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 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 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 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 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 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 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 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 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 54 #pragma omp distribute simd firstprivate(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_mapper( 74 // LAMBDA: call void [[OFFLOADING_FUN:@.+]]( 75 76 // LAMBDA: define{{.+}} void [[OFFLOADING_FUN]]( 77 // LAMBDA: call {{.*}}void {{.+}} @__kmpc_fork_teams({{.+}}, {{.+}}, {{.+}}* [[OMP_OUTLINED:@.+]] to {{.+}}) 78 #pragma omp target 79 #pragma omp teams 80 #pragma omp distribute simd firstprivate(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 // Private alloca's for conversion 84 // LAMBDA: [[G_ADDR:%.+]] = alloca double*, 85 // LAMBDA: [[G1_ADDR:%.+]] = alloca double*, 86 // LAMBDA: [[SVAR_ADDR:%.+]] = alloca i{{[0-9]+}}*, 87 // LAMBDA: [[SFVAR_ADDR:%.+]] = alloca float*, 88 // LAMBDA: [[G1_REF:%.+]] = alloca double*, 89 90 // Actual private variables to be used in the body (tmp is used for the reference type) 91 // LAMBDA: [[G_PRIVATE:%.+]] = alloca double, 92 // LAMBDA: [[G1_PRIVATE:%.+]] = alloca double, 93 // LAMBDA: [[TMP_PRIVATE:%.+]] = alloca double*, 94 // LAMBDA: [[SVAR_PRIVATE:%.+]] = alloca i{{[0-9]+}}, 95 // LAMBDA: [[SFVAR_PRIVATE:%.+]] = alloca float, 96 97 // Store input parameter addresses into private alloca's for conversion 98 // LAMBDA: store double* [[G_IN]], double** [[G_ADDR]], 99 // LAMBDA: store double* [[G1_IN]], double** [[G1_ADDR]], 100 // LAMBDA: store i{{[0-9]+}}* [[SVAR_IN]], i{{[0-9]+}}** [[SVAR_ADDR]], 101 // LAMBDA: store float* [[SFVAR_IN]], float** [[SFVAR_ADDR]], 102 103 // LAMBDA-DAG: [[G_ADDR_VAL:%.+]] = load double*, double** [[G_ADDR]], 104 // LAMBDA-DAG: [[G1_ADDR_VAL:%.+]] = load double*, double** [[G1_ADDR]], 105 // LAMBDA-DAG: [[SVAR_ADDR_VAL:%.+]] = load i{{[0-9]+}}*, i{{[0-9]+}}** [[SVAR_ADDR]], 106 // LAMBDA-DAG: [[SFVAR_ADDR_VAL:%.+]] = load float*, float** [[SFVAR_ADDR]], 107 // LAMBDA-DAG: store double* [[G1_ADDR_VAL]], double** [[G1_REF]], 108 109 // LAMBDA-DAG: [[G_VAL:%.+]] = load{{.+}} double, double* [[G_ADDR_VAL]], 110 // LAMBDA-DAG: store double [[G_VAL]], double* [[G_PRIVATE]], 111 // LAMBDA-DAG: [[G1_VAL_REF:%.+]] = load double*, double** [[G1_REF]], 112 // LAMBDA-DAG: [[G1_VAL:%.+]] = load{{.+}} double, double* % 113 // LAMBDA-DAG: store double [[G1_VAL]], double* [[G1_PRIVATE]], 114 // LAMBDA-DAG: store double* [[G1_PRIVATE]], double** [[TMP_PRIVATE]], 115 // LAMBDA-DAG: [[SVAR_VAL:%.+]] = load i{{[0-9]+}}, i{{[0-9]+}}* [[SVAR_ADDR_VAL]], 116 // LAMBDA-DAG: store i{{[0-9]+}} [[SVAR_VAL]], i{{[0-9]+}}* [[SVAR_PRIVATE]], 117 // LAMBDA-DAG: [[SFVAR_VAL:%.+]] = load float, float* [[SFVAR_ADDR_VAL]], 118 // LAMBDA-DAG: store float [[SFVAR_VAL]], float* [[SFVAR_PRIVATE]], 119 120 // LAMBDA: call {{.*}}void @__kmpc_for_static_init_4( 121 g += 1; 122 g1 += 1; 123 svar += 3; 124 sfvar += 4.0; 125 // LAMBDA-DAG: [[G_VAL:%.+]] = load double, double* [[G_PRIVATE]], 126 // LAMBDA-DAG: [[G_NEXT:%.+]] = fadd double [[G_VAL]], 1.{{.+}} 127 // LAMBDA-DAG: store double [[G_NEXT]], double* [[G_PRIVATE]], 128 // LAMBDA-DAG: [[TMP_VAL1:%.+]] = load double*, double** [[TMP_PRIVATE]], 129 // LAMBDA-DAG: [[TMP_VAL_VAL1:%.+]] = load{{.*}} double, double* [[TMP_VAL1]], 130 // LAMBDA-DAG: [[TMP_ADD:%.+]] = fadd double [[TMP_VAL_VAL1]], 1.{{.+}} 131 // LAMBDA-DAG: store{{.*}} double [[TMP_ADD]], double* [[TMP_VAL1]], 132 // LAMBDA-DAG: [[SVAR_VAL:%.+]] = load i{{[0-9]+}}, i{{[0-9]+}}* [[SVAR_PRIVATE]], 133 // LAMBDA-DAG: [[SVAR_ADD:%.+]] = add{{.*}} i{{[0-9]+}} [[SVAR_VAL]], 3 134 // LAMBDA-DAG: store i{{[0-9]+}} [[SVAR_ADD]], i{{[0-9]+}}* [[SVAR_PRIVATE]], 135 // LAMBDA-DAG: [[SFVAR_VAL:%.+]] = load float, float* [[SFVAR_PRIVATE]], 136 // LAMBDA-DAG: [[SFVAR_CONV_VAL1:%.+]] = fpext float [[SFVAR_VAL]] to double 137 // LAMBDA-DAG: [[SFVAR_ADD:%.+]] = fadd double [[SFVAR_CONV_VAL1]], 4.{{.+}} 138 // LAMBDA-DAG: [[SFVAR_CONV_VAL2:%.+]] = fptrunc double [[SFVAR_ADD]] to float 139 // LAMBDA-DAG: store float [[SFVAR_CONV_VAL2:%.+]], float* [[SFVAR_PRIVATE]], 140 141 // call inner lambda (use refs to private alloca's) 142 // LAMBDA: [[GEP_0:%.+]] = getelementptr{{.+}}, i{{[0-9]+}} 0, i{{[0-9]+}} 0 143 // LAMBDA: store double* [[G_PRIVATE]], double** [[GEP_0]], 144 // LAMBDA: [[GEP_1:%.+]] = getelementptr{{.+}}, i{{[0-9]+}} 0, i{{[0-9]+}} 1 145 // LAMBDA: [[TMP_PAR:%.+]] = load double*, double** [[TMP_PRIVATE]], 146 // LAMBDA: store double* [[TMP_PAR]], double** [[GEP_1]], 147 // LAMBDA: [[GEP_2:%.+]] = getelementptr{{.+}}, i{{[0-9]+}} 0, i{{[0-9]+}} 2 148 // LAMBDA: store i{{[0-9]+}}* [[SVAR_PRIVATE]], i{{[0-9]+}}** [[GEP_2]], 149 // LAMBDA: [[GEP_3:%.+]] = getelementptr{{.+}}, i{{[0-9]+}} 0, i{{[0-9]+}} 3 150 // LAMBDA: store float* [[SFVAR_PRIVATE]], float** [[GEP_3]], 151 // LAMBDA: call{{.*}} void [[INNER_LAMBDA:@.+]](%{{.+}}* {{.+}}) 152 // LAMBDA: call {{.*}}void @__kmpc_for_static_fini( 153 [&]() { 154 // LAMBDA: define {{.+}} void [[INNER_LAMBDA]](%{{.+}}* [[ARG_PTR:%.+]]) 155 // LAMBDA: store %{{.+}}* [[ARG_PTR]], %{{.+}}** [[ARG_PTR_REF:%.+]], 156 g += 2; 157 g1 += 2; 158 svar += 4; 159 sfvar += 8.0; 160 // LAMBDA-DAG: [[ARG_PTR:%.+]] = load %{{.+}}*, %{{.+}}** [[ARG_PTR_REF]] 161 // LAMBDA-DAG: [[G_PTR_REF:%.+]] = getelementptr inbounds %{{.+}}, %{{.+}}* [[ARG_PTR]], i{{[0-9]+}} 0, i{{[0-9]+}} 0 162 // LAMBDA-DAG: [[G_REF:%.+]] = load double*, double** [[G_PTR_REF]], 163 // LAMBDA-DAG: [[G_REF_VAL:%.+]] = load double, double* [[G_REF]], 164 // LAMBDA-DAG: [[G_REF_ADD:%.+]] = fadd double [[G_REF_VAL]], 2.{{.+}} 165 // LAMBDA-DAG: store double [[G_REF_ADD]], double* [[G_REF]] 166 167 // LAMBDA-DAG: [[TMP_PTR_REF:%.+]] = getelementptr inbounds %{{.+}}, %{{.+}}* [[ARG_PTR]], i{{[0-9]+}} 0, i{{[0-9]+}} 1 168 // LAMBDA-DAG: [[G1_REF:%.+]] = load double*, double** [[TMP_PTR_REF]] 169 // LAMBDA-DAG: [[G1_REF_VAL:%.+]] = load double, double* [[G1_REF]], 170 // LAMBDA-DAG: [[G1_ADD:%.+]] = fadd double [[G1_REF_VAL]], 2.{{.+}} 171 // LAMBDA-DAG: store double [[G1_ADD]], double* [[G1_REF]], 172 173 // LAMBDA-DAG: [[SVAR_PTR_REF:%.+]] = getelementptr inbounds %{{.+}}, %{{.+}}* [[ARG_PTR]], i{{[0-9]+}} 0, i{{[0-9]+}} 2 174 // LAMBDA-DAG: [[SVAR_REF:%.+]] = load i{{[0-9]+}}*, i{{[0-9]+}}** [[SVAR_PTR_REF]] 175 // LAMBDA-DAG: [[SVAR_REF_VAL:%.+]] = load i{{[0-9]+}}, i{{[0-9]+}}* [[SVAR_REF]] 176 // LAMBDA-DAG: [[SVAR_ADD:%.+]] = add{{.*}} i{{[0-9]+}} [[SVAR_REF_VAL]], 4 177 // LAMBDA-DAG: store i{{[0-9]+}} [[SVAR_ADD]], i{{[0-9]+}}* [[SVAR_REF]] 178 179 // LAMBDA-DAG: [[SFVAR_PTR_REF:%.+]] = getelementptr inbounds %{{.+}}, %{{.+}}* [[ARG_PTR]], i{{[0-9]+}} 0, i{{[0-9]+}} 3 180 // LAMBDA-DAG: [[SFVAR_REF:%.+]] = load float*, float** [[SFVAR_PTR_REF]] 181 // LAMBDA-DAG: [[SFVAR_REF_VAL:%.+]] = load float, float* [[SFVAR_REF]] 182 // LAMBDA-DAG: [[SFVAR_REF_CONV:%.+]] = fpext float [[SFVAR_REF_VAL]] to double 183 // LAMBDA-DAG: [[SFVAR_ADD:%.+]] = fadd double [[SFVAR_REF_CONV]], 8.{{.+}} 184 // LAMBDA-DAG: [[SFVAR_ADD_CONV:%.+]] = fptrunc double [[SFVAR_ADD]] to float 185 // LAMBDA-DAG: store float [[SFVAR_ADD_CONV]], float* [[SFVAR_REF]], 186 }(); 187 } 188 }(); 189 return 0; 190 #else 191 S<float> test; 192 int t_var = 0; 193 int vec[] = {1, 2}; 194 S<float> s_arr[] = {1, 2}; 195 S<float> &var = test; 196 197 #pragma omp target 198 #pragma omp teams 199 #pragma omp distribute simd firstprivate(t_var, vec, s_arr, s_arr, var, var, svar) 200 for (int i = 0; i < 2; ++i) { 201 vec[i] = t_var; 202 s_arr[i] = var; 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_mapper( 212 // CHECK: call void [[OFFLOAD_FUN:@.+]]( 213 // CHECK: ret 214 215 // CHECK: define{{.+}} [[OFFLOAD_FUN]]( 216 // CHECK: call void (%{{.+}}*, i{{[0-9]+}}, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)*, ...) @__kmpc_fork_teams(%{{.+}}* @{{.+}}, i{{[0-9]+}} 5, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)* bitcast (void (i{{[0-9]+}}*, i{{[0-9]+}}*, i{{[0-9]+}}*, [2 x i{{[0-9]+}}]*, [2 x [[S_FLOAT_TY]]]*, [[S_FLOAT_TY]]*, i{{[0-9]+}}*)* [[OMP_OUTLINED:@.+]] to void 217 // CHECK: ret 218 // 219 // CHECK: define internal void [[OMP_OUTLINED]](i{{[0-9]+}}*{{.+}}, i{{[0-9]+}}*{{.+}}, 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]+}}*{{.+}} [[SVAR_IN:%.+]]) 220 221 // CHECK: alloca i{{[0-9]+}}*, 222 // CHECK: alloca i{{[0-9]+}}*, 223 // CHECK: [[T_VAR_ADDR:%.+]] = alloca i{{[0-9]+}}*, 224 // CHECK: [[VEC_ADDR:%.+]] = alloca [2 x i{{[0-9]+}}]*, 225 // CHECK: [[S_ARR_ADDR:%.+]] = alloca [2 x [[S_FLOAT_TY]]]*, 226 // CHECK: [[VAR_ADDR:%.+]] = alloca [[S_FLOAT_TY]]*, 227 // CHECK: [[SVAR_ADDR:%.+]] = alloca i{{[0-9]+}}*, 228 // CHECK: [[TMP:%.+]] = alloca [[S_FLOAT_TY]]*, 229 230 // discard omp loop variables 231 // CHECK: {{.*}} = alloca i{{[0-9]+}}, 232 // CHECK: {{.*}} = alloca i{{[0-9]+}}, 233 // CHECK: {{.*}} = alloca i{{[0-9]+}}, 234 // CHECK: {{.*}} = alloca i{{[0-9]+}}, 235 // CHECK: {{.*}} = alloca i{{[0-9]+}}, 236 // CHECK: {{.*}} = alloca i{{[0-9]+}}, 237 238 // CHECK-DAG: [[T_VAR_PRIV:%.+]] = alloca i{{[0-9]+}}, 239 // CHECK-DAG: [[VEC_PRIV:%.+]] = alloca [2 x i{{[0-9]+}}], 240 // CHECK-DAG: [[S_ARR_PRIV:%.+]] = alloca [2 x [[S_FLOAT_TY]]], 241 // CHECK-DAG: [[VAR_PRIV:%.+]] = alloca [[S_FLOAT_TY]], 242 // CHECK-DAG: [[TMP_PRIV:%.+]] = alloca [[S_FLOAT_TY]]*, 243 // CHECK: [[SVAR_PRIV:%.+]] = alloca i{{[0-9]+}}, 244 245 // CHECK: store i{{[0-9]+}}* [[T_VAR_IN]], i{{[0-9]+}}** [[T_VAR_ADDR]], 246 // CHECK: store [2 x i{{[0-9]+}}]* [[VEC_IN]], [2 x i{{[0-9]+}}]** [[VEC_ADDR]], 247 // CHECK: store [2 x [[S_FLOAT_TY]]]* [[S_ARR_IN]], [2 x [[S_FLOAT_TY]]]** [[S_ARR_ADDR]], 248 // CHECK: store [[S_FLOAT_TY]]* [[VAR_IN]], [[S_FLOAT_TY]]** [[VAR_ADDR]], 249 // CHECK: store i{{[0-9]+}}* [[SVAR_IN]], i{{[0-9]+}}** [[SVAR_ADDR]], 250 251 // init t_var 252 // CHECK-DAG: [[T_VAR_ADDR_CONV_VAL_REF:%.+]] = load i{{[0-9]+}}*, i{{[0-9]+}}** [[T_VAR_ADDR]], 253 // CHECK-DAG: [[T_VAR_ADDR_CONV_VAL:%.+]] = load i{{[0-9]+}}, i{{[0-9]+}}* [[T_VAR_ADDR_CONV_VAL_REF]], 254 // CHECK-DAG: store i{{[0-9]+}} [[T_VAR_ADDR_CONV_VAL]], i{{[0-9]+}}* [[T_VAR_PRIV]], 255 256 // init vec 257 // CHECK-DAG: [[VEC_ADDR_VAL:%.+]] = load [2 x i{{[0-9]+}}]*, [2 x i{{[0-9]+}}]** [[VEC_ADDR]], 258 // CHECK-DAG: [[VEC_ADDR_VAL_BCAST:%.+]] = bitcast [2 x i{{[0-9]+}}]* [[VEC_ADDR_VAL]] to i{{[0-9]+}}* 259 // CHECK-DAG: [[VEC_PRIV_BCAST:%.+]] = bitcast [2 x i{{[0-9]+}}]* [[VEC_PRIV]] to i{{[0-9]+}}* 260 // CHECK-DAG: call void @llvm.memcpy.{{.*}}(i{{[0-9]+}}* align {{[0-9]+}} [[VEC_PRIV_BCAST]], i{{[0-9]+}}* align {{[0-9]+}} [[VEC_ADDR_VAL_BCAST]],{{.+}}) 261 262 // init s_arr 263 // CHECK-DAG: [[S_ARR_ADDR_VAL:%.+]] = load [2 x [[S_FLOAT_TY]]]*, [2 x [[S_FLOAT_TY]]]** [[S_ARR_ADDR]], 264 // CHECK-DAG: [[S_ARR_ADDR_BCAST:%.+]] = bitcast [2 x [[S_FLOAT_TY]]]* [[S_ARR_ADDR_VAL]] to [[S_FLOAT_TY]]* 265 // CHECK-DAG: [[S_ARR_PRIV_BGN:%.+]] = getelementptr{{.+}} [2 x [[S_FLOAT_TY]]], [2 x [[S_FLOAT_TY]]]* [[S_ARR_PRIV]]{{.+}} 266 // CHECK-DAG: [[S_ARR_PRIV_NEXT:%.+]] = getelementptr [[S_FLOAT_TY]], [[S_FLOAT_TY]]* [[S_ARR_PRIV_BGN]]{{.+}} 267 // CHECK-DAG: [[S_ARR_IS_EMPTY:%.+]] = icmp eq [[S_FLOAT_TY]]* [[S_ARR_PRIV_BGN]], [[S_ARR_PRIV_NEXT]] 268 // CHECK-DAG: br i1 [[S_ARR_IS_EMPTY]], label %[[S_ARR_CPY_DONE:.+]], label %[[S_ARR_CPY_BODY:.+]] 269 270 // CHECK-DAG: [[S_ARR_CPY_BODY]]: 271 // CHECK-DAG: [[S_ARR_SRC_PAST:%.+]] = phi{{.+}} [ [[S_ARR_ADDR_BCAST]],{{.+}} ], [ [[S_ARR_SRC:%.+]],{{.+}} ] 272 // CHECK-DAG: [[S_ARR_DST_PAST:%.+]] = phi{{.+}} [ [[S_ARR_PRIV_BGN]],{{.+}} ], [ [[S_ARR_DST:%.+]],{{.+}} ] 273 // CHECK-DAG: [[S_ARR_SRC_BCAST:%.+]] = bitcast{{.+}} [[S_ARR_SRC_PAST]] to{{.+}} 274 // CHECK-DAG: [[S_ARR_DST_BCAST:%.+]] = bitcast{{.+}} [[S_ARR_DST_PAST]] to{{.+}} 275 // CHECK-DAG: call{{.+}} @llvm.memcpy.{{.+}}({{.+}}* align {{[0-9]+}} [[S_ARR_DST_BCAST]], {{.+}}* align {{[0-9]+}} [[S_ARR_SRC_BCAST]]{{.+}}) 276 // CHECK-DAG: [[S_ARR_SRC]] = getelementptr{{.+}} 277 // CHECK-DAG: [[S_ARR_DST]] = getelementptr{{.+}} 278 // CHECK-DAG: [[S_ARR_CPY_FIN:%.+]] = icmp{{.+}} [[S_ARR_DST]], [[S_ARR_PRIV_NEXT]] 279 // CHECK-DAG: br i1 [[S_ARR_CPY_FIN]], label %[[S_ARR_CPY_DONE]], label %[[S_ARR_CPY_BODY]] 280 // CHECK-DAG: [[S_ARR_CPY_DONE]]: 281 282 // init var 283 // CHECK-DAG: [[VAR_ADDR_VAL:%.+]] = load [[S_FLOAT_TY]]*, [[S_FLOAT_TY]]** [[VAR_ADDR]], 284 // CHECK-DAG: store{{.+}} [[VAR_ADDR_VAL]],{{.+}} [[TMP]], 285 // CHECK-DAG: [[TMP_VAL:%.+]] = load [[S_FLOAT_TY]]*, [[S_FLOAT_TY]]** [[TMP]], 286 // CHECK-DAG: [[VAR_PRIV_BCAST:%.+]] = bitcast [[S_FLOAT_TY]]* [[VAR_PRIV]] to{{.+}} 287 // CHECK-DAG: [[TMP_BCAST:%.+]] = bitcast [[S_FLOAT_TY]]* %{{.+}} to{{.+}} 288 // CHECK-DAG: call{{.+}} @llvm.memcpy.{{.+}}({{.+}}* align {{[0-9]+}} [[VAR_PRIV_BCAST]], {{.+}}* align {{[0-9]+}} [[TMP_BCAST]],{{.+}}) 289 // CHECK-DAG: store [[S_FLOAT_TY]]* [[VAR_PRIV]], [[S_FLOAT_TY]]** [[TMP_PRIV]], 290 291 // init svar 292 // CHECK-DAG: [[SVAR_CONV_VAL_REF:%.+]] = load{{.+}},{{.+}} [[SVAR_ADDR]], 293 // CHECK-DAG: [[SVAR_CONV_VAL:%.+]] = load{{.+}},{{.+}} [[SVAR_CONV_VAL_REF]], 294 // CHECK-DAG: store{{.+}} [[SVAR_CONV_VAL]],{{.+}} [[SVAR_PRIV]], 295 296 // CHECK-DAG: store i{{[0-9]+}} 0, i{{[0-9]+}}* %.omp{{.+}}, 297 // CHECK-DAG: store i{{[0-9]+}} 1, i{{[0-9]+}}* %.omp{{.+}}, 298 // CHECK-DAG: store i{{[0-9]+}} 1, i{{[0-9]+}}* %.omp{{.+}}, 299 // CHECK-DAG: store i{{[0-9]+}} 0, i{{[0-9]+}}* %.omp{{.+}}, 300 301 // CHECK: call void @__kmpc_for_static_init_4( 302 // CHECK: call void @__kmpc_for_static_fini( 303 // CHECK: ret void 304 305 // Template 306 // CHECK: define{{.*}} i{{[0-9]+}} [[TMAIN_INT:@.+]]() 307 // CHECK: [[TEST:%.+]] = alloca [[S_INT_TY]], 308 // CHECK: call {{.*}} [[S_INT_TY_DEF_CONSTR:@.+]]([[S_INT_TY]]* [[TEST]]) 309 // CHECK: call i{{[0-9]+}} @__tgt_target_teams_mapper( 310 // CHECK: call void [[OFFLOAD_FUN_1:@.+]]( 311 // CHECK: ret 312 313 // CHECK: define{{.+}} [[OFFLOAD_FUN_1]]( 314 // CHECK: call void (%{{.+}}*, i{{[0-9]+}}, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)*, ...) @__kmpc_fork_teams(%{{.+}}* @{{.+}}, i{{[0-9]+}} 4, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)* bitcast (void (i{{[0-9]+}}*, i{{[0-9]+}}*, i{{[0-9]+}}*, [2 x i{{[0-9]+}}]*, [2 x [[S_INT_TY]]]*, [[S_INT_TY]]*)* [[OMP_OUTLINED_1:@.+]] to void 315 // CHECK: ret 316 // 317 // CHECK: define internal void [[OMP_OUTLINED_1]](i{{[0-9]+}}*{{.+}}, i{{[0-9]+}}*{{.+}}, i{{[0-9]+}}*{{.+}} [[T_VAR_IN:%.+]], [2 x i{{[0-9]+}}]*{{.*}} [[VEC_IN:%.+]], [2 x [[S_INT_TY]]]*{{.*}} [[S_ARR_IN:%.+]], [[S_INT_TY]]*{{.*}} [[VAR_IN:%.+]]) 318 319 // CHECK: alloca i{{[0-9]+}}*, 320 // CHECK: alloca i{{[0-9]+}}*, 321 // CHECK: [[T_VAR_ADDR:%.+]] = alloca i{{[0-9]+}}*, 322 // CHECK: [[VEC_ADDR:%.+]] = alloca [2 x i{{[0-9]+}}]*, 323 // CHECK: [[S_ARR_ADDR:%.+]] = alloca [2 x [[S_INT_TY]]]*, 324 // CHECK: [[VAR_ADDR:%.+]] = alloca [[S_INT_TY]]*, 325 // CHECK: [[TMP:%.+]] = alloca [[S_INT_TY]]*, 326 327 // discard omp loop variables 328 // CHECK: {{.*}} = alloca i{{[0-9]+}}, 329 // CHECK: {{.*}} = alloca i{{[0-9]+}}, 330 // CHECK: {{.*}} = alloca i{{[0-9]+}}, 331 // CHECK: {{.*}} = alloca i{{[0-9]+}}, 332 // CHECK: {{.*}} = alloca i{{[0-9]+}}, 333 // CHECK: {{.*}} = alloca i{{[0-9]+}}, 334 335 // CHECK-DAG: [[T_VAR_PRIV:%.+]] = alloca i{{[0-9]+}}, 336 // CHECK-DAG: [[VEC_PRIV:%.+]] = alloca [2 x i{{[0-9]+}}], 337 // CHECK-DAG: [[S_ARR_PRIV:%.+]] = alloca [2 x [[S_INT_TY]]], 338 // CHECK-DAG: [[VAR_PRIV:%.+]] = alloca [[S_INT_TY]], 339 // CHECK-DAG: [[TMP_PRIV:%.+]] = alloca [[S_INT_TY]]*, 340 341 // CHECK: store i{{[0-9]+}}* [[T_VAR_IN]], i{{[0-9]+}}** [[T_VAR_ADDR]], 342 // CHECK: store [2 x i{{[0-9]+}}]* [[VEC_IN]], [2 x i{{[0-9]+}}]** [[VEC_ADDR]], 343 // CHECK: store [2 x [[S_INT_TY]]]* [[S_ARR_IN]], [2 x [[S_INT_TY]]]** [[S_ARR_ADDR]], 344 // CHECK: store [[S_INT_TY]]* [[VAR_IN]], [[S_INT_TY]]** [[VAR_ADDR]], 345 346 // init t_var 347 // CHECK-DAG: [[T_VAR_ADDR_CONV_VAL_REF:%.+]] = load i{{[0-9]+}}*, i{{[0-9]+}}** [[T_VAR_ADDR]], 348 // CHECK-DAG: [[T_VAR_ADDR_CONV_VAL:%.+]] = load i{{[0-9]+}}, i{{[0-9]+}}* [[T_VAR_ADDR_CONV_VAL_REF]], 349 // CHECK-DAG: store i{{[0-9]+}} [[T_VAR_ADDR_CONV_VAL]], i{{[0-9]+}}* [[T_VAR_PRIV]], 350 351 // init vec 352 // CHECK-DAG: [[VEC_ADDR_VAL:%.+]] = load [2 x i{{[0-9]+}}]*, [2 x i{{[0-9]+}}]** [[VEC_ADDR]], 353 // CHECK-DAG: [[VEC_ADDR_VAL_BCAST:%.+]] = bitcast [2 x i{{[0-9]+}}]* [[VEC_ADDR_VAL]] to i{{[0-9]+}}* 354 // CHECK-DAG: [[VEC_PRIV_BCAST:%.+]] = bitcast [2 x i{{[0-9]+}}]* [[VEC_PRIV]] to i{{[0-9]+}}* 355 // CHECK-DAG: call void @llvm.memcpy.{{.*}}(i{{[0-9]+}}* align {{[0-9]+}} [[VEC_PRIV_BCAST]], i{{[0-9]+}}* align {{[0-9]+}} [[VEC_ADDR_VAL_BCAST]],{{.+}}) 356 357 // init s_arr 358 // CHECK-DAG: [[S_ARR_ADDR_VAL:%.+]] = load [2 x [[S_INT_TY]]]*, [2 x [[S_INT_TY]]]** [[S_ARR_ADDR]], 359 // CHECK-DAG: [[S_ARR_ADDR_BCAST:%.+]] = bitcast [2 x [[S_INT_TY]]]* [[S_ARR_ADDR_VAL]] to [[S_INT_TY]]* 360 // CHECK-DAG: [[S_ARR_PRIV_BGN:%.+]] = getelementptr{{.+}} [2 x [[S_INT_TY]]], [2 x [[S_INT_TY]]]* [[S_ARR_PRIV]]{{.+}} 361 // CHECK-DAG: [[S_ARR_PRIV_NEXT:%.+]] = getelementptr [[S_INT_TY]], [[S_INT_TY]]* [[S_ARR_PRIV_BGN]]{{.+}} 362 // CHECK-DAG: [[S_ARR_IS_EMPTY:%.+]] = icmp eq [[S_INT_TY]]* [[S_ARR_PRIV_BGN]], [[S_ARR_PRIV_NEXT]] 363 // CHECK-DAG: br i1 [[S_ARR_IS_EMPTY]], label %[[S_ARR_CPY_DONE:.+]], label %[[S_ARR_CPY_BODY:.+]] 364 365 // CHECK-DAG: [[S_ARR_CPY_BODY]]: 366 // CHECK-DAG: [[S_ARR_SRC_PAST:%.+]] = phi{{.+}} [ [[S_ARR_ADDR_BCAST]],{{.+}} ], [ [[S_ARR_SRC:%.+]],{{.+}} ] 367 // CHECK-DAG: [[S_ARR_DST_PAST:%.+]] = phi{{.+}} [ [[S_ARR_PRIV_BGN]],{{.+}} ], [ [[S_ARR_DST:%.+]],{{.+}} ] 368 // CHECK-DAG: [[S_ARR_SRC_BCAST:%.+]] = bitcast{{.+}} [[S_ARR_SRC_PAST]] to{{.+}} 369 // CHECK-DAG: [[S_ARR_DST_BCAST:%.+]] = bitcast{{.+}} [[S_ARR_DST_PAST]] to{{.+}} 370 // CHECK-DAG: call{{.+}} @llvm.memcpy.{{.+}}({{.+}}* align {{[0-9]+}} [[S_ARR_DST_BCAST]], {{.+}}* align {{[0-9]+}} [[S_ARR_SRC_BCAST]]{{.+}}) 371 // CHECK-DAG: [[S_ARR_SRC]] = getelementptr{{.+}} 372 // CHECK-DAG: [[S_ARR_DST]] = getelementptr{{.+}} 373 // CHECK-DAG: [[S_ARR_CPY_FIN:%.+]] = icmp{{.+}} [[S_ARR_DST]], [[S_ARR_PRIV_NEXT]] 374 // CHECK-DAG: br i1 [[S_ARR_CPY_FIN]], label %[[S_ARR_CPY_DONE]], label %[[S_ARR_CPY_BODY]] 375 // CHECK-DAG: [[S_ARR_CPY_DONE]]: 376 377 // init var 378 // CHECK-DAG: [[VAR_ADDR_VAL:%.+]] = load [[S_INT_TY]]*, [[S_INT_TY]]** [[VAR_ADDR]], 379 // CHECK-DAG: store{{.+}} [[VAR_ADDR_VAL]],{{.+}} [[TMP]], 380 // CHECK-DAG: [[TMP_VAL:%.+]] = load [[S_INT_TY]]*, [[S_INT_TY]]** [[TMP]], 381 // CHECK-DAG: [[VAR_PRIV_BCAST:%.+]] = bitcast [[S_INT_TY]]* [[VAR_PRIV]] to{{.+}} 382 // CHECK-DAG: [[TMP_BCAST:%.+]] = bitcast [[S_INT_TY]]* %{{.+}} to{{.+}} 383 // CHECK-DAG: call{{.+}} @llvm.memcpy.{{.+}}({{.+}}* align {{[0-9]+}} [[VAR_PRIV_BCAST]], {{.+}}* align {{[0-9]+}} [[TMP_BCAST]],{{.+}}) 384 // CHECK-DAG: store [[S_INT_TY]]* [[VAR_PRIV]], [[S_INT_TY]]** [[TMP_PRIV]], 385 386 // CHECK-DAG: store i{{[0-9]+}} 0, i{{[0-9]+}}* %.omp{{.+}}, 387 // CHECK-DAG: store i{{[0-9]+}} 1, i{{[0-9]+}}* %.omp{{.+}}, 388 // CHECK-DAG: store i{{[0-9]+}} 1, i{{[0-9]+}}* %.omp{{.+}}, 389 // CHECK-DAG: store i{{[0-9]+}} 0, i{{[0-9]+}}* %.omp{{.+}}, 390 391 // CHECK: call void @__kmpc_for_static_init_4( 392 // CHECK: call void @__kmpc_for_static_fini( 393 // CHECK: ret void 394 395 // CHECK: !{!"llvm.loop.vectorize.enable", i1 true} 396 #endif 397