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-target | 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 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-target | 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-target | 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 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-target | 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-target | 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 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-target | 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-target | 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 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-target | 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-target | 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 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-target | 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-target | 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 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-target | 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-target | 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 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-target | 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-target | 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 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-target | 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 54 #pragma omp distribute simd lastprivate(t_var, vec, s_arr, s_arr, var, var) 55 for (int i = 0; i < 2; ++i) { 56 vec[i] = t_var; 57 s_arr[i] = var; 58 } 59 return T(); 60 } 61 62 int main() { 63 static int svar; 64 volatile double g; 65 volatile double &g1 = g; 66 67 #ifdef LAMBDA 68 // LAMBDA-LABEL: @main 69 // LAMBDA: call{{.*}} void [[OUTER_LAMBDA:@.+]]( 70 [&]() { 71 static float sfvar; 72 // LAMBDA: define{{.*}} internal{{.*}} void [[OUTER_LAMBDA]]( 73 // LAMBDA: call i{{[0-9]+}} @__tgt_target_teams( 74 // LAMBDA: call void [[OFFLOADING_FUN:@.+]]( 75 76 // LAMBDA: define{{.+}} void [[OFFLOADING_FUN]]( 77 // LAMBDA: call {{.*}}void {{.+}} @__kmpc_fork_teams({{.+}}, i32 4, {{.+}}* [[OMP_OUTLINED:@.+]] to {{.+}}) 78 #pragma omp target 79 #pragma omp teams 80 #pragma omp distribute simd lastprivate(g, g1, svar, sfvar) 81 for (int i = 0; i < 2; ++i) { 82 // LAMBDA: define{{.*}} internal{{.*}} void [[OMP_OUTLINED]](i32* noalias %{{.+}}, i32* noalias %{{.+}}, double*{{.+}} [[G_IN:%.+]], double*{{.+}} [[G1_IN:%.+]], i{{[0-9]+}}*{{.+}} [[SVAR_IN:%.+]], float*{{.+}} [[SFVAR_IN:%.+]]) 83 // LAMBDA: [[G_PRIVATE_ADDR:%.+]] = alloca double*, 84 // LAMBDA: [[G1_PRIVATE_ADDR:%.+]] = alloca double*, 85 // LAMBDA: [[SVAR_PRIVATE_ADDR:%.+]] = alloca i{{[0-9]+}}*, 86 // LAMBDA: [[SFVAR_PRIVATE_ADDR:%.+]] = alloca float*, 87 // LAMBDA: [[TMP_G1:%.+]] = alloca double*, 88 // loop variables 89 // LAMBDA: {{.+}} = alloca i{{[0-9]+}}, 90 // LAMBDA: {{.+}} = alloca i{{[0-9]+}}, 91 // LAMBDA: {{.+}} = alloca i{{[0-9]+}}, 92 // LAMBDA: {{.+}} = alloca i{{[0-9]+}}, 93 // LAMBDA: {{.+}} = alloca i{{[0-9]+}}, 94 // LAMBDA: [[OMP_IS_LAST:%.+]] = alloca i{{[0-9]+}}, 95 // LAMBDA: [[G_PRIVATE:%.+]] = alloca double, 96 // LAMBDA: [[G1_PRIVATE:%.+]] = alloca double, 97 // LAMBDA: [[TMP_G1_PRIVATE:%.+]] = alloca double*, 98 // LAMBDA: [[SVAR_PRIVATE:%.+]] = alloca i{{[0-9]+}}, 99 // LAMBDA: [[SFVAR_PRIVATE:%.+]] = alloca float, 100 // LAMBDA: store double* [[G_IN]], double** [[G_PRIVATE_ADDR]], 101 // LAMBDA: store double* [[G1_IN]], double** [[G1_PRIVATE_ADDR]], 102 // LAMBDA: store i{{[0-9]+}}* [[SVAR_IN]], i{{[0-9]+}}** [[SVAR_PRIVATE_ADDR]], 103 // LAMBDA: store float* [[SFVAR_IN]], float** [[SFVAR_PRIVATE_ADDR]], 104 105 // init private variables 106 // LAMBDA: [[G_IN_REF:%.+]] = load double*, double** [[G_PRIVATE_ADDR]], 107 // LAMBDA: [[SVAR_IN_REF:%.+]] = load i{{[0-9]+}}*, i{{[0-9]+}}** [[SVAR_PRIVATE_ADDR]], 108 // LAMBDA: [[SFVAR_IN_REF:%.+]] = load float*, float** [[SFVAR_PRIVATE_ADDR]], 109 // LAMBDA: [[G1_IN_REF:%.+]] = load double*, double** [[G1_PRIVATE_ADDR]], 110 // LAMBDA: store double* [[G1_IN_REF]], double** [[TMP_G1]], 111 // LAMBDA: [[TMP_G1_VAL:%.+]] = load double*, double** [[TMP_G1]], 112 // LAMBDA: store double* [[G1_PRIVATE]], double** [[TMP_G1_PRIVATE]], 113 g = 1; 114 g1 = 1; 115 svar = 3; 116 sfvar = 4.0; 117 // LAMBDA: call {{.*}}void @__kmpc_for_static_init_4( 118 // LAMBDA: store double 1.0{{.+}}, double* [[G_PRIVATE]], 119 // LAMBDA: [[TMP_G1_REF:%.+]] = load double*, double** [[TMP_G1_PRIVATE]], 120 // LAMBDA: store{{.+}} double 1.0{{.+}}, double* [[TMP_G1_REF]], 121 // LAMBDA: store i{{[0-9]+}} 3, i{{[0-9]+}}* [[SVAR_PRIVATE]], 122 // LAMBDA: store float 4.0{{.+}}, float* [[SFVAR_PRIVATE]], 123 // LAMBDA: [[G_PRIVATE_ADDR_REF:%.+]] = getelementptr inbounds %{{.+}}, %{{.+}}* [[ARG:%.+]], i{{[0-9]+}} 0, i{{[0-9]+}} 0 124 // LAMBDA: store double* [[G_PRIVATE]], double** [[G_PRIVATE_ADDR_REF]], 125 // LAMBDA: [[TMP_PRIVATE_ADDR_REF:%.+]] = getelementptr inbounds %{{.+}}, %{{.+}}* [[ARG:%.+]], i{{[0-9]+}} 0, i{{[0-9]+}} 1 126 // LAMBDA: [[G1_PRIVATE_ADDR_FROM_TMP:%.+]] = load double*, double** [[TMP_G1_PRIVATE]], 127 // LAMBDA: store double* [[G1_PRIVATE_ADDR_FROM_TMP]], double** [[TMP_PRIVATE_ADDR_REF]], 128 // LAMBDA: [[SVAR_PRIVATE_ADDR_REF:%.+]] = getelementptr inbounds %{{.+}}, %{{.+}}* [[ARG:%.+]], i{{[0-9]+}} 0, i{{[0-9]+}} 2 129 // LAMBDA: store i{{[0-9]+}}* [[SVAR_PRIVATE]], i{{[0-9]+}}** [[SVAR_PRIVATE_ADDR_REF]] 130 // LAMBDA: [[SFVAR_PRIVATE_ADDR_REF:%.+]] = getelementptr inbounds %{{.+}}, %{{.+}}* [[ARG:%.+]], i{{[0-9]+}} 0, i{{[0-9]+}} 3 131 // LAMBDA: store float* [[SFVAR_PRIVATE]], float** [[SFVAR_PRIVATE_ADDR_REF]] 132 // LAMBDA: call{{.*}} void [[INNER_LAMBDA:@.+]](%{{.+}}* [[ARG]]) 133 // LAMBDA: call {{.*}}void @__kmpc_for_static_fini( 134 135 // linear counter 136 // LAMBDA: [[OMP_IS_LAST_VAL:%.+]] = load i{{[0-9]+}}, i{{[0-9]+}}* [[OMP_IS_LAST]], 137 // LAMBDA: [[IS_LAST_IT:%.+]] = icmp ne i{{[0-9]+}} [[OMP_IS_LAST_VAL]], 0 138 // LAMBDA: br i1 [[IS_LAST_IT]], label %[[OMP_COUNTER_BLOCK:.+]], label %[[OMP_COUNTER_DONE:.+]] 139 // LAMBDA: [[OMP_COUNTER_BLOCK]]: 140 // LAMBDA: store i{{[0-9]+}} 2, i{{[0-9]+}}* % 141 // LAMBDA: br label %[[OMP_COUNTER_DONE]] 142 // LAMBDA: [[OMP_COUNTER_DONE]]: 143 144 145 // LAMBDA: [[OMP_IS_LAST_VAL:%.+]] = load i{{[0-9]+}}, i{{[0-9]+}}* [[OMP_IS_LAST]], 146 // LAMBDA: [[IS_LAST_IT:%.+]] = icmp ne i{{[0-9]+}} [[OMP_IS_LAST_VAL]], 0 147 // LAMBDA: br i1 [[IS_LAST_IT]], label %[[OMP_LASTPRIV_BLOCK:.+]], label %[[OMP_LASTPRIV_DONE:.+]] 148 149 // LAMBDA: [[OMP_LASTPRIV_BLOCK]]: 150 // LAMBDA: [[G_PRIV_VAL:%.+]] = load double, double* [[G_PRIVATE]], 151 // LAMBDA: store{{.*}} double [[G_PRIV_VAL]], double* [[G_IN_REF]], 152 // LAMBDA: [[TMP_G1_PRIV_REF:%.+]] = load double*, double** [[TMP_G1_PRIVATE]], 153 // LAMBDA: [[TMP_G1_PRIV_VAL:%.+]] = load double, double* [[TMP_G1_PRIV_REF]], 154 // LAMBDA: store{{.*}} double [[TMP_G1_PRIV_VAL]], double* [[TMP_G1_VAL]], 155 156 // LAMBDA: [[SVAR_PRIV_VAL:%.+]] = load i{{[0-9]+}}, i{{[0-9]+}}* [[SVAR_PRIVATE]], 157 // LAMBDA: store i{{[0-9]+}} [[SVAR_PRIV_VAL]], i{{[0-9]+}}* [[SVAR_IN_REF]], 158 // LAMBDA: [[SFVAR_PRIV_VAL:%.+]] = load float, float* [[SFVAR_PRIVATE]], 159 // LAMBDA: store float [[SFVAR_PRIV_VAL]], float* [[SFVAR_IN_REF]], 160 // LAMBDA: br label %[[OMP_LASTPRIV_DONE]] 161 // LAMBDA: [[OMP_LASTPRIV_DONE]]: 162 // LAMBDA: ret 163 [&]() { 164 // LAMBDA: define {{.+}} void [[INNER_LAMBDA]](%{{.+}}* [[ARG_PTR:%.+]]) 165 // LAMBDA: store %{{.+}}* [[ARG_PTR]], %{{.+}}** [[ARG_PTR_REF:%.+]], 166 g = 2; 167 g1 = 2; 168 svar = 4; 169 sfvar = 8.0; 170 // LAMBDA: [[ARG_PTR:%.+]] = load %{{.+}}*, %{{.+}}** [[ARG_PTR_REF]] 171 // LAMBDA: [[G_PTR_REF:%.+]] = getelementptr inbounds %{{.+}}, %{{.+}}* [[ARG_PTR]], i{{[0-9]+}} 0, i{{[0-9]+}} 0 172 // LAMBDA: [[G_REF:%.+]] = load double*, double** [[G_PTR_REF]] 173 // LAMBDA: store double 2.0{{.+}}, double* [[G_REF]] 174 175 // LAMBDA: [[TMP_PTR_REF:%.+]] = getelementptr inbounds %{{.+}}, %{{.+}}* [[ARG_PTR]], i{{[0-9]+}} 0, i{{[0-9]+}} 1 176 // LAMBDA: [[G1_REF:%.+]] = load double*, double** [[TMP_PTR_REF]] 177 // LAMBDA: store double 2.0{{.+}}, double* [[G1_REF]], 178 // LAMBDA: [[SVAR_PTR_REF:%.+]] = getelementptr inbounds %{{.+}}, %{{.+}}* [[ARG_PTR]], i{{[0-9]+}} 0, i{{[0-9]+}} 2 179 // LAMBDA: [[SVAR_REF:%.+]] = load i{{[0-9]+}}*, i{{[0-9]+}}** [[SVAR_PTR_REF]] 180 // LAMBDA: store i{{[0-9]+}} 4, i{{[0-9]+}}* [[SVAR_REF]] 181 // LAMBDA: [[SFVAR_PTR_REF:%.+]] = getelementptr inbounds %{{.+}}, %{{.+}}* [[ARG_PTR]], i{{[0-9]+}} 0, i{{[0-9]+}} 3 182 // LAMBDA: [[SFVAR_REF:%.+]] = load float*, float** [[SFVAR_PTR_REF]] 183 // LAMBDA: store float 8.0{{.+}}, float* [[SFVAR_REF]] 184 }(); 185 } 186 }(); 187 return 0; 188 #else 189 S<float> test; 190 int t_var = 0; 191 int vec[] = {1, 2}; 192 S<float> s_arr[] = {1, 2}; 193 S<float> &var = test; 194 195 #pragma omp target 196 #pragma omp teams 197 #pragma omp distribute simd lastprivate(t_var, vec, s_arr, s_arr, var, var, svar) 198 for (int i = 0; i < 2; ++i) { 199 vec[i] = t_var; 200 s_arr[i] = var; 201 } 202 int i; 203 204 return tmain<int>(); 205 #endif 206 } 207 208 // CHECK: define{{.*}} i{{[0-9]+}} @main() 209 // CHECK: [[TEST:%.+]] = alloca [[S_FLOAT_TY]], 210 // CHECK: call {{.*}} [[S_FLOAT_TY_DEF_CONSTR:@.+]]([[S_FLOAT_TY]]* [[TEST]]) 211 // CHECK: call i{{[0-9]+}} @__tgt_target_teams( 212 // CHECK: call void [[OFFLOAD_FUN:@.+]](i{{[0-9]+}} {{.+}}, [2 x i{{[0-9]+}}]* {{.+}}, [2 x [[S_FLOAT_TY]]]* {{.+}}, [[S_FLOAT_TY]]* {{.+}}, i{{[0-9]+}} {{.+}}) 213 // CHECK: ret 214 215 // CHECK: define{{.+}} [[OFFLOAD_FUN]](i{{[0-9]+}} {{.+}}, [2 x i{{[0-9]+}}]*{{.+}} {{.+}}, [2 x [[S_FLOAT_TY]]]*{{.+}} {{.+}}, [[S_FLOAT_TY]]*{{.+}} {{.+}}, i{{[0-9]+}} {{.+}}) 216 // CHECK: call void (%{{.+}}*, i{{[0-9]+}}, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)*, ...) @__kmpc_fork_teams( 217 // CHECK: ret 218 // 219 // CHECK: define internal void [[OMP_OUTLINED:@.+]](i{{[0-9]+}}* noalias [[GTID_ADDR:%.+]], i{{[0-9]+}}* noalias %{{.+}}, i{{[0-9]+}}*{{.+}} [[T_VAR_IN:%.+]], [2 x i{{[0-9]+}}]*{{.+}} [[VEC_IN:%.+]], [2 x [[S_FLOAT_TY]]]*{{.+}} [[S_ARR_IN:%.+]], [[S_FLOAT_TY]]*{{.+}} [[VAR_IN:%.+]], i{{[0-9]+}}*{{.*}} [[S_VAR_IN:%.+]]) 220 // CHECK: {{.+}} = alloca i{{[0-9]+}}*, 221 // CHECK: {{.+}} = alloca i{{[0-9]+}}*, 222 // CHECK: [[T_VAR_ADDR:%.+]] = alloca i{{[0-9]+}}*, 223 // CHECK: [[VEC_ADDR:%.+]] = alloca [2 x i{{[0-9]+}}]*, 224 // CHECK: [[S_ARR_ADDR:%.+]] = alloca [2 x [[S_FLOAT_TY]]]*, 225 // CHECK: [[VAR_ADDR:%.+]] = alloca [[S_FLOAT_TY]]*, 226 // CHECK: [[SVAR_ADDR:%.+]] = alloca i{{[0-9]+}}*, 227 // CHECK: [[TMP:%.*]] = alloca [[S_FLOAT_TY]]*, 228 // skip loop variables 229 // CHECK: {{.+}} = alloca i{{[0-9]+}}, 230 // CHECK: {{.+}} = alloca i{{[0-9]+}}, 231 // CHECK: {{.+}} = alloca i{{[0-9]+}}, 232 // CHECK: {{.+}} = alloca i{{[0-9]+}}, 233 // CHECK: {{.+}} = alloca i{{[0-9]+}}, 234 // CHECK: [[OMP_IS_LAST:%.+]] = alloca i{{[0-9]+}}, 235 // CHECK: [[T_VAR_PRIV:%.+]] = alloca i{{[0-9]+}}, 236 // CHECK: [[VEC_PRIV:%.+]] = alloca [2 x i{{[0-9]+}}], 237 // CHECK: [[S_ARR_PRIV:%.+]] = alloca [2 x [[S_FLOAT_TY]]], 238 // CHECK: [[VAR_PRIV:%.+]] = alloca [[S_FLOAT_TY]], 239 // CHECK: [[TMP_PRIV:%.+]] = alloca [[S_FLOAT_TY]]*, 240 // CHECK: [[S_VAR_PRIV:%.+]] = alloca i{{[0-9]+}}, 241 242 // copy from parameters to local address variables 243 // CHECK: store i{{[0-9]+}}* [[T_VAR_IN]], i{{[0-9]+}}** [[T_VAR_ADDR]], 244 // CHECK: store [2 x i{{[0-9]+}}]* [[VEC_IN]], [2 x i{{[0-9]+}}]** [[VEC_ADDR]], 245 // CHECK: store [2 x [[S_FLOAT_TY]]]* [[S_ARR_IN]], [2 x [[S_FLOAT_TY]]]** [[S_ARR_ADDR]], 246 // CHECK: store [[S_FLOAT_TY]]* [[VAR_IN]], [[S_FLOAT_TY]]** [[VAR_ADDR]], 247 // CHECK: store i{{[0-9]+}}* [[S_VAR_IN]], i{{[0-9]+}}** [[SVAR_ADDR]], 248 249 // load content of local address variables 250 // CHECK: [[T_VAR_ADDR_REF:%.+]] = load i{{[0-9]+}}*, i{{[0-9]+}}** [[T_VAR_ADDR]], 251 // CHECK: [[VEC_ADDR_REF:%.+]] = load [2 x i{{[0-9]+}}]*, [2 x i{{[0-9]+}}]** [[VEC_ADDR]], 252 // CHECK: [[S_ARR_ADDR_REF:%.+]] = load [2 x [[S_FLOAT_TY]]]*, [2 x [[S_FLOAT_TY]]]** [[S_ARR_ADDR]], 253 // CHECK: [[SVAR_ADDR_REF:%.+]] = load i{{[0-9]+}}*, i{{[0-9]+}}** [[SVAR_ADDR]], 254 // CHECK: [[VAR_ADDR_REF:%.+]] = load [[S_FLOAT_TY]]*, [[S_FLOAT_TY]]** [[VAR_ADDR]], 255 // CHECK: store [[S_FLOAT_TY]]* [[VAR_ADDR_REF]], [[S_FLOAT_TY]]** [[TMP]], 256 // CHECK: store i{{[0-9]+}} 0, i{{[0-9]+}}* [[OMP_IS_LAST]], 257 // CHECK: [[TMP_REF:%.+]] = load [[S_FLOAT_TY]]*, [[S_FLOAT_TY]]** [[TMP]], 258 // the distribute loop 259 // CHECK: call void @__kmpc_for_static_init_4( 260 // assignment: vec[i] = t_var; 261 // CHECK: [[T_VAR_PRIV_VAL:%.+]] = load i{{[0-9]+}}, i{{[0-9]+}}* [[T_VAR_PRIV]], 262 // CHECK: [[VEC_PTR:%.+]] = getelementptr inbounds [2 x i{{[0-9]+}}], [2 x i{{[0-9]+}}]* [[VEC_PRIV]], i{{[0-9]+}} 0, i{{[0-9]+}} {{.+}} 263 // CHECK: store i{{[0-9]+}} [[T_VAR_PRIV_VAL]], i{{[0-9]+}}* [[VEC_PTR]], 264 265 // assignment: s_arr[i] = var; 266 // CHECK-DAG: [[S_ARR_PTR:%.+]] = getelementptr inbounds [2 x [[S_FLOAT_TY]]], [2 x [[S_FLOAT_TY]]]* [[S_ARR_PRIV]], 267 // CHECK-DAG: [[TMP_VAL:%.+]] = load [[S_FLOAT_TY]]*, [[S_FLOAT_TY]]** [[TMP_PRIV]], 268 // CHECK-DAG: [[S_ARR_PTR_BCAST:%.+]] = bitcast [[S_FLOAT_TY]]* [[S_ARR_PTR]] to i8* 269 // CHECK-DAG: [[TMP_VAL_BCAST:%.+]] = bitcast [[S_FLOAT_TY]]* [[TMP_VAL]] to i8* 270 // CHECK: call void @llvm.memcpy.{{.+}}(i8* align {{[0-9]+}} [[S_ARR_PTR_BCAST]], i8* align {{[0-9]+}} [[TMP_VAL_BCAST]], 271 // CHECK: call void @__kmpc_for_static_fini( 272 273 // lastprivates 274 // CHECK: [[OMP_IS_LAST_VAL:%.+]] = load i{{[0-9]+}}, i{{[0-9]+}}* [[OMP_IS_LAST]], 275 // CHECK: [[IS_LAST_IT:%.+]] = icmp ne i{{[0-9]+}} [[OMP_IS_LAST_VAL]], 0 276 // CHECK: br i1 [[IS_LAST_IT]], label %[[OMP_LASTPRIV_BLOCK:.+]], label %[[OMP_LASTPRIV_DONE:.+]] 277 278 // CHECK: [[OMP_LASTPRIV_BLOCK]]: 279 // CHECK: [[T_VAR_VAL:%.+]] = load i{{[0-9]+}}, i{{[0-9]+}}* [[T_VAR_PRIV]], 280 // CHECK: store i{{[0-9]+}} [[T_VAR_VAL]], i{{[0-9]+}}* [[T_VAR_ADDR_REF]], 281 // CHECK: [[BCAST_VEC_ADDR_REF:%.+]] = bitcast [2 x i{{[0-9]+}}]* [[VEC_ADDR_REF]] to i8* 282 // CHECK: [[BCAST_VEC_PRIV:%.+]] = bitcast [2 x i{{[0-9]+}}]* [[VEC_PRIV]] to i8* 283 // CHECK: call void @llvm.memcpy.{{.+}}(i8* align {{[0-9]+}} [[BCAST_VEC_ADDR_REF]], i8* align {{[0-9]+}} [[BCAST_VEC_PRIV]], 284 // 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]+}} 0 285 // CHECK: [[S_ARR_PRIV_BCAST:%.+]] = bitcast [2 x [[S_FLOAT_TY]]]* [[S_ARR_PRIV]] to [[S_FLOAT_TY]]* 286 // CHECK: [[S_ARR_BEGIN_GEP:%.+]] = getelementptr [[S_FLOAT_TY]], [[S_FLOAT_TY]]* [[S_ARR_BEGIN]], i{{[0-9]+}} 2 287 // CHECK: [[S_ARR_IS_EMPTY:%.+]] = icmp eq [[S_FLOAT_TY]]* [[S_ARR_BEGIN]], [[S_ARR_BEGIN_GEP]] 288 // CHECK: br i1 [[S_ARR_IS_EMPTY]], label %[[S_ARR_COPY_DONE:.+]], label %[[S_ARR_COPY_BLOCK:.+]] 289 // CHECK: [[S_ARR_COPY_BLOCK]]: 290 // CHECK: [[S_ARR_SRC_EL:%.+]] = phi [[S_FLOAT_TY]]*{{.+}} 291 // CHECK: [[S_ARR_DST_EL:%.+]] = phi [[S_FLOAT_TY]]*{{.+}} 292 // CHECK: [[S_ARR_DST_BCAST:%.+]] = bitcast [[S_FLOAT_TY]]* [[S_ARR_DST_EL]] to i8* 293 // CHECK: [[S_ARR_SRC_BCAST:%.+]] = bitcast [[S_FLOAT_TY]]* [[S_ARR_SRC_EL]] to i8* 294 // CHECK: call void @llvm.memcpy.{{.+}}(i8* align {{[0-9]+}} [[S_ARR_DST_BCAST]], i8* align {{[0-9]+}} [[S_ARR_SRC_BCAST]]{{.+}}) 295 // CHECK: [[S_ARR_DST_NEXT:%.+]] = getelementptr [[S_FLOAT_TY]], [[S_FLOAT_TY]]* [[S_ARR_DST_EL]], i{{[0-9]+}} 1 296 // CHECK: [[S_ARR_SRC_NEXT:%.+]] = getelementptr{{.+}} 297 // CHECK: [[CPY_IS_FINISHED:%.+]] = icmp eq [[S_FLOAT_TY]]* [[S_ARR_DST_NEXT]], [[S_ARR_BEGIN_GEP]] 298 // CHECK: br i1 [[CPY_IS_FINISHED]], label %[[S_ARR_COPY_DONE]], label %[[S_ARR_COPY_BLOCK]] 299 // CHECK: [[S_ARR_COPY_DONE]]: 300 // CHECK: [[TMP_VAL1:%.+]] = load [[S_FLOAT_TY]]*, [[S_FLOAT_TY]]** [[TMP_PRIV]], 301 // CHECK: [[VAR_ADDR_REF_BCAST:%.+]] = bitcast [[S_FLOAT_TY]]* [[TMP_REF]] to i8* 302 // CHECK: [[TMP_VAL1_BCAST:%.+]] = bitcast [[S_FLOAT_TY]]* [[TMP_VAL1]] to i8* 303 // CHECK: call void @llvm.memcpy.{{.+}}(i8* align {{[0-9]+}} [[VAR_ADDR_REF_BCAST]], i8* align {{[0-9]+}} [[TMP_VAL1_BCAST]],{{.+}}) 304 // CHECK: [[SVAR_VAL:%.+]] = load i{{[0-9]+}}, i{{[0-9]+}}* [[S_VAR_PRIV]], 305 // CHECK: store i{{[0-9]+}} [[SVAR_VAL]], i{{[0-9]+}}* [[SVAR_ADDR_REF]], 306 // CHECK: ret void 307 308 // template tmain 309 // CHECK: define{{.*}} i{{[0-9]+}} [[TMAIN_INT:@.+]]() 310 // CHECK: [[TEST:%.+]] = alloca [[S_INT_TY]], 311 // CHECK: call {{.*}} [[S_INT_TY_DEF_CONSTR:@.+]]([[S_INT_TY]]* [[TEST]]) 312 // CHECK: call i{{[0-9]+}} @__tgt_target_teams( 313 // CHECK: call void [[OFFLOAD_FUN_1:@.+]](i{{[0-9]+}} {{.+}}, [2 x i{{[0-9]+}}]* {{.+}}, [2 x [[S_INT_TY]]]* {{.+}}, [[S_INT_TY]]* {{.+}}) 314 // CHECK: ret 315 316 317 // CHECK: define internal void [[OFFLOAD_FUN_1]]( 318 // CHECK: call void (%{{.+}}*, i{{[0-9]+}}, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)*, ...) @__kmpc_fork_teams(%{{.+}}* @{{.+}}, i{{[0-9]+}} 4, 319 // CHECK: ret 320 321 // CHECK: define internal void [[OMP_OUTLINED_1:@.+]](i{{[0-9]+}}* noalias [[GTID_ADDR1:%.+]], i{{[0-9]+}}* noalias %{{.+}}, i{{[0-9]+}}*{{.+}} [[T_VAR_IN1:%.+]], [2 x i{{[0-9]+}}]*{{.+}} [[VEC_IN1:%.+]], [2 x [[S_INT_TY]]]*{{.+}} [[S_ARR_IN1:%.+]], [[S_INT_TY]]*{{.+}} [[VAR_IN1:%.+]]) 322 // skip alloca of global_tid and bound_tid 323 // CHECK: {{.+}} = alloca i{{[0-9]+}}*, 324 // CHECK: {{.+}} = alloca i{{[0-9]+}}*, 325 // CHECK: [[T_VAR_ADDR1:%.+]] = alloca i{{[0-9]+}}*, 326 // CHECK: [[VEC_ADDR1:%.+]] = alloca [2 x i{{[0-9]+}}]*, 327 // CHECK: [[S_ARR_ADDR1:%.+]] = alloca [2 x [[S_INT_TY]]]*, 328 // CHECK: [[VAR_ADDR1:%.+]] = alloca [[S_INT_TY]]*, 329 // CHECK: [[TMP:%.+]] = alloca [[S_INT_TY]]*, 330 // skip loop variables 331 // CHECK: {{.+}} = alloca i{{[0-9]+}}, 332 // CHECK: {{.+}} = alloca i{{[0-9]+}}, 333 // CHECK: {{.+}} = alloca i{{[0-9]+}}, 334 // CHECK: {{.+}} = alloca i{{[0-9]+}}, 335 // CHECK: {{.+}} = alloca i{{[0-9]+}}, 336 // CHECK: [[OMP_IS_LAST1:%.+]] = alloca i{{[0-9]+}}, 337 // CHECK: [[T_VAR_PRIV1:%.+]] = alloca i{{[0-9]+}}, 338 // CHECK: [[VEC_PRIV1:%.+]] = alloca [2 x i{{[0-9]+}}], 339 // CHECK: [[S_ARR_PRIV1:%.+]] = alloca [2 x [[S_INT_TY]]], 340 // CHECK: [[VAR_PRIV1:%.+]] = alloca [[S_INT_TY]], 341 // CHECK: [[TMP_PRIV1:%.+]] = alloca [[S_INT_TY]]*, 342 343 // skip init of bound and global tid 344 // CHECK: store i{{[0-9]+}}* {{.*}}, 345 // CHECK: store i{{[0-9]+}}* {{.*}}, 346 // copy from parameters to local address variables 347 // CHECK: store i{{[0-9]+}}* [[T_VAR_IN1]], i{{[0-9]+}}** [[T_VAR_ADDR1]], 348 // CHECK: store [2 x i{{[0-9]+}}]* [[VEC_IN1]], [2 x i{{[0-9]+}}]** [[VEC_ADDR1]], 349 // CHECK: store [2 x [[S_INT_TY]]]* [[S_ARR_IN1]], [2 x [[S_INT_TY]]]** [[S_ARR_ADDR1]], 350 // CHECK: store [[S_INT_TY]]* [[VAR_IN1]], [[S_INT_TY]]** [[VAR_ADDR1]], 351 352 // load content of local address variables 353 // CHECK: [[T_VAR_ADDR_REF1:%.+]] = load i{{[0-9]+}}*, i{{[0-9]+}}** [[T_VAR_ADDR1]], 354 // CHECK: [[VEC_ADDR_REF1:%.+]] = load [2 x i{{[0-9]+}}]*, [2 x i{{[0-9]+}}]** [[VEC_ADDR1]], 355 // CHECK: [[S_ARR_ADDR_REF1:%.+]] = load [2 x [[S_INT_TY]]]*, [2 x [[S_INT_TY]]]** [[S_ARR_ADDR1]], 356 // CHECK: [[VAR_ADDR1_REF:%.+]] = load [[S_INT_TY]]*, [[S_INT_TY]]** [[VAR_ADDR1]], 357 // CHECK-DAG: store [[S_INT_TY]]* [[VAR_ADDR1_REF]], [[S_INT_TY]]** [[TMP]], 358 // CHECK-DAG: store i{{[0-9]+}} 0, i{{[0-9]+}}* [[OMP_IS_LAST1]], 359 // CHECK-DAG: [[TMP_REF:%.+]] = load [[S_INT_TY]]*, [[S_INT_TY]]** [[TMP]], 360 // CHECK: call void @__kmpc_for_static_init_4( 361 // assignment: vec[i] = t_var; 362 // CHECK: [[IV_VAL1:%.+]] = 363 // CHECK: [[T_VAR_PRIV_VAL1:%.+]] = load i{{[0-9]+}}, i{{[0-9]+}}* [[T_VAR_PRIV1]], 364 // CHECK: [[VEC_PTR1:%.+]] = getelementptr inbounds [2 x i{{[0-9]+}}], [2 x i{{[0-9]+}}]* [[VEC_PRIV1]], i{{[0-9]+}} 0, i{{[0-9]+}} {{.+}} 365 // CHECK: store i{{[0-9]+}} [[T_VAR_PRIV_VAL1]], i{{[0-9]+}}* [[VEC_PTR1]], 366 367 // assignment: s_arr[i] = var; 368 // CHECK-DAG: [[S_ARR_PTR1:%.+]] = getelementptr inbounds [2 x [[S_INT_TY]]], [2 x [[S_INT_TY]]]* [[S_ARR_PRIV1]], 369 // CHECK-DAG: [[TMP_VAL1:%.+]] = load [[S_INT_TY]]*, [[S_INT_TY]]** [[TMP_PRIV1]], 370 // CHECK-DAG: [[S_ARR_PTR_BCAST1:%.+]] = bitcast [[S_INT_TY]]* [[S_ARR_PTR1]] to i8* 371 // CHECK-DAG: [[TMP_VAL_BCAST1:%.+]] = bitcast [[S_INT_TY]]* [[TMP_VAL1]] to i8* 372 // CHECK-DAG: call void @llvm.memcpy.{{.+}}(i8* align {{[0-9]+}} [[S_ARR_PTR_BCAST1]], i8* align {{[0-9]+}} [[TMP_VAL_BCAST1]], 373 // CHECK: call void @__kmpc_for_static_fini( 374 375 // lastprivates 376 // CHECK: [[OMP_IS_LAST_VAL:%.+]] = load i{{[0-9]+}}, i{{[0-9]+}}* [[OMP_IS_LAST1]], 377 // CHECK: [[IS_LAST_IT:%.+]] = icmp ne i{{[0-9]+}} [[OMP_IS_LAST_VAL]], 0 378 // CHECK: br i1 [[IS_LAST_IT]], label %[[OMP_LASTPRIV_BLOCK:.+]], label %[[OMP_LASTPRIV_DONE:.+]] 379 380 // CHECK: [[OMP_LASTPRIV_BLOCK]]: 381 // CHECK: [[T_VAR_VAL:%.+]] = load i{{[0-9]+}}, i{{[0-9]+}}* [[T_VAR_PRIV1]], 382 // CHECK: store i{{[0-9]+}} [[T_VAR_VAL]], i{{[0-9]+}}* [[T_VAR_ADDR_REF1]], 383 // CHECK: [[BCAST_VEC_ADDR_REF:%.+]] = bitcast [2 x i{{[0-9]+}}]* [[VEC_ADDR_REF1]] to i8* 384 // CHECK: [[BCAST_VEC_PRIV:%.+]] = bitcast [2 x i{{[0-9]+}}]* [[VEC_PRIV1]] to i8* 385 // CHECK: call void @llvm.memcpy.{{.+}}(i8* align {{[0-9]+}} [[BCAST_VEC_ADDR_REF]], i8* align {{[0-9]+}} [[BCAST_VEC_PRIV]], 386 // 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]+}} 0 387 // CHECK: [[S_ARR_PRIV_BCAST:%.+]] = bitcast [2 x [[S_INT_TY]]]* [[S_ARR_PRIV1]] to [[S_INT_TY]]* 388 // CHECK: [[S_ARR_BEGIN_GEP:%.+]] = getelementptr [[S_INT_TY]], [[S_INT_TY]]* [[S_ARR_BEGIN]], i{{[0-9]+}} 2 389 // CHECK: [[S_ARR_IS_EMPTY:%.+]] = icmp eq [[S_INT_TY]]* [[S_ARR_BEGIN]], [[S_ARR_BEGIN_GEP]] 390 // CHECK: br i1 [[S_ARR_IS_EMPTY]], label %[[S_ARR_COPY_DONE:.+]], label %[[S_ARR_COPY_BLOCK:.+]] 391 // CHECK: [[S_ARR_COPY_BLOCK]]: 392 // CHECK: [[S_ARR_SRC_EL:%.+]] = phi [[S_INT_TY]]*{{.+}} 393 // CHECK: [[S_ARR_DST_EL:%.+]] = phi [[S_INT_TY]]*{{.+}} 394 // CHECK: [[S_ARR_DST_BCAST:%.+]] = bitcast [[S_INT_TY]]* [[S_ARR_DST_EL]] to i8* 395 // CHECK: [[S_ARR_SRC_BCAST:%.+]] = bitcast [[S_INT_TY]]* [[S_ARR_SRC_EL]] to i8* 396 // CHECK: call void @llvm.memcpy.{{.+}}(i8* align {{[0-9]+}} [[S_ARR_DST_BCAST]], i8* align {{[0-9]+}} [[S_ARR_SRC_BCAST]]{{.+}}) 397 // CHECK: [[S_ARR_DST_NEXT:%.+]] = getelementptr [[S_INT_TY]], [[S_INT_TY]]* [[S_ARR_DST_EL]], i{{[0-9]+}} 1 398 // CHECK: [[S_ARR_SRC_NEXT:%.+]] = getelementptr{{.+}} 399 // CHECK: [[CPY_IS_FINISHED:%.+]] = icmp eq [[S_INT_TY]]* [[S_ARR_DST_NEXT]], [[S_ARR_BEGIN_GEP]] 400 // CHECK: br i1 [[CPY_IS_FINISHED]], label %[[S_ARR_COPY_DONE]], label %[[S_ARR_COPY_BLOCK]] 401 // CHECK: [[S_ARR_COPY_DONE]]: 402 // CHECK: [[TMP_VAL1:%.+]] = load [[S_INT_TY]]*, [[S_INT_TY]]** [[TMP_PRIV1]], 403 // CHECK: [[VAR_ADDR_REF_BCAST:%.+]] = bitcast [[S_INT_TY]]* [[TMP_REF]] to i8* 404 // CHECK: [[TMP_VAL1_BCAST:%.+]] = bitcast [[S_INT_TY]]* [[TMP_VAL1]] to i8* 405 // CHECK: call void @llvm.memcpy.{{.+}}(i8* align {{[0-9]+}} [[VAR_ADDR_REF_BCAST]], i8* align {{[0-9]+}} [[TMP_VAL1_BCAST]],{{.+}}) 406 // CHECK: ret void 407 408 // CHECK: !{!"llvm.loop.vectorize.enable", i1 true} 409 #endif 410