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 20 template <class T> 21 struct S { 22 T f; 23 S(T a) : f(a + g) {} 24 S() : f(g) {} 25 S(const S &s, St t = St()) : f(s.f + t.a) {} 26 operator T() { return T(); } 27 ~S() {} 28 }; 29 30 // CHECK-DAG: [[S_FLOAT_TY:%.+]] = type { float } 31 // CHECK-DAG: [[S_INT_TY:%.+]] = type { i{{[0-9]+}} } 32 // CHECK-DAG: [[ST_TY:%.+]] = type { i{{[0-9]+}}, i{{[0-9]+}} } 33 // CHECK-DAG: [[CAP_TMAIN_TY:%.+]] = type { i{{[0-9]+}}*, [2 x i{{[0-9]+}}]*, [2 x [[S_INT_TY]]]*, [[S_INT_TY]]* } 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(3); 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 1, {{.+}}* [[OMP_REGION:@.+]] to {{.+}}, i8* %{{.+}}) 73 #pragma omp parallel 74 #pragma omp for firstprivate(g) 75 for (int i = 0; i < 2; ++i) { 76 // LAMBDA: define{{.*}} internal{{.*}} void [[OMP_REGION]](i32* %{{.+}}, i32* %{{.+}}, %{{.+}}* [[ARG:%.+]]) 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 // LAMBDA: call void @__kmpc_for_static_init_4( 89 // LAMBDA: store volatile i{{[0-9]+}} 1, i{{[0-9]+}}* [[G_PRIVATE_ADDR]], 90 // LAMBDA: [[G_PRIVATE_ADDR_REF:%.+]] = getelementptr inbounds %{{.+}}, %{{.+}}* [[ARG:%.+]], i{{[0-9]+}} 0, i{{[0-9]+}} 0 91 // LAMBDA: store i{{[0-9]+}}* [[G_PRIVATE_ADDR]], i{{[0-9]+}}** [[G_PRIVATE_ADDR_REF]] 92 // LAMBDA: call void [[INNER_LAMBDA:@.+]](%{{.+}}* [[ARG]]) 93 // LAMBDA: call void @__kmpc_for_static_fini( 94 // LAMBDA: call i32 @__kmpc_cancel_barrier( 95 [&]() { 96 // LAMBDA: define {{.+}} void [[INNER_LAMBDA]](%{{.+}}* [[ARG_PTR:%.+]]) 97 // LAMBDA: store %{{.+}}* [[ARG_PTR]], %{{.+}}** [[ARG_PTR_REF:%.+]], 98 g = 2; 99 // LAMBDA: [[ARG_PTR:%.+]] = load %{{.+}}*, %{{.+}}** [[ARG_PTR_REF]] 100 // LAMBDA: [[G_PTR_REF:%.+]] = getelementptr inbounds %{{.+}}, %{{.+}}* [[ARG_PTR]], i{{[0-9]+}} 0, i{{[0-9]+}} 0 101 // LAMBDA: [[G_REF:%.+]] = load i{{[0-9]+}}*, i{{[0-9]+}}** [[G_PTR_REF]] 102 // LAMBDA: store volatile i{{[0-9]+}} 2, i{{[0-9]+}}* [[G_REF]] 103 }(); 104 } 105 }(); 106 return 0; 107 #elif defined(BLOCKS) 108 // BLOCKS: [[G:@.+]] = global i{{[0-9]+}} 1212, 109 // BLOCKS-LABEL: @main 110 // BLOCKS: call void {{%.+}}(i8 111 ^{ 112 // BLOCKS: define{{.*}} internal{{.*}} void {{.+}}(i8* 113 // BLOCKS: call void {{.+}} @__kmpc_fork_call({{.+}}, i32 1, {{.+}}* [[OMP_REGION:@.+]] to {{.+}}, i8* %{{.+}}) 114 #pragma omp parallel 115 #pragma omp for firstprivate(g) 116 for (int i = 0; i < 2; ++i) { 117 // BLOCKS: define{{.*}} internal{{.*}} void [[OMP_REGION]](i32* %{{.+}}, i32* %{{.+}}, %{{.+}}* [[ARG:%.+]]) 118 // Skip temp vars for loop 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: alloca i{{[0-9]+}}, 124 // BLOCKS: [[G_PRIVATE_ADDR:%.+]] = alloca i{{[0-9]+}}, 125 // BLOCKS: [[G_VAL:%.+]] = load volatile i{{[0-9]+}}, i{{[0-9]+}}* [[G]] 126 // BLOCKS: store i{{[0-9]+}} [[G_VAL]], i{{[0-9]+}}* [[G_PRIVATE_ADDR]] 127 // BLOCKS: call void @__kmpc_barrier( 128 g = 1; 129 // BLOCKS: call void @__kmpc_for_static_init_4( 130 // BLOCKS: store volatile i{{[0-9]+}} 1, i{{[0-9]+}}* [[G_PRIVATE_ADDR]], 131 // BLOCKS-NOT: [[G]]{{[[^:word:]]}} 132 // BLOCKS: i{{[0-9]+}}* [[G_PRIVATE_ADDR]] 133 // BLOCKS-NOT: [[G]]{{[[^:word:]]}} 134 // BLOCKS: call void {{%.+}}(i8 135 // BLOCKS: call void @__kmpc_for_static_fini( 136 // BLOCKS: call i32 @__kmpc_cancel_barrier( 137 ^{ 138 // BLOCKS: define {{.+}} void {{@.+}}(i8* 139 g = 2; 140 // BLOCKS-NOT: [[G]]{{[[^:word:]]}} 141 // BLOCKS: store volatile i{{[0-9]+}} 2, i{{[0-9]+}}* 142 // BLOCKS-NOT: [[G]]{{[[^:word:]]}} 143 // BLOCKS: ret 144 }(); 145 } 146 }(); 147 return 0; 148 #else 149 #pragma omp for firstprivate(t_var, vec, s_arr, var) 150 for (int i = 0; i < 2; ++i) { 151 vec[i] = t_var; 152 s_arr[i] = var; 153 } 154 return tmain<int>(); 155 #endif 156 } 157 158 // CHECK: define {{.*}}i{{[0-9]+}} @main() 159 // CHECK: alloca i{{[0-9]+}}, 160 // Skip temp vars for loop 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: alloca i{{[0-9]+}}, 166 // CHECK: [[T_VAR_PRIV:%.+]] = alloca i{{[0-9]+}}, 167 // CHECK: [[VEC_PRIV:%.+]] = alloca [2 x i{{[0-9]+}}], 168 // CHECK: [[S_ARR_PRIV:%.+]] = alloca [2 x [[S_FLOAT_TY]]], 169 // CHECK: [[VAR_PRIV:%.+]] = alloca [[S_FLOAT_TY]], 170 // CHECK: [[GTID:%.+]] = call i32 @__kmpc_global_thread_num( 171 172 // firstprivate t_var(t_var) 173 // CHECK: [[T_VAR_VAL:%.+]] = load i{{[0-9]+}}, i{{[0-9]+}}* [[T_VAR]], 174 // CHECK: store i{{[0-9]+}} [[T_VAR_VAL]], i{{[0-9]+}}* [[T_VAR_PRIV]], 175 176 // firstprivate vec(vec) 177 // CHECK: [[VEC_DEST:%.+]] = bitcast [2 x i{{[0-9]+}}]* [[VEC_PRIV]] to i8* 178 // CHECK: call void @llvm.memcpy.{{.+}}(i8* [[VEC_DEST]], i8* bitcast ([2 x i{{[0-9]+}}]* [[VEC]] to i8*), 179 180 // firstprivate s_arr(s_arr) 181 // 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]+}} 0 182 // CHECK: [[S_ARR_PRIV_END:%.+]] = getelementptr [[S_FLOAT_TY]], [[S_FLOAT_TY]]* [[S_ARR_PRIV_BEGIN]], i{{[0-9]+}} 2 183 // CHECK: [[IS_EMPTY:%.+]] = icmp eq [[S_FLOAT_TY]]* [[S_ARR_PRIV_BEGIN]], [[S_ARR_PRIV_END]] 184 // CHECK: br i1 [[IS_EMPTY]], label %[[S_ARR_BODY_DONE:.+]], label %[[S_ARR_BODY:.+]] 185 // CHECK: [[S_ARR_BODY]] 186 // CHECK: getelementptr inbounds ([2 x [[S_FLOAT_TY]]], [2 x [[S_FLOAT_TY]]]* [[S_ARR]], i{{[0-9]+}} 0, i{{[0-9]+}} 0) 187 // CHECK: call {{.*}} [[ST_TY_DEFAULT_CONSTR:@.+]]([[ST_TY]]* [[ST_TY_TEMP:%.+]]) 188 // CHECK: call {{.*}} [[S_FLOAT_TY_COPY_CONSTR:@.+]]([[S_FLOAT_TY]]* {{.+}}, [[S_FLOAT_TY]]* {{.+}}, [[ST_TY]]* [[ST_TY_TEMP]]) 189 // CHECK: call {{.*}} [[ST_TY_DESTR:@.+]]([[ST_TY]]* [[ST_TY_TEMP]]) 190 // CHECK: br i1 {{.+}}, label %{{.+}}, label %[[S_ARR_BODY]] 191 192 // firstprivate var(var) 193 // CHECK: call {{.*}} [[ST_TY_DEFAULT_CONSTR]]([[ST_TY]]* [[ST_TY_TEMP:%.+]]) 194 // CHECK: call {{.*}} [[S_FLOAT_TY_COPY_CONSTR]]([[S_FLOAT_TY]]* [[VAR_PRIV]], [[S_FLOAT_TY]]* {{.*}} [[VAR]], [[ST_TY]]* [[ST_TY_TEMP]]) 195 // CHECK: call {{.*}} [[ST_TY_DESTR]]([[ST_TY]]* [[ST_TY_TEMP]]) 196 197 // Synchronization for initialization. 198 // CHECK: call void @__kmpc_barrier(%{{.+}}* [[IMPLICIT_BARRIER_LOC]], i{{[0-9]+}} [[GTID]]) 199 200 // CHECK: call void @__kmpc_for_static_init_4( 201 // CHECK: call void @__kmpc_for_static_fini( 202 203 // ~(firstprivate var), ~(firstprivate s_arr) 204 // CHECK-DAG: call {{.*}} [[S_FLOAT_TY_DESTR]]([[S_FLOAT_TY]]* [[VAR_PRIV]]) 205 // CHECK-DAG: call {{.*}} [[S_FLOAT_TY_DESTR]]([[S_FLOAT_TY]]* 206 // CHECK: call void @__kmpc_barrier(%{{.+}}* [[IMPLICIT_BARRIER_LOC]], i{{[0-9]+}} [[GTID]]) 207 208 // CHECK: = call {{.*}}i{{.+}} [[TMAIN_INT:@.+]]() 209 210 // CHECK: ret void 211 212 // CHECK: define {{.*}} i{{[0-9]+}} [[TMAIN_INT]]() 213 // CHECK: [[TEST:%.+]] = alloca [[S_INT_TY]], 214 // CHECK: call {{.*}} [[S_INT_TY_DEF_CONSTR:@.+]]([[S_INT_TY]]* [[TEST]]) 215 // 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 216 // CHECK: call {{.*}} [[S_INT_TY_DESTR:@.+]]([[S_INT_TY]]* 217 // CHECK: ret 218 // 219 // CHECK: define internal void [[TMAIN_MICROTASK]](i{{[0-9]+}}* [[GTID_ADDR:%.+]], i{{[0-9]+}}* %{{.+}}, [[CAP_TMAIN_TY]]* %{{.+}}) 220 // Skip temp vars for loop 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: alloca i{{[0-9]+}}, 226 // CHECK: [[T_VAR_PRIV:%.+]] = alloca i{{[0-9]+}}, 227 // CHECK: [[VEC_PRIV:%.+]] = alloca [2 x i{{[0-9]+}}], 228 // CHECK: [[S_ARR_PRIV:%.+]] = alloca [2 x [[S_INT_TY]]], 229 // CHECK: [[VAR_PRIV:%.+]] = alloca [[S_INT_TY]], 230 // CHECK: store i{{[0-9]+}}* [[GTID_ADDR]], i{{[0-9]+}}** [[GTID_ADDR_ADDR:%.+]], 231 232 // firstprivate t_var(t_var) 233 // CHECK: [[T_VAR_PTR_REF:%.+]] = getelementptr inbounds [[CAP_TMAIN_TY]], [[CAP_TMAIN_TY]]* %{{.+}}, i{{[0-9]+}} 0, i{{[0-9]+}} 0 234 // CHECK: [[T_VAR_REF:%.+]] = load i{{[0-9]+}}*, i{{[0-9]+}}** [[T_VAR_PTR_REF]], 235 // CHECK: [[T_VAR_VAL:%.+]] = load i{{[0-9]+}}, i{{[0-9]+}}* [[T_VAR_REF]], 236 // CHECK: store i{{[0-9]+}} [[T_VAR_VAL]], i{{[0-9]+}}* [[T_VAR_PRIV]], 237 238 // firstprivate vec(vec) 239 // CHECK: [[VEC_PTR_REF:%.+]] = getelementptr inbounds [[CAP_TMAIN_TY]], [[CAP_TMAIN_TY]]* %{{.+}}, i{{[0-9]+}} 0, i{{[0-9]+}} 1 240 // CHECK: [[VEC_REF:%.+]] = load [2 x i{{[0-9]+}}]*, [2 x i{{[0-9]+}}]** [[VEC_PTR_REF:%.+]], 241 // CHECK: [[VEC_DEST:%.+]] = bitcast [2 x i{{[0-9]+}}]* [[VEC_PRIV]] to i8* 242 // CHECK: [[VEC_SRC:%.+]] = bitcast [2 x i{{[0-9]+}}]* [[VEC_REF]] to i8* 243 // CHECK: call void @llvm.memcpy.{{.+}}(i8* [[VEC_DEST]], i8* [[VEC_SRC]], 244 245 // firstprivate s_arr(s_arr) 246 // CHECK: [[S_ARR_REF:%.+]] = getelementptr inbounds [[CAP_TMAIN_TY]], [[CAP_TMAIN_TY]]* %{{.+}}, i{{[0-9]+}} 0, i{{[0-9]+}} 2 247 // CHECK: [[S_ARR:%.+]] = load [2 x [[S_INT_TY]]]*, [2 x [[S_INT_TY]]]** [[S_ARR_REF]], 248 // 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]+}} 0 249 // CHECK: [[S_ARR_PRIV_END:%.+]] = getelementptr [[S_INT_TY]], [[S_INT_TY]]* [[S_ARR_PRIV_BEGIN]], i{{[0-9]+}} 2 250 // CHECK: [[IS_EMPTY:%.+]] = icmp eq [[S_INT_TY]]* [[S_ARR_PRIV_BEGIN]], [[S_ARR_PRIV_END]] 251 // CHECK: br i1 [[IS_EMPTY]], label %[[S_ARR_BODY_DONE:.+]], label %[[S_ARR_BODY:.+]] 252 // CHECK: [[S_ARR_BODY]] 253 // CHECK: call {{.*}} [[ST_TY_DEFAULT_CONSTR:@.+]]([[ST_TY]]* [[ST_TY_TEMP:%.+]]) 254 // CHECK: call {{.*}} [[S_INT_TY_COPY_CONSTR:@.+]]([[S_INT_TY]]* {{.+}}, [[S_INT_TY]]* {{.+}}, [[ST_TY]]* [[ST_TY_TEMP]]) 255 // CHECK: call {{.*}} [[ST_TY_DESTR:@.+]]([[ST_TY]]* [[ST_TY_TEMP]]) 256 // CHECK: br i1 {{.+}}, label %{{.+}}, label %[[S_ARR_BODY]] 257 258 // firstprivate var(var) 259 // CHECK: [[VAR_REF_PTR:%.+]] = getelementptr inbounds [[CAP_TMAIN_TY]], [[CAP_TMAIN_TY]]* %{{.+}}, i{{[0-9]+}} 0, i{{[0-9]+}} 3 260 // CHECK: [[VAR_REF:%.+]] = load [[S_INT_TY]]*, [[S_INT_TY]]** [[VAR_REF_PTR]], 261 // CHECK: call {{.*}} [[ST_TY_DEFAULT_CONSTR]]([[ST_TY]]* [[ST_TY_TEMP:%.+]]) 262 // CHECK: call {{.*}} [[S_INT_TY_COPY_CONSTR]]([[S_INT_TY]]* [[VAR_PRIV]], [[S_INT_TY]]* {{.*}} [[VAR_REF]], [[ST_TY]]* [[ST_TY_TEMP]]) 263 // CHECK: call {{.*}} [[ST_TY_DESTR]]([[ST_TY]]* [[ST_TY_TEMP]]) 264 265 // Synchronization for initialization. 266 // CHECK: [[GTID_REF:%.+]] = load i{{[0-9]+}}*, i{{[0-9]+}}** [[GTID_ADDR_ADDR]] 267 // CHECK: [[GTID:%.+]] = load i{{[0-9]+}}, i{{[0-9]+}}* [[GTID_REF]] 268 // CHECK: call void @__kmpc_barrier(%{{.+}}* [[IMPLICIT_BARRIER_LOC]], i{{[0-9]+}} [[GTID]]) 269 270 // CHECK: call void @__kmpc_for_static_init_4( 271 // CHECK: call void @__kmpc_for_static_fini( 272 273 // ~(firstprivate var), ~(firstprivate s_arr) 274 // CHECK-DAG: call {{.*}} [[S_INT_TY_DESTR]]([[S_INT_TY]]* [[VAR_PRIV]]) 275 // CHECK-DAG: call {{.*}} [[S_INT_TY_DESTR]]([[S_INT_TY]]* 276 // CHECK: [[GTID_REF:%.+]] = load i{{[0-9]+}}*, i{{[0-9]+}}** [[GTID_ADDR_ADDR]] 277 // CHECK: [[GTID:%.+]] = load i{{[0-9]+}}, i{{[0-9]+}}* [[GTID_REF]] 278 // CHECK: call i32 @__kmpc_cancel_barrier(%{{.+}}* [[IMPLICIT_BARRIER_LOC]], i{{[0-9]+}} [[GTID]]) 279 // CHECK: ret void 280 #endif 281 282