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 template <class T> 12 struct S { 13 T f; 14 S(T a) : f(a) {} 15 S() : f() {} 16 S<T> &operator=(const S<T> &); 17 operator T() { return T(); } 18 ~S() {} 19 }; 20 21 volatile int g = 1212; 22 volatile int &g1 = g; 23 float f; 24 char cnt; 25 26 struct SS { 27 int a; 28 int b : 4; 29 int &c; 30 SS(int &d) : a(0), b(0), c(d) { 31 #pragma omp parallel 32 #pragma omp for linear(a, b, c) 33 for (int i = 0; i < 2; ++i) 34 #ifdef LAMBDA 35 [&]() { 36 ++this->a, --b, (this)->c /= 1; 37 #pragma omp parallel 38 #pragma omp for linear(a, b) linear(ref(c)) 39 for (int i = 0; i < 2; ++i) 40 ++(this)->a, --b, this->c /= 1; 41 }(); 42 #elif defined(BLOCKS) 43 ^{ 44 ++a; 45 --this->b; 46 (this)->c /= 1; 47 #pragma omp parallel 48 #pragma omp for linear(a, b) linear(uval(c)) 49 for (int i = 0; i < 2; ++i) 50 ++(this)->a, --b, this->c /= 1; 51 }(); 52 #else 53 ++this->a, --b, c /= 1; 54 #endif 55 } 56 }; 57 58 template <typename T> 59 struct SST { 60 T a; 61 SST() : a(T()) { 62 #pragma omp parallel 63 #pragma omp for linear(a) 64 for (int i = 0; i < 2; ++i) 65 #ifdef LAMBDA 66 [&]() { 67 [&]() { 68 ++this->a; 69 #pragma omp parallel 70 #pragma omp for linear(a) 71 for (int i = 0; i < 2; ++i) 72 ++(this)->a; 73 }(); 74 }(); 75 #elif defined(BLOCKS) 76 ^{ 77 ^{ 78 ++a; 79 #pragma omp parallel 80 #pragma omp for linear(a) 81 for (int i = 0; i < 2; ++i) 82 ++(this)->a; 83 }(); 84 }(); 85 #else 86 ++(this)->a; 87 #endif 88 } 89 }; 90 91 // CHECK: [[SS_TY:%.+]] = type { i{{[0-9]+}}, i8 92 // LAMBDA: [[SS_TY:%.+]] = type { i{{[0-9]+}}, i8 93 // BLOCKS: [[SS_TY:%.+]] = type { i{{[0-9]+}}, i8 94 // CHECK: [[S_FLOAT_TY:%.+]] = type { float } 95 // CHECK: [[S_INT_TY:%.+]] = type { i32 } 96 // CHECK-DAG: [[IMPLICIT_BARRIER_LOC:@.+]] = private unnamed_addr constant %{{.+}} { i32 0, i32 66, i32 0, i32 0, i8* 97 // CHECK-DAG: [[F:@.+]] = global float 0.0 98 // CHECK-DAG: [[CNT:@.+]] = global i8 0 99 template <typename T> 100 T tmain() { 101 S<T> test; 102 SST<T> sst; 103 T *pvar = &test.f; 104 T &lvar = test.f; 105 #pragma omp parallel 106 #pragma omp for linear(pvar, lvar) 107 for (int i = 0; i < 2; ++i) { 108 ++pvar, ++lvar; 109 } 110 return T(); 111 } 112 113 int main() { 114 static int sivar; 115 SS ss(sivar); 116 #ifdef LAMBDA 117 // LAMBDA: [[G:@.+]] = global i{{[0-9]+}} 1212, 118 // LAMBDA-LABEL: @main 119 // LAMBDA: alloca [[SS_TY]], 120 // LAMBDA: alloca [[CAP_TY:%.+]], 121 // LAMBDA: call void [[OUTER_LAMBDA:@.+]]([[CAP_TY]]* 122 [&]() { 123 // LAMBDA: define{{.*}} internal{{.*}} void [[OUTER_LAMBDA]]( 124 // LAMBDA: call void {{.+}} @__kmpc_fork_call({{.+}}, i32 0, {{.+}}* [[OMP_REGION:@.+]] to {{.+}}) 125 #pragma omp parallel 126 #pragma omp for linear(g, g1:5) 127 for (int i = 0; i < 2; ++i) { 128 // LAMBDA: define {{.+}} @{{.+}}([[SS_TY]]* 129 // LAMBDA: getelementptr inbounds [[SS_TY]], [[SS_TY]]* %{{.+}}, i32 0, i32 0 130 // LAMBDA: store i{{[0-9]+}} 0, i{{[0-9]+}}* % 131 // LAMBDA: getelementptr inbounds [[SS_TY]], [[SS_TY]]* %{{.+}}, i32 0, i32 1 132 // LAMBDA: store i8 133 // LAMBDA: getelementptr inbounds [[SS_TY]], [[SS_TY]]* %{{.+}}, i32 0, i32 2 134 // LAMBDA: 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]+}}*, [[SS_TY]]*)* [[SS_MICROTASK:@.+]] to void 135 // LAMBDA: ret 136 137 // LAMBDA: define internal void [[SS_MICROTASK]](i{{[0-9]+}}* noalias [[GTID_ADDR:%.+]], i{{[0-9]+}}* noalias %{{.+}}, [[SS_TY]]* %{{.+}}) 138 // LAMBDA: getelementptr {{.*}}[[SS_TY]], [[SS_TY]]* %{{.*}}, i32 0, i32 0 139 // LAMBDA-NOT: getelementptr {{.*}}[[SS_TY]], [[SS_TY]]* %{{.*}}, i32 0, i32 1 140 // LAMBDA: getelementptr {{.*}}[[SS_TY]], [[SS_TY]]* %{{.*}}, i32 0, i32 2 141 // LAMBDA: call void @__kmpc_for_static_init_4( 142 // LAMBDA-NOT: getelementptr {{.*}}[[SS_TY]], [[SS_TY]]* 143 // LAMBDA: call{{.*}} void 144 // LAMBDA: call void @__kmpc_for_static_fini( 145 // LAMBDA: br i1 146 // LAMBDA: [[B_REF:%.+]] = getelementptr {{.*}}[[SS_TY]], [[SS_TY]]* %{{.*}}, i32 0, i32 1 147 // LAMBDA: store i8 %{{.+}}, i8* [[B_REF]], 148 // LAMBDA: br label 149 // LAMBDA: ret void 150 151 // LAMBDA: define internal void @{{.+}}(i{{[0-9]+}}* noalias [[GTID_ADDR:%.+]], i{{[0-9]+}}* noalias %{{.+}}, [[SS_TY]]* %{{.+}}, i32* {{.+}}, i32* {{.+}}, i32* {{.+}}) 152 // LAMBDA: alloca i{{[0-9]+}}, 153 // LAMBDA: alloca i{{[0-9]+}}, 154 // LAMBDA: alloca i{{[0-9]+}}, 155 // LAMBDA: alloca i{{[0-9]+}}, 156 // LAMBDA: alloca i{{[0-9]+}}, 157 // LAMBDA: alloca i{{[0-9]+}}, 158 // LAMBDA: alloca i{{[0-9]+}}, 159 // LAMBDA: alloca i{{[0-9]+}}, 160 // LAMBDA: alloca i{{[0-9]+}}, 161 // LAMBDA: [[A_PRIV:%.+]] = alloca i{{[0-9]+}}, 162 // LAMBDA: [[B_PRIV:%.+]] = alloca i{{[0-9]+}}, 163 // LAMBDA: [[C_PRIV:%.+]] = alloca i{{[0-9]+}}, 164 // LAMBDA: store i{{[0-9]+}}* [[A_PRIV]], i{{[0-9]+}}** [[REFA:%.+]], 165 // LAMBDA: store i{{[0-9]+}}* [[C_PRIV]], i{{[0-9]+}}** [[REFC:%.+]], 166 // LAMBDA: call void @__kmpc_for_static_init_4( 167 // LAMBDA: [[A_PRIV:%.+]] = load i{{[0-9]+}}*, i{{[0-9]+}}** [[REFA]], 168 // LAMBDA-NEXT: [[A_VAL:%.+]] = load i{{[0-9]+}}, i{{[0-9]+}}* [[A_PRIV]], 169 // LAMBDA-NEXT: [[INC:%.+]] = add nsw i{{[0-9]+}} [[A_VAL]], 1 170 // LAMBDA-NEXT: store i{{[0-9]+}} [[INC]], i{{[0-9]+}}* [[A_PRIV]], 171 // LAMBDA-NEXT: [[B_VAL:%.+]] = load i{{[0-9]+}}, i{{[0-9]+}}* [[B_PRIV]], 172 // LAMBDA-NEXT: [[DEC:%.+]] = add nsw i{{[0-9]+}} [[B_VAL]], -1 173 // LAMBDA-NEXT: store i{{[0-9]+}} [[DEC]], i{{[0-9]+}}* [[B_PRIV]], 174 // LAMBDA-NEXT: [[C_PRIV:%.+]] = load i{{[0-9]+}}*, i{{[0-9]+}}** [[REFC]], 175 // LAMBDA-NEXT: [[C_VAL:%.+]] = load i{{[0-9]+}}, i{{[0-9]+}}* [[C_PRIV]], 176 // LAMBDA-NEXT: [[DIV:%.+]] = sdiv i{{[0-9]+}} [[C_VAL]], 1 177 // LAMBDA-NEXT: store i{{[0-9]+}} [[DIV]], i{{[0-9]+}}* [[C_PRIV]], 178 // LAMBDA: call void @__kmpc_for_static_fini( 179 // LAMBDA: br i1 180 // LAMBDA: br label 181 // LAMBDA: ret void 182 183 // LAMBDA: define{{.*}} internal{{.*}} void [[OMP_REGION]](i32* noalias %{{.+}}, i32* noalias %{{.+}}) 184 // LAMBDA: alloca i{{[0-9]+}}, 185 // LAMBDA: [[G_START_ADDR:%.+]] = alloca i{{[0-9]+}}, 186 // LAMBDA: alloca i{{[0-9]+}}, 187 // LAMBDA: alloca i{{[0-9]+}}, 188 // LAMBDA: alloca i{{[0-9]+}}, 189 // LAMBDA: alloca i{{[0-9]+}}, 190 // LAMBDA: alloca i{{[0-9]+}}, 191 // LAMBDA: alloca i{{[0-9]+}}, 192 // LAMBDA: [[G_PRIVATE_ADDR:%.+]] = alloca i{{[0-9]+}}, 193 // LAMBDA: [[GTID_REF:%.+]] = load i{{[0-9]+}}*, i{{[0-9]+}}** %{{.+}} 194 // LAMBDA: [[GTID:%.+]] = load i{{[0-9]+}}, i{{[0-9]+}}* [[GTID_REF]] 195 // LAMBDA: call {{.+}} @__kmpc_for_static_init_4(%{{.+}}* @{{.+}}, i32 [[GTID]], i32 34, i32* [[IS_LAST_ADDR:%.+]], i32* %{{.+}}, i32* %{{.+}}, i32* %{{.+}}, i32 1, i32 1) 196 // LAMBDA: [[VAL:%.+]] = load i32, i32* [[G_START_ADDR]] 197 // LAMBDA: [[CNT:%.+]] = load i32, i32* 198 // LAMBDA: [[MUL:%.+]] = mul nsw i32 [[CNT]], 5 199 // LAMBDA: [[ADD:%.+]] = add nsw i32 [[VAL]], [[MUL]] 200 // LAMBDA: store i32 [[ADD]], i32* [[G_PRIVATE_ADDR]], 201 // LAMBDA: [[VAL:%.+]] = load i32, i32* [[G_PRIVATE_ADDR]], 202 // LAMBDA: [[ADD:%.+]] = add nsw i32 [[VAL]], 5 203 // LAMBDA: store i32 [[ADD]], i32* [[G_PRIVATE_ADDR]], 204 // LAMBDA: [[G_PRIVATE_ADDR_REF:%.+]] = getelementptr inbounds %{{.+}}, %{{.+}}* [[ARG:%.+]], i{{[0-9]+}} 0, i{{[0-9]+}} 0 205 // LAMBDA: store i{{[0-9]+}}* [[G_PRIVATE_ADDR]], i{{[0-9]+}}** [[G_PRIVATE_ADDR_REF]] 206 // LAMBDA: call void [[INNER_LAMBDA:@.+]](%{{.+}}* [[ARG]]) 207 // LAMBDA: call void @__kmpc_for_static_fini(%{{.+}}* @{{.+}}, i32 [[GTID]]) 208 g += 5; 209 g1 += 5; 210 // LAMBDA: call void @__kmpc_barrier(%{{.+}}* @{{.+}}, i{{[0-9]+}} [[GTID]]) 211 [&]() { 212 // LAMBDA: define {{.+}} void [[INNER_LAMBDA]](%{{.+}}* [[ARG_PTR:%.+]]) 213 // LAMBDA: store %{{.+}}* [[ARG_PTR]], %{{.+}}** [[ARG_PTR_REF:%.+]], 214 g = 2; 215 g1 = 2; 216 // LAMBDA: [[ARG_PTR:%.+]] = load %{{.+}}*, %{{.+}}** [[ARG_PTR_REF]] 217 // LAMBDA: [[G_PTR_REF:%.+]] = getelementptr inbounds %{{.+}}, %{{.+}}* [[ARG_PTR]], i{{[0-9]+}} 0, i{{[0-9]+}} 0 218 // LAMBDA: [[G_REF:%.+]] = load i{{[0-9]+}}*, i{{[0-9]+}}** [[G_PTR_REF]] 219 // LAMBDA: store i{{[0-9]+}} 2, i{{[0-9]+}}* [[G_REF]] 220 }(); 221 } 222 }(); 223 return 0; 224 #elif defined(BLOCKS) 225 // BLOCKS: [[G:@.+]] = global i{{[0-9]+}} 1212, 226 // BLOCKS-LABEL: @main 227 // BLOCKS: call 228 // BLOCKS: call void {{%.+}}(i8 229 ^{ 230 // BLOCKS: define{{.*}} internal{{.*}} void {{.+}}(i8* 231 // BLOCKS: call void {{.+}} @__kmpc_fork_call({{.+}}, i32 0, {{.+}}* [[OMP_REGION:@.+]] to {{.+}}) 232 #pragma omp parallel 233 #pragma omp for linear(g, g1:5) 234 for (int i = 0; i < 2; ++i) { 235 // BLOCKS: define{{.*}} internal{{.*}} void [[OMP_REGION]](i32* noalias %{{.+}}, i32* noalias %{{.+}}) 236 // BLOCKS: alloca i{{[0-9]+}}, 237 // BLOCKS: [[G_START_ADDR:%.+]] = alloca i{{[0-9]+}}, 238 // BLOCKS: alloca i{{[0-9]+}}, 239 // BLOCKS: alloca i{{[0-9]+}}, 240 // BLOCKS: alloca i{{[0-9]+}}, 241 // BLOCKS: alloca i{{[0-9]+}}, 242 // BLOCKS: alloca i{{[0-9]+}}, 243 // BLOCKS: alloca i{{[0-9]+}}, 244 // BLOCKS: [[G_PRIVATE_ADDR:%.+]] = alloca i{{[0-9]+}}, 245 // BLOCKS: [[GTID_REF:%.+]] = load i{{[0-9]+}}*, i{{[0-9]+}}** %{{.+}} 246 // BLOCKS: [[GTID:%.+]] = load i{{[0-9]+}}, i{{[0-9]+}}* [[GTID_REF]] 247 // BLOCKS: call {{.+}} @__kmpc_for_static_init_4(%{{.+}}* @{{.+}}, i32 [[GTID]], i32 34, i32* [[IS_LAST_ADDR:%.+]], i32* %{{.+}}, i32* %{{.+}}, i32* %{{.+}}, i32 1, i32 1) 248 // BLOCKS: [[VAL:%.+]] = load i32, i32* [[G_START_ADDR]] 249 // BLOCKS: [[CNT:%.+]] = load i32, i32* 250 // BLOCKS: [[MUL:%.+]] = mul nsw i32 [[CNT]], 5 251 // BLOCKS: [[ADD:%.+]] = add nsw i32 [[VAL]], [[MUL]] 252 // BLOCKS: store i32 [[ADD]], i32* [[G_PRIVATE_ADDR]], 253 // BLOCKS: [[VAL:%.+]] = load i32, i32* [[G_PRIVATE_ADDR]], 254 // BLOCKS: [[ADD:%.+]] = add nsw i32 [[VAL]], 5 255 // BLOCKS: store i32 [[ADD]], i32* [[G_PRIVATE_ADDR]], 256 // BLOCKS-NOT: [[G]]{{[[^:word:]]}} 257 // BLOCKS: i{{[0-9]+}}* [[G_PRIVATE_ADDR]] 258 // BLOCKS-NOT: [[G]]{{[[^:word:]]}} 259 // BLOCKS: call void {{%.+}}(i8 260 // BLOCKS: call void @__kmpc_for_static_fini(%{{.+}}* @{{.+}}, i32 [[GTID]]) 261 g += 5; 262 g1 += 5; 263 // BLOCKS: call void @__kmpc_barrier(%{{.+}}* @{{.+}}, i{{[0-9]+}} [[GTID]]) 264 g = 1; 265 g1 = 5; 266 ^{ 267 // BLOCKS: define {{.+}} void {{@.+}}(i8* 268 g = 2; 269 g1 = 2; 270 // BLOCKS-NOT: [[G]]{{[[^:word:]]}} 271 // BLOCKS: store i{{[0-9]+}} 2, i{{[0-9]+}}* 272 // BLOCKS-NOT: [[G]]{{[[^:word:]]}} 273 // BLOCKS: ret 274 }(); 275 } 276 }(); 277 return 0; 278 // BLOCKS: define {{.+}} @{{.+}}([[SS_TY]]* 279 // BLOCKS: getelementptr inbounds [[SS_TY]], [[SS_TY]]* %{{.+}}, i32 0, i32 0 280 // BLOCKS: store i{{[0-9]+}} 0, i{{[0-9]+}}* % 281 // BLOCKS: getelementptr inbounds [[SS_TY]], [[SS_TY]]* %{{.+}}, i32 0, i32 1 282 // BLOCKS: store i8 283 // BLOCKS: getelementptr inbounds [[SS_TY]], [[SS_TY]]* %{{.+}}, i32 0, i32 2 284 // BLOCKS: 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]+}}*, [[SS_TY]]*)* [[SS_MICROTASK:@.+]] to void 285 // BLOCKS: ret 286 287 // BLOCKS: define internal void [[SS_MICROTASK]](i{{[0-9]+}}* noalias [[GTID_ADDR:%.+]], i{{[0-9]+}}* noalias %{{.+}}, [[SS_TY]]* %{{.+}}) 288 // BLOCKS: getelementptr {{.*}}[[SS_TY]], [[SS_TY]]* %{{.*}}, i32 0, i32 0 289 // BLOCKS-NOT: getelementptr {{.*}}[[SS_TY]], [[SS_TY]]* %{{.*}}, i32 0, i32 1 290 // BLOCKS: getelementptr {{.*}}[[SS_TY]], [[SS_TY]]* %{{.*}}, i32 0, i32 2 291 // BLOCKS: call void @__kmpc_for_static_init_4( 292 // BLOCKS-NOT: getelementptr {{.*}}[[SS_TY]], [[SS_TY]]* 293 // BLOCKS: call{{.*}} void 294 // BLOCKS: call void @__kmpc_for_static_fini( 295 // BLOCKS: br i1 296 // BLOCKS: [[B_REF:%.+]] = getelementptr {{.*}}[[SS_TY]], [[SS_TY]]* %{{.*}}, i32 0, i32 1 297 // BLOCKS: store i8 %{{.+}}, i8* [[B_REF]], 298 // BLOCKS: br label 299 // BLOCKS: ret void 300 301 // BLOCKS: define internal void @{{.+}}(i{{[0-9]+}}* noalias [[GTID_ADDR:%.+]], i{{[0-9]+}}* noalias %{{.+}}, [[SS_TY]]* %{{.+}}, i32* {{.+}}, i32* {{.+}}, i32* {{.+}}) 302 // BLOCKS: alloca i{{[0-9]+}}, 303 // BLOCKS: alloca i{{[0-9]+}}, 304 // BLOCKS: alloca i{{[0-9]+}}, 305 // BLOCKS: alloca i{{[0-9]+}}, 306 // BLOCKS: alloca i{{[0-9]+}}, 307 // BLOCKS: alloca i{{[0-9]+}}, 308 // BLOCKS: alloca i{{[0-9]+}}, 309 // BLOCKS: alloca i{{[0-9]+}}, 310 // BLOCKS: alloca i{{[0-9]+}}, 311 // BLOCKS: [[A_PRIV:%.+]] = alloca i{{[0-9]+}}, 312 // BLOCKS: [[B_PRIV:%.+]] = alloca i{{[0-9]+}}, 313 // BLOCKS: [[C_PRIV:%.+]] = alloca i{{[0-9]+}}, 314 // BLOCKS: store i{{[0-9]+}}* [[A_PRIV]], i{{[0-9]+}}** [[REFA:%.+]], 315 // BLOCKS: store i{{[0-9]+}}* [[C_PRIV]], i{{[0-9]+}}** [[REFC:%.+]], 316 // BLOCKS: call void @__kmpc_for_static_init_4( 317 // BLOCKS: [[A_PRIV:%.+]] = load i{{[0-9]+}}*, i{{[0-9]+}}** [[REFA]], 318 // BLOCKS-NEXT: [[A_VAL:%.+]] = load i{{[0-9]+}}, i{{[0-9]+}}* [[A_PRIV]], 319 // BLOCKS-NEXT: [[INC:%.+]] = add nsw i{{[0-9]+}} [[A_VAL]], 1 320 // BLOCKS-NEXT: store i{{[0-9]+}} [[INC]], i{{[0-9]+}}* [[A_PRIV]], 321 // BLOCKS-NEXT: [[B_VAL:%.+]] = load i{{[0-9]+}}, i{{[0-9]+}}* [[B_PRIV]], 322 // BLOCKS-NEXT: [[DEC:%.+]] = add nsw i{{[0-9]+}} [[B_VAL]], -1 323 // BLOCKS-NEXT: store i{{[0-9]+}} [[DEC]], i{{[0-9]+}}* [[B_PRIV]], 324 // BLOCKS-NEXT: [[C_PRIV:%.+]] = load i{{[0-9]+}}*, i{{[0-9]+}}** [[REFC]], 325 // BLOCKS-NEXT: [[C_VAL:%.+]] = load i{{[0-9]+}}, i{{[0-9]+}}* [[C_PRIV]], 326 // BLOCKS-NEXT: [[DIV:%.+]] = sdiv i{{[0-9]+}} [[C_VAL]], 1 327 // BLOCKS-NEXT: store i{{[0-9]+}} [[DIV]], i{{[0-9]+}}* [[C_PRIV]], 328 // BLOCKS: call void @__kmpc_for_static_fini( 329 // BLOCKS: br i1 330 // BLOCKS: br label 331 // BLOCKS: ret void 332 #else 333 S<float> test; 334 float *pvar = &test.f; 335 long long lvar = 0; 336 #pragma omp parallel 337 #pragma omp for linear(pvar, lvar : 3) 338 for (int i = 0; i < 2; ++i) { 339 pvar += 3, lvar += 3; 340 } 341 return tmain<int>(); 342 #endif 343 } 344 345 // CHECK: define i{{[0-9]+}} @main() 346 // CHECK: [[TEST:%.+]] = alloca [[S_FLOAT_TY]], 347 // CHECK: call {{.*}} [[S_FLOAT_TY_DEF_CONSTR:@.+]]([[S_FLOAT_TY]]* [[TEST]]) 348 // CHECK: call void (%{{.+}}*, i{{[0-9]+}}, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)*, ...) @__kmpc_fork_call(%{{.+}}* @{{.+}}, i{{[0-9]+}} 2, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)* bitcast (void (i{{[0-9]+}}*, i{{[0-9]+}}*, float**, i64*)* [[MAIN_MICROTASK:@.+]] to void 349 // CHECK: = call {{.+}} [[TMAIN_INT:@.+]]() 350 // CHECK: call void [[S_FLOAT_TY_DESTR:@.+]]([[S_FLOAT_TY]]* 351 // CHECK: ret 352 353 // CHECK: define internal void [[MAIN_MICROTASK]](i{{[0-9]+}}* noalias [[GTID_ADDR:%.+]], i{{[0-9]+}}* noalias %{{.+}}, float** dereferenceable(8) %{{.+}}, i64* dereferenceable(8) %{{.+}}) 354 // CHECK: alloca i{{[0-9]+}}, 355 // CHECK: [[PVAR_START:%.+]] = alloca float*, 356 // CHECK: [[LVAR_START:%.+]] = alloca i64, 357 // CHECK: alloca i{{[0-9]+}}, 358 // CHECK: alloca i{{[0-9]+}}, 359 // CHECK: alloca i{{[0-9]+}}, 360 // CHECK: alloca i{{[0-9]+}}, 361 // CHECK: alloca i{{[0-9]+}}, 362 // CHECK: [[PVAR_PRIV:%.+]] = alloca float*, 363 // CHECK: [[LVAR_PRIV:%.+]] = alloca i64, 364 // CHECK: store i{{[0-9]+}}* [[GTID_ADDR]], i{{[0-9]+}}** [[GTID_ADDR_REF:%.+]] 365 366 // Check for default initialization. 367 // CHECK: [[PVAR_REF:%.+]] = load float**, float*** % 368 // CHECK: [[LVAR_REF:%.+]] = load i64*, i64** % 369 // CHECK: [[PVAR_VAL:%.+]] = load float*, float** [[PVAR_REF]], 370 // CHECK: store float* [[PVAR_VAL]], float** [[PVAR_START]], 371 // CHECK: [[LVAR_VAL:%.+]] = load i64, i64* [[LVAR_REF]], 372 // CHECK: store i64 [[LVAR_VAL]], i64* [[LVAR_START]], 373 // CHECK: call {{.+}} @__kmpc_for_static_init_4(%{{.+}}* @{{.+}}, i32 [[GTID:%.+]], i32 34, i32* [[IS_LAST_ADDR:%.+]], i32* %{{.+}}, i32* %{{.+}}, i32* %{{.+}}, i32 1, i32 1) 374 // CHECK: [[PVAR_VAL:%.+]] = load float*, float** [[PVAR_START]], 375 // CHECK: [[CNT:%.+]] = load i32, i32* 376 // CHECK: [[MUL:%.+]] = mul nsw i32 [[CNT]], 3 377 // CHECK: [[IDX:%.+]] = sext i32 [[MUL]] to i64 378 // CHECK: [[PTR:%.+]] = getelementptr inbounds float, float* [[PVAR_VAL]], i64 [[IDX]] 379 // CHECK: store float* [[PTR]], float** [[PVAR_PRIV]], 380 // CHECK: [[LVAR_VAL:%.+]] = load i64, i64* [[LVAR_START]], 381 // CHECK: [[CNT:%.+]] = load i32, i32* 382 // CHECK: [[MUL:%.+]] = mul nsw i32 [[CNT]], 3 383 // CHECK: [[CONV:%.+]] = sext i32 [[MUL]] to i64 384 // CHECK: [[VAL:%.+]] = add nsw i64 [[LVAR_VAL]], [[CONV]] 385 // CHECK: store i64 [[VAL]], i64* [[LVAR_PRIV]], 386 // CHECK: [[PVAR_VAL:%.+]] = load float*, float** [[PVAR_PRIV]] 387 // CHECK: [[PTR:%.+]] = getelementptr inbounds float, float* [[PVAR_VAL]], i64 3 388 // CHECK: store float* [[PTR]], float** [[PVAR_PRIV]], 389 // CHECK: [[LVAR_VAL:%.+]] = load i64, i64* [[LVAR_PRIV]], 390 // CHECK: [[ADD:%.+]] = add nsw i64 [[LVAR_VAL]], 3 391 // CHECK: store i64 [[ADD]], i64* [[LVAR_PRIV]], 392 // CHECK: call void @__kmpc_for_static_fini(%{{.+}}* @{{.+}}, i32 %{{.+}}) 393 // CHECK: call void @__kmpc_barrier(%{{.+}}* [[IMPLICIT_BARRIER_LOC]], i{{[0-9]+}} [[GTID]]) 394 // CHECK: ret void 395 396 // CHECK: define {{.*}} i{{[0-9]+}} [[TMAIN_INT]]() 397 // CHECK: [[TEST:%.+]] = alloca [[S_INT_TY]], 398 // CHECK: call {{.*}} [[S_INT_TY_DEF_CONSTR:@.+]]([[S_INT_TY]]* [[TEST]]) 399 // CHECK: call void (%{{.+}}*, i{{[0-9]+}}, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)*, ...) @__kmpc_fork_call(%{{.+}}* @{{.+}}, i{{[0-9]+}} 2, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)* bitcast (void (i{{[0-9]+}}*, i{{[0-9]+}}*, i32**, i32*)* [[TMAIN_MICROTASK:@.+]] to void 400 // CHECK: call void [[S_INT_TY_DESTR:@.+]]([[S_INT_TY]]* 401 // CHECK: ret 402 403 // CHECK: define {{.+}} @{{.+}}([[SS_TY]]* 404 // CHECK: getelementptr inbounds [[SS_TY]], [[SS_TY]]* %{{.+}}, i32 0, i32 0 405 // CHECK: store i{{[0-9]+}} 0, i{{[0-9]+}}* % 406 // CHECK: getelementptr inbounds [[SS_TY]], [[SS_TY]]* %{{.+}}, i32 0, i32 1 407 // CHECK: store i8 408 // CHECK: getelementptr inbounds [[SS_TY]], [[SS_TY]]* %{{.+}}, i32 0, i32 2 409 // 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]+}}*, [[SS_TY]]*)* [[SS_MICROTASK:@.+]] to void 410 // CHECK: ret 411 412 // CHECK: define internal void [[SS_MICROTASK]](i{{[0-9]+}}* noalias [[GTID_ADDR:%.+]], i{{[0-9]+}}* noalias %{{.+}}, [[SS_TY]]* %{{.+}}) 413 // CHECK: alloca i{{[0-9]+}}, 414 // CHECK: alloca i{{[0-9]+}}, 415 // CHECK: alloca i{{[0-9]+}}, 416 // CHECK: alloca i{{[0-9]+}}, 417 // CHECK: alloca i{{[0-9]+}}, 418 // CHECK: alloca i{{[0-9]+}}, 419 // CHECK: alloca i{{[0-9]+}}, 420 // CHECK: alloca i{{[0-9]+}}, 421 // CHECK: alloca i{{[0-9]+}}, 422 // CHECK: alloca i{{[0-9]+}}, 423 // CHECK: [[A_PRIV:%.+]] = alloca i{{[0-9]+}}, 424 // CHECK: [[B_PRIV:%.+]] = alloca i{{[0-9]+}}, 425 // CHECK: [[C_PRIV:%.+]] = alloca i{{[0-9]+}}, 426 // CHECK: store i{{[0-9]+}}* [[A_PRIV]], i{{[0-9]+}}** [[REFA:%.+]], 427 // CHECK: store i{{[0-9]+}}* [[C_PRIV]], i{{[0-9]+}}** [[REFC:%.+]], 428 // CHECK: call void @__kmpc_for_static_init_4( 429 // CHECK: [[A_PRIV:%.+]] = load i{{[0-9]+}}*, i{{[0-9]+}}** [[REFA]], 430 // CHECK-NEXT: [[A_VAL:%.+]] = load i{{[0-9]+}}, i{{[0-9]+}}* [[A_PRIV]], 431 // CHECK-NEXT: [[INC:%.+]] = add nsw i{{[0-9]+}} [[A_VAL]], 1 432 // CHECK-NEXT: store i{{[0-9]+}} [[INC]], i{{[0-9]+}}* [[A_PRIV]], 433 // CHECK-NEXT: [[B_VAL:%.+]] = load i{{[0-9]+}}, i{{[0-9]+}}* [[B_PRIV]], 434 // CHECK-NEXT: [[DEC:%.+]] = add nsw i{{[0-9]+}} [[B_VAL]], -1 435 // CHECK-NEXT: store i{{[0-9]+}} [[DEC]], i{{[0-9]+}}* [[B_PRIV]], 436 // CHECK-NEXT: [[C_PRIV:%.+]] = load i{{[0-9]+}}*, i{{[0-9]+}}** [[REFC]], 437 // CHECK-NEXT: [[C_VAL:%.+]] = load i{{[0-9]+}}, i{{[0-9]+}}* [[C_PRIV]], 438 // CHECK-NEXT: [[DIV:%.+]] = sdiv i{{[0-9]+}} [[C_VAL]], 1 439 // CHECK-NEXT: store i{{[0-9]+}} [[DIV]], i{{[0-9]+}}* [[C_PRIV]], 440 // CHECK: call void @__kmpc_for_static_fini( 441 // CHECK: br i1 442 // CHECK: [[B_REF:%.+]] = getelementptr {{.*}}[[SS_TY]], [[SS_TY]]* %{{.*}}, i32 0, i32 1 443 // CHECK: store i8 %{{.+}}, i8* [[B_REF]], 444 // CHECK: br label 445 // CHECK: ret void 446 447 // CHECK: define internal void [[TMAIN_MICROTASK]](i{{[0-9]+}}* noalias [[GTID_ADDR:%.+]], i{{[0-9]+}}* noalias %{{.+}}, i32** dereferenceable(8) %{{.+}}, i32* dereferenceable(4) %{{.+}}) 448 // CHECK: alloca i{{[0-9]+}}, 449 // CHECK: [[PVAR_START:%.+]] = alloca i32*, 450 // CHECK: [[LVAR_START:%.+]] = alloca i32, 451 // CHECK: alloca i{{[0-9]+}}, 452 // CHECK: alloca i{{[0-9]+}}, 453 // CHECK: alloca i{{[0-9]+}}, 454 // CHECK: alloca i{{[0-9]+}}, 455 // CHECK: alloca i{{[0-9]+}}, 456 // CHECK: [[PVAR_PRIV:%.+]] = alloca i32*, 457 // CHECK: [[LVAR_PRIV:%.+]] = alloca i32, 458 // CHECK: [[LVAR_PRIV_REF:%.+]] = alloca i32*, 459 // CHECK: store i{{[0-9]+}}* [[GTID_ADDR]], i{{[0-9]+}}** [[GTID_ADDR_REF:%.+]] 460 461 // Check for default initialization. 462 // CHECK: [[PVAR_REF:%.+]] = load i32**, i32*** % 463 // CHECK: [[PVAR_VAL:%.+]] = load i32*, i32** [[PVAR_REF]], 464 // CHECK: store i32* [[PVAR_VAL]], i32** [[PVAR_START]], 465 // CHECK: [[LVAR_REF:%.+]] = load i32*, i32** % 466 // CHECK: [[LVAR_VAL:%.+]] = load i32, i32* [[LVAR_REF]], 467 // CHECK: store i32 [[LVAR_VAL]], i32* [[LVAR_START]], 468 // CHECK: store i32* [[LVAR_PRIV]], i32** [[LVAR_PRIV_REF]], 469 470 // CHECK: call {{.+}} @__kmpc_for_static_init_4(%{{.+}}* @{{.+}}, i32 [[GTID:%.+]], i32 34, i32* [[IS_LAST_ADDR:%.+]], i32* %{{.+}}, i32* %{{.+}}, i32* %{{.+}}, i32 1, i32 1) 471 // CHECK: [[PVAR_VAL:%.+]] = load i32*, i32** [[PVAR_START]], 472 // CHECK: [[CNT:%.+]] = load i32, i32* 473 // CHECK: [[MUL:%.+]] = mul nsw i32 [[CNT]], 1 474 // CHECK: [[IDX:%.+]] = sext i32 [[MUL]] to i64 475 // CHECK: [[PTR:%.+]] = getelementptr inbounds i32, i32* [[PVAR_VAL]], i64 [[IDX]] 476 // CHECK: store i32* [[PTR]], i32** [[PVAR_PRIV]], 477 // CHECK: [[LVAR_VAL:%.+]] = load i32, i32* [[LVAR_START]], 478 // CHECK: [[CNT:%.+]] = load i32, i32* 479 // CHECK: [[MUL:%.+]] = mul nsw i32 [[CNT]], 1 480 // CHECK: [[VAL:%.+]] = add nsw i32 [[LVAR_VAL]], [[MUL]] 481 // CHECK: store i32 [[VAL]], i32* [[LVAR_PRIV]], 482 // CHECK: [[PVAR_VAL:%.+]] = load i32*, i32** [[PVAR_PRIV]] 483 // CHECK: [[PTR:%.+]] = getelementptr inbounds i32, i32* [[PVAR_VAL]], i32 1 484 // CHECK: store i32* [[PTR]], i32** [[PVAR_PRIV]], 485 // CHECK: [[LVAR_PRIV:%.+]] = load i32*, i32** [[LVAR_PRIV_REF]], 486 // CHECK: [[LVAR_VAL:%.+]] = load i32, i32* [[LVAR_PRIV]], 487 // CHECK: [[ADD:%.+]] = add nsw i32 [[LVAR_VAL]], 1 488 // CHECK: store i32 [[ADD]], i32* [[LVAR_PRIV]], 489 // CHECK: call void @__kmpc_for_static_fini(%{{.+}}* @{{.+}}, i32 %{{.+}}) 490 // CHECK: call void @__kmpc_barrier(%{{.+}}* [[IMPLICIT_BARRIER_LOC]], i{{[0-9]+}} [[GTID]]) 491 // CHECK: ret void 492 #endif 493 494
