1 // RUN: %clang_cc1 -verify -fopenmp=libiomp5 -x c++ -triple x86_64-apple-darwin10 -emit-llvm %s -o - | FileCheck %s 2 // RUN: %clang_cc1 -fopenmp=libiomp5 -x c++ -std=c++11 -triple x86_64-apple-darwin10 -emit-pch -o %t %s 3 // RUN: %clang_cc1 -fopenmp=libiomp5 -x c++ -triple x86_64-apple-darwin10 -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck %s 4 // RUN: %clang_cc1 -verify -fopenmp=libiomp5 -x c++ -std=c++11 -DLAMBDA -triple x86_64-apple-darwin10 -emit-llvm %s -o - | FileCheck -check-prefix=LAMBDA %s 5 // RUN: %clang_cc1 -verify -fopenmp=libiomp5 -x c++ -fblocks -DBLOCKS -triple x86_64-apple-darwin10 -emit-llvm %s -o - | FileCheck -check-prefix=BLOCKS %s 6 // expected-no-diagnostics 7 #ifndef HEADER 8 #define HEADER 9 10 struct St { 11 int a, b; 12 St() : a(0), b(0) {} 13 St(const St &st) : a(st.a + st.b), b(0) {} 14 ~St() {} 15 }; 16 17 volatile int g = 1212; 18 19 template <class T> 20 struct S { 21 T f; 22 S(T a) : f(a + g) {} 23 S() : f(g) {} 24 S(const S &s, St t = St()) : f(s.f + t.a) {} 25 operator T() { return T(); } 26 ~S() {} 27 }; 28 29 // CHECK-DAG: [[S_FLOAT_TY:%.+]] = type { float } 30 // CHECK-DAG: [[S_INT_TY:%.+]] = type { i{{[0-9]+}} } 31 // CHECK-DAG: [[ST_TY:%.+]] = type { i{{[0-9]+}}, i{{[0-9]+}} } 32 // CHECK-DAG: [[CAP_TMAIN_TY:%.+]] = type { i{{[0-9]+}}*, [2 x i{{[0-9]+}}]*, [2 x [[S_INT_TY]]]*, [[S_INT_TY]]* } 33 34 template <typename T> 35 T tmain() { 36 S<T> test; 37 T t_var = T(); 38 T vec[] = {1, 2}; 39 S<T> s_arr[] = {1, 2}; 40 S<T> var(3); 41 #pragma omp parallel 42 #pragma omp for firstprivate(t_var, vec, s_arr, var) 43 for (int i = 0; i < 0; ++i) { 44 vec[i] = t_var; 45 s_arr[i] = var; 46 } 47 return T(); 48 } 49 50 // CHECK: [[TEST:@.+]] = global [[S_FLOAT_TY]] zeroinitializer, 51 S<float> test; 52 // CHECK-DAG: [[T_VAR:@.+]] = global i{{[0-9]+}} 333, 53 int t_var = 333; 54 // CHECK-DAG: [[VEC:@.+]] = global [2 x i{{[0-9]+}}] [i{{[0-9]+}} 1, i{{[0-9]+}} 2], 55 int vec[] = {1, 2}; 56 // CHECK-DAG: [[S_ARR:@.+]] = global [2 x [[S_FLOAT_TY]]] zeroinitializer, 57 S<float> s_arr[] = {1, 2}; 58 // CHECK-DAG: [[VAR:@.+]] = global [[S_FLOAT_TY]] zeroinitializer, 59 S<float> var(3); 60 // CHECK-DAG: [[IMPLICIT_BARRIER_LOC:@.+]] = private unnamed_addr constant %{{.+}} { i32 0, i32 66, i32 0, i32 0, i8* 61 62 // CHECK: call {{.*}} [[S_FLOAT_TY_DEF_CONSTR:@.+]]([[S_FLOAT_TY]]* [[TEST]]) 63 // CHECK: ([[S_FLOAT_TY]]*)* [[S_FLOAT_TY_DESTR:@[^ ]+]] {{[^,]+}}, {{.+}}([[S_FLOAT_TY]]* [[TEST]] 64 int main() { 65 #ifdef LAMBDA 66 // LAMBDA: [[G:@.+]] = global i{{[0-9]+}} 1212, 67 // LAMBDA-LABEL: @main 68 // LAMBDA: call void [[OUTER_LAMBDA:@.+]]( 69 [&]() { 70 // LAMBDA: define{{.*}} internal{{.*}} void [[OUTER_LAMBDA]]( 71 // LAMBDA: call void {{.+}} @__kmpc_fork_call({{.+}}, i32 1, {{.+}}* [[OMP_REGION:@.+]] to {{.+}}, i8* %{{.+}}) 72 #pragma omp parallel 73 #pragma omp for firstprivate(g) 74 for (int i = 0; i < 2; ++i) { 75 // LAMBDA: define{{.*}} internal{{.*}} void [[OMP_REGION]](i32* %{{.+}}, i32* %{{.+}}, %{{.+}}* [[ARG:%.+]]) 76 // Skip temp vars for loop 77 // LAMBDA: alloca i{{[0-9]+}}, 78 // LAMBDA: alloca i{{[0-9]+}}, 79 // LAMBDA: alloca i{{[0-9]+}}, 80 // LAMBDA: alloca i{{[0-9]+}}, 81 // LAMBDA: alloca i{{[0-9]+}}, 82 // LAMBDA: [[G_PRIVATE_ADDR:%.+]] = alloca i{{[0-9]+}}, 83 // LAMBDA: [[G_VAL:%.+]] = load volatile i{{[0-9]+}}, i{{[0-9]+}}* [[G]] 84 // LAMBDA: store i{{[0-9]+}} [[G_VAL]], i{{[0-9]+}}* [[G_PRIVATE_ADDR]] 85 // LAMBDA: call i32 @__kmpc_cancel_barrier( 86 g = 1; 87 // LAMBDA: call void @__kmpc_for_static_init_4( 88 // LAMBDA: store volatile i{{[0-9]+}} 1, i{{[0-9]+}}* [[G_PRIVATE_ADDR]], 89 // LAMBDA: [[G_PRIVATE_ADDR_REF:%.+]] = getelementptr inbounds %{{.+}}, %{{.+}}* [[ARG:%.+]], i{{[0-9]+}} 0, i{{[0-9]+}} 0 90 // LAMBDA: store i{{[0-9]+}}* [[G_PRIVATE_ADDR]], i{{[0-9]+}}** [[G_PRIVATE_ADDR_REF]] 91 // LAMBDA: call void [[INNER_LAMBDA:@.+]](%{{.+}}* [[ARG]]) 92 // LAMBDA: call void @__kmpc_for_static_fini( 93 // LAMBDA: call i32 @__kmpc_cancel_barrier( 94 [&]() { 95 // LAMBDA: define {{.+}} void [[INNER_LAMBDA]](%{{.+}}* [[ARG_PTR:%.+]]) 96 // LAMBDA: store %{{.+}}* [[ARG_PTR]], %{{.+}}** [[ARG_PTR_REF:%.+]], 97 g = 2; 98 // LAMBDA: [[ARG_PTR:%.+]] = load %{{.+}}*, %{{.+}}** [[ARG_PTR_REF]] 99 // LAMBDA: [[G_PTR_REF:%.+]] = getelementptr inbounds %{{.+}}, %{{.+}}* [[ARG_PTR]], i{{[0-9]+}} 0, i{{[0-9]+}} 0 100 // LAMBDA: [[G_REF:%.+]] = load i{{[0-9]+}}*, i{{[0-9]+}}** [[G_PTR_REF]] 101 // LAMBDA: store volatile i{{[0-9]+}} 2, i{{[0-9]+}}* [[G_REF]] 102 }(); 103 } 104 }(); 105 return 0; 106 #elif defined(BLOCKS) 107 // BLOCKS: [[G:@.+]] = global i{{[0-9]+}} 1212, 108 // BLOCKS-LABEL: @main 109 // BLOCKS: call void {{%.+}}(i8 110 ^{ 111 // BLOCKS: define{{.*}} internal{{.*}} void {{.+}}(i8* 112 // BLOCKS: call void {{.+}} @__kmpc_fork_call({{.+}}, i32 1, {{.+}}* [[OMP_REGION:@.+]] to {{.+}}, i8* %{{.+}}) 113 #pragma omp parallel 114 #pragma omp for firstprivate(g) 115 for (int i = 0; i < 2; ++i) { 116 // BLOCKS: define{{.*}} internal{{.*}} void [[OMP_REGION]](i32* %{{.+}}, i32* %{{.+}}, %{{.+}}* [[ARG:%.+]]) 117 // Skip temp vars for loop 118 // BLOCKS: alloca i{{[0-9]+}}, 119 // BLOCKS: alloca i{{[0-9]+}}, 120 // BLOCKS: alloca i{{[0-9]+}}, 121 // BLOCKS: alloca i{{[0-9]+}}, 122 // BLOCKS: alloca i{{[0-9]+}}, 123 // BLOCKS: [[G_PRIVATE_ADDR:%.+]] = alloca i{{[0-9]+}}, 124 // BLOCKS: [[G_VAL:%.+]] = load volatile i{{[0-9]+}}, i{{[0-9]+}}* [[G]] 125 // BLOCKS: store i{{[0-9]+}} [[G_VAL]], i{{[0-9]+}}* [[G_PRIVATE_ADDR]] 126 // BLOCKS: call i32 @__kmpc_cancel_barrier( 127 g = 1; 128 // BLOCKS: call void @__kmpc_for_static_init_4( 129 // BLOCKS: store volatile i{{[0-9]+}} 1, i{{[0-9]+}}* [[G_PRIVATE_ADDR]], 130 // BLOCKS-NOT: [[G]]{{[[^:word:]]}} 131 // BLOCKS: i{{[0-9]+}}* [[G_PRIVATE_ADDR]] 132 // BLOCKS-NOT: [[G]]{{[[^:word:]]}} 133 // BLOCKS: call void {{%.+}}(i8 134 // BLOCKS: call void @__kmpc_for_static_fini( 135 // BLOCKS: call i32 @__kmpc_cancel_barrier( 136 ^{ 137 // BLOCKS: define {{.+}} void {{@.+}}(i8* 138 g = 2; 139 // BLOCKS-NOT: [[G]]{{[[^:word:]]}} 140 // BLOCKS: store volatile i{{[0-9]+}} 2, i{{[0-9]+}}* 141 // BLOCKS-NOT: [[G]]{{[[^:word:]]}} 142 // BLOCKS: ret 143 }(); 144 } 145 }(); 146 return 0; 147 #else 148 #pragma omp for firstprivate(t_var, vec, s_arr, var) 149 for (int i = 0; i < 0; ++i) { 150 vec[i] = t_var; 151 s_arr[i] = var; 152 } 153 return tmain<int>(); 154 #endif 155 } 156 157 // CHECK: define {{.*}}i{{[0-9]+}} @main() 158 // CHECK: alloca i{{[0-9]+}}, 159 // Skip temp vars for loop 160 // CHECK: alloca i{{[0-9]+}}, 161 // CHECK: alloca i{{[0-9]+}}, 162 // CHECK: alloca i{{[0-9]+}}, 163 // CHECK: alloca i{{[0-9]+}}, 164 // CHECK: alloca i{{[0-9]+}}, 165 // CHECK: [[T_VAR_PRIV:%.+]] = alloca i{{[0-9]+}}, 166 // CHECK: [[VEC_PRIV:%.+]] = alloca [2 x i{{[0-9]+}}], 167 // CHECK: [[S_ARR_PRIV:%.+]] = alloca [2 x [[S_FLOAT_TY]]], 168 // CHECK: [[VAR_PRIV:%.+]] = alloca [[S_FLOAT_TY]], 169 // CHECK: [[GTID:%.+]] = call i32 @__kmpc_global_thread_num( 170 171 // firstprivate t_var(t_var) 172 // CHECK: [[T_VAR_VAL:%.+]] = load i{{[0-9]+}}, i{{[0-9]+}}* [[T_VAR]], 173 // CHECK: store i{{[0-9]+}} [[T_VAR_VAL]], i{{[0-9]+}}* [[T_VAR_PRIV]], 174 175 // firstprivate vec(vec) 176 // CHECK: [[VEC_DEST:%.+]] = bitcast [2 x i{{[0-9]+}}]* [[VEC_PRIV]] to i8* 177 // CHECK: call void @llvm.memcpy.{{.+}}(i8* [[VEC_DEST]], i8* bitcast ([2 x i{{[0-9]+}}]* [[VEC]] to i8*), 178 179 // firstprivate s_arr(s_arr) 180 // CHECK: [[S_ARR_PRIV_BEGIN:%.+]] = getelementptr inbounds [2 x [[S_FLOAT_TY]]], [2 x [[S_FLOAT_TY]]]* [[S_ARR_PRIV]], i{{[0-9]+}} 0, i{{[0-9]+}} 0 181 // CHECK: [[S_ARR_PRIV_END:%.+]] = getelementptr [[S_FLOAT_TY]], [[S_FLOAT_TY]]* [[S_ARR_PRIV_BEGIN]], i{{[0-9]+}} 2 182 // CHECK: [[IS_EMPTY:%.+]] = icmp eq [[S_FLOAT_TY]]* [[S_ARR_PRIV_BEGIN]], [[S_ARR_PRIV_END]] 183 // CHECK: br i1 [[IS_EMPTY]], label %[[S_ARR_BODY_DONE:.+]], label %[[S_ARR_BODY:.+]] 184 // CHECK: [[S_ARR_BODY]] 185 // CHECK: getelementptr inbounds ([2 x [[S_FLOAT_TY]]], [2 x [[S_FLOAT_TY]]]* [[S_ARR]], i{{[0-9]+}} 0, i{{[0-9]+}} 0) 186 // CHECK: call {{.*}} [[ST_TY_DEFAULT_CONSTR:@.+]]([[ST_TY]]* [[ST_TY_TEMP:%.+]]) 187 // CHECK: call {{.*}} [[S_FLOAT_TY_COPY_CONSTR:@.+]]([[S_FLOAT_TY]]* {{.+}}, [[S_FLOAT_TY]]* {{.+}}, [[ST_TY]]* [[ST_TY_TEMP]]) 188 // CHECK: call {{.*}} [[ST_TY_DESTR:@.+]]([[ST_TY]]* [[ST_TY_TEMP]]) 189 // CHECK: br i1 {{.+}}, label %{{.+}}, label %[[S_ARR_BODY]] 190 191 // firstprivate var(var) 192 // CHECK: call {{.*}} [[ST_TY_DEFAULT_CONSTR]]([[ST_TY]]* [[ST_TY_TEMP:%.+]]) 193 // CHECK: call {{.*}} [[S_FLOAT_TY_COPY_CONSTR]]([[S_FLOAT_TY]]* [[VAR_PRIV]], [[S_FLOAT_TY]]* {{.*}} [[VAR]], [[ST_TY]]* [[ST_TY_TEMP]]) 194 // CHECK: call {{.*}} [[ST_TY_DESTR]]([[ST_TY]]* [[ST_TY_TEMP]]) 195 196 // Synchronization for initialization. 197 // CHECK: call i32 @__kmpc_cancel_barrier(%{{.+}}* [[IMPLICIT_BARRIER_LOC]], i{{[0-9]+}} [[GTID]]) 198 199 // CHECK: call void @__kmpc_for_static_init_4( 200 // CHECK: call void @__kmpc_for_static_fini( 201 202 // ~(firstprivate var), ~(firstprivate s_arr) 203 // CHECK-DAG: call {{.*}} [[S_FLOAT_TY_DESTR]]([[S_FLOAT_TY]]* [[VAR_PRIV]]) 204 // CHECK-DAG: call {{.*}} [[S_FLOAT_TY_DESTR]]([[S_FLOAT_TY]]* 205 // CHECK: call i32 @__kmpc_cancel_barrier(%{{.+}}* [[IMPLICIT_BARRIER_LOC]], i{{[0-9]+}} [[GTID]]) 206 207 // CHECK: = call {{.*}}i{{.+}} [[TMAIN_INT:@.+]]() 208 209 // CHECK: ret void 210 211 // CHECK: define {{.*}} i{{[0-9]+}} [[TMAIN_INT]]() 212 // CHECK: [[TEST:%.+]] = alloca [[S_INT_TY]], 213 // CHECK: call {{.*}} [[S_INT_TY_DEF_CONSTR:@.+]]([[S_INT_TY]]* [[TEST]]) 214 // CHECK: call void (%{{.+}}*, i{{[0-9]+}}, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)*, ...) @__kmpc_fork_call(%{{.+}}* @{{.+}}, i{{[0-9]+}} 1, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)* bitcast (void (i{{[0-9]+}}*, i{{[0-9]+}}*, [[CAP_TMAIN_TY]]*)* [[TMAIN_MICROTASK:@.+]] to void 215 // CHECK: call {{.*}} [[S_INT_TY_DESTR:@.+]]([[S_INT_TY]]* 216 // CHECK: ret 217 // 218 // CHECK: define internal void [[TMAIN_MICROTASK]](i{{[0-9]+}}* [[GTID_ADDR:%.+]], i{{[0-9]+}}* %{{.+}}, [[CAP_TMAIN_TY]]* %{{.+}}) 219 // Skip temp vars for loop 220 // CHECK: alloca i{{[0-9]+}}, 221 // CHECK: alloca i{{[0-9]+}}, 222 // CHECK: alloca i{{[0-9]+}}, 223 // CHECK: alloca i{{[0-9]+}}, 224 // CHECK: alloca i{{[0-9]+}}, 225 // CHECK: [[T_VAR_PRIV:%.+]] = alloca i{{[0-9]+}}, 226 // CHECK: [[VEC_PRIV:%.+]] = alloca [2 x i{{[0-9]+}}], 227 // CHECK: [[S_ARR_PRIV:%.+]] = alloca [2 x [[S_INT_TY]]], 228 // CHECK: [[VAR_PRIV:%.+]] = alloca [[S_INT_TY]], 229 // CHECK: store i{{[0-9]+}}* [[GTID_ADDR]], i{{[0-9]+}}** [[GTID_ADDR_ADDR:%.+]], 230 231 // firstprivate t_var(t_var) 232 // CHECK: [[T_VAR_PTR_REF:%.+]] = getelementptr inbounds [[CAP_TMAIN_TY]], [[CAP_TMAIN_TY]]* %{{.+}}, i{{[0-9]+}} 0, i{{[0-9]+}} 0 233 // CHECK: [[T_VAR_REF:%.+]] = load i{{[0-9]+}}*, i{{[0-9]+}}** [[T_VAR_PTR_REF]], 234 // CHECK: [[T_VAR_VAL:%.+]] = load i{{[0-9]+}}, i{{[0-9]+}}* [[T_VAR_REF]], 235 // CHECK: store i{{[0-9]+}} [[T_VAR_VAL]], i{{[0-9]+}}* [[T_VAR_PRIV]], 236 237 // firstprivate vec(vec) 238 // CHECK: [[VEC_PTR_REF:%.+]] = getelementptr inbounds [[CAP_TMAIN_TY]], [[CAP_TMAIN_TY]]* %{{.+}}, i{{[0-9]+}} 0, i{{[0-9]+}} 1 239 // CHECK: [[VEC_REF:%.+]] = load [2 x i{{[0-9]+}}]*, [2 x i{{[0-9]+}}]** [[VEC_PTR_REF:%.+]], 240 // CHECK: [[VEC_DEST:%.+]] = bitcast [2 x i{{[0-9]+}}]* [[VEC_PRIV]] to i8* 241 // CHECK: [[VEC_SRC:%.+]] = bitcast [2 x i{{[0-9]+}}]* [[VEC_REF]] to i8* 242 // CHECK: call void @llvm.memcpy.{{.+}}(i8* [[VEC_DEST]], i8* [[VEC_SRC]], 243 244 // firstprivate s_arr(s_arr) 245 // CHECK: [[S_ARR_REF:%.+]] = getelementptr inbounds [[CAP_TMAIN_TY]], [[CAP_TMAIN_TY]]* %{{.+}}, i{{[0-9]+}} 0, i{{[0-9]+}} 2 246 // CHECK: [[S_ARR:%.+]] = load [2 x [[S_INT_TY]]]*, [2 x [[S_INT_TY]]]** [[S_ARR_REF]], 247 // CHECK: [[S_ARR_PRIV_BEGIN:%.+]] = getelementptr inbounds [2 x [[S_INT_TY]]], [2 x [[S_INT_TY]]]* [[S_ARR_PRIV]], i{{[0-9]+}} 0, i{{[0-9]+}} 0 248 // CHECK: [[S_ARR_PRIV_END:%.+]] = getelementptr [[S_INT_TY]], [[S_INT_TY]]* [[S_ARR_PRIV_BEGIN]], i{{[0-9]+}} 2 249 // CHECK: [[IS_EMPTY:%.+]] = icmp eq [[S_INT_TY]]* [[S_ARR_PRIV_BEGIN]], [[S_ARR_PRIV_END]] 250 // CHECK: br i1 [[IS_EMPTY]], label %[[S_ARR_BODY_DONE:.+]], label %[[S_ARR_BODY:.+]] 251 // CHECK: [[S_ARR_BODY]] 252 // CHECK: call {{.*}} [[ST_TY_DEFAULT_CONSTR:@.+]]([[ST_TY]]* [[ST_TY_TEMP:%.+]]) 253 // CHECK: call {{.*}} [[S_INT_TY_COPY_CONSTR:@.+]]([[S_INT_TY]]* {{.+}}, [[S_INT_TY]]* {{.+}}, [[ST_TY]]* [[ST_TY_TEMP]]) 254 // CHECK: call {{.*}} [[ST_TY_DESTR:@.+]]([[ST_TY]]* [[ST_TY_TEMP]]) 255 // CHECK: br i1 {{.+}}, label %{{.+}}, label %[[S_ARR_BODY]] 256 257 // firstprivate var(var) 258 // CHECK: [[VAR_REF_PTR:%.+]] = getelementptr inbounds [[CAP_TMAIN_TY]], [[CAP_TMAIN_TY]]* %{{.+}}, i{{[0-9]+}} 0, i{{[0-9]+}} 3 259 // CHECK: [[VAR_REF:%.+]] = load [[S_INT_TY]]*, [[S_INT_TY]]** [[VAR_REF_PTR]], 260 // CHECK: call {{.*}} [[ST_TY_DEFAULT_CONSTR]]([[ST_TY]]* [[ST_TY_TEMP:%.+]]) 261 // CHECK: call {{.*}} [[S_INT_TY_COPY_CONSTR]]([[S_INT_TY]]* [[VAR_PRIV]], [[S_INT_TY]]* {{.*}} [[VAR_REF]], [[ST_TY]]* [[ST_TY_TEMP]]) 262 // CHECK: call {{.*}} [[ST_TY_DESTR]]([[ST_TY]]* [[ST_TY_TEMP]]) 263 264 // Synchronization for initialization. 265 // CHECK: [[GTID_REF:%.+]] = load i{{[0-9]+}}*, i{{[0-9]+}}** [[GTID_ADDR_ADDR]] 266 // CHECK: [[GTID:%.+]] = load i{{[0-9]+}}, i{{[0-9]+}}* [[GTID_REF]] 267 // CHECK: call i32 @__kmpc_cancel_barrier(%{{.+}}* [[IMPLICIT_BARRIER_LOC]], i{{[0-9]+}} [[GTID]]) 268 269 // CHECK: call void @__kmpc_for_static_init_4( 270 // CHECK: call void @__kmpc_for_static_fini( 271 272 // ~(firstprivate var), ~(firstprivate s_arr) 273 // CHECK-DAG: call {{.*}} [[S_INT_TY_DESTR]]([[S_INT_TY]]* [[VAR_PRIV]]) 274 // CHECK-DAG: call {{.*}} [[S_INT_TY_DESTR]]([[S_INT_TY]]* 275 // CHECK: [[GTID_REF:%.+]] = load i{{[0-9]+}}*, i{{[0-9]+}}** [[GTID_ADDR_ADDR]] 276 // CHECK: [[GTID:%.+]] = load i{{[0-9]+}}, i{{[0-9]+}}* [[GTID_REF]] 277 // CHECK: call i32 @__kmpc_cancel_barrier(%{{.+}}* [[IMPLICIT_BARRIER_LOC]], i{{[0-9]+}} [[GTID]]) 278 // CHECK: ret void 279 #endif 280 281
