]], align 4 659 // CHECK3-NEXT: store i32 [[TMP5]], i32* [[I_CASTED]], align 4 660 // CHECK3-NEXT: [[TMP6:%.*]] = load i32, i32* [[I_CASTED]], align 4 661 // CHECK3-NEXT: [[TMP7:%.*]] = load i32, i32* [[N_ADDR]], align 4 662 // CHECK3-NEXT: store i32 [[TMP7]], i32* [[N_CASTED]], align 4 663 // CHECK3-NEXT: [[TMP8:%.*]] = load i32, i32* [[N_CASTED]], align 4 664 // CHECK3-NEXT: [[TMP9:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 665 // CHECK3-NEXT: [[TMP10:%.*]] = bitcast i8** [[TMP9]] to i32* 666 // CHECK3-NEXT: store i32 [[TMP2]], i32* [[TMP10]], align 4 667 // CHECK3-NEXT: [[TMP11:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 668 // CHECK3-NEXT: [[TMP12:%.*]] = bitcast i8** [[TMP11]] to i32* 669 // CHECK3-NEXT: store i32 [[TMP2]], i32* [[TMP12]], align 4 670 // CHECK3-NEXT: [[TMP13:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0 671 // CHECK3-NEXT: store i8* null, i8** [[TMP13]], align 4 672 // CHECK3-NEXT: [[TMP14:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1 673 // CHECK3-NEXT: [[TMP15:%.*]] = bitcast i8** [[TMP14]] to i32* 674 // CHECK3-NEXT: store i32 [[TMP4]], i32* [[TMP15]], align 4 675 // CHECK3-NEXT: [[TMP16:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1 676 // CHECK3-NEXT: [[TMP17:%.*]] = bitcast i8** [[TMP16]] to i32* 677 // CHECK3-NEXT: store i32 [[TMP4]], i32* [[TMP17]], align 4 678 // CHECK3-NEXT: [[TMP18:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 1 679 // CHECK3-NEXT: store i8* null, i8** [[TMP18]], align 4 680 // CHECK3-NEXT: [[TMP19:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2 681 // CHECK3-NEXT: [[TMP20:%.*]] = bitcast i8** [[TMP19]] to [100 x i32]** 682 // CHECK3-NEXT: store [100 x i32]* @a, [100 x i32]** [[TMP20]], align 4 683 // CHECK3-NEXT: [[TMP21:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2 684 // CHECK3-NEXT: [[TMP22:%.*]] = bitcast i8** [[TMP21]] to [100 x i32]** 685 // CHECK3-NEXT: store [100 x i32]* @a, [100 x i32]** [[TMP22]], align 4 686 // CHECK3-NEXT: [[TMP23:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2 687 // CHECK3-NEXT: store i8* null, i8** [[TMP23]], align 4 688 // CHECK3-NEXT: [[TMP24:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 3 689 // CHECK3-NEXT: [[TMP25:%.*]] = bitcast i8** [[TMP24]] to i32* 690 // CHECK3-NEXT: store i32 [[TMP6]], i32* [[TMP25]], align 4 691 // CHECK3-NEXT: [[TMP26:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 3 692 // CHECK3-NEXT: [[TMP27:%.*]] = bitcast i8** [[TMP26]] to i32* 693 // CHECK3-NEXT: store i32 [[TMP6]], i32* [[TMP27]], align 4 694 // CHECK3-NEXT: [[TMP28:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 3 695 // CHECK3-NEXT: store i8* null, i8** [[TMP28]], align 4 696 // CHECK3-NEXT: [[TMP29:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 4 697 // CHECK3-NEXT: [[TMP30:%.*]] = bitcast i8** [[TMP29]] to i32* 698 // CHECK3-NEXT: store i32 [[TMP8]], i32* [[TMP30]], align 4 699 // CHECK3-NEXT: [[TMP31:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 4 700 // CHECK3-NEXT: [[TMP32:%.*]] = bitcast i8** [[TMP31]] to i32* 701 // CHECK3-NEXT: store i32 [[TMP8]], i32* [[TMP32]], align 4 702 // CHECK3-NEXT: [[TMP33:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 4 703 // CHECK3-NEXT: store i8* null, i8** [[TMP33]], align 4 704 // CHECK3-NEXT: [[TMP34:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 705 // CHECK3-NEXT: [[TMP35:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 706 // CHECK3-NEXT: [[TMP36:%.*]] = load i32, i32* [[TE]], align 4 707 // CHECK3-NEXT: [[TMP37:%.*]] = load i32, i32* [[N_ADDR]], align 4 708 // CHECK3-NEXT: store i32 [[TMP37]], i32* [[DOTCAPTURE_EXPR_]], align 4 709 // CHECK3-NEXT: [[TMP38:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 710 // CHECK3-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP38]], 0 711 // CHECK3-NEXT: [[DIV2:%.*]] = sdiv i32 [[SUB]], 1 712 // CHECK3-NEXT: [[SUB3:%.*]] = sub nsw i32 [[DIV2]], 1 713 // CHECK3-NEXT: store i32 [[SUB3]], i32* [[DOTCAPTURE_EXPR_1]], align 4 714 // CHECK3-NEXT: [[TMP39:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 715 // CHECK3-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP39]], 1 716 // CHECK3-NEXT: [[TMP40:%.*]] = zext i32 [[ADD]] to i64 717 // CHECK3-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2:[0-9]+]], i64 -1, i64 [[TMP40]]) 718 // CHECK3-NEXT: [[TMP41:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z21teams_argument_globali_l30.region_id, i32 5, i8** [[TMP34]], i8** [[TMP35]], i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_sizes, i32 0, i32 0), i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 [[TMP36]], i32 1) 719 // CHECK3-NEXT: [[TMP42:%.*]] = icmp ne i32 [[TMP41]], 0 720 // CHECK3-NEXT: br i1 [[TMP42]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] 721 // CHECK3: omp_offload.failed: 722 // CHECK3-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z21teams_argument_globali_l30(i32 [[TMP2]], i32 [[TMP4]], [100 x i32]* @a, i32 [[TMP6]], i32 [[TMP8]]) #[[ATTR3:[0-9]+]] 723 // CHECK3-NEXT: br label [[OMP_OFFLOAD_CONT]] 724 // CHECK3: omp_offload.cont: 725 // CHECK3-NEXT: [[TMP43:%.*]] = load i32, i32* [[N_ADDR]], align 4 726 // CHECK3-NEXT: store i32 [[TMP43]], i32* [[N_CASTED4]], align 4 727 // CHECK3-NEXT: [[TMP44:%.*]] = load i32, i32* [[N_CASTED4]], align 4 728 // CHECK3-NEXT: [[TMP45:%.*]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[DOTOFFLOAD_BASEPTRS5]], i32 0, i32 0 729 // CHECK3-NEXT: [[TMP46:%.*]] = bitcast i8** [[TMP45]] to i32* 730 // CHECK3-NEXT: store i32 [[TMP44]], i32* [[TMP46]], align 4 731 // CHECK3-NEXT: [[TMP47:%.*]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[DOTOFFLOAD_PTRS6]], i32 0, i32 0 732 // CHECK3-NEXT: [[TMP48:%.*]] = bitcast i8** [[TMP47]] to i32* 733 // CHECK3-NEXT: store i32 [[TMP44]], i32* [[TMP48]], align 4 734 // CHECK3-NEXT: [[TMP49:%.*]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[DOTOFFLOAD_MAPPERS7]], i32 0, i32 0 735 // CHECK3-NEXT: store i8* null, i8** [[TMP49]], align 4 736 // CHECK3-NEXT: [[TMP50:%.*]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[DOTOFFLOAD_BASEPTRS5]], i32 0, i32 1 737 // CHECK3-NEXT: [[TMP51:%.*]] = bitcast i8** [[TMP50]] to [100 x i32]** 738 // CHECK3-NEXT: store [100 x i32]* @a, [100 x i32]** [[TMP51]], align 4 739 // CHECK3-NEXT: [[TMP52:%.*]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[DOTOFFLOAD_PTRS6]], i32 0, i32 1 740 // CHECK3-NEXT: [[TMP53:%.*]] = bitcast i8** [[TMP52]] to [100 x i32]** 741 // CHECK3-NEXT: store [100 x i32]* @a, [100 x i32]** [[TMP53]], align 4 742 // CHECK3-NEXT: [[TMP54:%.*]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[DOTOFFLOAD_MAPPERS7]], i32 0, i32 1 743 // CHECK3-NEXT: store i8* null, i8** [[TMP54]], align 4 744 // CHECK3-NEXT: [[TMP55:%.*]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[DOTOFFLOAD_BASEPTRS5]], i32 0, i32 0 745 // CHECK3-NEXT: [[TMP56:%.*]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[DOTOFFLOAD_PTRS6]], i32 0, i32 0 746 // CHECK3-NEXT: [[TMP57:%.*]] = load i32, i32* [[N_ADDR]], align 4 747 // CHECK3-NEXT: store i32 [[TMP57]], i32* [[DOTCAPTURE_EXPR_9]], align 4 748 // CHECK3-NEXT: [[TMP58:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_9]], align 4 749 // CHECK3-NEXT: [[SUB11:%.*]] = sub nsw i32 [[TMP58]], 0 750 // CHECK3-NEXT: [[DIV12:%.*]] = sdiv i32 [[SUB11]], 1 751 // CHECK3-NEXT: [[SUB13:%.*]] = sub nsw i32 [[DIV12]], 1 752 // CHECK3-NEXT: store i32 [[SUB13]], i32* [[DOTCAPTURE_EXPR_10]], align 4 753 // CHECK3-NEXT: [[TMP59:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_10]], align 4 754 // CHECK3-NEXT: [[ADD14:%.*]] = add nsw i32 [[TMP59]], 1 755 // CHECK3-NEXT: [[TMP60:%.*]] = zext i32 [[ADD14]] to i64 756 // CHECK3-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 [[TMP60]]) 757 // CHECK3-NEXT: [[TMP61:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z21teams_argument_globali_l36.region_id, i32 2, i8** [[TMP55]], i8** [[TMP56]], i64* getelementptr inbounds ([2 x i64], [2 x i64]* @.offload_sizes.2, i32 0, i32 0), i64* getelementptr inbounds ([2 x i64], [2 x i64]* @.offload_maptypes.3, i32 0, i32 0), i8** null, i8** null, i32 0, i32 1) 758 // CHECK3-NEXT: [[TMP62:%.*]] = icmp ne i32 [[TMP61]], 0 759 // CHECK3-NEXT: br i1 [[TMP62]], label [[OMP_OFFLOAD_FAILED15:%.*]], label [[OMP_OFFLOAD_CONT16:%.*]] 760 // CHECK3: omp_offload.failed15: 761 // CHECK3-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z21teams_argument_globali_l36(i32 [[TMP44]], [100 x i32]* @a) #[[ATTR3]] 762 // CHECK3-NEXT: br label [[OMP_OFFLOAD_CONT16]] 763 // CHECK3: omp_offload.cont16: 764 // CHECK3-NEXT: [[TMP63:%.*]] = load i32, i32* getelementptr inbounds ([100 x i32], [100 x i32]* @a, i32 0, i32 0), align 4 765 // CHECK3-NEXT: ret i32 [[TMP63]] 766 // 767 // 768 // CHECK3-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z21teams_argument_globali_l30 769 // CHECK3-SAME: (i32 noundef [[TE:%.*]], i32 noundef [[TH:%.*]], [100 x i32]* noundef nonnull align 4 dereferenceable(400) [[A:%.*]], i32 noundef [[I:%.*]], i32 noundef [[N:%.*]]) #[[ATTR1:[0-9]+]] { 770 // CHECK3-NEXT: entry: 771 // CHECK3-NEXT: [[TE_ADDR:%.*]] = alloca i32, align 4 772 // CHECK3-NEXT: [[TH_ADDR:%.*]] = alloca i32, align 4 773 // CHECK3-NEXT: [[A_ADDR:%.*]] = alloca [100 x i32]*, align 4 774 // CHECK3-NEXT: [[I_ADDR:%.*]] = alloca i32, align 4 775 // CHECK3-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4 776 // CHECK3-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB2]]) 777 // CHECK3-NEXT: store i32 [[TE]], i32* [[TE_ADDR]], align 4 778 // CHECK3-NEXT: store i32 [[TH]], i32* [[TH_ADDR]], align 4 779 // CHECK3-NEXT: store [100 x i32]* [[A]], [100 x i32]** [[A_ADDR]], align 4 780 // CHECK3-NEXT: store i32 [[I]], i32* [[I_ADDR]], align 4 781 // CHECK3-NEXT: store i32 [[N]], i32* [[N_ADDR]], align 4 782 // CHECK3-NEXT: [[TMP1:%.*]] = load [100 x i32]*, [100 x i32]** [[A_ADDR]], align 4 783 // CHECK3-NEXT: [[TMP2:%.*]] = load i32, i32* [[TE_ADDR]], align 4 784 // CHECK3-NEXT: [[TMP3:%.*]] = load i32, i32* [[TH_ADDR]], align 4 785 // CHECK3-NEXT: call void @__kmpc_push_num_teams(%struct.ident_t* @[[GLOB2]], i32 [[TMP0]], i32 [[TMP2]], i32 [[TMP3]]) 786 // CHECK3-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2]], i32 3, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32*, i32*, [100 x i32]*)* @.omp_outlined. to void (i32*, i32*, ...)*), i32* [[I_ADDR]], i32* [[N_ADDR]], [100 x i32]* [[TMP1]]) 787 // CHECK3-NEXT: ret void 788 // 789 // 790 // CHECK3-LABEL: define {{[^@]+}}@.omp_outlined. 791 // CHECK3-SAME: (i32* noalias noundef [[DOTGLOBAL_TID_:%.*]], i32* noalias noundef [[DOTBOUND_TID_:%.*]], i32* noundef nonnull align 4 dereferenceable(4) [[I:%.*]], i32* noundef nonnull align 4 dereferenceable(4) [[N:%.*]], [100 x i32]* noundef nonnull align 4 dereferenceable(400) [[A:%.*]]) #[[ATTR1]] { 792 // CHECK3-NEXT: entry: 793 // CHECK3-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4 794 // CHECK3-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4 795 // CHECK3-NEXT: [[I_ADDR:%.*]] = alloca i32*, align 4 796 // CHECK3-NEXT: [[N_ADDR:%.*]] = alloca i32*, align 4 797 // CHECK3-NEXT: [[A_ADDR:%.*]] = alloca [100 x i32]*, align 4 798 // CHECK3-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 799 // CHECK3-NEXT: [[TMP:%.*]] = alloca i32, align 4 800 // CHECK3-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 801 // CHECK3-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 802 // CHECK3-NEXT: [[I3:%.*]] = alloca i32, align 4 803 // CHECK3-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 804 // CHECK3-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4 805 // CHECK3-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 806 // CHECK3-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 807 // CHECK3-NEXT: [[I4:%.*]] = alloca i32, align 4 808 // CHECK3-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 4 809 // CHECK3-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 4 810 // CHECK3-NEXT: store i32* [[I]], i32** [[I_ADDR]], align 4 811 // CHECK3-NEXT: store i32* [[N]], i32** [[N_ADDR]], align 4 812 // CHECK3-NEXT: store [100 x i32]* [[A]], [100 x i32]** [[A_ADDR]], align 4 813 // CHECK3-NEXT: [[TMP0:%.*]] = load i32*, i32** [[I_ADDR]], align 4 814 // CHECK3-NEXT: [[TMP1:%.*]] = load i32*, i32** [[N_ADDR]], align 4 815 // CHECK3-NEXT: [[TMP2:%.*]] = load [100 x i32]*, [100 x i32]** [[A_ADDR]], align 4 816 // CHECK3-NEXT: [[TMP3:%.*]] = load i32, i32* [[TMP1]], align 4 817 // CHECK3-NEXT: store i32 [[TMP3]], i32* [[DOTCAPTURE_EXPR_]], align 4 818 // CHECK3-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 819 // CHECK3-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP4]], 0 820 // CHECK3-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 821 // CHECK3-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1 822 // CHECK3-NEXT: store i32 [[SUB2]], i32* [[DOTCAPTURE_EXPR_1]], align 4 823 // CHECK3-NEXT: store i32 0, i32* [[I3]], align 4 824 // CHECK3-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 825 // CHECK3-NEXT: [[CMP:%.*]] = icmp slt i32 0, [[TMP5]] 826 // CHECK3-NEXT: br i1 [[CMP]], label [[OMP_PRECOND_THEN:%.*]], label [[OMP_PRECOND_END:%.*]] 827 // CHECK3: omp.precond.then: 828 // CHECK3-NEXT: [[ARRAYDECAY:%.*]] = getelementptr inbounds [100 x i32], [100 x i32]* [[TMP2]], i32 0, i32 0 829 // CHECK3-NEXT: call void @llvm.assume(i1 true) [ "align"(i32* [[ARRAYDECAY]], i32 16) ] 830 // CHECK3-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4 831 // CHECK3-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 832 // CHECK3-NEXT: store i32 [[TMP6]], i32* [[DOTOMP_UB]], align 4 833 // CHECK3-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4 834 // CHECK3-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4 835 // CHECK3-NEXT: [[TMP7:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 4 836 // CHECK3-NEXT: [[TMP8:%.*]] = load i32, i32* [[TMP7]], align 4 837 // CHECK3-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1:[0-9]+]], i32 [[TMP8]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1) 838 // CHECK3-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 839 // CHECK3-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 840 // CHECK3-NEXT: [[CMP5:%.*]] = icmp sgt i32 [[TMP9]], [[TMP10]] 841 // CHECK3-NEXT: br i1 [[CMP5]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] 842 // CHECK3: cond.true: 843 // CHECK3-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 844 // CHECK3-NEXT: br label [[COND_END:%.*]] 845 // CHECK3: cond.false: 846 // CHECK3-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 847 // CHECK3-NEXT: br label [[COND_END]] 848 // CHECK3: cond.end: 849 // CHECK3-NEXT: [[COND:%.*]] = phi i32 [ [[TMP11]], [[COND_TRUE]] ], [ [[TMP12]], [[COND_FALSE]] ] 850 // CHECK3-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4 851 // CHECK3-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4 852 // CHECK3-NEXT: store i32 [[TMP13]], i32* [[DOTOMP_IV]], align 4 853 // CHECK3-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] 854 // CHECK3: omp.inner.for.cond: 855 // CHECK3-NEXT: [[TMP14:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !6 856 // CHECK3-NEXT: [[TMP15:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4, !llvm.access.group !6 857 // CHECK3-NEXT: [[CMP6:%.*]] = icmp sle i32 [[TMP14]], [[TMP15]] 858 // CHECK3-NEXT: br i1 [[CMP6]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] 859 // CHECK3: omp.inner.for.body: 860 // CHECK3-NEXT: [[TMP16:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !6 861 // CHECK3-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP16]], 1 862 // CHECK3-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] 863 // CHECK3-NEXT: store i32 [[ADD]], i32* [[I4]], align 4, !llvm.access.group !6 864 // CHECK3-NEXT: [[TMP17:%.*]] = load i32, i32* [[I4]], align 4, !llvm.access.group !6 865 // CHECK3-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [100 x i32], [100 x i32]* [[TMP2]], i32 0, i32 [[TMP17]] 866 // CHECK3-NEXT: store i32 0, i32* [[ARRAYIDX]], align 4, !llvm.access.group !6 867 // CHECK3-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] 868 // CHECK3: omp.body.continue: 869 // CHECK3-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] 870 // CHECK3: omp.inner.for.inc: 871 // CHECK3-NEXT: [[TMP18:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !6 872 // CHECK3-NEXT: [[ADD7:%.*]] = add nsw i32 [[TMP18]], 1 873 // CHECK3-NEXT: store i32 [[ADD7]], i32* [[DOTOMP_IV]], align 4, !llvm.access.group !6 874 // CHECK3-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP7:![0-9]+]] 875 // CHECK3: omp.inner.for.end: 876 // CHECK3-NEXT: br label [[OMP_LOOP_EXIT:%.*]] 877 // CHECK3: omp.loop.exit: 878 // CHECK3-NEXT: [[TMP19:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 4 879 // CHECK3-NEXT: [[TMP20:%.*]] = load i32, i32* [[TMP19]], align 4 880 // CHECK3-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP20]]) 881 // CHECK3-NEXT: [[TMP21:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4 882 // CHECK3-NEXT: [[TMP22:%.*]] = icmp ne i32 [[TMP21]], 0 883 // CHECK3-NEXT: br i1 [[TMP22]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]] 884 // CHECK3: .omp.final.then: 885 // CHECK3-NEXT: [[TMP23:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 886 // CHECK3-NEXT: [[SUB8:%.*]] = sub nsw i32 [[TMP23]], 0 887 // CHECK3-NEXT: [[DIV9:%.*]] = sdiv i32 [[SUB8]], 1 888 // CHECK3-NEXT: [[MUL10:%.*]] = mul nsw i32 [[DIV9]], 1 889 // CHECK3-NEXT: [[ADD11:%.*]] = add nsw i32 0, [[MUL10]] 890 // CHECK3-NEXT: store i32 [[ADD11]], i32* [[TMP0]], align 4 891 // CHECK3-NEXT: br label [[DOTOMP_FINAL_DONE]] 892 // CHECK3: .omp.final.done: 893 // CHECK3-NEXT: br label [[OMP_PRECOND_END]] 894 // CHECK3: omp.precond.end: 895 // CHECK3-NEXT: ret void 896 // 897 // 898 // CHECK3-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z21teams_argument_globali_l36 899 // CHECK3-SAME: (i32 noundef [[N:%.*]], [100 x i32]* noundef nonnull align 4 dereferenceable(400) [[A:%.*]]) #[[ATTR1]] { 900 // CHECK3-NEXT: entry: 901 // CHECK3-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4 902 // CHECK3-NEXT: [[A_ADDR:%.*]] = alloca [100 x i32]*, align 4 903 // CHECK3-NEXT: store i32 [[N]], i32* [[N_ADDR]], align 4 904 // CHECK3-NEXT: store [100 x i32]* [[A]], [100 x i32]** [[A_ADDR]], align 4 905 // CHECK3-NEXT: [[TMP0:%.*]] = load [100 x i32]*, [100 x i32]** [[A_ADDR]], align 4 906 // CHECK3-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2]], i32 2, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32*, [100 x i32]*)* @.omp_outlined..1 to void (i32*, i32*, ...)*), i32* [[N_ADDR]], [100 x i32]* [[TMP0]]) 907 // CHECK3-NEXT: ret void 908 // 909 // 910 // CHECK3-LABEL: define {{[^@]+}}@.omp_outlined..1 911 // CHECK3-SAME: (i32* noalias noundef [[DOTGLOBAL_TID_:%.*]], i32* noalias noundef [[DOTBOUND_TID_:%.*]], i32* noundef nonnull align 4 dereferenceable(4) [[N:%.*]], [100 x i32]* noundef nonnull align 4 dereferenceable(400) [[A:%.*]]) #[[ATTR1]] { 912 // CHECK3-NEXT: entry: 913 // CHECK3-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4 914 // CHECK3-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4 915 // CHECK3-NEXT: [[N_ADDR:%.*]] = alloca i32*, align 4 916 // CHECK3-NEXT: [[A_ADDR:%.*]] = alloca [100 x i32]*, align 4 917 // CHECK3-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 918 // CHECK3-NEXT: [[TMP:%.*]] = alloca i32, align 4 919 // CHECK3-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 920 // CHECK3-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 921 // CHECK3-NEXT: [[I:%.*]] = alloca i32, align 4 922 // CHECK3-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 923 // CHECK3-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4 924 // CHECK3-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 925 // CHECK3-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 926 // CHECK3-NEXT: [[I3:%.*]] = alloca i32, align 4 927 // CHECK3-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 4 928 // CHECK3-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 4 929 // CHECK3-NEXT: store i32* [[N]], i32** [[N_ADDR]], align 4 930 // CHECK3-NEXT: store [100 x i32]* [[A]], [100 x i32]** [[A_ADDR]], align 4 931 // CHECK3-NEXT: [[TMP0:%.*]] = load i32*, i32** [[N_ADDR]], align 4 932 // CHECK3-NEXT: [[TMP1:%.*]] = load [100 x i32]*, [100 x i32]** [[A_ADDR]], align 4 933 // CHECK3-NEXT: [[TMP2:%.*]] = load i32, i32* [[TMP0]], align 4 934 // CHECK3-NEXT: store i32 [[TMP2]], i32* [[DOTCAPTURE_EXPR_]], align 4 935 // CHECK3-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 936 // CHECK3-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP3]], 0 937 // CHECK3-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 938 // CHECK3-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1 939 // CHECK3-NEXT: store i32 [[SUB2]], i32* [[DOTCAPTURE_EXPR_1]], align 4 940 // CHECK3-NEXT: store i32 0, i32* [[I]], align 4 941 // CHECK3-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 942 // CHECK3-NEXT: [[CMP:%.*]] = icmp slt i32 0, [[TMP4]] 943 // CHECK3-NEXT: br i1 [[CMP]], label [[OMP_PRECOND_THEN:%.*]], label [[OMP_PRECOND_END:%.*]] 944 // CHECK3: omp.precond.then: 945 // CHECK3-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4 946 // CHECK3-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 947 // CHECK3-NEXT: store i32 [[TMP5]], i32* [[DOTOMP_UB]], align 4 948 // CHECK3-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4 949 // CHECK3-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4 950 // CHECK3-NEXT: [[TMP6:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 4 951 // CHECK3-NEXT: [[TMP7:%.*]] = load i32, i32* [[TMP6]], align 4 952 // CHECK3-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1]], i32 [[TMP7]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1) 953 // CHECK3-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 954 // CHECK3-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 955 // CHECK3-NEXT: [[CMP4:%.*]] = icmp sgt i32 [[TMP8]], [[TMP9]] 956 // CHECK3-NEXT: br i1 [[CMP4]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] 957 // CHECK3: cond.true: 958 // CHECK3-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 959 // CHECK3-NEXT: br label [[COND_END:%.*]] 960 // CHECK3: cond.false: 961 // CHECK3-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 962 // CHECK3-NEXT: br label [[COND_END]] 963 // CHECK3: cond.end: 964 // CHECK3-NEXT: [[COND:%.*]] = phi i32 [ [[TMP10]], [[COND_TRUE]] ], [ [[TMP11]], [[COND_FALSE]] ] 965 // CHECK3-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4 966 // CHECK3-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4 967 // CHECK3-NEXT: store i32 [[TMP12]], i32* [[DOTOMP_IV]], align 4 968 // CHECK3-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] 969 // CHECK3: omp.inner.for.cond: 970 // CHECK3-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 971 // CHECK3-NEXT: [[TMP14:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 972 // CHECK3-NEXT: [[CMP5:%.*]] = icmp sle i32 [[TMP13]], [[TMP14]] 973 // CHECK3-NEXT: br i1 [[CMP5]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] 974 // CHECK3: omp.inner.for.body: 975 // CHECK3-NEXT: [[TMP15:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 976 // CHECK3-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP15]], 1 977 // CHECK3-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] 978 // CHECK3-NEXT: store i32 [[ADD]], i32* [[I3]], align 4 979 // CHECK3-NEXT: [[TMP16:%.*]] = load i32, i32* [[I3]], align 4 980 // CHECK3-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [100 x i32], [100 x i32]* [[TMP1]], i32 0, i32 [[TMP16]] 981 // CHECK3-NEXT: store i32 0, i32* [[ARRAYIDX]], align 4 982 // CHECK3-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] 983 // CHECK3: omp.body.continue: 984 // CHECK3-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] 985 // CHECK3: omp.inner.for.inc: 986 // CHECK3-NEXT: [[TMP17:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 987 // CHECK3-NEXT: [[ADD6:%.*]] = add nsw i32 [[TMP17]], 1 988 // CHECK3-NEXT: store i32 [[ADD6]], i32* [[DOTOMP_IV]], align 4 989 // CHECK3-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP13:![0-9]+]] 990 // CHECK3: omp.inner.for.end: 991 // CHECK3-NEXT: br label [[OMP_LOOP_EXIT:%.*]] 992 // CHECK3: omp.loop.exit: 993 // CHECK3-NEXT: [[TMP18:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 4 994 // CHECK3-NEXT: [[TMP19:%.*]] = load i32, i32* [[TMP18]], align 4 995 // CHECK3-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP19]]) 996 // CHECK3-NEXT: [[TMP20:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4 997 // CHECK3-NEXT: [[TMP21:%.*]] = icmp ne i32 [[TMP20]], 0 998 // CHECK3-NEXT: br i1 [[TMP21]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]] 999 // CHECK3: .omp.final.then: 1000 // CHECK3-NEXT: [[TMP22:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 1001 // CHECK3-NEXT: [[SUB7:%.*]] = sub nsw i32 [[TMP22]], 0 1002 // CHECK3-NEXT: [[DIV8:%.*]] = sdiv i32 [[SUB7]], 1 1003 // CHECK3-NEXT: [[MUL9:%.*]] = mul nsw i32 [[DIV8]], 1 1004 // CHECK3-NEXT: [[ADD10:%.*]] = add nsw i32 0, [[MUL9]] 1005 // CHECK3-NEXT: store i32 [[ADD10]], i32* [[I3]], align 4 1006 // CHECK3-NEXT: br label [[DOTOMP_FINAL_DONE]] 1007 // CHECK3: .omp.final.done: 1008 // CHECK3-NEXT: br label [[OMP_PRECOND_END]] 1009 // CHECK3: omp.precond.end: 1010 // CHECK3-NEXT: ret void 1011 // 1012 // 1013 // CHECK3-LABEL: define {{[^@]+}}@.omp_offloading.requires_reg 1014 // CHECK3-SAME: () #[[ATTR4:[0-9]+]] { 1015 // CHECK3-NEXT: entry: 1016 // CHECK3-NEXT: call void @__tgt_register_requires(i64 1) 1017 // CHECK3-NEXT: ret void 1018 // 1019 // 1020 // CHECK5-LABEL: define {{[^@]+}}@_Z21teams_argument_globali 1021 // CHECK5-SAME: (i32 noundef signext [[N:%.*]]) #[[ATTR0:[0-9]+]] { 1022 // CHECK5-NEXT: entry: 1023 // CHECK5-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4 1024 // CHECK5-NEXT: [[I:%.*]] = alloca i32, align 4 1025 // CHECK5-NEXT: [[TE:%.*]] = alloca i32, align 4 1026 // CHECK5-NEXT: [[TH:%.*]] = alloca i32, align 4 1027 // CHECK5-NEXT: [[TMP:%.*]] = alloca i32, align 4 1028 // CHECK5-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 1029 // CHECK5-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 1030 // CHECK5-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 1031 // CHECK5-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4 1032 // CHECK5-NEXT: [[I4:%.*]] = alloca i32, align 4 1033 // CHECK5-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 1034 // CHECK5-NEXT: [[DOTLINEAR_START:%.*]] = alloca i32, align 4 1035 // CHECK5-NEXT: [[I5:%.*]] = alloca i32, align 4 1036 // CHECK5-NEXT: [[I6:%.*]] = alloca i32, align 4 1037 // CHECK5-NEXT: [[_TMP13:%.*]] = alloca i32, align 4 1038 // CHECK5-NEXT: [[DOTCAPTURE_EXPR_14:%.*]] = alloca i32, align 4 1039 // CHECK5-NEXT: [[DOTCAPTURE_EXPR_15:%.*]] = alloca i32, align 4 1040 // CHECK5-NEXT: [[DOTOMP_LB19:%.*]] = alloca i32, align 4 1041 // CHECK5-NEXT: [[DOTOMP_UB20:%.*]] = alloca i32, align 4 1042 // CHECK5-NEXT: [[I21:%.*]] = alloca i32, align 4 1043 // CHECK5-NEXT: [[DOTOMP_IV24:%.*]] = alloca i32, align 4 1044 // CHECK5-NEXT: [[I25:%.*]] = alloca i32, align 4 1045 // CHECK5-NEXT: store i32 [[N]], i32* [[N_ADDR]], align 4 1046 // CHECK5-NEXT: [[TMP0:%.*]] = load i32, i32* [[N_ADDR]], align 4 1047 // CHECK5-NEXT: [[DIV:%.*]] = sdiv i32 [[TMP0]], 128 1048 // CHECK5-NEXT: store i32 [[DIV]], i32* [[TE]], align 4 1049 // CHECK5-NEXT: store i32 128, i32* [[TH]], align 4 1050 // CHECK5-NEXT: [[TMP1:%.*]] = load i32, i32* [[N_ADDR]], align 4 1051 // CHECK5-NEXT: store i32 [[TMP1]], i32* [[DOTCAPTURE_EXPR_]], align 4 1052 // CHECK5-NEXT: [[TMP2:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 1053 // CHECK5-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP2]], 0 1054 // CHECK5-NEXT: [[DIV2:%.*]] = sdiv i32 [[SUB]], 1 1055 // CHECK5-NEXT: [[SUB3:%.*]] = sub nsw i32 [[DIV2]], 1 1056 // CHECK5-NEXT: store i32 [[SUB3]], i32* [[DOTCAPTURE_EXPR_1]], align 4 1057 // CHECK5-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4 1058 // CHECK5-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 1059 // CHECK5-NEXT: store i32 [[TMP3]], i32* [[DOTOMP_UB]], align 4 1060 // CHECK5-NEXT: store i32 0, i32* [[I4]], align 4 1061 // CHECK5-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 1062 // CHECK5-NEXT: [[CMP:%.*]] = icmp slt i32 0, [[TMP4]] 1063 // CHECK5-NEXT: br i1 [[CMP]], label [[SIMD_IF_THEN:%.*]], label [[SIMD_IF_END:%.*]] 1064 // CHECK5: simd.if.then: 1065 // CHECK5-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4 1066 // CHECK5-NEXT: store i32 [[TMP5]], i32* [[DOTOMP_IV]], align 4 1067 // CHECK5-NEXT: call void @llvm.assume(i1 true) [ "align"(i32* getelementptr inbounds ([100 x i32], [100 x i32]* @a, i64 0, i64 0), i64 16) ] 1068 // CHECK5-NEXT: [[TMP6:%.*]] = load i32, i32* [[I]], align 4 1069 // CHECK5-NEXT: store i32 [[TMP6]], i32* [[DOTLINEAR_START]], align 4 1070 // CHECK5-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] 1071 // CHECK5: omp.inner.for.cond: 1072 // CHECK5-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !2 1073 // CHECK5-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4, !llvm.access.group !2 1074 // CHECK5-NEXT: [[CMP7:%.*]] = icmp sle i32 [[TMP7]], [[TMP8]] 1075 // CHECK5-NEXT: br i1 [[CMP7]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] 1076 // CHECK5: omp.inner.for.body: 1077 // CHECK5-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !2 1078 // CHECK5-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP9]], 1 1079 // CHECK5-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] 1080 // CHECK5-NEXT: store i32 [[ADD]], i32* [[I5]], align 4, !llvm.access.group !2 1081 // CHECK5-NEXT: [[TMP10:%.*]] = load i32, i32* [[I5]], align 4, !llvm.access.group !2 1082 // CHECK5-NEXT: [[IDXPROM:%.*]] = sext i32 [[TMP10]] to i64 1083 // CHECK5-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [100 x i32], [100 x i32]* @a, i64 0, i64 [[IDXPROM]] 1084 // CHECK5-NEXT: store i32 0, i32* [[ARRAYIDX]], align 4, !llvm.access.group !2 1085 // CHECK5-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] 1086 // CHECK5: omp.body.continue: 1087 // CHECK5-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] 1088 // CHECK5: omp.inner.for.inc: 1089 // CHECK5-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !2 1090 // CHECK5-NEXT: [[ADD8:%.*]] = add nsw i32 [[TMP11]], 1 1091 // CHECK5-NEXT: store i32 [[ADD8]], i32* [[DOTOMP_IV]], align 4, !llvm.access.group !2 1092 // CHECK5-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP3:![0-9]+]] 1093 // CHECK5: omp.inner.for.end: 1094 // CHECK5-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 1095 // CHECK5-NEXT: [[SUB9:%.*]] = sub nsw i32 [[TMP12]], 0 1096 // CHECK5-NEXT: [[DIV10:%.*]] = sdiv i32 [[SUB9]], 1 1097 // CHECK5-NEXT: [[MUL11:%.*]] = mul nsw i32 [[DIV10]], 1 1098 // CHECK5-NEXT: [[ADD12:%.*]] = add nsw i32 0, [[MUL11]] 1099 // CHECK5-NEXT: store i32 [[ADD12]], i32* [[I]], align 4 1100 // CHECK5-NEXT: br label [[SIMD_IF_END]] 1101 // CHECK5: simd.if.end: 1102 // CHECK5-NEXT: [[TMP13:%.*]] = load i32, i32* [[N_ADDR]], align 4 1103 // CHECK5-NEXT: store i32 [[TMP13]], i32* [[DOTCAPTURE_EXPR_14]], align 4 1104 // CHECK5-NEXT: [[TMP14:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_14]], align 4 1105 // CHECK5-NEXT: [[SUB16:%.*]] = sub nsw i32 [[TMP14]], 0 1106 // CHECK5-NEXT: [[DIV17:%.*]] = sdiv i32 [[SUB16]], 1 1107 // CHECK5-NEXT: [[SUB18:%.*]] = sub nsw i32 [[DIV17]], 1 1108 // CHECK5-NEXT: store i32 [[SUB18]], i32* [[DOTCAPTURE_EXPR_15]], align 4 1109 // CHECK5-NEXT: store i32 0, i32* [[DOTOMP_LB19]], align 4 1110 // CHECK5-NEXT: [[TMP15:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_15]], align 4 1111 // CHECK5-NEXT: store i32 [[TMP15]], i32* [[DOTOMP_UB20]], align 4 1112 // CHECK5-NEXT: store i32 0, i32* [[I21]], align 4 1113 // CHECK5-NEXT: [[TMP16:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_14]], align 4 1114 // CHECK5-NEXT: [[CMP22:%.*]] = icmp slt i32 0, [[TMP16]] 1115 // CHECK5-NEXT: br i1 [[CMP22]], label [[SIMD_IF_THEN23:%.*]], label [[SIMD_IF_END41:%.*]] 1116 // CHECK5: simd.if.then23: 1117 // CHECK5-NEXT: [[TMP17:%.*]] = load i32, i32* [[DOTOMP_LB19]], align 4 1118 // CHECK5-NEXT: store i32 [[TMP17]], i32* [[DOTOMP_IV24]], align 4 1119 // CHECK5-NEXT: br label [[OMP_INNER_FOR_COND26:%.*]] 1120 // CHECK5: omp.inner.for.cond26: 1121 // CHECK5-NEXT: [[TMP18:%.*]] = load i32, i32* [[DOTOMP_IV24]], align 4 1122 // CHECK5-NEXT: [[TMP19:%.*]] = load i32, i32* [[DOTOMP_UB20]], align 4 1123 // CHECK5-NEXT: [[CMP27:%.*]] = icmp sle i32 [[TMP18]], [[TMP19]] 1124 // CHECK5-NEXT: br i1 [[CMP27]], label [[OMP_INNER_FOR_BODY28:%.*]], label [[OMP_INNER_FOR_END36:%.*]] 1125 // CHECK5: omp.inner.for.body28: 1126 // CHECK5-NEXT: [[TMP20:%.*]] = load i32, i32* [[DOTOMP_IV24]], align 4 1127 // CHECK5-NEXT: [[MUL29:%.*]] = mul nsw i32 [[TMP20]], 1 1128 // CHECK5-NEXT: [[ADD30:%.*]] = add nsw i32 0, [[MUL29]] 1129 // CHECK5-NEXT: store i32 [[ADD30]], i32* [[I25]], align 4 1130 // CHECK5-NEXT: [[TMP21:%.*]] = load i32, i32* [[I25]], align 4 1131 // CHECK5-NEXT: [[IDXPROM31:%.*]] = sext i32 [[TMP21]] to i64 1132 // CHECK5-NEXT: [[ARRAYIDX32:%.*]] = getelementptr inbounds [100 x i32], [100 x i32]* @a, i64 0, i64 [[IDXPROM31]] 1133 // CHECK5-NEXT: store i32 0, i32* [[ARRAYIDX32]], align 4 1134 // CHECK5-NEXT: br label [[OMP_BODY_CONTINUE33:%.*]] 1135 // CHECK5: omp.body.continue33: 1136 // CHECK5-NEXT: br label [[OMP_INNER_FOR_INC34:%.*]] 1137 // CHECK5: omp.inner.for.inc34: 1138 // CHECK5-NEXT: [[TMP22:%.*]] = load i32, i32* [[DOTOMP_IV24]], align 4 1139 // CHECK5-NEXT: [[ADD35:%.*]] = add nsw i32 [[TMP22]], 1 1140 // CHECK5-NEXT: store i32 [[ADD35]], i32* [[DOTOMP_IV24]], align 4 1141 // CHECK5-NEXT: br label [[OMP_INNER_FOR_COND26]], !llvm.loop [[LOOP7:![0-9]+]] 1142 // CHECK5: omp.inner.for.end36: 1143 // CHECK5-NEXT: [[TMP23:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_14]], align 4 1144 // CHECK5-NEXT: [[SUB37:%.*]] = sub nsw i32 [[TMP23]], 0 1145 // CHECK5-NEXT: [[DIV38:%.*]] = sdiv i32 [[SUB37]], 1 1146 // CHECK5-NEXT: [[MUL39:%.*]] = mul nsw i32 [[DIV38]], 1 1147 // CHECK5-NEXT: [[ADD40:%.*]] = add nsw i32 0, [[MUL39]] 1148 // CHECK5-NEXT: store i32 [[ADD40]], i32* [[I25]], align 4 1149 // CHECK5-NEXT: br label [[SIMD_IF_END41]] 1150 // CHECK5: simd.if.end41: 1151 // CHECK5-NEXT: [[TMP24:%.*]] = load i32, i32* getelementptr inbounds ([100 x i32], [100 x i32]* @a, i64 0, i64 0), align 4 1152 // CHECK5-NEXT: ret i32 [[TMP24]] 1153 // 1154 // 1155 // CHECK7-LABEL: define {{[^@]+}}@_Z21teams_argument_globali 1156 // CHECK7-SAME: (i32 noundef [[N:%.*]]) #[[ATTR0:[0-9]+]] { 1157 // CHECK7-NEXT: entry: 1158 // CHECK7-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4 1159 // CHECK7-NEXT: [[I:%.*]] = alloca i32, align 4 1160 // CHECK7-NEXT: [[TE:%.*]] = alloca i32, align 4 1161 // CHECK7-NEXT: [[TH:%.*]] = alloca i32, align 4 1162 // CHECK7-NEXT: [[TMP:%.*]] = alloca i32, align 4 1163 // CHECK7-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 1164 // CHECK7-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 1165 // CHECK7-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 1166 // CHECK7-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4 1167 // CHECK7-NEXT: [[I4:%.*]] = alloca i32, align 4 1168 // CHECK7-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 1169 // CHECK7-NEXT: [[DOTLINEAR_START:%.*]] = alloca i32, align 4 1170 // CHECK7-NEXT: [[I5:%.*]] = alloca i32, align 4 1171 // CHECK7-NEXT: [[I6:%.*]] = alloca i32, align 4 1172 // CHECK7-NEXT: [[_TMP13:%.*]] = alloca i32, align 4 1173 // CHECK7-NEXT: [[DOTCAPTURE_EXPR_14:%.*]] = alloca i32, align 4 1174 // CHECK7-NEXT: [[DOTCAPTURE_EXPR_15:%.*]] = alloca i32, align 4 1175 // CHECK7-NEXT: [[DOTOMP_LB19:%.*]] = alloca i32, align 4 1176 // CHECK7-NEXT: [[DOTOMP_UB20:%.*]] = alloca i32, align 4 1177 // CHECK7-NEXT: [[I21:%.*]] = alloca i32, align 4 1178 // CHECK7-NEXT: [[DOTOMP_IV24:%.*]] = alloca i32, align 4 1179 // CHECK7-NEXT: [[I25:%.*]] = alloca i32, align 4 1180 // CHECK7-NEXT: store i32 [[N]], i32* [[N_ADDR]], align 4 1181 // CHECK7-NEXT: [[TMP0:%.*]] = load i32, i32* [[N_ADDR]], align 4 1182 // CHECK7-NEXT: [[DIV:%.*]] = sdiv i32 [[TMP0]], 128 1183 // CHECK7-NEXT: store i32 [[DIV]], i32* [[TE]], align 4 1184 // CHECK7-NEXT: store i32 128, i32* [[TH]], align 4 1185 // CHECK7-NEXT: [[TMP1:%.*]] = load i32, i32* [[N_ADDR]], align 4 1186 // CHECK7-NEXT: store i32 [[TMP1]], i32* [[DOTCAPTURE_EXPR_]], align 4 1187 // CHECK7-NEXT: [[TMP2:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 1188 // CHECK7-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP2]], 0 1189 // CHECK7-NEXT: [[DIV2:%.*]] = sdiv i32 [[SUB]], 1 1190 // CHECK7-NEXT: [[SUB3:%.*]] = sub nsw i32 [[DIV2]], 1 1191 // CHECK7-NEXT: store i32 [[SUB3]], i32* [[DOTCAPTURE_EXPR_1]], align 4 1192 // CHECK7-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4 1193 // CHECK7-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 1194 // CHECK7-NEXT: store i32 [[TMP3]], i32* [[DOTOMP_UB]], align 4 1195 // CHECK7-NEXT: store i32 0, i32* [[I4]], align 4 1196 // CHECK7-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 1197 // CHECK7-NEXT: [[CMP:%.*]] = icmp slt i32 0, [[TMP4]] 1198 // CHECK7-NEXT: br i1 [[CMP]], label [[SIMD_IF_THEN:%.*]], label [[SIMD_IF_END:%.*]] 1199 // CHECK7: simd.if.then: 1200 // CHECK7-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4 1201 // CHECK7-NEXT: store i32 [[TMP5]], i32* [[DOTOMP_IV]], align 4 1202 // CHECK7-NEXT: call void @llvm.assume(i1 true) [ "align"(i32* getelementptr inbounds ([100 x i32], [100 x i32]* @a, i32 0, i32 0), i32 16) ] 1203 // CHECK7-NEXT: [[TMP6:%.*]] = load i32, i32* [[I]], align 4 1204 // CHECK7-NEXT: store i32 [[TMP6]], i32* [[DOTLINEAR_START]], align 4 1205 // CHECK7-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] 1206 // CHECK7: omp.inner.for.cond: 1207 // CHECK7-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !3 1208 // CHECK7-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4, !llvm.access.group !3 1209 // CHECK7-NEXT: [[CMP7:%.*]] = icmp sle i32 [[TMP7]], [[TMP8]] 1210 // CHECK7-NEXT: br i1 [[CMP7]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] 1211 // CHECK7: omp.inner.for.body: 1212 // CHECK7-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !3 1213 // CHECK7-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP9]], 1 1214 // CHECK7-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] 1215 // CHECK7-NEXT: store i32 [[ADD]], i32* [[I5]], align 4, !llvm.access.group !3 1216 // CHECK7-NEXT: [[TMP10:%.*]] = load i32, i32* [[I5]], align 4, !llvm.access.group !3 1217 // CHECK7-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [100 x i32], [100 x i32]* @a, i32 0, i32 [[TMP10]] 1218 // CHECK7-NEXT: store i32 0, i32* [[ARRAYIDX]], align 4, !llvm.access.group !3 1219 // CHECK7-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] 1220 // CHECK7: omp.body.continue: 1221 // CHECK7-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] 1222 // CHECK7: omp.inner.for.inc: 1223 // CHECK7-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !3 1224 // CHECK7-NEXT: [[ADD8:%.*]] = add nsw i32 [[TMP11]], 1 1225 // CHECK7-NEXT: store i32 [[ADD8]], i32* [[DOTOMP_IV]], align 4, !llvm.access.group !3 1226 // CHECK7-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP4:![0-9]+]] 1227 // CHECK7: omp.inner.for.end: 1228 // CHECK7-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 1229 // CHECK7-NEXT: [[SUB9:%.*]] = sub nsw i32 [[TMP12]], 0 1230 // CHECK7-NEXT: [[DIV10:%.*]] = sdiv i32 [[SUB9]], 1 1231 // CHECK7-NEXT: [[MUL11:%.*]] = mul nsw i32 [[DIV10]], 1 1232 // CHECK7-NEXT: [[ADD12:%.*]] = add nsw i32 0, [[MUL11]] 1233 // CHECK7-NEXT: store i32 [[ADD12]], i32* [[I]], align 4 1234 // CHECK7-NEXT: br label [[SIMD_IF_END]] 1235 // CHECK7: simd.if.end: 1236 // CHECK7-NEXT: [[TMP13:%.*]] = load i32, i32* [[N_ADDR]], align 4 1237 // CHECK7-NEXT: store i32 [[TMP13]], i32* [[DOTCAPTURE_EXPR_14]], align 4 1238 // CHECK7-NEXT: [[TMP14:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_14]], align 4 1239 // CHECK7-NEXT: [[SUB16:%.*]] = sub nsw i32 [[TMP14]], 0 1240 // CHECK7-NEXT: [[DIV17:%.*]] = sdiv i32 [[SUB16]], 1 1241 // CHECK7-NEXT: [[SUB18:%.*]] = sub nsw i32 [[DIV17]], 1 1242 // CHECK7-NEXT: store i32 [[SUB18]], i32* [[DOTCAPTURE_EXPR_15]], align 4 1243 // CHECK7-NEXT: store i32 0, i32* [[DOTOMP_LB19]], align 4 1244 // CHECK7-NEXT: [[TMP15:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_15]], align 4 1245 // CHECK7-NEXT: store i32 [[TMP15]], i32* [[DOTOMP_UB20]], align 4 1246 // CHECK7-NEXT: store i32 0, i32* [[I21]], align 4 1247 // CHECK7-NEXT: [[TMP16:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_14]], align 4 1248 // CHECK7-NEXT: [[CMP22:%.*]] = icmp slt i32 0, [[TMP16]] 1249 // CHECK7-NEXT: br i1 [[CMP22]], label [[SIMD_IF_THEN23:%.*]], label [[SIMD_IF_END40:%.*]] 1250 // CHECK7: simd.if.then23: 1251 // CHECK7-NEXT: [[TMP17:%.*]] = load i32, i32* [[DOTOMP_LB19]], align 4 1252 // CHECK7-NEXT: store i32 [[TMP17]], i32* [[DOTOMP_IV24]], align 4 1253 // CHECK7-NEXT: br label [[OMP_INNER_FOR_COND26:%.*]] 1254 // CHECK7: omp.inner.for.cond26: 1255 // CHECK7-NEXT: [[TMP18:%.*]] = load i32, i32* [[DOTOMP_IV24]], align 4 1256 // CHECK7-NEXT: [[TMP19:%.*]] = load i32, i32* [[DOTOMP_UB20]], align 4 1257 // CHECK7-NEXT: [[CMP27:%.*]] = icmp sle i32 [[TMP18]], [[TMP19]] 1258 // CHECK7-NEXT: br i1 [[CMP27]], label [[OMP_INNER_FOR_BODY28:%.*]], label [[OMP_INNER_FOR_END35:%.*]] 1259 // CHECK7: omp.inner.for.body28: 1260 // CHECK7-NEXT: [[TMP20:%.*]] = load i32, i32* [[DOTOMP_IV24]], align 4 1261 // CHECK7-NEXT: [[MUL29:%.*]] = mul nsw i32 [[TMP20]], 1 1262 // CHECK7-NEXT: [[ADD30:%.*]] = add nsw i32 0, [[MUL29]] 1263 // CHECK7-NEXT: store i32 [[ADD30]], i32* [[I25]], align 4 1264 // CHECK7-NEXT: [[TMP21:%.*]] = load i32, i32* [[I25]], align 4 1265 // CHECK7-NEXT: [[ARRAYIDX31:%.*]] = getelementptr inbounds [100 x i32], [100 x i32]* @a, i32 0, i32 [[TMP21]] 1266 // CHECK7-NEXT: store i32 0, i32* [[ARRAYIDX31]], align 4 1267 // CHECK7-NEXT: br label [[OMP_BODY_CONTINUE32:%.*]] 1268 // CHECK7: omp.body.continue32: 1269 // CHECK7-NEXT: br label [[OMP_INNER_FOR_INC33:%.*]] 1270 // CHECK7: omp.inner.for.inc33: 1271 // CHECK7-NEXT: [[TMP22:%.*]] = load i32, i32* [[DOTOMP_IV24]], align 4 1272 // CHECK7-NEXT: [[ADD34:%.*]] = add nsw i32 [[TMP22]], 1 1273 // CHECK7-NEXT: store i32 [[ADD34]], i32* [[DOTOMP_IV24]], align 4 1274 // CHECK7-NEXT: br label [[OMP_INNER_FOR_COND26]], !llvm.loop [[LOOP8:![0-9]+]] 1275 // CHECK7: omp.inner.for.end35: 1276 // CHECK7-NEXT: [[TMP23:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_14]], align 4 1277 // CHECK7-NEXT: [[SUB36:%.*]] = sub nsw i32 [[TMP23]], 0 1278 // CHECK7-NEXT: [[DIV37:%.*]] = sdiv i32 [[SUB36]], 1 1279 // CHECK7-NEXT: [[MUL38:%.*]] = mul nsw i32 [[DIV37]], 1 1280 // CHECK7-NEXT: [[ADD39:%.*]] = add nsw i32 0, [[MUL38]] 1281 // CHECK7-NEXT: store i32 [[ADD39]], i32* [[I25]], align 4 1282 // CHECK7-NEXT: br label [[SIMD_IF_END40]] 1283 // CHECK7: simd.if.end40: 1284 // CHECK7-NEXT: [[TMP24:%.*]] = load i32, i32* getelementptr inbounds ([100 x i32], [100 x i32]* @a, i32 0, i32 0), align 4 1285 // CHECK7-NEXT: ret i32 [[TMP24]] 1286 // 1287 // 1288 // CHECK9-LABEL: define {{[^@]+}}@_Z15teams_local_argv 1289 // CHECK9-SAME: () #[[ATTR0:[0-9]+]] { 1290 // CHECK9-NEXT: entry: 1291 // CHECK9-NEXT: [[N:%.*]] = alloca i32, align 4 1292 // CHECK9-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 8 1293 // CHECK9-NEXT: [[__VLA_EXPR0:%.*]] = alloca i64, align 8 1294 // CHECK9-NEXT: [[N_CASTED:%.*]] = alloca i64, align 8 1295 // CHECK9-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [3 x i8*], align 8 1296 // CHECK9-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [3 x i8*], align 8 1297 // CHECK9-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [3 x i8*], align 8 1298 // CHECK9-NEXT: [[DOTOFFLOAD_SIZES:%.*]] = alloca [3 x i64], align 8 1299 // CHECK9-NEXT: [[TMP:%.*]] = alloca i32, align 4 1300 // CHECK9-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 1301 // CHECK9-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 1302 // CHECK9-NEXT: store i32 100, i32* [[N]], align 4 1303 // CHECK9-NEXT: [[TMP0:%.*]] = load i32, i32* [[N]], align 4 1304 // CHECK9-NEXT: [[TMP1:%.*]] = zext i32 [[TMP0]] to i64 1305 // CHECK9-NEXT: [[TMP2:%.*]] = call i8* @llvm.stacksave() 1306 // CHECK9-NEXT: store i8* [[TMP2]], i8** [[SAVED_STACK]], align 8 1307 // CHECK9-NEXT: [[VLA:%.*]] = alloca i32, i64 [[TMP1]], align 4 1308 // CHECK9-NEXT: store i64 [[TMP1]], i64* [[__VLA_EXPR0]], align 8 1309 // CHECK9-NEXT: [[TMP3:%.*]] = load i32, i32* [[N]], align 4 1310 // CHECK9-NEXT: [[CONV:%.*]] = bitcast i64* [[N_CASTED]] to i32* 1311 // CHECK9-NEXT: store i32 [[TMP3]], i32* [[CONV]], align 4 1312 // CHECK9-NEXT: [[TMP4:%.*]] = load i64, i64* [[N_CASTED]], align 8 1313 // CHECK9-NEXT: [[TMP5:%.*]] = mul nuw i64 [[TMP1]], 4 1314 // CHECK9-NEXT: [[TMP6:%.*]] = bitcast [3 x i64]* [[DOTOFFLOAD_SIZES]] to i8* 1315 // CHECK9-NEXT: call void @llvm.memcpy.p0i8.p0i8.i64(i8* align 8 [[TMP6]], i8* align 8 bitcast ([3 x i64]* @.offload_sizes to i8*), i64 24, i1 false) 1316 // CHECK9-NEXT: [[TMP7:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 1317 // CHECK9-NEXT: [[TMP8:%.*]] = bitcast i8** [[TMP7]] to i64* 1318 // CHECK9-NEXT: store i64 [[TMP4]], i64* [[TMP8]], align 8 1319 // CHECK9-NEXT: [[TMP9:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 1320 // CHECK9-NEXT: [[TMP10:%.*]] = bitcast i8** [[TMP9]] to i64* 1321 // CHECK9-NEXT: store i64 [[TMP4]], i64* [[TMP10]], align 8 1322 // CHECK9-NEXT: [[TMP11:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0 1323 // CHECK9-NEXT: store i8* null, i8** [[TMP11]], align 8 1324 // CHECK9-NEXT: [[TMP12:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1 1325 // CHECK9-NEXT: [[TMP13:%.*]] = bitcast i8** [[TMP12]] to i64* 1326 // CHECK9-NEXT: store i64 [[TMP1]], i64* [[TMP13]], align 8 1327 // CHECK9-NEXT: [[TMP14:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1 1328 // CHECK9-NEXT: [[TMP15:%.*]] = bitcast i8** [[TMP14]] to i64* 1329 // CHECK9-NEXT: store i64 [[TMP1]], i64* [[TMP15]], align 8 1330 // CHECK9-NEXT: [[TMP16:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 1 1331 // CHECK9-NEXT: store i8* null, i8** [[TMP16]], align 8 1332 // CHECK9-NEXT: [[TMP17:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2 1333 // CHECK9-NEXT: [[TMP18:%.*]] = bitcast i8** [[TMP17]] to i32** 1334 // CHECK9-NEXT: store i32* [[VLA]], i32** [[TMP18]], align 8 1335 // CHECK9-NEXT: [[TMP19:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2 1336 // CHECK9-NEXT: [[TMP20:%.*]] = bitcast i8** [[TMP19]] to i32** 1337 // CHECK9-NEXT: store i32* [[VLA]], i32** [[TMP20]], align 8 1338 // CHECK9-NEXT: [[TMP21:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 2 1339 // CHECK9-NEXT: store i64 [[TMP5]], i64* [[TMP21]], align 8 1340 // CHECK9-NEXT: [[TMP22:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2 1341 // CHECK9-NEXT: store i8* null, i8** [[TMP22]], align 8 1342 // CHECK9-NEXT: [[TMP23:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 1343 // CHECK9-NEXT: [[TMP24:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 1344 // CHECK9-NEXT: [[TMP25:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0 1345 // CHECK9-NEXT: [[TMP26:%.*]] = load i32, i32* [[N]], align 4 1346 // CHECK9-NEXT: store i32 [[TMP26]], i32* [[DOTCAPTURE_EXPR_]], align 4 1347 // CHECK9-NEXT: [[TMP27:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 1348 // CHECK9-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP27]], 0 1349 // CHECK9-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 1350 // CHECK9-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1 1351 // CHECK9-NEXT: store i32 [[SUB2]], i32* [[DOTCAPTURE_EXPR_1]], align 4 1352 // CHECK9-NEXT: [[TMP28:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 1353 // CHECK9-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP28]], 1 1354 // CHECK9-NEXT: [[TMP29:%.*]] = zext i32 [[ADD]] to i64 1355 // CHECK9-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2:[0-9]+]], i64 -1, i64 [[TMP29]]) 1356 // CHECK9-NEXT: [[TMP30:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z15teams_local_argv_l75.region_id, i32 3, i8** [[TMP23]], i8** [[TMP24]], i64* [[TMP25]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 0, i32 1) 1357 // CHECK9-NEXT: [[TMP31:%.*]] = icmp ne i32 [[TMP30]], 0 1358 // CHECK9-NEXT: br i1 [[TMP31]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] 1359 // CHECK9: omp_offload.failed: 1360 // CHECK9-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z15teams_local_argv_l75(i64 [[TMP4]], i64 [[TMP1]], i32* [[VLA]]) #[[ATTR3:[0-9]+]] 1361 // CHECK9-NEXT: br label [[OMP_OFFLOAD_CONT]] 1362 // CHECK9: omp_offload.cont: 1363 // CHECK9-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i32, i32* [[VLA]], i64 0 1364 // CHECK9-NEXT: [[TMP32:%.*]] = load i32, i32* [[ARRAYIDX]], align 4 1365 // CHECK9-NEXT: [[TMP33:%.*]] = load i8*, i8** [[SAVED_STACK]], align 8 1366 // CHECK9-NEXT: call void @llvm.stackrestore(i8* [[TMP33]]) 1367 // CHECK9-NEXT: ret i32 [[TMP32]] 1368 // 1369 // 1370 // CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z15teams_local_argv_l75 1371 // CHECK9-SAME: (i64 noundef [[N:%.*]], i64 noundef [[VLA:%.*]], i32* noundef nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2:[0-9]+]] { 1372 // CHECK9-NEXT: entry: 1373 // CHECK9-NEXT: [[N_ADDR:%.*]] = alloca i64, align 8 1374 // CHECK9-NEXT: [[VLA_ADDR:%.*]] = alloca i64, align 8 1375 // CHECK9-NEXT: [[A_ADDR:%.*]] = alloca i32*, align 8 1376 // CHECK9-NEXT: store i64 [[N]], i64* [[N_ADDR]], align 8 1377 // CHECK9-NEXT: store i64 [[VLA]], i64* [[VLA_ADDR]], align 8 1378 // CHECK9-NEXT: store i32* [[A]], i32** [[A_ADDR]], align 8 1379 // CHECK9-NEXT: [[CONV:%.*]] = bitcast i64* [[N_ADDR]] to i32* 1380 // CHECK9-NEXT: [[TMP0:%.*]] = load i64, i64* [[VLA_ADDR]], align 8 1381 // CHECK9-NEXT: [[TMP1:%.*]] = load i32*, i32** [[A_ADDR]], align 8 1382 // CHECK9-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2]], i32 3, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32*, i64, i32*)* @.omp_outlined. to void (i32*, i32*, ...)*), i32* [[CONV]], i64 [[TMP0]], i32* [[TMP1]]) 1383 // CHECK9-NEXT: ret void 1384 // 1385 // 1386 // CHECK9-LABEL: define {{[^@]+}}@.omp_outlined. 1387 // CHECK9-SAME: (i32* noalias noundef [[DOTGLOBAL_TID_:%.*]], i32* noalias noundef [[DOTBOUND_TID_:%.*]], i32* noundef nonnull align 4 dereferenceable(4) [[N:%.*]], i64 noundef [[VLA:%.*]], i32* noundef nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2]] { 1388 // CHECK9-NEXT: entry: 1389 // CHECK9-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 1390 // CHECK9-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 1391 // CHECK9-NEXT: [[N_ADDR:%.*]] = alloca i32*, align 8 1392 // CHECK9-NEXT: [[VLA_ADDR:%.*]] = alloca i64, align 8 1393 // CHECK9-NEXT: [[A_ADDR:%.*]] = alloca i32*, align 8 1394 // CHECK9-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 1395 // CHECK9-NEXT: [[TMP:%.*]] = alloca i32, align 4 1396 // CHECK9-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 1397 // CHECK9-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 1398 // CHECK9-NEXT: [[I:%.*]] = alloca i32, align 4 1399 // CHECK9-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 1400 // CHECK9-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4 1401 // CHECK9-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 1402 // CHECK9-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 1403 // CHECK9-NEXT: [[I3:%.*]] = alloca i32, align 4 1404 // CHECK9-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8 1405 // CHECK9-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8 1406 // CHECK9-NEXT: store i32* [[N]], i32** [[N_ADDR]], align 8 1407 // CHECK9-NEXT: store i64 [[VLA]], i64* [[VLA_ADDR]], align 8 1408 // CHECK9-NEXT: store i32* [[A]], i32** [[A_ADDR]], align 8 1409 // CHECK9-NEXT: [[TMP0:%.*]] = load i32*, i32** [[N_ADDR]], align 8 1410 // CHECK9-NEXT: [[TMP1:%.*]] = load i64, i64* [[VLA_ADDR]], align 8 1411 // CHECK9-NEXT: [[TMP2:%.*]] = load i32*, i32** [[A_ADDR]], align 8 1412 // CHECK9-NEXT: [[TMP3:%.*]] = load i32, i32* [[TMP0]], align 4 1413 // CHECK9-NEXT: store i32 [[TMP3]], i32* [[DOTCAPTURE_EXPR_]], align 4 1414 // CHECK9-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 1415 // CHECK9-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP4]], 0 1416 // CHECK9-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 1417 // CHECK9-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1 1418 // CHECK9-NEXT: store i32 [[SUB2]], i32* [[DOTCAPTURE_EXPR_1]], align 4 1419 // CHECK9-NEXT: store i32 0, i32* [[I]], align 4 1420 // CHECK9-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 1421 // CHECK9-NEXT: [[CMP:%.*]] = icmp slt i32 0, [[TMP5]] 1422 // CHECK9-NEXT: br i1 [[CMP]], label [[OMP_PRECOND_THEN:%.*]], label [[OMP_PRECOND_END:%.*]] 1423 // CHECK9: omp.precond.then: 1424 // CHECK9-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4 1425 // CHECK9-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 1426 // CHECK9-NEXT: store i32 [[TMP6]], i32* [[DOTOMP_UB]], align 4 1427 // CHECK9-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4 1428 // CHECK9-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4 1429 // CHECK9-NEXT: [[TMP7:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 1430 // CHECK9-NEXT: [[TMP8:%.*]] = load i32, i32* [[TMP7]], align 4 1431 // CHECK9-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1:[0-9]+]], i32 [[TMP8]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1) 1432 // CHECK9-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 1433 // CHECK9-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 1434 // CHECK9-NEXT: [[CMP4:%.*]] = icmp sgt i32 [[TMP9]], [[TMP10]] 1435 // CHECK9-NEXT: br i1 [[CMP4]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] 1436 // CHECK9: cond.true: 1437 // CHECK9-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 1438 // CHECK9-NEXT: br label [[COND_END:%.*]] 1439 // CHECK9: cond.false: 1440 // CHECK9-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 1441 // CHECK9-NEXT: br label [[COND_END]] 1442 // CHECK9: cond.end: 1443 // CHECK9-NEXT: [[COND:%.*]] = phi i32 [ [[TMP11]], [[COND_TRUE]] ], [ [[TMP12]], [[COND_FALSE]] ] 1444 // CHECK9-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4 1445 // CHECK9-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4 1446 // CHECK9-NEXT: store i32 [[TMP13]], i32* [[DOTOMP_IV]], align 4 1447 // CHECK9-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] 1448 // CHECK9: omp.inner.for.cond: 1449 // CHECK9-NEXT: [[TMP14:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !4 1450 // CHECK9-NEXT: [[TMP15:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4, !llvm.access.group !4 1451 // CHECK9-NEXT: [[CMP5:%.*]] = icmp sle i32 [[TMP14]], [[TMP15]] 1452 // CHECK9-NEXT: br i1 [[CMP5]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] 1453 // CHECK9: omp.inner.for.body: 1454 // CHECK9-NEXT: [[TMP16:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !4 1455 // CHECK9-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP16]], 1 1456 // CHECK9-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] 1457 // CHECK9-NEXT: store i32 [[ADD]], i32* [[I3]], align 4, !llvm.access.group !4 1458 // CHECK9-NEXT: [[TMP17:%.*]] = load i32, i32* [[I3]], align 4, !llvm.access.group !4 1459 // CHECK9-NEXT: [[IDXPROM:%.*]] = sext i32 [[TMP17]] to i64 1460 // CHECK9-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i32, i32* [[TMP2]], i64 [[IDXPROM]] 1461 // CHECK9-NEXT: store i32 0, i32* [[ARRAYIDX]], align 4, !llvm.access.group !4 1462 // CHECK9-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] 1463 // CHECK9: omp.body.continue: 1464 // CHECK9-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] 1465 // CHECK9: omp.inner.for.inc: 1466 // CHECK9-NEXT: [[TMP18:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !4 1467 // CHECK9-NEXT: [[ADD6:%.*]] = add nsw i32 [[TMP18]], 1 1468 // CHECK9-NEXT: store i32 [[ADD6]], i32* [[DOTOMP_IV]], align 4, !llvm.access.group !4 1469 // CHECK9-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP5:![0-9]+]] 1470 // CHECK9: omp.inner.for.end: 1471 // CHECK9-NEXT: br label [[OMP_LOOP_EXIT:%.*]] 1472 // CHECK9: omp.loop.exit: 1473 // CHECK9-NEXT: [[TMP19:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 1474 // CHECK9-NEXT: [[TMP20:%.*]] = load i32, i32* [[TMP19]], align 4 1475 // CHECK9-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP20]]) 1476 // CHECK9-NEXT: [[TMP21:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4 1477 // CHECK9-NEXT: [[TMP22:%.*]] = icmp ne i32 [[TMP21]], 0 1478 // CHECK9-NEXT: br i1 [[TMP22]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]] 1479 // CHECK9: .omp.final.then: 1480 // CHECK9-NEXT: [[TMP23:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 1481 // CHECK9-NEXT: [[SUB7:%.*]] = sub nsw i32 [[TMP23]], 0 1482 // CHECK9-NEXT: [[DIV8:%.*]] = sdiv i32 [[SUB7]], 1 1483 // CHECK9-NEXT: [[MUL9:%.*]] = mul nsw i32 [[DIV8]], 1 1484 // CHECK9-NEXT: [[ADD10:%.*]] = add nsw i32 0, [[MUL9]] 1485 // CHECK9-NEXT: store i32 [[ADD10]], i32* [[I3]], align 4 1486 // CHECK9-NEXT: br label [[DOTOMP_FINAL_DONE]] 1487 // CHECK9: .omp.final.done: 1488 // CHECK9-NEXT: br label [[OMP_PRECOND_END]] 1489 // CHECK9: omp.precond.end: 1490 // CHECK9-NEXT: ret void 1491 // 1492 // 1493 // CHECK9-LABEL: define {{[^@]+}}@.omp_offloading.requires_reg 1494 // CHECK9-SAME: () #[[ATTR5:[0-9]+]] { 1495 // CHECK9-NEXT: entry: 1496 // CHECK9-NEXT: call void @__tgt_register_requires(i64 1) 1497 // CHECK9-NEXT: ret void 1498 // 1499 // 1500 // CHECK11-LABEL: define {{[^@]+}}@_Z15teams_local_argv 1501 // CHECK11-SAME: () #[[ATTR0:[0-9]+]] { 1502 // CHECK11-NEXT: entry: 1503 // CHECK11-NEXT: [[N:%.*]] = alloca i32, align 4 1504 // CHECK11-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 4 1505 // CHECK11-NEXT: [[__VLA_EXPR0:%.*]] = alloca i32, align 4 1506 // CHECK11-NEXT: [[N_CASTED:%.*]] = alloca i32, align 4 1507 // CHECK11-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [3 x i8*], align 4 1508 // CHECK11-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [3 x i8*], align 4 1509 // CHECK11-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [3 x i8*], align 4 1510 // CHECK11-NEXT: [[DOTOFFLOAD_SIZES:%.*]] = alloca [3 x i64], align 4 1511 // CHECK11-NEXT: [[TMP:%.*]] = alloca i32, align 4 1512 // CHECK11-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 1513 // CHECK11-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 1514 // CHECK11-NEXT: store i32 100, i32* [[N]], align 4 1515 // CHECK11-NEXT: [[TMP0:%.*]] = load i32, i32* [[N]], align 4 1516 // CHECK11-NEXT: [[TMP1:%.*]] = call i8* @llvm.stacksave() 1517 // CHECK11-NEXT: store i8* [[TMP1]], i8** [[SAVED_STACK]], align 4 1518 // CHECK11-NEXT: [[VLA:%.*]] = alloca i32, i32 [[TMP0]], align 4 1519 // CHECK11-NEXT: store i32 [[TMP0]], i32* [[__VLA_EXPR0]], align 4 1520 // CHECK11-NEXT: [[TMP2:%.*]] = load i32, i32* [[N]], align 4 1521 // CHECK11-NEXT: store i32 [[TMP2]], i32* [[N_CASTED]], align 4 1522 // CHECK11-NEXT: [[TMP3:%.*]] = load i32, i32* [[N_CASTED]], align 4 1523 // CHECK11-NEXT: [[TMP4:%.*]] = mul nuw i32 [[TMP0]], 4 1524 // CHECK11-NEXT: [[TMP5:%.*]] = sext i32 [[TMP4]] to i64 1525 // CHECK11-NEXT: [[TMP6:%.*]] = bitcast [3 x i64]* [[DOTOFFLOAD_SIZES]] to i8* 1526 // CHECK11-NEXT: call void @llvm.memcpy.p0i8.p0i8.i32(i8* align 4 [[TMP6]], i8* align 4 bitcast ([3 x i64]* @.offload_sizes to i8*), i32 24, i1 false) 1527 // CHECK11-NEXT: [[TMP7:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 1528 // CHECK11-NEXT: [[TMP8:%.*]] = bitcast i8** [[TMP7]] to i32* 1529 // CHECK11-NEXT: store i32 [[TMP3]], i32* [[TMP8]], align 4 1530 // CHECK11-NEXT: [[TMP9:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 1531 // CHECK11-NEXT: [[TMP10:%.*]] = bitcast i8** [[TMP9]] to i32* 1532 // CHECK11-NEXT: store i32 [[TMP3]], i32* [[TMP10]], align 4 1533 // CHECK11-NEXT: [[TMP11:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0 1534 // CHECK11-NEXT: store i8* null, i8** [[TMP11]], align 4 1535 // CHECK11-NEXT: [[TMP12:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1 1536 // CHECK11-NEXT: [[TMP13:%.*]] = bitcast i8** [[TMP12]] to i32* 1537 // CHECK11-NEXT: store i32 [[TMP0]], i32* [[TMP13]], align 4 1538 // CHECK11-NEXT: [[TMP14:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1 1539 // CHECK11-NEXT: [[TMP15:%.*]] = bitcast i8** [[TMP14]] to i32* 1540 // CHECK11-NEXT: store i32 [[TMP0]], i32* [[TMP15]], align 4 1541 // CHECK11-NEXT: [[TMP16:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 1 1542 // CHECK11-NEXT: store i8* null, i8** [[TMP16]], align 4 1543 // CHECK11-NEXT: [[TMP17:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2 1544 // CHECK11-NEXT: [[TMP18:%.*]] = bitcast i8** [[TMP17]] to i32** 1545 // CHECK11-NEXT: store i32* [[VLA]], i32** [[TMP18]], align 4 1546 // CHECK11-NEXT: [[TMP19:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2 1547 // CHECK11-NEXT: [[TMP20:%.*]] = bitcast i8** [[TMP19]] to i32** 1548 // CHECK11-NEXT: store i32* [[VLA]], i32** [[TMP20]], align 4 1549 // CHECK11-NEXT: [[TMP21:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 2 1550 // CHECK11-NEXT: store i64 [[TMP5]], i64* [[TMP21]], align 4 1551 // CHECK11-NEXT: [[TMP22:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2 1552 // CHECK11-NEXT: store i8* null, i8** [[TMP22]], align 4 1553 // CHECK11-NEXT: [[TMP23:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 1554 // CHECK11-NEXT: [[TMP24:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 1555 // CHECK11-NEXT: [[TMP25:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0 1556 // CHECK11-NEXT: [[TMP26:%.*]] = load i32, i32* [[N]], align 4 1557 // CHECK11-NEXT: store i32 [[TMP26]], i32* [[DOTCAPTURE_EXPR_]], align 4 1558 // CHECK11-NEXT: [[TMP27:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 1559 // CHECK11-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP27]], 0 1560 // CHECK11-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 1561 // CHECK11-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1 1562 // CHECK11-NEXT: store i32 [[SUB2]], i32* [[DOTCAPTURE_EXPR_1]], align 4 1563 // CHECK11-NEXT: [[TMP28:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 1564 // CHECK11-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP28]], 1 1565 // CHECK11-NEXT: [[TMP29:%.*]] = zext i32 [[ADD]] to i64 1566 // CHECK11-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2:[0-9]+]], i64 -1, i64 [[TMP29]]) 1567 // CHECK11-NEXT: [[TMP30:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z15teams_local_argv_l75.region_id, i32 3, i8** [[TMP23]], i8** [[TMP24]], i64* [[TMP25]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 0, i32 1) 1568 // CHECK11-NEXT: [[TMP31:%.*]] = icmp ne i32 [[TMP30]], 0 1569 // CHECK11-NEXT: br i1 [[TMP31]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] 1570 // CHECK11: omp_offload.failed: 1571 // CHECK11-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z15teams_local_argv_l75(i32 [[TMP3]], i32 [[TMP0]], i32* [[VLA]]) #[[ATTR3:[0-9]+]] 1572 // CHECK11-NEXT: br label [[OMP_OFFLOAD_CONT]] 1573 // CHECK11: omp_offload.cont: 1574 // CHECK11-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i32, i32* [[VLA]], i32 0 1575 // CHECK11-NEXT: [[TMP32:%.*]] = load i32, i32* [[ARRAYIDX]], align 4 1576 // CHECK11-NEXT: [[TMP33:%.*]] = load i8*, i8** [[SAVED_STACK]], align 4 1577 // CHECK11-NEXT: call void @llvm.stackrestore(i8* [[TMP33]]) 1578 // CHECK11-NEXT: ret i32 [[TMP32]] 1579 // 1580 // 1581 // CHECK11-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z15teams_local_argv_l75 1582 // CHECK11-SAME: (i32 noundef [[N:%.*]], i32 noundef [[VLA:%.*]], i32* noundef nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2:[0-9]+]] { 1583 // CHECK11-NEXT: entry: 1584 // CHECK11-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4 1585 // CHECK11-NEXT: [[VLA_ADDR:%.*]] = alloca i32, align 4 1586 // CHECK11-NEXT: [[A_ADDR:%.*]] = alloca i32*, align 4 1587 // CHECK11-NEXT: store i32 [[N]], i32* [[N_ADDR]], align 4 1588 // CHECK11-NEXT: store i32 [[VLA]], i32* [[VLA_ADDR]], align 4 1589 // CHECK11-NEXT: store i32* [[A]], i32** [[A_ADDR]], align 4 1590 // CHECK11-NEXT: [[TMP0:%.*]] = load i32, i32* [[VLA_ADDR]], align 4 1591 // CHECK11-NEXT: [[TMP1:%.*]] = load i32*, i32** [[A_ADDR]], align 4 1592 // CHECK11-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2]], i32 3, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32*, i32, i32*)* @.omp_outlined. to void (i32*, i32*, ...)*), i32* [[N_ADDR]], i32 [[TMP0]], i32* [[TMP1]]) 1593 // CHECK11-NEXT: ret void 1594 // 1595 // 1596 // CHECK11-LABEL: define {{[^@]+}}@.omp_outlined. 1597 // CHECK11-SAME: (i32* noalias noundef [[DOTGLOBAL_TID_:%.*]], i32* noalias noundef [[DOTBOUND_TID_:%.*]], i32* noundef nonnull align 4 dereferenceable(4) [[N:%.*]], i32 noundef [[VLA:%.*]], i32* noundef nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2]] { 1598 // CHECK11-NEXT: entry: 1599 // CHECK11-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4 1600 // CHECK11-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4 1601 // CHECK11-NEXT: [[N_ADDR:%.*]] = alloca i32*, align 4 1602 // CHECK11-NEXT: [[VLA_ADDR:%.*]] = alloca i32, align 4 1603 // CHECK11-NEXT: [[A_ADDR:%.*]] = alloca i32*, align 4 1604 // CHECK11-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 1605 // CHECK11-NEXT: [[TMP:%.*]] = alloca i32, align 4 1606 // CHECK11-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 1607 // CHECK11-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 1608 // CHECK11-NEXT: [[I:%.*]] = alloca i32, align 4 1609 // CHECK11-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 1610 // CHECK11-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4 1611 // CHECK11-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 1612 // CHECK11-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 1613 // CHECK11-NEXT: [[I3:%.*]] = alloca i32, align 4 1614 // CHECK11-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 4 1615 // CHECK11-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 4 1616 // CHECK11-NEXT: store i32* [[N]], i32** [[N_ADDR]], align 4 1617 // CHECK11-NEXT: store i32 [[VLA]], i32* [[VLA_ADDR]], align 4 1618 // CHECK11-NEXT: store i32* [[A]], i32** [[A_ADDR]], align 4 1619 // CHECK11-NEXT: [[TMP0:%.*]] = load i32*, i32** [[N_ADDR]], align 4 1620 // CHECK11-NEXT: [[TMP1:%.*]] = load i32, i32* [[VLA_ADDR]], align 4 1621 // CHECK11-NEXT: [[TMP2:%.*]] = load i32*, i32** [[A_ADDR]], align 4 1622 // CHECK11-NEXT: [[TMP3:%.*]] = load i32, i32* [[TMP0]], align 4 1623 // CHECK11-NEXT: store i32 [[TMP3]], i32* [[DOTCAPTURE_EXPR_]], align 4 1624 // CHECK11-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 1625 // CHECK11-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP4]], 0 1626 // CHECK11-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 1627 // CHECK11-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1 1628 // CHECK11-NEXT: store i32 [[SUB2]], i32* [[DOTCAPTURE_EXPR_1]], align 4 1629 // CHECK11-NEXT: store i32 0, i32* [[I]], align 4 1630 // CHECK11-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 1631 // CHECK11-NEXT: [[CMP:%.*]] = icmp slt i32 0, [[TMP5]] 1632 // CHECK11-NEXT: br i1 [[CMP]], label [[OMP_PRECOND_THEN:%.*]], label [[OMP_PRECOND_END:%.*]] 1633 // CHECK11: omp.precond.then: 1634 // CHECK11-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4 1635 // CHECK11-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 1636 // CHECK11-NEXT: store i32 [[TMP6]], i32* [[DOTOMP_UB]], align 4 1637 // CHECK11-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4 1638 // CHECK11-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4 1639 // CHECK11-NEXT: [[TMP7:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 4 1640 // CHECK11-NEXT: [[TMP8:%.*]] = load i32, i32* [[TMP7]], align 4 1641 // CHECK11-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1:[0-9]+]], i32 [[TMP8]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1) 1642 // CHECK11-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 1643 // CHECK11-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 1644 // CHECK11-NEXT: [[CMP4:%.*]] = icmp sgt i32 [[TMP9]], [[TMP10]] 1645 // CHECK11-NEXT: br i1 [[CMP4]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] 1646 // CHECK11: cond.true: 1647 // CHECK11-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 1648 // CHECK11-NEXT: br label [[COND_END:%.*]] 1649 // CHECK11: cond.false: 1650 // CHECK11-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 1651 // CHECK11-NEXT: br label [[COND_END]] 1652 // CHECK11: cond.end: 1653 // CHECK11-NEXT: [[COND:%.*]] = phi i32 [ [[TMP11]], [[COND_TRUE]] ], [ [[TMP12]], [[COND_FALSE]] ] 1654 // CHECK11-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4 1655 // CHECK11-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4 1656 // CHECK11-NEXT: store i32 [[TMP13]], i32* [[DOTOMP_IV]], align 4 1657 // CHECK11-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] 1658 // CHECK11: omp.inner.for.cond: 1659 // CHECK11-NEXT: [[TMP14:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !5 1660 // CHECK11-NEXT: [[TMP15:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4, !llvm.access.group !5 1661 // CHECK11-NEXT: [[CMP5:%.*]] = icmp sle i32 [[TMP14]], [[TMP15]] 1662 // CHECK11-NEXT: br i1 [[CMP5]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] 1663 // CHECK11: omp.inner.for.body: 1664 // CHECK11-NEXT: [[TMP16:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !5 1665 // CHECK11-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP16]], 1 1666 // CHECK11-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] 1667 // CHECK11-NEXT: store i32 [[ADD]], i32* [[I3]], align 4, !llvm.access.group !5 1668 // CHECK11-NEXT: [[TMP17:%.*]] = load i32, i32* [[I3]], align 4, !llvm.access.group !5 1669 // CHECK11-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i32, i32* [[TMP2]], i32 [[TMP17]] 1670 // CHECK11-NEXT: store i32 0, i32* [[ARRAYIDX]], align 4, !llvm.access.group !5 1671 // CHECK11-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] 1672 // CHECK11: omp.body.continue: 1673 // CHECK11-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] 1674 // CHECK11: omp.inner.for.inc: 1675 // CHECK11-NEXT: [[TMP18:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !5 1676 // CHECK11-NEXT: [[ADD6:%.*]] = add nsw i32 [[TMP18]], 1 1677 // CHECK11-NEXT: store i32 [[ADD6]], i32* [[DOTOMP_IV]], align 4, !llvm.access.group !5 1678 // CHECK11-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP6:![0-9]+]] 1679 // CHECK11: omp.inner.for.end: 1680 // CHECK11-NEXT: br label [[OMP_LOOP_EXIT:%.*]] 1681 // CHECK11: omp.loop.exit: 1682 // CHECK11-NEXT: [[TMP19:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 4 1683 // CHECK11-NEXT: [[TMP20:%.*]] = load i32, i32* [[TMP19]], align 4 1684 // CHECK11-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP20]]) 1685 // CHECK11-NEXT: [[TMP21:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4 1686 // CHECK11-NEXT: [[TMP22:%.*]] = icmp ne i32 [[TMP21]], 0 1687 // CHECK11-NEXT: br i1 [[TMP22]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]] 1688 // CHECK11: .omp.final.then: 1689 // CHECK11-NEXT: [[TMP23:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 1690 // CHECK11-NEXT: [[SUB7:%.*]] = sub nsw i32 [[TMP23]], 0 1691 // CHECK11-NEXT: [[DIV8:%.*]] = sdiv i32 [[SUB7]], 1 1692 // CHECK11-NEXT: [[MUL9:%.*]] = mul nsw i32 [[DIV8]], 1 1693 // CHECK11-NEXT: [[ADD10:%.*]] = add nsw i32 0, [[MUL9]] 1694 // CHECK11-NEXT: store i32 [[ADD10]], i32* [[I3]], align 4 1695 // CHECK11-NEXT: br label [[DOTOMP_FINAL_DONE]] 1696 // CHECK11: .omp.final.done: 1697 // CHECK11-NEXT: br label [[OMP_PRECOND_END]] 1698 // CHECK11: omp.precond.end: 1699 // CHECK11-NEXT: ret void 1700 // 1701 // 1702 // CHECK11-LABEL: define {{[^@]+}}@.omp_offloading.requires_reg 1703 // CHECK11-SAME: () #[[ATTR5:[0-9]+]] { 1704 // CHECK11-NEXT: entry: 1705 // CHECK11-NEXT: call void @__tgt_register_requires(i64 1) 1706 // CHECK11-NEXT: ret void 1707 // 1708 // 1709 // CHECK13-LABEL: define {{[^@]+}}@_Z15teams_local_argv 1710 // CHECK13-SAME: () #[[ATTR0:[0-9]+]] { 1711 // CHECK13-NEXT: entry: 1712 // CHECK13-NEXT: [[N:%.*]] = alloca i32, align 4 1713 // CHECK13-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 8 1714 // CHECK13-NEXT: [[__VLA_EXPR0:%.*]] = alloca i64, align 8 1715 // CHECK13-NEXT: [[TMP:%.*]] = alloca i32, align 4 1716 // CHECK13-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 1717 // CHECK13-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 1718 // CHECK13-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 1719 // CHECK13-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4 1720 // CHECK13-NEXT: [[I:%.*]] = alloca i32, align 4 1721 // CHECK13-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 1722 // CHECK13-NEXT: [[I3:%.*]] = alloca i32, align 4 1723 // CHECK13-NEXT: store i32 100, i32* [[N]], align 4 1724 // CHECK13-NEXT: [[TMP0:%.*]] = load i32, i32* [[N]], align 4 1725 // CHECK13-NEXT: [[TMP1:%.*]] = zext i32 [[TMP0]] to i64 1726 // CHECK13-NEXT: [[TMP2:%.*]] = call i8* @llvm.stacksave() 1727 // CHECK13-NEXT: store i8* [[TMP2]], i8** [[SAVED_STACK]], align 8 1728 // CHECK13-NEXT: [[VLA:%.*]] = alloca i32, i64 [[TMP1]], align 4 1729 // CHECK13-NEXT: store i64 [[TMP1]], i64* [[__VLA_EXPR0]], align 8 1730 // CHECK13-NEXT: [[TMP3:%.*]] = load i32, i32* [[N]], align 4 1731 // CHECK13-NEXT: store i32 [[TMP3]], i32* [[DOTCAPTURE_EXPR_]], align 4 1732 // CHECK13-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 1733 // CHECK13-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP4]], 0 1734 // CHECK13-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 1735 // CHECK13-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1 1736 // CHECK13-NEXT: store i32 [[SUB2]], i32* [[DOTCAPTURE_EXPR_1]], align 4 1737 // CHECK13-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4 1738 // CHECK13-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 1739 // CHECK13-NEXT: store i32 [[TMP5]], i32* [[DOTOMP_UB]], align 4 1740 // CHECK13-NEXT: store i32 0, i32* [[I]], align 4 1741 // CHECK13-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 1742 // CHECK13-NEXT: [[CMP:%.*]] = icmp slt i32 0, [[TMP6]] 1743 // CHECK13-NEXT: br i1 [[CMP]], label [[SIMD_IF_THEN:%.*]], label [[SIMD_IF_END:%.*]] 1744 // CHECK13: simd.if.then: 1745 // CHECK13-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4 1746 // CHECK13-NEXT: store i32 [[TMP7]], i32* [[DOTOMP_IV]], align 4 1747 // CHECK13-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] 1748 // CHECK13: omp.inner.for.cond: 1749 // CHECK13-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !2 1750 // CHECK13-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4, !llvm.access.group !2 1751 // CHECK13-NEXT: [[CMP4:%.*]] = icmp sle i32 [[TMP8]], [[TMP9]] 1752 // CHECK13-NEXT: br i1 [[CMP4]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] 1753 // CHECK13: omp.inner.for.body: 1754 // CHECK13-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !2 1755 // CHECK13-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP10]], 1 1756 // CHECK13-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] 1757 // CHECK13-NEXT: store i32 [[ADD]], i32* [[I3]], align 4, !llvm.access.group !2 1758 // CHECK13-NEXT: [[TMP11:%.*]] = load i32, i32* [[I3]], align 4, !llvm.access.group !2 1759 // CHECK13-NEXT: [[IDXPROM:%.*]] = sext i32 [[TMP11]] to i64 1760 // CHECK13-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i32, i32* [[VLA]], i64 [[IDXPROM]] 1761 // CHECK13-NEXT: store i32 0, i32* [[ARRAYIDX]], align 4, !llvm.access.group !2 1762 // CHECK13-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] 1763 // CHECK13: omp.body.continue: 1764 // CHECK13-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] 1765 // CHECK13: omp.inner.for.inc: 1766 // CHECK13-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !2 1767 // CHECK13-NEXT: [[ADD5:%.*]] = add nsw i32 [[TMP12]], 1 1768 // CHECK13-NEXT: store i32 [[ADD5]], i32* [[DOTOMP_IV]], align 4, !llvm.access.group !2 1769 // CHECK13-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP3:![0-9]+]] 1770 // CHECK13: omp.inner.for.end: 1771 // CHECK13-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 1772 // CHECK13-NEXT: [[SUB6:%.*]] = sub nsw i32 [[TMP13]], 0 1773 // CHECK13-NEXT: [[DIV7:%.*]] = sdiv i32 [[SUB6]], 1 1774 // CHECK13-NEXT: [[MUL8:%.*]] = mul nsw i32 [[DIV7]], 1 1775 // CHECK13-NEXT: [[ADD9:%.*]] = add nsw i32 0, [[MUL8]] 1776 // CHECK13-NEXT: store i32 [[ADD9]], i32* [[I3]], align 4 1777 // CHECK13-NEXT: br label [[SIMD_IF_END]] 1778 // CHECK13: simd.if.end: 1779 // CHECK13-NEXT: [[ARRAYIDX10:%.*]] = getelementptr inbounds i32, i32* [[VLA]], i64 0 1780 // CHECK13-NEXT: [[TMP14:%.*]] = load i32, i32* [[ARRAYIDX10]], align 4 1781 // CHECK13-NEXT: [[TMP15:%.*]] = load i8*, i8** [[SAVED_STACK]], align 8 1782 // CHECK13-NEXT: call void @llvm.stackrestore(i8* [[TMP15]]) 1783 // CHECK13-NEXT: ret i32 [[TMP14]] 1784 // 1785 // 1786 // CHECK15-LABEL: define {{[^@]+}}@_Z15teams_local_argv 1787 // CHECK15-SAME: () #[[ATTR0:[0-9]+]] { 1788 // CHECK15-NEXT: entry: 1789 // CHECK15-NEXT: [[N:%.*]] = alloca i32, align 4 1790 // CHECK15-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 4 1791 // CHECK15-NEXT: [[__VLA_EXPR0:%.*]] = alloca i32, align 4 1792 // CHECK15-NEXT: [[TMP:%.*]] = alloca i32, align 4 1793 // CHECK15-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 1794 // CHECK15-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 1795 // CHECK15-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 1796 // CHECK15-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4 1797 // CHECK15-NEXT: [[I:%.*]] = alloca i32, align 4 1798 // CHECK15-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 1799 // CHECK15-NEXT: [[I3:%.*]] = alloca i32, align 4 1800 // CHECK15-NEXT: store i32 100, i32* [[N]], align 4 1801 // CHECK15-NEXT: [[TMP0:%.*]] = load i32, i32* [[N]], align 4 1802 // CHECK15-NEXT: [[TMP1:%.*]] = call i8* @llvm.stacksave() 1803 // CHECK15-NEXT: store i8* [[TMP1]], i8** [[SAVED_STACK]], align 4 1804 // CHECK15-NEXT: [[VLA:%.*]] = alloca i32, i32 [[TMP0]], align 4 1805 // CHECK15-NEXT: store i32 [[TMP0]], i32* [[__VLA_EXPR0]], align 4 1806 // CHECK15-NEXT: [[TMP2:%.*]] = load i32, i32* [[N]], align 4 1807 // CHECK15-NEXT: store i32 [[TMP2]], i32* [[DOTCAPTURE_EXPR_]], align 4 1808 // CHECK15-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 1809 // CHECK15-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP3]], 0 1810 // CHECK15-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 1811 // CHECK15-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1 1812 // CHECK15-NEXT: store i32 [[SUB2]], i32* [[DOTCAPTURE_EXPR_1]], align 4 1813 // CHECK15-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4 1814 // CHECK15-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 1815 // CHECK15-NEXT: store i32 [[TMP4]], i32* [[DOTOMP_UB]], align 4 1816 // CHECK15-NEXT: store i32 0, i32* [[I]], align 4 1817 // CHECK15-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 1818 // CHECK15-NEXT: [[CMP:%.*]] = icmp slt i32 0, [[TMP5]] 1819 // CHECK15-NEXT: br i1 [[CMP]], label [[SIMD_IF_THEN:%.*]], label [[SIMD_IF_END:%.*]] 1820 // CHECK15: simd.if.then: 1821 // CHECK15-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4 1822 // CHECK15-NEXT: store i32 [[TMP6]], i32* [[DOTOMP_IV]], align 4 1823 // CHECK15-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] 1824 // CHECK15: omp.inner.for.cond: 1825 // CHECK15-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !3 1826 // CHECK15-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4, !llvm.access.group !3 1827 // CHECK15-NEXT: [[CMP4:%.*]] = icmp sle i32 [[TMP7]], [[TMP8]] 1828 // CHECK15-NEXT: br i1 [[CMP4]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] 1829 // CHECK15: omp.inner.for.body: 1830 // CHECK15-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !3 1831 // CHECK15-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP9]], 1 1832 // CHECK15-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] 1833 // CHECK15-NEXT: store i32 [[ADD]], i32* [[I3]], align 4, !llvm.access.group !3 1834 // CHECK15-NEXT: [[TMP10:%.*]] = load i32, i32* [[I3]], align 4, !llvm.access.group !3 1835 // CHECK15-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i32, i32* [[VLA]], i32 [[TMP10]] 1836 // CHECK15-NEXT: store i32 0, i32* [[ARRAYIDX]], align 4, !llvm.access.group !3 1837 // CHECK15-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] 1838 // CHECK15: omp.body.continue: 1839 // CHECK15-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] 1840 // CHECK15: omp.inner.for.inc: 1841 // CHECK15-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !3 1842 // CHECK15-NEXT: [[ADD5:%.*]] = add nsw i32 [[TMP11]], 1 1843 // CHECK15-NEXT: store i32 [[ADD5]], i32* [[DOTOMP_IV]], align 4, !llvm.access.group !3 1844 // CHECK15-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP4:![0-9]+]] 1845 // CHECK15: omp.inner.for.end: 1846 // CHECK15-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 1847 // CHECK15-NEXT: [[SUB6:%.*]] = sub nsw i32 [[TMP12]], 0 1848 // CHECK15-NEXT: [[DIV7:%.*]] = sdiv i32 [[SUB6]], 1 1849 // CHECK15-NEXT: [[MUL8:%.*]] = mul nsw i32 [[DIV7]], 1 1850 // CHECK15-NEXT: [[ADD9:%.*]] = add nsw i32 0, [[MUL8]] 1851 // CHECK15-NEXT: store i32 [[ADD9]], i32* [[I3]], align 4 1852 // CHECK15-NEXT: br label [[SIMD_IF_END]] 1853 // CHECK15: simd.if.end: 1854 // CHECK15-NEXT: [[ARRAYIDX10:%.*]] = getelementptr inbounds i32, i32* [[VLA]], i32 0 1855 // CHECK15-NEXT: [[TMP13:%.*]] = load i32, i32* [[ARRAYIDX10]], align 4 1856 // CHECK15-NEXT: [[TMP14:%.*]] = load i8*, i8** [[SAVED_STACK]], align 4 1857 // CHECK15-NEXT: call void @llvm.stackrestore(i8* [[TMP14]]) 1858 // CHECK15-NEXT: ret i32 [[TMP13]] 1859 // 1860 // 1861 // CHECK17-LABEL: define {{[^@]+}}@_Z21teams_template_structv 1862 // CHECK17-SAME: () #[[ATTR0:[0-9]+]] { 1863 // CHECK17-NEXT: entry: 1864 // CHECK17-NEXT: [[V:%.*]] = alloca [[STRUCT_SS:%.*]], align 4 1865 // CHECK17-NEXT: [[CALL:%.*]] = call noundef signext i32 @_ZN2SSIiLi123ELx456EE3fooEv(%struct.SS* noundef nonnull align 4 dereferenceable(496) [[V]]) 1866 // CHECK17-NEXT: ret i32 [[CALL]] 1867 // 1868 // 1869 // CHECK17-LABEL: define {{[^@]+}}@_ZN2SSIiLi123ELx456EE3fooEv 1870 // CHECK17-SAME: (%struct.SS* noundef nonnull align 4 dereferenceable(496) [[THIS:%.*]]) #[[ATTR0]] comdat align 2 { 1871 // CHECK17-NEXT: entry: 1872 // CHECK17-NEXT: [[THIS_ADDR:%.*]] = alloca %struct.SS*, align 8 1873 // CHECK17-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [3 x i8*], align 8 1874 // CHECK17-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [3 x i8*], align 8 1875 // CHECK17-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [3 x i8*], align 8 1876 // CHECK17-NEXT: [[DOTOFFLOAD_SIZES:%.*]] = alloca [3 x i64], align 8 1877 // CHECK17-NEXT: [[TMP:%.*]] = alloca i32, align 4 1878 // CHECK17-NEXT: store %struct.SS* [[THIS]], %struct.SS** [[THIS_ADDR]], align 8 1879 // CHECK17-NEXT: [[THIS1:%.*]] = load %struct.SS*, %struct.SS** [[THIS_ADDR]], align 8 1880 // CHECK17-NEXT: [[B:%.*]] = getelementptr inbounds [[STRUCT_SS:%.*]], %struct.SS* [[THIS1]], i32 0, i32 1 1881 // CHECK17-NEXT: [[A:%.*]] = getelementptr inbounds [[STRUCT_SS]], %struct.SS* [[THIS1]], i32 0, i32 0 1882 // CHECK17-NEXT: [[TMP0:%.*]] = getelementptr float, float* [[B]], i32 1 1883 // CHECK17-NEXT: [[TMP1:%.*]] = bitcast [123 x i32]* [[A]] to i8* 1884 // CHECK17-NEXT: [[TMP2:%.*]] = bitcast float* [[TMP0]] to i8* 1885 // CHECK17-NEXT: [[TMP3:%.*]] = ptrtoint i8* [[TMP2]] to i64 1886 // CHECK17-NEXT: [[TMP4:%.*]] = ptrtoint i8* [[TMP1]] to i64 1887 // CHECK17-NEXT: [[TMP5:%.*]] = sub i64 [[TMP3]], [[TMP4]] 1888 // CHECK17-NEXT: [[TMP6:%.*]] = sdiv exact i64 [[TMP5]], ptrtoint (i8* getelementptr (i8, i8* null, i32 1) to i64) 1889 // CHECK17-NEXT: [[TMP7:%.*]] = bitcast [3 x i64]* [[DOTOFFLOAD_SIZES]] to i8* 1890 // CHECK17-NEXT: call void @llvm.memcpy.p0i8.p0i8.i64(i8* align 8 [[TMP7]], i8* align 8 bitcast ([3 x i64]* @.offload_sizes to i8*), i64 24, i1 false) 1891 // CHECK17-NEXT: [[TMP8:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 1892 // CHECK17-NEXT: [[TMP9:%.*]] = bitcast i8** [[TMP8]] to %struct.SS** 1893 // CHECK17-NEXT: store %struct.SS* [[THIS1]], %struct.SS** [[TMP9]], align 8 1894 // CHECK17-NEXT: [[TMP10:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 1895 // CHECK17-NEXT: [[TMP11:%.*]] = bitcast i8** [[TMP10]] to [123 x i32]** 1896 // CHECK17-NEXT: store [123 x i32]* [[A]], [123 x i32]** [[TMP11]], align 8 1897 // CHECK17-NEXT: [[TMP12:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0 1898 // CHECK17-NEXT: store i64 [[TMP6]], i64* [[TMP12]], align 8 1899 // CHECK17-NEXT: [[TMP13:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0 1900 // CHECK17-NEXT: store i8* null, i8** [[TMP13]], align 8 1901 // CHECK17-NEXT: [[TMP14:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1 1902 // CHECK17-NEXT: [[TMP15:%.*]] = bitcast i8** [[TMP14]] to %struct.SS** 1903 // CHECK17-NEXT: store %struct.SS* [[THIS1]], %struct.SS** [[TMP15]], align 8 1904 // CHECK17-NEXT: [[TMP16:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1 1905 // CHECK17-NEXT: [[TMP17:%.*]] = bitcast i8** [[TMP16]] to float** 1906 // CHECK17-NEXT: store float* [[B]], float** [[TMP17]], align 8 1907 // CHECK17-NEXT: [[TMP18:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 1 1908 // CHECK17-NEXT: store i8* null, i8** [[TMP18]], align 8 1909 // CHECK17-NEXT: [[TMP19:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2 1910 // CHECK17-NEXT: [[TMP20:%.*]] = bitcast i8** [[TMP19]] to %struct.SS** 1911 // CHECK17-NEXT: store %struct.SS* [[THIS1]], %struct.SS** [[TMP20]], align 8 1912 // CHECK17-NEXT: [[TMP21:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2 1913 // CHECK17-NEXT: [[TMP22:%.*]] = bitcast i8** [[TMP21]] to [123 x i32]** 1914 // CHECK17-NEXT: store [123 x i32]* [[A]], [123 x i32]** [[TMP22]], align 8 1915 // CHECK17-NEXT: [[TMP23:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2 1916 // CHECK17-NEXT: store i8* null, i8** [[TMP23]], align 8 1917 // CHECK17-NEXT: [[TMP24:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 1918 // CHECK17-NEXT: [[TMP25:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 1919 // CHECK17-NEXT: [[TMP26:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0 1920 // CHECK17-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2:[0-9]+]], i64 -1, i64 123) 1921 // CHECK17-NEXT: [[TMP27:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l123.region_id, i32 3, i8** [[TMP24]], i8** [[TMP25]], i64* [[TMP26]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 0, i32 1) 1922 // CHECK17-NEXT: [[TMP28:%.*]] = icmp ne i32 [[TMP27]], 0 1923 // CHECK17-NEXT: br i1 [[TMP28]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] 1924 // CHECK17: omp_offload.failed: 1925 // CHECK17-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l123(%struct.SS* [[THIS1]]) #[[ATTR2:[0-9]+]] 1926 // CHECK17-NEXT: br label [[OMP_OFFLOAD_CONT]] 1927 // CHECK17: omp_offload.cont: 1928 // CHECK17-NEXT: [[A2:%.*]] = getelementptr inbounds [[STRUCT_SS]], %struct.SS* [[THIS1]], i32 0, i32 0 1929 // CHECK17-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [123 x i32], [123 x i32]* [[A2]], i64 0, i64 0 1930 // CHECK17-NEXT: [[TMP29:%.*]] = load i32, i32* [[ARRAYIDX]], align 4 1931 // CHECK17-NEXT: ret i32 [[TMP29]] 1932 // 1933 // 1934 // CHECK17-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l123 1935 // CHECK17-SAME: (%struct.SS* noundef [[THIS:%.*]]) #[[ATTR1:[0-9]+]] { 1936 // CHECK17-NEXT: entry: 1937 // CHECK17-NEXT: [[THIS_ADDR:%.*]] = alloca %struct.SS*, align 8 1938 // CHECK17-NEXT: store %struct.SS* [[THIS]], %struct.SS** [[THIS_ADDR]], align 8 1939 // CHECK17-NEXT: [[TMP0:%.*]] = load %struct.SS*, %struct.SS** [[THIS_ADDR]], align 8 1940 // CHECK17-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, %struct.SS*)* @.omp_outlined. to void (i32*, i32*, ...)*), %struct.SS* [[TMP0]]) 1941 // CHECK17-NEXT: ret void 1942 // 1943 // 1944 // CHECK17-LABEL: define {{[^@]+}}@.omp_outlined. 1945 // CHECK17-SAME: (i32* noalias noundef [[DOTGLOBAL_TID_:%.*]], i32* noalias noundef [[DOTBOUND_TID_:%.*]], %struct.SS* noundef [[THIS:%.*]]) #[[ATTR1]] { 1946 // CHECK17-NEXT: entry: 1947 // CHECK17-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 1948 // CHECK17-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 1949 // CHECK17-NEXT: [[THIS_ADDR:%.*]] = alloca %struct.SS*, align 8 1950 // CHECK17-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 1951 // CHECK17-NEXT: [[TMP:%.*]] = alloca i32, align 4 1952 // CHECK17-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 1953 // CHECK17-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4 1954 // CHECK17-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 1955 // CHECK17-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 1956 // CHECK17-NEXT: [[I:%.*]] = alloca i32, align 4 1957 // CHECK17-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8 1958 // CHECK17-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8 1959 // CHECK17-NEXT: store %struct.SS* [[THIS]], %struct.SS** [[THIS_ADDR]], align 8 1960 // CHECK17-NEXT: [[TMP0:%.*]] = load %struct.SS*, %struct.SS** [[THIS_ADDR]], align 8 1961 // CHECK17-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4 1962 // CHECK17-NEXT: store i32 122, i32* [[DOTOMP_UB]], align 4 1963 // CHECK17-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4 1964 // CHECK17-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4 1965 // CHECK17-NEXT: [[TMP1:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 1966 // CHECK17-NEXT: [[TMP2:%.*]] = load i32, i32* [[TMP1]], align 4 1967 // CHECK17-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1:[0-9]+]], i32 [[TMP2]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1) 1968 // CHECK17-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 1969 // CHECK17-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP3]], 122 1970 // CHECK17-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] 1971 // CHECK17: cond.true: 1972 // CHECK17-NEXT: br label [[COND_END:%.*]] 1973 // CHECK17: cond.false: 1974 // CHECK17-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 1975 // CHECK17-NEXT: br label [[COND_END]] 1976 // CHECK17: cond.end: 1977 // CHECK17-NEXT: [[COND:%.*]] = phi i32 [ 122, [[COND_TRUE]] ], [ [[TMP4]], [[COND_FALSE]] ] 1978 // CHECK17-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4 1979 // CHECK17-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4 1980 // CHECK17-NEXT: store i32 [[TMP5]], i32* [[DOTOMP_IV]], align 4 1981 // CHECK17-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] 1982 // CHECK17: omp.inner.for.cond: 1983 // CHECK17-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !4 1984 // CHECK17-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4, !llvm.access.group !4 1985 // CHECK17-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP6]], [[TMP7]] 1986 // CHECK17-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] 1987 // CHECK17: omp.inner.for.body: 1988 // CHECK17-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !4 1989 // CHECK17-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP8]], 1 1990 // CHECK17-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] 1991 // CHECK17-NEXT: store i32 [[ADD]], i32* [[I]], align 4, !llvm.access.group !4 1992 // CHECK17-NEXT: [[B:%.*]] = getelementptr inbounds [[STRUCT_SS:%.*]], %struct.SS* [[TMP0]], i32 0, i32 1 1993 // CHECK17-NEXT: [[TMP9:%.*]] = load float, float* [[B]], align 4, !llvm.access.group !4 1994 // CHECK17-NEXT: [[CONV:%.*]] = fptosi float [[TMP9]] to i32 1995 // CHECK17-NEXT: [[A:%.*]] = getelementptr inbounds [[STRUCT_SS]], %struct.SS* [[TMP0]], i32 0, i32 0 1996 // CHECK17-NEXT: [[TMP10:%.*]] = load i32, i32* [[I]], align 4, !llvm.access.group !4 1997 // CHECK17-NEXT: [[IDXPROM:%.*]] = sext i32 [[TMP10]] to i64 1998 // CHECK17-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [123 x i32], [123 x i32]* [[A]], i64 0, i64 [[IDXPROM]] 1999 // CHECK17-NEXT: store i32 [[CONV]], i32* [[ARRAYIDX]], align 4, !llvm.access.group !4 2000 // CHECK17-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] 2001 // CHECK17: omp.body.continue: 2002 // CHECK17-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] 2003 // CHECK17: omp.inner.for.inc: 2004 // CHECK17-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !4 2005 // CHECK17-NEXT: [[ADD2:%.*]] = add nsw i32 [[TMP11]], 1 2006 // CHECK17-NEXT: store i32 [[ADD2]], i32* [[DOTOMP_IV]], align 4, !llvm.access.group !4 2007 // CHECK17-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP5:![0-9]+]] 2008 // CHECK17: omp.inner.for.end: 2009 // CHECK17-NEXT: br label [[OMP_LOOP_EXIT:%.*]] 2010 // CHECK17: omp.loop.exit: 2011 // CHECK17-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP2]]) 2012 // CHECK17-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4 2013 // CHECK17-NEXT: [[TMP13:%.*]] = icmp ne i32 [[TMP12]], 0 2014 // CHECK17-NEXT: br i1 [[TMP13]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]] 2015 // CHECK17: .omp.final.then: 2016 // CHECK17-NEXT: store i32 123, i32* [[I]], align 4 2017 // CHECK17-NEXT: br label [[DOTOMP_FINAL_DONE]] 2018 // CHECK17: .omp.final.done: 2019 // CHECK17-NEXT: ret void 2020 // 2021 // 2022 // CHECK17-LABEL: define {{[^@]+}}@.omp_offloading.requires_reg 2023 // CHECK17-SAME: () #[[ATTR4:[0-9]+]] { 2024 // CHECK17-NEXT: entry: 2025 // CHECK17-NEXT: call void @__tgt_register_requires(i64 1) 2026 // CHECK17-NEXT: ret void 2027 // 2028 // 2029 // CHECK19-LABEL: define {{[^@]+}}@_Z21teams_template_structv 2030 // CHECK19-SAME: () #[[ATTR0:[0-9]+]] { 2031 // CHECK19-NEXT: entry: 2032 // CHECK19-NEXT: [[V:%.*]] = alloca [[STRUCT_SS:%.*]], align 4 2033 // CHECK19-NEXT: [[CALL:%.*]] = call noundef i32 @_ZN2SSIiLi123ELx456EE3fooEv(%struct.SS* noundef nonnull align 4 dereferenceable(496) [[V]]) 2034 // CHECK19-NEXT: ret i32 [[CALL]] 2035 // 2036 // 2037 // CHECK19-LABEL: define {{[^@]+}}@_ZN2SSIiLi123ELx456EE3fooEv 2038 // CHECK19-SAME: (%struct.SS* noundef nonnull align 4 dereferenceable(496) [[THIS:%.*]]) #[[ATTR0]] comdat align 2 { 2039 // CHECK19-NEXT: entry: 2040 // CHECK19-NEXT: [[THIS_ADDR:%.*]] = alloca %struct.SS*, align 4 2041 // CHECK19-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [3 x i8*], align 4 2042 // CHECK19-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [3 x i8*], align 4 2043 // CHECK19-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [3 x i8*], align 4 2044 // CHECK19-NEXT: [[DOTOFFLOAD_SIZES:%.*]] = alloca [3 x i64], align 4 2045 // CHECK19-NEXT: [[TMP:%.*]] = alloca i32, align 4 2046 // CHECK19-NEXT: store %struct.SS* [[THIS]], %struct.SS** [[THIS_ADDR]], align 4 2047 // CHECK19-NEXT: [[THIS1:%.*]] = load %struct.SS*, %struct.SS** [[THIS_ADDR]], align 4 2048 // CHECK19-NEXT: [[B:%.*]] = getelementptr inbounds [[STRUCT_SS:%.*]], %struct.SS* [[THIS1]], i32 0, i32 1 2049 // CHECK19-NEXT: [[A:%.*]] = getelementptr inbounds [[STRUCT_SS]], %struct.SS* [[THIS1]], i32 0, i32 0 2050 // CHECK19-NEXT: [[TMP0:%.*]] = getelementptr float, float* [[B]], i32 1 2051 // CHECK19-NEXT: [[TMP1:%.*]] = bitcast [123 x i32]* [[A]] to i8* 2052 // CHECK19-NEXT: [[TMP2:%.*]] = bitcast float* [[TMP0]] to i8* 2053 // CHECK19-NEXT: [[TMP3:%.*]] = ptrtoint i8* [[TMP2]] to i64 2054 // CHECK19-NEXT: [[TMP4:%.*]] = ptrtoint i8* [[TMP1]] to i64 2055 // CHECK19-NEXT: [[TMP5:%.*]] = sub i64 [[TMP3]], [[TMP4]] 2056 // CHECK19-NEXT: [[TMP6:%.*]] = sdiv exact i64 [[TMP5]], ptrtoint (i8* getelementptr (i8, i8* null, i32 1) to i64) 2057 // CHECK19-NEXT: [[TMP7:%.*]] = bitcast [3 x i64]* [[DOTOFFLOAD_SIZES]] to i8* 2058 // CHECK19-NEXT: call void @llvm.memcpy.p0i8.p0i8.i32(i8* align 4 [[TMP7]], i8* align 4 bitcast ([3 x i64]* @.offload_sizes to i8*), i32 24, i1 false) 2059 // CHECK19-NEXT: [[TMP8:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 2060 // CHECK19-NEXT: [[TMP9:%.*]] = bitcast i8** [[TMP8]] to %struct.SS** 2061 // CHECK19-NEXT: store %struct.SS* [[THIS1]], %struct.SS** [[TMP9]], align 4 2062 // CHECK19-NEXT: [[TMP10:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 2063 // CHECK19-NEXT: [[TMP11:%.*]] = bitcast i8** [[TMP10]] to [123 x i32]** 2064 // CHECK19-NEXT: store [123 x i32]* [[A]], [123 x i32]** [[TMP11]], align 4 2065 // CHECK19-NEXT: [[TMP12:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0 2066 // CHECK19-NEXT: store i64 [[TMP6]], i64* [[TMP12]], align 4 2067 // CHECK19-NEXT: [[TMP13:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0 2068 // CHECK19-NEXT: store i8* null, i8** [[TMP13]], align 4 2069 // CHECK19-NEXT: [[TMP14:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1 2070 // CHECK19-NEXT: [[TMP15:%.*]] = bitcast i8** [[TMP14]] to %struct.SS** 2071 // CHECK19-NEXT: store %struct.SS* [[THIS1]], %struct.SS** [[TMP15]], align 4 2072 // CHECK19-NEXT: [[TMP16:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1 2073 // CHECK19-NEXT: [[TMP17:%.*]] = bitcast i8** [[TMP16]] to float** 2074 // CHECK19-NEXT: store float* [[B]], float** [[TMP17]], align 4 2075 // CHECK19-NEXT: [[TMP18:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 1 2076 // CHECK19-NEXT: store i8* null, i8** [[TMP18]], align 4 2077 // CHECK19-NEXT: [[TMP19:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2 2078 // CHECK19-NEXT: [[TMP20:%.*]] = bitcast i8** [[TMP19]] to %struct.SS** 2079 // CHECK19-NEXT: store %struct.SS* [[THIS1]], %struct.SS** [[TMP20]], align 4 2080 // CHECK19-NEXT: [[TMP21:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2 2081 // CHECK19-NEXT: [[TMP22:%.*]] = bitcast i8** [[TMP21]] to [123 x i32]** 2082 // CHECK19-NEXT: store [123 x i32]* [[A]], [123 x i32]** [[TMP22]], align 4 2083 // CHECK19-NEXT: [[TMP23:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2 2084 // CHECK19-NEXT: store i8* null, i8** [[TMP23]], align 4 2085 // CHECK19-NEXT: [[TMP24:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 2086 // CHECK19-NEXT: [[TMP25:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 2087 // CHECK19-NEXT: [[TMP26:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0 2088 // CHECK19-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2:[0-9]+]], i64 -1, i64 123) 2089 // CHECK19-NEXT: [[TMP27:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l123.region_id, i32 3, i8** [[TMP24]], i8** [[TMP25]], i64* [[TMP26]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 0, i32 1) 2090 // CHECK19-NEXT: [[TMP28:%.*]] = icmp ne i32 [[TMP27]], 0 2091 // CHECK19-NEXT: br i1 [[TMP28]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] 2092 // CHECK19: omp_offload.failed: 2093 // CHECK19-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l123(%struct.SS* [[THIS1]]) #[[ATTR2:[0-9]+]] 2094 // CHECK19-NEXT: br label [[OMP_OFFLOAD_CONT]] 2095 // CHECK19: omp_offload.cont: 2096 // CHECK19-NEXT: [[A2:%.*]] = getelementptr inbounds [[STRUCT_SS]], %struct.SS* [[THIS1]], i32 0, i32 0 2097 // CHECK19-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [123 x i32], [123 x i32]* [[A2]], i32 0, i32 0 2098 // CHECK19-NEXT: [[TMP29:%.*]] = load i32, i32* [[ARRAYIDX]], align 4 2099 // CHECK19-NEXT: ret i32 [[TMP29]] 2100 // 2101 // 2102 // CHECK19-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l123 2103 // CHECK19-SAME: (%struct.SS* noundef [[THIS:%.*]]) #[[ATTR1:[0-9]+]] { 2104 // CHECK19-NEXT: entry: 2105 // CHECK19-NEXT: [[THIS_ADDR:%.*]] = alloca %struct.SS*, align 4 2106 // CHECK19-NEXT: store %struct.SS* [[THIS]], %struct.SS** [[THIS_ADDR]], align 4 2107 // CHECK19-NEXT: [[TMP0:%.*]] = load %struct.SS*, %struct.SS** [[THIS_ADDR]], align 4 2108 // CHECK19-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, %struct.SS*)* @.omp_outlined. to void (i32*, i32*, ...)*), %struct.SS* [[TMP0]]) 2109 // CHECK19-NEXT: ret void 2110 // 2111 // 2112 // CHECK19-LABEL: define {{[^@]+}}@.omp_outlined. 2113 // CHECK19-SAME: (i32* noalias noundef [[DOTGLOBAL_TID_:%.*]], i32* noalias noundef [[DOTBOUND_TID_:%.*]], %struct.SS* noundef [[THIS:%.*]]) #[[ATTR1]] { 2114 // CHECK19-NEXT: entry: 2115 // CHECK19-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4 2116 // CHECK19-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4 2117 // CHECK19-NEXT: [[THIS_ADDR:%.*]] = alloca %struct.SS*, align 4 2118 // CHECK19-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 2119 // CHECK19-NEXT: [[TMP:%.*]] = alloca i32, align 4 2120 // CHECK19-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 2121 // CHECK19-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4 2122 // CHECK19-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 2123 // CHECK19-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 2124 // CHECK19-NEXT: [[I:%.*]] = alloca i32, align 4 2125 // CHECK19-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 4 2126 // CHECK19-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 4 2127 // CHECK19-NEXT: store %struct.SS* [[THIS]], %struct.SS** [[THIS_ADDR]], align 4 2128 // CHECK19-NEXT: [[TMP0:%.*]] = load %struct.SS*, %struct.SS** [[THIS_ADDR]], align 4 2129 // CHECK19-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4 2130 // CHECK19-NEXT: store i32 122, i32* [[DOTOMP_UB]], align 4 2131 // CHECK19-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4 2132 // CHECK19-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4 2133 // CHECK19-NEXT: [[TMP1:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 4 2134 // CHECK19-NEXT: [[TMP2:%.*]] = load i32, i32* [[TMP1]], align 4 2135 // CHECK19-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1:[0-9]+]], i32 [[TMP2]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1) 2136 // CHECK19-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 2137 // CHECK19-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP3]], 122 2138 // CHECK19-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] 2139 // CHECK19: cond.true: 2140 // CHECK19-NEXT: br label [[COND_END:%.*]] 2141 // CHECK19: cond.false: 2142 // CHECK19-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 2143 // CHECK19-NEXT: br label [[COND_END]] 2144 // CHECK19: cond.end: 2145 // CHECK19-NEXT: [[COND:%.*]] = phi i32 [ 122, [[COND_TRUE]] ], [ [[TMP4]], [[COND_FALSE]] ] 2146 // CHECK19-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4 2147 // CHECK19-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4 2148 // CHECK19-NEXT: store i32 [[TMP5]], i32* [[DOTOMP_IV]], align 4 2149 // CHECK19-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] 2150 // CHECK19: omp.inner.for.cond: 2151 // CHECK19-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !5 2152 // CHECK19-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4, !llvm.access.group !5 2153 // CHECK19-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP6]], [[TMP7]] 2154 // CHECK19-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] 2155 // CHECK19: omp.inner.for.body: 2156 // CHECK19-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !5 2157 // CHECK19-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP8]], 1 2158 // CHECK19-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] 2159 // CHECK19-NEXT: store i32 [[ADD]], i32* [[I]], align 4, !llvm.access.group !5 2160 // CHECK19-NEXT: [[B:%.*]] = getelementptr inbounds [[STRUCT_SS:%.*]], %struct.SS* [[TMP0]], i32 0, i32 1 2161 // CHECK19-NEXT: [[TMP9:%.*]] = load float, float* [[B]], align 4, !llvm.access.group !5 2162 // CHECK19-NEXT: [[CONV:%.*]] = fptosi float [[TMP9]] to i32 2163 // CHECK19-NEXT: [[A:%.*]] = getelementptr inbounds [[STRUCT_SS]], %struct.SS* [[TMP0]], i32 0, i32 0 2164 // CHECK19-NEXT: [[TMP10:%.*]] = load i32, i32* [[I]], align 4, !llvm.access.group !5 2165 // CHECK19-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [123 x i32], [123 x i32]* [[A]], i32 0, i32 [[TMP10]] 2166 // CHECK19-NEXT: store i32 [[CONV]], i32* [[ARRAYIDX]], align 4, !llvm.access.group !5 2167 // CHECK19-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] 2168 // CHECK19: omp.body.continue: 2169 // CHECK19-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] 2170 // CHECK19: omp.inner.for.inc: 2171 // CHECK19-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !5 2172 // CHECK19-NEXT: [[ADD2:%.*]] = add nsw i32 [[TMP11]], 1 2173 // CHECK19-NEXT: store i32 [[ADD2]], i32* [[DOTOMP_IV]], align 4, !llvm.access.group !5 2174 // CHECK19-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP6:![0-9]+]] 2175 // CHECK19: omp.inner.for.end: 2176 // CHECK19-NEXT: br label [[OMP_LOOP_EXIT:%.*]] 2177 // CHECK19: omp.loop.exit: 2178 // CHECK19-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP2]]) 2179 // CHECK19-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4 2180 // CHECK19-NEXT: [[TMP13:%.*]] = icmp ne i32 [[TMP12]], 0 2181 // CHECK19-NEXT: br i1 [[TMP13]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]] 2182 // CHECK19: .omp.final.then: 2183 // CHECK19-NEXT: store i32 123, i32* [[I]], align 4 2184 // CHECK19-NEXT: br label [[DOTOMP_FINAL_DONE]] 2185 // CHECK19: .omp.final.done: 2186 // CHECK19-NEXT: ret void 2187 // 2188 // 2189 // CHECK19-LABEL: define {{[^@]+}}@.omp_offloading.requires_reg 2190 // CHECK19-SAME: () #[[ATTR4:[0-9]+]] { 2191 // CHECK19-NEXT: entry: 2192 // CHECK19-NEXT: call void @__tgt_register_requires(i64 1) 2193 // CHECK19-NEXT: ret void 2194 // 2195 // 2196 // CHECK21-LABEL: define {{[^@]+}}@_Z21teams_template_structv 2197 // CHECK21-SAME: () #[[ATTR0:[0-9]+]] { 2198 // CHECK21-NEXT: entry: 2199 // CHECK21-NEXT: [[V:%.*]] = alloca [[STRUCT_SS:%.*]], align 4 2200 // CHECK21-NEXT: [[CALL:%.*]] = call noundef signext i32 @_ZN2SSIiLi123ELx456EE3fooEv(%struct.SS* noundef nonnull align 4 dereferenceable(496) [[V]]) 2201 // CHECK21-NEXT: ret i32 [[CALL]] 2202 // 2203 // 2204 // CHECK21-LABEL: define {{[^@]+}}@_ZN2SSIiLi123ELx456EE3fooEv 2205 // CHECK21-SAME: (%struct.SS* noundef nonnull align 4 dereferenceable(496) [[THIS:%.*]]) #[[ATTR0]] comdat align 2 { 2206 // CHECK21-NEXT: entry: 2207 // CHECK21-NEXT: [[THIS_ADDR:%.*]] = alloca %struct.SS*, align 8 2208 // CHECK21-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [3 x i8*], align 8 2209 // CHECK21-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [3 x i8*], align 8 2210 // CHECK21-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [3 x i8*], align 8 2211 // CHECK21-NEXT: [[DOTOFFLOAD_SIZES:%.*]] = alloca [3 x i64], align 8 2212 // CHECK21-NEXT: [[TMP:%.*]] = alloca i32, align 4 2213 // CHECK21-NEXT: store %struct.SS* [[THIS]], %struct.SS** [[THIS_ADDR]], align 8 2214 // CHECK21-NEXT: [[THIS1:%.*]] = load %struct.SS*, %struct.SS** [[THIS_ADDR]], align 8 2215 // CHECK21-NEXT: [[B:%.*]] = getelementptr inbounds [[STRUCT_SS:%.*]], %struct.SS* [[THIS1]], i32 0, i32 1 2216 // CHECK21-NEXT: [[A:%.*]] = getelementptr inbounds [[STRUCT_SS]], %struct.SS* [[THIS1]], i32 0, i32 0 2217 // CHECK21-NEXT: [[TMP0:%.*]] = getelementptr float, float* [[B]], i32 1 2218 // CHECK21-NEXT: [[TMP1:%.*]] = bitcast [123 x i32]* [[A]] to i8* 2219 // CHECK21-NEXT: [[TMP2:%.*]] = bitcast float* [[TMP0]] to i8* 2220 // CHECK21-NEXT: [[TMP3:%.*]] = ptrtoint i8* [[TMP2]] to i64 2221 // CHECK21-NEXT: [[TMP4:%.*]] = ptrtoint i8* [[TMP1]] to i64 2222 // CHECK21-NEXT: [[TMP5:%.*]] = sub i64 [[TMP3]], [[TMP4]] 2223 // CHECK21-NEXT: [[TMP6:%.*]] = sdiv exact i64 [[TMP5]], ptrtoint (i8* getelementptr (i8, i8* null, i32 1) to i64) 2224 // CHECK21-NEXT: [[TMP7:%.*]] = bitcast [3 x i64]* [[DOTOFFLOAD_SIZES]] to i8* 2225 // CHECK21-NEXT: call void @llvm.memcpy.p0i8.p0i8.i64(i8* align 8 [[TMP7]], i8* align 8 bitcast ([3 x i64]* @.offload_sizes to i8*), i64 24, i1 false) 2226 // CHECK21-NEXT: [[TMP8:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 2227 // CHECK21-NEXT: [[TMP9:%.*]] = bitcast i8** [[TMP8]] to %struct.SS** 2228 // CHECK21-NEXT: store %struct.SS* [[THIS1]], %struct.SS** [[TMP9]], align 8 2229 // CHECK21-NEXT: [[TMP10:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 2230 // CHECK21-NEXT: [[TMP11:%.*]] = bitcast i8** [[TMP10]] to [123 x i32]** 2231 // CHECK21-NEXT: store [123 x i32]* [[A]], [123 x i32]** [[TMP11]], align 8 2232 // CHECK21-NEXT: [[TMP12:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0 2233 // CHECK21-NEXT: store i64 [[TMP6]], i64* [[TMP12]], align 8 2234 // CHECK21-NEXT: [[TMP13:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0 2235 // CHECK21-NEXT: store i8* null, i8** [[TMP13]], align 8 2236 // CHECK21-NEXT: [[TMP14:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1 2237 // CHECK21-NEXT: [[TMP15:%.*]] = bitcast i8** [[TMP14]] to %struct.SS** 2238 // CHECK21-NEXT: store %struct.SS* [[THIS1]], %struct.SS** [[TMP15]], align 8 2239 // CHECK21-NEXT: [[TMP16:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1 2240 // CHECK21-NEXT: [[TMP17:%.*]] = bitcast i8** [[TMP16]] to float** 2241 // CHECK21-NEXT: store float* [[B]], float** [[TMP17]], align 8 2242 // CHECK21-NEXT: [[TMP18:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 1 2243 // CHECK21-NEXT: store i8* null, i8** [[TMP18]], align 8 2244 // CHECK21-NEXT: [[TMP19:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2 2245 // CHECK21-NEXT: [[TMP20:%.*]] = bitcast i8** [[TMP19]] to %struct.SS** 2246 // CHECK21-NEXT: store %struct.SS* [[THIS1]], %struct.SS** [[TMP20]], align 8 2247 // CHECK21-NEXT: [[TMP21:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2 2248 // CHECK21-NEXT: [[TMP22:%.*]] = bitcast i8** [[TMP21]] to [123 x i32]** 2249 // CHECK21-NEXT: store [123 x i32]* [[A]], [123 x i32]** [[TMP22]], align 8 2250 // CHECK21-NEXT: [[TMP23:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2 2251 // CHECK21-NEXT: store i8* null, i8** [[TMP23]], align 8 2252 // CHECK21-NEXT: [[TMP24:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 2253 // CHECK21-NEXT: [[TMP25:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 2254 // CHECK21-NEXT: [[TMP26:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0 2255 // CHECK21-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2:[0-9]+]], i64 -1, i64 123) 2256 // CHECK21-NEXT: [[TMP27:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l123.region_id, i32 3, i8** [[TMP24]], i8** [[TMP25]], i64* [[TMP26]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 0, i32 1) 2257 // CHECK21-NEXT: [[TMP28:%.*]] = icmp ne i32 [[TMP27]], 0 2258 // CHECK21-NEXT: br i1 [[TMP28]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] 2259 // CHECK21: omp_offload.failed: 2260 // CHECK21-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l123(%struct.SS* [[THIS1]]) #[[ATTR2:[0-9]+]] 2261 // CHECK21-NEXT: br label [[OMP_OFFLOAD_CONT]] 2262 // CHECK21: omp_offload.cont: 2263 // CHECK21-NEXT: [[A2:%.*]] = getelementptr inbounds [[STRUCT_SS]], %struct.SS* [[THIS1]], i32 0, i32 0 2264 // CHECK21-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [123 x i32], [123 x i32]* [[A2]], i64 0, i64 0 2265 // CHECK21-NEXT: [[TMP29:%.*]] = load i32, i32* [[ARRAYIDX]], align 4 2266 // CHECK21-NEXT: ret i32 [[TMP29]] 2267 // 2268 // 2269 // CHECK21-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l123 2270 // CHECK21-SAME: (%struct.SS* noundef [[THIS:%.*]]) #[[ATTR1:[0-9]+]] { 2271 // CHECK21-NEXT: entry: 2272 // CHECK21-NEXT: [[THIS_ADDR:%.*]] = alloca %struct.SS*, align 8 2273 // CHECK21-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i8, align 1 2274 // CHECK21-NEXT: [[DOTCAPTURE_EXPR__CASTED:%.*]] = alloca i64, align 8 2275 // CHECK21-NEXT: store %struct.SS* [[THIS]], %struct.SS** [[THIS_ADDR]], align 8 2276 // CHECK21-NEXT: [[TMP0:%.*]] = load %struct.SS*, %struct.SS** [[THIS_ADDR]], align 8 2277 // CHECK21-NEXT: [[B:%.*]] = getelementptr inbounds [[STRUCT_SS:%.*]], %struct.SS* [[TMP0]], i32 0, i32 1 2278 // CHECK21-NEXT: [[TMP1:%.*]] = load float, float* [[B]], align 4 2279 // CHECK21-NEXT: [[TOBOOL:%.*]] = fcmp une float [[TMP1]], 0.000000e+00 2280 // CHECK21-NEXT: [[FROMBOOL:%.*]] = zext i1 [[TOBOOL]] to i8 2281 // CHECK21-NEXT: store i8 [[FROMBOOL]], i8* [[DOTCAPTURE_EXPR_]], align 1 2282 // CHECK21-NEXT: [[TMP2:%.*]] = load i8, i8* [[DOTCAPTURE_EXPR_]], align 1 2283 // CHECK21-NEXT: [[TOBOOL1:%.*]] = trunc i8 [[TMP2]] to i1 2284 // CHECK21-NEXT: [[CONV:%.*]] = bitcast i64* [[DOTCAPTURE_EXPR__CASTED]] to i8* 2285 // CHECK21-NEXT: [[FROMBOOL2:%.*]] = zext i1 [[TOBOOL1]] to i8 2286 // CHECK21-NEXT: store i8 [[FROMBOOL2]], i8* [[CONV]], align 1 2287 // CHECK21-NEXT: [[TMP3:%.*]] = load i64, i64* [[DOTCAPTURE_EXPR__CASTED]], align 8 2288 // CHECK21-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2]], i32 2, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, %struct.SS*, i64)* @.omp_outlined. to void (i32*, i32*, ...)*), %struct.SS* [[TMP0]], i64 [[TMP3]]) 2289 // CHECK21-NEXT: ret void 2290 // 2291 // 2292 // CHECK21-LABEL: define {{[^@]+}}@.omp_outlined. 2293 // CHECK21-SAME: (i32* noalias noundef [[DOTGLOBAL_TID_:%.*]], i32* noalias noundef [[DOTBOUND_TID_:%.*]], %struct.SS* noundef [[THIS:%.*]], i64 noundef [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR1]] { 2294 // CHECK21-NEXT: entry: 2295 // CHECK21-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 2296 // CHECK21-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 2297 // CHECK21-NEXT: [[THIS_ADDR:%.*]] = alloca %struct.SS*, align 8 2298 // CHECK21-NEXT: [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i64, align 8 2299 // CHECK21-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 2300 // CHECK21-NEXT: [[TMP:%.*]] = alloca i32, align 4 2301 // CHECK21-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 2302 // CHECK21-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4 2303 // CHECK21-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 2304 // CHECK21-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 2305 // CHECK21-NEXT: [[I:%.*]] = alloca i32, align 4 2306 // CHECK21-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8 2307 // CHECK21-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8 2308 // CHECK21-NEXT: store %struct.SS* [[THIS]], %struct.SS** [[THIS_ADDR]], align 8 2309 // CHECK21-NEXT: store i64 [[DOTCAPTURE_EXPR_]], i64* [[DOTCAPTURE_EXPR__ADDR]], align 8 2310 // CHECK21-NEXT: [[TMP0:%.*]] = load %struct.SS*, %struct.SS** [[THIS_ADDR]], align 8 2311 // CHECK21-NEXT: [[CONV:%.*]] = bitcast i64* [[DOTCAPTURE_EXPR__ADDR]] to i8* 2312 // CHECK21-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4 2313 // CHECK21-NEXT: store i32 122, i32* [[DOTOMP_UB]], align 4 2314 // CHECK21-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4 2315 // CHECK21-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4 2316 // CHECK21-NEXT: [[TMP1:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 2317 // CHECK21-NEXT: [[TMP2:%.*]] = load i32, i32* [[TMP1]], align 4 2318 // CHECK21-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1:[0-9]+]], i32 [[TMP2]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1) 2319 // CHECK21-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 2320 // CHECK21-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP3]], 122 2321 // CHECK21-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] 2322 // CHECK21: cond.true: 2323 // CHECK21-NEXT: br label [[COND_END:%.*]] 2324 // CHECK21: cond.false: 2325 // CHECK21-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 2326 // CHECK21-NEXT: br label [[COND_END]] 2327 // CHECK21: cond.end: 2328 // CHECK21-NEXT: [[COND:%.*]] = phi i32 [ 122, [[COND_TRUE]] ], [ [[TMP4]], [[COND_FALSE]] ] 2329 // CHECK21-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4 2330 // CHECK21-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4 2331 // CHECK21-NEXT: store i32 [[TMP5]], i32* [[DOTOMP_IV]], align 4 2332 // CHECK21-NEXT: [[TMP6:%.*]] = load i8, i8* [[CONV]], align 1 2333 // CHECK21-NEXT: [[TOBOOL:%.*]] = trunc i8 [[TMP6]] to i1 2334 // CHECK21-NEXT: br i1 [[TOBOOL]], label [[OMP_IF_THEN:%.*]], label [[OMP_IF_ELSE:%.*]] 2335 // CHECK21: omp_if.then: 2336 // CHECK21-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] 2337 // CHECK21: omp.inner.for.cond: 2338 // CHECK21-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !4 2339 // CHECK21-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4, !llvm.access.group !4 2340 // CHECK21-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP7]], [[TMP8]] 2341 // CHECK21-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] 2342 // CHECK21: omp.inner.for.body: 2343 // CHECK21-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !4 2344 // CHECK21-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP9]], 1 2345 // CHECK21-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] 2346 // CHECK21-NEXT: store i32 [[ADD]], i32* [[I]], align 4, !llvm.access.group !4 2347 // CHECK21-NEXT: [[B:%.*]] = getelementptr inbounds [[STRUCT_SS:%.*]], %struct.SS* [[TMP0]], i32 0, i32 1 2348 // CHECK21-NEXT: [[TMP10:%.*]] = load float, float* [[B]], align 4, !nontemporal !5, !llvm.access.group !4 2349 // CHECK21-NEXT: [[CONV2:%.*]] = fptosi float [[TMP10]] to i32 2350 // CHECK21-NEXT: [[A:%.*]] = getelementptr inbounds [[STRUCT_SS]], %struct.SS* [[TMP0]], i32 0, i32 0 2351 // CHECK21-NEXT: [[TMP11:%.*]] = load i32, i32* [[I]], align 4, !llvm.access.group !4 2352 // CHECK21-NEXT: [[IDXPROM:%.*]] = sext i32 [[TMP11]] to i64 2353 // CHECK21-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [123 x i32], [123 x i32]* [[A]], i64 0, i64 [[IDXPROM]] 2354 // CHECK21-NEXT: store i32 [[CONV2]], i32* [[ARRAYIDX]], align 4, !llvm.access.group !4 2355 // CHECK21-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] 2356 // CHECK21: omp.body.continue: 2357 // CHECK21-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] 2358 // CHECK21: omp.inner.for.inc: 2359 // CHECK21-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !4 2360 // CHECK21-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP12]], 1 2361 // CHECK21-NEXT: store i32 [[ADD3]], i32* [[DOTOMP_IV]], align 4, !llvm.access.group !4 2362 // CHECK21-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP6:![0-9]+]] 2363 // CHECK21: omp.inner.for.end: 2364 // CHECK21-NEXT: br label [[OMP_IF_END:%.*]] 2365 // CHECK21: omp_if.else: 2366 // CHECK21-NEXT: br label [[OMP_INNER_FOR_COND4:%.*]] 2367 // CHECK21: omp.inner.for.cond4: 2368 // CHECK21-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 2369 // CHECK21-NEXT: [[TMP14:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 2370 // CHECK21-NEXT: [[CMP5:%.*]] = icmp sle i32 [[TMP13]], [[TMP14]] 2371 // CHECK21-NEXT: br i1 [[CMP5]], label [[OMP_INNER_FOR_BODY6:%.*]], label [[OMP_INNER_FOR_END17:%.*]] 2372 // CHECK21: omp.inner.for.body6: 2373 // CHECK21-NEXT: [[TMP15:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 2374 // CHECK21-NEXT: [[MUL7:%.*]] = mul nsw i32 [[TMP15]], 1 2375 // CHECK21-NEXT: [[ADD8:%.*]] = add nsw i32 0, [[MUL7]] 2376 // CHECK21-NEXT: store i32 [[ADD8]], i32* [[I]], align 4 2377 // CHECK21-NEXT: [[B9:%.*]] = getelementptr inbounds [[STRUCT_SS]], %struct.SS* [[TMP0]], i32 0, i32 1 2378 // CHECK21-NEXT: [[TMP16:%.*]] = load float, float* [[B9]], align 4 2379 // CHECK21-NEXT: [[CONV10:%.*]] = fptosi float [[TMP16]] to i32 2380 // CHECK21-NEXT: [[A11:%.*]] = getelementptr inbounds [[STRUCT_SS]], %struct.SS* [[TMP0]], i32 0, i32 0 2381 // CHECK21-NEXT: [[TMP17:%.*]] = load i32, i32* [[I]], align 4 2382 // CHECK21-NEXT: [[IDXPROM12:%.*]] = sext i32 [[TMP17]] to i64 2383 // CHECK21-NEXT: [[ARRAYIDX13:%.*]] = getelementptr inbounds [123 x i32], [123 x i32]* [[A11]], i64 0, i64 [[IDXPROM12]] 2384 // CHECK21-NEXT: store i32 [[CONV10]], i32* [[ARRAYIDX13]], align 4 2385 // CHECK21-NEXT: br label [[OMP_BODY_CONTINUE14:%.*]] 2386 // CHECK21: omp.body.continue14: 2387 // CHECK21-NEXT: br label [[OMP_INNER_FOR_INC15:%.*]] 2388 // CHECK21: omp.inner.for.inc15: 2389 // CHECK21-NEXT: [[TMP18:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 2390 // CHECK21-NEXT: [[ADD16:%.*]] = add nsw i32 [[TMP18]], 1 2391 // CHECK21-NEXT: store i32 [[ADD16]], i32* [[DOTOMP_IV]], align 4 2392 // CHECK21-NEXT: br label [[OMP_INNER_FOR_COND4]], !llvm.loop [[LOOP9:![0-9]+]] 2393 // CHECK21: omp.inner.for.end17: 2394 // CHECK21-NEXT: br label [[OMP_IF_END]] 2395 // CHECK21: omp_if.end: 2396 // CHECK21-NEXT: br label [[OMP_LOOP_EXIT:%.*]] 2397 // CHECK21: omp.loop.exit: 2398 // CHECK21-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP2]]) 2399 // CHECK21-NEXT: [[TMP19:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4 2400 // CHECK21-NEXT: [[TMP20:%.*]] = icmp ne i32 [[TMP19]], 0 2401 // CHECK21-NEXT: br i1 [[TMP20]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]] 2402 // CHECK21: .omp.final.then: 2403 // CHECK21-NEXT: store i32 123, i32* [[I]], align 4 2404 // CHECK21-NEXT: br label [[DOTOMP_FINAL_DONE]] 2405 // CHECK21: .omp.final.done: 2406 // CHECK21-NEXT: ret void 2407 // 2408 // 2409 // CHECK21-LABEL: define {{[^@]+}}@.omp_offloading.requires_reg 2410 // CHECK21-SAME: () #[[ATTR4:[0-9]+]] { 2411 // CHECK21-NEXT: entry: 2412 // CHECK21-NEXT: call void @__tgt_register_requires(i64 1) 2413 // CHECK21-NEXT: ret void 2414 // 2415 // 2416 // CHECK23-LABEL: define {{[^@]+}}@_Z21teams_template_structv 2417 // CHECK23-SAME: () #[[ATTR0:[0-9]+]] { 2418 // CHECK23-NEXT: entry: 2419 // CHECK23-NEXT: [[V:%.*]] = alloca [[STRUCT_SS:%.*]], align 4 2420 // CHECK23-NEXT: [[CALL:%.*]] = call noundef i32 @_ZN2SSIiLi123ELx456EE3fooEv(%struct.SS* noundef nonnull align 4 dereferenceable(496) [[V]]) 2421 // CHECK23-NEXT: ret i32 [[CALL]] 2422 // 2423 // 2424 // CHECK23-LABEL: define {{[^@]+}}@_ZN2SSIiLi123ELx456EE3fooEv 2425 // CHECK23-SAME: (%struct.SS* noundef nonnull align 4 dereferenceable(496) [[THIS:%.*]]) #[[ATTR0]] comdat align 2 { 2426 // CHECK23-NEXT: entry: 2427 // CHECK23-NEXT: [[THIS_ADDR:%.*]] = alloca %struct.SS*, align 4 2428 // CHECK23-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [3 x i8*], align 4 2429 // CHECK23-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [3 x i8*], align 4 2430 // CHECK23-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [3 x i8*], align 4 2431 // CHECK23-NEXT: [[DOTOFFLOAD_SIZES:%.*]] = alloca [3 x i64], align 4 2432 // CHECK23-NEXT: [[TMP:%.*]] = alloca i32, align 4 2433 // CHECK23-NEXT: store %struct.SS* [[THIS]], %struct.SS** [[THIS_ADDR]], align 4 2434 // CHECK23-NEXT: [[THIS1:%.*]] = load %struct.SS*, %struct.SS** [[THIS_ADDR]], align 4 2435 // CHECK23-NEXT: [[B:%.*]] = getelementptr inbounds [[STRUCT_SS:%.*]], %struct.SS* [[THIS1]], i32 0, i32 1 2436 // CHECK23-NEXT: [[A:%.*]] = getelementptr inbounds [[STRUCT_SS]], %struct.SS* [[THIS1]], i32 0, i32 0 2437 // CHECK23-NEXT: [[TMP0:%.*]] = getelementptr float, float* [[B]], i32 1 2438 // CHECK23-NEXT: [[TMP1:%.*]] = bitcast [123 x i32]* [[A]] to i8* 2439 // CHECK23-NEXT: [[TMP2:%.*]] = bitcast float* [[TMP0]] to i8* 2440 // CHECK23-NEXT: [[TMP3:%.*]] = ptrtoint i8* [[TMP2]] to i64 2441 // CHECK23-NEXT: [[TMP4:%.*]] = ptrtoint i8* [[TMP1]] to i64 2442 // CHECK23-NEXT: [[TMP5:%.*]] = sub i64 [[TMP3]], [[TMP4]] 2443 // CHECK23-NEXT: [[TMP6:%.*]] = sdiv exact i64 [[TMP5]], ptrtoint (i8* getelementptr (i8, i8* null, i32 1) to i64) 2444 // CHECK23-NEXT: [[TMP7:%.*]] = bitcast [3 x i64]* [[DOTOFFLOAD_SIZES]] to i8* 2445 // CHECK23-NEXT: call void @llvm.memcpy.p0i8.p0i8.i32(i8* align 4 [[TMP7]], i8* align 4 bitcast ([3 x i64]* @.offload_sizes to i8*), i32 24, i1 false) 2446 // CHECK23-NEXT: [[TMP8:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 2447 // CHECK23-NEXT: [[TMP9:%.*]] = bitcast i8** [[TMP8]] to %struct.SS** 2448 // CHECK23-NEXT: store %struct.SS* [[THIS1]], %struct.SS** [[TMP9]], align 4 2449 // CHECK23-NEXT: [[TMP10:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 2450 // CHECK23-NEXT: [[TMP11:%.*]] = bitcast i8** [[TMP10]] to [123 x i32]** 2451 // CHECK23-NEXT: store [123 x i32]* [[A]], [123 x i32]** [[TMP11]], align 4 2452 // CHECK23-NEXT: [[TMP12:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0 2453 // CHECK23-NEXT: store i64 [[TMP6]], i64* [[TMP12]], align 4 2454 // CHECK23-NEXT: [[TMP13:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0 2455 // CHECK23-NEXT: store i8* null, i8** [[TMP13]], align 4 2456 // CHECK23-NEXT: [[TMP14:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1 2457 // CHECK23-NEXT: [[TMP15:%.*]] = bitcast i8** [[TMP14]] to %struct.SS** 2458 // CHECK23-NEXT: store %struct.SS* [[THIS1]], %struct.SS** [[TMP15]], align 4 2459 // CHECK23-NEXT: [[TMP16:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1 2460 // CHECK23-NEXT: [[TMP17:%.*]] = bitcast i8** [[TMP16]] to float** 2461 // CHECK23-NEXT: store float* [[B]], float** [[TMP17]], align 4 2462 // CHECK23-NEXT: [[TMP18:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 1 2463 // CHECK23-NEXT: store i8* null, i8** [[TMP18]], align 4 2464 // CHECK23-NEXT: [[TMP19:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2 2465 // CHECK23-NEXT: [[TMP20:%.*]] = bitcast i8** [[TMP19]] to %struct.SS** 2466 // CHECK23-NEXT: store %struct.SS* [[THIS1]], %struct.SS** [[TMP20]], align 4 2467 // CHECK23-NEXT: [[TMP21:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2 2468 // CHECK23-NEXT: [[TMP22:%.*]] = bitcast i8** [[TMP21]] to [123 x i32]** 2469 // CHECK23-NEXT: store [123 x i32]* [[A]], [123 x i32]** [[TMP22]], align 4 2470 // CHECK23-NEXT: [[TMP23:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2 2471 // CHECK23-NEXT: store i8* null, i8** [[TMP23]], align 4 2472 // CHECK23-NEXT: [[TMP24:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 2473 // CHECK23-NEXT: [[TMP25:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 2474 // CHECK23-NEXT: [[TMP26:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0 2475 // CHECK23-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2:[0-9]+]], i64 -1, i64 123) 2476 // CHECK23-NEXT: [[TMP27:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l123.region_id, i32 3, i8** [[TMP24]], i8** [[TMP25]], i64* [[TMP26]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 0, i32 1) 2477 // CHECK23-NEXT: [[TMP28:%.*]] = icmp ne i32 [[TMP27]], 0 2478 // CHECK23-NEXT: br i1 [[TMP28]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] 2479 // CHECK23: omp_offload.failed: 2480 // CHECK23-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l123(%struct.SS* [[THIS1]]) #[[ATTR2:[0-9]+]] 2481 // CHECK23-NEXT: br label [[OMP_OFFLOAD_CONT]] 2482 // CHECK23: omp_offload.cont: 2483 // CHECK23-NEXT: [[A2:%.*]] = getelementptr inbounds [[STRUCT_SS]], %struct.SS* [[THIS1]], i32 0, i32 0 2484 // CHECK23-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [123 x i32], [123 x i32]* [[A2]], i32 0, i32 0 2485 // CHECK23-NEXT: [[TMP29:%.*]] = load i32, i32* [[ARRAYIDX]], align 4 2486 // CHECK23-NEXT: ret i32 [[TMP29]] 2487 // 2488 // 2489 // CHECK23-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l123 2490 // CHECK23-SAME: (%struct.SS* noundef [[THIS:%.*]]) #[[ATTR1:[0-9]+]] { 2491 // CHECK23-NEXT: entry: 2492 // CHECK23-NEXT: [[THIS_ADDR:%.*]] = alloca %struct.SS*, align 4 2493 // CHECK23-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i8, align 1 2494 // CHECK23-NEXT: [[DOTCAPTURE_EXPR__CASTED:%.*]] = alloca i32, align 4 2495 // CHECK23-NEXT: store %struct.SS* [[THIS]], %struct.SS** [[THIS_ADDR]], align 4 2496 // CHECK23-NEXT: [[TMP0:%.*]] = load %struct.SS*, %struct.SS** [[THIS_ADDR]], align 4 2497 // CHECK23-NEXT: [[B:%.*]] = getelementptr inbounds [[STRUCT_SS:%.*]], %struct.SS* [[TMP0]], i32 0, i32 1 2498 // CHECK23-NEXT: [[TMP1:%.*]] = load float, float* [[B]], align 4 2499 // CHECK23-NEXT: [[TOBOOL:%.*]] = fcmp une float [[TMP1]], 0.000000e+00 2500 // CHECK23-NEXT: [[FROMBOOL:%.*]] = zext i1 [[TOBOOL]] to i8 2501 // CHECK23-NEXT: store i8 [[FROMBOOL]], i8* [[DOTCAPTURE_EXPR_]], align 1 2502 // CHECK23-NEXT: [[TMP2:%.*]] = load i8, i8* [[DOTCAPTURE_EXPR_]], align 1 2503 // CHECK23-NEXT: [[TOBOOL1:%.*]] = trunc i8 [[TMP2]] to i1 2504 // CHECK23-NEXT: [[CONV:%.*]] = bitcast i32* [[DOTCAPTURE_EXPR__CASTED]] to i8* 2505 // CHECK23-NEXT: [[FROMBOOL2:%.*]] = zext i1 [[TOBOOL1]] to i8 2506 // CHECK23-NEXT: store i8 [[FROMBOOL2]], i8* [[CONV]], align 1 2507 // CHECK23-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR__CASTED]], align 4 2508 // CHECK23-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2]], i32 2, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, %struct.SS*, i32)* @.omp_outlined. to void (i32*, i32*, ...)*), %struct.SS* [[TMP0]], i32 [[TMP3]]) 2509 // CHECK23-NEXT: ret void 2510 // 2511 // 2512 // CHECK23-LABEL: define {{[^@]+}}@.omp_outlined. 2513 // CHECK23-SAME: (i32* noalias noundef [[DOTGLOBAL_TID_:%.*]], i32* noalias noundef [[DOTBOUND_TID_:%.*]], %struct.SS* noundef [[THIS:%.*]], i32 noundef [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR1]] { 2514 // CHECK23-NEXT: entry: 2515 // CHECK23-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4 2516 // CHECK23-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4 2517 // CHECK23-NEXT: [[THIS_ADDR:%.*]] = alloca %struct.SS*, align 4 2518 // CHECK23-NEXT: [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i32, align 4 2519 // CHECK23-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 2520 // CHECK23-NEXT: [[TMP:%.*]] = alloca i32, align 4 2521 // CHECK23-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 2522 // CHECK23-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4 2523 // CHECK23-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 2524 // CHECK23-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 2525 // CHECK23-NEXT: [[I:%.*]] = alloca i32, align 4 2526 // CHECK23-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 4 2527 // CHECK23-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 4 2528 // CHECK23-NEXT: store %struct.SS* [[THIS]], %struct.SS** [[THIS_ADDR]], align 4 2529 // CHECK23-NEXT: store i32 [[DOTCAPTURE_EXPR_]], i32* [[DOTCAPTURE_EXPR__ADDR]], align 4 2530 // CHECK23-NEXT: [[TMP0:%.*]] = load %struct.SS*, %struct.SS** [[THIS_ADDR]], align 4 2531 // CHECK23-NEXT: [[CONV:%.*]] = bitcast i32* [[DOTCAPTURE_EXPR__ADDR]] to i8* 2532 // CHECK23-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4 2533 // CHECK23-NEXT: store i32 122, i32* [[DOTOMP_UB]], align 4 2534 // CHECK23-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4 2535 // CHECK23-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4 2536 // CHECK23-NEXT: [[TMP1:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 4 2537 // CHECK23-NEXT: [[TMP2:%.*]] = load i32, i32* [[TMP1]], align 4 2538 // CHECK23-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1:[0-9]+]], i32 [[TMP2]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1) 2539 // CHECK23-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 2540 // CHECK23-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP3]], 122 2541 // CHECK23-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] 2542 // CHECK23: cond.true: 2543 // CHECK23-NEXT: br label [[COND_END:%.*]] 2544 // CHECK23: cond.false: 2545 // CHECK23-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 2546 // CHECK23-NEXT: br label [[COND_END]] 2547 // CHECK23: cond.end: 2548 // CHECK23-NEXT: [[COND:%.*]] = phi i32 [ 122, [[COND_TRUE]] ], [ [[TMP4]], [[COND_FALSE]] ] 2549 // CHECK23-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4 2550 // CHECK23-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4 2551 // CHECK23-NEXT: store i32 [[TMP5]], i32* [[DOTOMP_IV]], align 4 2552 // CHECK23-NEXT: [[TMP6:%.*]] = load i8, i8* [[CONV]], align 1 2553 // CHECK23-NEXT: [[TOBOOL:%.*]] = trunc i8 [[TMP6]] to i1 2554 // CHECK23-NEXT: br i1 [[TOBOOL]], label [[OMP_IF_THEN:%.*]], label [[OMP_IF_ELSE:%.*]] 2555 // CHECK23: omp_if.then: 2556 // CHECK23-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] 2557 // CHECK23: omp.inner.for.cond: 2558 // CHECK23-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !5 2559 // CHECK23-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4, !llvm.access.group !5 2560 // CHECK23-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP7]], [[TMP8]] 2561 // CHECK23-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] 2562 // CHECK23: omp.inner.for.body: 2563 // CHECK23-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !5 2564 // CHECK23-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP9]], 1 2565 // CHECK23-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] 2566 // CHECK23-NEXT: store i32 [[ADD]], i32* [[I]], align 4, !llvm.access.group !5 2567 // CHECK23-NEXT: [[B:%.*]] = getelementptr inbounds [[STRUCT_SS:%.*]], %struct.SS* [[TMP0]], i32 0, i32 1 2568 // CHECK23-NEXT: [[TMP10:%.*]] = load float, float* [[B]], align 4, !nontemporal !6, !llvm.access.group !5 2569 // CHECK23-NEXT: [[CONV2:%.*]] = fptosi float [[TMP10]] to i32 2570 // CHECK23-NEXT: [[A:%.*]] = getelementptr inbounds [[STRUCT_SS]], %struct.SS* [[TMP0]], i32 0, i32 0 2571 // CHECK23-NEXT: [[TMP11:%.*]] = load i32, i32* [[I]], align 4, !llvm.access.group !5 2572 // CHECK23-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [123 x i32], [123 x i32]* [[A]], i32 0, i32 [[TMP11]] 2573 // CHECK23-NEXT: store i32 [[CONV2]], i32* [[ARRAYIDX]], align 4, !llvm.access.group !5 2574 // CHECK23-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] 2575 // CHECK23: omp.body.continue: 2576 // CHECK23-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] 2577 // CHECK23: omp.inner.for.inc: 2578 // CHECK23-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !5 2579 // CHECK23-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP12]], 1 2580 // CHECK23-NEXT: store i32 [[ADD3]], i32* [[DOTOMP_IV]], align 4, !llvm.access.group !5 2581 // CHECK23-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP7:![0-9]+]] 2582 // CHECK23: omp.inner.for.end: 2583 // CHECK23-NEXT: br label [[OMP_IF_END:%.*]] 2584 // CHECK23: omp_if.else: 2585 // CHECK23-NEXT: br label [[OMP_INNER_FOR_COND4:%.*]] 2586 // CHECK23: omp.inner.for.cond4: 2587 // CHECK23-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 2588 // CHECK23-NEXT: [[TMP14:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 2589 // CHECK23-NEXT: [[CMP5:%.*]] = icmp sle i32 [[TMP13]], [[TMP14]] 2590 // CHECK23-NEXT: br i1 [[CMP5]], label [[OMP_INNER_FOR_BODY6:%.*]], label [[OMP_INNER_FOR_END16:%.*]] 2591 // CHECK23: omp.inner.for.body6: 2592 // CHECK23-NEXT: [[TMP15:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 2593 // CHECK23-NEXT: [[MUL7:%.*]] = mul nsw i32 [[TMP15]], 1 2594 // CHECK23-NEXT: [[ADD8:%.*]] = add nsw i32 0, [[MUL7]] 2595 // CHECK23-NEXT: store i32 [[ADD8]], i32* [[I]], align 4 2596 // CHECK23-NEXT: [[B9:%.*]] = getelementptr inbounds [[STRUCT_SS]], %struct.SS* [[TMP0]], i32 0, i32 1 2597 // CHECK23-NEXT: [[TMP16:%.*]] = load float, float* [[B9]], align 4 2598 // CHECK23-NEXT: [[CONV10:%.*]] = fptosi float [[TMP16]] to i32 2599 // CHECK23-NEXT: [[A11:%.*]] = getelementptr inbounds [[STRUCT_SS]], %struct.SS* [[TMP0]], i32 0, i32 0 2600 // CHECK23-NEXT: [[TMP17:%.*]] = load i32, i32* [[I]], align 4 2601 // CHECK23-NEXT: [[ARRAYIDX12:%.*]] = getelementptr inbounds [123 x i32], [123 x i32]* [[A11]], i32 0, i32 [[TMP17]] 2602 // CHECK23-NEXT: store i32 [[CONV10]], i32* [[ARRAYIDX12]], align 4 2603 // CHECK23-NEXT: br label [[OMP_BODY_CONTINUE13:%.*]] 2604 // CHECK23: omp.body.continue13: 2605 // CHECK23-NEXT: br label [[OMP_INNER_FOR_INC14:%.*]] 2606 // CHECK23: omp.inner.for.inc14: 2607 // CHECK23-NEXT: [[TMP18:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 2608 // CHECK23-NEXT: [[ADD15:%.*]] = add nsw i32 [[TMP18]], 1 2609 // CHECK23-NEXT: store i32 [[ADD15]], i32* [[DOTOMP_IV]], align 4 2610 // CHECK23-NEXT: br label [[OMP_INNER_FOR_COND4]], !llvm.loop [[LOOP10:![0-9]+]] 2611 // CHECK23: omp.inner.for.end16: 2612 // CHECK23-NEXT: br label [[OMP_IF_END]] 2613 // CHECK23: omp_if.end: 2614 // CHECK23-NEXT: br label [[OMP_LOOP_EXIT:%.*]] 2615 // CHECK23: omp.loop.exit: 2616 // CHECK23-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP2]]) 2617 // CHECK23-NEXT: [[TMP19:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4 2618 // CHECK23-NEXT: [[TMP20:%.*]] = icmp ne i32 [[TMP19]], 0 2619 // CHECK23-NEXT: br i1 [[TMP20]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]] 2620 // CHECK23: .omp.final.then: 2621 // CHECK23-NEXT: store i32 123, i32* [[I]], align 4 2622 // CHECK23-NEXT: br label [[DOTOMP_FINAL_DONE]] 2623 // CHECK23: .omp.final.done: 2624 // CHECK23-NEXT: ret void 2625 // 2626 // 2627 // CHECK23-LABEL: define {{[^@]+}}@.omp_offloading.requires_reg 2628 // CHECK23-SAME: () #[[ATTR4:[0-9]+]] { 2629 // CHECK23-NEXT: entry: 2630 // CHECK23-NEXT: call void @__tgt_register_requires(i64 1) 2631 // CHECK23-NEXT: ret void 2632 // 2633 // 2634 // CHECK25-LABEL: define {{[^@]+}}@_Z21teams_template_structv 2635 // CHECK25-SAME: () #[[ATTR0:[0-9]+]] { 2636 // CHECK25-NEXT: entry: 2637 // CHECK25-NEXT: [[V:%.*]] = alloca [[STRUCT_SS:%.*]], align 4 2638 // CHECK25-NEXT: [[CALL:%.*]] = call noundef signext i32 @_ZN2SSIiLi123ELx456EE3fooEv(%struct.SS* noundef nonnull align 4 dereferenceable(496) [[V]]) 2639 // CHECK25-NEXT: ret i32 [[CALL]] 2640 // 2641 // 2642 // CHECK25-LABEL: define {{[^@]+}}@_ZN2SSIiLi123ELx456EE3fooEv 2643 // CHECK25-SAME: (%struct.SS* noundef nonnull align 4 dereferenceable(496) [[THIS:%.*]]) #[[ATTR0]] comdat align 2 { 2644 // CHECK25-NEXT: entry: 2645 // CHECK25-NEXT: [[THIS_ADDR:%.*]] = alloca %struct.SS*, align 8 2646 // CHECK25-NEXT: [[TMP:%.*]] = alloca i32, align 4 2647 // CHECK25-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 2648 // CHECK25-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4 2649 // CHECK25-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 2650 // CHECK25-NEXT: [[I:%.*]] = alloca i32, align 4 2651 // CHECK25-NEXT: store %struct.SS* [[THIS]], %struct.SS** [[THIS_ADDR]], align 8 2652 // CHECK25-NEXT: [[THIS1:%.*]] = load %struct.SS*, %struct.SS** [[THIS_ADDR]], align 8 2653 // CHECK25-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4 2654 // CHECK25-NEXT: store i32 122, i32* [[DOTOMP_UB]], align 4 2655 // CHECK25-NEXT: [[TMP0:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4 2656 // CHECK25-NEXT: store i32 [[TMP0]], i32* [[DOTOMP_IV]], align 4 2657 // CHECK25-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] 2658 // CHECK25: omp.inner.for.cond: 2659 // CHECK25-NEXT: [[TMP1:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !2 2660 // CHECK25-NEXT: [[TMP2:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4, !llvm.access.group !2 2661 // CHECK25-NEXT: [[CMP:%.*]] = icmp sle i32 [[TMP1]], [[TMP2]] 2662 // CHECK25-NEXT: br i1 [[CMP]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] 2663 // CHECK25: omp.inner.for.body: 2664 // CHECK25-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !2 2665 // CHECK25-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP3]], 1 2666 // CHECK25-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] 2667 // CHECK25-NEXT: store i32 [[ADD]], i32* [[I]], align 4, !llvm.access.group !2 2668 // CHECK25-NEXT: [[B:%.*]] = getelementptr inbounds [[STRUCT_SS:%.*]], %struct.SS* [[THIS1]], i32 0, i32 1 2669 // CHECK25-NEXT: [[TMP4:%.*]] = load float, float* [[B]], align 4, !llvm.access.group !2 2670 // CHECK25-NEXT: [[CONV:%.*]] = fptosi float [[TMP4]] to i32 2671 // CHECK25-NEXT: [[A:%.*]] = getelementptr inbounds [[STRUCT_SS]], %struct.SS* [[THIS1]], i32 0, i32 0 2672 // CHECK25-NEXT: [[TMP5:%.*]] = load i32, i32* [[I]], align 4, !llvm.access.group !2 2673 // CHECK25-NEXT: [[IDXPROM:%.*]] = sext i32 [[TMP5]] to i64 2674 // CHECK25-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [123 x i32], [123 x i32]* [[A]], i64 0, i64 [[IDXPROM]] 2675 // CHECK25-NEXT: store i32 [[CONV]], i32* [[ARRAYIDX]], align 4, !llvm.access.group !2 2676 // CHECK25-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] 2677 // CHECK25: omp.body.continue: 2678 // CHECK25-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] 2679 // CHECK25: omp.inner.for.inc: 2680 // CHECK25-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !2 2681 // CHECK25-NEXT: [[ADD2:%.*]] = add nsw i32 [[TMP6]], 1 2682 // CHECK25-NEXT: store i32 [[ADD2]], i32* [[DOTOMP_IV]], align 4, !llvm.access.group !2 2683 // CHECK25-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP3:![0-9]+]] 2684 // CHECK25: omp.inner.for.end: 2685 // CHECK25-NEXT: store i32 123, i32* [[I]], align 4 2686 // CHECK25-NEXT: [[A3:%.*]] = getelementptr inbounds [[STRUCT_SS]], %struct.SS* [[THIS1]], i32 0, i32 0 2687 // CHECK25-NEXT: [[ARRAYIDX4:%.*]] = getelementptr inbounds [123 x i32], [123 x i32]* [[A3]], i64 0, i64 0 2688 // CHECK25-NEXT: [[TMP7:%.*]] = load i32, i32* [[ARRAYIDX4]], align 4 2689 // CHECK25-NEXT: ret i32 [[TMP7]] 2690 // 2691 // 2692 // CHECK27-LABEL: define {{[^@]+}}@_Z21teams_template_structv 2693 // CHECK27-SAME: () #[[ATTR0:[0-9]+]] { 2694 // CHECK27-NEXT: entry: 2695 // CHECK27-NEXT: [[V:%.*]] = alloca [[STRUCT_SS:%.*]], align 4 2696 // CHECK27-NEXT: [[CALL:%.*]] = call noundef i32 @_ZN2SSIiLi123ELx456EE3fooEv(%struct.SS* noundef nonnull align 4 dereferenceable(496) [[V]]) 2697 // CHECK27-NEXT: ret i32 [[CALL]] 2698 // 2699 // 2700 // CHECK27-LABEL: define {{[^@]+}}@_ZN2SSIiLi123ELx456EE3fooEv 2701 // CHECK27-SAME: (%struct.SS* noundef nonnull align 4 dereferenceable(496) [[THIS:%.*]]) #[[ATTR0]] comdat align 2 { 2702 // CHECK27-NEXT: entry: 2703 // CHECK27-NEXT: [[THIS_ADDR:%.*]] = alloca %struct.SS*, align 4 2704 // CHECK27-NEXT: [[TMP:%.*]] = alloca i32, align 4 2705 // CHECK27-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 2706 // CHECK27-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4 2707 // CHECK27-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 2708 // CHECK27-NEXT: [[I:%.*]] = alloca i32, align 4 2709 // CHECK27-NEXT: store %struct.SS* [[THIS]], %struct.SS** [[THIS_ADDR]], align 4 2710 // CHECK27-NEXT: [[THIS1:%.*]] = load %struct.SS*, %struct.SS** [[THIS_ADDR]], align 4 2711 // CHECK27-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4 2712 // CHECK27-NEXT: store i32 122, i32* [[DOTOMP_UB]], align 4 2713 // CHECK27-NEXT: [[TMP0:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4 2714 // CHECK27-NEXT: store i32 [[TMP0]], i32* [[DOTOMP_IV]], align 4 2715 // CHECK27-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] 2716 // CHECK27: omp.inner.for.cond: 2717 // CHECK27-NEXT: [[TMP1:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !3 2718 // CHECK27-NEXT: [[TMP2:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4, !llvm.access.group !3 2719 // CHECK27-NEXT: [[CMP:%.*]] = icmp sle i32 [[TMP1]], [[TMP2]] 2720 // CHECK27-NEXT: br i1 [[CMP]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] 2721 // CHECK27: omp.inner.for.body: 2722 // CHECK27-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !3 2723 // CHECK27-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP3]], 1 2724 // CHECK27-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] 2725 // CHECK27-NEXT: store i32 [[ADD]], i32* [[I]], align 4, !llvm.access.group !3 2726 // CHECK27-NEXT: [[B:%.*]] = getelementptr inbounds [[STRUCT_SS:%.*]], %struct.SS* [[THIS1]], i32 0, i32 1 2727 // CHECK27-NEXT: [[TMP4:%.*]] = load float, float* [[B]], align 4, !llvm.access.group !3 2728 // CHECK27-NEXT: [[CONV:%.*]] = fptosi float [[TMP4]] to i32 2729 // CHECK27-NEXT: [[A:%.*]] = getelementptr inbounds [[STRUCT_SS]], %struct.SS* [[THIS1]], i32 0, i32 0 2730 // CHECK27-NEXT: [[TMP5:%.*]] = load i32, i32* [[I]], align 4, !llvm.access.group !3 2731 // CHECK27-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [123 x i32], [123 x i32]* [[A]], i32 0, i32 [[TMP5]] 2732 // CHECK27-NEXT: store i32 [[CONV]], i32* [[ARRAYIDX]], align 4, !llvm.access.group !3 2733 // CHECK27-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] 2734 // CHECK27: omp.body.continue: 2735 // CHECK27-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] 2736 // CHECK27: omp.inner.for.inc: 2737 // CHECK27-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !3 2738 // CHECK27-NEXT: [[ADD2:%.*]] = add nsw i32 [[TMP6]], 1 2739 // CHECK27-NEXT: store i32 [[ADD2]], i32* [[DOTOMP_IV]], align 4, !llvm.access.group !3 2740 // CHECK27-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP4:![0-9]+]] 2741 // CHECK27: omp.inner.for.end: 2742 // CHECK27-NEXT: store i32 123, i32* [[I]], align 4 2743 // CHECK27-NEXT: [[A3:%.*]] = getelementptr inbounds [[STRUCT_SS]], %struct.SS* [[THIS1]], i32 0, i32 0 2744 // CHECK27-NEXT: [[ARRAYIDX4:%.*]] = getelementptr inbounds [123 x i32], [123 x i32]* [[A3]], i32 0, i32 0 2745 // CHECK27-NEXT: [[TMP7:%.*]] = load i32, i32* [[ARRAYIDX4]], align 4 2746 // CHECK27-NEXT: ret i32 [[TMP7]] 2747 // 2748 // 2749 // CHECK29-LABEL: define {{[^@]+}}@_Z21teams_template_structv 2750 // CHECK29-SAME: () #[[ATTR0:[0-9]+]] { 2751 // CHECK29-NEXT: entry: 2752 // CHECK29-NEXT: [[V:%.*]] = alloca [[STRUCT_SS:%.*]], align 4 2753 // CHECK29-NEXT: [[CALL:%.*]] = call noundef signext i32 @_ZN2SSIiLi123ELx456EE3fooEv(%struct.SS* noundef nonnull align 4 dereferenceable(496) [[V]]) 2754 // CHECK29-NEXT: ret i32 [[CALL]] 2755 // 2756 // 2757 // CHECK29-LABEL: define {{[^@]+}}@_ZN2SSIiLi123ELx456EE3fooEv 2758 // CHECK29-SAME: (%struct.SS* noundef nonnull align 4 dereferenceable(496) [[THIS:%.*]]) #[[ATTR0]] comdat align 2 { 2759 // CHECK29-NEXT: entry: 2760 // CHECK29-NEXT: [[THIS_ADDR:%.*]] = alloca %struct.SS*, align 8 2761 // CHECK29-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i8, align 1 2762 // CHECK29-NEXT: [[TMP:%.*]] = alloca i32, align 4 2763 // CHECK29-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 2764 // CHECK29-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4 2765 // CHECK29-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 2766 // CHECK29-NEXT: [[I:%.*]] = alloca i32, align 4 2767 // CHECK29-NEXT: store %struct.SS* [[THIS]], %struct.SS** [[THIS_ADDR]], align 8 2768 // CHECK29-NEXT: [[THIS1:%.*]] = load %struct.SS*, %struct.SS** [[THIS_ADDR]], align 8 2769 // CHECK29-NEXT: [[B:%.*]] = getelementptr inbounds [[STRUCT_SS:%.*]], %struct.SS* [[THIS1]], i32 0, i32 1 2770 // CHECK29-NEXT: [[TMP0:%.*]] = load float, float* [[B]], align 4 2771 // CHECK29-NEXT: [[TOBOOL:%.*]] = fcmp une float [[TMP0]], 0.000000e+00 2772 // CHECK29-NEXT: [[FROMBOOL:%.*]] = zext i1 [[TOBOOL]] to i8 2773 // CHECK29-NEXT: store i8 [[FROMBOOL]], i8* [[DOTCAPTURE_EXPR_]], align 1 2774 // CHECK29-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4 2775 // CHECK29-NEXT: store i32 122, i32* [[DOTOMP_UB]], align 4 2776 // CHECK29-NEXT: [[TMP1:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4 2777 // CHECK29-NEXT: store i32 [[TMP1]], i32* [[DOTOMP_IV]], align 4 2778 // CHECK29-NEXT: [[TMP2:%.*]] = load i8, i8* [[DOTCAPTURE_EXPR_]], align 1 2779 // CHECK29-NEXT: [[TOBOOL2:%.*]] = trunc i8 [[TMP2]] to i1 2780 // CHECK29-NEXT: br i1 [[TOBOOL2]], label [[OMP_IF_THEN:%.*]], label [[OMP_IF_ELSE:%.*]] 2781 // CHECK29: omp_if.then: 2782 // CHECK29-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] 2783 // CHECK29: omp.inner.for.cond: 2784 // CHECK29-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !2 2785 // CHECK29-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4, !llvm.access.group !2 2786 // CHECK29-NEXT: [[CMP:%.*]] = icmp sle i32 [[TMP3]], [[TMP4]] 2787 // CHECK29-NEXT: br i1 [[CMP]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] 2788 // CHECK29: omp.inner.for.body: 2789 // CHECK29-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !2 2790 // CHECK29-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP5]], 1 2791 // CHECK29-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] 2792 // CHECK29-NEXT: store i32 [[ADD]], i32* [[I]], align 4, !llvm.access.group !2 2793 // CHECK29-NEXT: [[B3:%.*]] = getelementptr inbounds [[STRUCT_SS]], %struct.SS* [[THIS1]], i32 0, i32 1 2794 // CHECK29-NEXT: [[TMP6:%.*]] = load float, float* [[B3]], align 4, !nontemporal !3, !llvm.access.group !2 2795 // CHECK29-NEXT: [[CONV:%.*]] = fptosi float [[TMP6]] to i32 2796 // CHECK29-NEXT: [[A:%.*]] = getelementptr inbounds [[STRUCT_SS]], %struct.SS* [[THIS1]], i32 0, i32 0 2797 // CHECK29-NEXT: [[TMP7:%.*]] = load i32, i32* [[I]], align 4, !llvm.access.group !2 2798 // CHECK29-NEXT: [[IDXPROM:%.*]] = sext i32 [[TMP7]] to i64 2799 // CHECK29-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [123 x i32], [123 x i32]* [[A]], i64 0, i64 [[IDXPROM]] 2800 // CHECK29-NEXT: store i32 [[CONV]], i32* [[ARRAYIDX]], align 4, !llvm.access.group !2 2801 // CHECK29-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] 2802 // CHECK29: omp.body.continue: 2803 // CHECK29-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] 2804 // CHECK29: omp.inner.for.inc: 2805 // CHECK29-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !2 2806 // CHECK29-NEXT: [[ADD4:%.*]] = add nsw i32 [[TMP8]], 1 2807 // CHECK29-NEXT: store i32 [[ADD4]], i32* [[DOTOMP_IV]], align 4, !llvm.access.group !2 2808 // CHECK29-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP4:![0-9]+]] 2809 // CHECK29: omp.inner.for.end: 2810 // CHECK29-NEXT: br label [[OMP_IF_END:%.*]] 2811 // CHECK29: omp_if.else: 2812 // CHECK29-NEXT: br label [[OMP_INNER_FOR_COND5:%.*]] 2813 // CHECK29: omp.inner.for.cond5: 2814 // CHECK29-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 2815 // CHECK29-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 2816 // CHECK29-NEXT: [[CMP6:%.*]] = icmp sle i32 [[TMP9]], [[TMP10]] 2817 // CHECK29-NEXT: br i1 [[CMP6]], label [[OMP_INNER_FOR_BODY7:%.*]], label [[OMP_INNER_FOR_END18:%.*]] 2818 // CHECK29: omp.inner.for.body7: 2819 // CHECK29-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 2820 // CHECK29-NEXT: [[MUL8:%.*]] = mul nsw i32 [[TMP11]], 1 2821 // CHECK29-NEXT: [[ADD9:%.*]] = add nsw i32 0, [[MUL8]] 2822 // CHECK29-NEXT: store i32 [[ADD9]], i32* [[I]], align 4 2823 // CHECK29-NEXT: [[B10:%.*]] = getelementptr inbounds [[STRUCT_SS]], %struct.SS* [[THIS1]], i32 0, i32 1 2824 // CHECK29-NEXT: [[TMP12:%.*]] = load float, float* [[B10]], align 4 2825 // CHECK29-NEXT: [[CONV11:%.*]] = fptosi float [[TMP12]] to i32 2826 // CHECK29-NEXT: [[A12:%.*]] = getelementptr inbounds [[STRUCT_SS]], %struct.SS* [[THIS1]], i32 0, i32 0 2827 // CHECK29-NEXT: [[TMP13:%.*]] = load i32, i32* [[I]], align 4 2828 // CHECK29-NEXT: [[IDXPROM13:%.*]] = sext i32 [[TMP13]] to i64 2829 // CHECK29-NEXT: [[ARRAYIDX14:%.*]] = getelementptr inbounds [123 x i32], [123 x i32]* [[A12]], i64 0, i64 [[IDXPROM13]] 2830 // CHECK29-NEXT: store i32 [[CONV11]], i32* [[ARRAYIDX14]], align 4 2831 // CHECK29-NEXT: br label [[OMP_BODY_CONTINUE15:%.*]] 2832 // CHECK29: omp.body.continue15: 2833 // CHECK29-NEXT: br label [[OMP_INNER_FOR_INC16:%.*]] 2834 // CHECK29: omp.inner.for.inc16: 2835 // CHECK29-NEXT: [[TMP14:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 2836 // CHECK29-NEXT: [[ADD17:%.*]] = add nsw i32 [[TMP14]], 1 2837 // CHECK29-NEXT: store i32 [[ADD17]], i32* [[DOTOMP_IV]], align 4 2838 // CHECK29-NEXT: br label [[OMP_INNER_FOR_COND5]], !llvm.loop [[LOOP7:![0-9]+]] 2839 // CHECK29: omp.inner.for.end18: 2840 // CHECK29-NEXT: br label [[OMP_IF_END]] 2841 // CHECK29: omp_if.end: 2842 // CHECK29-NEXT: store i32 123, i32* [[I]], align 4 2843 // CHECK29-NEXT: [[A19:%.*]] = getelementptr inbounds [[STRUCT_SS]], %struct.SS* [[THIS1]], i32 0, i32 0 2844 // CHECK29-NEXT: [[ARRAYIDX20:%.*]] = getelementptr inbounds [123 x i32], [123 x i32]* [[A19]], i64 0, i64 0 2845 // CHECK29-NEXT: [[TMP15:%.*]] = load i32, i32* [[ARRAYIDX20]], align 4 2846 // CHECK29-NEXT: ret i32 [[TMP15]] 2847 // 2848 // 2849 // CHECK31-LABEL: define {{[^@]+}}@_Z21teams_template_structv 2850 // CHECK31-SAME: () #[[ATTR0:[0-9]+]] { 2851 // CHECK31-NEXT: entry: 2852 // CHECK31-NEXT: [[V:%.*]] = alloca [[STRUCT_SS:%.*]], align 4 2853 // CHECK31-NEXT: [[CALL:%.*]] = call noundef i32 @_ZN2SSIiLi123ELx456EE3fooEv(%struct.SS* noundef nonnull align 4 dereferenceable(496) [[V]]) 2854 // CHECK31-NEXT: ret i32 [[CALL]] 2855 // 2856 // 2857 // CHECK31-LABEL: define {{[^@]+}}@_ZN2SSIiLi123ELx456EE3fooEv 2858 // CHECK31-SAME: (%struct.SS* noundef nonnull align 4 dereferenceable(496) [[THIS:%.*]]) #[[ATTR0]] comdat align 2 { 2859 // CHECK31-NEXT: entry: 2860 // CHECK31-NEXT: [[THIS_ADDR:%.*]] = alloca %struct.SS*, align 4 2861 // CHECK31-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i8, align 1 2862 // CHECK31-NEXT: [[TMP:%.*]] = alloca i32, align 4 2863 // CHECK31-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 2864 // CHECK31-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4 2865 // CHECK31-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 2866 // CHECK31-NEXT: [[I:%.*]] = alloca i32, align 4 2867 // CHECK31-NEXT: store %struct.SS* [[THIS]], %struct.SS** [[THIS_ADDR]], align 4 2868 // CHECK31-NEXT: [[THIS1:%.*]] = load %struct.SS*, %struct.SS** [[THIS_ADDR]], align 4 2869 // CHECK31-NEXT: [[B:%.*]] = getelementptr inbounds [[STRUCT_SS:%.*]], %struct.SS* [[THIS1]], i32 0, i32 1 2870 // CHECK31-NEXT: [[TMP0:%.*]] = load float, float* [[B]], align 4 2871 // CHECK31-NEXT: [[TOBOOL:%.*]] = fcmp une float [[TMP0]], 0.000000e+00 2872 // CHECK31-NEXT: [[FROMBOOL:%.*]] = zext i1 [[TOBOOL]] to i8 2873 // CHECK31-NEXT: store i8 [[FROMBOOL]], i8* [[DOTCAPTURE_EXPR_]], align 1 2874 // CHECK31-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4 2875 // CHECK31-NEXT: store i32 122, i32* [[DOTOMP_UB]], align 4 2876 // CHECK31-NEXT: [[TMP1:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4 2877 // CHECK31-NEXT: store i32 [[TMP1]], i32* [[DOTOMP_IV]], align 4 2878 // CHECK31-NEXT: [[TMP2:%.*]] = load i8, i8* [[DOTCAPTURE_EXPR_]], align 1 2879 // CHECK31-NEXT: [[TOBOOL2:%.*]] = trunc i8 [[TMP2]] to i1 2880 // CHECK31-NEXT: br i1 [[TOBOOL2]], label [[OMP_IF_THEN:%.*]], label [[OMP_IF_ELSE:%.*]] 2881 // CHECK31: omp_if.then: 2882 // CHECK31-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] 2883 // CHECK31: omp.inner.for.cond: 2884 // CHECK31-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !3 2885 // CHECK31-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4, !llvm.access.group !3 2886 // CHECK31-NEXT: [[CMP:%.*]] = icmp sle i32 [[TMP3]], [[TMP4]] 2887 // CHECK31-NEXT: br i1 [[CMP]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] 2888 // CHECK31: omp.inner.for.body: 2889 // CHECK31-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !3 2890 // CHECK31-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP5]], 1 2891 // CHECK31-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] 2892 // CHECK31-NEXT: store i32 [[ADD]], i32* [[I]], align 4, !llvm.access.group !3 2893 // CHECK31-NEXT: [[B3:%.*]] = getelementptr inbounds [[STRUCT_SS]], %struct.SS* [[THIS1]], i32 0, i32 1 2894 // CHECK31-NEXT: [[TMP6:%.*]] = load float, float* [[B3]], align 4, !nontemporal !4, !llvm.access.group !3 2895 // CHECK31-NEXT: [[CONV:%.*]] = fptosi float [[TMP6]] to i32 2896 // CHECK31-NEXT: [[A:%.*]] = getelementptr inbounds [[STRUCT_SS]], %struct.SS* [[THIS1]], i32 0, i32 0 2897 // CHECK31-NEXT: [[TMP7:%.*]] = load i32, i32* [[I]], align 4, !llvm.access.group !3 2898 // CHECK31-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [123 x i32], [123 x i32]* [[A]], i32 0, i32 [[TMP7]] 2899 // CHECK31-NEXT: store i32 [[CONV]], i32* [[ARRAYIDX]], align 4, !llvm.access.group !3 2900 // CHECK31-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] 2901 // CHECK31: omp.body.continue: 2902 // CHECK31-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] 2903 // CHECK31: omp.inner.for.inc: 2904 // CHECK31-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !3 2905 // CHECK31-NEXT: [[ADD4:%.*]] = add nsw i32 [[TMP8]], 1 2906 // CHECK31-NEXT: store i32 [[ADD4]], i32* [[DOTOMP_IV]], align 4, !llvm.access.group !3 2907 // CHECK31-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP5:![0-9]+]] 2908 // CHECK31: omp.inner.for.end: 2909 // CHECK31-NEXT: br label [[OMP_IF_END:%.*]] 2910 // CHECK31: omp_if.else: 2911 // CHECK31-NEXT: br label [[OMP_INNER_FOR_COND5:%.*]] 2912 // CHECK31: omp.inner.for.cond5: 2913 // CHECK31-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 2914 // CHECK31-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 2915 // CHECK31-NEXT: [[CMP6:%.*]] = icmp sle i32 [[TMP9]], [[TMP10]] 2916 // CHECK31-NEXT: br i1 [[CMP6]], label [[OMP_INNER_FOR_BODY7:%.*]], label [[OMP_INNER_FOR_END17:%.*]] 2917 // CHECK31: omp.inner.for.body7: 2918 // CHECK31-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 2919 // CHECK31-NEXT: [[MUL8:%.*]] = mul nsw i32 [[TMP11]], 1 2920 // CHECK31-NEXT: [[ADD9:%.*]] = add nsw i32 0, [[MUL8]] 2921 // CHECK31-NEXT: store i32 [[ADD9]], i32* [[I]], align 4 2922 // CHECK31-NEXT: [[B10:%.*]] = getelementptr inbounds [[STRUCT_SS]], %struct.SS* [[THIS1]], i32 0, i32 1 2923 // CHECK31-NEXT: [[TMP12:%.*]] = load float, float* [[B10]], align 4 2924 // CHECK31-NEXT: [[CONV11:%.*]] = fptosi float [[TMP12]] to i32 2925 // CHECK31-NEXT: [[A12:%.*]] = getelementptr inbounds [[STRUCT_SS]], %struct.SS* [[THIS1]], i32 0, i32 0 2926 // CHECK31-NEXT: [[TMP13:%.*]] = load i32, i32* [[I]], align 4 2927 // CHECK31-NEXT: [[ARRAYIDX13:%.*]] = getelementptr inbounds [123 x i32], [123 x i32]* [[A12]], i32 0, i32 [[TMP13]] 2928 // CHECK31-NEXT: store i32 [[CONV11]], i32* [[ARRAYIDX13]], align 4 2929 // CHECK31-NEXT: br label [[OMP_BODY_CONTINUE14:%.*]] 2930 // CHECK31: omp.body.continue14: 2931 // CHECK31-NEXT: br label [[OMP_INNER_FOR_INC15:%.*]] 2932 // CHECK31: omp.inner.for.inc15: 2933 // CHECK31-NEXT: [[TMP14:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 2934 // CHECK31-NEXT: [[ADD16:%.*]] = add nsw i32 [[TMP14]], 1 2935 // CHECK31-NEXT: store i32 [[ADD16]], i32* [[DOTOMP_IV]], align 4 2936 // CHECK31-NEXT: br label [[OMP_INNER_FOR_COND5]], !llvm.loop [[LOOP8:![0-9]+]] 2937 // CHECK31: omp.inner.for.end17: 2938 // CHECK31-NEXT: br label [[OMP_IF_END]] 2939 // CHECK31: omp_if.end: 2940 // CHECK31-NEXT: store i32 123, i32* [[I]], align 4 2941 // CHECK31-NEXT: [[A18:%.*]] = getelementptr inbounds [[STRUCT_SS]], %struct.SS* [[THIS1]], i32 0, i32 0 2942 // CHECK31-NEXT: [[ARRAYIDX19:%.*]] = getelementptr inbounds [123 x i32], [123 x i32]* [[A18]], i32 0, i32 0 2943 // CHECK31-NEXT: [[TMP15:%.*]] = load i32, i32* [[ARRAYIDX19]], align 4 2944 // CHECK31-NEXT: ret i32 [[TMP15]] 2945 // 2946 // 2947 // CHECK33-LABEL: define {{[^@]+}}@main 2948 // CHECK33-SAME: (i32 noundef signext [[ARGC:%.*]], i8** noundef [[ARGV:%.*]]) #[[ATTR0:[0-9]+]] { 2949 // CHECK33-NEXT: entry: 2950 // CHECK33-NEXT: [[RETVAL:%.*]] = alloca i32, align 4 2951 // CHECK33-NEXT: [[ARGC_ADDR:%.*]] = alloca i32, align 4 2952 // CHECK33-NEXT: [[ARGV_ADDR:%.*]] = alloca i8**, align 8 2953 // CHECK33-NEXT: [[N:%.*]] = alloca i32, align 4 2954 // CHECK33-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 8 2955 // CHECK33-NEXT: [[__VLA_EXPR0:%.*]] = alloca i64, align 8 2956 // CHECK33-NEXT: [[N_CASTED:%.*]] = alloca i64, align 8 2957 // CHECK33-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [3 x i8*], align 8 2958 // CHECK33-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [3 x i8*], align 8 2959 // CHECK33-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [3 x i8*], align 8 2960 // CHECK33-NEXT: [[DOTOFFLOAD_SIZES:%.*]] = alloca [3 x i64], align 8 2961 // CHECK33-NEXT: [[TMP:%.*]] = alloca i32, align 4 2962 // CHECK33-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 2963 // CHECK33-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 2964 // CHECK33-NEXT: store i32 0, i32* [[RETVAL]], align 4 2965 // CHECK33-NEXT: store i32 [[ARGC]], i32* [[ARGC_ADDR]], align 4 2966 // CHECK33-NEXT: store i8** [[ARGV]], i8*** [[ARGV_ADDR]], align 8 2967 // CHECK33-NEXT: store i32 100, i32* [[N]], align 4 2968 // CHECK33-NEXT: [[TMP0:%.*]] = load i32, i32* [[N]], align 4 2969 // CHECK33-NEXT: [[TMP1:%.*]] = zext i32 [[TMP0]] to i64 2970 // CHECK33-NEXT: [[TMP2:%.*]] = call i8* @llvm.stacksave() 2971 // CHECK33-NEXT: store i8* [[TMP2]], i8** [[SAVED_STACK]], align 8 2972 // CHECK33-NEXT: [[VLA:%.*]] = alloca i32, i64 [[TMP1]], align 4 2973 // CHECK33-NEXT: store i64 [[TMP1]], i64* [[__VLA_EXPR0]], align 8 2974 // CHECK33-NEXT: [[TMP3:%.*]] = load i32, i32* [[N]], align 4 2975 // CHECK33-NEXT: [[CONV:%.*]] = bitcast i64* [[N_CASTED]] to i32* 2976 // CHECK33-NEXT: store i32 [[TMP3]], i32* [[CONV]], align 4 2977 // CHECK33-NEXT: [[TMP4:%.*]] = load i64, i64* [[N_CASTED]], align 8 2978 // CHECK33-NEXT: [[TMP5:%.*]] = mul nuw i64 [[TMP1]], 4 2979 // CHECK33-NEXT: [[TMP6:%.*]] = bitcast [3 x i64]* [[DOTOFFLOAD_SIZES]] to i8* 2980 // CHECK33-NEXT: call void @llvm.memcpy.p0i8.p0i8.i64(i8* align 8 [[TMP6]], i8* align 8 bitcast ([3 x i64]* @.offload_sizes to i8*), i64 24, i1 false) 2981 // CHECK33-NEXT: [[TMP7:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 2982 // CHECK33-NEXT: [[TMP8:%.*]] = bitcast i8** [[TMP7]] to i64* 2983 // CHECK33-NEXT: store i64 [[TMP4]], i64* [[TMP8]], align 8 2984 // CHECK33-NEXT: [[TMP9:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 2985 // CHECK33-NEXT: [[TMP10:%.*]] = bitcast i8** [[TMP9]] to i64* 2986 // CHECK33-NEXT: store i64 [[TMP4]], i64* [[TMP10]], align 8 2987 // CHECK33-NEXT: [[TMP11:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0 2988 // CHECK33-NEXT: store i8* null, i8** [[TMP11]], align 8 2989 // CHECK33-NEXT: [[TMP12:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1 2990 // CHECK33-NEXT: [[TMP13:%.*]] = bitcast i8** [[TMP12]] to i64* 2991 // CHECK33-NEXT: store i64 [[TMP1]], i64* [[TMP13]], align 8 2992 // CHECK33-NEXT: [[TMP14:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1 2993 // CHECK33-NEXT: [[TMP15:%.*]] = bitcast i8** [[TMP14]] to i64* 2994 // CHECK33-NEXT: store i64 [[TMP1]], i64* [[TMP15]], align 8 2995 // CHECK33-NEXT: [[TMP16:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 1 2996 // CHECK33-NEXT: store i8* null, i8** [[TMP16]], align 8 2997 // CHECK33-NEXT: [[TMP17:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2 2998 // CHECK33-NEXT: [[TMP18:%.*]] = bitcast i8** [[TMP17]] to i32** 2999 // CHECK33-NEXT: store i32* [[VLA]], i32** [[TMP18]], align 8 3000 // CHECK33-NEXT: [[TMP19:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2 3001 // CHECK33-NEXT: [[TMP20:%.*]] = bitcast i8** [[TMP19]] to i32** 3002 // CHECK33-NEXT: store i32* [[VLA]], i32** [[TMP20]], align 8 3003 // CHECK33-NEXT: [[TMP21:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 2 3004 // CHECK33-NEXT: store i64 [[TMP5]], i64* [[TMP21]], align 8 3005 // CHECK33-NEXT: [[TMP22:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2 3006 // CHECK33-NEXT: store i8* null, i8** [[TMP22]], align 8 3007 // CHECK33-NEXT: [[TMP23:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 3008 // CHECK33-NEXT: [[TMP24:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 3009 // CHECK33-NEXT: [[TMP25:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0 3010 // CHECK33-NEXT: [[TMP26:%.*]] = load i32, i32* [[N]], align 4 3011 // CHECK33-NEXT: store i32 [[TMP26]], i32* [[DOTCAPTURE_EXPR_]], align 4 3012 // CHECK33-NEXT: [[TMP27:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 3013 // CHECK33-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP27]], 0 3014 // CHECK33-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 3015 // CHECK33-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1 3016 // CHECK33-NEXT: store i32 [[SUB2]], i32* [[DOTCAPTURE_EXPR_1]], align 4 3017 // CHECK33-NEXT: [[TMP28:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 3018 // CHECK33-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP28]], 1 3019 // CHECK33-NEXT: [[TMP29:%.*]] = zext i32 [[ADD]] to i64 3020 // CHECK33-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2:[0-9]+]], i64 -1, i64 [[TMP29]]) 3021 // CHECK33-NEXT: [[TMP30:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l192.region_id, i32 3, i8** [[TMP23]], i8** [[TMP24]], i64* [[TMP25]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 0, i32 1) 3022 // CHECK33-NEXT: [[TMP31:%.*]] = icmp ne i32 [[TMP30]], 0 3023 // CHECK33-NEXT: br i1 [[TMP31]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] 3024 // CHECK33: omp_offload.failed: 3025 // CHECK33-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l192(i64 [[TMP4]], i64 [[TMP1]], i32* [[VLA]]) #[[ATTR3:[0-9]+]] 3026 // CHECK33-NEXT: br label [[OMP_OFFLOAD_CONT]] 3027 // CHECK33: omp_offload.cont: 3028 // CHECK33-NEXT: [[TMP32:%.*]] = load i32, i32* [[ARGC_ADDR]], align 4 3029 // CHECK33-NEXT: [[CALL:%.*]] = call noundef signext i32 @_Z5tmainIiLi10EEiT_(i32 noundef signext [[TMP32]]) 3030 // CHECK33-NEXT: store i32 [[CALL]], i32* [[RETVAL]], align 4 3031 // CHECK33-NEXT: [[TMP33:%.*]] = load i8*, i8** [[SAVED_STACK]], align 8 3032 // CHECK33-NEXT: call void @llvm.stackrestore(i8* [[TMP33]]) 3033 // CHECK33-NEXT: [[TMP34:%.*]] = load i32, i32* [[RETVAL]], align 4 3034 // CHECK33-NEXT: ret i32 [[TMP34]] 3035 // 3036 // 3037 // CHECK33-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l192 3038 // CHECK33-SAME: (i64 noundef [[N:%.*]], i64 noundef [[VLA:%.*]], i32* noundef nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2:[0-9]+]] { 3039 // CHECK33-NEXT: entry: 3040 // CHECK33-NEXT: [[N_ADDR:%.*]] = alloca i64, align 8 3041 // CHECK33-NEXT: [[VLA_ADDR:%.*]] = alloca i64, align 8 3042 // CHECK33-NEXT: [[A_ADDR:%.*]] = alloca i32*, align 8 3043 // CHECK33-NEXT: store i64 [[N]], i64* [[N_ADDR]], align 8 3044 // CHECK33-NEXT: store i64 [[VLA]], i64* [[VLA_ADDR]], align 8 3045 // CHECK33-NEXT: store i32* [[A]], i32** [[A_ADDR]], align 8 3046 // CHECK33-NEXT: [[CONV:%.*]] = bitcast i64* [[N_ADDR]] to i32* 3047 // CHECK33-NEXT: [[TMP0:%.*]] = load i64, i64* [[VLA_ADDR]], align 8 3048 // CHECK33-NEXT: [[TMP1:%.*]] = load i32*, i32** [[A_ADDR]], align 8 3049 // CHECK33-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2]], i32 3, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32*, i64, i32*)* @.omp_outlined. to void (i32*, i32*, ...)*), i32* [[CONV]], i64 [[TMP0]], i32* [[TMP1]]) 3050 // CHECK33-NEXT: ret void 3051 // 3052 // 3053 // CHECK33-LABEL: define {{[^@]+}}@.omp_outlined. 3054 // CHECK33-SAME: (i32* noalias noundef [[DOTGLOBAL_TID_:%.*]], i32* noalias noundef [[DOTBOUND_TID_:%.*]], i32* noundef nonnull align 4 dereferenceable(4) [[N:%.*]], i64 noundef [[VLA:%.*]], i32* noundef nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2]] { 3055 // CHECK33-NEXT: entry: 3056 // CHECK33-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 3057 // CHECK33-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 3058 // CHECK33-NEXT: [[N_ADDR:%.*]] = alloca i32*, align 8 3059 // CHECK33-NEXT: [[VLA_ADDR:%.*]] = alloca i64, align 8 3060 // CHECK33-NEXT: [[A_ADDR:%.*]] = alloca i32*, align 8 3061 // CHECK33-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 3062 // CHECK33-NEXT: [[TMP:%.*]] = alloca i32, align 4 3063 // CHECK33-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 3064 // CHECK33-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 3065 // CHECK33-NEXT: [[I:%.*]] = alloca i32, align 4 3066 // CHECK33-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 3067 // CHECK33-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4 3068 // CHECK33-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 3069 // CHECK33-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 3070 // CHECK33-NEXT: [[I3:%.*]] = alloca i32, align 4 3071 // CHECK33-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8 3072 // CHECK33-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8 3073 // CHECK33-NEXT: store i32* [[N]], i32** [[N_ADDR]], align 8 3074 // CHECK33-NEXT: store i64 [[VLA]], i64* [[VLA_ADDR]], align 8 3075 // CHECK33-NEXT: store i32* [[A]], i32** [[A_ADDR]], align 8 3076 // CHECK33-NEXT: [[TMP0:%.*]] = load i32*, i32** [[N_ADDR]], align 8 3077 // CHECK33-NEXT: [[TMP1:%.*]] = load i64, i64* [[VLA_ADDR]], align 8 3078 // CHECK33-NEXT: [[TMP2:%.*]] = load i32*, i32** [[A_ADDR]], align 8 3079 // CHECK33-NEXT: [[TMP3:%.*]] = load i32, i32* [[TMP0]], align 4 3080 // CHECK33-NEXT: store i32 [[TMP3]], i32* [[DOTCAPTURE_EXPR_]], align 4 3081 // CHECK33-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 3082 // CHECK33-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP4]], 0 3083 // CHECK33-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 3084 // CHECK33-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1 3085 // CHECK33-NEXT: store i32 [[SUB2]], i32* [[DOTCAPTURE_EXPR_1]], align 4 3086 // CHECK33-NEXT: store i32 0, i32* [[I]], align 4 3087 // CHECK33-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 3088 // CHECK33-NEXT: [[CMP:%.*]] = icmp slt i32 0, [[TMP5]] 3089 // CHECK33-NEXT: br i1 [[CMP]], label [[OMP_PRECOND_THEN:%.*]], label [[OMP_PRECOND_END:%.*]] 3090 // CHECK33: omp.precond.then: 3091 // CHECK33-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4 3092 // CHECK33-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 3093 // CHECK33-NEXT: store i32 [[TMP6]], i32* [[DOTOMP_UB]], align 4 3094 // CHECK33-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4 3095 // CHECK33-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4 3096 // CHECK33-NEXT: [[TMP7:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 3097 // CHECK33-NEXT: [[TMP8:%.*]] = load i32, i32* [[TMP7]], align 4 3098 // CHECK33-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1:[0-9]+]], i32 [[TMP8]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1) 3099 // CHECK33-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 3100 // CHECK33-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 3101 // CHECK33-NEXT: [[CMP4:%.*]] = icmp sgt i32 [[TMP9]], [[TMP10]] 3102 // CHECK33-NEXT: br i1 [[CMP4]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] 3103 // CHECK33: cond.true: 3104 // CHECK33-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 3105 // CHECK33-NEXT: br label [[COND_END:%.*]] 3106 // CHECK33: cond.false: 3107 // CHECK33-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 3108 // CHECK33-NEXT: br label [[COND_END]] 3109 // CHECK33: cond.end: 3110 // CHECK33-NEXT: [[COND:%.*]] = phi i32 [ [[TMP11]], [[COND_TRUE]] ], [ [[TMP12]], [[COND_FALSE]] ] 3111 // CHECK33-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4 3112 // CHECK33-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4 3113 // CHECK33-NEXT: store i32 [[TMP13]], i32* [[DOTOMP_IV]], align 4 3114 // CHECK33-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] 3115 // CHECK33: omp.inner.for.cond: 3116 // CHECK33-NEXT: [[TMP14:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !5 3117 // CHECK33-NEXT: [[TMP15:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4, !llvm.access.group !5 3118 // CHECK33-NEXT: [[CMP5:%.*]] = icmp sle i32 [[TMP14]], [[TMP15]] 3119 // CHECK33-NEXT: br i1 [[CMP5]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] 3120 // CHECK33: omp.inner.for.body: 3121 // CHECK33-NEXT: [[TMP16:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !5 3122 // CHECK33-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP16]], 1 3123 // CHECK33-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] 3124 // CHECK33-NEXT: store i32 [[ADD]], i32* [[I3]], align 4, !llvm.access.group !5 3125 // CHECK33-NEXT: [[TMP17:%.*]] = load i32, i32* [[I3]], align 4, !llvm.access.group !5 3126 // CHECK33-NEXT: [[IDXPROM:%.*]] = sext i32 [[TMP17]] to i64 3127 // CHECK33-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i32, i32* [[TMP2]], i64 [[IDXPROM]] 3128 // CHECK33-NEXT: store i32 0, i32* [[ARRAYIDX]], align 4, !llvm.access.group !5 3129 // CHECK33-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] 3130 // CHECK33: omp.body.continue: 3131 // CHECK33-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] 3132 // CHECK33: omp.inner.for.inc: 3133 // CHECK33-NEXT: [[TMP18:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !5 3134 // CHECK33-NEXT: [[ADD6:%.*]] = add nsw i32 [[TMP18]], 1 3135 // CHECK33-NEXT: store i32 [[ADD6]], i32* [[DOTOMP_IV]], align 4, !llvm.access.group !5 3136 // CHECK33-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP6:![0-9]+]] 3137 // CHECK33: omp.inner.for.end: 3138 // CHECK33-NEXT: br label [[OMP_LOOP_EXIT:%.*]] 3139 // CHECK33: omp.loop.exit: 3140 // CHECK33-NEXT: [[TMP19:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 3141 // CHECK33-NEXT: [[TMP20:%.*]] = load i32, i32* [[TMP19]], align 4 3142 // CHECK33-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP20]]) 3143 // CHECK33-NEXT: [[TMP21:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4 3144 // CHECK33-NEXT: [[TMP22:%.*]] = icmp ne i32 [[TMP21]], 0 3145 // CHECK33-NEXT: br i1 [[TMP22]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]] 3146 // CHECK33: .omp.final.then: 3147 // CHECK33-NEXT: [[TMP23:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 3148 // CHECK33-NEXT: [[SUB7:%.*]] = sub nsw i32 [[TMP23]], 0 3149 // CHECK33-NEXT: [[DIV8:%.*]] = sdiv i32 [[SUB7]], 1 3150 // CHECK33-NEXT: [[MUL9:%.*]] = mul nsw i32 [[DIV8]], 1 3151 // CHECK33-NEXT: [[ADD10:%.*]] = add nsw i32 0, [[MUL9]] 3152 // CHECK33-NEXT: store i32 [[ADD10]], i32* [[I3]], align 4 3153 // CHECK33-NEXT: br label [[DOTOMP_FINAL_DONE]] 3154 // CHECK33: .omp.final.done: 3155 // CHECK33-NEXT: br label [[OMP_PRECOND_END]] 3156 // CHECK33: omp.precond.end: 3157 // CHECK33-NEXT: ret void 3158 // 3159 // 3160 // CHECK33-LABEL: define {{[^@]+}}@_Z5tmainIiLi10EEiT_ 3161 // CHECK33-SAME: (i32 noundef signext [[ARGC:%.*]]) #[[ATTR5:[0-9]+]] comdat { 3162 // CHECK33-NEXT: entry: 3163 // CHECK33-NEXT: [[ARGC_ADDR:%.*]] = alloca i32, align 4 3164 // CHECK33-NEXT: [[A:%.*]] = alloca [10 x i32], align 4 3165 // CHECK33-NEXT: [[TE:%.*]] = alloca i32, align 4 3166 // CHECK33-NEXT: [[TH:%.*]] = alloca i32, align 4 3167 // CHECK33-NEXT: [[TE_CASTED:%.*]] = alloca i64, align 8 3168 // CHECK33-NEXT: [[TH_CASTED:%.*]] = alloca i64, align 8 3169 // CHECK33-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [3 x i8*], align 8 3170 // CHECK33-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [3 x i8*], align 8 3171 // CHECK33-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [3 x i8*], align 8 3172 // CHECK33-NEXT: [[TMP:%.*]] = alloca i32, align 4 3173 // CHECK33-NEXT: store i32 [[ARGC]], i32* [[ARGC_ADDR]], align 4 3174 // CHECK33-NEXT: store i32 0, i32* [[TE]], align 4 3175 // CHECK33-NEXT: store i32 128, i32* [[TH]], align 4 3176 // CHECK33-NEXT: [[TMP0:%.*]] = load i32, i32* [[TE]], align 4 3177 // CHECK33-NEXT: [[CONV:%.*]] = bitcast i64* [[TE_CASTED]] to i32* 3178 // CHECK33-NEXT: store i32 [[TMP0]], i32* [[CONV]], align 4 3179 // CHECK33-NEXT: [[TMP1:%.*]] = load i64, i64* [[TE_CASTED]], align 8 3180 // CHECK33-NEXT: [[TMP2:%.*]] = load i32, i32* [[TH]], align 4 3181 // CHECK33-NEXT: [[CONV1:%.*]] = bitcast i64* [[TH_CASTED]] to i32* 3182 // CHECK33-NEXT: store i32 [[TMP2]], i32* [[CONV1]], align 4 3183 // CHECK33-NEXT: [[TMP3:%.*]] = load i64, i64* [[TH_CASTED]], align 8 3184 // CHECK33-NEXT: [[TMP4:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 3185 // CHECK33-NEXT: [[TMP5:%.*]] = bitcast i8** [[TMP4]] to i64* 3186 // CHECK33-NEXT: store i64 [[TMP1]], i64* [[TMP5]], align 8 3187 // CHECK33-NEXT: [[TMP6:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 3188 // CHECK33-NEXT: [[TMP7:%.*]] = bitcast i8** [[TMP6]] to i64* 3189 // CHECK33-NEXT: store i64 [[TMP1]], i64* [[TMP7]], align 8 3190 // CHECK33-NEXT: [[TMP8:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0 3191 // CHECK33-NEXT: store i8* null, i8** [[TMP8]], align 8 3192 // CHECK33-NEXT: [[TMP9:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1 3193 // CHECK33-NEXT: [[TMP10:%.*]] = bitcast i8** [[TMP9]] to i64* 3194 // CHECK33-NEXT: store i64 [[TMP3]], i64* [[TMP10]], align 8 3195 // CHECK33-NEXT: [[TMP11:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1 3196 // CHECK33-NEXT: [[TMP12:%.*]] = bitcast i8** [[TMP11]] to i64* 3197 // CHECK33-NEXT: store i64 [[TMP3]], i64* [[TMP12]], align 8 3198 // CHECK33-NEXT: [[TMP13:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 1 3199 // CHECK33-NEXT: store i8* null, i8** [[TMP13]], align 8 3200 // CHECK33-NEXT: [[TMP14:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2 3201 // CHECK33-NEXT: [[TMP15:%.*]] = bitcast i8** [[TMP14]] to [10 x i32]** 3202 // CHECK33-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP15]], align 8 3203 // CHECK33-NEXT: [[TMP16:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2 3204 // CHECK33-NEXT: [[TMP17:%.*]] = bitcast i8** [[TMP16]] to [10 x i32]** 3205 // CHECK33-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP17]], align 8 3206 // CHECK33-NEXT: [[TMP18:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2 3207 // CHECK33-NEXT: store i8* null, i8** [[TMP18]], align 8 3208 // CHECK33-NEXT: [[TMP19:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 3209 // CHECK33-NEXT: [[TMP20:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 3210 // CHECK33-NEXT: [[TMP21:%.*]] = load i32, i32* [[TE]], align 4 3211 // CHECK33-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 10) 3212 // CHECK33-NEXT: [[TMP22:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l181.region_id, i32 3, i8** [[TMP19]], i8** [[TMP20]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes.2, i32 0, i32 0), i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes.3, i32 0, i32 0), i8** null, i8** null, i32 [[TMP21]], i32 1) 3213 // CHECK33-NEXT: [[TMP23:%.*]] = icmp ne i32 [[TMP22]], 0 3214 // CHECK33-NEXT: br i1 [[TMP23]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] 3215 // CHECK33: omp_offload.failed: 3216 // CHECK33-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l181(i64 [[TMP1]], i64 [[TMP3]], [10 x i32]* [[A]]) #[[ATTR3]] 3217 // CHECK33-NEXT: br label [[OMP_OFFLOAD_CONT]] 3218 // CHECK33: omp_offload.cont: 3219 // CHECK33-NEXT: ret i32 0 3220 // 3221 // 3222 // CHECK33-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l181 3223 // CHECK33-SAME: (i64 noundef [[TE:%.*]], i64 noundef [[TH:%.*]], [10 x i32]* noundef nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] { 3224 // CHECK33-NEXT: entry: 3225 // CHECK33-NEXT: [[TE_ADDR:%.*]] = alloca i64, align 8 3226 // CHECK33-NEXT: [[TH_ADDR:%.*]] = alloca i64, align 8 3227 // CHECK33-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 8 3228 // CHECK33-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB2]]) 3229 // CHECK33-NEXT: store i64 [[TE]], i64* [[TE_ADDR]], align 8 3230 // CHECK33-NEXT: store i64 [[TH]], i64* [[TH_ADDR]], align 8 3231 // CHECK33-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[A_ADDR]], align 8 3232 // CHECK33-NEXT: [[CONV:%.*]] = bitcast i64* [[TE_ADDR]] to i32* 3233 // CHECK33-NEXT: [[CONV1:%.*]] = bitcast i64* [[TH_ADDR]] to i32* 3234 // CHECK33-NEXT: [[TMP1:%.*]] = load [10 x i32]*, [10 x i32]** [[A_ADDR]], align 8 3235 // CHECK33-NEXT: [[TMP2:%.*]] = load i32, i32* [[CONV]], align 4 3236 // CHECK33-NEXT: [[TMP3:%.*]] = load i32, i32* [[CONV1]], align 4 3237 // CHECK33-NEXT: call void @__kmpc_push_num_teams(%struct.ident_t* @[[GLOB2]], i32 [[TMP0]], i32 [[TMP2]], i32 [[TMP3]]) 3238 // CHECK33-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, [10 x i32]*)* @.omp_outlined..1 to void (i32*, i32*, ...)*), [10 x i32]* [[TMP1]]) 3239 // CHECK33-NEXT: ret void 3240 // 3241 // 3242 // CHECK33-LABEL: define {{[^@]+}}@.omp_outlined..1 3243 // CHECK33-SAME: (i32* noalias noundef [[DOTGLOBAL_TID_:%.*]], i32* noalias noundef [[DOTBOUND_TID_:%.*]], [10 x i32]* noundef nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] { 3244 // CHECK33-NEXT: entry: 3245 // CHECK33-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 3246 // CHECK33-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 3247 // CHECK33-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 8 3248 // CHECK33-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 3249 // CHECK33-NEXT: [[TMP:%.*]] = alloca i32, align 4 3250 // CHECK33-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 3251 // CHECK33-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4 3252 // CHECK33-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 3253 // CHECK33-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 3254 // CHECK33-NEXT: [[I:%.*]] = alloca i32, align 4 3255 // CHECK33-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8 3256 // CHECK33-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8 3257 // CHECK33-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[A_ADDR]], align 8 3258 // CHECK33-NEXT: [[TMP0:%.*]] = load [10 x i32]*, [10 x i32]** [[A_ADDR]], align 8 3259 // CHECK33-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4 3260 // CHECK33-NEXT: store i32 9, i32* [[DOTOMP_UB]], align 4 3261 // CHECK33-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4 3262 // CHECK33-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4 3263 // CHECK33-NEXT: [[TMP1:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 3264 // CHECK33-NEXT: [[TMP2:%.*]] = load i32, i32* [[TMP1]], align 4 3265 // CHECK33-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1]], i32 [[TMP2]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1) 3266 // CHECK33-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 3267 // CHECK33-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP3]], 9 3268 // CHECK33-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] 3269 // CHECK33: cond.true: 3270 // CHECK33-NEXT: br label [[COND_END:%.*]] 3271 // CHECK33: cond.false: 3272 // CHECK33-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 3273 // CHECK33-NEXT: br label [[COND_END]] 3274 // CHECK33: cond.end: 3275 // CHECK33-NEXT: [[COND:%.*]] = phi i32 [ 9, [[COND_TRUE]] ], [ [[TMP4]], [[COND_FALSE]] ] 3276 // CHECK33-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4 3277 // CHECK33-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4 3278 // CHECK33-NEXT: store i32 [[TMP5]], i32* [[DOTOMP_IV]], align 4 3279 // CHECK33-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] 3280 // CHECK33: omp.inner.for.cond: 3281 // CHECK33-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !11 3282 // CHECK33-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4, !llvm.access.group !11 3283 // CHECK33-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP6]], [[TMP7]] 3284 // CHECK33-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] 3285 // CHECK33: omp.inner.for.body: 3286 // CHECK33-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !11 3287 // CHECK33-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP8]], 1 3288 // CHECK33-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] 3289 // CHECK33-NEXT: store i32 [[ADD]], i32* [[I]], align 4, !llvm.access.group !11 3290 // CHECK33-NEXT: [[TMP9:%.*]] = load i32, i32* [[I]], align 4, !llvm.access.group !11 3291 // CHECK33-NEXT: [[IDXPROM:%.*]] = sext i32 [[TMP9]] to i64 3292 // CHECK33-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [10 x i32], [10 x i32]* [[TMP0]], i64 0, i64 [[IDXPROM]] 3293 // CHECK33-NEXT: store i32 0, i32* [[ARRAYIDX]], align 4, !llvm.access.group !11 3294 // CHECK33-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] 3295 // CHECK33: omp.body.continue: 3296 // CHECK33-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] 3297 // CHECK33: omp.inner.for.inc: 3298 // CHECK33-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !11 3299 // CHECK33-NEXT: [[ADD2:%.*]] = add nsw i32 [[TMP10]], 1 3300 // CHECK33-NEXT: store i32 [[ADD2]], i32* [[DOTOMP_IV]], align 4, !llvm.access.group !11 3301 // CHECK33-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP12:![0-9]+]] 3302 // CHECK33: omp.inner.for.end: 3303 // CHECK33-NEXT: br label [[OMP_LOOP_EXIT:%.*]] 3304 // CHECK33: omp.loop.exit: 3305 // CHECK33-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP2]]) 3306 // CHECK33-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4 3307 // CHECK33-NEXT: [[TMP12:%.*]] = icmp ne i32 [[TMP11]], 0 3308 // CHECK33-NEXT: br i1 [[TMP12]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]] 3309 // CHECK33: .omp.final.then: 3310 // CHECK33-NEXT: store i32 10, i32* [[I]], align 4 3311 // CHECK33-NEXT: br label [[DOTOMP_FINAL_DONE]] 3312 // CHECK33: .omp.final.done: 3313 // CHECK33-NEXT: ret void 3314 // 3315 // 3316 // CHECK33-LABEL: define {{[^@]+}}@.omp_offloading.requires_reg 3317 // CHECK33-SAME: () #[[ATTR6:[0-9]+]] { 3318 // CHECK33-NEXT: entry: 3319 // CHECK33-NEXT: call void @__tgt_register_requires(i64 1) 3320 // CHECK33-NEXT: ret void 3321 // 3322 // 3323 // CHECK35-LABEL: define {{[^@]+}}@main 3324 // CHECK35-SAME: (i32 noundef [[ARGC:%.*]], i8** noundef [[ARGV:%.*]]) #[[ATTR0:[0-9]+]] { 3325 // CHECK35-NEXT: entry: 3326 // CHECK35-NEXT: [[RETVAL:%.*]] = alloca i32, align 4 3327 // CHECK35-NEXT: [[ARGC_ADDR:%.*]] = alloca i32, align 4 3328 // CHECK35-NEXT: [[ARGV_ADDR:%.*]] = alloca i8**, align 4 3329 // CHECK35-NEXT: [[N:%.*]] = alloca i32, align 4 3330 // CHECK35-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 4 3331 // CHECK35-NEXT: [[__VLA_EXPR0:%.*]] = alloca i32, align 4 3332 // CHECK35-NEXT: [[N_CASTED:%.*]] = alloca i32, align 4 3333 // CHECK35-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [3 x i8*], align 4 3334 // CHECK35-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [3 x i8*], align 4 3335 // CHECK35-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [3 x i8*], align 4 3336 // CHECK35-NEXT: [[DOTOFFLOAD_SIZES:%.*]] = alloca [3 x i64], align 4 3337 // CHECK35-NEXT: [[TMP:%.*]] = alloca i32, align 4 3338 // CHECK35-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 3339 // CHECK35-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 3340 // CHECK35-NEXT: store i32 0, i32* [[RETVAL]], align 4 3341 // CHECK35-NEXT: store i32 [[ARGC]], i32* [[ARGC_ADDR]], align 4 3342 // CHECK35-NEXT: store i8** [[ARGV]], i8*** [[ARGV_ADDR]], align 4 3343 // CHECK35-NEXT: store i32 100, i32* [[N]], align 4 3344 // CHECK35-NEXT: [[TMP0:%.*]] = load i32, i32* [[N]], align 4 3345 // CHECK35-NEXT: [[TMP1:%.*]] = call i8* @llvm.stacksave() 3346 // CHECK35-NEXT: store i8* [[TMP1]], i8** [[SAVED_STACK]], align 4 3347 // CHECK35-NEXT: [[VLA:%.*]] = alloca i32, i32 [[TMP0]], align 4 3348 // CHECK35-NEXT: store i32 [[TMP0]], i32* [[__VLA_EXPR0]], align 4 3349 // CHECK35-NEXT: [[TMP2:%.*]] = load i32, i32* [[N]], align 4 3350 // CHECK35-NEXT: store i32 [[TMP2]], i32* [[N_CASTED]], align 4 3351 // CHECK35-NEXT: [[TMP3:%.*]] = load i32, i32* [[N_CASTED]], align 4 3352 // CHECK35-NEXT: [[TMP4:%.*]] = mul nuw i32 [[TMP0]], 4 3353 // CHECK35-NEXT: [[TMP5:%.*]] = sext i32 [[TMP4]] to i64 3354 // CHECK35-NEXT: [[TMP6:%.*]] = bitcast [3 x i64]* [[DOTOFFLOAD_SIZES]] to i8* 3355 // CHECK35-NEXT: call void @llvm.memcpy.p0i8.p0i8.i32(i8* align 4 [[TMP6]], i8* align 4 bitcast ([3 x i64]* @.offload_sizes to i8*), i32 24, i1 false) 3356 // CHECK35-NEXT: [[TMP7:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 3357 // CHECK35-NEXT: [[TMP8:%.*]] = bitcast i8** [[TMP7]] to i32* 3358 // CHECK35-NEXT: store i32 [[TMP3]], i32* [[TMP8]], align 4 3359 // CHECK35-NEXT: [[TMP9:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 3360 // CHECK35-NEXT: [[TMP10:%.*]] = bitcast i8** [[TMP9]] to i32* 3361 // CHECK35-NEXT: store i32 [[TMP3]], i32* [[TMP10]], align 4 3362 // CHECK35-NEXT: [[TMP11:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0 3363 // CHECK35-NEXT: store i8* null, i8** [[TMP11]], align 4 3364 // CHECK35-NEXT: [[TMP12:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1 3365 // CHECK35-NEXT: [[TMP13:%.*]] = bitcast i8** [[TMP12]] to i32* 3366 // CHECK35-NEXT: store i32 [[TMP0]], i32* [[TMP13]], align 4 3367 // CHECK35-NEXT: [[TMP14:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1 3368 // CHECK35-NEXT: [[TMP15:%.*]] = bitcast i8** [[TMP14]] to i32* 3369 // CHECK35-NEXT: store i32 [[TMP0]], i32* [[TMP15]], align 4 3370 // CHECK35-NEXT: [[TMP16:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 1 3371 // CHECK35-NEXT: store i8* null, i8** [[TMP16]], align 4 3372 // CHECK35-NEXT: [[TMP17:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2 3373 // CHECK35-NEXT: [[TMP18:%.*]] = bitcast i8** [[TMP17]] to i32** 3374 // CHECK35-NEXT: store i32* [[VLA]], i32** [[TMP18]], align 4 3375 // CHECK35-NEXT: [[TMP19:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2 3376 // CHECK35-NEXT: [[TMP20:%.*]] = bitcast i8** [[TMP19]] to i32** 3377 // CHECK35-NEXT: store i32* [[VLA]], i32** [[TMP20]], align 4 3378 // CHECK35-NEXT: [[TMP21:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 2 3379 // CHECK35-NEXT: store i64 [[TMP5]], i64* [[TMP21]], align 4 3380 // CHECK35-NEXT: [[TMP22:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2 3381 // CHECK35-NEXT: store i8* null, i8** [[TMP22]], align 4 3382 // CHECK35-NEXT: [[TMP23:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 3383 // CHECK35-NEXT: [[TMP24:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 3384 // CHECK35-NEXT: [[TMP25:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0 3385 // CHECK35-NEXT: [[TMP26:%.*]] = load i32, i32* [[N]], align 4 3386 // CHECK35-NEXT: store i32 [[TMP26]], i32* [[DOTCAPTURE_EXPR_]], align 4 3387 // CHECK35-NEXT: [[TMP27:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 3388 // CHECK35-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP27]], 0 3389 // CHECK35-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 3390 // CHECK35-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1 3391 // CHECK35-NEXT: store i32 [[SUB2]], i32* [[DOTCAPTURE_EXPR_1]], align 4 3392 // CHECK35-NEXT: [[TMP28:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 3393 // CHECK35-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP28]], 1 3394 // CHECK35-NEXT: [[TMP29:%.*]] = zext i32 [[ADD]] to i64 3395 // CHECK35-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2:[0-9]+]], i64 -1, i64 [[TMP29]]) 3396 // CHECK35-NEXT: [[TMP30:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l192.region_id, i32 3, i8** [[TMP23]], i8** [[TMP24]], i64* [[TMP25]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 0, i32 1) 3397 // CHECK35-NEXT: [[TMP31:%.*]] = icmp ne i32 [[TMP30]], 0 3398 // CHECK35-NEXT: br i1 [[TMP31]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] 3399 // CHECK35: omp_offload.failed: 3400 // CHECK35-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l192(i32 [[TMP3]], i32 [[TMP0]], i32* [[VLA]]) #[[ATTR3:[0-9]+]] 3401 // CHECK35-NEXT: br label [[OMP_OFFLOAD_CONT]] 3402 // CHECK35: omp_offload.cont: 3403 // CHECK35-NEXT: [[TMP32:%.*]] = load i32, i32* [[ARGC_ADDR]], align 4 3404 // CHECK35-NEXT: [[CALL:%.*]] = call noundef i32 @_Z5tmainIiLi10EEiT_(i32 noundef [[TMP32]]) 3405 // CHECK35-NEXT: store i32 [[CALL]], i32* [[RETVAL]], align 4 3406 // CHECK35-NEXT: [[TMP33:%.*]] = load i8*, i8** [[SAVED_STACK]], align 4 3407 // CHECK35-NEXT: call void @llvm.stackrestore(i8* [[TMP33]]) 3408 // CHECK35-NEXT: [[TMP34:%.*]] = load i32, i32* [[RETVAL]], align 4 3409 // CHECK35-NEXT: ret i32 [[TMP34]] 3410 // 3411 // 3412 // CHECK35-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l192 3413 // CHECK35-SAME: (i32 noundef [[N:%.*]], i32 noundef [[VLA:%.*]], i32* noundef nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2:[0-9]+]] { 3414 // CHECK35-NEXT: entry: 3415 // CHECK35-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4 3416 // CHECK35-NEXT: [[VLA_ADDR:%.*]] = alloca i32, align 4 3417 // CHECK35-NEXT: [[A_ADDR:%.*]] = alloca i32*, align 4 3418 // CHECK35-NEXT: store i32 [[N]], i32* [[N_ADDR]], align 4 3419 // CHECK35-NEXT: store i32 [[VLA]], i32* [[VLA_ADDR]], align 4 3420 // CHECK35-NEXT: store i32* [[A]], i32** [[A_ADDR]], align 4 3421 // CHECK35-NEXT: [[TMP0:%.*]] = load i32, i32* [[VLA_ADDR]], align 4 3422 // CHECK35-NEXT: [[TMP1:%.*]] = load i32*, i32** [[A_ADDR]], align 4 3423 // CHECK35-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2]], i32 3, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32*, i32, i32*)* @.omp_outlined. to void (i32*, i32*, ...)*), i32* [[N_ADDR]], i32 [[TMP0]], i32* [[TMP1]]) 3424 // CHECK35-NEXT: ret void 3425 // 3426 // 3427 // CHECK35-LABEL: define {{[^@]+}}@.omp_outlined. 3428 // CHECK35-SAME: (i32* noalias noundef [[DOTGLOBAL_TID_:%.*]], i32* noalias noundef [[DOTBOUND_TID_:%.*]], i32* noundef nonnull align 4 dereferenceable(4) [[N:%.*]], i32 noundef [[VLA:%.*]], i32* noundef nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2]] { 3429 // CHECK35-NEXT: entry: 3430 // CHECK35-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4 3431 // CHECK35-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4 3432 // CHECK35-NEXT: [[N_ADDR:%.*]] = alloca i32*, align 4 3433 // CHECK35-NEXT: [[VLA_ADDR:%.*]] = alloca i32, align 4 3434 // CHECK35-NEXT: [[A_ADDR:%.*]] = alloca i32*, align 4 3435 // CHECK35-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 3436 // CHECK35-NEXT: [[TMP:%.*]] = alloca i32, align 4 3437 // CHECK35-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 3438 // CHECK35-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 3439 // CHECK35-NEXT: [[I:%.*]] = alloca i32, align 4 3440 // CHECK35-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 3441 // CHECK35-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4 3442 // CHECK35-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 3443 // CHECK35-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 3444 // CHECK35-NEXT: [[I3:%.*]] = alloca i32, align 4 3445 // CHECK35-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 4 3446 // CHECK35-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 4 3447 // CHECK35-NEXT: store i32* [[N]], i32** [[N_ADDR]], align 4 3448 // CHECK35-NEXT: store i32 [[VLA]], i32* [[VLA_ADDR]], align 4 3449 // CHECK35-NEXT: store i32* [[A]], i32** [[A_ADDR]], align 4 3450 // CHECK35-NEXT: [[TMP0:%.*]] = load i32*, i32** [[N_ADDR]], align 4 3451 // CHECK35-NEXT: [[TMP1:%.*]] = load i32, i32* [[VLA_ADDR]], align 4 3452 // CHECK35-NEXT: [[TMP2:%.*]] = load i32*, i32** [[A_ADDR]], align 4 3453 // CHECK35-NEXT: [[TMP3:%.*]] = load i32, i32* [[TMP0]], align 4 3454 // CHECK35-NEXT: store i32 [[TMP3]], i32* [[DOTCAPTURE_EXPR_]], align 4 3455 // CHECK35-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 3456 // CHECK35-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP4]], 0 3457 // CHECK35-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 3458 // CHECK35-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1 3459 // CHECK35-NEXT: store i32 [[SUB2]], i32* [[DOTCAPTURE_EXPR_1]], align 4 3460 // CHECK35-NEXT: store i32 0, i32* [[I]], align 4 3461 // CHECK35-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 3462 // CHECK35-NEXT: [[CMP:%.*]] = icmp slt i32 0, [[TMP5]] 3463 // CHECK35-NEXT: br i1 [[CMP]], label [[OMP_PRECOND_THEN:%.*]], label [[OMP_PRECOND_END:%.*]] 3464 // CHECK35: omp.precond.then: 3465 // CHECK35-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4 3466 // CHECK35-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 3467 // CHECK35-NEXT: store i32 [[TMP6]], i32* [[DOTOMP_UB]], align 4 3468 // CHECK35-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4 3469 // CHECK35-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4 3470 // CHECK35-NEXT: [[TMP7:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 4 3471 // CHECK35-NEXT: [[TMP8:%.*]] = load i32, i32* [[TMP7]], align 4 3472 // CHECK35-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1:[0-9]+]], i32 [[TMP8]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1) 3473 // CHECK35-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 3474 // CHECK35-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 3475 // CHECK35-NEXT: [[CMP4:%.*]] = icmp sgt i32 [[TMP9]], [[TMP10]] 3476 // CHECK35-NEXT: br i1 [[CMP4]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] 3477 // CHECK35: cond.true: 3478 // CHECK35-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 3479 // CHECK35-NEXT: br label [[COND_END:%.*]] 3480 // CHECK35: cond.false: 3481 // CHECK35-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 3482 // CHECK35-NEXT: br label [[COND_END]] 3483 // CHECK35: cond.end: 3484 // CHECK35-NEXT: [[COND:%.*]] = phi i32 [ [[TMP11]], [[COND_TRUE]] ], [ [[TMP12]], [[COND_FALSE]] ] 3485 // CHECK35-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4 3486 // CHECK35-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4 3487 // CHECK35-NEXT: store i32 [[TMP13]], i32* [[DOTOMP_IV]], align 4 3488 // CHECK35-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] 3489 // CHECK35: omp.inner.for.cond: 3490 // CHECK35-NEXT: [[TMP14:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !6 3491 // CHECK35-NEXT: [[TMP15:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4, !llvm.access.group !6 3492 // CHECK35-NEXT: [[CMP5:%.*]] = icmp sle i32 [[TMP14]], [[TMP15]] 3493 // CHECK35-NEXT: br i1 [[CMP5]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] 3494 // CHECK35: omp.inner.for.body: 3495 // CHECK35-NEXT: [[TMP16:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !6 3496 // CHECK35-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP16]], 1 3497 // CHECK35-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] 3498 // CHECK35-NEXT: store i32 [[ADD]], i32* [[I3]], align 4, !llvm.access.group !6 3499 // CHECK35-NEXT: [[TMP17:%.*]] = load i32, i32* [[I3]], align 4, !llvm.access.group !6 3500 // CHECK35-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i32, i32* [[TMP2]], i32 [[TMP17]] 3501 // CHECK35-NEXT: store i32 0, i32* [[ARRAYIDX]], align 4, !llvm.access.group !6 3502 // CHECK35-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] 3503 // CHECK35: omp.body.continue: 3504 // CHECK35-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] 3505 // CHECK35: omp.inner.for.inc: 3506 // CHECK35-NEXT: [[TMP18:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !6 3507 // CHECK35-NEXT: [[ADD6:%.*]] = add nsw i32 [[TMP18]], 1 3508 // CHECK35-NEXT: store i32 [[ADD6]], i32* [[DOTOMP_IV]], align 4, !llvm.access.group !6 3509 // CHECK35-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP7:![0-9]+]] 3510 // CHECK35: omp.inner.for.end: 3511 // CHECK35-NEXT: br label [[OMP_LOOP_EXIT:%.*]] 3512 // CHECK35: omp.loop.exit: 3513 // CHECK35-NEXT: [[TMP19:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 4 3514 // CHECK35-NEXT: [[TMP20:%.*]] = load i32, i32* [[TMP19]], align 4 3515 // CHECK35-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP20]]) 3516 // CHECK35-NEXT: [[TMP21:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4 3517 // CHECK35-NEXT: [[TMP22:%.*]] = icmp ne i32 [[TMP21]], 0 3518 // CHECK35-NEXT: br i1 [[TMP22]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]] 3519 // CHECK35: .omp.final.then: 3520 // CHECK35-NEXT: [[TMP23:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 3521 // CHECK35-NEXT: [[SUB7:%.*]] = sub nsw i32 [[TMP23]], 0 3522 // CHECK35-NEXT: [[DIV8:%.*]] = sdiv i32 [[SUB7]], 1 3523 // CHECK35-NEXT: [[MUL9:%.*]] = mul nsw i32 [[DIV8]], 1 3524 // CHECK35-NEXT: [[ADD10:%.*]] = add nsw i32 0, [[MUL9]] 3525 // CHECK35-NEXT: store i32 [[ADD10]], i32* [[I3]], align 4 3526 // CHECK35-NEXT: br label [[DOTOMP_FINAL_DONE]] 3527 // CHECK35: .omp.final.done: 3528 // CHECK35-NEXT: br label [[OMP_PRECOND_END]] 3529 // CHECK35: omp.precond.end: 3530 // CHECK35-NEXT: ret void 3531 // 3532 // 3533 // CHECK35-LABEL: define {{[^@]+}}@_Z5tmainIiLi10EEiT_ 3534 // CHECK35-SAME: (i32 noundef [[ARGC:%.*]]) #[[ATTR5:[0-9]+]] comdat { 3535 // CHECK35-NEXT: entry: 3536 // CHECK35-NEXT: [[ARGC_ADDR:%.*]] = alloca i32, align 4 3537 // CHECK35-NEXT: [[A:%.*]] = alloca [10 x i32], align 4 3538 // CHECK35-NEXT: [[TE:%.*]] = alloca i32, align 4 3539 // CHECK35-NEXT: [[TH:%.*]] = alloca i32, align 4 3540 // CHECK35-NEXT: [[TE_CASTED:%.*]] = alloca i32, align 4 3541 // CHECK35-NEXT: [[TH_CASTED:%.*]] = alloca i32, align 4 3542 // CHECK35-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [3 x i8*], align 4 3543 // CHECK35-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [3 x i8*], align 4 3544 // CHECK35-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [3 x i8*], align 4 3545 // CHECK35-NEXT: [[TMP:%.*]] = alloca i32, align 4 3546 // CHECK35-NEXT: store i32 [[ARGC]], i32* [[ARGC_ADDR]], align 4 3547 // CHECK35-NEXT: store i32 0, i32* [[TE]], align 4 3548 // CHECK35-NEXT: store i32 128, i32* [[TH]], align 4 3549 // CHECK35-NEXT: [[TMP0:%.*]] = load i32, i32* [[TE]], align 4 3550 // CHECK35-NEXT: store i32 [[TMP0]], i32* [[TE_CASTED]], align 4 3551 // CHECK35-NEXT: [[TMP1:%.*]] = load i32, i32* [[TE_CASTED]], align 4 3552 // CHECK35-NEXT: [[TMP2:%.*]] = load i32, i32* [[TH]], align 4 3553 // CHECK35-NEXT: store i32 [[TMP2]], i32* [[TH_CASTED]], align 4 3554 // CHECK35-NEXT: [[TMP3:%.*]] = load i32, i32* [[TH_CASTED]], align 4 3555 // CHECK35-NEXT: [[TMP4:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 3556 // CHECK35-NEXT: [[TMP5:%.*]] = bitcast i8** [[TMP4]] to i32* 3557 // CHECK35-NEXT: store i32 [[TMP1]], i32* [[TMP5]], align 4 3558 // CHECK35-NEXT: [[TMP6:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 3559 // CHECK35-NEXT: [[TMP7:%.*]] = bitcast i8** [[TMP6]] to i32* 3560 // CHECK35-NEXT: store i32 [[TMP1]], i32* [[TMP7]], align 4 3561 // CHECK35-NEXT: [[TMP8:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0 3562 // CHECK35-NEXT: store i8* null, i8** [[TMP8]], align 4 3563 // CHECK35-NEXT: [[TMP9:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1 3564 // CHECK35-NEXT: [[TMP10:%.*]] = bitcast i8** [[TMP9]] to i32* 3565 // CHECK35-NEXT: store i32 [[TMP3]], i32* [[TMP10]], align 4 3566 // CHECK35-NEXT: [[TMP11:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1 3567 // CHECK35-NEXT: [[TMP12:%.*]] = bitcast i8** [[TMP11]] to i32* 3568 // CHECK35-NEXT: store i32 [[TMP3]], i32* [[TMP12]], align 4 3569 // CHECK35-NEXT: [[TMP13:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 1 3570 // CHECK35-NEXT: store i8* null, i8** [[TMP13]], align 4 3571 // CHECK35-NEXT: [[TMP14:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2 3572 // CHECK35-NEXT: [[TMP15:%.*]] = bitcast i8** [[TMP14]] to [10 x i32]** 3573 // CHECK35-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP15]], align 4 3574 // CHECK35-NEXT: [[TMP16:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2 3575 // CHECK35-NEXT: [[TMP17:%.*]] = bitcast i8** [[TMP16]] to [10 x i32]** 3576 // CHECK35-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP17]], align 4 3577 // CHECK35-NEXT: [[TMP18:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2 3578 // CHECK35-NEXT: store i8* null, i8** [[TMP18]], align 4 3579 // CHECK35-NEXT: [[TMP19:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 3580 // CHECK35-NEXT: [[TMP20:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 3581 // CHECK35-NEXT: [[TMP21:%.*]] = load i32, i32* [[TE]], align 4 3582 // CHECK35-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 10) 3583 // CHECK35-NEXT: [[TMP22:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l181.region_id, i32 3, i8** [[TMP19]], i8** [[TMP20]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes.2, i32 0, i32 0), i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes.3, i32 0, i32 0), i8** null, i8** null, i32 [[TMP21]], i32 1) 3584 // CHECK35-NEXT: [[TMP23:%.*]] = icmp ne i32 [[TMP22]], 0 3585 // CHECK35-NEXT: br i1 [[TMP23]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] 3586 // CHECK35: omp_offload.failed: 3587 // CHECK35-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l181(i32 [[TMP1]], i32 [[TMP3]], [10 x i32]* [[A]]) #[[ATTR3]] 3588 // CHECK35-NEXT: br label [[OMP_OFFLOAD_CONT]] 3589 // CHECK35: omp_offload.cont: 3590 // CHECK35-NEXT: ret i32 0 3591 // 3592 // 3593 // CHECK35-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l181 3594 // CHECK35-SAME: (i32 noundef [[TE:%.*]], i32 noundef [[TH:%.*]], [10 x i32]* noundef nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] { 3595 // CHECK35-NEXT: entry: 3596 // CHECK35-NEXT: [[TE_ADDR:%.*]] = alloca i32, align 4 3597 // CHECK35-NEXT: [[TH_ADDR:%.*]] = alloca i32, align 4 3598 // CHECK35-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 4 3599 // CHECK35-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB2]]) 3600 // CHECK35-NEXT: store i32 [[TE]], i32* [[TE_ADDR]], align 4 3601 // CHECK35-NEXT: store i32 [[TH]], i32* [[TH_ADDR]], align 4 3602 // CHECK35-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[A_ADDR]], align 4 3603 // CHECK35-NEXT: [[TMP1:%.*]] = load [10 x i32]*, [10 x i32]** [[A_ADDR]], align 4 3604 // CHECK35-NEXT: [[TMP2:%.*]] = load i32, i32* [[TE_ADDR]], align 4 3605 // CHECK35-NEXT: [[TMP3:%.*]] = load i32, i32* [[TH_ADDR]], align 4 3606 // CHECK35-NEXT: call void @__kmpc_push_num_teams(%struct.ident_t* @[[GLOB2]], i32 [[TMP0]], i32 [[TMP2]], i32 [[TMP3]]) 3607 // CHECK35-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, [10 x i32]*)* @.omp_outlined..1 to void (i32*, i32*, ...)*), [10 x i32]* [[TMP1]]) 3608 // CHECK35-NEXT: ret void 3609 // 3610 // 3611 // CHECK35-LABEL: define {{[^@]+}}@.omp_outlined..1 3612 // CHECK35-SAME: (i32* noalias noundef [[DOTGLOBAL_TID_:%.*]], i32* noalias noundef [[DOTBOUND_TID_:%.*]], [10 x i32]* noundef nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] { 3613 // CHECK35-NEXT: entry: 3614 // CHECK35-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4 3615 // CHECK35-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4 3616 // CHECK35-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 4 3617 // CHECK35-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 3618 // CHECK35-NEXT: [[TMP:%.*]] = alloca i32, align 4 3619 // CHECK35-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 3620 // CHECK35-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4 3621 // CHECK35-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 3622 // CHECK35-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 3623 // CHECK35-NEXT: [[I:%.*]] = alloca i32, align 4 3624 // CHECK35-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 4 3625 // CHECK35-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 4 3626 // CHECK35-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[A_ADDR]], align 4 3627 // CHECK35-NEXT: [[TMP0:%.*]] = load [10 x i32]*, [10 x i32]** [[A_ADDR]], align 4 3628 // CHECK35-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4 3629 // CHECK35-NEXT: store i32 9, i32* [[DOTOMP_UB]], align 4 3630 // CHECK35-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4 3631 // CHECK35-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4 3632 // CHECK35-NEXT: [[TMP1:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 4 3633 // CHECK35-NEXT: [[TMP2:%.*]] = load i32, i32* [[TMP1]], align 4 3634 // CHECK35-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1]], i32 [[TMP2]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1) 3635 // CHECK35-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 3636 // CHECK35-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP3]], 9 3637 // CHECK35-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] 3638 // CHECK35: cond.true: 3639 // CHECK35-NEXT: br label [[COND_END:%.*]] 3640 // CHECK35: cond.false: 3641 // CHECK35-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 3642 // CHECK35-NEXT: br label [[COND_END]] 3643 // CHECK35: cond.end: 3644 // CHECK35-NEXT: [[COND:%.*]] = phi i32 [ 9, [[COND_TRUE]] ], [ [[TMP4]], [[COND_FALSE]] ] 3645 // CHECK35-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4 3646 // CHECK35-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4 3647 // CHECK35-NEXT: store i32 [[TMP5]], i32* [[DOTOMP_IV]], align 4 3648 // CHECK35-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] 3649 // CHECK35: omp.inner.for.cond: 3650 // CHECK35-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !12 3651 // CHECK35-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4, !llvm.access.group !12 3652 // CHECK35-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP6]], [[TMP7]] 3653 // CHECK35-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] 3654 // CHECK35: omp.inner.for.body: 3655 // CHECK35-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !12 3656 // CHECK35-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP8]], 1 3657 // CHECK35-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] 3658 // CHECK35-NEXT: store i32 [[ADD]], i32* [[I]], align 4, !llvm.access.group !12 3659 // CHECK35-NEXT: [[TMP9:%.*]] = load i32, i32* [[I]], align 4, !llvm.access.group !12 3660 // CHECK35-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [10 x i32], [10 x i32]* [[TMP0]], i32 0, i32 [[TMP9]] 3661 // CHECK35-NEXT: store i32 0, i32* [[ARRAYIDX]], align 4, !llvm.access.group !12 3662 // CHECK35-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] 3663 // CHECK35: omp.body.continue: 3664 // CHECK35-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] 3665 // CHECK35: omp.inner.for.inc: 3666 // CHECK35-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !12 3667 // CHECK35-NEXT: [[ADD2:%.*]] = add nsw i32 [[TMP10]], 1 3668 // CHECK35-NEXT: store i32 [[ADD2]], i32* [[DOTOMP_IV]], align 4, !llvm.access.group !12 3669 // CHECK35-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP13:![0-9]+]] 3670 // CHECK35: omp.inner.for.end: 3671 // CHECK35-NEXT: br label [[OMP_LOOP_EXIT:%.*]] 3672 // CHECK35: omp.loop.exit: 3673 // CHECK35-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP2]]) 3674 // CHECK35-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4 3675 // CHECK35-NEXT: [[TMP12:%.*]] = icmp ne i32 [[TMP11]], 0 3676 // CHECK35-NEXT: br i1 [[TMP12]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]] 3677 // CHECK35: .omp.final.then: 3678 // CHECK35-NEXT: store i32 10, i32* [[I]], align 4 3679 // CHECK35-NEXT: br label [[DOTOMP_FINAL_DONE]] 3680 // CHECK35: .omp.final.done: 3681 // CHECK35-NEXT: ret void 3682 // 3683 // 3684 // CHECK35-LABEL: define {{[^@]+}}@.omp_offloading.requires_reg 3685 // CHECK35-SAME: () #[[ATTR6:[0-9]+]] { 3686 // CHECK35-NEXT: entry: 3687 // CHECK35-NEXT: call void @__tgt_register_requires(i64 1) 3688 // CHECK35-NEXT: ret void 3689 // 3690 // 3691 // CHECK37-LABEL: define {{[^@]+}}@main 3692 // CHECK37-SAME: (i32 noundef signext [[ARGC:%.*]], i8** noundef [[ARGV:%.*]]) #[[ATTR0:[0-9]+]] { 3693 // CHECK37-NEXT: entry: 3694 // CHECK37-NEXT: [[RETVAL:%.*]] = alloca i32, align 4 3695 // CHECK37-NEXT: [[ARGC_ADDR:%.*]] = alloca i32, align 4 3696 // CHECK37-NEXT: [[ARGV_ADDR:%.*]] = alloca i8**, align 8 3697 // CHECK37-NEXT: [[N:%.*]] = alloca i32, align 4 3698 // CHECK37-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 8 3699 // CHECK37-NEXT: [[__VLA_EXPR0:%.*]] = alloca i64, align 8 3700 // CHECK37-NEXT: [[ARGC_CASTED:%.*]] = alloca i64, align 8 3701 // CHECK37-NEXT: [[N_CASTED:%.*]] = alloca i64, align 8 3702 // CHECK37-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [4 x i8*], align 8 3703 // CHECK37-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [4 x i8*], align 8 3704 // CHECK37-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [4 x i8*], align 8 3705 // CHECK37-NEXT: [[DOTOFFLOAD_SIZES:%.*]] = alloca [4 x i64], align 8 3706 // CHECK37-NEXT: [[TMP:%.*]] = alloca i32, align 4 3707 // CHECK37-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 3708 // CHECK37-NEXT: [[DOTCAPTURE_EXPR_2:%.*]] = alloca i32, align 4 3709 // CHECK37-NEXT: store i32 0, i32* [[RETVAL]], align 4 3710 // CHECK37-NEXT: store i32 [[ARGC]], i32* [[ARGC_ADDR]], align 4 3711 // CHECK37-NEXT: store i8** [[ARGV]], i8*** [[ARGV_ADDR]], align 8 3712 // CHECK37-NEXT: store i32 100, i32* [[N]], align 4 3713 // CHECK37-NEXT: [[TMP0:%.*]] = load i32, i32* [[N]], align 4 3714 // CHECK37-NEXT: [[TMP1:%.*]] = zext i32 [[TMP0]] to i64 3715 // CHECK37-NEXT: [[TMP2:%.*]] = call i8* @llvm.stacksave() 3716 // CHECK37-NEXT: store i8* [[TMP2]], i8** [[SAVED_STACK]], align 8 3717 // CHECK37-NEXT: [[VLA:%.*]] = alloca i32, i64 [[TMP1]], align 4 3718 // CHECK37-NEXT: store i64 [[TMP1]], i64* [[__VLA_EXPR0]], align 8 3719 // CHECK37-NEXT: [[TMP3:%.*]] = load i32, i32* [[ARGC_ADDR]], align 4 3720 // CHECK37-NEXT: [[CONV:%.*]] = bitcast i64* [[ARGC_CASTED]] to i32* 3721 // CHECK37-NEXT: store i32 [[TMP3]], i32* [[CONV]], align 4 3722 // CHECK37-NEXT: [[TMP4:%.*]] = load i64, i64* [[ARGC_CASTED]], align 8 3723 // CHECK37-NEXT: [[TMP5:%.*]] = load i32, i32* [[N]], align 4 3724 // CHECK37-NEXT: [[CONV1:%.*]] = bitcast i64* [[N_CASTED]] to i32* 3725 // CHECK37-NEXT: store i32 [[TMP5]], i32* [[CONV1]], align 4 3726 // CHECK37-NEXT: [[TMP6:%.*]] = load i64, i64* [[N_CASTED]], align 8 3727 // CHECK37-NEXT: [[TMP7:%.*]] = mul nuw i64 [[TMP1]], 4 3728 // CHECK37-NEXT: [[TMP8:%.*]] = bitcast [4 x i64]* [[DOTOFFLOAD_SIZES]] to i8* 3729 // CHECK37-NEXT: call void @llvm.memcpy.p0i8.p0i8.i64(i8* align 8 [[TMP8]], i8* align 8 bitcast ([4 x i64]* @.offload_sizes to i8*), i64 32, i1 false) 3730 // CHECK37-NEXT: [[TMP9:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 3731 // CHECK37-NEXT: [[TMP10:%.*]] = bitcast i8** [[TMP9]] to i64* 3732 // CHECK37-NEXT: store i64 [[TMP4]], i64* [[TMP10]], align 8 3733 // CHECK37-NEXT: [[TMP11:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 3734 // CHECK37-NEXT: [[TMP12:%.*]] = bitcast i8** [[TMP11]] to i64* 3735 // CHECK37-NEXT: store i64 [[TMP4]], i64* [[TMP12]], align 8 3736 // CHECK37-NEXT: [[TMP13:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0 3737 // CHECK37-NEXT: store i8* null, i8** [[TMP13]], align 8 3738 // CHECK37-NEXT: [[TMP14:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1 3739 // CHECK37-NEXT: [[TMP15:%.*]] = bitcast i8** [[TMP14]] to i64* 3740 // CHECK37-NEXT: store i64 [[TMP6]], i64* [[TMP15]], align 8 3741 // CHECK37-NEXT: [[TMP16:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1 3742 // CHECK37-NEXT: [[TMP17:%.*]] = bitcast i8** [[TMP16]] to i64* 3743 // CHECK37-NEXT: store i64 [[TMP6]], i64* [[TMP17]], align 8 3744 // CHECK37-NEXT: [[TMP18:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 1 3745 // CHECK37-NEXT: store i8* null, i8** [[TMP18]], align 8 3746 // CHECK37-NEXT: [[TMP19:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2 3747 // CHECK37-NEXT: [[TMP20:%.*]] = bitcast i8** [[TMP19]] to i64* 3748 // CHECK37-NEXT: store i64 [[TMP1]], i64* [[TMP20]], align 8 3749 // CHECK37-NEXT: [[TMP21:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2 3750 // CHECK37-NEXT: [[TMP22:%.*]] = bitcast i8** [[TMP21]] to i64* 3751 // CHECK37-NEXT: store i64 [[TMP1]], i64* [[TMP22]], align 8 3752 // CHECK37-NEXT: [[TMP23:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2 3753 // CHECK37-NEXT: store i8* null, i8** [[TMP23]], align 8 3754 // CHECK37-NEXT: [[TMP24:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 3 3755 // CHECK37-NEXT: [[TMP25:%.*]] = bitcast i8** [[TMP24]] to i32** 3756 // CHECK37-NEXT: store i32* [[VLA]], i32** [[TMP25]], align 8 3757 // CHECK37-NEXT: [[TMP26:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 3 3758 // CHECK37-NEXT: [[TMP27:%.*]] = bitcast i8** [[TMP26]] to i32** 3759 // CHECK37-NEXT: store i32* [[VLA]], i32** [[TMP27]], align 8 3760 // CHECK37-NEXT: [[TMP28:%.*]] = getelementptr inbounds [4 x i64], [4 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 3 3761 // CHECK37-NEXT: store i64 [[TMP7]], i64* [[TMP28]], align 8 3762 // CHECK37-NEXT: [[TMP29:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 3 3763 // CHECK37-NEXT: store i8* null, i8** [[TMP29]], align 8 3764 // CHECK37-NEXT: [[TMP30:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 3765 // CHECK37-NEXT: [[TMP31:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 3766 // CHECK37-NEXT: [[TMP32:%.*]] = getelementptr inbounds [4 x i64], [4 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0 3767 // CHECK37-NEXT: [[TMP33:%.*]] = load i32, i32* [[N]], align 4 3768 // CHECK37-NEXT: store i32 [[TMP33]], i32* [[DOTCAPTURE_EXPR_]], align 4 3769 // CHECK37-NEXT: [[TMP34:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 3770 // CHECK37-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP34]], 0 3771 // CHECK37-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 3772 // CHECK37-NEXT: [[SUB3:%.*]] = sub nsw i32 [[DIV]], 1 3773 // CHECK37-NEXT: store i32 [[SUB3]], i32* [[DOTCAPTURE_EXPR_2]], align 4 3774 // CHECK37-NEXT: [[TMP35:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_2]], align 4 3775 // CHECK37-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP35]], 1 3776 // CHECK37-NEXT: [[TMP36:%.*]] = zext i32 [[ADD]] to i64 3777 // CHECK37-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2:[0-9]+]], i64 -1, i64 [[TMP36]]) 3778 // CHECK37-NEXT: [[TMP37:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l192.region_id, i32 4, i8** [[TMP30]], i8** [[TMP31]], i64* [[TMP32]], i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 0, i32 1) 3779 // CHECK37-NEXT: [[TMP38:%.*]] = icmp ne i32 [[TMP37]], 0 3780 // CHECK37-NEXT: br i1 [[TMP38]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] 3781 // CHECK37: omp_offload.failed: 3782 // CHECK37-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l192(i64 [[TMP4]], i64 [[TMP6]], i64 [[TMP1]], i32* [[VLA]]) #[[ATTR3:[0-9]+]] 3783 // CHECK37-NEXT: br label [[OMP_OFFLOAD_CONT]] 3784 // CHECK37: omp_offload.cont: 3785 // CHECK37-NEXT: [[TMP39:%.*]] = load i32, i32* [[ARGC_ADDR]], align 4 3786 // CHECK37-NEXT: [[CALL:%.*]] = call noundef signext i32 @_Z5tmainIiLi10EEiT_(i32 noundef signext [[TMP39]]) 3787 // CHECK37-NEXT: store i32 [[CALL]], i32* [[RETVAL]], align 4 3788 // CHECK37-NEXT: [[TMP40:%.*]] = load i8*, i8** [[SAVED_STACK]], align 8 3789 // CHECK37-NEXT: call void @llvm.stackrestore(i8* [[TMP40]]) 3790 // CHECK37-NEXT: [[TMP41:%.*]] = load i32, i32* [[RETVAL]], align 4 3791 // CHECK37-NEXT: ret i32 [[TMP41]] 3792 // 3793 // 3794 // CHECK37-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l192 3795 // CHECK37-SAME: (i64 noundef [[ARGC:%.*]], i64 noundef [[N:%.*]], i64 noundef [[VLA:%.*]], i32* noundef nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2:[0-9]+]] { 3796 // CHECK37-NEXT: entry: 3797 // CHECK37-NEXT: [[ARGC_ADDR:%.*]] = alloca i64, align 8 3798 // CHECK37-NEXT: [[N_ADDR:%.*]] = alloca i64, align 8 3799 // CHECK37-NEXT: [[VLA_ADDR:%.*]] = alloca i64, align 8 3800 // CHECK37-NEXT: [[A_ADDR:%.*]] = alloca i32*, align 8 3801 // CHECK37-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i8, align 1 3802 // CHECK37-NEXT: [[DOTCAPTURE_EXPR__CASTED:%.*]] = alloca i64, align 8 3803 // CHECK37-NEXT: store i64 [[ARGC]], i64* [[ARGC_ADDR]], align 8 3804 // CHECK37-NEXT: store i64 [[N]], i64* [[N_ADDR]], align 8 3805 // CHECK37-NEXT: store i64 [[VLA]], i64* [[VLA_ADDR]], align 8 3806 // CHECK37-NEXT: store i32* [[A]], i32** [[A_ADDR]], align 8 3807 // CHECK37-NEXT: [[CONV:%.*]] = bitcast i64* [[ARGC_ADDR]] to i32* 3808 // CHECK37-NEXT: [[CONV1:%.*]] = bitcast i64* [[N_ADDR]] to i32* 3809 // CHECK37-NEXT: [[TMP0:%.*]] = load i64, i64* [[VLA_ADDR]], align 8 3810 // CHECK37-NEXT: [[TMP1:%.*]] = load i32*, i32** [[A_ADDR]], align 8 3811 // CHECK37-NEXT: [[TMP2:%.*]] = load i32, i32* [[CONV]], align 4 3812 // CHECK37-NEXT: [[TOBOOL:%.*]] = icmp ne i32 [[TMP2]], 0 3813 // CHECK37-NEXT: [[FROMBOOL:%.*]] = zext i1 [[TOBOOL]] to i8 3814 // CHECK37-NEXT: store i8 [[FROMBOOL]], i8* [[DOTCAPTURE_EXPR_]], align 1 3815 // CHECK37-NEXT: [[TMP3:%.*]] = load i8, i8* [[DOTCAPTURE_EXPR_]], align 1 3816 // CHECK37-NEXT: [[TOBOOL2:%.*]] = trunc i8 [[TMP3]] to i1 3817 // CHECK37-NEXT: [[CONV3:%.*]] = bitcast i64* [[DOTCAPTURE_EXPR__CASTED]] to i8* 3818 // CHECK37-NEXT: [[FROMBOOL4:%.*]] = zext i1 [[TOBOOL2]] to i8 3819 // CHECK37-NEXT: store i8 [[FROMBOOL4]], i8* [[CONV3]], align 1 3820 // CHECK37-NEXT: [[TMP4:%.*]] = load i64, i64* [[DOTCAPTURE_EXPR__CASTED]], align 8 3821 // CHECK37-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2]], i32 4, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32*, i64, i32*, i64)* @.omp_outlined. to void (i32*, i32*, ...)*), i32* [[CONV1]], i64 [[TMP0]], i32* [[TMP1]], i64 [[TMP4]]) 3822 // CHECK37-NEXT: ret void 3823 // 3824 // 3825 // CHECK37-LABEL: define {{[^@]+}}@.omp_outlined. 3826 // CHECK37-SAME: (i32* noalias noundef [[DOTGLOBAL_TID_:%.*]], i32* noalias noundef [[DOTBOUND_TID_:%.*]], i32* noundef nonnull align 4 dereferenceable(4) [[N:%.*]], i64 noundef [[VLA:%.*]], i32* noundef nonnull align 4 dereferenceable(4) [[A:%.*]], i64 noundef [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR2]] { 3827 // CHECK37-NEXT: entry: 3828 // CHECK37-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 3829 // CHECK37-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 3830 // CHECK37-NEXT: [[N_ADDR:%.*]] = alloca i32*, align 8 3831 // CHECK37-NEXT: [[VLA_ADDR:%.*]] = alloca i64, align 8 3832 // CHECK37-NEXT: [[A_ADDR:%.*]] = alloca i32*, align 8 3833 // CHECK37-NEXT: [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i64, align 8 3834 // CHECK37-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 3835 // CHECK37-NEXT: [[TMP:%.*]] = alloca i32, align 4 3836 // CHECK37-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 3837 // CHECK37-NEXT: [[DOTCAPTURE_EXPR_2:%.*]] = alloca i32, align 4 3838 // CHECK37-NEXT: [[I:%.*]] = alloca i32, align 4 3839 // CHECK37-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 3840 // CHECK37-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4 3841 // CHECK37-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 3842 // CHECK37-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 3843 // CHECK37-NEXT: [[I4:%.*]] = alloca i32, align 4 3844 // CHECK37-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8 3845 // CHECK37-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8 3846 // CHECK37-NEXT: store i32* [[N]], i32** [[N_ADDR]], align 8 3847 // CHECK37-NEXT: store i64 [[VLA]], i64* [[VLA_ADDR]], align 8 3848 // CHECK37-NEXT: store i32* [[A]], i32** [[A_ADDR]], align 8 3849 // CHECK37-NEXT: store i64 [[DOTCAPTURE_EXPR_]], i64* [[DOTCAPTURE_EXPR__ADDR]], align 8 3850 // CHECK37-NEXT: [[TMP0:%.*]] = load i32*, i32** [[N_ADDR]], align 8 3851 // CHECK37-NEXT: [[TMP1:%.*]] = load i64, i64* [[VLA_ADDR]], align 8 3852 // CHECK37-NEXT: [[TMP2:%.*]] = load i32*, i32** [[A_ADDR]], align 8 3853 // CHECK37-NEXT: [[CONV:%.*]] = bitcast i64* [[DOTCAPTURE_EXPR__ADDR]] to i8* 3854 // CHECK37-NEXT: [[TMP3:%.*]] = load i32, i32* [[TMP0]], align 4 3855 // CHECK37-NEXT: store i32 [[TMP3]], i32* [[DOTCAPTURE_EXPR_1]], align 4 3856 // CHECK37-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 3857 // CHECK37-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP4]], 0 3858 // CHECK37-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 3859 // CHECK37-NEXT: [[SUB3:%.*]] = sub nsw i32 [[DIV]], 1 3860 // CHECK37-NEXT: store i32 [[SUB3]], i32* [[DOTCAPTURE_EXPR_2]], align 4 3861 // CHECK37-NEXT: store i32 0, i32* [[I]], align 4 3862 // CHECK37-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 3863 // CHECK37-NEXT: [[CMP:%.*]] = icmp slt i32 0, [[TMP5]] 3864 // CHECK37-NEXT: br i1 [[CMP]], label [[OMP_PRECOND_THEN:%.*]], label [[OMP_PRECOND_END:%.*]] 3865 // CHECK37: omp.precond.then: 3866 // CHECK37-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4 3867 // CHECK37-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_2]], align 4 3868 // CHECK37-NEXT: store i32 [[TMP6]], i32* [[DOTOMP_UB]], align 4 3869 // CHECK37-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4 3870 // CHECK37-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4 3871 // CHECK37-NEXT: [[TMP7:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 3872 // CHECK37-NEXT: [[TMP8:%.*]] = load i32, i32* [[TMP7]], align 4 3873 // CHECK37-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1:[0-9]+]], i32 [[TMP8]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1) 3874 // CHECK37-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 3875 // CHECK37-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_2]], align 4 3876 // CHECK37-NEXT: [[CMP5:%.*]] = icmp sgt i32 [[TMP9]], [[TMP10]] 3877 // CHECK37-NEXT: br i1 [[CMP5]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] 3878 // CHECK37: cond.true: 3879 // CHECK37-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_2]], align 4 3880 // CHECK37-NEXT: br label [[COND_END:%.*]] 3881 // CHECK37: cond.false: 3882 // CHECK37-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 3883 // CHECK37-NEXT: br label [[COND_END]] 3884 // CHECK37: cond.end: 3885 // CHECK37-NEXT: [[COND:%.*]] = phi i32 [ [[TMP11]], [[COND_TRUE]] ], [ [[TMP12]], [[COND_FALSE]] ] 3886 // CHECK37-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4 3887 // CHECK37-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4 3888 // CHECK37-NEXT: store i32 [[TMP13]], i32* [[DOTOMP_IV]], align 4 3889 // CHECK37-NEXT: [[TMP14:%.*]] = load i8, i8* [[CONV]], align 1 3890 // CHECK37-NEXT: [[TOBOOL:%.*]] = trunc i8 [[TMP14]] to i1 3891 // CHECK37-NEXT: br i1 [[TOBOOL]], label [[OMP_IF_THEN:%.*]], label [[OMP_IF_ELSE:%.*]] 3892 // CHECK37: omp_if.then: 3893 // CHECK37-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] 3894 // CHECK37: omp.inner.for.cond: 3895 // CHECK37-NEXT: [[TMP15:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !5 3896 // CHECK37-NEXT: [[TMP16:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4, !llvm.access.group !5 3897 // CHECK37-NEXT: [[CMP6:%.*]] = icmp sle i32 [[TMP15]], [[TMP16]] 3898 // CHECK37-NEXT: br i1 [[CMP6]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] 3899 // CHECK37: omp.inner.for.body: 3900 // CHECK37-NEXT: [[TMP17:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !5 3901 // CHECK37-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP17]], 1 3902 // CHECK37-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] 3903 // CHECK37-NEXT: store i32 [[ADD]], i32* [[I4]], align 4, !llvm.access.group !5 3904 // CHECK37-NEXT: [[TMP18:%.*]] = load i32, i32* [[I4]], align 4, !llvm.access.group !5 3905 // CHECK37-NEXT: [[IDXPROM:%.*]] = sext i32 [[TMP18]] to i64 3906 // CHECK37-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i32, i32* [[TMP2]], i64 [[IDXPROM]] 3907 // CHECK37-NEXT: store i32 0, i32* [[ARRAYIDX]], align 4, !llvm.access.group !5 3908 // CHECK37-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] 3909 // CHECK37: omp.body.continue: 3910 // CHECK37-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] 3911 // CHECK37: omp.inner.for.inc: 3912 // CHECK37-NEXT: [[TMP19:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !5 3913 // CHECK37-NEXT: [[ADD7:%.*]] = add nsw i32 [[TMP19]], 1 3914 // CHECK37-NEXT: store i32 [[ADD7]], i32* [[DOTOMP_IV]], align 4, !llvm.access.group !5 3915 // CHECK37-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP6:![0-9]+]] 3916 // CHECK37: omp.inner.for.end: 3917 // CHECK37-NEXT: br label [[OMP_IF_END:%.*]] 3918 // CHECK37: omp_if.else: 3919 // CHECK37-NEXT: br label [[OMP_INNER_FOR_COND8:%.*]] 3920 // CHECK37: omp.inner.for.cond8: 3921 // CHECK37-NEXT: [[TMP20:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 3922 // CHECK37-NEXT: [[TMP21:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 3923 // CHECK37-NEXT: [[CMP9:%.*]] = icmp sle i32 [[TMP20]], [[TMP21]] 3924 // CHECK37-NEXT: br i1 [[CMP9]], label [[OMP_INNER_FOR_BODY10:%.*]], label [[OMP_INNER_FOR_END18:%.*]] 3925 // CHECK37: omp.inner.for.body10: 3926 // CHECK37-NEXT: [[TMP22:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 3927 // CHECK37-NEXT: [[MUL11:%.*]] = mul nsw i32 [[TMP22]], 1 3928 // CHECK37-NEXT: [[ADD12:%.*]] = add nsw i32 0, [[MUL11]] 3929 // CHECK37-NEXT: store i32 [[ADD12]], i32* [[I4]], align 4 3930 // CHECK37-NEXT: [[TMP23:%.*]] = load i32, i32* [[I4]], align 4 3931 // CHECK37-NEXT: [[IDXPROM13:%.*]] = sext i32 [[TMP23]] to i64 3932 // CHECK37-NEXT: [[ARRAYIDX14:%.*]] = getelementptr inbounds i32, i32* [[TMP2]], i64 [[IDXPROM13]] 3933 // CHECK37-NEXT: store i32 0, i32* [[ARRAYIDX14]], align 4 3934 // CHECK37-NEXT: br label [[OMP_BODY_CONTINUE15:%.*]] 3935 // CHECK37: omp.body.continue15: 3936 // CHECK37-NEXT: br label [[OMP_INNER_FOR_INC16:%.*]] 3937 // CHECK37: omp.inner.for.inc16: 3938 // CHECK37-NEXT: [[TMP24:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 3939 // CHECK37-NEXT: [[ADD17:%.*]] = add nsw i32 [[TMP24]], 1 3940 // CHECK37-NEXT: store i32 [[ADD17]], i32* [[DOTOMP_IV]], align 4 3941 // CHECK37-NEXT: br label [[OMP_INNER_FOR_COND8]], !llvm.loop [[LOOP9:![0-9]+]] 3942 // CHECK37: omp.inner.for.end18: 3943 // CHECK37-NEXT: br label [[OMP_IF_END]] 3944 // CHECK37: omp_if.end: 3945 // CHECK37-NEXT: br label [[OMP_LOOP_EXIT:%.*]] 3946 // CHECK37: omp.loop.exit: 3947 // CHECK37-NEXT: [[TMP25:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 3948 // CHECK37-NEXT: [[TMP26:%.*]] = load i32, i32* [[TMP25]], align 4 3949 // CHECK37-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP26]]) 3950 // CHECK37-NEXT: [[TMP27:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4 3951 // CHECK37-NEXT: [[TMP28:%.*]] = icmp ne i32 [[TMP27]], 0 3952 // CHECK37-NEXT: br i1 [[TMP28]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]] 3953 // CHECK37: .omp.final.then: 3954 // CHECK37-NEXT: [[TMP29:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 3955 // CHECK37-NEXT: [[SUB19:%.*]] = sub nsw i32 [[TMP29]], 0 3956 // CHECK37-NEXT: [[DIV20:%.*]] = sdiv i32 [[SUB19]], 1 3957 // CHECK37-NEXT: [[MUL21:%.*]] = mul nsw i32 [[DIV20]], 1 3958 // CHECK37-NEXT: [[ADD22:%.*]] = add nsw i32 0, [[MUL21]] 3959 // CHECK37-NEXT: store i32 [[ADD22]], i32* [[I4]], align 4 3960 // CHECK37-NEXT: br label [[DOTOMP_FINAL_DONE]] 3961 // CHECK37: .omp.final.done: 3962 // CHECK37-NEXT: br label [[OMP_PRECOND_END]] 3963 // CHECK37: omp.precond.end: 3964 // CHECK37-NEXT: ret void 3965 // 3966 // 3967 // CHECK37-LABEL: define {{[^@]+}}@_Z5tmainIiLi10EEiT_ 3968 // CHECK37-SAME: (i32 noundef signext [[ARGC:%.*]]) #[[ATTR5:[0-9]+]] comdat { 3969 // CHECK37-NEXT: entry: 3970 // CHECK37-NEXT: [[ARGC_ADDR:%.*]] = alloca i32, align 4 3971 // CHECK37-NEXT: [[A:%.*]] = alloca [10 x i32], align 4 3972 // CHECK37-NEXT: [[TE:%.*]] = alloca i32, align 4 3973 // CHECK37-NEXT: [[TH:%.*]] = alloca i32, align 4 3974 // CHECK37-NEXT: [[TE_CASTED:%.*]] = alloca i64, align 8 3975 // CHECK37-NEXT: [[TH_CASTED:%.*]] = alloca i64, align 8 3976 // CHECK37-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [3 x i8*], align 8 3977 // CHECK37-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [3 x i8*], align 8 3978 // CHECK37-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [3 x i8*], align 8 3979 // CHECK37-NEXT: [[TMP:%.*]] = alloca i32, align 4 3980 // CHECK37-NEXT: store i32 [[ARGC]], i32* [[ARGC_ADDR]], align 4 3981 // CHECK37-NEXT: store i32 0, i32* [[TE]], align 4 3982 // CHECK37-NEXT: store i32 128, i32* [[TH]], align 4 3983 // CHECK37-NEXT: [[TMP0:%.*]] = load i32, i32* [[TE]], align 4 3984 // CHECK37-NEXT: [[CONV:%.*]] = bitcast i64* [[TE_CASTED]] to i32* 3985 // CHECK37-NEXT: store i32 [[TMP0]], i32* [[CONV]], align 4 3986 // CHECK37-NEXT: [[TMP1:%.*]] = load i64, i64* [[TE_CASTED]], align 8 3987 // CHECK37-NEXT: [[TMP2:%.*]] = load i32, i32* [[TH]], align 4 3988 // CHECK37-NEXT: [[CONV1:%.*]] = bitcast i64* [[TH_CASTED]] to i32* 3989 // CHECK37-NEXT: store i32 [[TMP2]], i32* [[CONV1]], align 4 3990 // CHECK37-NEXT: [[TMP3:%.*]] = load i64, i64* [[TH_CASTED]], align 8 3991 // CHECK37-NEXT: [[TMP4:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 3992 // CHECK37-NEXT: [[TMP5:%.*]] = bitcast i8** [[TMP4]] to i64* 3993 // CHECK37-NEXT: store i64 [[TMP1]], i64* [[TMP5]], align 8 3994 // CHECK37-NEXT: [[TMP6:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 3995 // CHECK37-NEXT: [[TMP7:%.*]] = bitcast i8** [[TMP6]] to i64* 3996 // CHECK37-NEXT: store i64 [[TMP1]], i64* [[TMP7]], align 8 3997 // CHECK37-NEXT: [[TMP8:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0 3998 // CHECK37-NEXT: store i8* null, i8** [[TMP8]], align 8 3999 // CHECK37-NEXT: [[TMP9:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1 4000 // CHECK37-NEXT: [[TMP10:%.*]] = bitcast i8** [[TMP9]] to i64* 4001 // CHECK37-NEXT: store i64 [[TMP3]], i64* [[TMP10]], align 8 4002 // CHECK37-NEXT: [[TMP11:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1 4003 // CHECK37-NEXT: [[TMP12:%.*]] = bitcast i8** [[TMP11]] to i64* 4004 // CHECK37-NEXT: store i64 [[TMP3]], i64* [[TMP12]], align 8 4005 // CHECK37-NEXT: [[TMP13:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 1 4006 // CHECK37-NEXT: store i8* null, i8** [[TMP13]], align 8 4007 // CHECK37-NEXT: [[TMP14:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2 4008 // CHECK37-NEXT: [[TMP15:%.*]] = bitcast i8** [[TMP14]] to [10 x i32]** 4009 // CHECK37-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP15]], align 8 4010 // CHECK37-NEXT: [[TMP16:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2 4011 // CHECK37-NEXT: [[TMP17:%.*]] = bitcast i8** [[TMP16]] to [10 x i32]** 4012 // CHECK37-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP17]], align 8 4013 // CHECK37-NEXT: [[TMP18:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2 4014 // CHECK37-NEXT: store i8* null, i8** [[TMP18]], align 8 4015 // CHECK37-NEXT: [[TMP19:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 4016 // CHECK37-NEXT: [[TMP20:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 4017 // CHECK37-NEXT: [[TMP21:%.*]] = load i32, i32* [[TE]], align 4 4018 // CHECK37-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 10) 4019 // CHECK37-NEXT: [[TMP22:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l181.region_id, i32 3, i8** [[TMP19]], i8** [[TMP20]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes.2, i32 0, i32 0), i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes.3, i32 0, i32 0), i8** null, i8** null, i32 [[TMP21]], i32 1) 4020 // CHECK37-NEXT: [[TMP23:%.*]] = icmp ne i32 [[TMP22]], 0 4021 // CHECK37-NEXT: br i1 [[TMP23]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] 4022 // CHECK37: omp_offload.failed: 4023 // CHECK37-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l181(i64 [[TMP1]], i64 [[TMP3]], [10 x i32]* [[A]]) #[[ATTR3]] 4024 // CHECK37-NEXT: br label [[OMP_OFFLOAD_CONT]] 4025 // CHECK37: omp_offload.cont: 4026 // CHECK37-NEXT: ret i32 0 4027 // 4028 // 4029 // CHECK37-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l181 4030 // CHECK37-SAME: (i64 noundef [[TE:%.*]], i64 noundef [[TH:%.*]], [10 x i32]* noundef nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] { 4031 // CHECK37-NEXT: entry: 4032 // CHECK37-NEXT: [[TE_ADDR:%.*]] = alloca i64, align 8 4033 // CHECK37-NEXT: [[TH_ADDR:%.*]] = alloca i64, align 8 4034 // CHECK37-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 8 4035 // CHECK37-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB2]]) 4036 // CHECK37-NEXT: store i64 [[TE]], i64* [[TE_ADDR]], align 8 4037 // CHECK37-NEXT: store i64 [[TH]], i64* [[TH_ADDR]], align 8 4038 // CHECK37-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[A_ADDR]], align 8 4039 // CHECK37-NEXT: [[CONV:%.*]] = bitcast i64* [[TE_ADDR]] to i32* 4040 // CHECK37-NEXT: [[CONV1:%.*]] = bitcast i64* [[TH_ADDR]] to i32* 4041 // CHECK37-NEXT: [[TMP1:%.*]] = load [10 x i32]*, [10 x i32]** [[A_ADDR]], align 8 4042 // CHECK37-NEXT: [[TMP2:%.*]] = load i32, i32* [[CONV]], align 4 4043 // CHECK37-NEXT: [[TMP3:%.*]] = load i32, i32* [[CONV1]], align 4 4044 // CHECK37-NEXT: call void @__kmpc_push_num_teams(%struct.ident_t* @[[GLOB2]], i32 [[TMP0]], i32 [[TMP2]], i32 [[TMP3]]) 4045 // CHECK37-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, [10 x i32]*)* @.omp_outlined..1 to void (i32*, i32*, ...)*), [10 x i32]* [[TMP1]]) 4046 // CHECK37-NEXT: ret void 4047 // 4048 // 4049 // CHECK37-LABEL: define {{[^@]+}}@.omp_outlined..1 4050 // CHECK37-SAME: (i32* noalias noundef [[DOTGLOBAL_TID_:%.*]], i32* noalias noundef [[DOTBOUND_TID_:%.*]], [10 x i32]* noundef nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] { 4051 // CHECK37-NEXT: entry: 4052 // CHECK37-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 4053 // CHECK37-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 4054 // CHECK37-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 8 4055 // CHECK37-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 4056 // CHECK37-NEXT: [[TMP:%.*]] = alloca i32, align 4 4057 // CHECK37-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 4058 // CHECK37-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4 4059 // CHECK37-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 4060 // CHECK37-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 4061 // CHECK37-NEXT: [[I:%.*]] = alloca i32, align 4 4062 // CHECK37-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8 4063 // CHECK37-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8 4064 // CHECK37-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[A_ADDR]], align 8 4065 // CHECK37-NEXT: [[TMP0:%.*]] = load [10 x i32]*, [10 x i32]** [[A_ADDR]], align 8 4066 // CHECK37-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4 4067 // CHECK37-NEXT: store i32 9, i32* [[DOTOMP_UB]], align 4 4068 // CHECK37-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4 4069 // CHECK37-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4 4070 // CHECK37-NEXT: [[TMP1:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 4071 // CHECK37-NEXT: [[TMP2:%.*]] = load i32, i32* [[TMP1]], align 4 4072 // CHECK37-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1]], i32 [[TMP2]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1) 4073 // CHECK37-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 4074 // CHECK37-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP3]], 9 4075 // CHECK37-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] 4076 // CHECK37: cond.true: 4077 // CHECK37-NEXT: br label [[COND_END:%.*]] 4078 // CHECK37: cond.false: 4079 // CHECK37-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 4080 // CHECK37-NEXT: br label [[COND_END]] 4081 // CHECK37: cond.end: 4082 // CHECK37-NEXT: [[COND:%.*]] = phi i32 [ 9, [[COND_TRUE]] ], [ [[TMP4]], [[COND_FALSE]] ] 4083 // CHECK37-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4 4084 // CHECK37-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4 4085 // CHECK37-NEXT: store i32 [[TMP5]], i32* [[DOTOMP_IV]], align 4 4086 // CHECK37-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] 4087 // CHECK37: omp.inner.for.cond: 4088 // CHECK37-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !13 4089 // CHECK37-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4, !llvm.access.group !13 4090 // CHECK37-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP6]], [[TMP7]] 4091 // CHECK37-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] 4092 // CHECK37: omp.inner.for.body: 4093 // CHECK37-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !13 4094 // CHECK37-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP8]], 1 4095 // CHECK37-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] 4096 // CHECK37-NEXT: store i32 [[ADD]], i32* [[I]], align 4, !llvm.access.group !13 4097 // CHECK37-NEXT: [[TMP9:%.*]] = load i32, i32* [[I]], align 4, !llvm.access.group !13 4098 // CHECK37-NEXT: [[IDXPROM:%.*]] = sext i32 [[TMP9]] to i64 4099 // CHECK37-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [10 x i32], [10 x i32]* [[TMP0]], i64 0, i64 [[IDXPROM]] 4100 // CHECK37-NEXT: store i32 0, i32* [[ARRAYIDX]], align 4, !llvm.access.group !13 4101 // CHECK37-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] 4102 // CHECK37: omp.body.continue: 4103 // CHECK37-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] 4104 // CHECK37: omp.inner.for.inc: 4105 // CHECK37-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !13 4106 // CHECK37-NEXT: [[ADD2:%.*]] = add nsw i32 [[TMP10]], 1 4107 // CHECK37-NEXT: store i32 [[ADD2]], i32* [[DOTOMP_IV]], align 4, !llvm.access.group !13 4108 // CHECK37-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP14:![0-9]+]] 4109 // CHECK37: omp.inner.for.end: 4110 // CHECK37-NEXT: br label [[OMP_LOOP_EXIT:%.*]] 4111 // CHECK37: omp.loop.exit: 4112 // CHECK37-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP2]]) 4113 // CHECK37-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4 4114 // CHECK37-NEXT: [[TMP12:%.*]] = icmp ne i32 [[TMP11]], 0 4115 // CHECK37-NEXT: br i1 [[TMP12]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]] 4116 // CHECK37: .omp.final.then: 4117 // CHECK37-NEXT: store i32 10, i32* [[I]], align 4 4118 // CHECK37-NEXT: br label [[DOTOMP_FINAL_DONE]] 4119 // CHECK37: .omp.final.done: 4120 // CHECK37-NEXT: ret void 4121 // 4122 // 4123 // CHECK37-LABEL: define {{[^@]+}}@.omp_offloading.requires_reg 4124 // CHECK37-SAME: () #[[ATTR6:[0-9]+]] { 4125 // CHECK37-NEXT: entry: 4126 // CHECK37-NEXT: call void @__tgt_register_requires(i64 1) 4127 // CHECK37-NEXT: ret void 4128 // 4129 // 4130 // CHECK39-LABEL: define {{[^@]+}}@main 4131 // CHECK39-SAME: (i32 noundef [[ARGC:%.*]], i8** noundef [[ARGV:%.*]]) #[[ATTR0:[0-9]+]] { 4132 // CHECK39-NEXT: entry: 4133 // CHECK39-NEXT: [[RETVAL:%.*]] = alloca i32, align 4 4134 // CHECK39-NEXT: [[ARGC_ADDR:%.*]] = alloca i32, align 4 4135 // CHECK39-NEXT: [[ARGV_ADDR:%.*]] = alloca i8**, align 4 4136 // CHECK39-NEXT: [[N:%.*]] = alloca i32, align 4 4137 // CHECK39-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 4 4138 // CHECK39-NEXT: [[__VLA_EXPR0:%.*]] = alloca i32, align 4 4139 // CHECK39-NEXT: [[ARGC_CASTED:%.*]] = alloca i32, align 4 4140 // CHECK39-NEXT: [[N_CASTED:%.*]] = alloca i32, align 4 4141 // CHECK39-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [4 x i8*], align 4 4142 // CHECK39-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [4 x i8*], align 4 4143 // CHECK39-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [4 x i8*], align 4 4144 // CHECK39-NEXT: [[DOTOFFLOAD_SIZES:%.*]] = alloca [4 x i64], align 4 4145 // CHECK39-NEXT: [[TMP:%.*]] = alloca i32, align 4 4146 // CHECK39-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 4147 // CHECK39-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 4148 // CHECK39-NEXT: store i32 0, i32* [[RETVAL]], align 4 4149 // CHECK39-NEXT: store i32 [[ARGC]], i32* [[ARGC_ADDR]], align 4 4150 // CHECK39-NEXT: store i8** [[ARGV]], i8*** [[ARGV_ADDR]], align 4 4151 // CHECK39-NEXT: store i32 100, i32* [[N]], align 4 4152 // CHECK39-NEXT: [[TMP0:%.*]] = load i32, i32* [[N]], align 4 4153 // CHECK39-NEXT: [[TMP1:%.*]] = call i8* @llvm.stacksave() 4154 // CHECK39-NEXT: store i8* [[TMP1]], i8** [[SAVED_STACK]], align 4 4155 // CHECK39-NEXT: [[VLA:%.*]] = alloca i32, i32 [[TMP0]], align 4 4156 // CHECK39-NEXT: store i32 [[TMP0]], i32* [[__VLA_EXPR0]], align 4 4157 // CHECK39-NEXT: [[TMP2:%.*]] = load i32, i32* [[ARGC_ADDR]], align 4 4158 // CHECK39-NEXT: store i32 [[TMP2]], i32* [[ARGC_CASTED]], align 4 4159 // CHECK39-NEXT: [[TMP3:%.*]] = load i32, i32* [[ARGC_CASTED]], align 4 4160 // CHECK39-NEXT: [[TMP4:%.*]] = load i32, i32* [[N]], align 4 4161 // CHECK39-NEXT: store i32 [[TMP4]], i32* [[N_CASTED]], align 4 4162 // CHECK39-NEXT: [[TMP5:%.*]] = load i32, i32* [[N_CASTED]], align 4 4163 // CHECK39-NEXT: [[TMP6:%.*]] = mul nuw i32 [[TMP0]], 4 4164 // CHECK39-NEXT: [[TMP7:%.*]] = sext i32 [[TMP6]] to i64 4165 // CHECK39-NEXT: [[TMP8:%.*]] = bitcast [4 x i64]* [[DOTOFFLOAD_SIZES]] to i8* 4166 // CHECK39-NEXT: call void @llvm.memcpy.p0i8.p0i8.i32(i8* align 4 [[TMP8]], i8* align 4 bitcast ([4 x i64]* @.offload_sizes to i8*), i32 32, i1 false) 4167 // CHECK39-NEXT: [[TMP9:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 4168 // CHECK39-NEXT: [[TMP10:%.*]] = bitcast i8** [[TMP9]] to i32* 4169 // CHECK39-NEXT: store i32 [[TMP3]], i32* [[TMP10]], align 4 4170 // CHECK39-NEXT: [[TMP11:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 4171 // CHECK39-NEXT: [[TMP12:%.*]] = bitcast i8** [[TMP11]] to i32* 4172 // CHECK39-NEXT: store i32 [[TMP3]], i32* [[TMP12]], align 4 4173 // CHECK39-NEXT: [[TMP13:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0 4174 // CHECK39-NEXT: store i8* null, i8** [[TMP13]], align 4 4175 // CHECK39-NEXT: [[TMP14:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1 4176 // CHECK39-NEXT: [[TMP15:%.*]] = bitcast i8** [[TMP14]] to i32* 4177 // CHECK39-NEXT: store i32 [[TMP5]], i32* [[TMP15]], align 4 4178 // CHECK39-NEXT: [[TMP16:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1 4179 // CHECK39-NEXT: [[TMP17:%.*]] = bitcast i8** [[TMP16]] to i32* 4180 // CHECK39-NEXT: store i32 [[TMP5]], i32* [[TMP17]], align 4 4181 // CHECK39-NEXT: [[TMP18:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 1 4182 // CHECK39-NEXT: store i8* null, i8** [[TMP18]], align 4 4183 // CHECK39-NEXT: [[TMP19:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2 4184 // CHECK39-NEXT: [[TMP20:%.*]] = bitcast i8** [[TMP19]] to i32* 4185 // CHECK39-NEXT: store i32 [[TMP0]], i32* [[TMP20]], align 4 4186 // CHECK39-NEXT: [[TMP21:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2 4187 // CHECK39-NEXT: [[TMP22:%.*]] = bitcast i8** [[TMP21]] to i32* 4188 // CHECK39-NEXT: store i32 [[TMP0]], i32* [[TMP22]], align 4 4189 // CHECK39-NEXT: [[TMP23:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2 4190 // CHECK39-NEXT: store i8* null, i8** [[TMP23]], align 4 4191 // CHECK39-NEXT: [[TMP24:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 3 4192 // CHECK39-NEXT: [[TMP25:%.*]] = bitcast i8** [[TMP24]] to i32** 4193 // CHECK39-NEXT: store i32* [[VLA]], i32** [[TMP25]], align 4 4194 // CHECK39-NEXT: [[TMP26:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 3 4195 // CHECK39-NEXT: [[TMP27:%.*]] = bitcast i8** [[TMP26]] to i32** 4196 // CHECK39-NEXT: store i32* [[VLA]], i32** [[TMP27]], align 4 4197 // CHECK39-NEXT: [[TMP28:%.*]] = getelementptr inbounds [4 x i64], [4 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 3 4198 // CHECK39-NEXT: store i64 [[TMP7]], i64* [[TMP28]], align 4 4199 // CHECK39-NEXT: [[TMP29:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 3 4200 // CHECK39-NEXT: store i8* null, i8** [[TMP29]], align 4 4201 // CHECK39-NEXT: [[TMP30:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 4202 // CHECK39-NEXT: [[TMP31:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 4203 // CHECK39-NEXT: [[TMP32:%.*]] = getelementptr inbounds [4 x i64], [4 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0 4204 // CHECK39-NEXT: [[TMP33:%.*]] = load i32, i32* [[N]], align 4 4205 // CHECK39-NEXT: store i32 [[TMP33]], i32* [[DOTCAPTURE_EXPR_]], align 4 4206 // CHECK39-NEXT: [[TMP34:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 4207 // CHECK39-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP34]], 0 4208 // CHECK39-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 4209 // CHECK39-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1 4210 // CHECK39-NEXT: store i32 [[SUB2]], i32* [[DOTCAPTURE_EXPR_1]], align 4 4211 // CHECK39-NEXT: [[TMP35:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 4212 // CHECK39-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP35]], 1 4213 // CHECK39-NEXT: [[TMP36:%.*]] = zext i32 [[ADD]] to i64 4214 // CHECK39-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2:[0-9]+]], i64 -1, i64 [[TMP36]]) 4215 // CHECK39-NEXT: [[TMP37:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l192.region_id, i32 4, i8** [[TMP30]], i8** [[TMP31]], i64* [[TMP32]], i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 0, i32 1) 4216 // CHECK39-NEXT: [[TMP38:%.*]] = icmp ne i32 [[TMP37]], 0 4217 // CHECK39-NEXT: br i1 [[TMP38]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] 4218 // CHECK39: omp_offload.failed: 4219 // CHECK39-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l192(i32 [[TMP3]], i32 [[TMP5]], i32 [[TMP0]], i32* [[VLA]]) #[[ATTR3:[0-9]+]] 4220 // CHECK39-NEXT: br label [[OMP_OFFLOAD_CONT]] 4221 // CHECK39: omp_offload.cont: 4222 // CHECK39-NEXT: [[TMP39:%.*]] = load i32, i32* [[ARGC_ADDR]], align 4 4223 // CHECK39-NEXT: [[CALL:%.*]] = call noundef i32 @_Z5tmainIiLi10EEiT_(i32 noundef [[TMP39]]) 4224 // CHECK39-NEXT: store i32 [[CALL]], i32* [[RETVAL]], align 4 4225 // CHECK39-NEXT: [[TMP40:%.*]] = load i8*, i8** [[SAVED_STACK]], align 4 4226 // CHECK39-NEXT: call void @llvm.stackrestore(i8* [[TMP40]]) 4227 // CHECK39-NEXT: [[TMP41:%.*]] = load i32, i32* [[RETVAL]], align 4 4228 // CHECK39-NEXT: ret i32 [[TMP41]] 4229 // 4230 // 4231 // CHECK39-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l192 4232 // CHECK39-SAME: (i32 noundef [[ARGC:%.*]], i32 noundef [[N:%.*]], i32 noundef [[VLA:%.*]], i32* noundef nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2:[0-9]+]] { 4233 // CHECK39-NEXT: entry: 4234 // CHECK39-NEXT: [[ARGC_ADDR:%.*]] = alloca i32, align 4 4235 // CHECK39-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4 4236 // CHECK39-NEXT: [[VLA_ADDR:%.*]] = alloca i32, align 4 4237 // CHECK39-NEXT: [[A_ADDR:%.*]] = alloca i32*, align 4 4238 // CHECK39-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i8, align 1 4239 // CHECK39-NEXT: [[DOTCAPTURE_EXPR__CASTED:%.*]] = alloca i32, align 4 4240 // CHECK39-NEXT: store i32 [[ARGC]], i32* [[ARGC_ADDR]], align 4 4241 // CHECK39-NEXT: store i32 [[N]], i32* [[N_ADDR]], align 4 4242 // CHECK39-NEXT: store i32 [[VLA]], i32* [[VLA_ADDR]], align 4 4243 // CHECK39-NEXT: store i32* [[A]], i32** [[A_ADDR]], align 4 4244 // CHECK39-NEXT: [[TMP0:%.*]] = load i32, i32* [[VLA_ADDR]], align 4 4245 // CHECK39-NEXT: [[TMP1:%.*]] = load i32*, i32** [[A_ADDR]], align 4 4246 // CHECK39-NEXT: [[TMP2:%.*]] = load i32, i32* [[ARGC_ADDR]], align 4 4247 // CHECK39-NEXT: [[TOBOOL:%.*]] = icmp ne i32 [[TMP2]], 0 4248 // CHECK39-NEXT: [[FROMBOOL:%.*]] = zext i1 [[TOBOOL]] to i8 4249 // CHECK39-NEXT: store i8 [[FROMBOOL]], i8* [[DOTCAPTURE_EXPR_]], align 1 4250 // CHECK39-NEXT: [[TMP3:%.*]] = load i8, i8* [[DOTCAPTURE_EXPR_]], align 1 4251 // CHECK39-NEXT: [[TOBOOL1:%.*]] = trunc i8 [[TMP3]] to i1 4252 // CHECK39-NEXT: [[CONV:%.*]] = bitcast i32* [[DOTCAPTURE_EXPR__CASTED]] to i8* 4253 // CHECK39-NEXT: [[FROMBOOL2:%.*]] = zext i1 [[TOBOOL1]] to i8 4254 // CHECK39-NEXT: store i8 [[FROMBOOL2]], i8* [[CONV]], align 1 4255 // CHECK39-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR__CASTED]], align 4 4256 // CHECK39-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2]], i32 4, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32*, i32, i32*, i32)* @.omp_outlined. to void (i32*, i32*, ...)*), i32* [[N_ADDR]], i32 [[TMP0]], i32* [[TMP1]], i32 [[TMP4]]) 4257 // CHECK39-NEXT: ret void 4258 // 4259 // 4260 // CHECK39-LABEL: define {{[^@]+}}@.omp_outlined. 4261 // CHECK39-SAME: (i32* noalias noundef [[DOTGLOBAL_TID_:%.*]], i32* noalias noundef [[DOTBOUND_TID_:%.*]], i32* noundef nonnull align 4 dereferenceable(4) [[N:%.*]], i32 noundef [[VLA:%.*]], i32* noundef nonnull align 4 dereferenceable(4) [[A:%.*]], i32 noundef [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR2]] { 4262 // CHECK39-NEXT: entry: 4263 // CHECK39-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4 4264 // CHECK39-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4 4265 // CHECK39-NEXT: [[N_ADDR:%.*]] = alloca i32*, align 4 4266 // CHECK39-NEXT: [[VLA_ADDR:%.*]] = alloca i32, align 4 4267 // CHECK39-NEXT: [[A_ADDR:%.*]] = alloca i32*, align 4 4268 // CHECK39-NEXT: [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i32, align 4 4269 // CHECK39-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 4270 // CHECK39-NEXT: [[TMP:%.*]] = alloca i32, align 4 4271 // CHECK39-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 4272 // CHECK39-NEXT: [[DOTCAPTURE_EXPR_2:%.*]] = alloca i32, align 4 4273 // CHECK39-NEXT: [[I:%.*]] = alloca i32, align 4 4274 // CHECK39-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 4275 // CHECK39-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4 4276 // CHECK39-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 4277 // CHECK39-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 4278 // CHECK39-NEXT: [[I4:%.*]] = alloca i32, align 4 4279 // CHECK39-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 4 4280 // CHECK39-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 4 4281 // CHECK39-NEXT: store i32* [[N]], i32** [[N_ADDR]], align 4 4282 // CHECK39-NEXT: store i32 [[VLA]], i32* [[VLA_ADDR]], align 4 4283 // CHECK39-NEXT: store i32* [[A]], i32** [[A_ADDR]], align 4 4284 // CHECK39-NEXT: store i32 [[DOTCAPTURE_EXPR_]], i32* [[DOTCAPTURE_EXPR__ADDR]], align 4 4285 // CHECK39-NEXT: [[TMP0:%.*]] = load i32*, i32** [[N_ADDR]], align 4 4286 // CHECK39-NEXT: [[TMP1:%.*]] = load i32, i32* [[VLA_ADDR]], align 4 4287 // CHECK39-NEXT: [[TMP2:%.*]] = load i32*, i32** [[A_ADDR]], align 4 4288 // CHECK39-NEXT: [[CONV:%.*]] = bitcast i32* [[DOTCAPTURE_EXPR__ADDR]] to i8* 4289 // CHECK39-NEXT: [[TMP3:%.*]] = load i32, i32* [[TMP0]], align 4 4290 // CHECK39-NEXT: store i32 [[TMP3]], i32* [[DOTCAPTURE_EXPR_1]], align 4 4291 // CHECK39-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 4292 // CHECK39-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP4]], 0 4293 // CHECK39-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 4294 // CHECK39-NEXT: [[SUB3:%.*]] = sub nsw i32 [[DIV]], 1 4295 // CHECK39-NEXT: store i32 [[SUB3]], i32* [[DOTCAPTURE_EXPR_2]], align 4 4296 // CHECK39-NEXT: store i32 0, i32* [[I]], align 4 4297 // CHECK39-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 4298 // CHECK39-NEXT: [[CMP:%.*]] = icmp slt i32 0, [[TMP5]] 4299 // CHECK39-NEXT: br i1 [[CMP]], label [[OMP_PRECOND_THEN:%.*]], label [[OMP_PRECOND_END:%.*]] 4300 // CHECK39: omp.precond.then: 4301 // CHECK39-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4 4302 // CHECK39-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_2]], align 4 4303 // CHECK39-NEXT: store i32 [[TMP6]], i32* [[DOTOMP_UB]], align 4 4304 // CHECK39-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4 4305 // CHECK39-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4 4306 // CHECK39-NEXT: [[TMP7:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 4 4307 // CHECK39-NEXT: [[TMP8:%.*]] = load i32, i32* [[TMP7]], align 4 4308 // CHECK39-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1:[0-9]+]], i32 [[TMP8]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1) 4309 // CHECK39-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 4310 // CHECK39-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_2]], align 4 4311 // CHECK39-NEXT: [[CMP5:%.*]] = icmp sgt i32 [[TMP9]], [[TMP10]] 4312 // CHECK39-NEXT: br i1 [[CMP5]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] 4313 // CHECK39: cond.true: 4314 // CHECK39-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_2]], align 4 4315 // CHECK39-NEXT: br label [[COND_END:%.*]] 4316 // CHECK39: cond.false: 4317 // CHECK39-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 4318 // CHECK39-NEXT: br label [[COND_END]] 4319 // CHECK39: cond.end: 4320 // CHECK39-NEXT: [[COND:%.*]] = phi i32 [ [[TMP11]], [[COND_TRUE]] ], [ [[TMP12]], [[COND_FALSE]] ] 4321 // CHECK39-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4 4322 // CHECK39-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4 4323 // CHECK39-NEXT: store i32 [[TMP13]], i32* [[DOTOMP_IV]], align 4 4324 // CHECK39-NEXT: [[TMP14:%.*]] = load i8, i8* [[CONV]], align 1 4325 // CHECK39-NEXT: [[TOBOOL:%.*]] = trunc i8 [[TMP14]] to i1 4326 // CHECK39-NEXT: br i1 [[TOBOOL]], label [[OMP_IF_THEN:%.*]], label [[OMP_IF_ELSE:%.*]] 4327 // CHECK39: omp_if.then: 4328 // CHECK39-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] 4329 // CHECK39: omp.inner.for.cond: 4330 // CHECK39-NEXT: [[TMP15:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !6 4331 // CHECK39-NEXT: [[TMP16:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4, !llvm.access.group !6 4332 // CHECK39-NEXT: [[CMP6:%.*]] = icmp sle i32 [[TMP15]], [[TMP16]] 4333 // CHECK39-NEXT: br i1 [[CMP6]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] 4334 // CHECK39: omp.inner.for.body: 4335 // CHECK39-NEXT: [[TMP17:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !6 4336 // CHECK39-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP17]], 1 4337 // CHECK39-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] 4338 // CHECK39-NEXT: store i32 [[ADD]], i32* [[I4]], align 4, !llvm.access.group !6 4339 // CHECK39-NEXT: [[TMP18:%.*]] = load i32, i32* [[I4]], align 4, !llvm.access.group !6 4340 // CHECK39-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i32, i32* [[TMP2]], i32 [[TMP18]] 4341 // CHECK39-NEXT: store i32 0, i32* [[ARRAYIDX]], align 4, !llvm.access.group !6 4342 // CHECK39-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] 4343 // CHECK39: omp.body.continue: 4344 // CHECK39-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] 4345 // CHECK39: omp.inner.for.inc: 4346 // CHECK39-NEXT: [[TMP19:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !6 4347 // CHECK39-NEXT: [[ADD7:%.*]] = add nsw i32 [[TMP19]], 1 4348 // CHECK39-NEXT: store i32 [[ADD7]], i32* [[DOTOMP_IV]], align 4, !llvm.access.group !6 4349 // CHECK39-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP7:![0-9]+]] 4350 // CHECK39: omp.inner.for.end: 4351 // CHECK39-NEXT: br label [[OMP_IF_END:%.*]] 4352 // CHECK39: omp_if.else: 4353 // CHECK39-NEXT: br label [[OMP_INNER_FOR_COND8:%.*]] 4354 // CHECK39: omp.inner.for.cond8: 4355 // CHECK39-NEXT: [[TMP20:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 4356 // CHECK39-NEXT: [[TMP21:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 4357 // CHECK39-NEXT: [[CMP9:%.*]] = icmp sle i32 [[TMP20]], [[TMP21]] 4358 // CHECK39-NEXT: br i1 [[CMP9]], label [[OMP_INNER_FOR_BODY10:%.*]], label [[OMP_INNER_FOR_END17:%.*]] 4359 // CHECK39: omp.inner.for.body10: 4360 // CHECK39-NEXT: [[TMP22:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 4361 // CHECK39-NEXT: [[MUL11:%.*]] = mul nsw i32 [[TMP22]], 1 4362 // CHECK39-NEXT: [[ADD12:%.*]] = add nsw i32 0, [[MUL11]] 4363 // CHECK39-NEXT: store i32 [[ADD12]], i32* [[I4]], align 4 4364 // CHECK39-NEXT: [[TMP23:%.*]] = load i32, i32* [[I4]], align 4 4365 // CHECK39-NEXT: [[ARRAYIDX13:%.*]] = getelementptr inbounds i32, i32* [[TMP2]], i32 [[TMP23]] 4366 // CHECK39-NEXT: store i32 0, i32* [[ARRAYIDX13]], align 4 4367 // CHECK39-NEXT: br label [[OMP_BODY_CONTINUE14:%.*]] 4368 // CHECK39: omp.body.continue14: 4369 // CHECK39-NEXT: br label [[OMP_INNER_FOR_INC15:%.*]] 4370 // CHECK39: omp.inner.for.inc15: 4371 // CHECK39-NEXT: [[TMP24:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 4372 // CHECK39-NEXT: [[ADD16:%.*]] = add nsw i32 [[TMP24]], 1 4373 // CHECK39-NEXT: store i32 [[ADD16]], i32* [[DOTOMP_IV]], align 4 4374 // CHECK39-NEXT: br label [[OMP_INNER_FOR_COND8]], !llvm.loop [[LOOP10:![0-9]+]] 4375 // CHECK39: omp.inner.for.end17: 4376 // CHECK39-NEXT: br label [[OMP_IF_END]] 4377 // CHECK39: omp_if.end: 4378 // CHECK39-NEXT: br label [[OMP_LOOP_EXIT:%.*]] 4379 // CHECK39: omp.loop.exit: 4380 // CHECK39-NEXT: [[TMP25:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 4 4381 // CHECK39-NEXT: [[TMP26:%.*]] = load i32, i32* [[TMP25]], align 4 4382 // CHECK39-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP26]]) 4383 // CHECK39-NEXT: [[TMP27:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4 4384 // CHECK39-NEXT: [[TMP28:%.*]] = icmp ne i32 [[TMP27]], 0 4385 // CHECK39-NEXT: br i1 [[TMP28]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]] 4386 // CHECK39: .omp.final.then: 4387 // CHECK39-NEXT: [[TMP29:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 4388 // CHECK39-NEXT: [[SUB18:%.*]] = sub nsw i32 [[TMP29]], 0 4389 // CHECK39-NEXT: [[DIV19:%.*]] = sdiv i32 [[SUB18]], 1 4390 // CHECK39-NEXT: [[MUL20:%.*]] = mul nsw i32 [[DIV19]], 1 4391 // CHECK39-NEXT: [[ADD21:%.*]] = add nsw i32 0, [[MUL20]] 4392 // CHECK39-NEXT: store i32 [[ADD21]], i32* [[I4]], align 4 4393 // CHECK39-NEXT: br label [[DOTOMP_FINAL_DONE]] 4394 // CHECK39: .omp.final.done: 4395 // CHECK39-NEXT: br label [[OMP_PRECOND_END]] 4396 // CHECK39: omp.precond.end: 4397 // CHECK39-NEXT: ret void 4398 // 4399 // 4400 // CHECK39-LABEL: define {{[^@]+}}@_Z5tmainIiLi10EEiT_ 4401 // CHECK39-SAME: (i32 noundef [[ARGC:%.*]]) #[[ATTR5:[0-9]+]] comdat { 4402 // CHECK39-NEXT: entry: 4403 // CHECK39-NEXT: [[ARGC_ADDR:%.*]] = alloca i32, align 4 4404 // CHECK39-NEXT: [[A:%.*]] = alloca [10 x i32], align 4 4405 // CHECK39-NEXT: [[TE:%.*]] = alloca i32, align 4 4406 // CHECK39-NEXT: [[TH:%.*]] = alloca i32, align 4 4407 // CHECK39-NEXT: [[TE_CASTED:%.*]] = alloca i32, align 4 4408 // CHECK39-NEXT: [[TH_CASTED:%.*]] = alloca i32, align 4 4409 // CHECK39-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [3 x i8*], align 4 4410 // CHECK39-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [3 x i8*], align 4 4411 // CHECK39-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [3 x i8*], align 4 4412 // CHECK39-NEXT: [[TMP:%.*]] = alloca i32, align 4 4413 // CHECK39-NEXT: store i32 [[ARGC]], i32* [[ARGC_ADDR]], align 4 4414 // CHECK39-NEXT: store i32 0, i32* [[TE]], align 4 4415 // CHECK39-NEXT: store i32 128, i32* [[TH]], align 4 4416 // CHECK39-NEXT: [[TMP0:%.*]] = load i32, i32* [[TE]], align 4 4417 // CHECK39-NEXT: store i32 [[TMP0]], i32* [[TE_CASTED]], align 4 4418 // CHECK39-NEXT: [[TMP1:%.*]] = load i32, i32* [[TE_CASTED]], align 4 4419 // CHECK39-NEXT: [[TMP2:%.*]] = load i32, i32* [[TH]], align 4 4420 // CHECK39-NEXT: store i32 [[TMP2]], i32* [[TH_CASTED]], align 4 4421 // CHECK39-NEXT: [[TMP3:%.*]] = load i32, i32* [[TH_CASTED]], align 4 4422 // CHECK39-NEXT: [[TMP4:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 4423 // CHECK39-NEXT: [[TMP5:%.*]] = bitcast i8** [[TMP4]] to i32* 4424 // CHECK39-NEXT: store i32 [[TMP1]], i32* [[TMP5]], align 4 4425 // CHECK39-NEXT: [[TMP6:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 4426 // CHECK39-NEXT: [[TMP7:%.*]] = bitcast i8** [[TMP6]] to i32* 4427 // CHECK39-NEXT: store i32 [[TMP1]], i32* [[TMP7]], align 4 4428 // CHECK39-NEXT: [[TMP8:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0 4429 // CHECK39-NEXT: store i8* null, i8** [[TMP8]], align 4 4430 // CHECK39-NEXT: [[TMP9:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1 4431 // CHECK39-NEXT: [[TMP10:%.*]] = bitcast i8** [[TMP9]] to i32* 4432 // CHECK39-NEXT: store i32 [[TMP3]], i32* [[TMP10]], align 4 4433 // CHECK39-NEXT: [[TMP11:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1 4434 // CHECK39-NEXT: [[TMP12:%.*]] = bitcast i8** [[TMP11]] to i32* 4435 // CHECK39-NEXT: store i32 [[TMP3]], i32* [[TMP12]], align 4 4436 // CHECK39-NEXT: [[TMP13:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 1 4437 // CHECK39-NEXT: store i8* null, i8** [[TMP13]], align 4 4438 // CHECK39-NEXT: [[TMP14:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2 4439 // CHECK39-NEXT: [[TMP15:%.*]] = bitcast i8** [[TMP14]] to [10 x i32]** 4440 // CHECK39-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP15]], align 4 4441 // CHECK39-NEXT: [[TMP16:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2 4442 // CHECK39-NEXT: [[TMP17:%.*]] = bitcast i8** [[TMP16]] to [10 x i32]** 4443 // CHECK39-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP17]], align 4 4444 // CHECK39-NEXT: [[TMP18:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2 4445 // CHECK39-NEXT: store i8* null, i8** [[TMP18]], align 4 4446 // CHECK39-NEXT: [[TMP19:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 4447 // CHECK39-NEXT: [[TMP20:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 4448 // CHECK39-NEXT: [[TMP21:%.*]] = load i32, i32* [[TE]], align 4 4449 // CHECK39-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 10) 4450 // CHECK39-NEXT: [[TMP22:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l181.region_id, i32 3, i8** [[TMP19]], i8** [[TMP20]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes.2, i32 0, i32 0), i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes.3, i32 0, i32 0), i8** null, i8** null, i32 [[TMP21]], i32 1) 4451 // CHECK39-NEXT: [[TMP23:%.*]] = icmp ne i32 [[TMP22]], 0 4452 // CHECK39-NEXT: br i1 [[TMP23]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] 4453 // CHECK39: omp_offload.failed: 4454 // CHECK39-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l181(i32 [[TMP1]], i32 [[TMP3]], [10 x i32]* [[A]]) #[[ATTR3]] 4455 // CHECK39-NEXT: br label [[OMP_OFFLOAD_CONT]] 4456 // CHECK39: omp_offload.cont: 4457 // CHECK39-NEXT: ret i32 0 4458 // 4459 // 4460 // CHECK39-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l181 4461 // CHECK39-SAME: (i32 noundef [[TE:%.*]], i32 noundef [[TH:%.*]], [10 x i32]* noundef nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] { 4462 // CHECK39-NEXT: entry: 4463 // CHECK39-NEXT: [[TE_ADDR:%.*]] = alloca i32, align 4 4464 // CHECK39-NEXT: [[TH_ADDR:%.*]] = alloca i32, align 4 4465 // CHECK39-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 4 4466 // CHECK39-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB2]]) 4467 // CHECK39-NEXT: store i32 [[TE]], i32* [[TE_ADDR]], align 4 4468 // CHECK39-NEXT: store i32 [[TH]], i32* [[TH_ADDR]], align 4 4469 // CHECK39-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[A_ADDR]], align 4 4470 // CHECK39-NEXT: [[TMP1:%.*]] = load [10 x i32]*, [10 x i32]** [[A_ADDR]], align 4 4471 // CHECK39-NEXT: [[TMP2:%.*]] = load i32, i32* [[TE_ADDR]], align 4 4472 // CHECK39-NEXT: [[TMP3:%.*]] = load i32, i32* [[TH_ADDR]], align 4 4473 // CHECK39-NEXT: call void @__kmpc_push_num_teams(%struct.ident_t* @[[GLOB2]], i32 [[TMP0]], i32 [[TMP2]], i32 [[TMP3]]) 4474 // CHECK39-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, [10 x i32]*)* @.omp_outlined..1 to void (i32*, i32*, ...)*), [10 x i32]* [[TMP1]]) 4475 // CHECK39-NEXT: ret void 4476 // 4477 // 4478 // CHECK39-LABEL: define {{[^@]+}}@.omp_outlined..1 4479 // CHECK39-SAME: (i32* noalias noundef [[DOTGLOBAL_TID_:%.*]], i32* noalias noundef [[DOTBOUND_TID_:%.*]], [10 x i32]* noundef nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] { 4480 // CHECK39-NEXT: entry: 4481 // CHECK39-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4 4482 // CHECK39-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4 4483 // CHECK39-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 4 4484 // CHECK39-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 4485 // CHECK39-NEXT: [[TMP:%.*]] = alloca i32, align 4 4486 // CHECK39-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 4487 // CHECK39-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4 4488 // CHECK39-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 4489 // CHECK39-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 4490 // CHECK39-NEXT: [[I:%.*]] = alloca i32, align 4 4491 // CHECK39-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 4 4492 // CHECK39-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 4 4493 // CHECK39-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[A_ADDR]], align 4 4494 // CHECK39-NEXT: [[TMP0:%.*]] = load [10 x i32]*, [10 x i32]** [[A_ADDR]], align 4 4495 // CHECK39-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4 4496 // CHECK39-NEXT: store i32 9, i32* [[DOTOMP_UB]], align 4 4497 // CHECK39-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4 4498 // CHECK39-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4 4499 // CHECK39-NEXT: [[TMP1:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 4 4500 // CHECK39-NEXT: [[TMP2:%.*]] = load i32, i32* [[TMP1]], align 4 4501 // CHECK39-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1]], i32 [[TMP2]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1) 4502 // CHECK39-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 4503 // CHECK39-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP3]], 9 4504 // CHECK39-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] 4505 // CHECK39: cond.true: 4506 // CHECK39-NEXT: br label [[COND_END:%.*]] 4507 // CHECK39: cond.false: 4508 // CHECK39-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 4509 // CHECK39-NEXT: br label [[COND_END]] 4510 // CHECK39: cond.end: 4511 // CHECK39-NEXT: [[COND:%.*]] = phi i32 [ 9, [[COND_TRUE]] ], [ [[TMP4]], [[COND_FALSE]] ] 4512 // CHECK39-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4 4513 // CHECK39-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4 4514 // CHECK39-NEXT: store i32 [[TMP5]], i32* [[DOTOMP_IV]], align 4 4515 // CHECK39-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] 4516 // CHECK39: omp.inner.for.cond: 4517 // CHECK39-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !14 4518 // CHECK39-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4, !llvm.access.group !14 4519 // CHECK39-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP6]], [[TMP7]] 4520 // CHECK39-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] 4521 // CHECK39: omp.inner.for.body: 4522 // CHECK39-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !14 4523 // CHECK39-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP8]], 1 4524 // CHECK39-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] 4525 // CHECK39-NEXT: store i32 [[ADD]], i32* [[I]], align 4, !llvm.access.group !14 4526 // CHECK39-NEXT: [[TMP9:%.*]] = load i32, i32* [[I]], align 4, !llvm.access.group !14 4527 // CHECK39-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [10 x i32], [10 x i32]* [[TMP0]], i32 0, i32 [[TMP9]] 4528 // CHECK39-NEXT: store i32 0, i32* [[ARRAYIDX]], align 4, !llvm.access.group !14 4529 // CHECK39-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] 4530 // CHECK39: omp.body.continue: 4531 // CHECK39-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] 4532 // CHECK39: omp.inner.for.inc: 4533 // CHECK39-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !14 4534 // CHECK39-NEXT: [[ADD2:%.*]] = add nsw i32 [[TMP10]], 1 4535 // CHECK39-NEXT: store i32 [[ADD2]], i32* [[DOTOMP_IV]], align 4, !llvm.access.group !14 4536 // CHECK39-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP15:![0-9]+]] 4537 // CHECK39: omp.inner.for.end: 4538 // CHECK39-NEXT: br label [[OMP_LOOP_EXIT:%.*]] 4539 // CHECK39: omp.loop.exit: 4540 // CHECK39-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP2]]) 4541 // CHECK39-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4 4542 // CHECK39-NEXT: [[TMP12:%.*]] = icmp ne i32 [[TMP11]], 0 4543 // CHECK39-NEXT: br i1 [[TMP12]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]] 4544 // CHECK39: .omp.final.then: 4545 // CHECK39-NEXT: store i32 10, i32* [[I]], align 4 4546 // CHECK39-NEXT: br label [[DOTOMP_FINAL_DONE]] 4547 // CHECK39: .omp.final.done: 4548 // CHECK39-NEXT: ret void 4549 // 4550 // 4551 // CHECK39-LABEL: define {{[^@]+}}@.omp_offloading.requires_reg 4552 // CHECK39-SAME: () #[[ATTR6:[0-9]+]] { 4553 // CHECK39-NEXT: entry: 4554 // CHECK39-NEXT: call void @__tgt_register_requires(i64 1) 4555 // CHECK39-NEXT: ret void 4556 // 4557 // 4558 // CHECK41-LABEL: define {{[^@]+}}@main 4559 // CHECK41-SAME: (i32 noundef signext [[ARGC:%.*]], i8** noundef [[ARGV:%.*]]) #[[ATTR0:[0-9]+]] { 4560 // CHECK41-NEXT: entry: 4561 // CHECK41-NEXT: [[RETVAL:%.*]] = alloca i32, align 4 4562 // CHECK41-NEXT: [[ARGC_ADDR:%.*]] = alloca i32, align 4 4563 // CHECK41-NEXT: [[ARGV_ADDR:%.*]] = alloca i8**, align 8 4564 // CHECK41-NEXT: [[N:%.*]] = alloca i32, align 4 4565 // CHECK41-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 8 4566 // CHECK41-NEXT: [[__VLA_EXPR0:%.*]] = alloca i64, align 8 4567 // CHECK41-NEXT: [[TMP:%.*]] = alloca i32, align 4 4568 // CHECK41-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 4569 // CHECK41-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 4570 // CHECK41-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 4571 // CHECK41-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4 4572 // CHECK41-NEXT: [[I:%.*]] = alloca i32, align 4 4573 // CHECK41-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 4574 // CHECK41-NEXT: [[I3:%.*]] = alloca i32, align 4 4575 // CHECK41-NEXT: store i32 0, i32* [[RETVAL]], align 4 4576 // CHECK41-NEXT: store i32 [[ARGC]], i32* [[ARGC_ADDR]], align 4 4577 // CHECK41-NEXT: store i8** [[ARGV]], i8*** [[ARGV_ADDR]], align 8 4578 // CHECK41-NEXT: store i32 100, i32* [[N]], align 4 4579 // CHECK41-NEXT: [[TMP0:%.*]] = load i32, i32* [[N]], align 4 4580 // CHECK41-NEXT: [[TMP1:%.*]] = zext i32 [[TMP0]] to i64 4581 // CHECK41-NEXT: [[TMP2:%.*]] = call i8* @llvm.stacksave() 4582 // CHECK41-NEXT: store i8* [[TMP2]], i8** [[SAVED_STACK]], align 8 4583 // CHECK41-NEXT: [[VLA:%.*]] = alloca i32, i64 [[TMP1]], align 4 4584 // CHECK41-NEXT: store i64 [[TMP1]], i64* [[__VLA_EXPR0]], align 8 4585 // CHECK41-NEXT: [[TMP3:%.*]] = load i32, i32* [[N]], align 4 4586 // CHECK41-NEXT: store i32 [[TMP3]], i32* [[DOTCAPTURE_EXPR_]], align 4 4587 // CHECK41-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 4588 // CHECK41-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP4]], 0 4589 // CHECK41-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 4590 // CHECK41-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1 4591 // CHECK41-NEXT: store i32 [[SUB2]], i32* [[DOTCAPTURE_EXPR_1]], align 4 4592 // CHECK41-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4 4593 // CHECK41-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 4594 // CHECK41-NEXT: store i32 [[TMP5]], i32* [[DOTOMP_UB]], align 4 4595 // CHECK41-NEXT: store i32 0, i32* [[I]], align 4 4596 // CHECK41-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 4597 // CHECK41-NEXT: [[CMP:%.*]] = icmp slt i32 0, [[TMP6]] 4598 // CHECK41-NEXT: br i1 [[CMP]], label [[SIMD_IF_THEN:%.*]], label [[SIMD_IF_END:%.*]] 4599 // CHECK41: simd.if.then: 4600 // CHECK41-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4 4601 // CHECK41-NEXT: store i32 [[TMP7]], i32* [[DOTOMP_IV]], align 4 4602 // CHECK41-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] 4603 // CHECK41: omp.inner.for.cond: 4604 // CHECK41-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !2 4605 // CHECK41-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4, !llvm.access.group !2 4606 // CHECK41-NEXT: [[CMP4:%.*]] = icmp sle i32 [[TMP8]], [[TMP9]] 4607 // CHECK41-NEXT: br i1 [[CMP4]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] 4608 // CHECK41: omp.inner.for.body: 4609 // CHECK41-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !2 4610 // CHECK41-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP10]], 1 4611 // CHECK41-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] 4612 // CHECK41-NEXT: store i32 [[ADD]], i32* [[I3]], align 4, !llvm.access.group !2 4613 // CHECK41-NEXT: [[TMP11:%.*]] = load i32, i32* [[I3]], align 4, !llvm.access.group !2 4614 // CHECK41-NEXT: [[IDXPROM:%.*]] = sext i32 [[TMP11]] to i64 4615 // CHECK41-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i32, i32* [[VLA]], i64 [[IDXPROM]] 4616 // CHECK41-NEXT: store i32 0, i32* [[ARRAYIDX]], align 4, !llvm.access.group !2 4617 // CHECK41-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] 4618 // CHECK41: omp.body.continue: 4619 // CHECK41-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] 4620 // CHECK41: omp.inner.for.inc: 4621 // CHECK41-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !2 4622 // CHECK41-NEXT: [[ADD5:%.*]] = add nsw i32 [[TMP12]], 1 4623 // CHECK41-NEXT: store i32 [[ADD5]], i32* [[DOTOMP_IV]], align 4, !llvm.access.group !2 4624 // CHECK41-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP3:![0-9]+]] 4625 // CHECK41: omp.inner.for.end: 4626 // CHECK41-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 4627 // CHECK41-NEXT: [[SUB6:%.*]] = sub nsw i32 [[TMP13]], 0 4628 // CHECK41-NEXT: [[DIV7:%.*]] = sdiv i32 [[SUB6]], 1 4629 // CHECK41-NEXT: [[MUL8:%.*]] = mul nsw i32 [[DIV7]], 1 4630 // CHECK41-NEXT: [[ADD9:%.*]] = add nsw i32 0, [[MUL8]] 4631 // CHECK41-NEXT: store i32 [[ADD9]], i32* [[I3]], align 4 4632 // CHECK41-NEXT: br label [[SIMD_IF_END]] 4633 // CHECK41: simd.if.end: 4634 // CHECK41-NEXT: [[TMP14:%.*]] = load i32, i32* [[ARGC_ADDR]], align 4 4635 // CHECK41-NEXT: [[CALL:%.*]] = call noundef signext i32 @_Z5tmainIiLi10EEiT_(i32 noundef signext [[TMP14]]) 4636 // CHECK41-NEXT: store i32 [[CALL]], i32* [[RETVAL]], align 4 4637 // CHECK41-NEXT: [[TMP15:%.*]] = load i8*, i8** [[SAVED_STACK]], align 8 4638 // CHECK41-NEXT: call void @llvm.stackrestore(i8* [[TMP15]]) 4639 // CHECK41-NEXT: [[TMP16:%.*]] = load i32, i32* [[RETVAL]], align 4 4640 // CHECK41-NEXT: ret i32 [[TMP16]] 4641 // 4642 // 4643 // CHECK41-LABEL: define {{[^@]+}}@_Z5tmainIiLi10EEiT_ 4644 // CHECK41-SAME: (i32 noundef signext [[ARGC:%.*]]) #[[ATTR2:[0-9]+]] comdat { 4645 // CHECK41-NEXT: entry: 4646 // CHECK41-NEXT: [[ARGC_ADDR:%.*]] = alloca i32, align 4 4647 // CHECK41-NEXT: [[A:%.*]] = alloca [10 x i32], align 4 4648 // CHECK41-NEXT: [[TE:%.*]] = alloca i32, align 4 4649 // CHECK41-NEXT: [[TH:%.*]] = alloca i32, align 4 4650 // CHECK41-NEXT: [[TMP:%.*]] = alloca i32, align 4 4651 // CHECK41-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 4652 // CHECK41-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4 4653 // CHECK41-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 4654 // CHECK41-NEXT: [[I:%.*]] = alloca i32, align 4 4655 // CHECK41-NEXT: store i32 [[ARGC]], i32* [[ARGC_ADDR]], align 4 4656 // CHECK41-NEXT: store i32 0, i32* [[TE]], align 4 4657 // CHECK41-NEXT: store i32 128, i32* [[TH]], align 4 4658 // CHECK41-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4 4659 // CHECK41-NEXT: store i32 9, i32* [[DOTOMP_UB]], align 4 4660 // CHECK41-NEXT: [[TMP0:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4 4661 // CHECK41-NEXT: store i32 [[TMP0]], i32* [[DOTOMP_IV]], align 4 4662 // CHECK41-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] 4663 // CHECK41: omp.inner.for.cond: 4664 // CHECK41-NEXT: [[TMP1:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !6 4665 // CHECK41-NEXT: [[TMP2:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4, !llvm.access.group !6 4666 // CHECK41-NEXT: [[CMP:%.*]] = icmp sle i32 [[TMP1]], [[TMP2]] 4667 // CHECK41-NEXT: br i1 [[CMP]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] 4668 // CHECK41: omp.inner.for.body: 4669 // CHECK41-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !6 4670 // CHECK41-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP3]], 1 4671 // CHECK41-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] 4672 // CHECK41-NEXT: store i32 [[ADD]], i32* [[I]], align 4, !llvm.access.group !6 4673 // CHECK41-NEXT: [[TMP4:%.*]] = load i32, i32* [[I]], align 4, !llvm.access.group !6 4674 // CHECK41-NEXT: [[IDXPROM:%.*]] = sext i32 [[TMP4]] to i64 4675 // CHECK41-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [10 x i32], [10 x i32]* [[A]], i64 0, i64 [[IDXPROM]] 4676 // CHECK41-NEXT: store i32 0, i32* [[ARRAYIDX]], align 4, !llvm.access.group !6 4677 // CHECK41-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] 4678 // CHECK41: omp.body.continue: 4679 // CHECK41-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] 4680 // CHECK41: omp.inner.for.inc: 4681 // CHECK41-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !6 4682 // CHECK41-NEXT: [[ADD1:%.*]] = add nsw i32 [[TMP5]], 1 4683 // CHECK41-NEXT: store i32 [[ADD1]], i32* [[DOTOMP_IV]], align 4, !llvm.access.group !6 4684 // CHECK41-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP7:![0-9]+]] 4685 // CHECK41: omp.inner.for.end: 4686 // CHECK41-NEXT: store i32 10, i32* [[I]], align 4 4687 // CHECK41-NEXT: ret i32 0 4688 // 4689 // 4690 // CHECK43-LABEL: define {{[^@]+}}@main 4691 // CHECK43-SAME: (i32 noundef [[ARGC:%.*]], i8** noundef [[ARGV:%.*]]) #[[ATTR0:[0-9]+]] { 4692 // CHECK43-NEXT: entry: 4693 // CHECK43-NEXT: [[RETVAL:%.*]] = alloca i32, align 4 4694 // CHECK43-NEXT: [[ARGC_ADDR:%.*]] = alloca i32, align 4 4695 // CHECK43-NEXT: [[ARGV_ADDR:%.*]] = alloca i8**, align 4 4696 // CHECK43-NEXT: [[N:%.*]] = alloca i32, align 4 4697 // CHECK43-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 4 4698 // CHECK43-NEXT: [[__VLA_EXPR0:%.*]] = alloca i32, align 4 4699 // CHECK43-NEXT: [[TMP:%.*]] = alloca i32, align 4 4700 // CHECK43-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 4701 // CHECK43-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 4702 // CHECK43-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 4703 // CHECK43-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4 4704 // CHECK43-NEXT: [[I:%.*]] = alloca i32, align 4 4705 // CHECK43-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 4706 // CHECK43-NEXT: [[I3:%.*]] = alloca i32, align 4 4707 // CHECK43-NEXT: store i32 0, i32* [[RETVAL]], align 4 4708 // CHECK43-NEXT: store i32 [[ARGC]], i32* [[ARGC_ADDR]], align 4 4709 // CHECK43-NEXT: store i8** [[ARGV]], i8*** [[ARGV_ADDR]], align 4 4710 // CHECK43-NEXT: store i32 100, i32* [[N]], align 4 4711 // CHECK43-NEXT: [[TMP0:%.*]] = load i32, i32* [[N]], align 4 4712 // CHECK43-NEXT: [[TMP1:%.*]] = call i8* @llvm.stacksave() 4713 // CHECK43-NEXT: store i8* [[TMP1]], i8** [[SAVED_STACK]], align 4 4714 // CHECK43-NEXT: [[VLA:%.*]] = alloca i32, i32 [[TMP0]], align 4 4715 // CHECK43-NEXT: store i32 [[TMP0]], i32* [[__VLA_EXPR0]], align 4 4716 // CHECK43-NEXT: [[TMP2:%.*]] = load i32, i32* [[N]], align 4 4717 // CHECK43-NEXT: store i32 [[TMP2]], i32* [[DOTCAPTURE_EXPR_]], align 4 4718 // CHECK43-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 4719 // CHECK43-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP3]], 0 4720 // CHECK43-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 4721 // CHECK43-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1 4722 // CHECK43-NEXT: store i32 [[SUB2]], i32* [[DOTCAPTURE_EXPR_1]], align 4 4723 // CHECK43-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4 4724 // CHECK43-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 4725 // CHECK43-NEXT: store i32 [[TMP4]], i32* [[DOTOMP_UB]], align 4 4726 // CHECK43-NEXT: store i32 0, i32* [[I]], align 4 4727 // CHECK43-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 4728 // CHECK43-NEXT: [[CMP:%.*]] = icmp slt i32 0, [[TMP5]] 4729 // CHECK43-NEXT: br i1 [[CMP]], label [[SIMD_IF_THEN:%.*]], label [[SIMD_IF_END:%.*]] 4730 // CHECK43: simd.if.then: 4731 // CHECK43-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4 4732 // CHECK43-NEXT: store i32 [[TMP6]], i32* [[DOTOMP_IV]], align 4 4733 // CHECK43-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] 4734 // CHECK43: omp.inner.for.cond: 4735 // CHECK43-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !3 4736 // CHECK43-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4, !llvm.access.group !3 4737 // CHECK43-NEXT: [[CMP4:%.*]] = icmp sle i32 [[TMP7]], [[TMP8]] 4738 // CHECK43-NEXT: br i1 [[CMP4]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] 4739 // CHECK43: omp.inner.for.body: 4740 // CHECK43-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !3 4741 // CHECK43-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP9]], 1 4742 // CHECK43-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] 4743 // CHECK43-NEXT: store i32 [[ADD]], i32* [[I3]], align 4, !llvm.access.group !3 4744 // CHECK43-NEXT: [[TMP10:%.*]] = load i32, i32* [[I3]], align 4, !llvm.access.group !3 4745 // CHECK43-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i32, i32* [[VLA]], i32 [[TMP10]] 4746 // CHECK43-NEXT: store i32 0, i32* [[ARRAYIDX]], align 4, !llvm.access.group !3 4747 // CHECK43-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] 4748 // CHECK43: omp.body.continue: 4749 // CHECK43-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] 4750 // CHECK43: omp.inner.for.inc: 4751 // CHECK43-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !3 4752 // CHECK43-NEXT: [[ADD5:%.*]] = add nsw i32 [[TMP11]], 1 4753 // CHECK43-NEXT: store i32 [[ADD5]], i32* [[DOTOMP_IV]], align 4, !llvm.access.group !3 4754 // CHECK43-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP4:![0-9]+]] 4755 // CHECK43: omp.inner.for.end: 4756 // CHECK43-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 4757 // CHECK43-NEXT: [[SUB6:%.*]] = sub nsw i32 [[TMP12]], 0 4758 // CHECK43-NEXT: [[DIV7:%.*]] = sdiv i32 [[SUB6]], 1 4759 // CHECK43-NEXT: [[MUL8:%.*]] = mul nsw i32 [[DIV7]], 1 4760 // CHECK43-NEXT: [[ADD9:%.*]] = add nsw i32 0, [[MUL8]] 4761 // CHECK43-NEXT: store i32 [[ADD9]], i32* [[I3]], align 4 4762 // CHECK43-NEXT: br label [[SIMD_IF_END]] 4763 // CHECK43: simd.if.end: 4764 // CHECK43-NEXT: [[TMP13:%.*]] = load i32, i32* [[ARGC_ADDR]], align 4 4765 // CHECK43-NEXT: [[CALL:%.*]] = call noundef i32 @_Z5tmainIiLi10EEiT_(i32 noundef [[TMP13]]) 4766 // CHECK43-NEXT: store i32 [[CALL]], i32* [[RETVAL]], align 4 4767 // CHECK43-NEXT: [[TMP14:%.*]] = load i8*, i8** [[SAVED_STACK]], align 4 4768 // CHECK43-NEXT: call void @llvm.stackrestore(i8* [[TMP14]]) 4769 // CHECK43-NEXT: [[TMP15:%.*]] = load i32, i32* [[RETVAL]], align 4 4770 // CHECK43-NEXT: ret i32 [[TMP15]] 4771 // 4772 // 4773 // CHECK43-LABEL: define {{[^@]+}}@_Z5tmainIiLi10EEiT_ 4774 // CHECK43-SAME: (i32 noundef [[ARGC:%.*]]) #[[ATTR2:[0-9]+]] comdat { 4775 // CHECK43-NEXT: entry: 4776 // CHECK43-NEXT: [[ARGC_ADDR:%.*]] = alloca i32, align 4 4777 // CHECK43-NEXT: [[A:%.*]] = alloca [10 x i32], align 4 4778 // CHECK43-NEXT: [[TE:%.*]] = alloca i32, align 4 4779 // CHECK43-NEXT: [[TH:%.*]] = alloca i32, align 4 4780 // CHECK43-NEXT: [[TMP:%.*]] = alloca i32, align 4 4781 // CHECK43-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 4782 // CHECK43-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4 4783 // CHECK43-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 4784 // CHECK43-NEXT: [[I:%.*]] = alloca i32, align 4 4785 // CHECK43-NEXT: store i32 [[ARGC]], i32* [[ARGC_ADDR]], align 4 4786 // CHECK43-NEXT: store i32 0, i32* [[TE]], align 4 4787 // CHECK43-NEXT: store i32 128, i32* [[TH]], align 4 4788 // CHECK43-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4 4789 // CHECK43-NEXT: store i32 9, i32* [[DOTOMP_UB]], align 4 4790 // CHECK43-NEXT: [[TMP0:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4 4791 // CHECK43-NEXT: store i32 [[TMP0]], i32* [[DOTOMP_IV]], align 4 4792 // CHECK43-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] 4793 // CHECK43: omp.inner.for.cond: 4794 // CHECK43-NEXT: [[TMP1:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !7 4795 // CHECK43-NEXT: [[TMP2:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4, !llvm.access.group !7 4796 // CHECK43-NEXT: [[CMP:%.*]] = icmp sle i32 [[TMP1]], [[TMP2]] 4797 // CHECK43-NEXT: br i1 [[CMP]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] 4798 // CHECK43: omp.inner.for.body: 4799 // CHECK43-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !7 4800 // CHECK43-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP3]], 1 4801 // CHECK43-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] 4802 // CHECK43-NEXT: store i32 [[ADD]], i32* [[I]], align 4, !llvm.access.group !7 4803 // CHECK43-NEXT: [[TMP4:%.*]] = load i32, i32* [[I]], align 4, !llvm.access.group !7 4804 // CHECK43-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [10 x i32], [10 x i32]* [[A]], i32 0, i32 [[TMP4]] 4805 // CHECK43-NEXT: store i32 0, i32* [[ARRAYIDX]], align 4, !llvm.access.group !7 4806 // CHECK43-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] 4807 // CHECK43: omp.body.continue: 4808 // CHECK43-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] 4809 // CHECK43: omp.inner.for.inc: 4810 // CHECK43-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !7 4811 // CHECK43-NEXT: [[ADD1:%.*]] = add nsw i32 [[TMP5]], 1 4812 // CHECK43-NEXT: store i32 [[ADD1]], i32* [[DOTOMP_IV]], align 4, !llvm.access.group !7 4813 // CHECK43-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP8:![0-9]+]] 4814 // CHECK43: omp.inner.for.end: 4815 // CHECK43-NEXT: store i32 10, i32* [[I]], align 4 4816 // CHECK43-NEXT: ret i32 0 4817 // 4818 // 4819 // CHECK45-LABEL: define {{[^@]+}}@main 4820 // CHECK45-SAME: (i32 noundef signext [[ARGC:%.*]], i8** noundef [[ARGV:%.*]]) #[[ATTR0:[0-9]+]] { 4821 // CHECK45-NEXT: entry: 4822 // CHECK45-NEXT: [[RETVAL:%.*]] = alloca i32, align 4 4823 // CHECK45-NEXT: [[ARGC_ADDR:%.*]] = alloca i32, align 4 4824 // CHECK45-NEXT: [[ARGV_ADDR:%.*]] = alloca i8**, align 8 4825 // CHECK45-NEXT: [[N:%.*]] = alloca i32, align 4 4826 // CHECK45-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 8 4827 // CHECK45-NEXT: [[__VLA_EXPR0:%.*]] = alloca i64, align 8 4828 // CHECK45-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i8, align 1 4829 // CHECK45-NEXT: [[TMP:%.*]] = alloca i32, align 4 4830 // CHECK45-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 4831 // CHECK45-NEXT: [[DOTCAPTURE_EXPR_2:%.*]] = alloca i32, align 4 4832 // CHECK45-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 4833 // CHECK45-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4 4834 // CHECK45-NEXT: [[I:%.*]] = alloca i32, align 4 4835 // CHECK45-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 4836 // CHECK45-NEXT: [[I4:%.*]] = alloca i32, align 4 4837 // CHECK45-NEXT: store i32 0, i32* [[RETVAL]], align 4 4838 // CHECK45-NEXT: store i32 [[ARGC]], i32* [[ARGC_ADDR]], align 4 4839 // CHECK45-NEXT: store i8** [[ARGV]], i8*** [[ARGV_ADDR]], align 8 4840 // CHECK45-NEXT: store i32 100, i32* [[N]], align 4 4841 // CHECK45-NEXT: [[TMP0:%.*]] = load i32, i32* [[N]], align 4 4842 // CHECK45-NEXT: [[TMP1:%.*]] = zext i32 [[TMP0]] to i64 4843 // CHECK45-NEXT: [[TMP2:%.*]] = call i8* @llvm.stacksave() 4844 // CHECK45-NEXT: store i8* [[TMP2]], i8** [[SAVED_STACK]], align 8 4845 // CHECK45-NEXT: [[VLA:%.*]] = alloca i32, i64 [[TMP1]], align 4 4846 // CHECK45-NEXT: store i64 [[TMP1]], i64* [[__VLA_EXPR0]], align 8 4847 // CHECK45-NEXT: [[TMP3:%.*]] = load i32, i32* [[ARGC_ADDR]], align 4 4848 // CHECK45-NEXT: [[TOBOOL:%.*]] = icmp ne i32 [[TMP3]], 0 4849 // CHECK45-NEXT: [[FROMBOOL:%.*]] = zext i1 [[TOBOOL]] to i8 4850 // CHECK45-NEXT: store i8 [[FROMBOOL]], i8* [[DOTCAPTURE_EXPR_]], align 1 4851 // CHECK45-NEXT: [[TMP4:%.*]] = load i32, i32* [[N]], align 4 4852 // CHECK45-NEXT: store i32 [[TMP4]], i32* [[DOTCAPTURE_EXPR_1]], align 4 4853 // CHECK45-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 4854 // CHECK45-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP5]], 0 4855 // CHECK45-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 4856 // CHECK45-NEXT: [[SUB3:%.*]] = sub nsw i32 [[DIV]], 1 4857 // CHECK45-NEXT: store i32 [[SUB3]], i32* [[DOTCAPTURE_EXPR_2]], align 4 4858 // CHECK45-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4 4859 // CHECK45-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_2]], align 4 4860 // CHECK45-NEXT: store i32 [[TMP6]], i32* [[DOTOMP_UB]], align 4 4861 // CHECK45-NEXT: store i32 0, i32* [[I]], align 4 4862 // CHECK45-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 4863 // CHECK45-NEXT: [[CMP:%.*]] = icmp slt i32 0, [[TMP7]] 4864 // CHECK45-NEXT: br i1 [[CMP]], label [[SIMD_IF_THEN:%.*]], label [[SIMD_IF_END:%.*]] 4865 // CHECK45: simd.if.then: 4866 // CHECK45-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4 4867 // CHECK45-NEXT: store i32 [[TMP8]], i32* [[DOTOMP_IV]], align 4 4868 // CHECK45-NEXT: [[TMP9:%.*]] = load i8, i8* [[DOTCAPTURE_EXPR_]], align 1 4869 // CHECK45-NEXT: [[TOBOOL5:%.*]] = trunc i8 [[TMP9]] to i1 4870 // CHECK45-NEXT: br i1 [[TOBOOL5]], label [[OMP_IF_THEN:%.*]], label [[OMP_IF_ELSE:%.*]] 4871 // CHECK45: omp_if.then: 4872 // CHECK45-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] 4873 // CHECK45: omp.inner.for.cond: 4874 // CHECK45-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !2 4875 // CHECK45-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4, !llvm.access.group !2 4876 // CHECK45-NEXT: [[CMP6:%.*]] = icmp sle i32 [[TMP10]], [[TMP11]] 4877 // CHECK45-NEXT: br i1 [[CMP6]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] 4878 // CHECK45: omp.inner.for.body: 4879 // CHECK45-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !2 4880 // CHECK45-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP12]], 1 4881 // CHECK45-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] 4882 // CHECK45-NEXT: store i32 [[ADD]], i32* [[I4]], align 4, !llvm.access.group !2 4883 // CHECK45-NEXT: [[TMP13:%.*]] = load i32, i32* [[I4]], align 4, !llvm.access.group !2 4884 // CHECK45-NEXT: [[IDXPROM:%.*]] = sext i32 [[TMP13]] to i64 4885 // CHECK45-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i32, i32* [[VLA]], i64 [[IDXPROM]] 4886 // CHECK45-NEXT: store i32 0, i32* [[ARRAYIDX]], align 4, !llvm.access.group !2 4887 // CHECK45-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] 4888 // CHECK45: omp.body.continue: 4889 // CHECK45-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] 4890 // CHECK45: omp.inner.for.inc: 4891 // CHECK45-NEXT: [[TMP14:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !2 4892 // CHECK45-NEXT: [[ADD7:%.*]] = add nsw i32 [[TMP14]], 1 4893 // CHECK45-NEXT: store i32 [[ADD7]], i32* [[DOTOMP_IV]], align 4, !llvm.access.group !2 4894 // CHECK45-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP3:![0-9]+]] 4895 // CHECK45: omp.inner.for.end: 4896 // CHECK45-NEXT: br label [[OMP_IF_END:%.*]] 4897 // CHECK45: omp_if.else: 4898 // CHECK45-NEXT: br label [[OMP_INNER_FOR_COND8:%.*]] 4899 // CHECK45: omp.inner.for.cond8: 4900 // CHECK45-NEXT: [[TMP15:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 4901 // CHECK45-NEXT: [[TMP16:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 4902 // CHECK45-NEXT: [[CMP9:%.*]] = icmp sle i32 [[TMP15]], [[TMP16]] 4903 // CHECK45-NEXT: br i1 [[CMP9]], label [[OMP_INNER_FOR_BODY10:%.*]], label [[OMP_INNER_FOR_END18:%.*]] 4904 // CHECK45: omp.inner.for.body10: 4905 // CHECK45-NEXT: [[TMP17:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 4906 // CHECK45-NEXT: [[MUL11:%.*]] = mul nsw i32 [[TMP17]], 1 4907 // CHECK45-NEXT: [[ADD12:%.*]] = add nsw i32 0, [[MUL11]] 4908 // CHECK45-NEXT: store i32 [[ADD12]], i32* [[I4]], align 4 4909 // CHECK45-NEXT: [[TMP18:%.*]] = load i32, i32* [[I4]], align 4 4910 // CHECK45-NEXT: [[IDXPROM13:%.*]] = sext i32 [[TMP18]] to i64 4911 // CHECK45-NEXT: [[ARRAYIDX14:%.*]] = getelementptr inbounds i32, i32* [[VLA]], i64 [[IDXPROM13]] 4912 // CHECK45-NEXT: store i32 0, i32* [[ARRAYIDX14]], align 4 4913 // CHECK45-NEXT: br label [[OMP_BODY_CONTINUE15:%.*]] 4914 // CHECK45: omp.body.continue15: 4915 // CHECK45-NEXT: br label [[OMP_INNER_FOR_INC16:%.*]] 4916 // CHECK45: omp.inner.for.inc16: 4917 // CHECK45-NEXT: [[TMP19:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 4918 // CHECK45-NEXT: [[ADD17:%.*]] = add nsw i32 [[TMP19]], 1 4919 // CHECK45-NEXT: store i32 [[ADD17]], i32* [[DOTOMP_IV]], align 4 4920 // CHECK45-NEXT: br label [[OMP_INNER_FOR_COND8]], !llvm.loop [[LOOP6:![0-9]+]] 4921 // CHECK45: omp.inner.for.end18: 4922 // CHECK45-NEXT: br label [[OMP_IF_END]] 4923 // CHECK45: omp_if.end: 4924 // CHECK45-NEXT: [[TMP20:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 4925 // CHECK45-NEXT: [[SUB19:%.*]] = sub nsw i32 [[TMP20]], 0 4926 // CHECK45-NEXT: [[DIV20:%.*]] = sdiv i32 [[SUB19]], 1 4927 // CHECK45-NEXT: [[MUL21:%.*]] = mul nsw i32 [[DIV20]], 1 4928 // CHECK45-NEXT: [[ADD22:%.*]] = add nsw i32 0, [[MUL21]] 4929 // CHECK45-NEXT: store i32 [[ADD22]], i32* [[I4]], align 4 4930 // CHECK45-NEXT: br label [[SIMD_IF_END]] 4931 // CHECK45: simd.if.end: 4932 // CHECK45-NEXT: [[TMP21:%.*]] = load i32, i32* [[ARGC_ADDR]], align 4 4933 // CHECK45-NEXT: [[CALL:%.*]] = call noundef signext i32 @_Z5tmainIiLi10EEiT_(i32 noundef signext [[TMP21]]) 4934 // CHECK45-NEXT: store i32 [[CALL]], i32* [[RETVAL]], align 4 4935 // CHECK45-NEXT: [[TMP22:%.*]] = load i8*, i8** [[SAVED_STACK]], align 8 4936 // CHECK45-NEXT: call void @llvm.stackrestore(i8* [[TMP22]]) 4937 // CHECK45-NEXT: [[TMP23:%.*]] = load i32, i32* [[RETVAL]], align 4 4938 // CHECK45-NEXT: ret i32 [[TMP23]] 4939 // 4940 // 4941 // CHECK45-LABEL: define {{[^@]+}}@_Z5tmainIiLi10EEiT_ 4942 // CHECK45-SAME: (i32 noundef signext [[ARGC:%.*]]) #[[ATTR2:[0-9]+]] comdat { 4943 // CHECK45-NEXT: entry: 4944 // CHECK45-NEXT: [[ARGC_ADDR:%.*]] = alloca i32, align 4 4945 // CHECK45-NEXT: [[A:%.*]] = alloca [10 x i32], align 4 4946 // CHECK45-NEXT: [[TE:%.*]] = alloca i32, align 4 4947 // CHECK45-NEXT: [[TH:%.*]] = alloca i32, align 4 4948 // CHECK45-NEXT: [[TMP:%.*]] = alloca i32, align 4 4949 // CHECK45-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 4950 // CHECK45-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4 4951 // CHECK45-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 4952 // CHECK45-NEXT: [[I:%.*]] = alloca i32, align 4 4953 // CHECK45-NEXT: store i32 [[ARGC]], i32* [[ARGC_ADDR]], align 4 4954 // CHECK45-NEXT: store i32 0, i32* [[TE]], align 4 4955 // CHECK45-NEXT: store i32 128, i32* [[TH]], align 4 4956 // CHECK45-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4 4957 // CHECK45-NEXT: store i32 9, i32* [[DOTOMP_UB]], align 4 4958 // CHECK45-NEXT: [[TMP0:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4 4959 // CHECK45-NEXT: store i32 [[TMP0]], i32* [[DOTOMP_IV]], align 4 4960 // CHECK45-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] 4961 // CHECK45: omp.inner.for.cond: 4962 // CHECK45-NEXT: [[TMP1:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !8 4963 // CHECK45-NEXT: [[TMP2:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4, !llvm.access.group !8 4964 // CHECK45-NEXT: [[CMP:%.*]] = icmp sle i32 [[TMP1]], [[TMP2]] 4965 // CHECK45-NEXT: br i1 [[CMP]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] 4966 // CHECK45: omp.inner.for.body: 4967 // CHECK45-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !8 4968 // CHECK45-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP3]], 1 4969 // CHECK45-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] 4970 // CHECK45-NEXT: store i32 [[ADD]], i32* [[I]], align 4, !llvm.access.group !8 4971 // CHECK45-NEXT: [[TMP4:%.*]] = load i32, i32* [[I]], align 4, !llvm.access.group !8 4972 // CHECK45-NEXT: [[IDXPROM:%.*]] = sext i32 [[TMP4]] to i64 4973 // CHECK45-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [10 x i32], [10 x i32]* [[A]], i64 0, i64 [[IDXPROM]] 4974 // CHECK45-NEXT: store i32 0, i32* [[ARRAYIDX]], align 4, !llvm.access.group !8 4975 // CHECK45-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] 4976 // CHECK45: omp.body.continue: 4977 // CHECK45-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] 4978 // CHECK45: omp.inner.for.inc: 4979 // CHECK45-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !8 4980 // CHECK45-NEXT: [[ADD1:%.*]] = add nsw i32 [[TMP5]], 1 4981 // CHECK45-NEXT: store i32 [[ADD1]], i32* [[DOTOMP_IV]], align 4, !llvm.access.group !8 4982 // CHECK45-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP9:![0-9]+]] 4983 // CHECK45: omp.inner.for.end: 4984 // CHECK45-NEXT: store i32 10, i32* [[I]], align 4 4985 // CHECK45-NEXT: ret i32 0 4986 // 4987 // 4988 // CHECK47-LABEL: define {{[^@]+}}@main 4989 // CHECK47-SAME: (i32 noundef [[ARGC:%.*]], i8** noundef [[ARGV:%.*]]) #[[ATTR0:[0-9]+]] { 4990 // CHECK47-NEXT: entry: 4991 // CHECK47-NEXT: [[RETVAL:%.*]] = alloca i32, align 4 4992 // CHECK47-NEXT: [[ARGC_ADDR:%.*]] = alloca i32, align 4 4993 // CHECK47-NEXT: [[ARGV_ADDR:%.*]] = alloca i8**, align 4 4994 // CHECK47-NEXT: [[N:%.*]] = alloca i32, align 4 4995 // CHECK47-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 4 4996 // CHECK47-NEXT: [[__VLA_EXPR0:%.*]] = alloca i32, align 4 4997 // CHECK47-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i8, align 1 4998 // CHECK47-NEXT: [[TMP:%.*]] = alloca i32, align 4 4999 // CHECK47-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 5000 // CHECK47-NEXT: [[DOTCAPTURE_EXPR_2:%.*]] = alloca i32, align 4 5001 // CHECK47-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 5002 // CHECK47-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4 5003 // CHECK47-NEXT: [[I:%.*]] = alloca i32, align 4 5004 // CHECK47-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 5005 // CHECK47-NEXT: [[I4:%.*]] = alloca i32, align 4 5006 // CHECK47-NEXT: store i32 0, i32* [[RETVAL]], align 4 5007 // CHECK47-NEXT: store i32 [[ARGC]], i32* [[ARGC_ADDR]], align 4 5008 // CHECK47-NEXT: store i8** [[ARGV]], i8*** [[ARGV_ADDR]], align 4 5009 // CHECK47-NEXT: store i32 100, i32* [[N]], align 4 5010 // CHECK47-NEXT: [[TMP0:%.*]] = load i32, i32* [[N]], align 4 5011 // CHECK47-NEXT: [[TMP1:%.*]] = call i8* @llvm.stacksave() 5012 // CHECK47-NEXT: store i8* [[TMP1]], i8** [[SAVED_STACK]], align 4 5013 // CHECK47-NEXT: [[VLA:%.*]] = alloca i32, i32 [[TMP0]], align 4 5014 // CHECK47-NEXT: store i32 [[TMP0]], i32* [[__VLA_EXPR0]], align 4 5015 // CHECK47-NEXT: [[TMP2:%.*]] = load i32, i32* [[ARGC_ADDR]], align 4 5016 // CHECK47-NEXT: [[TOBOOL:%.*]] = icmp ne i32 [[TMP2]], 0 5017 // CHECK47-NEXT: [[FROMBOOL:%.*]] = zext i1 [[TOBOOL]] to i8 5018 // CHECK47-NEXT: store i8 [[FROMBOOL]], i8* [[DOTCAPTURE_EXPR_]], align 1 5019 // CHECK47-NEXT: [[TMP3:%.*]] = load i32, i32* [[N]], align 4 5020 // CHECK47-NEXT: store i32 [[TMP3]], i32* [[DOTCAPTURE_EXPR_1]], align 4 5021 // CHECK47-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 5022 // CHECK47-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP4]], 0 5023 // CHECK47-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 5024 // CHECK47-NEXT: [[SUB3:%.*]] = sub nsw i32 [[DIV]], 1 5025 // CHECK47-NEXT: store i32 [[SUB3]], i32* [[DOTCAPTURE_EXPR_2]], align 4 5026 // CHECK47-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4 5027 // CHECK47-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_2]], align 4 5028 // CHECK47-NEXT: store i32 [[TMP5]], i32* [[DOTOMP_UB]], align 4 5029 // CHECK47-NEXT: store i32 0, i32* [[I]], align 4 5030 // CHECK47-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 5031 // CHECK47-NEXT: [[CMP:%.*]] = icmp slt i32 0, [[TMP6]] 5032 // CHECK47-NEXT: br i1 [[CMP]], label [[SIMD_IF_THEN:%.*]], label [[SIMD_IF_END:%.*]] 5033 // CHECK47: simd.if.then: 5034 // CHECK47-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4 5035 // CHECK47-NEXT: store i32 [[TMP7]], i32* [[DOTOMP_IV]], align 4 5036 // CHECK47-NEXT: [[TMP8:%.*]] = load i8, i8* [[DOTCAPTURE_EXPR_]], align 1 5037 // CHECK47-NEXT: [[TOBOOL5:%.*]] = trunc i8 [[TMP8]] to i1 5038 // CHECK47-NEXT: br i1 [[TOBOOL5]], label [[OMP_IF_THEN:%.*]], label [[OMP_IF_ELSE:%.*]] 5039 // CHECK47: omp_if.then: 5040 // CHECK47-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] 5041 // CHECK47: omp.inner.for.cond: 5042 // CHECK47-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !3 5043 // CHECK47-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4, !llvm.access.group !3 5044 // CHECK47-NEXT: [[CMP6:%.*]] = icmp sle i32 [[TMP9]], [[TMP10]] 5045 // CHECK47-NEXT: br i1 [[CMP6]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] 5046 // CHECK47: omp.inner.for.body: 5047 // CHECK47-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !3 5048 // CHECK47-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP11]], 1 5049 // CHECK47-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] 5050 // CHECK47-NEXT: store i32 [[ADD]], i32* [[I4]], align 4, !llvm.access.group !3 5051 // CHECK47-NEXT: [[TMP12:%.*]] = load i32, i32* [[I4]], align 4, !llvm.access.group !3 5052 // CHECK47-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i32, i32* [[VLA]], i32 [[TMP12]] 5053 // CHECK47-NEXT: store i32 0, i32* [[ARRAYIDX]], align 4, !llvm.access.group !3 5054 // CHECK47-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] 5055 // CHECK47: omp.body.continue: 5056 // CHECK47-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] 5057 // CHECK47: omp.inner.for.inc: 5058 // CHECK47-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !3 5059 // CHECK47-NEXT: [[ADD7:%.*]] = add nsw i32 [[TMP13]], 1 5060 // CHECK47-NEXT: store i32 [[ADD7]], i32* [[DOTOMP_IV]], align 4, !llvm.access.group !3 5061 // CHECK47-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP4:![0-9]+]] 5062 // CHECK47: omp.inner.for.end: 5063 // CHECK47-NEXT: br label [[OMP_IF_END:%.*]] 5064 // CHECK47: omp_if.else: 5065 // CHECK47-NEXT: br label [[OMP_INNER_FOR_COND8:%.*]] 5066 // CHECK47: omp.inner.for.cond8: 5067 // CHECK47-NEXT: [[TMP14:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 5068 // CHECK47-NEXT: [[TMP15:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 5069 // CHECK47-NEXT: [[CMP9:%.*]] = icmp sle i32 [[TMP14]], [[TMP15]] 5070 // CHECK47-NEXT: br i1 [[CMP9]], label [[OMP_INNER_FOR_BODY10:%.*]], label [[OMP_INNER_FOR_END17:%.*]] 5071 // CHECK47: omp.inner.for.body10: 5072 // CHECK47-NEXT: [[TMP16:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 5073 // CHECK47-NEXT: [[MUL11:%.*]] = mul nsw i32 [[TMP16]], 1 5074 // CHECK47-NEXT: [[ADD12:%.*]] = add nsw i32 0, [[MUL11]] 5075 // CHECK47-NEXT: store i32 [[ADD12]], i32* [[I4]], align 4 5076 // CHECK47-NEXT: [[TMP17:%.*]] = load i32, i32* [[I4]], align 4 5077 // CHECK47-NEXT: [[ARRAYIDX13:%.*]] = getelementptr inbounds i32, i32* [[VLA]], i32 [[TMP17]] 5078 // CHECK47-NEXT: store i32 0, i32* [[ARRAYIDX13]], align 4 5079 // CHECK47-NEXT: br label [[OMP_BODY_CONTINUE14:%.*]] 5080 // CHECK47: omp.body.continue14: 5081 // CHECK47-NEXT: br label [[OMP_INNER_FOR_INC15:%.*]] 5082 // CHECK47: omp.inner.for.inc15: 5083 // CHECK47-NEXT: [[TMP18:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 5084 // CHECK47-NEXT: [[ADD16:%.*]] = add nsw i32 [[TMP18]], 1 5085 // CHECK47-NEXT: store i32 [[ADD16]], i32* [[DOTOMP_IV]], align 4 5086 // CHECK47-NEXT: br label [[OMP_INNER_FOR_COND8]], !llvm.loop [[LOOP7:![0-9]+]] 5087 // CHECK47: omp.inner.for.end17: 5088 // CHECK47-NEXT: br label [[OMP_IF_END]] 5089 // CHECK47: omp_if.end: 5090 // CHECK47-NEXT: [[TMP19:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 5091 // CHECK47-NEXT: [[SUB18:%.*]] = sub nsw i32 [[TMP19]], 0 5092 // CHECK47-NEXT: [[DIV19:%.*]] = sdiv i32 [[SUB18]], 1 5093 // CHECK47-NEXT: [[MUL20:%.*]] = mul nsw i32 [[DIV19]], 1 5094 // CHECK47-NEXT: [[ADD21:%.*]] = add nsw i32 0, [[MUL20]] 5095 // CHECK47-NEXT: store i32 [[ADD21]], i32* [[I4]], align 4 5096 // CHECK47-NEXT: br label [[SIMD_IF_END]] 5097 // CHECK47: simd.if.end: 5098 // CHECK47-NEXT: [[TMP20:%.*]] = load i32, i32* [[ARGC_ADDR]], align 4 5099 // CHECK47-NEXT: [[CALL:%.*]] = call noundef i32 @_Z5tmainIiLi10EEiT_(i32 noundef [[TMP20]]) 5100 // CHECK47-NEXT: store i32 [[CALL]], i32* [[RETVAL]], align 4 5101 // CHECK47-NEXT: [[TMP21:%.*]] = load i8*, i8** [[SAVED_STACK]], align 4 5102 // CHECK47-NEXT: call void @llvm.stackrestore(i8* [[TMP21]]) 5103 // CHECK47-NEXT: [[TMP22:%.*]] = load i32, i32* [[RETVAL]], align 4 5104 // CHECK47-NEXT: ret i32 [[TMP22]] 5105 // 5106 // 5107 // CHECK47-LABEL: define {{[^@]+}}@_Z5tmainIiLi10EEiT_ 5108 // CHECK47-SAME: (i32 noundef [[ARGC:%.*]]) #[[ATTR2:[0-9]+]] comdat { 5109 // CHECK47-NEXT: entry: 5110 // CHECK47-NEXT: [[ARGC_ADDR:%.*]] = alloca i32, align 4 5111 // CHECK47-NEXT: [[A:%.*]] = alloca [10 x i32], align 4 5112 // CHECK47-NEXT: [[TE:%.*]] = alloca i32, align 4 5113 // CHECK47-NEXT: [[TH:%.*]] = alloca i32, align 4 5114 // CHECK47-NEXT: [[TMP:%.*]] = alloca i32, align 4 5115 // CHECK47-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 5116 // CHECK47-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4 5117 // CHECK47-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 5118 // CHECK47-NEXT: [[I:%.*]] = alloca i32, align 4 5119 // CHECK47-NEXT: store i32 [[ARGC]], i32* [[ARGC_ADDR]], align 4 5120 // CHECK47-NEXT: store i32 0, i32* [[TE]], align 4 5121 // CHECK47-NEXT: store i32 128, i32* [[TH]], align 4 5122 // CHECK47-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4 5123 // CHECK47-NEXT: store i32 9, i32* [[DOTOMP_UB]], align 4 5124 // CHECK47-NEXT: [[TMP0:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4 5125 // CHECK47-NEXT: store i32 [[TMP0]], i32* [[DOTOMP_IV]], align 4 5126 // CHECK47-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] 5127 // CHECK47: omp.inner.for.cond: 5128 // CHECK47-NEXT: [[TMP1:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !9 5129 // CHECK47-NEXT: [[TMP2:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4, !llvm.access.group !9 5130 // CHECK47-NEXT: [[CMP:%.*]] = icmp sle i32 [[TMP1]], [[TMP2]] 5131 // CHECK47-NEXT: br i1 [[CMP]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] 5132 // CHECK47: omp.inner.for.body: 5133 // CHECK47-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !9 5134 // CHECK47-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP3]], 1 5135 // CHECK47-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] 5136 // CHECK47-NEXT: store i32 [[ADD]], i32* [[I]], align 4, !llvm.access.group !9 5137 // CHECK47-NEXT: [[TMP4:%.*]] = load i32, i32* [[I]], align 4, !llvm.access.group !9 5138 // CHECK47-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [10 x i32], [10 x i32]* [[A]], i32 0, i32 [[TMP4]] 5139 // CHECK47-NEXT: store i32 0, i32* [[ARRAYIDX]], align 4, !llvm.access.group !9 5140 // CHECK47-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] 5141 // CHECK47: omp.body.continue: 5142 // CHECK47-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] 5143 // CHECK47: omp.inner.for.inc: 5144 // CHECK47-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !9 5145 // CHECK47-NEXT: [[ADD1:%.*]] = add nsw i32 [[TMP5]], 1 5146 // CHECK47-NEXT: store i32 [[ADD1]], i32* [[DOTOMP_IV]], align 4, !llvm.access.group !9 5147 // CHECK47-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP10:![0-9]+]] 5148 // CHECK47: omp.inner.for.end: 5149 // CHECK47-NEXT: store i32 10, i32* [[I]], align 4 5150 // CHECK47-NEXT: ret i32 0 5151 // 5152