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 struct SS { 31 int a; 32 int b : 4; 33 int &c; 34 SS(int &d) : a(0), b(0), c(d) { 35 #pragma omp parallel 36 #pragma omp for lastprivate(a, b, c) 37 for (int i = 0; i < 2; ++i) 38 #ifdef LAMBDA 39 [&]() { 40 ++this->a, --b, (this)->c /= 1; 41 #pragma omp parallel 42 #pragma omp for lastprivate(a, b, c) 43 for (int i = 0; i < 2; ++i) 44 ++(this)->a, --b, this->c /= 1; 45 }(); 46 #elif defined(BLOCKS) 47 ^{ 48 ++a; 49 --this->b; 50 (this)->c /= 1; 51 #pragma omp parallel 52 #pragma omp for lastprivate(a, b, c) 53 for (int i = 0; i < 2; ++i) 54 ++(this)->a, --b, this->c /= 1; 55 }(); 56 #else 57 ++this->a, --b, c /= 1; 58 #endif 59 #pragma omp for 60 for (a = 0; a < 2; ++a) 61 #ifdef LAMBDA 62 [&]() { 63 --this->a, ++b, (this)->c *= 2; 64 #pragma omp parallel 65 #pragma omp for lastprivate(b) 66 for (b = 0; b < 2; ++b) 67 ++(this)->a, --b, this->c /= 1; 68 }(); 69 #elif defined(BLOCKS) 70 ^{ 71 ++a; 72 --this->b; 73 (this)->c /= 1; 74 #pragma omp parallel 75 #pragma omp for 76 for (c = 0; c < 2; ++c) 77 ++(this)->a, --b, this->c /= 1; 78 }(); 79 #else 80 ++this->a, --b, c /= 1; 81 #endif 82 } 83 }; 84 85 template <typename T> 86 struct SST { 87 T a; 88 SST() : a(T()) { 89 #pragma omp parallel 90 #pragma omp for lastprivate(a) 91 for (int i = 0; i < 2; ++i) 92 #ifdef LAMBDA 93 [&]() { 94 [&]() { 95 ++this->a; 96 #pragma omp parallel 97 #pragma omp for lastprivate(a) 98 for (int i = 0; i < 2; ++i) 99 ++(this)->a; 100 }(); 101 }(); 102 #elif defined(BLOCKS) 103 ^{ 104 ^{ 105 ++a; 106 #pragma omp parallel 107 #pragma omp for lastprivate(a) 108 for (int i = 0; i < 2; ++i) 109 ++(this)->a; 110 }(); 111 }(); 112 #else 113 ++(this)->a; 114 #endif 115 #pragma omp for 116 for (a = 0; a < 2; ++a) 117 #ifdef LAMBDA 118 [&]() { 119 ++this->a; 120 #pragma omp parallel 121 #pragma omp for 122 for (a = 0; a < 2; ++(this)->a) 123 ++(this)->a; 124 }(); 125 #elif defined(BLOCKS) 126 ^{ 127 ++a; 128 #pragma omp parallel 129 #pragma omp for 130 for (this->a = 0; a < 2; ++a) 131 ++(this)->a; 132 }(); 133 #else 134 ++(this)->a; 135 #endif 136 } 137 }; 138 139 template <class T> 140 struct S { 141 T f; 142 S(T a) : f(a) {} 143 S() : f() {} 144 S<T> &operator=(const S<T> &); 145 operator T() { return T(); } 146 ~S() {} 147 }; 148 149 volatile int g __attribute__((aligned(128)))= 1212; 150 volatile int &g1 = g; 151 float f; 152 char cnt; 153 154 // CHECK: [[SS_TY:%.+]] = type { i{{[0-9]+}}, i8 155 // LAMBDA: [[SS_TY:%.+]] = type { i{{[0-9]+}}, i8 156 // BLOCKS: [[SS_TY:%.+]] = type { i{{[0-9]+}}, i8 157 // CHECK: [[S_FLOAT_TY:%.+]] = type { float } 158 // CHECK: [[S_INT_TY:%.+]] = type { i32 } 159 // CHECK-DAG: [[IMPLICIT_BARRIER_LOC:@.+]] = private unnamed_addr global %{{.+}} { i32 0, i32 66, i32 0, i32 0, i8* 160 // CHECK-DAG: [[X:@.+]] = global double 0.0 161 // CHECK-DAG: [[F:@.+]] = global float 0.0 162 // CHECK-DAG: [[CNT:@.+]] = global i8 0 163 template <typename T> 164 T tmain() { 165 S<T> test; 166 SST<T> sst; 167 T t_var __attribute__((aligned(128))) = T(); 168 T vec[] __attribute__((aligned(128))) = {1, 2}; 169 S<T> s_arr[] __attribute__((aligned(128))) = {1, 2}; 170 S<T> &var __attribute__((aligned(128))) = test; 171 #pragma omp parallel 172 #pragma omp for lastprivate(t_var, vec, s_arr, var) 173 for (int i = 0; i < 2; ++i) { 174 vec[i] = t_var; 175 s_arr[i] = var; 176 } 177 return T(); 178 } 179 180 namespace A { 181 double x; 182 } 183 namespace B { 184 using A::x; 185 } 186 187 int main() { 188 static int sivar; 189 SS ss(sivar); 190 #ifdef LAMBDA 191 // LAMBDA: [[G:@.+]] = global i{{[0-9]+}} 1212, 192 // LAMBDA: [[SIVAR:@.+]] = internal global i{{[0-9]+}} 0, 193 // LAMBDA-LABEL: @main 194 // LAMBDA: alloca [[SS_TY]], 195 // LAMBDA: alloca [[CAP_TY:%.+]], 196 // LAMBDA: call void [[OUTER_LAMBDA:@.+]]([[CAP_TY]]* 197 [&]() { 198 // LAMBDA: define{{.*}} internal{{.*}} void [[OUTER_LAMBDA]]( 199 // LAMBDA: call void {{.+}} @__kmpc_fork_call({{.+}}, i32 1, {{.+}}* [[OMP_REGION:@.+]] to {{.+}}, i32* %{{.+}}) 200 #pragma omp parallel 201 #pragma omp for lastprivate(g, g1, sivar) 202 for (int i = 0; i < 2; ++i) { 203 // LAMBDA: define {{.+}} @{{.+}}([[SS_TY]]* 204 // LAMBDA: getelementptr inbounds [[SS_TY]], [[SS_TY]]* %{{.+}}, i32 0, i32 0 205 // LAMBDA: store i{{[0-9]+}} 0, i{{[0-9]+}}* % 206 // LAMBDA: getelementptr inbounds [[SS_TY]], [[SS_TY]]* %{{.+}}, i32 0, i32 1 207 // LAMBDA: store i8 208 // LAMBDA: getelementptr inbounds [[SS_TY]], [[SS_TY]]* %{{.+}}, i32 0, i32 2 209 // 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 210 // LAMBDA: call void @__kmpc_for_static_init_4( 211 // LAMBDA-NOT: getelementptr inbounds [[SS_TY]], [[SS_TY]]* %{{.+}}, i32 0, i32 0 212 // LAMBDA: call{{.*}} void [[SS_LAMBDA1:@[^ ]+]] 213 // LAMBDA: call void @__kmpc_for_static_fini(% 214 // LAMBDA: ret 215 216 // LAMBDA: define internal void [[SS_MICROTASK]](i{{[0-9]+}}* noalias [[GTID_ADDR:%.+]], i{{[0-9]+}}* noalias %{{.+}}, [[SS_TY]]* %{{.+}}) 217 // LAMBDA: getelementptr {{.*}}[[SS_TY]], [[SS_TY]]* %{{.*}}, i32 0, i32 0 218 // LAMBDA-NOT: getelementptr {{.*}}[[SS_TY]], [[SS_TY]]* %{{.*}}, i32 0, i32 1 219 // LAMBDA: getelementptr {{.*}}[[SS_TY]], [[SS_TY]]* %{{.*}}, i32 0, i32 2 220 // LAMBDA: call void @__kmpc_for_static_init_4( 221 // LAMBDA-NOT: getelementptr {{.*}}[[SS_TY]], [[SS_TY]]* 222 // LAMBDA: call{{.*}} void [[SS_LAMBDA:@[^ ]+]] 223 // LAMBDA: call void @__kmpc_for_static_fini( 224 // LAMBDA: br i1 225 // LAMBDA: [[B_REF:%.+]] = getelementptr {{.*}}[[SS_TY]], [[SS_TY]]* %{{.*}}, i32 0, i32 1 226 // LAMBDA: store i8 %{{.+}}, i8* [[B_REF]], 227 // LAMBDA: br label 228 // LAMBDA: ret void 229 230 // LAMBDA: define internal void @{{.+}}(i{{[0-9]+}}* noalias [[GTID_ADDR:%.+]], i{{[0-9]+}}* noalias %{{.+}}, [[SS_TY]]* %{{.+}}, i32* {{.+}}, i32* {{.+}}, i32* {{.+}}) 231 // LAMBDA: alloca i{{[0-9]+}}, 232 // LAMBDA: alloca i{{[0-9]+}}, 233 // LAMBDA: alloca i{{[0-9]+}}, 234 // LAMBDA: alloca i{{[0-9]+}}, 235 // LAMBDA: alloca i{{[0-9]+}}, 236 // LAMBDA: alloca i{{[0-9]+}}, 237 // LAMBDA: [[A_PRIV:%.+]] = alloca i{{[0-9]+}}, 238 // LAMBDA: [[B_PRIV:%.+]] = alloca i{{[0-9]+}}, 239 // LAMBDA: [[C_PRIV:%.+]] = alloca i{{[0-9]+}}, 240 // LAMBDA: store i{{[0-9]+}}* [[A_PRIV]], i{{[0-9]+}}** [[REFA:%.+]], 241 // LAMBDA: store i{{[0-9]+}}* [[C_PRIV]], i{{[0-9]+}}** [[REFC:%.+]], 242 // LAMBDA: call void @__kmpc_for_static_init_4( 243 // LAMBDA: [[A_PRIV:%.+]] = load i{{[0-9]+}}*, i{{[0-9]+}}** [[REFA]], 244 // LAMBDA-NEXT: [[A_VAL:%.+]] = load i{{[0-9]+}}, i{{[0-9]+}}* [[A_PRIV]], 245 // LAMBDA-NEXT: [[INC:%.+]] = add nsw i{{[0-9]+}} [[A_VAL]], 1 246 // LAMBDA-NEXT: store i{{[0-9]+}} [[INC]], i{{[0-9]+}}* [[A_PRIV]], 247 // LAMBDA-NEXT: [[B_VAL:%.+]] = load i{{[0-9]+}}, i{{[0-9]+}}* [[B_PRIV]], 248 // LAMBDA-NEXT: [[DEC:%.+]] = add nsw i{{[0-9]+}} [[B_VAL]], -1 249 // LAMBDA-NEXT: store i{{[0-9]+}} [[DEC]], i{{[0-9]+}}* [[B_PRIV]], 250 // LAMBDA-NEXT: [[C_PRIV:%.+]] = load i{{[0-9]+}}*, i{{[0-9]+}}** [[REFC]], 251 // LAMBDA-NEXT: [[C_VAL:%.+]] = load i{{[0-9]+}}, i{{[0-9]+}}* [[C_PRIV]], 252 // LAMBDA-NEXT: [[DIV:%.+]] = sdiv i{{[0-9]+}} [[C_VAL]], 1 253 // LAMBDA-NEXT: store i{{[0-9]+}} [[DIV]], i{{[0-9]+}}* [[C_PRIV]], 254 // LAMBDA: call void @__kmpc_for_static_fini( 255 // LAMBDA: br i1 256 // LAMBDA: br label 257 // LAMBDA: ret void 258 259 // LAMBDA: define internal void @{{.+}}(i{{[0-9]+}}* noalias [[GTID_ADDR:%.+]], i{{[0-9]+}}* noalias %{{.+}}, [[SS_TY]]* %{{.+}}) 260 // LAMBDA: ret void 261 262 // LAMBDA: define{{.*}} internal{{.*}} void [[OMP_REGION]](i32* noalias %{{.+}}, i32* noalias %{{.+}}, i32* dereferenceable(4) [[SIVAR:%.+]]) 263 // LAMBDA: alloca i{{[0-9]+}}, 264 // LAMBDA: alloca i{{[0-9]+}}, 265 // LAMBDA: alloca i{{[0-9]+}}, 266 // LAMBDA: alloca i{{[0-9]+}}, 267 // LAMBDA: alloca i{{[0-9]+}}, 268 // LAMBDA: [[G_PRIVATE_ADDR:%.+]] = alloca i{{[0-9]+}}, align 128 269 // LAMBDA: [[G1_PRIVATE_ADDR:%.+]] = alloca i{{[0-9]+}}, 270 // LAMBDA: [[G1_PRIVATE_REF:%.+]] = alloca i{{[0-9]+}}*, 271 // LAMBDA: [[SIVAR_PRIVATE_ADDR:%.+]] = alloca i{{[0-9]+}}, 272 // LAMBDA: [[SIVAR_PRIVATE_ADDR_REF:%.+]] = load i{{[0-9]+}}*, i{{[0-9]+}}** %{{.+}}, 273 274 // LAMBDA: [[GTID_REF:%.+]] = load i{{[0-9]+}}*, i{{[0-9]+}}** %{{.+}} 275 // LAMBDA: [[GTID:%.+]] = load i{{[0-9]+}}, i{{[0-9]+}}* [[GTID_REF]] 276 277 // LAMBDA: call {{.+}} @__kmpc_for_static_init_4(%{{.+}}* @{{.+}}, i32 [[GTID]], i32 34, i32* [[IS_LAST_ADDR:%.+]], i32* %{{.+}}, i32* %{{.+}}, i32* %{{.+}}, i32 1, i32 1) 278 // LAMBDA: store i{{[0-9]+}} 1, i{{[0-9]+}}* [[G_PRIVATE_ADDR]], 279 // LAMBDA: [[G1_PRIVATE_ADDR:%.+]] = load i{{[0-9]+}}*, i{{[0-9]+}}** [[G1_PRIVATE_REF]], 280 // LAMBDA: store volatile i{{[0-9]+}} 1, i{{[0-9]+}}* [[G1_PRIVATE_ADDR]], 281 // LAMBDA: store i{{[0-9]+}} 2, i{{[0-9]+}}* [[SIVAR_PRIVATE_ADDR]], 282 // LAMBDA: [[G_PRIVATE_ADDR_REF:%.+]] = getelementptr inbounds %{{.+}}, %{{.+}}* [[ARG:%.+]], i{{[0-9]+}} 0, i{{[0-9]+}} 0 283 // LAMBDA: store i{{[0-9]+}}* [[G_PRIVATE_ADDR]], i{{[0-9]+}}** [[G_PRIVATE_ADDR_REF]] 284 // LAMBDA: [[G1_PRIVATE_ADDR_REF:%.+]] = getelementptr inbounds %{{.+}}, %{{.+}}* [[ARG:%.+]], i{{[0-9]+}} 0, i{{[0-9]+}} 1 285 // LAMBDA: [[G1_PRIVATE_ADDR:%.+]] = load i{{[0-9]+}}*, i{{[0-9]+}}** [[G1_PRIVATE_REF]], 286 // LAMBDA: store i{{[0-9]+}}* [[G1_PRIVATE_ADDR]], i{{[0-9]+}}** [[G1_PRIVATE_ADDR_REF]] 287 // LAMBDA: [[SIVAR_PRIVATE_ADDR_REF:%.+]] = getelementptr inbounds %{{.+}}, %{{.+}}* [[ARG:%.+]], i{{[0-9]+}} 0, i{{[0-9]+}} 2 288 // LAMBDA: store i{{[0-9]+}}* [[SIVAR_PRIVATE_ADDR]], i{{[0-9]+}}** [[SIVAR_PRIVATE_ADDR_REF]] 289 // LAMBDA: call void [[INNER_LAMBDA:@.+]](%{{.+}}* [[ARG]]) 290 // LAMBDA: call void @__kmpc_for_static_fini(%{{.+}}* @{{.+}}, i32 [[GTID]]) 291 g = 1; 292 g1 = 1; 293 sivar = 2; 294 // Check for final copying of private values back to original vars. 295 // LAMBDA: [[IS_LAST_VAL:%.+]] = load i32, i32* [[IS_LAST_ADDR]], 296 // LAMBDA: [[IS_LAST_ITER:%.+]] = icmp ne i32 [[IS_LAST_VAL]], 0 297 // LAMBDA: br i1 [[IS_LAST_ITER:%.+]], label %[[LAST_THEN:.+]], label %[[LAST_DONE:.+]] 298 // LAMBDA: [[LAST_THEN]] 299 // Actual copying. 300 301 // original g=private_g; 302 // LAMBDA: [[G_VAL:%.+]] = load i{{[0-9]+}}, i{{[0-9]+}}* [[G_PRIVATE_ADDR]], 303 // LAMBDA: store volatile i{{[0-9]+}} [[G_VAL]], i{{[0-9]+}}* [[G]], 304 305 // original sivar=private_sivar; 306 // LAMBDA: [[SIVAR_VAL:%.+]] = load i{{[0-9]+}}, i{{[0-9]+}}* [[SIVAR_PRIVATE_ADDR]], 307 // LAMBDA: store i{{[0-9]+}} [[SIVAR_VAL]], i{{[0-9]+}}* %{{.+}}, 308 // LAMBDA: br label %[[LAST_DONE]] 309 // LAMBDA: [[LAST_DONE]] 310 // LAMBDA: call void @__kmpc_barrier(%{{.+}}* @{{.+}}, i{{[0-9]+}} [[GTID]]) 311 [&]() { 312 // LAMBDA: define {{.+}} void [[INNER_LAMBDA]](%{{.+}}* [[ARG_PTR:%.+]]) 313 // LAMBDA: store %{{.+}}* [[ARG_PTR]], %{{.+}}** [[ARG_PTR_REF:%.+]], 314 g = 2; 315 g1 = 2; 316 sivar = 4; 317 // LAMBDA: [[ARG_PTR:%.+]] = load %{{.+}}*, %{{.+}}** [[ARG_PTR_REF]] 318 // LAMBDA: [[G_PTR_REF:%.+]] = getelementptr inbounds %{{.+}}, %{{.+}}* [[ARG_PTR]], i{{[0-9]+}} 0, i{{[0-9]+}} 0 319 // LAMBDA: [[G_REF:%.+]] = load i{{[0-9]+}}*, i{{[0-9]+}}** [[G_PTR_REF]] 320 // LAMBDA: store i{{[0-9]+}} 2, i{{[0-9]+}}* [[G_REF]] 321 // LAMBDA: [[G1_PTR_REF:%.+]] = getelementptr inbounds %{{.+}}, %{{.+}}* [[ARG_PTR]], i{{[0-9]+}} 0, i{{[0-9]+}} 1 322 // LAMBDA: [[G1_REF:%.+]] = load i{{[0-9]+}}*, i{{[0-9]+}}** [[G1_PTR_REF]] 323 // LAMBDA: store i{{[0-9]+}} 2, i{{[0-9]+}}* [[G1_REF]] 324 // LAMBDA: [[SIVAR_PTR_REF:%.+]] = getelementptr inbounds %{{.+}}, %{{.+}}* [[ARG_PTR]], i{{[0-9]+}} 0, i{{[0-9]+}} 2 325 // LAMBDA: [[SIVAR_REF:%.+]] = load i{{[0-9]+}}*, i{{[0-9]+}}** [[SIVAR_PTR_REF]] 326 // LAMBDA: store i{{[0-9]+}} 4, i{{[0-9]+}}* [[SIVAR_REF]] 327 }(); 328 } 329 }(); 330 return 0; 331 #elif defined(BLOCKS) 332 // BLOCKS: [[G:@.+]] = global i{{[0-9]+}} 1212, 333 // BLOCKS-LABEL: @main 334 // BLOCKS: call 335 // BLOCKS: call void {{%.+}}(i8 336 ^{ 337 // BLOCKS: define{{.*}} internal{{.*}} void {{.+}}(i8* 338 // BLOCKS: call void {{.+}} @__kmpc_fork_call({{.+}}, i32 1, {{.+}}* [[OMP_REGION:@.+]] to {{.+}}) 339 #pragma omp parallel 340 #pragma omp for lastprivate(g, g1, sivar) 341 for (int i = 0; i < 2; ++i) { 342 // BLOCKS: define{{.*}} internal{{.*}} void [[OMP_REGION]](i32* noalias %{{.+}}, i32* noalias %{{.+}}, i32* dereferenceable(4) [[SIVAR:%.+]]) 343 // BLOCKS: alloca i{{[0-9]+}}, 344 // BLOCKS: alloca i{{[0-9]+}}, 345 // BLOCKS: alloca i{{[0-9]+}}, 346 // BLOCKS: alloca i{{[0-9]+}}, 347 // BLOCKS: alloca i{{[0-9]+}}, 348 // BLOCKS: [[G_PRIVATE_ADDR:%.+]] = alloca i{{[0-9]+}}, align 128 349 // BLOCKS: [[G1_PRIVATE_ADDR:%.+]] = alloca i{{[0-9]+}}, align 4 350 // BLOCKS: [[SIVAR_PRIVATE_ADDR:%.+]] = alloca i{{[0-9]+}}, 351 // BLOCKS: store i{{[0-9]+}}* [[SIVAR]], i{{[0-9]+}}** [[SIVAR_ADDR:%.+]], 352 // BLOCKS: {{.+}} = load i{{[0-9]+}}*, i{{[0-9]+}}** [[SIVAR_ADDR]] 353 // BLOCKS: [[GTID_REF:%.+]] = load i{{[0-9]+}}*, i{{[0-9]+}}** %{{.+}} 354 // BLOCKS: [[GTID:%.+]] = load i{{[0-9]+}}, i{{[0-9]+}}* [[GTID_REF]] 355 // BLOCKS: call {{.+}} @__kmpc_for_static_init_4(%{{.+}}* @{{.+}}, i32 [[GTID]], i32 34, i32* [[IS_LAST_ADDR:%.+]], i32* %{{.+}}, i32* %{{.+}}, i32* %{{.+}}, i32 1, i32 1) 356 // BLOCKS: store i{{[0-9]+}} 1, i{{[0-9]+}}* [[G_PRIVATE_ADDR]], 357 // BLOCKS-NOT: [[G]]{{[[^:word:]]}} 358 // BLOCKS: i{{[0-9]+}}* [[G_PRIVATE_ADDR]] 359 // BLOCKS-NOT: [[G]]{{[[^:word:]]}} 360 // BLOCKS: call void {{%.+}}(i8 361 // BLOCKS: call void @__kmpc_for_static_fini(%{{.+}}* @{{.+}}, i32 [[GTID]]) 362 g = 1; 363 g1 = 1; 364 sivar = 2; 365 // Check for final copying of private values back to original vars. 366 // BLOCKS: [[IS_LAST_VAL:%.+]] = load i32, i32* [[IS_LAST_ADDR]], 367 // BLOCKS: [[IS_LAST_ITER:%.+]] = icmp ne i32 [[IS_LAST_VAL]], 0 368 // BLOCKS: br i1 [[IS_LAST_ITER:%.+]], label %[[LAST_THEN:.+]], label %[[LAST_DONE:.+]] 369 // BLOCKS: [[LAST_THEN]] 370 // Actual copying. 371 372 // original g=private_g; 373 // BLOCKS: [[G_VAL:%.+]] = load i{{[0-9]+}}, i{{[0-9]+}}* [[G_PRIVATE_ADDR]], 374 // BLOCKS: store volatile i{{[0-9]+}} [[G_VAL]], i{{[0-9]+}}* [[G]], 375 // BLOCKS: [[SIVAR_VAL:%.+]] = load i{{[0-9]+}}, i{{[0-9]+}}* [[SIVAR_PRIVATE_ADDR]], 376 // BLOCKS: store i{{[0-9]+}} [[SIVAR_VAL]], i{{[0-9]+}}* %{{.+}}, 377 // BLOCKS: br label %[[LAST_DONE]] 378 // BLOCKS: [[LAST_DONE]] 379 // BLOCKS: call void @__kmpc_barrier(%{{.+}}* @{{.+}}, i{{[0-9]+}} [[GTID]]) 380 g = 1; 381 g1 = 1; 382 ^{ 383 // BLOCKS: define {{.+}} void {{@.+}}(i8* 384 g = 2; 385 g1 = 1; 386 sivar = 4; 387 // BLOCKS-NOT: [[G]]{{[[^:word:]]}} 388 // BLOCKS: store i{{[0-9]+}} 2, i{{[0-9]+}}* 389 // BLOCKS-NOT: [[G]]{{[[^:word:]]}} 390 // BLOCKS-NOT: [[SIVAR]]{{[[^:word:]]}} 391 // BLOCKS: store i{{[0-9]+}} 4, i{{[0-9]+}}* 392 // BLOCKS-NOT: [[SIVAR]]{{[[^:word:]]}} 393 // BLOCKS: ret 394 }(); 395 } 396 }(); 397 return 0; 398 // BLOCKS: define {{.+}} @{{.+}}([[SS_TY]]* 399 // BLOCKS: getelementptr inbounds [[SS_TY]], [[SS_TY]]* %{{.+}}, i32 0, i32 0 400 // BLOCKS: store i{{[0-9]+}} 0, i{{[0-9]+}}* % 401 // BLOCKS: getelementptr inbounds [[SS_TY]], [[SS_TY]]* %{{.+}}, i32 0, i32 1 402 // BLOCKS: store i8 403 // BLOCKS: getelementptr inbounds [[SS_TY]], [[SS_TY]]* %{{.+}}, i32 0, i32 2 404 // 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 405 // BLOCKS: call void @__kmpc_for_static_init_4( 406 // BLOCKS-NOT: getelementptr inbounds [[SS_TY]], [[SS_TY]]* %{{.+}}, i32 0, i32 0 407 // BLOCKS: call void 408 // BLOCKS: call void @__kmpc_for_static_fini(% 409 // BLOCKS: ret 410 411 // BLOCKS: define internal void [[SS_MICROTASK]](i{{[0-9]+}}* noalias [[GTID_ADDR:%.+]], i{{[0-9]+}}* noalias %{{.+}}, [[SS_TY]]* %{{.+}}) 412 // BLOCKS: getelementptr {{.*}}[[SS_TY]], [[SS_TY]]* %{{.*}}, i32 0, i32 0 413 // BLOCKS-NOT: getelementptr {{.*}}[[SS_TY]], [[SS_TY]]* %{{.*}}, i32 0, i32 1 414 // BLOCKS: getelementptr {{.*}}[[SS_TY]], [[SS_TY]]* %{{.*}}, i32 0, i32 2 415 // BLOCKS: call void @__kmpc_for_static_init_4( 416 // BLOCKS-NOT: getelementptr {{.*}}[[SS_TY]], [[SS_TY]]* 417 // BLOCKS: call{{.*}} void 418 // BLOCKS: call void @__kmpc_for_static_fini( 419 // BLOCKS: br i1 420 // BLOCKS: [[B_REF:%.+]] = getelementptr {{.*}}[[SS_TY]], [[SS_TY]]* %{{.*}}, i32 0, i32 1 421 // BLOCKS: store i8 %{{.+}}, i8* [[B_REF]], 422 // BLOCKS: br label 423 // BLOCKS: ret void 424 425 // BLOCKS: define internal void @{{.+}}(i{{[0-9]+}}* noalias [[GTID_ADDR:%.+]], i{{[0-9]+}}* noalias %{{.+}}, [[SS_TY]]* %{{.+}}, i32* {{.+}}, i32* {{.+}}, i32* {{.+}}) 426 // BLOCKS: alloca i{{[0-9]+}}, 427 // BLOCKS: alloca i{{[0-9]+}}, 428 // BLOCKS: alloca i{{[0-9]+}}, 429 // BLOCKS: alloca i{{[0-9]+}}, 430 // BLOCKS: alloca i{{[0-9]+}}, 431 // BLOCKS: alloca i{{[0-9]+}}, 432 // BLOCKS: [[A_PRIV:%.+]] = alloca i{{[0-9]+}}, 433 // BLOCKS: [[B_PRIV:%.+]] = alloca i{{[0-9]+}}, 434 // BLOCKS: [[C_PRIV:%.+]] = alloca i{{[0-9]+}}, 435 // BLOCKS: store i{{[0-9]+}}* [[A_PRIV]], i{{[0-9]+}}** [[REFA:%.+]], 436 // BLOCKS: store i{{[0-9]+}}* [[C_PRIV]], i{{[0-9]+}}** [[REFC:%.+]], 437 // BLOCKS: call void @__kmpc_for_static_init_4( 438 // BLOCKS: [[A_PRIV:%.+]] = load i{{[0-9]+}}*, i{{[0-9]+}}** [[REFA]], 439 // BLOCKS-NEXT: [[A_VAL:%.+]] = load i{{[0-9]+}}, i{{[0-9]+}}* [[A_PRIV]], 440 // BLOCKS-NEXT: [[INC:%.+]] = add nsw i{{[0-9]+}} [[A_VAL]], 1 441 // BLOCKS-NEXT: store i{{[0-9]+}} [[INC]], i{{[0-9]+}}* [[A_PRIV]], 442 // BLOCKS-NEXT: [[B_VAL:%.+]] = load i{{[0-9]+}}, i{{[0-9]+}}* [[B_PRIV]], 443 // BLOCKS-NEXT: [[DEC:%.+]] = add nsw i{{[0-9]+}} [[B_VAL]], -1 444 // BLOCKS-NEXT: store i{{[0-9]+}} [[DEC]], i{{[0-9]+}}* [[B_PRIV]], 445 // BLOCKS-NEXT: [[C_PRIV:%.+]] = load i{{[0-9]+}}*, i{{[0-9]+}}** [[REFC]], 446 // BLOCKS-NEXT: [[C_VAL:%.+]] = load i{{[0-9]+}}, i{{[0-9]+}}* [[C_PRIV]], 447 // BLOCKS-NEXT: [[DIV:%.+]] = sdiv i{{[0-9]+}} [[C_VAL]], 1 448 // BLOCKS-NEXT: store i{{[0-9]+}} [[DIV]], i{{[0-9]+}}* [[C_PRIV]], 449 // BLOCKS: call void @__kmpc_for_static_fini( 450 // BLOCKS: br i1 451 // BLOCKS: br label 452 // BLOCKS: ret void 453 #else 454 S<float> test; 455 int t_var = 0; 456 int vec[] = {1, 2}; 457 S<float> s_arr[] = {1, 2}; 458 S<float> var(3); 459 #pragma omp parallel 460 #pragma omp for lastprivate(t_var, vec, s_arr, var, sivar) 461 for (int i = 0; i < 2; ++i) { 462 vec[i] = t_var; 463 s_arr[i] = var; 464 sivar += i; 465 } 466 #pragma omp parallel 467 #pragma omp for lastprivate(A::x, B::x) firstprivate(f) lastprivate(f) 468 for (int i = 0; i < 2; ++i) { 469 A::x++; 470 } 471 #pragma omp parallel 472 #pragma omp for allocate(omp_const_mem_alloc: f) firstprivate(f) lastprivate(f) 473 for (int i = 0; i < 2; ++i) { 474 A::x++; 475 } 476 #pragma omp parallel 477 #pragma omp for allocate(omp_const_mem_alloc :cnt) lastprivate(cnt) 478 for (cnt = 0; cnt < 2; ++cnt) { 479 A::x++; 480 } 481 return tmain<int>(); 482 #endif 483 } 484 485 // CHECK: define i{{[0-9]+}} @main() 486 // CHECK: [[TEST:%.+]] = alloca [[S_FLOAT_TY]], 487 // CHECK: call {{.*}} [[S_FLOAT_TY_DEF_CONSTR:@.+]]([[S_FLOAT_TY]]* [[TEST]]) 488 // CHECK: call void (%{{.+}}*, i{{[0-9]+}}, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)*, ...) @__kmpc_fork_call(%{{.+}}* @{{.+}}, i{{[0-9]+}} 5, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)* bitcast (void (i{{[0-9]+}}*, i{{[0-9]+}}*, i32*, [2 x i32]*, [2 x [[S_FLOAT_TY]]]*, [[S_FLOAT_TY]]*, i32*)* [[MAIN_MICROTASK:@.+]] to void 489 // CHECK: call void (%{{.+}}*, i{{[0-9]+}}, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)*, ...) @__kmpc_fork_call(%{{.+}}* @{{.+}}, i{{[0-9]+}} 0, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)* bitcast (void (i{{[0-9]+}}*, i{{[0-9]+}}*)* [[MAIN_MICROTASK1:@.+]] to void 490 // CHECK: call void (%{{.+}}*, i{{[0-9]+}}, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)*, ...) @__kmpc_fork_call(%{{.+}}* @{{.+}}, i{{[0-9]+}} 0, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)* bitcast (void (i{{[0-9]+}}*, i{{[0-9]+}}*)* [[MAIN_MICROTASK2:@.+]] to void 491 // CHECK: call void (%{{.+}}*, i{{[0-9]+}}, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)*, ...) @__kmpc_fork_call(%{{.+}}* @{{.+}}, i{{[0-9]+}} 0, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)* bitcast (void (i{{[0-9]+}}*, i{{[0-9]+}}*)* [[MAIN_MICROTASK3:@.+]] to void 492 // CHECK: = call {{.+}} [[TMAIN_INT:@.+]]() 493 // CHECK: call void [[S_FLOAT_TY_DESTR:@.+]]([[S_FLOAT_TY]]* 494 // CHECK: ret 495 496 // CHECK: define internal void [[MAIN_MICROTASK]](i32* noalias [[GTID_ADDR:%.+]], i32* noalias %{{.+}}, i32* dereferenceable(4) %{{.+}}, [2 x i32]* dereferenceable(8) %{{.+}}, [2 x [[S_FLOAT_TY]]]* dereferenceable(8) %{{.+}}, [[S_FLOAT_TY]]* dereferenceable(4) %{{.+}}) 497 // CHECK: alloca i{{[0-9]+}}, 498 // CHECK: alloca i{{[0-9]+}}, 499 // CHECK: alloca i{{[0-9]+}}, 500 // CHECK: alloca i{{[0-9]+}}, 501 // CHECK: alloca i{{[0-9]+}}, 502 // CHECK: alloca i{{[0-9]+}}, 503 // CHECK: [[T_VAR_PRIV:%.+]] = alloca i{{[0-9]+}}, 504 // CHECK: [[VEC_PRIV:%.+]] = alloca [2 x i{{[0-9]+}}], 505 // CHECK: [[S_ARR_PRIV:%.+]] = alloca [2 x [[S_FLOAT_TY]]], 506 // CHECK: [[VAR_PRIV:%.+]] = alloca [[S_FLOAT_TY]], 507 // CHECK: [[SIVAR_PRIV:%.+]] = alloca i{{[0-9]+}}, 508 // CHECK: store i{{[0-9]+}}* [[GTID_ADDR]], i{{[0-9]+}}** [[GTID_ADDR_REF:%.+]] 509 510 // CHECK: [[T_VAR_REF:%.+]] = load i{{[0-9]+}}*, i{{[0-9]+}}** % 511 // CHECK: [[VEC_REF:%.+]] = load [2 x i32]*, [2 x i32]** % 512 // CHECK: [[S_ARR_REF:%.+]] = load [2 x [[S_FLOAT_TY]]]*, [2 x [[S_FLOAT_TY]]]** % 513 // CHECK: [[VAR_REF:%.+]] = load [[S_FLOAT_TY]]*, [[S_FLOAT_TY]]** % 514 515 // Check for default initialization. 516 // CHECK-NOT: [[T_VAR_PRIV]] 517 // CHECK-NOT: [[VEC_PRIV]] 518 // CHECK: [[S_ARR_PRIV_ITEM:%.+]] = phi [[S_FLOAT_TY]]* 519 // CHECK: call {{.*}} [[S_FLOAT_TY_DEF_CONSTR]]([[S_FLOAT_TY]]* [[S_ARR_PRIV_ITEM]]) 520 // CHECK: call {{.*}} [[S_FLOAT_TY_DEF_CONSTR]]([[S_FLOAT_TY]]* [[VAR_PRIV]]) 521 // CHECK: call {{.+}} @__kmpc_for_static_init_4(%{{.+}}* @{{.+}}, i32 %{{.+}}, i32 34, i32* [[IS_LAST_ADDR:%.+]], i32* %{{.+}}, i32* %{{.+}}, i32* %{{.+}}, i32 1, i32 1) 522 // <Skip loop body> 523 // CHECK: call void @__kmpc_for_static_fini(%{{.+}}* @{{.+}}, i32 %{{.+}}) 524 525 // Check for final copying of private values back to original vars. 526 // CHECK: [[IS_LAST_VAL:%.+]] = load i32, i32* [[IS_LAST_ADDR]], 527 // CHECK: [[IS_LAST_ITER:%.+]] = icmp ne i32 [[IS_LAST_VAL]], 0 528 // CHECK: br i1 [[IS_LAST_ITER:%.+]], label %[[LAST_THEN:.+]], label %[[LAST_DONE:.+]] 529 // CHECK: [[LAST_THEN]] 530 // Actual copying. 531 532 // original t_var=private_t_var; 533 // CHECK: [[T_VAR_VAL:%.+]] = load i{{[0-9]+}}, i{{[0-9]+}}* [[T_VAR_PRIV]], 534 // CHECK: store i{{[0-9]+}} [[T_VAR_VAL]], i{{[0-9]+}}* [[T_VAR_REF]], 535 536 // original vec[]=private_vec[]; 537 // CHECK: [[VEC_DEST:%.+]] = bitcast [2 x i{{[0-9]+}}]* [[VEC_REF]] to i8* 538 // CHECK: [[VEC_SRC:%.+]] = bitcast [2 x i{{[0-9]+}}]* [[VEC_PRIV]] to i8* 539 // CHECK: call void @llvm.memcpy.{{.+}}(i8* align {{[0-9]+}} [[VEC_DEST]], i8* align {{[0-9]+}} [[VEC_SRC]], 540 541 // original s_arr[]=private_s_arr[]; 542 // CHECK: [[S_ARR_BEGIN:%.+]] = getelementptr inbounds [2 x [[S_FLOAT_TY]]], [2 x [[S_FLOAT_TY]]]* [[S_ARR_REF]], i{{[0-9]+}} 0, i{{[0-9]+}} 0 543 // CHECK: [[S_ARR_PRIV_BEGIN:%.+]] = bitcast [2 x [[S_FLOAT_TY]]]* [[S_ARR_PRIV]] to [[S_FLOAT_TY]]* 544 // CHECK: [[S_ARR_END:%.+]] = getelementptr [[S_FLOAT_TY]], [[S_FLOAT_TY]]* [[S_ARR_BEGIN]], i{{[0-9]+}} 2 545 // CHECK: [[IS_EMPTY:%.+]] = icmp eq [[S_FLOAT_TY]]* [[S_ARR_BEGIN]], [[S_ARR_END]] 546 // CHECK: br i1 [[IS_EMPTY]], label %[[S_ARR_BODY_DONE:.+]], label %[[S_ARR_BODY:.+]] 547 // CHECK: [[S_ARR_BODY]] 548 // CHECK: call {{.*}} [[S_FLOAT_TY_COPY_ASSIGN:@.+]]([[S_FLOAT_TY]]* {{.+}}, [[S_FLOAT_TY]]* {{.+}}) 549 // CHECK: br i1 {{.+}}, label %[[S_ARR_BODY_DONE]], label %[[S_ARR_BODY]] 550 // CHECK: [[S_ARR_BODY_DONE]] 551 552 // original var=private_var; 553 // CHECK: call {{.*}} [[S_FLOAT_TY_COPY_ASSIGN:@.+]]([[S_FLOAT_TY]]* [[VAR_REF]], [[S_FLOAT_TY]]* {{.*}} [[VAR_PRIV]]) 554 // CHECK: [[SIVAR_VAL:%.+]] = load i{{[0-9]+}}, i{{[0-9]+}}* [[SIVAR_PRIV]], 555 // CHECK: br label %[[LAST_DONE]] 556 // CHECK: [[LAST_DONE]] 557 // CHECK-DAG: call void [[S_FLOAT_TY_DESTR]]([[S_FLOAT_TY]]* [[VAR_PRIV]]) 558 // CHECK-DAG: call void [[S_FLOAT_TY_DESTR]]([[S_FLOAT_TY]]* 559 // CHECK: [[GTID_REF:%.+]] = load i{{[0-9]+}}*, i{{[0-9]+}}** [[GTID_ADDR_REF]] 560 // CHECK: [[GTID:%.+]] = load i{{[0-9]+}}, i{{[0-9]+}}* [[GTID_REF]] 561 // CHECK: call void @__kmpc_barrier(%{{.+}}* [[IMPLICIT_BARRIER_LOC]], i{{[0-9]+}} [[GTID]]) 562 // CHECK: ret void 563 564 // 565 // CHECK: define internal void [[MAIN_MICROTASK1]](i{{[0-9]+}}* noalias [[GTID_ADDR:%.+]], i{{[0-9]+}}* noalias %{{.+}}) 566 // CHECK: [[F_PRIV:%.+]] = alloca float, 567 // CHECK-NOT: alloca float 568 // CHECK: [[X_PRIV:%.+]] = alloca double, 569 // CHECK-NOT: alloca float 570 // CHECK-NOT: alloca double 571 572 // Check for default initialization. 573 // CHECK-NOT: [[X_PRIV]] 574 // CHECK: [[F_VAL:%.+]] = load float, float* [[F]], 575 // CHECK: store float [[F_VAL]], float* [[F_PRIV]], 576 // CHECK-NOT: [[X_PRIV]] 577 578 // CHECK: [[GTID_REF:%.+]] = load i{{[0-9]+}}*, i{{[0-9]+}}** [[GTID_ADDR_REF]] 579 // CHECK: [[GTID:%.+]] = load i{{[0-9]+}}, i{{[0-9]+}}* [[GTID_REF]] 580 // CHECK: call {{.+}} @__kmpc_for_static_init_4(%{{.+}}* @{{.+}}, i32 [[GTID]], i32 34, i32* [[IS_LAST_ADDR:%.+]], i32* %{{.+}}, i32* %{{.+}}, i32* %{{.+}}, i32 1, i32 1) 581 // <Skip loop body> 582 // CHECK: call void @__kmpc_for_static_fini(%{{.+}}* @{{.+}}, i32 [[GTID]]) 583 584 // Check for final copying of private values back to original vars. 585 // CHECK: [[IS_LAST_VAL:%.+]] = load i32, i32* [[IS_LAST_ADDR]], 586 // CHECK: [[IS_LAST_ITER:%.+]] = icmp ne i32 [[IS_LAST_VAL]], 0 587 // CHECK: br i1 [[IS_LAST_ITER:%.+]], label %[[LAST_THEN:.+]], label %[[LAST_DONE:.+]] 588 // CHECK: [[LAST_THEN]] 589 // Actual copying. 590 591 // original x=private_x; 592 // CHECK: [[X_VAL:%.+]] = load double, double* [[X_PRIV]], 593 // CHECK: store double [[X_VAL]], double* [[X]], 594 595 // original f=private_f; 596 // CHECK: [[F_VAL:%.+]] = load float, float* [[F_PRIV]], 597 // CHECK: store float [[F_VAL]], float* [[F]], 598 599 // CHECK-NEXT: br label %[[LAST_DONE]] 600 // CHECK: [[LAST_DONE]] 601 602 // CHECK: call void @__kmpc_barrier(%{{.+}}* [[IMPLICIT_BARRIER_LOC]], i{{[0-9]+}} [[GTID]]) 603 // CHECK: ret void 604 605 // CHECK: define internal void [[MAIN_MICROTASK2]](i{{[0-9]+}}* noalias [[GTID_ADDR:%.+]], i{{[0-9]+}}* noalias %{{.+}}) 606 // CHECK-NOT: alloca float 607 608 // Check for default initialization. 609 // CHECK: [[GTID_REF:%.+]] = load i{{[0-9]+}}*, i{{[0-9]+}}** [[GTID_ADDR_REF]] 610 // CHECK: [[GTID:%.+]] = load i{{[0-9]+}}, i{{[0-9]+}}* [[GTID_REF]] 611 // CHECK: [[F_VOID_PTR:%.+]] = call i8* @__kmpc_alloc(i32 [[GTID]], i64 4, i8* inttoptr (i64 3 to i8*)) 612 // CHECK: [[F_PRIV:%.+]] = bitcast i8* [[F_VOID_PTR]] to float* 613 // CHECK: [[F_VAL:%.+]] = load float, float* [[F]], 614 // CHECK: store float [[F_VAL]], float* [[F_PRIV]], 615 616 // CHECK: call {{.+}} @__kmpc_for_static_init_4(%{{.+}}* @{{.+}}, i32 [[GTID]], i32 34, i32* [[IS_LAST_ADDR:%.+]], i32* %{{.+}}, i32* %{{.+}}, i32* %{{.+}}, i32 1, i32 1) 617 // <Skip loop body> 618 // CHECK: call void @__kmpc_for_static_fini(%{{.+}}* @{{.+}}, i32 [[GTID]]) 619 620 // Check for final copying of private values back to original vars. 621 // CHECK: [[IS_LAST_VAL:%.+]] = load i32, i32* [[IS_LAST_ADDR]], 622 // CHECK: [[IS_LAST_ITER:%.+]] = icmp ne i32 [[IS_LAST_VAL]], 0 623 // CHECK: br i1 [[IS_LAST_ITER:%.+]], label %[[LAST_THEN:.+]], label %[[LAST_DONE:.+]] 624 // CHECK: [[LAST_THEN]] 625 // Actual copying. 626 627 // original f=private_f; 628 // CHECK: [[F_VAL:%.+]] = load float, float* [[F_PRIV]], 629 // CHECK: store float [[F_VAL]], float* [[F]], 630 631 // CHECK-NEXT: br label %[[LAST_DONE]] 632 // CHECK: [[LAST_DONE]] 633 634 // CHECK: call void @__kmpc_free(i32 [[GTID]], i8* [[F_VOID_PTR]], i8* inttoptr (i64 3 to i8*)) 635 // CHECK-NEXT: call void @__kmpc_barrier(%{{.+}}* [[IMPLICIT_BARRIER_LOC]], i{{[0-9]+}} [[GTID]]) 636 // CHECK-NEXT: ret void 637 638 // CHECK: define internal void [[MAIN_MICROTASK3]](i{{[0-9]+}}* noalias [[GTID_ADDR:%.+]], i{{[0-9]+}}* noalias %{{.+}}) 639 640 // CHECK: [[GTID_REF:%.+]] = load i{{[0-9]+}}*, i{{[0-9]+}}** [[GTID_ADDR_REF]] 641 // CHECK: [[GTID:%.+]] = load i{{[0-9]+}}, i{{[0-9]+}}* [[GTID_REF]] 642 // CHECK: [[CNT_PRIV:%.+]] = call i8* @__kmpc_alloc(i32 [[GTID]], i64 1, i8* inttoptr (i64 3 to i8*)) 643 // CHECK: call {{.+}} @__kmpc_for_static_init_4(%{{.+}}* @{{.+}}, i32 [[GTID]], i32 34, i32* [[IS_LAST_ADDR:%.+]], i32* [[OMP_LB:%[^,]+]], i32* [[OMP_UB:%[^,]+]], i32* [[OMP_ST:%[^,]+]], i32 1, i32 1) 644 // UB = min(UB, GlobalUB) 645 // CHECK-NEXT: [[UB:%.+]] = load i32, i32* [[OMP_UB]] 646 // CHECK-NEXT: [[UBCMP:%.+]] = icmp sgt i32 [[UB]], 1 647 // CHECK-NEXT: br i1 [[UBCMP]], label [[UB_TRUE:%[^,]+]], label [[UB_FALSE:%[^,]+]] 648 // CHECK: [[UBRESULT:%.+]] = phi i32 [ 1, [[UB_TRUE]] ], [ [[UBVAL:%[^,]+]], [[UB_FALSE]] ] 649 // CHECK-NEXT: store i32 [[UBRESULT]], i32* [[OMP_UB]] 650 // CHECK-NEXT: [[LB:%.+]] = load i32, i32* [[OMP_LB]] 651 // CHECK-NEXT: store i32 [[LB]], i32* [[OMP_IV:[^,]+]] 652 // <Skip loop body> 653 // CHECK: call void @__kmpc_for_static_fini(%{{.+}}* @{{.+}}, i32 [[GTID]]) 654 655 // Check for final copying of private values back to original vars. 656 // CHECK: [[IS_LAST_VAL:%.+]] = load i32, i32* [[IS_LAST_ADDR]], 657 // CHECK: [[IS_LAST_ITER:%.+]] = icmp ne i32 [[IS_LAST_VAL]], 0 658 // CHECK: br i1 [[IS_LAST_ITER:%.+]], label %[[LAST_THEN:.+]], label %[[LAST_DONE:.+]] 659 // CHECK: [[LAST_THEN]] 660 661 // Calculate private cnt value. 662 // CHECK: store i8 2, i8* [[CNT_PRIV]] 663 // original cnt=private_cnt; 664 // CHECK: [[CNT_VAL:%.+]] = load i8, i8* [[CNT_PRIV]], 665 // CHECK: store i8 [[CNT_VAL]], i8* [[CNT]], 666 667 // CHECK-NEXT: br label %[[LAST_DONE]] 668 // CHECK: [[LAST_DONE]] 669 670 // CHECK: call void @__kmpc_free(i32 [[GTID]], i8* [[CNT_PRIV]], i8* inttoptr (i64 3 to i8*)) 671 // CHECK-NEXT: call void @__kmpc_barrier(%{{.+}}* [[IMPLICIT_BARRIER_LOC]], i{{[0-9]+}} [[GTID]]) 672 // CHECK-NEXT: ret void 673 674 // CHECK: define {{.*}} i{{[0-9]+}} [[TMAIN_INT]]() 675 // CHECK: [[TEST:%.+]] = alloca [[S_INT_TY]], 676 // CHECK: call {{.*}} [[S_INT_TY_DEF_CONSTR:@.+]]([[S_INT_TY]]* [[TEST]]) 677 // 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 678 // CHECK: call void [[S_INT_TY_DESTR:@.+]]([[S_INT_TY]]* 679 // CHECK: ret 680 681 // CHECK: define {{.+}} @{{.+}}([[SS_TY]]* 682 // CHECK: getelementptr inbounds [[SS_TY]], [[SS_TY]]* %{{.+}}, i32 0, i32 0 683 // CHECK: store i{{[0-9]+}} 0, i{{[0-9]+}}* % 684 // CHECK: getelementptr inbounds [[SS_TY]], [[SS_TY]]* %{{.+}}, i32 0, i32 1 685 // CHECK: store i8 686 // CHECK: getelementptr inbounds [[SS_TY]], [[SS_TY]]* %{{.+}}, i32 0, i32 2 687 // 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 688 // CHECK: call void @__kmpc_for_static_init_4( 689 // CHECK-NOT: getelementptr inbounds [[SS_TY]], [[SS_TY]]* %{{.+}}, i32 0, i32 0 690 // CHECK: getelementptr inbounds [[SS_TY]], [[SS_TY]]* %{{.+}}, i32 0, i32 1 691 // CHECK: getelementptr inbounds [[SS_TY]], [[SS_TY]]* %{{.+}}, i32 0, i32 2 692 // CHECK: call void @__kmpc_for_static_fini(% 693 // CHECK: ret 694 695 // CHECK: define internal void [[SS_MICROTASK]](i{{[0-9]+}}* noalias [[GTID_ADDR:%.+]], i{{[0-9]+}}* noalias %{{.+}}, [[SS_TY]]* %{{.+}}) 696 // CHECK: alloca i{{[0-9]+}}, 697 // CHECK: alloca i{{[0-9]+}}, 698 // CHECK: alloca i{{[0-9]+}}, 699 // CHECK: alloca i{{[0-9]+}}, 700 // CHECK: alloca i{{[0-9]+}}, 701 // CHECK: alloca i{{[0-9]+}}, 702 // CHECK: alloca i{{[0-9]+}}, 703 // CHECK: [[A_PRIV:%.+]] = alloca i{{[0-9]+}}, 704 // CHECK: [[B_PRIV:%.+]] = alloca i{{[0-9]+}}, 705 // CHECK: [[C_PRIV:%.+]] = alloca i{{[0-9]+}}, 706 // CHECK: store i{{[0-9]+}}* [[A_PRIV]], i{{[0-9]+}}** [[REFA:%.+]], 707 // CHECK: store i{{[0-9]+}}* [[C_PRIV]], i{{[0-9]+}}** [[REFC:%.+]], 708 // CHECK: call void @__kmpc_for_static_init_4( 709 // CHECK: [[A_PRIV:%.+]] = load i{{[0-9]+}}*, i{{[0-9]+}}** [[REFA]], 710 // CHECK-NEXT: [[A_VAL:%.+]] = load i{{[0-9]+}}, i{{[0-9]+}}* [[A_PRIV]], 711 // CHECK-NEXT: [[INC:%.+]] = add nsw i{{[0-9]+}} [[A_VAL]], 1 712 // CHECK-NEXT: store i{{[0-9]+}} [[INC]], i{{[0-9]+}}* [[A_PRIV]], 713 // CHECK-NEXT: [[B_VAL:%.+]] = load i{{[0-9]+}}, i{{[0-9]+}}* [[B_PRIV]], 714 // CHECK-NEXT: [[DEC:%.+]] = add nsw i{{[0-9]+}} [[B_VAL]], -1 715 // CHECK-NEXT: store i{{[0-9]+}} [[DEC]], i{{[0-9]+}}* [[B_PRIV]], 716 // CHECK-NEXT: [[C_PRIV:%.+]] = load i{{[0-9]+}}*, i{{[0-9]+}}** [[REFC]], 717 // CHECK-NEXT: [[C_VAL:%.+]] = load i{{[0-9]+}}, i{{[0-9]+}}* [[C_PRIV]], 718 // CHECK-NEXT: [[DIV:%.+]] = sdiv i{{[0-9]+}} [[C_VAL]], 1 719 // CHECK-NEXT: store i{{[0-9]+}} [[DIV]], i{{[0-9]+}}* [[C_PRIV]], 720 // CHECK: call void @__kmpc_for_static_fini( 721 // CHECK: br i1 722 // CHECK: [[B_REF:%.+]] = getelementptr {{.*}}[[SS_TY]], [[SS_TY]]* %{{.*}}, i32 0, i32 1 723 // CHECK: store i8 %{{.+}}, i8* [[B_REF]], 724 // CHECK: br label 725 // CHECK: ret void 726 727 // 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) %{{.+}}) 728 // CHECK: alloca i{{[0-9]+}}, 729 // CHECK: alloca i{{[0-9]+}}, 730 // CHECK: alloca i{{[0-9]+}}, 731 // CHECK: alloca i{{[0-9]+}}, 732 // CHECK: alloca i{{[0-9]+}}, 733 // CHECK: [[T_VAR_PRIV:%.+]] = alloca i{{[0-9]+}}, align 128 734 // CHECK: [[VEC_PRIV:%.+]] = alloca [2 x i{{[0-9]+}}], align 128 735 // CHECK: [[S_ARR_PRIV:%.+]] = alloca [2 x [[S_INT_TY]]], align 128 736 // CHECK: [[VAR_PRIV:%.+]] = alloca [[S_INT_TY]], align 128 737 // CHECK: [[VAR_PRIV_REF:%.+]] = alloca [[S_INT_TY]]*, 738 // CHECK: store i{{[0-9]+}}* [[GTID_ADDR]], i{{[0-9]+}}** [[GTID_ADDR_REF:%.+]] 739 740 // CHECK: [[T_VAR_REF:%.+]] = load i{{[0-9]+}}*, i{{[0-9]+}}** % 741 // CHECK: [[VEC_REF:%.+]] = load [2 x i{{[0-9]+}}]*, [2 x i{{[0-9]+}}]** % 742 // CHECK: [[S_ARR_REF:%.+]] = load [2 x [[S_INT_TY]]]*, [2 x [[S_INT_TY]]]** % 743 744 // Check for default initialization. 745 // CHECK-NOT: [[T_VAR_PRIV]] 746 // CHECK-NOT: [[VEC_PRIV]] 747 // CHECK: [[S_ARR_PRIV_ITEM:%.+]] = phi [[S_INT_TY]]* 748 // CHECK: call {{.*}} [[S_INT_TY_DEF_CONSTR]]([[S_INT_TY]]* [[S_ARR_PRIV_ITEM]]) 749 // CHECK: [[VAR_REF:%.+]] = load [[S_INT_TY]]*, [[S_INT_TY]]** % 750 // CHECK: call {{.*}} [[S_INT_TY_DEF_CONSTR]]([[S_INT_TY]]* [[VAR_PRIV]]) 751 // CHECK: store [[S_INT_TY]]* [[VAR_PRIV]], [[S_INT_TY]]** [[VAR_PRIV_REF]] 752 // CHECK: call {{.+}} @__kmpc_for_static_init_4(%{{.+}}* @{{.+}}, i32 %{{.+}}, i32 34, i32* [[IS_LAST_ADDR:%.+]], i32* %{{.+}}, i32* %{{.+}}, i32* %{{.+}}, i32 1, i32 1) 753 // <Skip loop body> 754 // CHECK: call void @__kmpc_for_static_fini(%{{.+}}* @{{.+}}, i32 %{{.+}}) 755 756 // Check for final copying of private values back to original vars. 757 // CHECK: [[IS_LAST_VAL:%.+]] = load i32, i32* [[IS_LAST_ADDR]], 758 // CHECK: [[IS_LAST_ITER:%.+]] = icmp ne i32 [[IS_LAST_VAL]], 0 759 // CHECK: br i1 [[IS_LAST_ITER:%.+]], label %[[LAST_THEN:.+]], label %[[LAST_DONE:.+]] 760 // CHECK: [[LAST_THEN]] 761 // Actual copying. 762 763 // original t_var=private_t_var; 764 // CHECK: [[T_VAR_VAL:%.+]] = load i{{[0-9]+}}, i{{[0-9]+}}* [[T_VAR_PRIV]], 765 // CHECK: store i{{[0-9]+}} [[T_VAR_VAL]], i{{[0-9]+}}* [[T_VAR_REF]], 766 767 // original vec[]=private_vec[]; 768 // CHECK: [[VEC_DEST:%.+]] = bitcast [2 x i{{[0-9]+}}]* [[VEC_REF]] to i8* 769 // CHECK: [[VEC_SRC:%.+]] = bitcast [2 x i{{[0-9]+}}]* [[VEC_PRIV]] to i8* 770 // CHECK: call void @llvm.memcpy.{{.+}}(i8* align {{[0-9]+}} [[VEC_DEST]], i8* align {{[0-9]+}} [[VEC_SRC]], 771 772 // original s_arr[]=private_s_arr[]; 773 // CHECK: [[S_ARR_BEGIN:%.+]] = getelementptr inbounds [2 x [[S_INT_TY]]], [2 x [[S_INT_TY]]]* [[S_ARR_REF]], i{{[0-9]+}} 0, i{{[0-9]+}} 0 774 // CHECK: [[S_ARR_PRIV_BEGIN:%.+]] = bitcast [2 x [[S_INT_TY]]]* [[S_ARR_PRIV]] to [[S_INT_TY]]* 775 // CHECK: [[S_ARR_END:%.+]] = getelementptr [[S_INT_TY]], [[S_INT_TY]]* [[S_ARR_BEGIN]], i{{[0-9]+}} 2 776 // CHECK: [[IS_EMPTY:%.+]] = icmp eq [[S_INT_TY]]* [[S_ARR_BEGIN]], [[S_ARR_END]] 777 // CHECK: br i1 [[IS_EMPTY]], label %[[S_ARR_BODY_DONE:.+]], label %[[S_ARR_BODY:.+]] 778 // CHECK: [[S_ARR_BODY]] 779 // CHECK: call {{.*}} [[S_INT_TY_COPY_ASSIGN:@.+]]([[S_INT_TY]]* {{.+}}, [[S_INT_TY]]* {{.+}}) 780 // CHECK: br i1 {{.+}}, label %[[S_ARR_BODY_DONE]], label %[[S_ARR_BODY]] 781 // CHECK: [[S_ARR_BODY_DONE]] 782 783 // original var=private_var; 784 // CHECK: [[VAR_PRIV1:%.+]] = load [[S_INT_TY]]*, [[S_INT_TY]]** [[VAR_PRIV_REF]], 785 // CHECK: call {{.*}} [[S_INT_TY_COPY_ASSIGN:@.+]]([[S_INT_TY]]* [[VAR_REF]], [[S_INT_TY]]* {{.*}} [[VAR_PRIV1]]) 786 // CHECK: br label %[[LAST_DONE]] 787 // CHECK: [[LAST_DONE]] 788 // CHECK-DAG: call void [[S_INT_TY_DESTR]]([[S_INT_TY]]* [[VAR_PRIV]]) 789 // CHECK-DAG: call void [[S_INT_TY_DESTR]]([[S_INT_TY]]* 790 // CHECK: [[GTID_REF:%.+]] = load i{{[0-9]+}}*, i{{[0-9]+}}** [[GTID_ADDR_REF]] 791 // CHECK: [[GTID:%.+]] = load i{{[0-9]+}}, i{{[0-9]+}}* [[GTID_REF]] 792 // CHECK: call void @__kmpc_barrier(%{{.+}}* [[IMPLICIT_BARRIER_LOC]], i{{[0-9]+}} [[GTID]]) 793 // CHECK: ret void 794 #endif 795 796