1 // RUN: %clang_cc1 -verify -fopenmp -x c++ -triple x86_64-apple-darwin10 -emit-llvm %s -o - | FileCheck %s 2 // RUN: %clang_cc1 -fopenmp -x c++ -std=c++11 -triple x86_64-apple-darwin10 -emit-pch -o %t %s 3 // RUN: %clang_cc1 -fopenmp -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 -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 -x c++ -fblocks -DBLOCKS -triple x86_64-apple-darwin10 -emit-llvm %s -o - | FileCheck -check-prefix=BLOCKS %s 6 // expected-no-diagnostics 7 // REQUIRES: x86-registered-target 8 #ifndef HEADER 9 #define HEADER 10 11 struct St { 12 int a, b; 13 St() : a(0), b(0) {} 14 St(const St &st) : a(st.a + st.b), b(0) {} 15 ~St() {} 16 }; 17 18 volatile int g = 1212; 19 volatile int &g1 = g; 20 21 template <class T> 22 struct S { 23 T f; 24 S(T a) : f(a + g) {} 25 S() : f(g) {} 26 S(const S &s, St t = St()) : f(s.f + t.a) {} 27 operator T() { return T(); } 28 ~S() {} 29 }; 30 31 // CHECK-DAG: [[S_FLOAT_TY:%.+]] = type { float } 32 // CHECK-DAG: [[S_INT_TY:%.+]] = type { i{{[0-9]+}} } 33 // CHECK-DAG: [[ST_TY:%.+]] = type { i{{[0-9]+}}, i{{[0-9]+}} } 34 35 template <typename T> 36 T tmain() { 37 S<T> test; 38 T t_var = T(); 39 T vec[] = {1, 2}; 40 S<T> s_arr[] = {1, 2}; 41 S<T> &var = test; 42 #pragma omp parallel 43 #pragma omp for firstprivate(t_var, vec, s_arr, var) 44 for (int i = 0; i < 2; ++i) { 45 vec[i] = t_var; 46 s_arr[i] = var; 47 } 48 return T(); 49 } 50 51 // CHECK: [[TEST:@.+]] = global [[S_FLOAT_TY]] zeroinitializer, 52 S<float> test; 53 // CHECK-DAG: [[T_VAR:@.+]] = global i{{[0-9]+}} 333, 54 int t_var = 333; 55 // CHECK-DAG: [[VEC:@.+]] = global [2 x i{{[0-9]+}}] [i{{[0-9]+}} 1, i{{[0-9]+}} 2], 56 int vec[] = {1, 2}; 57 // CHECK-DAG: [[S_ARR:@.+]] = global [2 x [[S_FLOAT_TY]]] zeroinitializer, 58 S<float> s_arr[] = {1, 2}; 59 // CHECK-DAG: [[VAR:@.+]] = global [[S_FLOAT_TY]] zeroinitializer, 60 S<float> var(3); 61 // CHECK-DAG: [[IMPLICIT_BARRIER_LOC:@.+]] = private unnamed_addr constant %{{.+}} { i32 0, i32 66, i32 0, i32 0, i8* 62 63 // CHECK: call {{.*}} [[S_FLOAT_TY_DEF_CONSTR:@.+]]([[S_FLOAT_TY]]* [[TEST]]) 64 // CHECK: ([[S_FLOAT_TY]]*)* [[S_FLOAT_TY_DESTR:@[^ ]+]] {{[^,]+}}, {{.+}}([[S_FLOAT_TY]]* [[TEST]] 65 int main() { 66 #ifdef LAMBDA 67 // LAMBDA: [[G:@.+]] = global i{{[0-9]+}} 1212, 68 // LAMBDA-LABEL: @main 69 // LAMBDA: call void [[OUTER_LAMBDA:@.+]]( 70 [&]() { 71 // LAMBDA: define{{.*}} internal{{.*}} void [[OUTER_LAMBDA]]( 72 // LAMBDA: call void {{.+}} @__kmpc_fork_call({{.+}}, i32 0, {{.+}}* [[OMP_REGION:@.+]] to {{.+}}) 73 #pragma omp parallel 74 #pragma omp for firstprivate(g, g1) 75 for (int i = 0; i < 2; ++i) { 76 // LAMBDA: define{{.*}} internal{{.*}} void [[OMP_REGION]](i32* noalias %{{.+}}, i32* noalias %{{.+}}) 77 // Skip temp vars for loop 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: alloca i{{[0-9]+}}, 83 // LAMBDA: [[G_PRIVATE_ADDR:%.+]] = alloca i{{[0-9]+}}, 84 // LAMBDA: [[G_VAL:%.+]] = load volatile i{{[0-9]+}}, i{{[0-9]+}}* [[G]] 85 // LAMBDA: store i{{[0-9]+}} [[G_VAL]], i{{[0-9]+}}* [[G_PRIVATE_ADDR]] 86 // LAMBDA: call void @__kmpc_barrier( 87 g = 1; 88 g1 = 1; 89 // LAMBDA: call void @__kmpc_for_static_init_4( 90 // LAMBDA: store i{{[0-9]+}} 1, i{{[0-9]+}}* [[G_PRIVATE_ADDR]], 91 // LAMBDA: [[G_PRIVATE_ADDR_REF:%.+]] = getelementptr inbounds %{{.+}}, %{{.+}}* [[ARG:%.+]], i{{[0-9]+}} 0, i{{[0-9]+}} 0 92 // LAMBDA: store i{{[0-9]+}}* [[G_PRIVATE_ADDR]], i{{[0-9]+}}** [[G_PRIVATE_ADDR_REF]] 93 // LAMBDA: call void [[INNER_LAMBDA:@.+]](%{{.+}}* [[ARG]]) 94 // LAMBDA: call void @__kmpc_for_static_fini( 95 // LAMBDA: call i32 @__kmpc_cancel_barrier( 96 [&]() { 97 // LAMBDA: define {{.+}} void [[INNER_LAMBDA]](%{{.+}}* [[ARG_PTR:%.+]]) 98 // LAMBDA: store %{{.+}}* [[ARG_PTR]], %{{.+}}** [[ARG_PTR_REF:%.+]], 99 g = 2; 100 g1 = 2; 101 // LAMBDA: [[ARG_PTR:%.+]] = load %{{.+}}*, %{{.+}}** [[ARG_PTR_REF]] 102 // LAMBDA: [[G_PTR_REF:%.+]] = getelementptr inbounds %{{.+}}, %{{.+}}* [[ARG_PTR]], i{{[0-9]+}} 0, i{{[0-9]+}} 0 103 // LAMBDA: [[G_REF:%.+]] = load i{{[0-9]+}}*, i{{[0-9]+}}** [[G_PTR_REF]] 104 // LAMBDA: store i{{[0-9]+}} 2, i{{[0-9]+}}* [[G_REF]] 105 }(); 106 } 107 }(); 108 return 0; 109 #elif defined(BLOCKS) 110 // BLOCKS: [[G:@.+]] = global i{{[0-9]+}} 1212, 111 // BLOCKS-LABEL: @main 112 // BLOCKS: call void {{%.+}}(i8 113 ^{ 114 // BLOCKS: define{{.*}} internal{{.*}} void {{.+}}(i8* 115 // BLOCKS: call void {{.+}} @__kmpc_fork_call({{.+}}, i32 0, {{.+}}* [[OMP_REGION:@.+]] to {{.+}}) 116 #pragma omp parallel 117 #pragma omp for firstprivate(g, g1) 118 for (int i = 0; i < 2; ++i) { 119 // BLOCKS: define{{.*}} internal{{.*}} void [[OMP_REGION]](i32* noalias %{{.+}}, i32* noalias %{{.+}}) 120 // Skip temp vars for loop 121 // BLOCKS: alloca i{{[0-9]+}}, 122 // BLOCKS: alloca i{{[0-9]+}}, 123 // BLOCKS: alloca i{{[0-9]+}}, 124 // BLOCKS: alloca i{{[0-9]+}}, 125 // BLOCKS: alloca i{{[0-9]+}}, 126 // BLOCKS: [[G_PRIVATE_ADDR:%.+]] = alloca i{{[0-9]+}}, 127 // BLOCKS: [[G_VAL:%.+]] = load volatile i{{[0-9]+}}, i{{[0-9]+}}* [[G]] 128 // BLOCKS: store i{{[0-9]+}} [[G_VAL]], i{{[0-9]+}}* [[G_PRIVATE_ADDR]] 129 // BLOCKS: call void @__kmpc_barrier( 130 g = 1; 131 g1 =1; 132 // BLOCKS: call void @__kmpc_for_static_init_4( 133 // BLOCKS: store i{{[0-9]+}} 1, i{{[0-9]+}}* [[G_PRIVATE_ADDR]], 134 // BLOCKS-NOT: [[G]]{{[[^:word:]]}} 135 // BLOCKS: i{{[0-9]+}}* [[G_PRIVATE_ADDR]] 136 // BLOCKS-NOT: [[G]]{{[[^:word:]]}} 137 // BLOCKS: call void {{%.+}}(i8 138 // BLOCKS: call void @__kmpc_for_static_fini( 139 // BLOCKS: call i32 @__kmpc_cancel_barrier( 140 ^{ 141 // BLOCKS: define {{.+}} void {{@.+}}(i8* 142 g = 2; 143 g1 = 2; 144 // BLOCKS-NOT: [[G]]{{[[^:word:]]}} 145 // BLOCKS: store i{{[0-9]+}} 2, i{{[0-9]+}}* 146 // BLOCKS-NOT: [[G]]{{[[^:word:]]}} 147 // BLOCKS: ret 148 }(); 149 } 150 }(); 151 return 0; 152 #else 153 #pragma omp for firstprivate(t_var, vec, s_arr, var) 154 for (int i = 0; i < 2; ++i) { 155 vec[i] = t_var; 156 s_arr[i] = var; 157 } 158 return tmain<int>(); 159 #endif 160 } 161 162 // CHECK: define {{.*}}i{{[0-9]+}} @main() 163 // CHECK: alloca i{{[0-9]+}}, 164 // Skip temp vars for loop 165 // CHECK: alloca i{{[0-9]+}}, 166 // CHECK: alloca i{{[0-9]+}}, 167 // CHECK: alloca i{{[0-9]+}}, 168 // CHECK: alloca i{{[0-9]+}}, 169 // CHECK: alloca i{{[0-9]+}}, 170 // CHECK: [[T_VAR_PRIV:%.+]] = alloca i{{[0-9]+}}, 171 // CHECK: [[VEC_PRIV:%.+]] = alloca [2 x i{{[0-9]+}}], 172 // CHECK: [[S_ARR_PRIV:%.+]] = alloca [2 x [[S_FLOAT_TY]]], 173 // CHECK: [[VAR_PRIV:%.+]] = alloca [[S_FLOAT_TY]], 174 // CHECK: [[GTID:%.+]] = call i32 @__kmpc_global_thread_num( 175 176 // firstprivate t_var(t_var) 177 // CHECK: [[T_VAR_VAL:%.+]] = load i{{[0-9]+}}, i{{[0-9]+}}* [[T_VAR]], 178 // CHECK: store i{{[0-9]+}} [[T_VAR_VAL]], i{{[0-9]+}}* [[T_VAR_PRIV]], 179 180 // firstprivate vec(vec) 181 // CHECK: [[VEC_DEST:%.+]] = bitcast [2 x i{{[0-9]+}}]* [[VEC_PRIV]] to i8* 182 // CHECK: call void @llvm.memcpy.{{.+}}(i8* [[VEC_DEST]], i8* bitcast ([2 x i{{[0-9]+}}]* [[VEC]] to i8*), 183 184 // firstprivate s_arr(s_arr) 185 // 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]+}} 0 186 // CHECK: [[S_ARR_PRIV_END:%.+]] = getelementptr [[S_FLOAT_TY]], [[S_FLOAT_TY]]* [[S_ARR_PRIV_BEGIN]], i{{[0-9]+}} 2 187 // CHECK: [[IS_EMPTY:%.+]] = icmp eq [[S_FLOAT_TY]]* [[S_ARR_PRIV_BEGIN]], [[S_ARR_PRIV_END]] 188 // CHECK: br i1 [[IS_EMPTY]], label %[[S_ARR_BODY_DONE:.+]], label %[[S_ARR_BODY:.+]] 189 // CHECK: [[S_ARR_BODY]] 190 // CHECK: getelementptr inbounds ([2 x [[S_FLOAT_TY]]], [2 x [[S_FLOAT_TY]]]* [[S_ARR]], i{{[0-9]+}} 0, i{{[0-9]+}} 0) 191 // CHECK: call {{.*}} [[ST_TY_DEFAULT_CONSTR:@.+]]([[ST_TY]]* [[ST_TY_TEMP:%.+]]) 192 // CHECK: call {{.*}} [[S_FLOAT_TY_COPY_CONSTR:@.+]]([[S_FLOAT_TY]]* {{.+}}, [[S_FLOAT_TY]]* {{.+}}, [[ST_TY]]* [[ST_TY_TEMP]]) 193 // CHECK: call {{.*}} [[ST_TY_DESTR:@.+]]([[ST_TY]]* [[ST_TY_TEMP]]) 194 // CHECK: br i1 {{.+}}, label %{{.+}}, label %[[S_ARR_BODY]] 195 196 // firstprivate var(var) 197 // CHECK: call {{.*}} [[ST_TY_DEFAULT_CONSTR]]([[ST_TY]]* [[ST_TY_TEMP:%.+]]) 198 // CHECK: call {{.*}} [[S_FLOAT_TY_COPY_CONSTR]]([[S_FLOAT_TY]]* [[VAR_PRIV]], [[S_FLOAT_TY]]* {{.*}} [[VAR]], [[ST_TY]]* [[ST_TY_TEMP]]) 199 // CHECK: call {{.*}} [[ST_TY_DESTR]]([[ST_TY]]* [[ST_TY_TEMP]]) 200 201 // Synchronization for initialization. 202 // CHECK: call void @__kmpc_barrier(%{{.+}}* [[IMPLICIT_BARRIER_LOC]], i{{[0-9]+}} [[GTID]]) 203 204 // CHECK: call void @__kmpc_for_static_init_4( 205 // CHECK: call void @__kmpc_for_static_fini( 206 207 // ~(firstprivate var), ~(firstprivate s_arr) 208 // CHECK-DAG: call {{.*}} [[S_FLOAT_TY_DESTR]]([[S_FLOAT_TY]]* [[VAR_PRIV]]) 209 // CHECK-DAG: call {{.*}} [[S_FLOAT_TY_DESTR]]([[S_FLOAT_TY]]* 210 // CHECK: call void @__kmpc_barrier(%{{.+}}* [[IMPLICIT_BARRIER_LOC]], i{{[0-9]+}} [[GTID]]) 211 212 // CHECK: = call {{.*}}i{{.+}} [[TMAIN_INT:@.+]]() 213 214 // CHECK: ret void 215 216 // CHECK: define {{.*}} i{{[0-9]+}} [[TMAIN_INT]]() 217 // CHECK: [[TEST:%.+]] = alloca [[S_INT_TY]], 218 // CHECK: call {{.*}} [[S_INT_TY_DEF_CONSTR:@.+]]([[S_INT_TY]]* [[TEST]]) 219 // CHECK: call void (%{{.+}}*, i{{[0-9]+}}, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)*, ...) @__kmpc_fork_call(%{{.+}}* @{{.+}}, i{{[0-9]+}} 4, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)* bitcast (void (i{{[0-9]+}}*, i{{[0-9]+}}*, i32*, [2 x i32]*, [2 x [[S_INT_TY]]]*, [[S_INT_TY]]*)* [[TMAIN_MICROTASK:@.+]] to void 220 // CHECK: call {{.*}} [[S_INT_TY_DESTR:@.+]]([[S_INT_TY]]* 221 // CHECK: ret 222 // 223 // CHECK: define internal void [[TMAIN_MICROTASK]](i{{[0-9]+}}* noalias [[GTID_ADDR:%.+]], i{{[0-9]+}}* noalias %{{.+}}, i32* dereferenceable(4) %{{.+}}, [2 x i32]* dereferenceable(8) %{{.+}}, [2 x [[S_INT_TY]]]* dereferenceable(8) %{{.+}}, [[S_INT_TY]]* dereferenceable(4) %{{.+}}) 224 // Skip temp vars for loop 225 // CHECK: alloca i{{[0-9]+}}, 226 // CHECK: alloca i{{[0-9]+}}, 227 // CHECK: alloca i{{[0-9]+}}, 228 // CHECK: alloca i{{[0-9]+}}, 229 // CHECK: alloca i{{[0-9]+}}, 230 // CHECK: [[T_VAR_PRIV:%.+]] = alloca i{{[0-9]+}}, 231 // CHECK: [[VEC_PRIV:%.+]] = alloca [2 x i{{[0-9]+}}], 232 // CHECK: [[S_ARR_PRIV:%.+]] = alloca [2 x [[S_INT_TY]]], 233 // CHECK: [[VAR_PRIV:%.+]] = alloca [[S_INT_TY]], 234 // CHECK: store i{{[0-9]+}}* [[GTID_ADDR]], i{{[0-9]+}}** [[GTID_ADDR_ADDR:%.+]], 235 236 // CHECK: [[T_VAR_REF:%.+]] = load i{{[0-9]+}}*, i{{[0-9]+}}** % 237 // CHECK: [[VEC_REF:%.+]] = load [2 x i{{[0-9]+}}]*, [2 x i{{[0-9]+}}]** % 238 // CHECK: [[S_ARR:%.+]] = load [2 x [[S_INT_TY]]]*, [2 x [[S_INT_TY]]]** % 239 240 // firstprivate t_var(t_var) 241 // CHECK: [[T_VAR_VAL:%.+]] = load i{{[0-9]+}}, i{{[0-9]+}}* [[T_VAR_REF]], 242 // CHECK: store i{{[0-9]+}} [[T_VAR_VAL]], i{{[0-9]+}}* [[T_VAR_PRIV]], 243 244 // firstprivate vec(vec) 245 // CHECK: [[VEC_DEST:%.+]] = bitcast [2 x i{{[0-9]+}}]* [[VEC_PRIV]] to i8* 246 // CHECK: [[VEC_SRC:%.+]] = bitcast [2 x i{{[0-9]+}}]* [[VEC_REF]] to i8* 247 // CHECK: call void @llvm.memcpy.{{.+}}(i8* [[VEC_DEST]], i8* [[VEC_SRC]], 248 249 // firstprivate s_arr(s_arr) 250 // 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]+}} 0 251 // CHECK: [[S_ARR_PRIV_END:%.+]] = getelementptr [[S_INT_TY]], [[S_INT_TY]]* [[S_ARR_PRIV_BEGIN]], i{{[0-9]+}} 2 252 // CHECK: [[IS_EMPTY:%.+]] = icmp eq [[S_INT_TY]]* [[S_ARR_PRIV_BEGIN]], [[S_ARR_PRIV_END]] 253 // CHECK: br i1 [[IS_EMPTY]], label %[[S_ARR_BODY_DONE:.+]], label %[[S_ARR_BODY:.+]] 254 // CHECK: [[S_ARR_BODY]] 255 // CHECK: call {{.*}} [[ST_TY_DEFAULT_CONSTR:@.+]]([[ST_TY]]* [[ST_TY_TEMP:%.+]]) 256 // CHECK: call {{.*}} [[S_INT_TY_COPY_CONSTR:@.+]]([[S_INT_TY]]* {{.+}}, [[S_INT_TY]]* {{.+}}, [[ST_TY]]* [[ST_TY_TEMP]]) 257 // CHECK: call {{.*}} [[ST_TY_DESTR:@.+]]([[ST_TY]]* [[ST_TY_TEMP]]) 258 // CHECK: br i1 {{.+}}, label %{{.+}}, label %[[S_ARR_BODY]] 259 260 // firstprivate var(var) 261 // CHECK: [[VAR_REF:%.+]] = load [[S_INT_TY]]*, [[S_INT_TY]]** % 262 // CHECK: call {{.*}} [[ST_TY_DEFAULT_CONSTR]]([[ST_TY]]* [[ST_TY_TEMP:%.+]]) 263 // CHECK: call {{.*}} [[S_INT_TY_COPY_CONSTR]]([[S_INT_TY]]* [[VAR_PRIV]], [[S_INT_TY]]* {{.*}} [[VAR_REF]], [[ST_TY]]* [[ST_TY_TEMP]]) 264 // CHECK: call {{.*}} [[ST_TY_DESTR]]([[ST_TY]]* [[ST_TY_TEMP]]) 265 266 // Synchronization for initialization. 267 // CHECK: [[GTID_REF:%.+]] = load i{{[0-9]+}}*, i{{[0-9]+}}** [[GTID_ADDR_ADDR]] 268 // CHECK: [[GTID:%.+]] = load i{{[0-9]+}}, i{{[0-9]+}}* [[GTID_REF]] 269 // CHECK: call void @__kmpc_barrier(%{{.+}}* [[IMPLICIT_BARRIER_LOC]], i{{[0-9]+}} [[GTID]]) 270 271 // CHECK: call void @__kmpc_for_static_init_4( 272 // CHECK: call void @__kmpc_for_static_fini( 273 274 // ~(firstprivate var), ~(firstprivate s_arr) 275 // CHECK-DAG: call {{.*}} [[S_INT_TY_DESTR]]([[S_INT_TY]]* [[VAR_PRIV]]) 276 // CHECK-DAG: call {{.*}} [[S_INT_TY_DESTR]]([[S_INT_TY]]* 277 // CHECK: [[GTID_REF:%.+]] = load i{{[0-9]+}}*, i{{[0-9]+}}** [[GTID_ADDR_ADDR]] 278 // CHECK: [[GTID:%.+]] = load i{{[0-9]+}}, i{{[0-9]+}}* [[GTID_REF]] 279 // CHECK: call i32 @__kmpc_cancel_barrier(%{{.+}}* [[IMPLICIT_BARRIER_LOC]], i{{[0-9]+}} [[GTID]]) 280 // CHECK: ret void 281 #endif 282 283