1 // RUN: %clang_cc1 -DLAMBDA -verify -fopenmp -x c++ -std=c++11 -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -emit-llvm %s -o - -Wno-openmp-mapping | FileCheck %s --check-prefix LAMBDA --check-prefix LAMBDA-64 2 // RUN: %clang_cc1 -DLAMBDA -fopenmp -x c++ -std=c++11 -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -emit-pch -o %t %s -Wno-openmp-mapping 3 // RUN: %clang_cc1 -DLAMBDA -fopenmp -x c++ -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -std=c++11 -include-pch %t -verify %s -emit-llvm -o - -Wno-openmp-mapping | FileCheck %s --check-prefix LAMBDA --check-prefix LAMBDA-64 4 // RUN: %clang_cc1 -DLAMBDA -verify -fopenmp -x c++ -std=c++11 -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -emit-llvm %s -o - -Wno-openmp-mapping | FileCheck %s --check-prefix LAMBDA --check-prefix LAMBDA-32 5 // RUN: %clang_cc1 -DLAMBDA -fopenmp -x c++ -std=c++11 -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -emit-pch -o %t %s -Wno-openmp-mapping 6 // RUN: %clang_cc1 -DLAMBDA -fopenmp -x c++ -std=c++11 -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -std=c++11 -include-pch %t -verify %s -emit-llvm -o - -Wno-openmp-mapping | FileCheck %s --check-prefix LAMBDA --check-prefix LAMBDA-32 7 8 // RUN: %clang_cc1 -DLAMBDA -verify -fopenmp-simd -x c++ -std=c++11 -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -emit-llvm %s -o - -Wno-openmp-mapping | FileCheck --check-prefix SIMD-ONLY0 %s 9 // RUN: %clang_cc1 -DLAMBDA -fopenmp-simd -x c++ -std=c++11 -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -emit-pch -o %t %s -Wno-openmp-mapping 10 // RUN: %clang_cc1 -DLAMBDA -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 - -Wno-openmp-mapping | FileCheck --check-prefix SIMD-ONLY0 %s 11 // RUN: %clang_cc1 -DLAMBDA -verify -fopenmp-simd -x c++ -std=c++11 -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -emit-llvm %s -o - -Wno-openmp-mapping | FileCheck --check-prefix SIMD-ONLY0 %s 12 // RUN: %clang_cc1 -DLAMBDA -fopenmp-simd -x c++ -std=c++11 -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -emit-pch -o %t %s -Wno-openmp-mapping 13 // RUN: %clang_cc1 -DLAMBDA -fopenmp-simd -x c++ -std=c++11 -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -std=c++11 -include-pch %t -verify %s -emit-llvm -o - -Wno-openmp-mapping | FileCheck --check-prefix SIMD-ONLY0 %s 14 // SIMD-ONLY0-NOT: {{__kmpc|__tgt}} 15 16 // RUN: %clang_cc1 -verify -fopenmp -x c++ -std=c++11 -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -emit-llvm %s -o - -Wno-openmp-mapping | FileCheck %s --check-prefix CHECK --check-prefix CHECK-64 17 // RUN: %clang_cc1 -fopenmp -x c++ -std=c++11 -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -emit-pch -o %t %s -Wno-openmp-mapping 18 // RUN: %clang_cc1 -fopenmp -x c++ -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -std=c++11 -include-pch %t -verify %s -emit-llvm -o - -Wno-openmp-mapping | FileCheck %s --check-prefix CHECK --check-prefix CHECK-64 19 // RUN: %clang_cc1 -verify -fopenmp -x c++ -std=c++11 -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -emit-llvm %s -o - -Wno-openmp-mapping | FileCheck %s --check-prefix CHECK --check-prefix CHECK-32 20 // RUN: %clang_cc1 -fopenmp -x c++ -std=c++11 -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -emit-pch -o %t %s -Wno-openmp-mapping 21 // RUN: %clang_cc1 -fopenmp -x c++ -std=c++11 -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -std=c++11 -include-pch %t -verify %s -emit-llvm -o - -Wno-openmp-mapping | FileCheck %s --check-prefix CHECK --check-prefix CHECK-32 22 23 // RUN: %clang_cc1 -verify -fopenmp-simd -x c++ -std=c++11 -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -emit-llvm %s -o - -Wno-openmp-mapping | FileCheck --check-prefix SIMD-ONLY1 %s 24 // RUN: %clang_cc1 -fopenmp-simd -x c++ -std=c++11 -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -emit-pch -o %t %s -Wno-openmp-mapping 25 // RUN: %clang_cc1 -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 - -Wno-openmp-mapping | FileCheck --check-prefix SIMD-ONLY1 %s 26 // RUN: %clang_cc1 -verify -fopenmp-simd -x c++ -std=c++11 -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -emit-llvm %s -o - -Wno-openmp-mapping | FileCheck --check-prefix SIMD-ONLY1 %s 27 // RUN: %clang_cc1 -fopenmp-simd -x c++ -std=c++11 -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -emit-pch -o %t %s -Wno-openmp-mapping 28 // RUN: %clang_cc1 -fopenmp-simd -x c++ -std=c++11 -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -std=c++11 -include-pch %t -verify %s -emit-llvm -o - -Wno-openmp-mapping | FileCheck --check-prefix SIMD-ONLY1 %s 29 // SIMD-ONLY1-NOT: {{__kmpc|__tgt}} 30 // expected-no-diagnostics 31 #ifndef HEADER 32 #define HEADER 33 34 template <class T> 35 struct S { 36 T f; 37 S(T a) : f(a) {} 38 S() : f() {} 39 operator T() { return T(); } 40 ~S() {} 41 }; 42 43 // CHECK: [[S_FLOAT_TY:%.+]] = type { float } 44 // CHECK: [[S_INT_TY:%.+]] = type { i{{[0-9]+}} } 45 template <typename T> 46 T tmain() { 47 S<T> test; 48 T t_var = T(); 49 T vec[] = {1, 2}; 50 S<T> s_arr[] = {1, 2}; 51 S<T> &var = test; 52 #pragma omp target 53 #pragma omp teams distribute parallel for lastprivate(t_var, vec, s_arr, s_arr, var, var) 54 for (int i = 0; i < 2; ++i) { 55 vec[i] = t_var; 56 s_arr[i] = var; 57 } 58 return T(); 59 } 60 61 int main() { 62 static int svar; 63 volatile double g; 64 volatile double &g1 = g; 65 66 #ifdef LAMBDA 67 // LAMBDA-LABEL: @main 68 // LAMBDA: call{{.*}} void [[OUTER_LAMBDA:@.+]]( 69 [&]() { 70 static float sfvar; 71 // LAMBDA: define{{.*}} internal{{.*}} void [[OUTER_LAMBDA]]( 72 // LAMBDA: call i{{[0-9]+}} @__tgt_target_teams_mapper(%struct.ident_t* @{{.+}}, 73 // LAMBDA: call void [[OFFLOADING_FUN:@.+]]( 74 75 // LAMBDA: define{{.+}} void [[OFFLOADING_FUN]]( 76 // LAMBDA: call {{.*}}void {{.+}} @__kmpc_fork_teams({{.+}}, i32 4, {{.+}}* [[OMP_OUTLINED:@.+]] to {{.+}}) 77 #pragma omp target 78 #pragma omp teams distribute parallel for lastprivate(g, g1, svar, sfvar) 79 for (int i = 0; i < 2; ++i) { 80 // LAMBDA: define{{.*}} internal{{.*}} void [[OMP_OUTLINED]](i32* noalias %{{.+}}, i32* noalias %{{.+}}, double*{{.+}} [[G_IN:%.+]], double*{{.+}} [[G1_IN:%.+]], i{{[0-9]+}}*{{.+}} [[SVAR_IN:%.+]], float*{{.+}} [[SFVAR_IN:%.+]]) 81 // LAMBDA: [[G_PRIVATE_ADDR:%.+]] = alloca double*, 82 // LAMBDA: [[G1_PRIVATE_ADDR:%.+]] = alloca double*, 83 // LAMBDA: [[SVAR_PRIVATE_ADDR:%.+]] = alloca i{{[0-9]+}}*, 84 // LAMBDA: [[SFVAR_PRIVATE_ADDR:%.+]] = alloca float*, 85 // loop variables 86 // LAMBDA: {{.+}} = alloca i{{[0-9]+}}, 87 // LAMBDA: {{.+}} = alloca i{{[0-9]+}}, 88 // LAMBDA: {{.+}} = alloca i{{[0-9]+}}, 89 // LAMBDA: {{.+}} = alloca i{{[0-9]+}}, 90 // LAMBDA: {{.+}} = alloca i{{[0-9]+}}, 91 // LAMBDA: [[OMP_IS_LAST:%.+]] = alloca i{{[0-9]+}}, 92 // LAMBDA: [[G_PRIVATE:%.+]] = alloca double, 93 // LAMBDA: [[G1_PRIVATE:%.+]] = alloca double, 94 // LAMBDA: [[TMP_G1_PRIVATE:%.+]] = alloca double*, 95 // LAMBDA: [[SVAR_PRIVATE:%.+]] = alloca i{{[0-9]+}}, 96 // LAMBDA: [[SFVAR_PRIVATE:%.+]] = alloca float, 97 // LAMBDA: store double* [[G_IN]], double** [[G_PRIVATE_ADDR]], 98 // LAMBDA: store double* [[G1_IN]], double** [[G1_PRIVATE_ADDR]], 99 // LAMBDA: store i{{[0-9]+}}* [[SVAR_IN]], i{{[0-9]+}}** [[SVAR_PRIVATE_ADDR]], 100 // LAMBDA: store float* [[SFVAR_IN]], float** [[SFVAR_PRIVATE_ADDR]], 101 102 // init private variables 103 // LAMBDA: [[G_IN_REF:%.+]] = load double*, double** [[G_PRIVATE_ADDR]], 104 // LAMBDA: [[G1_IN_REF:%.+]] = load double*, double** [[G1_PRIVATE_ADDR]], 105 // LAMBDA: [[SVAR_IN_REF:%.+]] = load i{{[0-9]+}}*, i{{[0-9]+}}** [[SVAR_PRIVATE_ADDR]], 106 // LAMBDA: [[SFVAR_IN_REF:%.+]] = load float*, float** [[SFVAR_PRIVATE_ADDR]], 107 // LAMBDA: [[G1_IN_REF:%.+]] = load double*, double** % 108 // LAMBDA: store double* [[G1_PRIVATE]], double** [[TMP_G1_PRIVATE]], 109 g = 1; 110 g1 = 1; 111 svar = 3; 112 sfvar = 4.0; 113 // LAMBDA: call {{.*}}void @__kmpc_for_static_init_4( 114 // LAMBDA: call void {{.*}} @__kmpc_fork_call({{.+}}, {{.+}}, {{.+}} @[[LPAR_OUTL:.+]] to 115 // LAMBDA: call {{.*}}void @__kmpc_for_static_fini( 116 117 // LAMBDA: [[OMP_IS_LAST_VAL:%.+]] = load i{{[0-9]+}}, i{{[0-9]+}}* [[OMP_IS_LAST]], 118 // LAMBDA: [[IS_LAST_IT:%.+]] = icmp ne i{{[0-9]+}} [[OMP_IS_LAST_VAL]], 0 119 // LAMBDA: br i1 [[IS_LAST_IT]], label %[[OMP_LASTPRIV_BLOCK:.+]], label %[[OMP_LASTPRIV_DONE:.+]] 120 121 // LAMBDA: [[OMP_LASTPRIV_BLOCK]]: 122 // LAMBDA: [[G_PRIV_VAL:%.+]] = load double, double* [[G_PRIVATE]], 123 // LAMBDA: store{{.*}} double [[G_PRIV_VAL]], double* [[G_IN_REF]], 124 // LAMBDA: [[TMP_G1_PRIV_REF:%.+]] = load double*, double** [[TMP_G1_PRIVATE]], 125 // LAMBDA: [[TMP_G1_PRIV_VAL:%.+]] = load double, double* [[TMP_G1_PRIV_REF]], 126 // LAMBDA: store{{.*}} double [[TMP_G1_PRIV_VAL]], double* [[G1_IN_REF]], 127 128 // LAMBDA: [[SVAR_PRIV_VAL:%.+]] = load i{{[0-9]+}}, i{{[0-9]+}}* [[SVAR_PRIVATE]], 129 // LAMBDA: store i{{[0-9]+}} [[SVAR_PRIV_VAL]], i{{[0-9]+}}* [[SVAR_IN_REF]], 130 // LAMBDA: [[SFVAR_PRIV_VAL:%.+]] = load float, float* [[SFVAR_PRIVATE]], 131 // LAMBDA: store float [[SFVAR_PRIV_VAL]], float* [[SFVAR_IN_REF]], 132 // LAMBDA: br label %[[OMP_LASTPRIV_DONE]] 133 // LAMBDA: [[OMP_LASTPRIV_DONE]]: 134 // LAMBDA: ret 135 136 // LAMBDA: define{{.*}} internal{{.*}} void @[[LPAR_OUTL]](i32* noalias %{{.+}}, i32* noalias %{{.+}}, {{.+}}, {{.+}}, double*{{.+}} [[G_IN:%.+]], double*{{.+}} [[G1_IN:%.+]], i{{[0-9]+}}*{{.+}} [[SVAR_IN:%.+]], float*{{.+}} [[SFVAR_IN:%.+]]) 137 // LAMBDA: [[G_PRIVATE_ADDR:%.+]] = alloca double*, 138 // LAMBDA: [[G1_PRIVATE_ADDR:%.+]] = alloca double*, 139 // LAMBDA: [[SVAR_PRIVATE_ADDR:%.+]] = alloca i{{[0-9]+}}*, 140 // LAMBDA: [[SFVAR_PRIVATE_ADDR:%.+]] = alloca float*, 141 // loop variables 142 // LAMBDA: {{.+}} = alloca i{{[0-9]+}}, 143 // LAMBDA: {{.+}} = alloca i{{[0-9]+}}, 144 // LAMBDA: {{.+}} = alloca i{{[0-9]+}}, 145 // LAMBDA: {{.+}} = alloca i{{[0-9]+}}, 146 // LAMBDA: {{.+}} = alloca i{{[0-9]+}}, 147 // LAMBDA: [[OMP_IS_LAST:%.+]] = alloca i{{[0-9]+}}, 148 // LAMBDA: [[G_PRIVATE:%.+]] = alloca double, 149 // LAMBDA: [[G1_PRIVATE:%.+]] = alloca double, 150 // LAMBDA: [[TMP_G1_PRIVATE:%.+]] = alloca double*, 151 // LAMBDA: [[SVAR_PRIVATE:%.+]] = alloca i{{[0-9]+}}, 152 // LAMBDA: [[SFVAR_PRIVATE:%.+]] = alloca float, 153 // LAMBDA: store double* [[G_IN]], double** [[G_PRIVATE_ADDR]], 154 // LAMBDA: store double* [[G1_IN]], double** [[G1_PRIVATE_ADDR]], 155 // LAMBDA: store i{{[0-9]+}}* [[SVAR_IN]], i{{[0-9]+}}** [[SVAR_PRIVATE_ADDR]], 156 // LAMBDA: store float* [[SFVAR_IN]], float** [[SFVAR_PRIVATE_ADDR]], 157 158 // init private variables 159 // LAMBDA: [[G_IN_REF:%.+]] = load double*, double** [[G_PRIVATE_ADDR]], 160 // LAMBDA: [[G1_IN_REF:%.+]] = load double*, double** [[G1_PRIVATE_ADDR]], 161 // LAMBDA: [[SVAR_IN_REF:%.+]] = load i{{[0-9]+}}*, i{{[0-9]+}}** [[SVAR_PRIVATE_ADDR]], 162 // LAMBDA: [[SFVAR_IN_REF:%.+]] = load float*, float** [[SFVAR_PRIVATE_ADDR]], 163 // LAMBDA: [[G1_IN_REF:%.+]] = load double*, double** % 164 // LAMBDA: store double* [[G1_PRIVATE]], double** [[TMP_G1_PRIVATE]], 165 166 // LAMBDA: call {{.*}}void @__kmpc_for_static_init_4( 167 // LAMBDA: store double 1.0{{.+}}, double* [[G_PRIVATE]], 168 // LAMBDA: [[TMP_G1_REF:%.+]] = load double*, double** [[TMP_G1_PRIVATE]], 169 // LAMBDA: store{{.+}} double 1.0{{.+}}, double* [[TMP_G1_REF]], 170 // LAMBDA: store i{{[0-9]+}} 3, i{{[0-9]+}}* [[SVAR_PRIVATE]], 171 // LAMBDA: store float 4.0{{.+}}, float* [[SFVAR_PRIVATE]], 172 // LAMBDA: [[G_PRIVATE_ADDR_REF:%.+]] = getelementptr inbounds %{{.+}}, %{{.+}}* [[ARG:%.+]], i{{[0-9]+}} 0, i{{[0-9]+}} 0 173 // LAMBDA: store double* [[G_PRIVATE]], double** [[G_PRIVATE_ADDR_REF]], 174 // LAMBDA: [[TMP_PRIVATE_ADDR_REF:%.+]] = getelementptr inbounds %{{.+}}, %{{.+}}* [[ARG:%.+]], i{{[0-9]+}} 0, i{{[0-9]+}} 1 175 // LAMBDA: [[G1_PRIVATE_ADDR_FROM_TMP:%.+]] = load double*, double** [[TMP_G1_PRIVATE]], 176 // LAMBDA: store double* [[G1_PRIVATE_ADDR_FROM_TMP]], double** [[TMP_PRIVATE_ADDR_REF]], 177 // LAMBDA: [[SVAR_PRIVATE_ADDR_REF:%.+]] = getelementptr inbounds %{{.+}}, %{{.+}}* [[ARG:%.+]], i{{[0-9]+}} 0, i{{[0-9]+}} 2 178 // LAMBDA: store i{{[0-9]+}}* [[SVAR_PRIVATE]], i{{[0-9]+}}** [[SVAR_PRIVATE_ADDR_REF]] 179 // LAMBDA: [[SFVAR_PRIVATE_ADDR_REF:%.+]] = getelementptr inbounds %{{.+}}, %{{.+}}* [[ARG:%.+]], i{{[0-9]+}} 0, i{{[0-9]+}} 3 180 // LAMBDA: store float* [[SFVAR_PRIVATE]], float** [[SFVAR_PRIVATE_ADDR_REF]] 181 // LAMBDA: call{{.*}} void [[INNER_LAMBDA:@.+]]({{.+}} [[ARG]]) 182 // LAMBDA: call {{.*}}void @__kmpc_for_static_fini( 183 184 // LAMBDA: [[OMP_IS_LAST_VAL:%.+]] = load i{{[0-9]+}}, i{{[0-9]+}}* [[OMP_IS_LAST]], 185 // LAMBDA: [[IS_LAST_IT:%.+]] = icmp ne i{{[0-9]+}} [[OMP_IS_LAST_VAL]], 0 186 // LAMBDA: br i1 [[IS_LAST_IT]], label %[[OMP_LASTPRIV_BLOCK:.+]], label %[[OMP_LASTPRIV_DONE:.+]] 187 188 // LAMBDA: [[OMP_LASTPRIV_BLOCK]]: 189 // LAMBDA: [[G_PRIV_VAL:%.+]] = load double, double* [[G_PRIVATE]], 190 // LAMBDA: store{{.*}} double [[G_PRIV_VAL]], double* [[G_IN_REF]], 191 // LAMBDA: [[TMP_G1_PRIV_REF:%.+]] = load double*, double** [[TMP_G1_PRIVATE]], 192 // LAMBDA: [[TMP_G1_PRIV_VAL:%.+]] = load double, double* [[TMP_G1_PRIV_REF]], 193 // LAMBDA: store{{.*}} double [[TMP_G1_PRIV_VAL]], double* [[G1_IN_REF]], 194 195 // LAMBDA: [[SVAR_PRIV_VAL:%.+]] = load i{{[0-9]+}}, i{{[0-9]+}}* [[SVAR_PRIVATE]], 196 // LAMBDA: store i{{[0-9]+}} [[SVAR_PRIV_VAL]], i{{[0-9]+}}* [[SVAR_IN_REF]], 197 // LAMBDA: [[SFVAR_PRIV_VAL:%.+]] = load float, float* [[SFVAR_PRIVATE]], 198 // LAMBDA: store float [[SFVAR_PRIV_VAL]], float* [[SFVAR_IN_REF]], 199 // LAMBDA: br label %[[OMP_LASTPRIV_DONE]] 200 // LAMBDA: [[OMP_LASTPRIV_DONE]]: 201 // LAMBDA: ret 202 203 [&]() { 204 // LAMBDA: define {{.+}} void [[INNER_LAMBDA]]({{.+}} [[ARG_PTR:%.+]]) 205 // LAMBDA: store %{{.+}}* [[ARG_PTR]], %{{.+}}** [[ARG_PTR_REF:%.+]], 206 g = 2; 207 g1 = 2; 208 svar = 4; 209 sfvar = 8.0; 210 // LAMBDA: [[ARG_PTR:%.+]] = load %{{.+}}*, %{{.+}}** [[ARG_PTR_REF]] 211 // LAMBDA: [[G_PTR_REF:%.+]] = getelementptr inbounds %{{.+}}, %{{.+}}* [[ARG_PTR]], i{{[0-9]+}} 0, i{{[0-9]+}} 0 212 // LAMBDA: [[G_REF:%.+]] = load double*, double** [[G_PTR_REF]] 213 // LAMBDA: store double 2.0{{.+}}, double* [[G_REF]] 214 215 // LAMBDA: [[TMP_PTR_REF:%.+]] = getelementptr inbounds %{{.+}}, %{{.+}}* [[ARG_PTR]], i{{[0-9]+}} 0, i{{[0-9]+}} 1 216 // LAMBDA: [[G1_REF:%.+]] = load double*, double** [[TMP_PTR_REF]] 217 // LAMBDA: store double 2.0{{.+}}, double* [[G1_REF]], 218 // LAMBDA: [[SVAR_PTR_REF:%.+]] = getelementptr inbounds %{{.+}}, %{{.+}}* [[ARG_PTR]], i{{[0-9]+}} 0, i{{[0-9]+}} 2 219 // LAMBDA: [[SVAR_REF:%.+]] = load i{{[0-9]+}}*, i{{[0-9]+}}** [[SVAR_PTR_REF]] 220 // LAMBDA: store i{{[0-9]+}} 4, i{{[0-9]+}}* [[SVAR_REF]] 221 // LAMBDA: [[SFVAR_PTR_REF:%.+]] = getelementptr inbounds %{{.+}}, %{{.+}}* [[ARG_PTR]], i{{[0-9]+}} 0, i{{[0-9]+}} 3 222 // LAMBDA: [[SFVAR_REF:%.+]] = load float*, float** [[SFVAR_PTR_REF]] 223 // LAMBDA: store float 8.0{{.+}}, float* [[SFVAR_REF]] 224 }(); 225 } 226 }(); 227 return 0; 228 #else 229 S<float> test; 230 int t_var = 0; 231 int vec[] = {1, 2}; 232 S<float> s_arr[] = {1, 2}; 233 S<float> &var = test; 234 235 #pragma omp target 236 #pragma omp teams distribute parallel for lastprivate(t_var, vec, s_arr, s_arr, var, var, svar) 237 for (int i = 0; i < 2; ++i) { 238 vec[i] = t_var; 239 s_arr[i] = var; 240 } 241 int i; 242 243 return tmain<int>(); 244 #endif 245 } 246 247 // CHECK: define{{.*}} i{{[0-9]+}} @main() 248 // CHECK: [[TEST:%.+]] = alloca [[S_FLOAT_TY]], 249 // CHECK: call {{.*}} [[S_FLOAT_TY_DEF_CONSTR:@.+]]([[S_FLOAT_TY]]* {{[^,]*}} [[TEST]]) 250 // CHECK: call i{{[0-9]+}} @__tgt_target_teams_mapper(%struct.ident_t* @{{.+}}, 251 // CHECK: call void [[OFFLOAD_FUN:@.+]](i{{[0-9]+}} {{.+}}, [2 x i{{[0-9]+}}]* {{.+}}, [2 x [[S_FLOAT_TY]]]* {{.+}}, [[S_FLOAT_TY]]* {{.+}}, i{{[0-9]+}} {{.+}}) 252 // CHECK: ret 253 254 // CHECK: define{{.+}} [[OFFLOAD_FUN]](i{{[0-9]+}} {{.+}}, [2 x i{{[0-9]+}}]*{{.+}} {{.+}}, [2 x [[S_FLOAT_TY]]]*{{.+}} {{.+}}, [[S_FLOAT_TY]]*{{.+}} {{.+}}, i{{[0-9]+}} {{.+}}) 255 // CHECK: call void (%{{.+}}*, i{{[0-9]+}}, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)*, ...) @__kmpc_fork_teams( 256 // CHECK: ret 257 // 258 // CHECK: define internal void [[OMP_OUTLINED:@.+]](i{{[0-9]+}}* noalias [[GTID_ADDR:%.+]], i{{[0-9]+}}* noalias %{{.+}}, [2 x i{{[0-9]+}}]*{{.+}} [[VEC_IN:%.+]], i{{[0-9]+}}*{{.+}} [[T_VAR_IN:%.+]], [2 x [[S_FLOAT_TY]]]*{{.+}} [[S_ARR_IN:%.+]], [[S_FLOAT_TY]]*{{.+}} [[VAR_IN:%.+]], i{{[0-9]+}}*{{.*}} [[S_VAR_IN:%.+]]) 259 // CHECK: {{.+}} = alloca i{{[0-9]+}}*, 260 // CHECK: {{.+}} = alloca i{{[0-9]+}}*, 261 // CHECK: [[VEC_ADDR:%.+]] = alloca [2 x i{{[0-9]+}}]*, 262 // CHECK: [[T_VAR_ADDR:%.+]] = alloca i{{[0-9]+}}*, 263 // CHECK: [[S_ARR_ADDR:%.+]] = alloca [2 x [[S_FLOAT_TY]]]*, 264 // CHECK: [[VAR_ADDR:%.+]] = alloca [[S_FLOAT_TY]]*, 265 // CHECK: [[SVAR_ADDR:%.+]] = alloca i{{[0-9]+}}*, 266 // skip loop variables 267 // CHECK: {{.+}} = alloca i{{[0-9]+}}, 268 // CHECK: {{.+}} = alloca i{{[0-9]+}}, 269 // CHECK: {{.+}} = alloca i{{[0-9]+}}, 270 // CHECK: {{.+}} = alloca i{{[0-9]+}}, 271 // CHECK: {{.+}} = alloca i{{[0-9]+}}, 272 // CHECK: [[OMP_IS_LAST:%.+]] = alloca i{{[0-9]+}}, 273 // CHECK: [[T_VAR_PRIV:%.+]] = alloca i{{[0-9]+}}, 274 // CHECK: [[VEC_PRIV:%.+]] = alloca [2 x i{{[0-9]+}}], 275 // CHECK: [[S_ARR_PRIV:%.+]] = alloca [2 x [[S_FLOAT_TY]]], 276 // CHECK: [[VAR_PRIV:%.+]] = alloca [[S_FLOAT_TY]], 277 // CHECK: [[TMP_PRIV:%.+]] = alloca [[S_FLOAT_TY]]*, 278 // CHECK: [[S_VAR_PRIV:%.+]] = alloca i{{[0-9]+}}, 279 280 // copy from parameters to local address variables 281 // CHECK: store [2 x i{{[0-9]+}}]* [[VEC_IN]], [2 x i{{[0-9]+}}]** [[VEC_ADDR]], 282 // CHECK: store i{{[0-9]+}}* [[T_VAR_IN]], i{{[0-9]+}}** [[T_VAR_ADDR]], 283 // CHECK: store [2 x [[S_FLOAT_TY]]]* [[S_ARR_IN]], [2 x [[S_FLOAT_TY]]]** [[S_ARR_ADDR]], 284 // CHECK: store [[S_FLOAT_TY]]* [[VAR_IN]], [[S_FLOAT_TY]]** [[VAR_ADDR]], 285 // CHECK: store i{{[0-9]+}}* [[S_VAR_IN]], i{{[0-9]+}}** [[SVAR_ADDR]], 286 287 // load content of local address variables 288 // CHECK: [[VEC_ADDR_REF:%.+]] = load [2 x i{{[0-9]+}}]*, [2 x i{{[0-9]+}}]** [[VEC_ADDR]], 289 // CHECK: [[T_VAR_ADDR_REF:%.+]] = load i{{[0-9]+}}*, i{{[0-9]+}}** [[T_VAR_ADDR]], 290 // CHECK: [[S_ARR_ADDR_REF:%.+]] = load [2 x [[S_FLOAT_TY]]]*, [2 x [[S_FLOAT_TY]]]** [[S_ARR_ADDR]], 291 // CHECK: [[VAR_ADDR_REF:%.+]] = load {{.+}}, {{.+}} [[VAR_ADDR]], 292 // CHECK: [[SVAR_ADDR_REF:%.+]] = load i{{[0-9]+}}*, i{{[0-9]+}}** [[SVAR_ADDR]], 293 // CHECK: store i{{[0-9]+}} 0, i{{[0-9]+}}* [[OMP_IS_LAST]], 294 // CHECK: [[VAR_ADDR_REF:%.+]] = load [[S_FLOAT_TY]]*, [[S_FLOAT_TY]]** % 295 // the distribute loop 296 // CHECK: call void @__kmpc_for_static_init_4( 297 // CHECK: call void {{.*}} @__kmpc_fork_call({{.+}}, {{.+}}, {{.+}} @[[PAR_OUTL:.+]] to 298 // CHECK: call void @__kmpc_for_static_fini( 299 300 // lastprivates 301 // CHECK: [[OMP_IS_LAST_VAL:%.+]] = load i{{[0-9]+}}, i{{[0-9]+}}* [[OMP_IS_LAST]], 302 // CHECK: [[IS_LAST_IT:%.+]] = icmp ne i{{[0-9]+}} [[OMP_IS_LAST_VAL]], 0 303 // CHECK: br i1 [[IS_LAST_IT]], label %[[OMP_LASTPRIV_BLOCK:.+]], label %[[OMP_LASTPRIV_DONE:.+]] 304 305 // CHECK: [[OMP_LASTPRIV_BLOCK]]: 306 // CHECK: [[T_VAR_VAL:%.+]] = load i{{[0-9]+}}, i{{[0-9]+}}* [[T_VAR_PRIV]], 307 // CHECK: store i{{[0-9]+}} [[T_VAR_VAL]], i{{[0-9]+}}* [[T_VAR_ADDR_REF]], 308 // CHECK: [[BCAST_VEC_ADDR_REF:%.+]] = bitcast [2 x i{{[0-9]+}}]* [[VEC_ADDR_REF]] to i8* 309 // CHECK: [[BCAST_VEC_PRIV:%.+]] = bitcast [2 x i{{[0-9]+}}]* [[VEC_PRIV]] to i8* 310 // CHECK: call void @llvm.memcpy.{{.+}}(i8* align {{[0-9]+}} [[BCAST_VEC_ADDR_REF]], i8* align {{[0-9]+}} [[BCAST_VEC_PRIV]], 311 // CHECK: [[S_ARR_BEGIN:%.+]] = getelementptr inbounds [2 x [[S_FLOAT_TY]]], [2 x [[S_FLOAT_TY]]]* [[S_ARR_ADDR_REF]], i{{[0-9]+}} 0, i{{[0-9]+}} 0 312 // CHECK: [[S_ARR_PRIV_BCAST:%.+]] = bitcast [2 x [[S_FLOAT_TY]]]* [[S_ARR_PRIV]] to [[S_FLOAT_TY]]* 313 // CHECK: [[S_ARR_BEGIN_GEP:%.+]] = getelementptr [[S_FLOAT_TY]], [[S_FLOAT_TY]]* [[S_ARR_BEGIN]], i{{[0-9]+}} 2 314 // CHECK: [[S_ARR_IS_EMPTY:%.+]] = icmp eq [[S_FLOAT_TY]]* [[S_ARR_BEGIN]], [[S_ARR_BEGIN_GEP]] 315 // CHECK: br i1 [[S_ARR_IS_EMPTY]], label %[[S_ARR_COPY_DONE:.+]], label %[[S_ARR_COPY_BLOCK:.+]] 316 // CHECK: [[S_ARR_COPY_BLOCK]]: 317 // CHECK: [[S_ARR_SRC_EL:%.+]] = phi [[S_FLOAT_TY]]*{{.+}} 318 // CHECK: [[S_ARR_DST_EL:%.+]] = phi [[S_FLOAT_TY]]*{{.+}} 319 // CHECK: [[S_ARR_DST_BCAST:%.+]] = bitcast [[S_FLOAT_TY]]* [[S_ARR_DST_EL]] to i8* 320 // CHECK: [[S_ARR_SRC_BCAST:%.+]] = bitcast [[S_FLOAT_TY]]* [[S_ARR_SRC_EL]] to i8* 321 // CHECK: call void @llvm.memcpy.{{.+}}(i8* align {{[0-9]+}} [[S_ARR_DST_BCAST]], i8* align {{[0-9]+}} [[S_ARR_SRC_BCAST]]{{.+}}) 322 // CHECK: [[S_ARR_DST_NEXT:%.+]] = getelementptr [[S_FLOAT_TY]], [[S_FLOAT_TY]]* [[S_ARR_DST_EL]], i{{[0-9]+}} 1 323 // CHECK: [[S_ARR_SRC_NEXT:%.+]] = getelementptr{{.+}} 324 // CHECK: [[CPY_IS_FINISHED:%.+]] = icmp eq [[S_FLOAT_TY]]* [[S_ARR_DST_NEXT]], [[S_ARR_BEGIN_GEP]] 325 // CHECK: br i1 [[CPY_IS_FINISHED]], label %[[S_ARR_COPY_DONE]], label %[[S_ARR_COPY_BLOCK]] 326 // CHECK: [[S_ARR_COPY_DONE]]: 327 // CHECK: [[TMP_VAL1:%.+]] = load [[S_FLOAT_TY]]*, [[S_FLOAT_TY]]** [[TMP_PRIV]], 328 // CHECK: [[VAR_ADDR_REF_BCAST:%.+]] = bitcast [[S_FLOAT_TY]]* [[VAR_ADDR_REF]] to i8* 329 // CHECK: [[TMP_VAL1_BCAST:%.+]] = bitcast [[S_FLOAT_TY]]* [[TMP_VAL1]] to i8* 330 // CHECK: call void @llvm.memcpy.{{.+}}(i8* align {{[0-9]+}} [[VAR_ADDR_REF_BCAST]], i8* align {{[0-9]+}} [[TMP_VAL1_BCAST]],{{.+}}) 331 // CHECK: [[SVAR_VAL:%.+]] = load i{{[0-9]+}}, i{{[0-9]+}}* [[S_VAR_PRIV]], 332 // CHECK: store i{{[0-9]+}} [[SVAR_VAL]], i{{[0-9]+}}* [[SVAR_ADDR_REF]], 333 // CHECK: ret void 334 335 // CHECK: define internal void [[OMP_OUTLINED:@.+]](i{{[0-9]+}}* noalias [[GTID_ADDR:%.+]], i{{[0-9]+}}* noalias %{{.+}}, {{.+}}, {{.+}}, [2 x i{{[0-9]+}}]*{{.+}} [[VEC_IN:%.+]], i{{[0-9]+}}*{{.+}} [[T_VAR_IN:%.+]], [2 x [[S_FLOAT_TY]]]*{{.+}} [[S_ARR_IN:%.+]], [[S_FLOAT_TY]]*{{.+}} [[VAR_IN:%.+]], i{{[0-9]+}}*{{.*}} [[S_VAR_IN:%.+]]) 336 // gbl and bound tid vars, prev lb and ub vars 337 // CHECK: {{.+}} = alloca i{{[0-9]+}}*, 338 // CHECK: {{.+}} = alloca i{{[0-9]+}}*, 339 // CHECK: alloca i{{[0-9]+}}, 340 // CHECK: alloca i{{[0-9]+}}, 341 // CHECK: [[VEC_ADDR:%.+]] = alloca [2 x i{{[0-9]+}}]*, 342 // CHECK: [[T_VAR_ADDR:%.+]] = alloca i{{[0-9]+}}*, 343 // CHECK: [[S_ARR_ADDR:%.+]] = alloca [2 x [[S_FLOAT_TY]]]*, 344 // CHECK: [[VAR_ADDR:%.+]] = alloca [[S_FLOAT_TY]]*, 345 // CHECK: [[SVAR_ADDR:%.+]] = alloca i{{[0-9]+}}*, 346 // skip loop variables 347 // CHECK: {{.+}} = alloca i{{[0-9]+}}, 348 // CHECK: {{.+}} = alloca i{{[0-9]+}}, 349 // CHECK: {{.+}} = alloca i{{[0-9]+}}, 350 // CHECK: {{.+}} = alloca i{{[0-9]+}}, 351 // CHECK: {{.+}} = alloca i{{[0-9]+}}, 352 // CHECK: [[OMP_IS_LAST:%.+]] = alloca i{{[0-9]+}}, 353 // CHECK: [[T_VAR_PRIV:%.+]] = alloca i{{[0-9]+}}, 354 // CHECK: [[VEC_PRIV:%.+]] = alloca [2 x i{{[0-9]+}}], 355 // CHECK: [[S_ARR_PRIV:%.+]] = alloca [2 x [[S_FLOAT_TY]]], 356 // CHECK: [[VAR_PRIV:%.+]] = alloca [[S_FLOAT_TY]], 357 // CHECK: [[TMP_PRIV:%.+]] = alloca [[S_FLOAT_TY]]*, 358 // CHECK: [[S_VAR_PRIV:%.+]] = alloca i{{[0-9]+}}, 359 360 // copy from parameters to local address variables 361 // CHECK: store [2 x i{{[0-9]+}}]* [[VEC_IN]], [2 x i{{[0-9]+}}]** [[VEC_ADDR]], 362 // CHECK: store i{{[0-9]+}}* [[T_VAR_IN]], i{{[0-9]+}}** [[T_VAR_ADDR]], 363 // CHECK: store [2 x [[S_FLOAT_TY]]]* [[S_ARR_IN]], [2 x [[S_FLOAT_TY]]]** [[S_ARR_ADDR]], 364 // CHECK: store [[S_FLOAT_TY]]* [[VAR_IN]], [[S_FLOAT_TY]]** [[VAR_ADDR]], 365 // CHECK: store i{{[0-9]+}}* [[S_VAR_IN]], i{{[0-9]+}}** [[SVAR_ADDR]], 366 367 // load content of local address variables 368 // CHECK: [[VEC_ADDR_REF:%.+]] = load [2 x i{{[0-9]+}}]*, [2 x i{{[0-9]+}}]** [[VEC_ADDR]], 369 // CHECK: [[T_VAR_ADDR_REF:%.+]] = load i{{[0-9]+}}*, i{{[0-9]+}}** [[T_VAR_ADDR]], 370 // CHECK: [[S_ARR_ADDR_REF:%.+]] = load [2 x [[S_FLOAT_TY]]]*, [2 x [[S_FLOAT_TY]]]** [[S_ARR_ADDR]], 371 // CHECK: [[VAR_ADDR_REF:%.+]] = load {{.+}}, {{.+}} [[VAR_ADDR]], 372 // CHECK: [[SVAR_ADDR_REF:%.+]] = load i{{[0-9]+}}*, i{{[0-9]+}}** [[SVAR_ADDR]], 373 // CHECK: store i{{[0-9]+}} 0, i{{[0-9]+}}* [[OMP_IS_LAST]], 374 // CHECK: [[VAR_ADDR_REF:%.+]] = load [[S_FLOAT_TY]]*, [[S_FLOAT_TY]]** % 375 // the distribute loop 376 // CHECK: call void @__kmpc_for_static_init_4( 377 378 // assignment: vec[i] = t_var; 379 // CHECK: [[T_VAR_PRIV_VAL:%.+]] = load i{{[0-9]+}}, i{{[0-9]+}}* [[T_VAR_PRIV]], 380 // CHECK: [[VEC_PTR:%.+]] = getelementptr inbounds [2 x i{{[0-9]+}}], [2 x i{{[0-9]+}}]* [[VEC_PRIV]], i{{[0-9]+}} 0, i{{[0-9]+}} {{.+}} 381 // CHECK: store i{{[0-9]+}} [[T_VAR_PRIV_VAL]], i{{[0-9]+}}* [[VEC_PTR]], 382 383 // assignment: s_arr[i] = var; 384 // CHECK-DAG: [[S_ARR_PTR:%.+]] = getelementptr inbounds [2 x [[S_FLOAT_TY]]], [2 x [[S_FLOAT_TY]]]* [[S_ARR_PRIV]], 385 // CHECK-DAG: [[TMP_VAL:%.+]] = load [[S_FLOAT_TY]]*, [[S_FLOAT_TY]]** [[TMP_PRIV]], 386 // CHECK-DAG: [[S_ARR_PTR_BCAST:%.+]] = bitcast [[S_FLOAT_TY]]* [[S_ARR_PTR]] to i8* 387 // CHECK-DAG: [[TMP_VAL_BCAST:%.+]] = bitcast [[S_FLOAT_TY]]* [[TMP_VAL]] to i8* 388 // CHECK: call void @llvm.memcpy.{{.+}}(i8* align {{[0-9]+}} [[S_ARR_PTR_BCAST]], i8* align {{[0-9]+}} [[TMP_VAL_BCAST]], 389 390 // CHECK: call void @__kmpc_for_static_fini( 391 392 // lastprivates 393 // CHECK: [[OMP_IS_LAST_VAL:%.+]] = load i{{[0-9]+}}, i{{[0-9]+}}* [[OMP_IS_LAST]], 394 // CHECK: [[IS_LAST_IT:%.+]] = icmp ne i{{[0-9]+}} [[OMP_IS_LAST_VAL]], 0 395 // CHECK: br i1 [[IS_LAST_IT]], label %[[OMP_LASTPRIV_BLOCK:.+]], label %[[OMP_LASTPRIV_DONE:.+]] 396 397 // CHECK: [[OMP_LASTPRIV_BLOCK]]: 398 // CHECK: [[T_VAR_VAL:%.+]] = load i{{[0-9]+}}, i{{[0-9]+}}* [[T_VAR_PRIV]], 399 // CHECK: store i{{[0-9]+}} [[T_VAR_VAL]], i{{[0-9]+}}* [[T_VAR_ADDR_REF]], 400 // CHECK: [[BCAST_VEC_ADDR_REF:%.+]] = bitcast [2 x i{{[0-9]+}}]* [[VEC_ADDR_REF]] to i8* 401 // CHECK: [[BCAST_VEC_PRIV:%.+]] = bitcast [2 x i{{[0-9]+}}]* [[VEC_PRIV]] to i8* 402 // CHECK: call void @llvm.memcpy.{{.+}}(i8* align {{[0-9]+}} [[BCAST_VEC_ADDR_REF]], i8* align {{[0-9]+}} [[BCAST_VEC_PRIV]], 403 // CHECK: [[S_ARR_BEGIN:%.+]] = getelementptr inbounds [2 x [[S_FLOAT_TY]]], [2 x [[S_FLOAT_TY]]]* [[S_ARR_ADDR_REF]], i{{[0-9]+}} 0, i{{[0-9]+}} 0 404 // CHECK: [[S_ARR_PRIV_BCAST:%.+]] = bitcast [2 x [[S_FLOAT_TY]]]* [[S_ARR_PRIV]] to [[S_FLOAT_TY]]* 405 // CHECK: [[S_ARR_BEGIN_GEP:%.+]] = getelementptr [[S_FLOAT_TY]], [[S_FLOAT_TY]]* [[S_ARR_BEGIN]], i{{[0-9]+}} 2 406 // CHECK: [[S_ARR_IS_EMPTY:%.+]] = icmp eq [[S_FLOAT_TY]]* [[S_ARR_BEGIN]], [[S_ARR_BEGIN_GEP]] 407 // CHECK: br i1 [[S_ARR_IS_EMPTY]], label %[[S_ARR_COPY_DONE:.+]], label %[[S_ARR_COPY_BLOCK:.+]] 408 // CHECK: [[S_ARR_COPY_BLOCK]]: 409 // CHECK: [[S_ARR_SRC_EL:%.+]] = phi [[S_FLOAT_TY]]*{{.+}} 410 // CHECK: [[S_ARR_DST_EL:%.+]] = phi [[S_FLOAT_TY]]*{{.+}} 411 // CHECK: [[S_ARR_DST_BCAST:%.+]] = bitcast [[S_FLOAT_TY]]* [[S_ARR_DST_EL]] to i8* 412 // CHECK: [[S_ARR_SRC_BCAST:%.+]] = bitcast [[S_FLOAT_TY]]* [[S_ARR_SRC_EL]] to i8* 413 // CHECK: call void @llvm.memcpy.{{.+}}(i8* align {{[0-9]+}} [[S_ARR_DST_BCAST]], i8* align {{[0-9]+}} [[S_ARR_SRC_BCAST]]{{.+}}) 414 // CHECK: [[S_ARR_DST_NEXT:%.+]] = getelementptr [[S_FLOAT_TY]], [[S_FLOAT_TY]]* [[S_ARR_DST_EL]], i{{[0-9]+}} 1 415 // CHECK: [[S_ARR_SRC_NEXT:%.+]] = getelementptr{{.+}} 416 // CHECK: [[CPY_IS_FINISHED:%.+]] = icmp eq [[S_FLOAT_TY]]* [[S_ARR_DST_NEXT]], [[S_ARR_BEGIN_GEP]] 417 // CHECK: br i1 [[CPY_IS_FINISHED]], label %[[S_ARR_COPY_DONE]], label %[[S_ARR_COPY_BLOCK]] 418 // CHECK: [[S_ARR_COPY_DONE]]: 419 // CHECK: [[TMP_VAL1:%.+]] = load [[S_FLOAT_TY]]*, [[S_FLOAT_TY]]** [[TMP_PRIV]], 420 // CHECK: [[VAR_ADDR_REF_BCAST:%.+]] = bitcast [[S_FLOAT_TY]]* [[VAR_ADDR_REF]] to i8* 421 // CHECK: [[TMP_VAL1_BCAST:%.+]] = bitcast [[S_FLOAT_TY]]* [[TMP_VAL1]] to i8* 422 // CHECK: call void @llvm.memcpy.{{.+}}(i8* align {{[0-9]+}} [[VAR_ADDR_REF_BCAST]], i8* align {{[0-9]+}} [[TMP_VAL1_BCAST]],{{.+}}) 423 // CHECK: [[SVAR_VAL:%.+]] = load i{{[0-9]+}}, i{{[0-9]+}}* [[S_VAR_PRIV]], 424 // CHECK: store i{{[0-9]+}} [[SVAR_VAL]], i{{[0-9]+}}* [[SVAR_ADDR_REF]], 425 // CHECK: ret void 426 427 // template tmain 428 // CHECK: define{{.*}} i{{[0-9]+}} [[TMAIN_INT:@.+]]() 429 // CHECK: [[TEST:%.+]] = alloca [[S_INT_TY]], 430 // CHECK: call {{.*}} [[S_INT_TY_DEF_CONSTR:@.+]]([[S_INT_TY]]* {{[^,]*}} [[TEST]]) 431 // CHECK: call i{{[0-9]+}} @__tgt_target_teams_mapper(%struct.ident_t* @{{.+}}, 432 // CHECK: call void [[OFFLOAD_FUN_1:@.+]](i{{[0-9]+}} {{.+}}, [2 x i{{[0-9]+}}]* {{.+}}, [2 x [[S_INT_TY]]]* {{.+}}, [[S_INT_TY]]* {{.+}}) 433 // CHECK: ret 434 435 // CHECK: define internal void [[OFFLOAD_FUN_1]]( 436 // CHECK: call void (%{{.+}}*, i{{[0-9]+}}, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)*, ...) @__kmpc_fork_teams(%{{.+}}* @{{.+}}, i{{[0-9]+}} 4, 437 // CHECK: ret 438 439 // CHECK: define internal void [[OMP_OUTLINED_1:@.+]](i{{[0-9]+}}* noalias [[GTID_ADDR1:%.+]], i{{[0-9]+}}* noalias %{{.+}}, [2 x i{{[0-9]+}}]*{{.+}} [[VEC_IN1:%.+]], i{{[0-9]+}}*{{.+}} [[T_VAR_IN1:%.+]], [2 x [[S_INT_TY]]]*{{.+}} [[S_ARR_IN1:%.+]], [[S_INT_TY]]*{{.+}} [[VAR_IN1:%.+]]) 440 // skip alloca of global_tid and bound_tid 441 // CHECK: {{.+}} = alloca i{{[0-9]+}}*, 442 // CHECK: {{.+}} = alloca i{{[0-9]+}}*, 443 // CHECK: [[VEC_ADDR1:%.+]] = alloca [2 x i{{[0-9]+}}]*, 444 // CHECK: [[T_VAR_ADDR1:%.+]] = alloca i{{[0-9]+}}*, 445 // CHECK: [[S_ARR_ADDR1:%.+]] = alloca [2 x [[S_INT_TY]]]*, 446 // CHECK: [[VAR_ADDR1:%.+]] = alloca [[S_INT_TY]]*, 447 // skip loop variables 448 // CHECK: {{.+}} = alloca i{{[0-9]+}}, 449 // CHECK: {{.+}} = alloca i{{[0-9]+}}, 450 // CHECK: {{.+}} = alloca i{{[0-9]+}}, 451 // CHECK: {{.+}} = alloca i{{[0-9]+}}, 452 // CHECK: {{.+}} = alloca i{{[0-9]+}}, 453 // CHECK: [[OMP_IS_LAST1:%.+]] = alloca i{{[0-9]+}}, 454 // CHECK: [[T_VAR_PRIV1:%.+]] = alloca i{{[0-9]+}}, 455 // CHECK: [[VEC_PRIV1:%.+]] = alloca [2 x i{{[0-9]+}}], 456 // CHECK: [[S_ARR_PRIV1:%.+]] = alloca [2 x [[S_INT_TY]]], 457 // CHECK: [[VAR_PRIV1:%.+]] = alloca [[S_INT_TY]], 458 // CHECK: [[TMP_PRIV1:%.+]] = alloca [[S_INT_TY]]*, 459 460 // skip init of bound and global tid 461 // CHECK: store i{{[0-9]+}}* {{.*}}, 462 // CHECK: store i{{[0-9]+}}* {{.*}}, 463 // copy from parameters to local address variables 464 // CHECK: store [2 x i{{[0-9]+}}]* [[VEC_IN1]], [2 x i{{[0-9]+}}]** [[VEC_ADDR1]], 465 // CHECK: store i{{[0-9]+}}* [[T_VAR_IN1]], i{{[0-9]+}}** [[T_VAR_ADDR1]], 466 // CHECK: store [2 x [[S_INT_TY]]]* [[S_ARR_IN1]], [2 x [[S_INT_TY]]]** [[S_ARR_ADDR1]], 467 // CHECK: store [[S_INT_TY]]* [[VAR_IN1]], [[S_INT_TY]]** [[VAR_ADDR1]], 468 469 // load content of local address variables 470 // CHECK: [[VEC_ADDR_REF1:%.+]] = load [2 x i{{[0-9]+}}]*, [2 x i{{[0-9]+}}]** [[VEC_ADDR1]], 471 // CHECK: [[T_VAR_ADDR_REF1:%.+]] = load i{{[0-9]+}}*, i{{[0-9]+}}** [[T_VAR_ADDR1]], 472 // CHECK: [[S_ARR_ADDR_REF1:%.+]] = load [2 x [[S_INT_TY]]]*, [2 x [[S_INT_TY]]]** [[S_ARR_ADDR1]], 473 // CHECK: [[VAR_ADDR1_REF:%.+]] = load [[S_INT_TY]]*, [[S_INT_TY]]** [[VAR_ADDR1]], 474 // CHECK-DAG: store i{{[0-9]+}} 0, i{{[0-9]+}}* [[OMP_IS_LAST1]], 475 // CHECK: [[VAR_ADDR1_REF:%.+]] = load [[S_INT_TY]]*, [[S_INT_TY]]** % 476 // CHECK-DAG: store [[S_INT_TY]]* [[VAR_PRIV1]], [[S_INT_TY]]** [[TMP_PRIV1]], 477 478 // CHECK: call void @__kmpc_for_static_init_4( 479 // CHECK: call void {{.*}} @__kmpc_fork_call({{.+}}, {{.+}}, {{.+}} @[[TPAR_OUTL:.+]] to 480 // CHECK: call void @__kmpc_for_static_fini( 481 482 // lastprivates 483 // CHECK: [[OMP_IS_LAST_VAL:%.+]] = load i{{[0-9]+}}, i{{[0-9]+}}* [[OMP_IS_LAST1]], 484 // CHECK: [[IS_LAST_IT:%.+]] = icmp ne i{{[0-9]+}} [[OMP_IS_LAST_VAL]], 0 485 // CHECK: br i1 [[IS_LAST_IT]], label %[[OMP_LASTPRIV_BLOCK:.+]], label %[[OMP_LASTPRIV_DONE:.+]] 486 487 // CHECK: [[OMP_LASTPRIV_BLOCK]]: 488 // CHECK: [[T_VAR_VAL:%.+]] = load i{{[0-9]+}}, i{{[0-9]+}}* [[T_VAR_PRIV1]], 489 // CHECK: store i{{[0-9]+}} [[T_VAR_VAL]], i{{[0-9]+}}* [[T_VAR_ADDR_REF1]], 490 // CHECK: [[BCAST_VEC_ADDR_REF:%.+]] = bitcast [2 x i{{[0-9]+}}]* [[VEC_ADDR_REF1]] to i8* 491 // CHECK: [[BCAST_VEC_PRIV:%.+]] = bitcast [2 x i{{[0-9]+}}]* [[VEC_PRIV1]] to i8* 492 // CHECK: call void @llvm.memcpy.{{.+}}(i8* align {{[0-9]+}} [[BCAST_VEC_ADDR_REF]], i8* align {{[0-9]+}} [[BCAST_VEC_PRIV]], 493 // CHECK: [[S_ARR_BEGIN:%.+]] = getelementptr inbounds [2 x [[S_INT_TY]]], [2 x [[S_INT_TY]]]* [[S_ARR_ADDR_REF]], i{{[0-9]+}} 0, i{{[0-9]+}} 0 494 // CHECK: [[S_ARR_PRIV_BCAST:%.+]] = bitcast [2 x [[S_INT_TY]]]* [[S_ARR_PRIV1]] to [[S_INT_TY]]* 495 // CHECK: [[S_ARR_BEGIN_GEP:%.+]] = getelementptr [[S_INT_TY]], [[S_INT_TY]]* [[S_ARR_BEGIN]], i{{[0-9]+}} 2 496 // CHECK: [[S_ARR_IS_EMPTY:%.+]] = icmp eq [[S_INT_TY]]* [[S_ARR_BEGIN]], [[S_ARR_BEGIN_GEP]] 497 // CHECK: br i1 [[S_ARR_IS_EMPTY]], label %[[S_ARR_COPY_DONE:.+]], label %[[S_ARR_COPY_BLOCK:.+]] 498 // CHECK: [[S_ARR_COPY_BLOCK]]: 499 // CHECK: [[S_ARR_SRC_EL:%.+]] = phi [[S_INT_TY]]*{{.+}} 500 // CHECK: [[S_ARR_DST_EL:%.+]] = phi [[S_INT_TY]]*{{.+}} 501 // CHECK: [[S_ARR_DST_BCAST:%.+]] = bitcast [[S_INT_TY]]* [[S_ARR_DST_EL]] to i8* 502 // CHECK: [[S_ARR_SRC_BCAST:%.+]] = bitcast [[S_INT_TY]]* [[S_ARR_SRC_EL]] to i8* 503 // CHECK: call void @llvm.memcpy.{{.+}}(i8* align {{[0-9]+}} [[S_ARR_DST_BCAST]], i8* align {{[0-9]+}} [[S_ARR_SRC_BCAST]]{{.+}}) 504 // CHECK: [[S_ARR_DST_NEXT:%.+]] = getelementptr [[S_INT_TY]], [[S_INT_TY]]* [[S_ARR_DST_EL]], i{{[0-9]+}} 1 505 // CHECK: [[S_ARR_SRC_NEXT:%.+]] = getelementptr{{.+}} 506 // CHECK: [[CPY_IS_FINISHED:%.+]] = icmp eq [[S_INT_TY]]* [[S_ARR_DST_NEXT]], [[S_ARR_BEGIN_GEP]] 507 // CHECK: br i1 [[CPY_IS_FINISHED]], label %[[S_ARR_COPY_DONE]], label %[[S_ARR_COPY_BLOCK]] 508 // CHECK: [[S_ARR_COPY_DONE]]: 509 // CHECK: [[TMP_VAL1:%.+]] = load [[S_INT_TY]]*, [[S_INT_TY]]** [[TMP_PRIV1]], 510 // CHECK: [[VAR_ADDR_REF_BCAST:%.+]] = bitcast [[S_INT_TY]]* [[VAR_ADDR1_REF]] to i8* 511 // CHECK: [[TMP_VAL1_BCAST:%.+]] = bitcast [[S_INT_TY]]* [[TMP_VAL1]] to i8* 512 // CHECK: call void @llvm.memcpy.{{.+}}(i8* align {{[0-9]+}} [[VAR_ADDR_REF_BCAST]], i8* align {{[0-9]+}} [[TMP_VAL1_BCAST]],{{.+}}) 513 // CHECK: ret void 514 515 // CHECK: define internal void [[TPAR_OUTL:@.+]](i{{[0-9]+}}* noalias [[GTID_ADDR1:%.+]], i{{[0-9]+}}* noalias %{{.+}}, {{.+}}, {{.+}}, [2 x i{{[0-9]+}}]*{{.+}} [[VEC_IN1:%.+]], i{{[0-9]+}}*{{.+}} [[T_VAR_IN1:%.+]], [2 x [[S_INT_TY]]]*{{.+}} [[S_ARR_IN1:%.+]], [[S_INT_TY]]*{{.+}} [[VAR_IN1:%.+]]) 516 // skip alloca of global_tid and bound_tid, and prev lb and ub vars 517 // CHECK: {{.+}} = alloca i{{[0-9]+}}*, 518 // CHECK: {{.+}} = alloca i{{[0-9]+}}*, 519 // CHECK: alloca i{{[0-9]+}}, 520 // CHECK: alloca i{{[0-9]+}}, 521 // CHECK: [[VEC_ADDR1:%.+]] = alloca [2 x i{{[0-9]+}}]*, 522 // CHECK: [[T_VAR_ADDR1:%.+]] = alloca i{{[0-9]+}}*, 523 // CHECK: [[S_ARR_ADDR1:%.+]] = alloca [2 x [[S_INT_TY]]]*, 524 // CHECK: [[VAR_ADDR1:%.+]] = alloca [[S_INT_TY]]*, 525 // skip loop variables 526 // CHECK: {{.+}} = alloca i{{[0-9]+}}, 527 // CHECK: {{.+}} = alloca i{{[0-9]+}}, 528 // CHECK: {{.+}} = alloca i{{[0-9]+}}, 529 // CHECK: {{.+}} = alloca i{{[0-9]+}}, 530 // CHECK: {{.+}} = alloca i{{[0-9]+}}, 531 // CHECK: [[OMP_IS_LAST1:%.+]] = alloca i{{[0-9]+}}, 532 // CHECK: [[T_VAR_PRIV1:%.+]] = alloca i{{[0-9]+}}, 533 // CHECK: [[VEC_PRIV1:%.+]] = alloca [2 x i{{[0-9]+}}], 534 // CHECK: [[S_ARR_PRIV1:%.+]] = alloca [2 x [[S_INT_TY]]], 535 // CHECK: [[VAR_PRIV1:%.+]] = alloca [[S_INT_TY]], 536 // CHECK: [[TMP_PRIV1:%.+]] = alloca [[S_INT_TY]]*, 537 538 // skip init of bound and global tid 539 // CHECK: store i{{[0-9]+}}* {{.*}}, 540 // CHECK: store i{{[0-9]+}}* {{.*}}, 541 // copy from parameters to local address variables 542 // CHECK: store [2 x i{{[0-9]+}}]* [[VEC_IN1]], [2 x i{{[0-9]+}}]** [[VEC_ADDR1]], 543 // CHECK: store i{{[0-9]+}}* [[T_VAR_IN1]], i{{[0-9]+}}** [[T_VAR_ADDR1]], 544 // CHECK: store [2 x [[S_INT_TY]]]* [[S_ARR_IN1]], [2 x [[S_INT_TY]]]** [[S_ARR_ADDR1]], 545 // CHECK: store [[S_INT_TY]]* [[VAR_IN1]], [[S_INT_TY]]** [[VAR_ADDR1]], 546 547 // load content of local address variables 548 // CHECK: [[VEC_ADDR_REF1:%.+]] = load [2 x i{{[0-9]+}}]*, [2 x i{{[0-9]+}}]** [[VEC_ADDR1]], 549 // CHECK: [[T_VAR_ADDR_REF1:%.+]] = load i{{[0-9]+}}*, i{{[0-9]+}}** [[T_VAR_ADDR1]], 550 // CHECK: [[S_ARR_ADDR_REF1:%.+]] = load [2 x [[S_INT_TY]]]*, [2 x [[S_INT_TY]]]** [[S_ARR_ADDR1]], 551 // CHECK: [[VAR_ADDR1_REF:%.+]] = load [[S_INT_TY]]*, [[S_INT_TY]]** [[VAR_ADDR1]], 552 // CHECK-DAG: store i{{[0-9]+}} 0, i{{[0-9]+}}* [[OMP_IS_LAST1]], 553 // CHECK: [[VAR_ADDR1_REF:%.+]] = load [[S_INT_TY]]*, [[S_INT_TY]]** % 554 // CHECK-DAG: store [[S_INT_TY]]* [[VAR_PRIV1]], [[S_INT_TY]]** [[TMP_PRIV1]], 555 556 // CHECK: call void @__kmpc_for_static_init_4( 557 558 // assignment: vec[i] = t_var; 559 // CHECK: [[IV_VAL1:%.+]] = 560 // CHECK: [[T_VAR_PRIV_VAL1:%.+]] = load i{{[0-9]+}}, i{{[0-9]+}}* [[T_VAR_PRIV1]], 561 // CHECK: [[VEC_PTR1:%.+]] = getelementptr inbounds [2 x i{{[0-9]+}}], [2 x i{{[0-9]+}}]* [[VEC_PRIV1]], i{{[0-9]+}} 0, i{{[0-9]+}} {{.+}} 562 // CHECK: store i{{[0-9]+}} [[T_VAR_PRIV_VAL1]], i{{[0-9]+}}* [[VEC_PTR1]], 563 564 // assignment: s_arr[i] = var; 565 // CHECK-DAG: [[S_ARR_PTR1:%.+]] = getelementptr inbounds [2 x [[S_INT_TY]]], [2 x [[S_INT_TY]]]* [[S_ARR_PRIV1]], 566 // CHECK-DAG: [[TMP_VAL1:%.+]] = load [[S_INT_TY]]*, [[S_INT_TY]]** [[TMP_PRIV1]], 567 // CHECK-DAG: [[S_ARR_PTR_BCAST1:%.+]] = bitcast [[S_INT_TY]]* [[S_ARR_PTR1]] to i8* 568 // CHECK-DAG: [[TMP_VAL_BCAST1:%.+]] = bitcast [[S_INT_TY]]* [[TMP_VAL1]] to i8* 569 // CHECK-DAG: call void @llvm.memcpy.{{.+}}(i8* align {{[0-9]+}} [[S_ARR_PTR_BCAST1]], i8* align {{[0-9]+}} [[TMP_VAL_BCAST1]], 570 571 // CHECK: call void @__kmpc_for_static_fini( 572 573 // lastprivates 574 // CHECK: [[OMP_IS_LAST_VAL:%.+]] = load i{{[0-9]+}}, i{{[0-9]+}}* [[OMP_IS_LAST1]], 575 // CHECK: [[IS_LAST_IT:%.+]] = icmp ne i{{[0-9]+}} [[OMP_IS_LAST_VAL]], 0 576 // CHECK: br i1 [[IS_LAST_IT]], label %[[OMP_LASTPRIV_BLOCK:.+]], label %[[OMP_LASTPRIV_DONE:.+]] 577 578 // CHECK: [[OMP_LASTPRIV_BLOCK]]: 579 // CHECK: [[T_VAR_VAL:%.+]] = load i{{[0-9]+}}, i{{[0-9]+}}* [[T_VAR_PRIV1]], 580 // CHECK: store i{{[0-9]+}} [[T_VAR_VAL]], i{{[0-9]+}}* [[T_VAR_ADDR_REF1]], 581 // CHECK: [[BCAST_VEC_ADDR_REF:%.+]] = bitcast [2 x i{{[0-9]+}}]* [[VEC_ADDR_REF1]] to i8* 582 // CHECK: [[BCAST_VEC_PRIV:%.+]] = bitcast [2 x i{{[0-9]+}}]* [[VEC_PRIV1]] to i8* 583 // CHECK: call void @llvm.memcpy.{{.+}}(i8* align {{[0-9]+}} [[BCAST_VEC_ADDR_REF]], i8* align {{[0-9]+}} [[BCAST_VEC_PRIV]], 584 // CHECK: [[S_ARR_BEGIN:%.+]] = getelementptr inbounds [2 x [[S_INT_TY]]], [2 x [[S_INT_TY]]]* [[S_ARR_ADDR_REF]], i{{[0-9]+}} 0, i{{[0-9]+}} 0 585 // CHECK: [[S_ARR_PRIV_BCAST:%.+]] = bitcast [2 x [[S_INT_TY]]]* [[S_ARR_PRIV1]] to [[S_INT_TY]]* 586 // CHECK: [[S_ARR_BEGIN_GEP:%.+]] = getelementptr [[S_INT_TY]], [[S_INT_TY]]* [[S_ARR_BEGIN]], i{{[0-9]+}} 2 587 // CHECK: [[S_ARR_IS_EMPTY:%.+]] = icmp eq [[S_INT_TY]]* [[S_ARR_BEGIN]], [[S_ARR_BEGIN_GEP]] 588 // CHECK: br i1 [[S_ARR_IS_EMPTY]], label %[[S_ARR_COPY_DONE:.+]], label %[[S_ARR_COPY_BLOCK:.+]] 589 // CHECK: [[S_ARR_COPY_BLOCK]]: 590 // CHECK: [[S_ARR_SRC_EL:%.+]] = phi [[S_INT_TY]]*{{.+}} 591 // CHECK: [[S_ARR_DST_EL:%.+]] = phi [[S_INT_TY]]*{{.+}} 592 // CHECK: [[S_ARR_DST_BCAST:%.+]] = bitcast [[S_INT_TY]]* [[S_ARR_DST_EL]] to i8* 593 // CHECK: [[S_ARR_SRC_BCAST:%.+]] = bitcast [[S_INT_TY]]* [[S_ARR_SRC_EL]] to i8* 594 // CHECK: call void @llvm.memcpy.{{.+}}(i8* align {{[0-9]+}} [[S_ARR_DST_BCAST]], i8* align {{[0-9]+}} [[S_ARR_SRC_BCAST]]{{.+}}) 595 // CHECK: [[S_ARR_DST_NEXT:%.+]] = getelementptr [[S_INT_TY]], [[S_INT_TY]]* [[S_ARR_DST_EL]], i{{[0-9]+}} 1 596 // CHECK: [[S_ARR_SRC_NEXT:%.+]] = getelementptr{{.+}} 597 // CHECK: [[CPY_IS_FINISHED:%.+]] = icmp eq [[S_INT_TY]]* [[S_ARR_DST_NEXT]], [[S_ARR_BEGIN_GEP]] 598 // CHECK: br i1 [[CPY_IS_FINISHED]], label %[[S_ARR_COPY_DONE]], label %[[S_ARR_COPY_BLOCK]] 599 // CHECK: [[S_ARR_COPY_DONE]]: 600 // CHECK: [[TMP_VAL1:%.+]] = load [[S_INT_TY]]*, [[S_INT_TY]]** [[TMP_PRIV1]], 601 // CHECK: [[VAR_ADDR_REF_BCAST:%.+]] = bitcast [[S_INT_TY]]* [[VAR_ADDR1_REF]] to i8* 602 // CHECK: [[TMP_VAL1_BCAST:%.+]] = bitcast [[S_INT_TY]]* [[TMP_VAL1]] to i8* 603 // CHECK: call void @llvm.memcpy.{{.+}}(i8* align {{[0-9]+}} [[VAR_ADDR_REF_BCAST]], i8* align {{[0-9]+}} [[TMP_VAL1_BCAST]],{{.+}}) 604 // CHECK: ret void 605 606 #endif 607