1 // RUN: %clang_cc1 -DCHECK -verify -fopenmp -x c++ -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -emit-llvm %s -o - | FileCheck -allow-deprecated-dag-overlap %s --check-prefix CHECK --check-prefix CHECK-64 2 // RUN: %clang_cc1 -DCHECK -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 -DCHECK -fopenmp -x c++ -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck -allow-deprecated-dag-overlap %s --check-prefix CHECK --check-prefix CHECK-64 4 // RUN: %clang_cc1 -DCHECK -verify -fopenmp -x c++ -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -emit-llvm %s -o - | FileCheck -allow-deprecated-dag-overlap %s --check-prefix CHECK --check-prefix CHECK-32 5 // RUN: %clang_cc1 -DCHECK -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 -DCHECK -fopenmp -x c++ -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck -allow-deprecated-dag-overlap %s --check-prefix CHECK --check-prefix CHECK-32 7 8 // RUN: %clang_cc1 -DCHECK -verify -fopenmp-simd -x c++ -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -emit-llvm %s -o - | FileCheck -allow-deprecated-dag-overlap --check-prefix SIMD-ONLY0 %s 9 // RUN: %clang_cc1 -DCHECK -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 -DCHECK -fopenmp-simd -x c++ -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck -allow-deprecated-dag-overlap --check-prefix SIMD-ONLY0 %s 11 // RUN: %clang_cc1 -DCHECK -verify -fopenmp-simd -x c++ -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -emit-llvm %s -o - | FileCheck -allow-deprecated-dag-overlap --check-prefix SIMD-ONLY0 %s 12 // RUN: %clang_cc1 -DCHECK -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 -DCHECK -fopenmp-simd -x c++ -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck -allow-deprecated-dag-overlap --check-prefix SIMD-ONLY0 %s 14 // SIMD-ONLY0-NOT: {{__kmpc|__tgt}} 15 16 // 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 - | FileCheck -allow-deprecated-dag-overlap %s --check-prefix LAMBDA --check-prefix LAMBDA-64 17 // 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 18 // RUN: %clang_cc1 -DLAMBDA -fopenmp -x c++ -std=c++11 -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck -allow-deprecated-dag-overlap %s --check-prefix LAMBDA --check-prefix LAMBDA-64 19 20 // 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 - | FileCheck -allow-deprecated-dag-overlap --check-prefix SIMD-ONLY1 %s 21 // 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 22 // RUN: %clang_cc1 -DLAMBDA -fopenmp-simd -x c++ -std=c++11 -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck -allow-deprecated-dag-overlap --check-prefix SIMD-ONLY1 %s 23 // SIMD-ONLY1-NOT: {{__kmpc|__tgt}} 24 25 // expected-no-diagnostics 26 #ifndef HEADER 27 #define HEADER 28 29 struct St { 30 int a, b; 31 St() : a(0), b(0) {} 32 St(const St &st) : a(st.a + st.b), b(0) {} 33 ~St() {} 34 }; 35 36 volatile int g = 1212; 37 volatile int &g1 = g; 38 39 template <class T> 40 struct S { 41 T f; 42 S(T a) : f(a + g) {} 43 S() : f(g) {} 44 S(const S &s, St t = St()) : f(s.f + t.a) {} 45 operator T() { return T(); } 46 ~S() {} 47 }; 48 49 // CHECK-DAG: [[S_FLOAT_TY:%.+]] = type { float } 50 // CHECK-DAG: [[S_INT_TY:%.+]] = type { i{{[0-9]+}} } 51 52 template <typename T> 53 T tmain() { 54 S<T> test; 55 T t_var = T(); 56 T vec[] = {1, 2}; 57 S<T> s_arr[] = {1, 2}; 58 S<T> &var = test; 59 #pragma omp target 60 #pragma omp teams distribute simd private(t_var, vec, s_arr, var) 61 for (int i = 0; i < 2; ++i) { 62 vec[i] = t_var; 63 s_arr[i] = var; 64 } 65 return T(); 66 } 67 68 // CHECK-DAG: [[TEST:@.+]] ={{.*}} global [[S_FLOAT_TY]] zeroinitializer, 69 S<float> test; 70 // CHECK-DAG: [[T_VAR:@.+]] ={{.*}} global i{{[0-9]+}} 333, 71 int t_var = 333; 72 // CHECK-DAG: [[VEC:@.+]] ={{.*}} global [2 x i{{[0-9]+}}] [i{{[0-9]+}} 1, i{{[0-9]+}} 2], 73 int vec[] = {1, 2}; 74 // CHECK-DAG: [[S_ARR:@.+]] ={{.*}} global [2 x [[S_FLOAT_TY]]] zeroinitializer, 75 S<float> s_arr[] = {1, 2}; 76 // CHECK-DAG: [[VAR:@.+]] ={{.*}} global [[S_FLOAT_TY]] zeroinitializer, 77 S<float> var(3); 78 // CHECK-DAG: [[SIVAR:@.+]] = internal global i{{[0-9]+}} 0, 79 80 int main() { 81 static int sivar; 82 #ifdef LAMBDA 83 // LAMBDA: [[G:@.+]] ={{.*}} global i{{[0-9]+}} 1212, 84 // LAMBDA-LABEL: @main 85 // LAMBDA: call void [[OUTER_LAMBDA:@.+]]( 86 [&]() { 87 // LAMBDA: define{{.*}} internal{{.*}} void [[OUTER_LAMBDA]]( 88 // LAMBDA: call i32 @__tgt_target_teams_mapper(%struct.ident_t* @{{.+}}, i64 -1, i8* @{{[^,]+}}, i32 1, i8** %{{[^,]+}}, i8** %{{[^,]+}}, i{{64|32}}* {{.+}}@{{[^,]+}}, i32 0, i32 0), i64* {{.+}}@{{[^,]+}}, i32 0, i32 0), i8** null, i8** null, i32 0, i32 1) 89 // LAMBDA: call void @[[LOFFL1:.+]]( 90 // LAMBDA: ret 91 #pragma omp target 92 #pragma omp teams distribute simd private(g, g1, sivar) 93 for (int i = 0; i < 2; ++i) { 94 // LAMBDA: define{{.*}} internal{{.*}} void @[[LOFFL1]](i{{64|32}} {{%.+}}) 95 // LAMBDA: call void {{.+}} @__kmpc_fork_teams({{.+}}, i32 0, {{.+}} @[[LOUTL1:.+]] to {{.+}}) 96 // LAMBDA: ret void 97 98 // LAMBDA: define internal void @[[LOUTL1]]({{.+}}) 99 // Skip global, bound tid and loop vars 100 // LAMBDA: {{.+}} = alloca i32*, 101 // LAMBDA: {{.+}} = alloca i32*, 102 // LAMBDA: alloca i32, 103 // LAMBDA: alloca i32, 104 // LAMBDA: alloca i32, 105 // LAMBDA: alloca i32, 106 // LAMBDA: alloca i32, 107 // LAMBDA: alloca i32, 108 // LAMBDA: [[G_PRIV:%.+]] = alloca i{{[0-9]+}}, 109 // LAMBDA: [[G1_PRIV:%.+]] = alloca i{{[0-9]+}} 110 // LAMBDA: [[TMP:%.+]] = alloca i{{[0-9]+}}*, 111 // LAMBDA: [[SIVAR_PRIV:%.+]] = alloca i{{[0-9]+}}, 112 // LAMBDA: store{{.+}} [[G1_PRIV]], {{.+}} [[TMP]], 113 114 g = 1; 115 g1 = 1; 116 sivar = 2; 117 // LAMBDA: call void @__kmpc_for_static_init_4( 118 // LAMBDA-DAG: store{{.+}} 1, {{.+}} [[G_PRIV]], 119 // LAMBDA-DAG: store{{.+}} 2, {{.+}} [[SIVAR_PRIV]], 120 // LAMBDA-DAG: [[G1_REF:%.+]] = load{{.+}}, {{.+}} [[TMP]], 121 // LAMBDA-DAG: store{{.+}} 1, {{.+}} [[G1_REF]], 122 // LAMBDA: call void [[INNER_LAMBDA:@.+]]( 123 // LAMBDA: call void @__kmpc_for_static_fini( 124 [&]() { 125 // LAMBDA: define {{.+}} void [[INNER_LAMBDA]]({{.+}} [[ARG_PTR:%.+]]) 126 // LAMBDA: store %{{.+}}* [[ARG_PTR]], %{{.+}}** [[ARG_PTR_REF:%.+]], 127 g = 2; 128 g1 = 2; 129 sivar = 4; 130 // LAMBDA: [[ARG_PTR:%.+]] = load %{{.+}}*, %{{.+}}** [[ARG_PTR_REF]] 131 132 // LAMBDA: [[G_PTR_REF:%.+]] = getelementptr inbounds %{{.+}}, %{{.+}}* [[ARG_PTR]], i{{[0-9]+}} 0, i{{[0-9]+}} 0 133 // LAMBDA: [[G_REF:%.+]] = load i{{[0-9]+}}*, i{{[0-9]+}}** [[G_PTR_REF]] 134 // LAMBDA: store i{{[0-9]+}} 2, i{{[0-9]+}}* [[G_REF]] 135 // LAMBDA: [[G1_PTR_REF:%.+]] = getelementptr inbounds %{{.+}}, %{{.+}}* [[ARG_PTR]], i{{[0-9]+}} 0, i{{[0-9]+}} 1 136 // LAMBDA: [[G1_REF:%.+]] = load i{{[0-9]+}}*, i{{[0-9]+}}** [[G1_PTR_REF]] 137 // LAMBDA: store i{{[0-9]+}} 2, i{{[0-9]+}}* [[G1_REF]] 138 // LAMBDA: [[SIVAR_PTR_REF:%.+]] = getelementptr inbounds %{{.+}}, %{{.+}}* [[ARG_PTR]], i{{[0-9]+}} 0, i{{[0-9]+}} 2 139 // LAMBDA: [[SIVAR_REF:%.+]] = load i{{[0-9]+}}*, i{{[0-9]+}}** [[SIVAR_PTR_REF]] 140 // LAMBDA: store i{{[0-9]+}} 4, i{{[0-9]+}}* [[SIVAR_REF]] 141 }(); 142 } 143 }(); 144 return 0; 145 #else 146 #pragma omp target 147 #pragma omp teams distribute simd private(t_var, vec, s_arr, var, sivar) 148 for (int i = 0; i < 2; ++i) { 149 vec[i] = t_var; 150 s_arr[i] = var; 151 sivar += i; 152 } 153 return tmain<int>(); 154 #endif 155 } 156 157 // CHECK: define {{.*}}i{{[0-9]+}} @main() 158 // CHECK: call i32 @__tgt_target_teams_mapper(%struct.ident_t* @{{.+}}, i64 -1, i8* @{{[^,]+}}, i32 0, i8** null, i8** null, i{{64|32}}* null, i64* null, i8** null, i8** null, i32 0, i32 1) 159 // CHECK: call void @[[OFFL1:.+]]() 160 // CHECK: {{%.+}} = call{{.*}} i32 @[[TMAIN_INT:.+]]() 161 // CHECK: ret 162 163 // CHECK: define{{.*}} void @[[OFFL1]]() 164 // CHECK: call void {{.+}} @__kmpc_fork_teams({{.+}}, i32 0, {{.+}} @[[OUTL1:.+]] to {{.+}}) 165 // CHECK: ret void 166 167 // CHECK: define internal void @[[OUTL1]]({{.+}}) 168 // Skip global, bound tid and loop vars 169 // CHECK: {{.+}} = alloca i32*, 170 // CHECK: {{.+}} = alloca i32*, 171 // CHECK: {{.+}} = alloca i32, 172 // CHECK: {{.+}} = alloca i32, 173 // CHECK: {{.+}} = alloca i32, 174 // CHECK: {{.+}} = alloca i32, 175 // CHECK: {{.+}} = alloca i32, 176 // CHECK: {{.+}} = alloca i32, 177 // CHECK-DAG: [[T_VAR_PRIV:%.+]] = alloca i{{[0-9]+}}, 178 // CHECK-DAG: [[VEC_PRIV:%.+]] = alloca [2 x i{{[0-9]+}}], 179 // CHECK-DAG: [[S_ARR_PRIV:%.+]] = alloca [2 x [[S_FLOAT_TY]]], 180 // CHECK-DAG: [[VAR_PRIV:%.+]] = alloca [[S_FLOAT_TY]], 181 // CHECK-DAG: [[SIVAR_PRIV:%.+]] = alloca i{{[0-9]+}}, 182 183 // private(s_arr) 184 // CHECK-DAG: [[S_ARR_PRIV_BGN:%.+]] = getelementptr{{.*}} [2 x [[S_FLOAT_TY]]], [2 x [[S_FLOAT_TY]]]* [[S_ARR_PRIV]], 185 // CHECK-DAG: [[S_ARR_PTR_ALLOC:%.+]] = phi{{.+}} [ [[S_ARR_PRIV_BGN]], {{.+}} ], [ [[S_ARR_NEXT:%.+]], {{.+}} ] 186 // CHECK-DAG: call void @{{.+}}({{.+}} [[S_ARR_PTR_ALLOC]]) 187 // CHECK-DAG: [[S_ARR_NEXT]] = getelementptr {{.+}} [[S_ARR_PTR_ALLOC]], 188 189 // private(var) 190 // CHECK-DAG: call void @{{.+}}({{.+}} [[VAR_PRIV]]) 191 192 // CHECK: call void @__kmpc_for_static_init_4( 193 // CHECK-DAG: {{.+}} = {{.+}} [[T_VAR_PRIV]] 194 // CHECK-DAG: {{.+}} = {{.+}} [[VEC_PRIV]] 195 // CHECK-DAG: {{.+}} = {{.+}} [[S_ARR_PRIV]] 196 // CHECK-DAG: {{.+}} = {{.+}} [[VAR_PRIV]] 197 // CHECK-DAG: {{.+}} = {{.+}} [[SIVAR_PRIV]] 198 // CHECK: call void @__kmpc_for_static_fini( 199 // CHECK: ret void 200 201 202 // CHECK: define{{.*}} i{{[0-9]+}} @[[TMAIN_INT]]() 203 // CHECK: call i32 @__tgt_target_teams_mapper(%struct.ident_t* @{{.+}}, i64 -1, i8* @{{[^,]+}}, i32 0, 204 // CHECK: call void @[[TOFFL1:.+]]() 205 // CHECK: ret 206 207 // CHECK: define {{.*}}void @[[TOFFL1]]() 208 // CHECK: call void {{.+}} @__kmpc_fork_teams({{.+}}, i32 0, {{.+}} @[[TOUTL1:.+]] to {{.+}}) 209 // CHECK: ret void 210 211 // CHECK: define internal void @[[TOUTL1]]({{.+}}) 212 // Skip global, bound tid and loop vars 213 // CHECK: {{.+}} = alloca i32*, 214 // CHECK: {{.+}} = alloca i32*, 215 // CHECK: alloca i{{[0-9]+}}, 216 // CHECK: alloca i{{[0-9]+}}, 217 // CHECK: alloca i{{[0-9]+}}, 218 // CHECK: alloca i{{[0-9]+}}, 219 // CHECK: alloca i{{[0-9]+}}, 220 // CHECK: alloca i{{[0-9]+}}, 221 // CHECK: [[T_VAR_PRIV:%.+]] = alloca i{{[0-9]+}}, 222 // CHECK: [[VEC_PRIV:%.+]] = alloca [2 x i{{[0-9]+}}], 223 // CHECK: [[S_ARR_PRIV:%.+]] = alloca [2 x [[S_INT_TY]]], 224 // CHECK: [[VAR_PRIV:%.+]] = alloca [[S_INT_TY]], 225 // CHECK: [[TMP:%.+]] = alloca [[S_INT_TY]]*, 226 227 // private(s_arr) 228 // CHECK-DAG: [[S_ARR_PRIV_BGN:%.+]] = getelementptr{{.*}} [2 x [[S_INT_TY]]], [2 x [[S_INT_TY]]]* [[S_ARR_PRIV]], 229 // CHECK-DAG: [[S_ARR_PTR_ALLOC:%.+]] = phi{{.+}} [ [[S_ARR_PRIV_BGN]], {{.+}} ], [ [[S_ARR_NEXT:%.+]], {{.+}} ] 230 // CHECK-DAG: call void @{{.+}}({{.+}} [[S_ARR_PTR_ALLOC]]) 231 // CHECK-DAG: [[S_ARR_NEXT]] = getelementptr {{.+}} [[S_ARR_PTR_ALLOC]], 232 233 // CHECK-DAG: [[S_ARR_PRIV_BGN:%.+]] = getelementptr{{.*}} [2 x [[S_INT_TY]]], [2 x [[S_INT_TY]]]* [[S_ARR_PRIV]], 234 // CHECK-DAG: [[S_ARR_PTR_ALLOC:%.+]] = phi{{.+}} [ [[S_ARR_PRIV_BGN]], {{.+}} ], [ [[S_ARR_NEXT:%.+]], {{.+}} ] 235 // CHECK-DAG: call void @{{.+}}({{.+}} [[S_ARR_PTR_ALLOC]]) 236 // CHECK-DAG: [[S_ARR_NEXT]] = getelementptr {{.+}} [[S_ARR_PTR_ALLOC]], 237 238 // private(var) 239 // CHECK-DAG: call void @{{.+}}({{.+}} [[VAR_PRIV]]) 240 // CHECK-DAG: store{{.+}} [[VAR_PRIV]], {{.+}} [[TMP]] 241 242 // CHECK: call void @__kmpc_for_static_init_4( 243 // CHECK-DAG: {{.+}} = {{.+}} [[T_VAR_PRIV]] 244 // CHECK-DAG: {{.+}} = {{.+}} [[VEC_PRIV]] 245 // CHECK-DAG: {{.+}} = {{.+}} [[S_ARR_PRIV]] 246 // CHECK-DAG: {{.+}} = {{.+}} [[TMP]] 247 // CHECK: call void @__kmpc_for_static_fini( 248 // CHECK: ret void 249 250 // CHECK: !{!"llvm.loop.vectorize.enable", i1 true} 251 #endif 252