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 7 // RUN: %clang_cc1 -verify -fopenmp-simd -x c++ -triple x86_64-apple-darwin10 -emit-llvm %s -o - | FileCheck --check-prefix SIMD-ONLY0 %s 8 // RUN: %clang_cc1 -fopenmp-simd -x c++ -std=c++11 -triple x86_64-apple-darwin10 -emit-pch -o %t %s 9 // RUN: %clang_cc1 -fopenmp-simd -x c++ -triple x86_64-apple-darwin10 -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck --check-prefix SIMD-ONLY0 %s 10 // RUN: %clang_cc1 -verify -fopenmp-simd -x c++ -std=c++11 -DLAMBDA -triple x86_64-apple-darwin10 -emit-llvm %s -o - | FileCheck --check-prefix SIMD-ONLY0 %s 11 // RUN: %clang_cc1 -verify -fopenmp-simd -x c++ -fblocks -DBLOCKS -triple x86_64-apple-darwin10 -emit-llvm %s -o - | FileCheck --check-prefix SIMD-ONLY0 %s 12 // SIMD-ONLY0-NOT: {{__kmpc|__tgt}} 13 // expected-no-diagnostics 14 #ifndef HEADER 15 #define HEADER 16 17 enum omp_allocator_handle_t { 18 omp_null_allocator = 0, 19 omp_default_mem_alloc = 1, 20 omp_large_cap_mem_alloc = 2, 21 omp_const_mem_alloc = 3, 22 omp_high_bw_mem_alloc = 4, 23 omp_low_lat_mem_alloc = 5, 24 omp_cgroup_mem_alloc = 6, 25 omp_pteam_mem_alloc = 7, 26 omp_thread_mem_alloc = 8, 27 KMP_ALLOCATOR_MAX_HANDLE = __UINTPTR_MAX__ 28 }; 29 30 template <class T> 31 struct S { 32 T f; 33 S(T a) : f(a) {} 34 S() : f() {} 35 S<T> &operator=(const S<T> &); 36 operator T() { return T(); } 37 ~S() {} 38 }; 39 40 volatile int g = 1212; 41 volatile int &g1 = g; 42 float f; 43 char cnt; 44 45 struct SS { 46 int a; 47 int b : 4; 48 int &c; 49 SS(int &d) : a(0), b(0), c(d) { 50 #pragma omp parallel 51 #pragma omp for linear(a, b, c) 52 for (int i = 0; i < 2; ++i) 53 #ifdef LAMBDA 54 [&]() { 55 ++this->a, --b, (this)->c /= 1; 56 #pragma omp parallel 57 #pragma omp for linear(a, b) linear(ref(c)) 58 for (int i = 0; i < 2; ++i) 59 ++(this)->a, --b, this->c /= 1; 60 }(); 61 #elif defined(BLOCKS) 62 ^{ 63 ++a; 64 --this->b; 65 (this)->c /= 1; 66 #pragma omp parallel 67 #pragma omp for linear(a, b) linear(uval(c)) 68 for (int i = 0; i < 2; ++i) 69 ++(this)->a, --b, this->c /= 1; 70 }(); 71 #else 72 ++this->a, --b, c /= 1; 73 #endif 74 } 75 }; 76 77 template <typename T> 78 struct SST { 79 T a; 80 SST() : a(T()) { 81 #pragma omp parallel 82 #pragma omp for linear(a) 83 for (int i = 0; i < 2; ++i) 84 #ifdef LAMBDA 85 [&]() { 86 [&]() { 87 ++this->a; 88 #pragma omp parallel 89 #pragma omp for linear(a) 90 for (int i = 0; i < 2; ++i) 91 ++(this)->a; 92 }(); 93 }(); 94 #elif defined(BLOCKS) 95 ^{ 96 ^{ 97 ++a; 98 #pragma omp parallel 99 #pragma omp for linear(a) 100 for (int i = 0; i < 2; ++i) 101 ++(this)->a; 102 }(); 103 }(); 104 #else 105 ++(this)->a; 106 #endif 107 } 108 }; 109 110 // CHECK: [[SS_TY:%.+]] = type { i{{[0-9]+}}, i8 111 // LAMBDA: [[SS_TY:%.+]] = type { i{{[0-9]+}}, i8 112 // BLOCKS: [[SS_TY:%.+]] = type { i{{[0-9]+}}, i8 113 // CHECK: [[S_FLOAT_TY:%.+]] = type { float } 114 // CHECK: [[S_INT_TY:%.+]] = type { i32 } 115 // CHECK-DAG: [[IMPLICIT_BARRIER_LOC:@.+]] = private unnamed_addr constant %{{.+}} { i32 0, i32 66, i32 0, i32 0, i8* 116 // CHECK-DAG: [[F:@.+]] = global float 0.0 117 // CHECK-DAG: [[CNT:@.+]] = global i8 0 118 template <typename T> 119 T tmain() { 120 S<T> test; 121 SST<T> sst; 122 T *pvar = &test.f; 123 T &lvar = test.f; 124 #pragma omp parallel 125 #pragma omp for linear(pvar, lvar) 126 for (int i = 0; i < 2; ++i) { 127 ++pvar, ++lvar; 128 } 129 return T(); 130 } 131 132 int main() { 133 static int sivar; 134 SS ss(sivar); 135 #ifdef LAMBDA 136 // LAMBDA: [[G:@.+]] = global i{{[0-9]+}} 1212, 137 // LAMBDA-LABEL: @main 138 // LAMBDA: alloca [[SS_TY]], 139 // LAMBDA: alloca [[CAP_TY:%.+]], 140 // LAMBDA: call void [[OUTER_LAMBDA:@.+]]([[CAP_TY]]* 141 [&]() { 142 // LAMBDA: define{{.*}} internal{{.*}} void [[OUTER_LAMBDA]]( 143 // LAMBDA: call void {{.+}} @__kmpc_fork_call({{.+}}, i32 0, {{.+}}* [[OMP_REGION:@.+]] to {{.+}}) 144 #pragma omp parallel 145 #pragma omp for linear(g, g1:5) 146 for (int i = 0; i < 2; ++i) { 147 // LAMBDA: define {{.+}} @{{.+}}([[SS_TY]]* 148 // LAMBDA: getelementptr inbounds [[SS_TY]], [[SS_TY]]* %{{.+}}, i32 0, i32 0 149 // LAMBDA: store i{{[0-9]+}} 0, i{{[0-9]+}}* % 150 // LAMBDA: getelementptr inbounds [[SS_TY]], [[SS_TY]]* %{{.+}}, i32 0, i32 1 151 // LAMBDA: store i8 152 // LAMBDA: getelementptr inbounds [[SS_TY]], [[SS_TY]]* %{{.+}}, i32 0, i32 2 153 // 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 154 // LAMBDA: ret 155 156 // LAMBDA: define internal void [[SS_MICROTASK]](i{{[0-9]+}}* noalias [[GTID_ADDR:%.+]], i{{[0-9]+}}* noalias %{{.+}}, [[SS_TY]]* %{{.+}}) 157 // LAMBDA: getelementptr {{.*}}[[SS_TY]], [[SS_TY]]* %{{.*}}, i32 0, i32 0 158 // LAMBDA-NOT: getelementptr {{.*}}[[SS_TY]], [[SS_TY]]* %{{.*}}, i32 0, i32 1 159 // LAMBDA: getelementptr {{.*}}[[SS_TY]], [[SS_TY]]* %{{.*}}, i32 0, i32 2 160 // LAMBDA: call void @__kmpc_for_static_init_4( 161 // LAMBDA-NOT: getelementptr {{.*}}[[SS_TY]], [[SS_TY]]* 162 // LAMBDA: call{{.*}} void 163 // LAMBDA: call void @__kmpc_for_static_fini( 164 // LAMBDA: br i1 165 // LAMBDA: [[B_REF:%.+]] = getelementptr {{.*}}[[SS_TY]], [[SS_TY]]* %{{.*}}, i32 0, i32 1 166 // LAMBDA: store i8 %{{.+}}, i8* [[B_REF]], 167 // LAMBDA: br label 168 // LAMBDA: ret void 169 170 // LAMBDA: define internal void @{{.+}}(i{{[0-9]+}}* noalias [[GTID_ADDR:%.+]], i{{[0-9]+}}* noalias %{{.+}}, [[SS_TY]]* %{{.+}}, i32* {{.+}}, i32* {{.+}}, i32* {{.+}}) 171 // LAMBDA: alloca i{{[0-9]+}}, 172 // LAMBDA: alloca i{{[0-9]+}}, 173 // LAMBDA: alloca i{{[0-9]+}}, 174 // LAMBDA: alloca i{{[0-9]+}}, 175 // LAMBDA: alloca i{{[0-9]+}}, 176 // LAMBDA: alloca i{{[0-9]+}}, 177 // LAMBDA: alloca i{{[0-9]+}}, 178 // LAMBDA: alloca i{{[0-9]+}}, 179 // LAMBDA: alloca i{{[0-9]+}}, 180 // LAMBDA: alloca i{{[0-9]+}}, 181 // LAMBDA: [[A_PRIV:%.+]] = alloca i{{[0-9]+}}, 182 // LAMBDA: [[B_PRIV:%.+]] = alloca i{{[0-9]+}}, 183 // LAMBDA: [[C_PRIV:%.+]] = alloca i{{[0-9]+}}, 184 // LAMBDA: store i{{[0-9]+}}* [[A_PRIV]], i{{[0-9]+}}** [[REFA:%.+]], 185 // LAMBDA: store i{{[0-9]+}}* [[C_PRIV]], i{{[0-9]+}}** [[REFC:%.+]], 186 // LAMBDA: call void @__kmpc_for_static_init_4( 187 // LAMBDA: [[A_PRIV:%.+]] = load i{{[0-9]+}}*, i{{[0-9]+}}** [[REFA]], 188 // LAMBDA-NEXT: [[A_VAL:%.+]] = load i{{[0-9]+}}, i{{[0-9]+}}* [[A_PRIV]], 189 // LAMBDA-NEXT: [[INC:%.+]] = add nsw i{{[0-9]+}} [[A_VAL]], 1 190 // LAMBDA-NEXT: store i{{[0-9]+}} [[INC]], i{{[0-9]+}}* [[A_PRIV]], 191 // LAMBDA-NEXT: [[B_VAL:%.+]] = load i{{[0-9]+}}, i{{[0-9]+}}* [[B_PRIV]], 192 // LAMBDA-NEXT: [[DEC:%.+]] = add nsw i{{[0-9]+}} [[B_VAL]], -1 193 // LAMBDA-NEXT: store i{{[0-9]+}} [[DEC]], i{{[0-9]+}}* [[B_PRIV]], 194 // LAMBDA-NEXT: [[C_PRIV:%.+]] = load i{{[0-9]+}}*, i{{[0-9]+}}** [[REFC]], 195 // LAMBDA-NEXT: [[C_VAL:%.+]] = load i{{[0-9]+}}, i{{[0-9]+}}* [[C_PRIV]], 196 // LAMBDA-NEXT: [[DIV:%.+]] = sdiv i{{[0-9]+}} [[C_VAL]], 1 197 // LAMBDA-NEXT: store i{{[0-9]+}} [[DIV]], i{{[0-9]+}}* [[C_PRIV]], 198 // LAMBDA: call void @__kmpc_for_static_fini( 199 // LAMBDA: br i1 200 // LAMBDA: br label 201 // LAMBDA: ret void 202 203 // LAMBDA: define{{.*}} internal{{.*}} void [[OMP_REGION]](i32* noalias %{{.+}}, i32* noalias %{{.+}}) 204 // LAMBDA: alloca i{{[0-9]+}}, 205 // LAMBDA: alloca i{{[0-9]+}}, 206 // LAMBDA: [[G_START_ADDR:%.+]] = alloca i{{[0-9]+}}, 207 // LAMBDA: alloca i{{[0-9]+}}, 208 // LAMBDA: alloca i{{[0-9]+}}, 209 // LAMBDA: alloca i{{[0-9]+}}, 210 // LAMBDA: alloca i{{[0-9]+}}, 211 // LAMBDA: alloca i{{[0-9]+}}, 212 // LAMBDA: alloca i{{[0-9]+}}, 213 // LAMBDA: [[G_PRIVATE_ADDR:%.+]] = alloca i{{[0-9]+}}, 214 // LAMBDA: [[GTID_REF:%.+]] = load i{{[0-9]+}}*, i{{[0-9]+}}** %{{.+}} 215 // LAMBDA: [[GTID:%.+]] = load i{{[0-9]+}}, i{{[0-9]+}}* [[GTID_REF]] 216 // LAMBDA: call {{.+}} @__kmpc_for_static_init_4(%{{.+}}* @{{.+}}, i32 [[GTID]], i32 34, i32* [[IS_LAST_ADDR:%.+]], i32* %{{.+}}, i32* %{{.+}}, i32* %{{.+}}, i32 1, i32 1) 217 // LAMBDA: [[VAL:%.+]] = load i32, i32* [[G_START_ADDR]] 218 // LAMBDA: [[CNT:%.+]] = load i32, i32* 219 // LAMBDA: [[MUL:%.+]] = mul nsw i32 [[CNT]], 5 220 // LAMBDA: [[ADD:%.+]] = add nsw i32 [[VAL]], [[MUL]] 221 // LAMBDA: store i32 [[ADD]], i32* [[G_PRIVATE_ADDR]], 222 // LAMBDA: [[VAL:%.+]] = load i32, i32* [[G_PRIVATE_ADDR]], 223 // LAMBDA: [[ADD:%.+]] = add nsw i32 [[VAL]], 5 224 // LAMBDA: store i32 [[ADD]], i32* [[G_PRIVATE_ADDR]], 225 // LAMBDA: [[G_PRIVATE_ADDR_REF:%.+]] = getelementptr inbounds %{{.+}}, %{{.+}}* [[ARG:%.+]], i{{[0-9]+}} 0, i{{[0-9]+}} 0 226 // LAMBDA: store i{{[0-9]+}}* [[G_PRIVATE_ADDR]], i{{[0-9]+}}** [[G_PRIVATE_ADDR_REF]] 227 // LAMBDA: call void [[INNER_LAMBDA:@.+]](%{{.+}}* [[ARG]]) 228 // LAMBDA: call void @__kmpc_for_static_fini(%{{.+}}* @{{.+}}, i32 [[GTID]]) 229 g += 5; 230 g1 += 5; 231 // LAMBDA: call void @__kmpc_barrier(%{{.+}}* @{{.+}}, i{{[0-9]+}} [[GTID]]) 232 [&]() { 233 // LAMBDA: define {{.+}} void [[INNER_LAMBDA]](%{{.+}}* [[ARG_PTR:%.+]]) 234 // LAMBDA: store %{{.+}}* [[ARG_PTR]], %{{.+}}** [[ARG_PTR_REF:%.+]], 235 g = 2; 236 g1 = 2; 237 // LAMBDA: [[ARG_PTR:%.+]] = load %{{.+}}*, %{{.+}}** [[ARG_PTR_REF]] 238 // LAMBDA: [[G_PTR_REF:%.+]] = getelementptr inbounds %{{.+}}, %{{.+}}* [[ARG_PTR]], i{{[0-9]+}} 0, i{{[0-9]+}} 0 239 // LAMBDA: [[G_REF:%.+]] = load i{{[0-9]+}}*, i{{[0-9]+}}** [[G_PTR_REF]] 240 // LAMBDA: store i{{[0-9]+}} 2, i{{[0-9]+}}* [[G_REF]] 241 }(); 242 } 243 }(); 244 return 0; 245 #elif defined(BLOCKS) 246 // BLOCKS: [[G:@.+]] = global i{{[0-9]+}} 1212, 247 // BLOCKS-LABEL: @main 248 // BLOCKS: call 249 // BLOCKS: call void {{%.+}}(i8 250 ^{ 251 // BLOCKS: define{{.*}} internal{{.*}} void {{.+}}(i8* 252 // BLOCKS: call void {{.+}} @__kmpc_fork_call({{.+}}, i32 0, {{.+}}* [[OMP_REGION:@.+]] to {{.+}}) 253 #pragma omp parallel 254 #pragma omp for linear(g, g1:5) 255 for (int i = 0; i < 2; ++i) { 256 // BLOCKS: define{{.*}} internal{{.*}} void [[OMP_REGION]](i32* noalias %{{.+}}, i32* noalias %{{.+}}) 257 // BLOCKS: alloca i{{[0-9]+}}, 258 // BLOCKS: alloca i{{[0-9]+}}, 259 // BLOCKS: [[G_START_ADDR:%.+]] = alloca i{{[0-9]+}}, 260 // BLOCKS: alloca i{{[0-9]+}}, 261 // BLOCKS: alloca i{{[0-9]+}}, 262 // BLOCKS: alloca i{{[0-9]+}}, 263 // BLOCKS: alloca i{{[0-9]+}}, 264 // BLOCKS: alloca i{{[0-9]+}}, 265 // BLOCKS: alloca i{{[0-9]+}}, 266 // BLOCKS: [[G_PRIVATE_ADDR:%.+]] = alloca i{{[0-9]+}}, 267 // BLOCKS: [[GTID_REF:%.+]] = load i{{[0-9]+}}*, i{{[0-9]+}}** %{{.+}} 268 // BLOCKS: [[GTID:%.+]] = load i{{[0-9]+}}, i{{[0-9]+}}* [[GTID_REF]] 269 // BLOCKS: call {{.+}} @__kmpc_for_static_init_4(%{{.+}}* @{{.+}}, i32 [[GTID]], i32 34, i32* [[IS_LAST_ADDR:%.+]], i32* %{{.+}}, i32* %{{.+}}, i32* %{{.+}}, i32 1, i32 1) 270 // BLOCKS: [[VAL:%.+]] = load i32, i32* [[G_START_ADDR]] 271 // BLOCKS: [[CNT:%.+]] = load i32, i32* 272 // BLOCKS: [[MUL:%.+]] = mul nsw i32 [[CNT]], 5 273 // BLOCKS: [[ADD:%.+]] = add nsw i32 [[VAL]], [[MUL]] 274 // BLOCKS: store i32 [[ADD]], i32* [[G_PRIVATE_ADDR]], 275 // BLOCKS: [[VAL:%.+]] = load i32, i32* [[G_PRIVATE_ADDR]], 276 // BLOCKS: [[ADD:%.+]] = add nsw i32 [[VAL]], 5 277 // BLOCKS: store i32 [[ADD]], i32* [[G_PRIVATE_ADDR]], 278 // BLOCKS-NOT: [[G]]{{[[^:word:]]}} 279 // BLOCKS: i{{[0-9]+}}* [[G_PRIVATE_ADDR]] 280 // BLOCKS-NOT: [[G]]{{[[^:word:]]}} 281 // BLOCKS: call void {{%.+}}(i8 282 // BLOCKS: call void @__kmpc_for_static_fini(%{{.+}}* @{{.+}}, i32 [[GTID]]) 283 g += 5; 284 g1 += 5; 285 // BLOCKS: call void @__kmpc_barrier(%{{.+}}* @{{.+}}, i{{[0-9]+}} [[GTID]]) 286 g = 1; 287 g1 = 5; 288 ^{ 289 // BLOCKS: define {{.+}} void {{@.+}}(i8* 290 g = 2; 291 g1 = 2; 292 // BLOCKS-NOT: [[G]]{{[[^:word:]]}} 293 // BLOCKS: store i{{[0-9]+}} 2, i{{[0-9]+}}* 294 // BLOCKS-NOT: [[G]]{{[[^:word:]]}} 295 // BLOCKS: ret 296 }(); 297 } 298 }(); 299 return 0; 300 // BLOCKS: define {{.+}} @{{.+}}([[SS_TY]]* 301 // BLOCKS: getelementptr inbounds [[SS_TY]], [[SS_TY]]* %{{.+}}, i32 0, i32 0 302 // BLOCKS: store i{{[0-9]+}} 0, i{{[0-9]+}}* % 303 // BLOCKS: getelementptr inbounds [[SS_TY]], [[SS_TY]]* %{{.+}}, i32 0, i32 1 304 // BLOCKS: store i8 305 // BLOCKS: getelementptr inbounds [[SS_TY]], [[SS_TY]]* %{{.+}}, i32 0, i32 2 306 // 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 307 // BLOCKS: ret 308 309 // BLOCKS: define internal void [[SS_MICROTASK]](i{{[0-9]+}}* noalias [[GTID_ADDR:%.+]], i{{[0-9]+}}* noalias %{{.+}}, [[SS_TY]]* %{{.+}}) 310 // BLOCKS: getelementptr {{.*}}[[SS_TY]], [[SS_TY]]* %{{.*}}, i32 0, i32 0 311 // BLOCKS-NOT: getelementptr {{.*}}[[SS_TY]], [[SS_TY]]* %{{.*}}, i32 0, i32 1 312 // BLOCKS: getelementptr {{.*}}[[SS_TY]], [[SS_TY]]* %{{.*}}, i32 0, i32 2 313 // BLOCKS: call void @__kmpc_for_static_init_4( 314 // BLOCKS-NOT: getelementptr {{.*}}[[SS_TY]], [[SS_TY]]* 315 // BLOCKS: call{{.*}} void 316 // BLOCKS: call void @__kmpc_for_static_fini( 317 // BLOCKS: br i1 318 // BLOCKS: [[B_REF:%.+]] = getelementptr {{.*}}[[SS_TY]], [[SS_TY]]* %{{.*}}, i32 0, i32 1 319 // BLOCKS: store i8 %{{.+}}, i8* [[B_REF]], 320 // BLOCKS: br label 321 // BLOCKS: ret void 322 323 // BLOCKS: define internal void @{{.+}}(i{{[0-9]+}}* noalias [[GTID_ADDR:%.+]], i{{[0-9]+}}* noalias %{{.+}}, [[SS_TY]]* %{{.+}}, i32* {{.+}}, i32* {{.+}}, i32* {{.+}}) 324 // BLOCKS: alloca i{{[0-9]+}}, 325 // BLOCKS: alloca i{{[0-9]+}}, 326 // BLOCKS: alloca i{{[0-9]+}}, 327 // BLOCKS: alloca i{{[0-9]+}}, 328 // BLOCKS: alloca i{{[0-9]+}}, 329 // BLOCKS: alloca i{{[0-9]+}}, 330 // BLOCKS: alloca i{{[0-9]+}}, 331 // BLOCKS: alloca i{{[0-9]+}}, 332 // BLOCKS: alloca i{{[0-9]+}}, 333 // BLOCKS: alloca i{{[0-9]+}}, 334 // BLOCKS: [[A_PRIV:%.+]] = alloca i{{[0-9]+}}, 335 // BLOCKS: [[B_PRIV:%.+]] = alloca i{{[0-9]+}}, 336 // BLOCKS: [[C_PRIV:%.+]] = alloca i{{[0-9]+}}, 337 // BLOCKS: store i{{[0-9]+}}* [[A_PRIV]], i{{[0-9]+}}** [[REFA:%.+]], 338 // BLOCKS: store i{{[0-9]+}}* [[C_PRIV]], i{{[0-9]+}}** [[REFC:%.+]], 339 // BLOCKS: call void @__kmpc_for_static_init_4( 340 // BLOCKS: [[A_PRIV:%.+]] = load i{{[0-9]+}}*, i{{[0-9]+}}** [[REFA]], 341 // BLOCKS-NEXT: [[A_VAL:%.+]] = load i{{[0-9]+}}, i{{[0-9]+}}* [[A_PRIV]], 342 // BLOCKS-NEXT: [[INC:%.+]] = add nsw i{{[0-9]+}} [[A_VAL]], 1 343 // BLOCKS-NEXT: store i{{[0-9]+}} [[INC]], i{{[0-9]+}}* [[A_PRIV]], 344 // BLOCKS-NEXT: [[B_VAL:%.+]] = load i{{[0-9]+}}, i{{[0-9]+}}* [[B_PRIV]], 345 // BLOCKS-NEXT: [[DEC:%.+]] = add nsw i{{[0-9]+}} [[B_VAL]], -1 346 // BLOCKS-NEXT: store i{{[0-9]+}} [[DEC]], i{{[0-9]+}}* [[B_PRIV]], 347 // BLOCKS-NEXT: [[C_PRIV:%.+]] = load i{{[0-9]+}}*, i{{[0-9]+}}** [[REFC]], 348 // BLOCKS-NEXT: [[C_VAL:%.+]] = load i{{[0-9]+}}, i{{[0-9]+}}* [[C_PRIV]], 349 // BLOCKS-NEXT: [[DIV:%.+]] = sdiv i{{[0-9]+}} [[C_VAL]], 1 350 // BLOCKS-NEXT: store i{{[0-9]+}} [[DIV]], i{{[0-9]+}}* [[C_PRIV]], 351 // BLOCKS: call void @__kmpc_for_static_fini( 352 // BLOCKS: br i1 353 // BLOCKS: br label 354 // BLOCKS: ret void 355 #else 356 S<float> test; 357 float *pvar = &test.f; 358 long long lvar = 0; 359 #pragma omp parallel 360 #pragma omp for linear(pvar, lvar : 3) allocate(omp_low_lat_mem_alloc: lvar) 361 for (int i = 0; i < 2; ++i) { 362 pvar += 3, lvar += 3; 363 } 364 return tmain<int>(); 365 #endif 366 } 367 368 // CHECK: define i{{[0-9]+}} @main() 369 // CHECK: [[TEST:%.+]] = alloca [[S_FLOAT_TY]], 370 // CHECK: call {{.*}} [[S_FLOAT_TY_DEF_CONSTR:@.+]]([[S_FLOAT_TY]]* [[TEST]]) 371 // 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 372 // CHECK: = call {{.+}} [[TMAIN_INT:@.+]]() 373 // CHECK: call void [[S_FLOAT_TY_DESTR:@.+]]([[S_FLOAT_TY]]* 374 // CHECK: ret 375 376 // CHECK: define internal void [[MAIN_MICROTASK]](i{{[0-9]+}}* noalias [[GTID_ADDR:%.+]], i{{[0-9]+}}* noalias %{{.+}}, float** nonnull align 8 dereferenceable(8) %{{.+}}, i64* nonnull align 8 dereferenceable(8) %{{.+}}) 377 // CHECK: alloca i{{[0-9]+}}, 378 // CHECK: alloca i{{[0-9]+}}, 379 // CHECK: [[PVAR_START:%.+]] = alloca float*, 380 // CHECK: [[LVAR_START:%.+]] = alloca i64, 381 // CHECK: alloca i{{[0-9]+}}, 382 // CHECK: alloca i{{[0-9]+}}, 383 // CHECK: alloca i{{[0-9]+}}, 384 // CHECK: alloca i{{[0-9]+}}, 385 // CHECK: [[PVAR_PRIV:%.+]] = alloca float*, 386 // CHECK: store i{{[0-9]+}}* [[GTID_ADDR]], i{{[0-9]+}}** [[GTID_ADDR_REF:%.+]] 387 388 // Check for default initialization. 389 // CHECK: [[PVAR_REF:%.+]] = load float**, float*** % 390 // CHECK: [[LVAR_REF:%.+]] = load i64*, i64** % 391 // CHECK: [[PVAR_VAL:%.+]] = load float*, float** [[PVAR_REF]], 392 // CHECK: store float* [[PVAR_VAL]], float** [[PVAR_START]], 393 // CHECK: [[LVAR_VAL:%.+]] = load i64, i64* [[LVAR_REF]], 394 // CHECK: store i64 [[LVAR_VAL]], i64* [[LVAR_START]], 395 // CHECK: [[LVAR_VOID_PTR:%.+]] = call i8* @__kmpc_alloc(i32 [[GTID:%.+]], i64 8, i8* inttoptr (i64 5 to i8*)) 396 // CHECK: [[LVAR_PRIV:%.+]] = bitcast i8* [[LVAR_VOID_PTR]] to i64* 397 // CHECK: call {{.+}} @__kmpc_for_static_init_4(%{{.+}}* @{{.+}}, i32 [[GTID]], i32 34, i32* [[IS_LAST_ADDR:%.+]], i32* %{{.+}}, i32* %{{.+}}, i32* %{{.+}}, i32 1, i32 1) 398 // CHECK: [[PVAR_VAL:%.+]] = load float*, float** [[PVAR_START]], 399 // CHECK: [[CNT:%.+]] = load i32, i32* 400 // CHECK: [[MUL:%.+]] = mul nsw i32 [[CNT]], 3 401 // CHECK: [[IDX:%.+]] = sext i32 [[MUL]] to i64 402 // CHECK: [[PTR:%.+]] = getelementptr inbounds float, float* [[PVAR_VAL]], i64 [[IDX]] 403 // CHECK: store float* [[PTR]], float** [[PVAR_PRIV]], 404 // CHECK: [[LVAR_VAL:%.+]] = load i64, i64* [[LVAR_START]], 405 // CHECK: [[CNT:%.+]] = load i32, i32* 406 // CHECK: [[MUL:%.+]] = mul nsw i32 [[CNT]], 3 407 // CHECK: [[CONV:%.+]] = sext i32 [[MUL]] to i64 408 // CHECK: [[VAL:%.+]] = add nsw i64 [[LVAR_VAL]], [[CONV]] 409 // CHECK: store i64 [[VAL]], i64* [[LVAR_PRIV]], 410 // CHECK: [[PVAR_VAL:%.+]] = load float*, float** [[PVAR_PRIV]] 411 // CHECK: [[PTR:%.+]] = getelementptr inbounds float, float* [[PVAR_VAL]], i64 3 412 // CHECK: store float* [[PTR]], float** [[PVAR_PRIV]], 413 // CHECK: [[LVAR_VAL:%.+]] = load i64, i64* [[LVAR_PRIV]], 414 // CHECK: [[ADD:%.+]] = add nsw i64 [[LVAR_VAL]], 3 415 // CHECK: store i64 [[ADD]], i64* [[LVAR_PRIV]], 416 // CHECK: call void @__kmpc_for_static_fini(%{{.+}}* @{{.+}}, i32 %{{.+}}) 417 // CHECK: [[LVAR_VOID_PTR:%.+]] = bitcast i64* [[LVAR_PRIV]] to i8* 418 // CHECK: call void @__kmpc_free(i32 [[GTID]], i8* [[LVAR_VOID_PTR]], i8* inttoptr (i64 5 to i8*)) 419 // CHECK: call void @__kmpc_barrier(%{{.+}}* [[IMPLICIT_BARRIER_LOC]], i{{[0-9]+}} [[GTID]]) 420 // CHECK: ret void 421 422 // CHECK: define {{.*}} i{{[0-9]+}} [[TMAIN_INT]]() 423 // CHECK: [[TEST:%.+]] = alloca [[S_INT_TY]], 424 // CHECK: call {{.*}} [[S_INT_TY_DEF_CONSTR:@.+]]([[S_INT_TY]]* [[TEST]]) 425 // 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 426 // CHECK: call void [[S_INT_TY_DESTR:@.+]]([[S_INT_TY]]* 427 // CHECK: ret 428 429 // CHECK: define {{.+}} @{{.+}}([[SS_TY]]* 430 // CHECK: getelementptr inbounds [[SS_TY]], [[SS_TY]]* %{{.+}}, i32 0, i32 0 431 // CHECK: store i{{[0-9]+}} 0, i{{[0-9]+}}* % 432 // CHECK: getelementptr inbounds [[SS_TY]], [[SS_TY]]* %{{.+}}, i32 0, i32 1 433 // CHECK: store i8 434 // CHECK: getelementptr inbounds [[SS_TY]], [[SS_TY]]* %{{.+}}, i32 0, i32 2 435 // 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 436 // CHECK: ret 437 438 // CHECK: define internal void [[SS_MICROTASK]](i{{[0-9]+}}* noalias [[GTID_ADDR:%.+]], i{{[0-9]+}}* noalias %{{.+}}, [[SS_TY]]* %{{.+}}) 439 // CHECK: alloca i{{[0-9]+}}, 440 // CHECK: alloca i{{[0-9]+}}, 441 // CHECK: alloca i{{[0-9]+}}, 442 // CHECK: alloca i{{[0-9]+}}, 443 // CHECK: alloca i{{[0-9]+}}, 444 // CHECK: alloca i{{[0-9]+}}, 445 // CHECK: alloca i{{[0-9]+}}, 446 // CHECK: alloca i{{[0-9]+}}, 447 // CHECK: alloca i{{[0-9]+}}, 448 // CHECK: alloca i{{[0-9]+}}, 449 // CHECK: alloca i{{[0-9]+}}, 450 // CHECK: [[A_PRIV:%.+]] = alloca i{{[0-9]+}}, 451 // CHECK: [[B_PRIV:%.+]] = alloca i{{[0-9]+}}, 452 // CHECK: [[C_PRIV:%.+]] = alloca i{{[0-9]+}}, 453 // CHECK: call void @__kmpc_barrier( 454 // CHECK: store i{{[0-9]+}}* [[A_PRIV]], i{{[0-9]+}}** [[REFA:%.+]], 455 // CHECK: store i{{[0-9]+}}* [[C_PRIV]], i{{[0-9]+}}** [[REFC:%.+]], 456 // CHECK: call void @__kmpc_for_static_init_4( 457 // CHECK: [[A_PRIV:%.+]] = load i{{[0-9]+}}*, i{{[0-9]+}}** [[REFA]], 458 // CHECK-NEXT: [[A_VAL:%.+]] = load i{{[0-9]+}}, i{{[0-9]+}}* [[A_PRIV]], 459 // CHECK-NEXT: [[INC:%.+]] = add nsw i{{[0-9]+}} [[A_VAL]], 1 460 // CHECK-NEXT: store i{{[0-9]+}} [[INC]], i{{[0-9]+}}* [[A_PRIV]], 461 // CHECK-NEXT: [[B_VAL:%.+]] = load i{{[0-9]+}}, i{{[0-9]+}}* [[B_PRIV]], 462 // CHECK-NEXT: [[DEC:%.+]] = add nsw i{{[0-9]+}} [[B_VAL]], -1 463 // CHECK-NEXT: store i{{[0-9]+}} [[DEC]], i{{[0-9]+}}* [[B_PRIV]], 464 // CHECK-NEXT: [[C_PRIV:%.+]] = load i{{[0-9]+}}*, i{{[0-9]+}}** [[REFC]], 465 // CHECK-NEXT: [[C_VAL:%.+]] = load i{{[0-9]+}}, i{{[0-9]+}}* [[C_PRIV]], 466 // CHECK-NEXT: [[DIV:%.+]] = sdiv i{{[0-9]+}} [[C_VAL]], 1 467 // CHECK-NEXT: store i{{[0-9]+}} [[DIV]], i{{[0-9]+}}* [[C_PRIV]], 468 // CHECK: call void @__kmpc_for_static_fini( 469 // CHECK: br i1 470 // CHECK: [[B_REF:%.+]] = getelementptr {{.*}}[[SS_TY]], [[SS_TY]]* %{{.*}}, i32 0, i32 1 471 // CHECK: store i8 %{{.+}}, i8* [[B_REF]], 472 // CHECK: br label 473 // CHECK: ret void 474 475 // CHECK: define internal void [[TMAIN_MICROTASK]](i{{[0-9]+}}* noalias [[GTID_ADDR:%.+]], i{{[0-9]+}}* noalias %{{.+}}, i32** nonnull align 8 dereferenceable(8) %{{.+}}, i32* nonnull align 4 dereferenceable(4) %{{.+}}) 476 // CHECK: alloca i{{[0-9]+}}, 477 // CHECK: alloca i{{[0-9]+}}, 478 // CHECK: [[PVAR_START:%.+]] = alloca i32*, 479 // CHECK: [[LVAR_START:%.+]] = alloca i32, 480 // CHECK: alloca i{{[0-9]+}}, 481 // CHECK: alloca i{{[0-9]+}}, 482 // CHECK: alloca i{{[0-9]+}}, 483 // CHECK: alloca i{{[0-9]+}}, 484 // CHECK: [[PVAR_PRIV:%.+]] = alloca i32*, 485 // CHECK: [[LVAR_PRIV:%.+]] = alloca i32, 486 // CHECK: [[LVAR_PRIV_REF:%.+]] = alloca i32*, 487 // CHECK: store i{{[0-9]+}}* [[GTID_ADDR]], i{{[0-9]+}}** [[GTID_ADDR_REF:%.+]] 488 489 // Check for default initialization. 490 // CHECK: [[PVAR_REF:%.+]] = load i32**, i32*** % 491 // CHECK: [[PVAR_VAL:%.+]] = load i32*, i32** [[PVAR_REF]], 492 // CHECK: store i32* [[PVAR_VAL]], i32** [[PVAR_START]], 493 // CHECK: [[LVAR_REF:%.+]] = load i32*, i32** % 494 // CHECK: [[LVAR_VAL:%.+]] = load i32, i32* [[LVAR_REF]], 495 // CHECK: store i32 [[LVAR_VAL]], i32* [[LVAR_START]], 496 // CHECK: store i32* [[LVAR_PRIV]], i32** [[LVAR_PRIV_REF]], 497 498 // CHECK: call {{.+}} @__kmpc_for_static_init_4(%{{.+}}* @{{.+}}, i32 [[GTID:%.+]], i32 34, i32* [[IS_LAST_ADDR:%.+]], i32* %{{.+}}, i32* %{{.+}}, i32* %{{.+}}, i32 1, i32 1) 499 // CHECK: [[PVAR_VAL:%.+]] = load i32*, i32** [[PVAR_START]], 500 // CHECK: [[CNT:%.+]] = load i32, i32* 501 // CHECK: [[MUL:%.+]] = mul nsw i32 [[CNT]], 1 502 // CHECK: [[IDX:%.+]] = sext i32 [[MUL]] to i64 503 // CHECK: [[PTR:%.+]] = getelementptr inbounds i32, i32* [[PVAR_VAL]], i64 [[IDX]] 504 // CHECK: store i32* [[PTR]], i32** [[PVAR_PRIV]], 505 // CHECK: [[LVAR_VAL:%.+]] = load i32, i32* [[LVAR_START]], 506 // CHECK: [[CNT:%.+]] = load i32, i32* 507 // CHECK: [[MUL:%.+]] = mul nsw i32 [[CNT]], 1 508 // CHECK: [[VAL:%.+]] = add nsw i32 [[LVAR_VAL]], [[MUL]] 509 // CHECK: store i32 [[VAL]], i32* [[LVAR_PRIV]], 510 // CHECK: [[PVAR_VAL:%.+]] = load i32*, i32** [[PVAR_PRIV]] 511 // CHECK: [[PTR:%.+]] = getelementptr inbounds i32, i32* [[PVAR_VAL]], i32 1 512 // CHECK: store i32* [[PTR]], i32** [[PVAR_PRIV]], 513 // CHECK: [[LVAR_PRIV:%.+]] = load i32*, i32** [[LVAR_PRIV_REF]], 514 // CHECK: [[LVAR_VAL:%.+]] = load i32, i32* [[LVAR_PRIV]], 515 // CHECK: [[ADD:%.+]] = add nsw i32 [[LVAR_VAL]], 1 516 // CHECK: store i32 [[ADD]], i32* [[LVAR_PRIV]], 517 // CHECK: call void @__kmpc_for_static_fini(%{{.+}}* @{{.+}}, i32 %{{.+}}) 518 // CHECK: call void @__kmpc_barrier(%{{.+}}* [[IMPLICIT_BARRIER_LOC]], i{{[0-9]+}} [[GTID]]) 519 // CHECK: ret void 520 #endif 521 522