1 // NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --function-signature --include-generated-funcs --replace-value-regex "__omp_offloading_[0-9a-z]+_[0-9a-z]+" "reduction_size[.].+[.]" "pl_cond[.].+[.|,]" --prefix-filecheck-ir-name _ 2 // RUN: %clang_cc1 -verify -fopenmp -fopenmp-version=50 -x c++ -triple x86_64-apple-darwin10 -emit-llvm %s -o - | FileCheck %s --check-prefix=CHECK1 3 // RUN: %clang_cc1 -fopenmp -fopenmp-version=50 -x c++ -std=c++11 -triple x86_64-apple-darwin10 -emit-pch -o %t %s 4 // RUN: %clang_cc1 -fopenmp -fopenmp-version=50 -x c++ -triple x86_64-apple-darwin10 -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck %s --check-prefix=CHECK2 5 // RUN: %clang_cc1 -verify -fopenmp -fopenmp-version=50 -x c++ -std=c++11 -DLAMBDA -triple x86_64-apple-darwin10 -emit-llvm %s -o - | FileCheck %s --check-prefix=CHECK3 6 // RUN: %clang_cc1 -verify -fopenmp -fopenmp-version=50 -x c++ -fblocks -DBLOCKS -triple x86_64-apple-darwin10 -emit-llvm %s -o - | FileCheck %s --check-prefix=CHECK4 7 8 // RUN: %clang_cc1 -verify -fopenmp-simd -fopenmp-version=50 -x c++ -triple x86_64-apple-darwin10 -emit-llvm %s -o - | FileCheck %s --implicit-check-not="{{__kmpc|__tgt}}" 9 // RUN: %clang_cc1 -fopenmp-simd -fopenmp-version=50 -x c++ -std=c++11 -triple x86_64-apple-darwin10 -emit-pch -o %t %s 10 // RUN: %clang_cc1 -fopenmp-simd -fopenmp-version=50 -x c++ -triple x86_64-apple-darwin10 -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck %s --implicit-check-not="{{__kmpc|__tgt}}" 11 // RUN: %clang_cc1 -verify -fopenmp-simd -fopenmp-version=50 -x c++ -std=c++11 -DLAMBDA -triple x86_64-apple-darwin10 -emit-llvm %s -o - | FileCheck %s --implicit-check-not="{{__kmpc|__tgt}}" 12 // RUN: %clang_cc1 -verify -fopenmp-simd -fopenmp-version=50 -x c++ -fblocks -DBLOCKS -triple x86_64-apple-darwin10 -emit-llvm %s -o - | FileCheck %s --implicit-check-not="{{__kmpc|__tgt}}" 13 // expected-no-diagnostics 14 #ifndef HEADER 15 #define HEADER 16 17 volatile int g __attribute__((aligned(128))) = 1212; 18 19 template <class T> 20 struct S { 21 T f; 22 S(T a) : f(a + g) {} 23 S() : f(g) {} 24 operator T() { return T(); } 25 S &operator&(const S &) { return *this; } 26 ~S() {} 27 }; 28 29 struct SS { 30 int a; 31 int b : 4; 32 int &c; 33 SS(int &d) : a(0), b(0), c(d) { 34 #pragma omp parallel reduction(default, +: a, b, c) 35 #ifdef LAMBDA 36 [&]() { 37 ++this->a, --b, (this)->c /= 1; 38 #pragma omp parallel reduction(&: a, b, c) 39 ++(this)->a, --b, this->c /= 1; 40 }(); 41 #elif defined(BLOCKS) 42 ^{ 43 ++a; 44 --this->b; 45 (this)->c /= 1; 46 #pragma omp parallel reduction(-: a, b, c) 47 ++(this)->a, --b, this->c /= 1; 48 }(); 49 #else 50 ++this->a, --b, c /= 1; 51 #endif 52 } 53 }; 54 55 template<typename T> 56 struct SST { 57 T a; 58 SST() : a(T()) { 59 #pragma omp parallel reduction(*: a) 60 #ifdef LAMBDA 61 [&]() { 62 [&]() { 63 ++this->a; 64 #pragma omp parallel reduction(&& :a) 65 ++(this)->a; 66 }(); 67 }(); 68 #elif defined(BLOCKS) 69 ^{ 70 ^{ 71 ++a; 72 #pragma omp parallel reduction(|: a) 73 ++(this)->a; 74 }(); 75 }(); 76 #else 77 ++(this)->a; 78 #endif 79 } 80 }; 81 82 83 void foo_array_sect(short x[1]) { 84 #pragma omp parallel reduction(default, + : x[:]) 85 {} 86 } 87 88 template <typename T> 89 T tmain() { 90 T t; 91 S<T> test; 92 SST<T> sst; 93 T t_var __attribute__((aligned(128))) = T(), t_var1 __attribute__((aligned(128))); 94 T vec[] = {1, 2}; 95 S<T> s_arr[] = {1, 2}; 96 S<T> var __attribute__((aligned(128))) (3), var1 __attribute__((aligned(128))); 97 #pragma omp parallel reduction(+:t_var) reduction(&:var) reduction(&& : var1) reduction(min: t_var1) 98 { 99 vec[0] = t_var; 100 s_arr[0] = var; 101 } 102 return T(); 103 } 104 105 int sivar; 106 int main() { 107 SS ss(sivar); 108 #ifdef LAMBDA 109 [&]() { 110 #pragma omp parallel reduction(+:g) 111 { 112 113 114 115 116 // Reduction list for runtime. 117 118 g = 1; 119 120 [&]() { 121 g = 2; 122 }(); 123 } 124 }(); 125 return 0; 126 #elif defined(BLOCKS) 127 ^{ 128 #pragma omp parallel reduction(-:g) 129 { 130 131 // Reduction list for runtime. 132 133 g = 1; 134 135 ^{ 136 g = 2; 137 }(); 138 } 139 }(); 140 return 0; 141 142 143 #else 144 S<float> test; 145 float t_var = 0, t_var1; 146 int vec[] = {1, 2}; 147 S<float> s_arr[] = {1, 2}; 148 S<float> var(3), var1; 149 float _Complex cf; 150 #pragma omp parallel reduction(+:t_var) reduction(&:var) reduction(&& : var1) reduction(min: t_var1) 151 { 152 vec[0] = t_var; 153 s_arr[0] = var; 154 } 155 if (var1) 156 #pragma omp parallel reduction(+ : t_var) reduction(& : var) reduction(&& : var1) reduction(min : t_var1) 157 while (1) { 158 vec[0] = t_var; 159 s_arr[0] = var; 160 } 161 #pragma omp parallel reduction(+ : cf) 162 ; 163 return tmain<int>(); 164 #endif 165 } 166 167 168 // Reduction list for runtime. 169 170 171 172 // For + reduction operation initial value of private variable is 0. 173 174 // For & reduction operation initial value of private variable is ones in all bits. 175 176 // For && reduction operation initial value of private variable is 1.0. 177 178 // For min reduction operation initial value of private variable is largest repesentable value. 179 180 // Skip checks for internal operations. 181 182 // void *RedList[<n>] = {<ReductionVars>[0], ..., <ReductionVars>[<n>-1]}; 183 184 185 // res = __kmpc_reduce_nowait(<loc>, <gtid>, <n>, sizeof(RedList), RedList, reduce_func, &<lock>); 186 187 188 // switch(res) 189 190 // case 1: 191 // t_var += t_var_reduction; 192 193 // var = var.operator &(var_reduction); 194 195 // var1 = var1.operator &&(var1_reduction); 196 197 // t_var1 = min(t_var1, t_var1_reduction); 198 199 // __kmpc_end_reduce_nowait(<loc>, <gtid>, &<lock>); 200 201 // break; 202 203 // case 2: 204 // t_var += t_var_reduction; 205 206 // var = var.operator &(var_reduction); 207 208 // var1 = var1.operator &&(var1_reduction); 209 210 // t_var1 = min(t_var1, t_var1_reduction); 211 212 // break; 213 214 215 // void reduce_func(void *lhs[<n>], void *rhs[<n>]) { 216 // *(Type0*)lhs[0] = ReductionOperation0(*(Type0*)lhs[0], *(Type0*)rhs[0]); 217 // ... 218 // *(Type<n>-1*)lhs[<n>-1] = ReductionOperation<n>-1(*(Type<n>-1*)lhs[<n>-1], 219 // *(Type<n>-1*)rhs[<n>-1]); 220 // } 221 // t_var_lhs = (float*)lhs[0]; 222 // t_var_rhs = (float*)rhs[0]; 223 224 // var_lhs = (S<float>*)lhs[1]; 225 // var_rhs = (S<float>*)rhs[1]; 226 227 // var1_lhs = (S<float>*)lhs[2]; 228 // var1_rhs = (S<float>*)rhs[2]; 229 230 // t_var1_lhs = (float*)lhs[3]; 231 // t_var1_rhs = (float*)rhs[3]; 232 233 // t_var_lhs += t_var_rhs; 234 235 // var_lhs = var_lhs.operator &(var_rhs); 236 237 // var1_lhs = var1_lhs.operator &&(var1_rhs); 238 239 // t_var1_lhs = min(t_var1_lhs, t_var1_rhs); 240 241 242 243 244 // For + reduction operation initial value of private variable is 0. 245 246 // For & reduction operation initial value of private variable is ones in all bits. 247 248 // For && reduction operation initial value of private variable is 1.0. 249 250 // For min reduction operation initial value of private variable is largest repesentable value. 251 252 253 254 255 256 257 // Reduction list for runtime. 258 259 260 261 // For + reduction operation initial value of private variable is 0. 262 263 // For & reduction operation initial value of private variable is ones in all bits. 264 265 // For && reduction operation initial value of private variable is 1.0. 266 267 // For min reduction operation initial value of private variable is largest repesentable value. 268 269 // Skip checks for internal operations. 270 271 // void *RedList[<n>] = {<ReductionVars>[0], ..., <ReductionVars>[<n>-1]}; 272 273 274 // res = __kmpc_reduce_nowait(<loc>, <gtid>, <n>, sizeof(RedList), RedList, reduce_func, &<lock>); 275 276 277 // switch(res) 278 279 // case 1: 280 // t_var += t_var_reduction; 281 282 // var = var.operator &(var_reduction); 283 284 // var1 = var1.operator &&(var1_reduction); 285 286 // t_var1 = min(t_var1, t_var1_reduction); 287 288 // __kmpc_end_reduce_nowait(<loc>, <gtid>, &<lock>); 289 290 // break; 291 292 // case 2: 293 // t_var += t_var_reduction; 294 295 // var = var.operator &(var_reduction); 296 297 // var1 = var1.operator &&(var1_reduction); 298 299 // t_var1 = min(t_var1, t_var1_reduction); 300 301 // break; 302 303 304 // void reduce_func(void *lhs[<n>], void *rhs[<n>]) { 305 // *(Type0*)lhs[0] = ReductionOperation0(*(Type0*)lhs[0], *(Type0*)rhs[0]); 306 // ... 307 // *(Type<n>-1*)lhs[<n>-1] = ReductionOperation<n>-1(*(Type<n>-1*)lhs[<n>-1], 308 // *(Type<n>-1*)rhs[<n>-1]); 309 // } 310 // t_var_lhs = (i{{[0-9]+}}*)lhs[0]; 311 // t_var_rhs = (i{{[0-9]+}}*)rhs[0]; 312 313 // var_lhs = (S<i{{[0-9]+}}>*)lhs[1]; 314 // var_rhs = (S<i{{[0-9]+}}>*)rhs[1]; 315 316 // var1_lhs = (S<i{{[0-9]+}}>*)lhs[2]; 317 // var1_rhs = (S<i{{[0-9]+}}>*)rhs[2]; 318 319 // t_var1_lhs = (i{{[0-9]+}}*)lhs[3]; 320 // t_var1_rhs = (i{{[0-9]+}}*)rhs[3]; 321 322 // t_var_lhs += t_var_rhs; 323 324 // var_lhs = var_lhs.operator &(var_rhs); 325 326 // var1_lhs = var1_lhs.operator &&(var1_rhs); 327 328 // t_var1_lhs = min(t_var1_lhs, t_var1_rhs); 329 330 #endif 331 // CHECK1-LABEL: define {{[^@]+}}@_Z14foo_array_sectPs 332 // CHECK1-SAME: (i16* [[X:%.*]]) #[[ATTR0:[0-9]+]] { 333 // CHECK1-NEXT: entry: 334 // CHECK1-NEXT: [[X_ADDR:%.*]] = alloca i16*, align 8 335 // CHECK1-NEXT: store i16* [[X]], i16** [[X_ADDR]], align 8 336 // CHECK1-NEXT: [[TMP0:%.*]] = load i16*, i16** [[X_ADDR]], align 8 337 // CHECK1-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB2:[0-9]+]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i16*)* @.omp_outlined. to void (i32*, i32*, ...)*), i16* [[TMP0]]) 338 // CHECK1-NEXT: ret void 339 // 340 // 341 // CHECK1-LABEL: define {{[^@]+}}@.omp_outlined. 342 // CHECK1-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i16* [[X:%.*]]) #[[ATTR1:[0-9]+]] { 343 // CHECK1-NEXT: entry: 344 // CHECK1-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 345 // CHECK1-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 346 // CHECK1-NEXT: [[X_ADDR:%.*]] = alloca i16*, align 8 347 // CHECK1-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 8 348 // CHECK1-NEXT: [[__VLA_EXPR0:%.*]] = alloca i64, align 8 349 // CHECK1-NEXT: [[TMP:%.*]] = alloca i16*, align 8 350 // CHECK1-NEXT: [[DOTOMP_REDUCTION_RED_LIST:%.*]] = alloca [2 x i8*], align 8 351 // CHECK1-NEXT: [[ATOMIC_TEMP:%.*]] = alloca i16, align 2 352 // CHECK1-NEXT: [[_TMP13:%.*]] = alloca i16, align 2 353 // CHECK1-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8 354 // CHECK1-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8 355 // CHECK1-NEXT: store i16* [[X]], i16** [[X_ADDR]], align 8 356 // CHECK1-NEXT: [[TMP0:%.*]] = load i16*, i16** [[X_ADDR]], align 8 357 // CHECK1-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i16, i16* [[TMP0]], i64 0 358 // CHECK1-NEXT: [[TMP1:%.*]] = load i16*, i16** [[X_ADDR]], align 8 359 // CHECK1-NEXT: [[ARRAYIDX1:%.*]] = getelementptr inbounds i16, i16* [[TMP1]], i64 0 360 // CHECK1-NEXT: [[TMP2:%.*]] = ptrtoint i16* [[ARRAYIDX1]] to i64 361 // CHECK1-NEXT: [[TMP3:%.*]] = ptrtoint i16* [[ARRAYIDX]] to i64 362 // CHECK1-NEXT: [[TMP4:%.*]] = sub i64 [[TMP2]], [[TMP3]] 363 // CHECK1-NEXT: [[TMP5:%.*]] = sdiv exact i64 [[TMP4]], ptrtoint (i16* getelementptr (i16, i16* null, i32 1) to i64) 364 // CHECK1-NEXT: [[TMP6:%.*]] = add nuw i64 [[TMP5]], 1 365 // CHECK1-NEXT: [[TMP7:%.*]] = mul nuw i64 [[TMP6]], ptrtoint (i16* getelementptr (i16, i16* null, i32 1) to i64) 366 // CHECK1-NEXT: [[TMP8:%.*]] = call i8* @llvm.stacksave() 367 // CHECK1-NEXT: store i8* [[TMP8]], i8** [[SAVED_STACK]], align 8 368 // CHECK1-NEXT: [[VLA:%.*]] = alloca i16, i64 [[TMP6]], align 16 369 // CHECK1-NEXT: store i64 [[TMP6]], i64* [[__VLA_EXPR0]], align 8 370 // CHECK1-NEXT: [[TMP9:%.*]] = getelementptr i16, i16* [[VLA]], i64 [[TMP6]] 371 // CHECK1-NEXT: [[OMP_ARRAYINIT_ISEMPTY:%.*]] = icmp eq i16* [[VLA]], [[TMP9]] 372 // CHECK1-NEXT: br i1 [[OMP_ARRAYINIT_ISEMPTY]], label [[OMP_ARRAYINIT_DONE:%.*]], label [[OMP_ARRAYINIT_BODY:%.*]] 373 // CHECK1: omp.arrayinit.body: 374 // CHECK1-NEXT: [[OMP_ARRAYCPY_DESTELEMENTPAST:%.*]] = phi i16* [ [[VLA]], [[ENTRY:%.*]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT:%.*]], [[OMP_ARRAYINIT_BODY]] ] 375 // CHECK1-NEXT: store i16 0, i16* [[OMP_ARRAYCPY_DESTELEMENTPAST]], align 2 376 // CHECK1-NEXT: [[OMP_ARRAYCPY_DEST_ELEMENT]] = getelementptr i16, i16* [[OMP_ARRAYCPY_DESTELEMENTPAST]], i32 1 377 // CHECK1-NEXT: [[OMP_ARRAYCPY_DONE:%.*]] = icmp eq i16* [[OMP_ARRAYCPY_DEST_ELEMENT]], [[TMP9]] 378 // CHECK1-NEXT: br i1 [[OMP_ARRAYCPY_DONE]], label [[OMP_ARRAYINIT_DONE]], label [[OMP_ARRAYINIT_BODY]] 379 // CHECK1: omp.arrayinit.done: 380 // CHECK1-NEXT: [[TMP10:%.*]] = load i16*, i16** [[X_ADDR]], align 8 381 // CHECK1-NEXT: [[TMP11:%.*]] = ptrtoint i16* [[TMP10]] to i64 382 // CHECK1-NEXT: [[TMP12:%.*]] = ptrtoint i16* [[ARRAYIDX]] to i64 383 // CHECK1-NEXT: [[TMP13:%.*]] = sub i64 [[TMP11]], [[TMP12]] 384 // CHECK1-NEXT: [[TMP14:%.*]] = sdiv exact i64 [[TMP13]], ptrtoint (i16* getelementptr (i16, i16* null, i32 1) to i64) 385 // CHECK1-NEXT: [[TMP15:%.*]] = getelementptr i16, i16* [[VLA]], i64 [[TMP14]] 386 // CHECK1-NEXT: store i16* [[TMP15]], i16** [[TMP]], align 8 387 // CHECK1-NEXT: [[TMP16:%.*]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[DOTOMP_REDUCTION_RED_LIST]], i64 0, i64 0 388 // CHECK1-NEXT: [[TMP17:%.*]] = bitcast i16* [[VLA]] to i8* 389 // CHECK1-NEXT: store i8* [[TMP17]], i8** [[TMP16]], align 8 390 // CHECK1-NEXT: [[TMP18:%.*]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[DOTOMP_REDUCTION_RED_LIST]], i64 0, i64 1 391 // CHECK1-NEXT: [[TMP19:%.*]] = inttoptr i64 [[TMP6]] to i8* 392 // CHECK1-NEXT: store i8* [[TMP19]], i8** [[TMP18]], align 8 393 // CHECK1-NEXT: [[TMP20:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 394 // CHECK1-NEXT: [[TMP21:%.*]] = load i32, i32* [[TMP20]], align 4 395 // CHECK1-NEXT: [[TMP22:%.*]] = bitcast [2 x i8*]* [[DOTOMP_REDUCTION_RED_LIST]] to i8* 396 // CHECK1-NEXT: [[TMP23:%.*]] = call i32 @__kmpc_reduce_nowait(%struct.ident_t* @[[GLOB1:[0-9]+]], i32 [[TMP21]], i32 1, i64 16, i8* [[TMP22]], void (i8*, i8*)* @.omp.reduction.reduction_func, [8 x i32]* @.gomp_critical_user_.reduction.var) 397 // CHECK1-NEXT: switch i32 [[TMP23]], label [[DOTOMP_REDUCTION_DEFAULT:%.*]] [ 398 // CHECK1-NEXT: i32 1, label [[DOTOMP_REDUCTION_CASE1:%.*]] 399 // CHECK1-NEXT: i32 2, label [[DOTOMP_REDUCTION_CASE2:%.*]] 400 // CHECK1-NEXT: ] 401 // CHECK1: .omp.reduction.case1: 402 // CHECK1-NEXT: [[TMP24:%.*]] = getelementptr i16, i16* [[ARRAYIDX]], i64 [[TMP6]] 403 // CHECK1-NEXT: [[OMP_ARRAYCPY_ISEMPTY:%.*]] = icmp eq i16* [[ARRAYIDX]], [[TMP24]] 404 // CHECK1-NEXT: br i1 [[OMP_ARRAYCPY_ISEMPTY]], label [[OMP_ARRAYCPY_DONE7:%.*]], label [[OMP_ARRAYCPY_BODY:%.*]] 405 // CHECK1: omp.arraycpy.body: 406 // CHECK1-NEXT: [[OMP_ARRAYCPY_SRCELEMENTPAST:%.*]] = phi i16* [ [[VLA]], [[DOTOMP_REDUCTION_CASE1]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ] 407 // CHECK1-NEXT: [[OMP_ARRAYCPY_DESTELEMENTPAST2:%.*]] = phi i16* [ [[ARRAYIDX]], [[DOTOMP_REDUCTION_CASE1]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT5:%.*]], [[OMP_ARRAYCPY_BODY]] ] 408 // CHECK1-NEXT: [[TMP25:%.*]] = load i16, i16* [[OMP_ARRAYCPY_DESTELEMENTPAST2]], align 2 409 // CHECK1-NEXT: [[CONV:%.*]] = sext i16 [[TMP25]] to i32 410 // CHECK1-NEXT: [[TMP26:%.*]] = load i16, i16* [[OMP_ARRAYCPY_SRCELEMENTPAST]], align 2 411 // CHECK1-NEXT: [[CONV3:%.*]] = sext i16 [[TMP26]] to i32 412 // CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 [[CONV]], [[CONV3]] 413 // CHECK1-NEXT: [[CONV4:%.*]] = trunc i32 [[ADD]] to i16 414 // CHECK1-NEXT: store i16 [[CONV4]], i16* [[OMP_ARRAYCPY_DESTELEMENTPAST2]], align 2 415 // CHECK1-NEXT: [[OMP_ARRAYCPY_DEST_ELEMENT5]] = getelementptr i16, i16* [[OMP_ARRAYCPY_DESTELEMENTPAST2]], i32 1 416 // CHECK1-NEXT: [[OMP_ARRAYCPY_SRC_ELEMENT]] = getelementptr i16, i16* [[OMP_ARRAYCPY_SRCELEMENTPAST]], i32 1 417 // CHECK1-NEXT: [[OMP_ARRAYCPY_DONE6:%.*]] = icmp eq i16* [[OMP_ARRAYCPY_DEST_ELEMENT5]], [[TMP24]] 418 // CHECK1-NEXT: br i1 [[OMP_ARRAYCPY_DONE6]], label [[OMP_ARRAYCPY_DONE7]], label [[OMP_ARRAYCPY_BODY]] 419 // CHECK1: omp.arraycpy.done7: 420 // CHECK1-NEXT: call void @__kmpc_end_reduce_nowait(%struct.ident_t* @[[GLOB1]], i32 [[TMP21]], [8 x i32]* @.gomp_critical_user_.reduction.var) 421 // CHECK1-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]] 422 // CHECK1: .omp.reduction.case2: 423 // CHECK1-NEXT: [[TMP27:%.*]] = getelementptr i16, i16* [[ARRAYIDX]], i64 [[TMP6]] 424 // CHECK1-NEXT: [[OMP_ARRAYCPY_ISEMPTY8:%.*]] = icmp eq i16* [[ARRAYIDX]], [[TMP27]] 425 // CHECK1-NEXT: br i1 [[OMP_ARRAYCPY_ISEMPTY8]], label [[OMP_ARRAYCPY_DONE21:%.*]], label [[OMP_ARRAYCPY_BODY9:%.*]] 426 // CHECK1: omp.arraycpy.body9: 427 // CHECK1-NEXT: [[OMP_ARRAYCPY_SRCELEMENTPAST10:%.*]] = phi i16* [ [[VLA]], [[DOTOMP_REDUCTION_CASE2]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT19:%.*]], [[ATOMIC_EXIT:%.*]] ] 428 // CHECK1-NEXT: [[OMP_ARRAYCPY_DESTELEMENTPAST11:%.*]] = phi i16* [ [[ARRAYIDX]], [[DOTOMP_REDUCTION_CASE2]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT18:%.*]], [[ATOMIC_EXIT]] ] 429 // CHECK1-NEXT: [[TMP28:%.*]] = load i16, i16* [[OMP_ARRAYCPY_SRCELEMENTPAST10]], align 2 430 // CHECK1-NEXT: [[CONV12:%.*]] = sext i16 [[TMP28]] to i32 431 // CHECK1-NEXT: [[ATOMIC_LOAD:%.*]] = load atomic i16, i16* [[OMP_ARRAYCPY_DESTELEMENTPAST11]] monotonic, align 2 432 // CHECK1-NEXT: br label [[ATOMIC_CONT:%.*]] 433 // CHECK1: atomic_cont: 434 // CHECK1-NEXT: [[TMP29:%.*]] = phi i16 [ [[ATOMIC_LOAD]], [[OMP_ARRAYCPY_BODY9]] ], [ [[TMP34:%.*]], [[ATOMIC_CONT]] ] 435 // CHECK1-NEXT: store i16 [[TMP29]], i16* [[_TMP13]], align 2 436 // CHECK1-NEXT: [[TMP30:%.*]] = load i16, i16* [[_TMP13]], align 2 437 // CHECK1-NEXT: [[CONV14:%.*]] = sext i16 [[TMP30]] to i32 438 // CHECK1-NEXT: [[TMP31:%.*]] = load i16, i16* [[OMP_ARRAYCPY_SRCELEMENTPAST10]], align 2 439 // CHECK1-NEXT: [[CONV15:%.*]] = sext i16 [[TMP31]] to i32 440 // CHECK1-NEXT: [[ADD16:%.*]] = add nsw i32 [[CONV14]], [[CONV15]] 441 // CHECK1-NEXT: [[CONV17:%.*]] = trunc i32 [[ADD16]] to i16 442 // CHECK1-NEXT: store i16 [[CONV17]], i16* [[ATOMIC_TEMP]], align 2 443 // CHECK1-NEXT: [[TMP32:%.*]] = load i16, i16* [[ATOMIC_TEMP]], align 2 444 // CHECK1-NEXT: [[TMP33:%.*]] = cmpxchg i16* [[OMP_ARRAYCPY_DESTELEMENTPAST11]], i16 [[TMP29]], i16 [[TMP32]] monotonic monotonic, align 2 445 // CHECK1-NEXT: [[TMP34]] = extractvalue { i16, i1 } [[TMP33]], 0 446 // CHECK1-NEXT: [[TMP35:%.*]] = extractvalue { i16, i1 } [[TMP33]], 1 447 // CHECK1-NEXT: br i1 [[TMP35]], label [[ATOMIC_EXIT]], label [[ATOMIC_CONT]] 448 // CHECK1: atomic_exit: 449 // CHECK1-NEXT: [[OMP_ARRAYCPY_DEST_ELEMENT18]] = getelementptr i16, i16* [[OMP_ARRAYCPY_DESTELEMENTPAST11]], i32 1 450 // CHECK1-NEXT: [[OMP_ARRAYCPY_SRC_ELEMENT19]] = getelementptr i16, i16* [[OMP_ARRAYCPY_SRCELEMENTPAST10]], i32 1 451 // CHECK1-NEXT: [[OMP_ARRAYCPY_DONE20:%.*]] = icmp eq i16* [[OMP_ARRAYCPY_DEST_ELEMENT18]], [[TMP27]] 452 // CHECK1-NEXT: br i1 [[OMP_ARRAYCPY_DONE20]], label [[OMP_ARRAYCPY_DONE21]], label [[OMP_ARRAYCPY_BODY9]] 453 // CHECK1: omp.arraycpy.done21: 454 // CHECK1-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]] 455 // CHECK1: .omp.reduction.default: 456 // CHECK1-NEXT: [[TMP36:%.*]] = load i8*, i8** [[SAVED_STACK]], align 8 457 // CHECK1-NEXT: call void @llvm.stackrestore(i8* [[TMP36]]) 458 // CHECK1-NEXT: ret void 459 // 460 // 461 // CHECK1-LABEL: define {{[^@]+}}@.omp.reduction.reduction_func 462 // CHECK1-SAME: (i8* [[TMP0:%.*]], i8* [[TMP1:%.*]]) #[[ATTR3:[0-9]+]] { 463 // CHECK1-NEXT: entry: 464 // CHECK1-NEXT: [[DOTADDR:%.*]] = alloca i8*, align 8 465 // CHECK1-NEXT: [[DOTADDR1:%.*]] = alloca i8*, align 8 466 // CHECK1-NEXT: store i8* [[TMP0]], i8** [[DOTADDR]], align 8 467 // CHECK1-NEXT: store i8* [[TMP1]], i8** [[DOTADDR1]], align 8 468 // CHECK1-NEXT: [[TMP2:%.*]] = load i8*, i8** [[DOTADDR]], align 8 469 // CHECK1-NEXT: [[TMP3:%.*]] = bitcast i8* [[TMP2]] to [2 x i8*]* 470 // CHECK1-NEXT: [[TMP4:%.*]] = load i8*, i8** [[DOTADDR1]], align 8 471 // CHECK1-NEXT: [[TMP5:%.*]] = bitcast i8* [[TMP4]] to [2 x i8*]* 472 // CHECK1-NEXT: [[TMP6:%.*]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[TMP5]], i64 0, i64 0 473 // CHECK1-NEXT: [[TMP7:%.*]] = load i8*, i8** [[TMP6]], align 8 474 // CHECK1-NEXT: [[TMP8:%.*]] = bitcast i8* [[TMP7]] to i16* 475 // CHECK1-NEXT: [[TMP9:%.*]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[TMP3]], i64 0, i64 0 476 // CHECK1-NEXT: [[TMP10:%.*]] = load i8*, i8** [[TMP9]], align 8 477 // CHECK1-NEXT: [[TMP11:%.*]] = bitcast i8* [[TMP10]] to i16* 478 // CHECK1-NEXT: [[TMP12:%.*]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[TMP3]], i64 0, i64 1 479 // CHECK1-NEXT: [[TMP13:%.*]] = load i8*, i8** [[TMP12]], align 8 480 // CHECK1-NEXT: [[TMP14:%.*]] = ptrtoint i8* [[TMP13]] to i64 481 // CHECK1-NEXT: [[TMP15:%.*]] = getelementptr i16, i16* [[TMP11]], i64 [[TMP14]] 482 // CHECK1-NEXT: [[OMP_ARRAYCPY_ISEMPTY:%.*]] = icmp eq i16* [[TMP11]], [[TMP15]] 483 // CHECK1-NEXT: br i1 [[OMP_ARRAYCPY_ISEMPTY]], label [[OMP_ARRAYCPY_DONE4:%.*]], label [[OMP_ARRAYCPY_BODY:%.*]] 484 // CHECK1: omp.arraycpy.body: 485 // CHECK1-NEXT: [[OMP_ARRAYCPY_SRCELEMENTPAST:%.*]] = phi i16* [ [[TMP8]], [[ENTRY:%.*]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ] 486 // CHECK1-NEXT: [[OMP_ARRAYCPY_DESTELEMENTPAST:%.*]] = phi i16* [ [[TMP11]], [[ENTRY]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ] 487 // CHECK1-NEXT: [[TMP16:%.*]] = load i16, i16* [[OMP_ARRAYCPY_DESTELEMENTPAST]], align 2 488 // CHECK1-NEXT: [[CONV:%.*]] = sext i16 [[TMP16]] to i32 489 // CHECK1-NEXT: [[TMP17:%.*]] = load i16, i16* [[OMP_ARRAYCPY_SRCELEMENTPAST]], align 2 490 // CHECK1-NEXT: [[CONV2:%.*]] = sext i16 [[TMP17]] to i32 491 // CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 [[CONV]], [[CONV2]] 492 // CHECK1-NEXT: [[CONV3:%.*]] = trunc i32 [[ADD]] to i16 493 // CHECK1-NEXT: store i16 [[CONV3]], i16* [[OMP_ARRAYCPY_DESTELEMENTPAST]], align 2 494 // CHECK1-NEXT: [[OMP_ARRAYCPY_DEST_ELEMENT]] = getelementptr i16, i16* [[OMP_ARRAYCPY_DESTELEMENTPAST]], i32 1 495 // CHECK1-NEXT: [[OMP_ARRAYCPY_SRC_ELEMENT]] = getelementptr i16, i16* [[OMP_ARRAYCPY_SRCELEMENTPAST]], i32 1 496 // CHECK1-NEXT: [[OMP_ARRAYCPY_DONE:%.*]] = icmp eq i16* [[OMP_ARRAYCPY_DEST_ELEMENT]], [[TMP15]] 497 // CHECK1-NEXT: br i1 [[OMP_ARRAYCPY_DONE]], label [[OMP_ARRAYCPY_DONE4]], label [[OMP_ARRAYCPY_BODY]] 498 // CHECK1: omp.arraycpy.done4: 499 // CHECK1-NEXT: ret void 500 // 501 // 502 // CHECK1-LABEL: define {{[^@]+}}@main 503 // CHECK1-SAME: () #[[ATTR6:[0-9]+]] { 504 // CHECK1-NEXT: entry: 505 // CHECK1-NEXT: [[RETVAL:%.*]] = alloca i32, align 4 506 // CHECK1-NEXT: [[SS:%.*]] = alloca [[STRUCT_SS:%.*]], align 8 507 // CHECK1-NEXT: [[TEST:%.*]] = alloca [[STRUCT_S:%.*]], align 4 508 // CHECK1-NEXT: [[T_VAR:%.*]] = alloca float, align 4 509 // CHECK1-NEXT: [[T_VAR1:%.*]] = alloca float, align 4 510 // CHECK1-NEXT: [[VEC:%.*]] = alloca [2 x i32], align 4 511 // CHECK1-NEXT: [[S_ARR:%.*]] = alloca [2 x %struct.S], align 4 512 // CHECK1-NEXT: [[VAR:%.*]] = alloca [[STRUCT_S]], align 4 513 // CHECK1-NEXT: [[VAR1:%.*]] = alloca [[STRUCT_S]], align 4 514 // CHECK1-NEXT: [[CF:%.*]] = alloca { float, float }, align 4 515 // CHECK1-NEXT: store i32 0, i32* [[RETVAL]], align 4 516 // CHECK1-NEXT: call void @_ZN2SSC1ERi(%struct.SS* nonnull align 8 dereferenceable(16) [[SS]], i32* nonnull align 4 dereferenceable(4) @sivar) 517 // CHECK1-NEXT: call void @_ZN1SIfEC1Ev(%struct.S* nonnull align 4 dereferenceable(4) [[TEST]]) 518 // CHECK1-NEXT: store float 0.000000e+00, float* [[T_VAR]], align 4 519 // CHECK1-NEXT: [[TMP0:%.*]] = bitcast [2 x i32]* [[VEC]] to i8* 520 // CHECK1-NEXT: call void @llvm.memcpy.p0i8.p0i8.i64(i8* align 4 [[TMP0]], i8* align 4 bitcast ([2 x i32]* @__const.main.vec to i8*), i64 8, i1 false) 521 // CHECK1-NEXT: [[ARRAYINIT_BEGIN:%.*]] = getelementptr inbounds [2 x %struct.S], [2 x %struct.S]* [[S_ARR]], i64 0, i64 0 522 // CHECK1-NEXT: call void @_ZN1SIfEC1Ef(%struct.S* nonnull align 4 dereferenceable(4) [[ARRAYINIT_BEGIN]], float 1.000000e+00) 523 // CHECK1-NEXT: [[ARRAYINIT_ELEMENT:%.*]] = getelementptr inbounds [[STRUCT_S]], %struct.S* [[ARRAYINIT_BEGIN]], i64 1 524 // CHECK1-NEXT: call void @_ZN1SIfEC1Ef(%struct.S* nonnull align 4 dereferenceable(4) [[ARRAYINIT_ELEMENT]], float 2.000000e+00) 525 // CHECK1-NEXT: call void @_ZN1SIfEC1Ef(%struct.S* nonnull align 4 dereferenceable(4) [[VAR]], float 3.000000e+00) 526 // CHECK1-NEXT: call void @_ZN1SIfEC1Ev(%struct.S* nonnull align 4 dereferenceable(4) [[VAR1]]) 527 // CHECK1-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB2]], i32 6, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, [2 x i32]*, float*, [2 x %struct.S]*, %struct.S*, %struct.S*, float*)* @.omp_outlined..1 to void (i32*, i32*, ...)*), [2 x i32]* [[VEC]], float* [[T_VAR]], [2 x %struct.S]* [[S_ARR]], %struct.S* [[VAR]], %struct.S* [[VAR1]], float* [[T_VAR1]]) 528 // CHECK1-NEXT: [[CALL:%.*]] = call float @_ZN1SIfEcvfEv(%struct.S* nonnull align 4 dereferenceable(4) [[VAR1]]) 529 // CHECK1-NEXT: [[TOBOOL:%.*]] = fcmp une float [[CALL]], 0.000000e+00 530 // CHECK1-NEXT: br i1 [[TOBOOL]], label [[IF_THEN:%.*]], label [[IF_END:%.*]] 531 // CHECK1: if.then: 532 // CHECK1-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB2]], i32 6, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, [2 x i32]*, float*, [2 x %struct.S]*, %struct.S*, %struct.S*, float*)* @.omp_outlined..3 to void (i32*, i32*, ...)*), [2 x i32]* [[VEC]], float* [[T_VAR]], [2 x %struct.S]* [[S_ARR]], %struct.S* [[VAR]], %struct.S* [[VAR1]], float* [[T_VAR1]]) 533 // CHECK1-NEXT: br label [[IF_END]] 534 // CHECK1: if.end: 535 // CHECK1-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB2]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, { float, float }*)* @.omp_outlined..4 to void (i32*, i32*, ...)*), { float, float }* [[CF]]) 536 // CHECK1-NEXT: [[CALL1:%.*]] = call i32 @_Z5tmainIiET_v() 537 // CHECK1-NEXT: store i32 [[CALL1]], i32* [[RETVAL]], align 4 538 // CHECK1-NEXT: call void @_ZN1SIfED1Ev(%struct.S* nonnull align 4 dereferenceable(4) [[VAR1]]) #[[ATTR5:[0-9]+]] 539 // CHECK1-NEXT: call void @_ZN1SIfED1Ev(%struct.S* nonnull align 4 dereferenceable(4) [[VAR]]) #[[ATTR5]] 540 // CHECK1-NEXT: [[ARRAY_BEGIN:%.*]] = getelementptr inbounds [2 x %struct.S], [2 x %struct.S]* [[S_ARR]], i32 0, i32 0 541 // CHECK1-NEXT: [[TMP1:%.*]] = getelementptr inbounds [[STRUCT_S]], %struct.S* [[ARRAY_BEGIN]], i64 2 542 // CHECK1-NEXT: br label [[ARRAYDESTROY_BODY:%.*]] 543 // CHECK1: arraydestroy.body: 544 // CHECK1-NEXT: [[ARRAYDESTROY_ELEMENTPAST:%.*]] = phi %struct.S* [ [[TMP1]], [[IF_END]] ], [ [[ARRAYDESTROY_ELEMENT:%.*]], [[ARRAYDESTROY_BODY]] ] 545 // CHECK1-NEXT: [[ARRAYDESTROY_ELEMENT]] = getelementptr inbounds [[STRUCT_S]], %struct.S* [[ARRAYDESTROY_ELEMENTPAST]], i64 -1 546 // CHECK1-NEXT: call void @_ZN1SIfED1Ev(%struct.S* nonnull align 4 dereferenceable(4) [[ARRAYDESTROY_ELEMENT]]) #[[ATTR5]] 547 // CHECK1-NEXT: [[ARRAYDESTROY_DONE:%.*]] = icmp eq %struct.S* [[ARRAYDESTROY_ELEMENT]], [[ARRAY_BEGIN]] 548 // CHECK1-NEXT: br i1 [[ARRAYDESTROY_DONE]], label [[ARRAYDESTROY_DONE2:%.*]], label [[ARRAYDESTROY_BODY]] 549 // CHECK1: arraydestroy.done2: 550 // CHECK1-NEXT: call void @_ZN1SIfED1Ev(%struct.S* nonnull align 4 dereferenceable(4) [[TEST]]) #[[ATTR5]] 551 // CHECK1-NEXT: [[TMP2:%.*]] = load i32, i32* [[RETVAL]], align 4 552 // CHECK1-NEXT: ret i32 [[TMP2]] 553 // 554 // 555 // CHECK1-LABEL: define {{[^@]+}}@_ZN2SSC1ERi 556 // CHECK1-SAME: (%struct.SS* nonnull align 8 dereferenceable(16) [[THIS:%.*]], i32* nonnull align 4 dereferenceable(4) [[D:%.*]]) unnamed_addr #[[ATTR7:[0-9]+]] align 2 { 557 // CHECK1-NEXT: entry: 558 // CHECK1-NEXT: [[THIS_ADDR:%.*]] = alloca %struct.SS*, align 8 559 // CHECK1-NEXT: [[D_ADDR:%.*]] = alloca i32*, align 8 560 // CHECK1-NEXT: store %struct.SS* [[THIS]], %struct.SS** [[THIS_ADDR]], align 8 561 // CHECK1-NEXT: store i32* [[D]], i32** [[D_ADDR]], align 8 562 // CHECK1-NEXT: [[THIS1:%.*]] = load %struct.SS*, %struct.SS** [[THIS_ADDR]], align 8 563 // CHECK1-NEXT: [[TMP0:%.*]] = load i32*, i32** [[D_ADDR]], align 8 564 // CHECK1-NEXT: call void @_ZN2SSC2ERi(%struct.SS* nonnull align 8 dereferenceable(16) [[THIS1]], i32* nonnull align 4 dereferenceable(4) [[TMP0]]) 565 // CHECK1-NEXT: ret void 566 // 567 // 568 // CHECK1-LABEL: define {{[^@]+}}@_ZN1SIfEC1Ev 569 // CHECK1-SAME: (%struct.S* nonnull align 4 dereferenceable(4) [[THIS:%.*]]) unnamed_addr #[[ATTR7]] align 2 { 570 // CHECK1-NEXT: entry: 571 // CHECK1-NEXT: [[THIS_ADDR:%.*]] = alloca %struct.S*, align 8 572 // CHECK1-NEXT: store %struct.S* [[THIS]], %struct.S** [[THIS_ADDR]], align 8 573 // CHECK1-NEXT: [[THIS1:%.*]] = load %struct.S*, %struct.S** [[THIS_ADDR]], align 8 574 // CHECK1-NEXT: call void @_ZN1SIfEC2Ev(%struct.S* nonnull align 4 dereferenceable(4) [[THIS1]]) 575 // CHECK1-NEXT: ret void 576 // 577 // 578 // CHECK1-LABEL: define {{[^@]+}}@_ZN1SIfEC1Ef 579 // CHECK1-SAME: (%struct.S* nonnull align 4 dereferenceable(4) [[THIS:%.*]], float [[A:%.*]]) unnamed_addr #[[ATTR7]] align 2 { 580 // CHECK1-NEXT: entry: 581 // CHECK1-NEXT: [[THIS_ADDR:%.*]] = alloca %struct.S*, align 8 582 // CHECK1-NEXT: [[A_ADDR:%.*]] = alloca float, align 4 583 // CHECK1-NEXT: store %struct.S* [[THIS]], %struct.S** [[THIS_ADDR]], align 8 584 // CHECK1-NEXT: store float [[A]], float* [[A_ADDR]], align 4 585 // CHECK1-NEXT: [[THIS1:%.*]] = load %struct.S*, %struct.S** [[THIS_ADDR]], align 8 586 // CHECK1-NEXT: [[TMP0:%.*]] = load float, float* [[A_ADDR]], align 4 587 // CHECK1-NEXT: call void @_ZN1SIfEC2Ef(%struct.S* nonnull align 4 dereferenceable(4) [[THIS1]], float [[TMP0]]) 588 // CHECK1-NEXT: ret void 589 // 590 // 591 // CHECK1-LABEL: define {{[^@]+}}@.omp_outlined..1 592 // CHECK1-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], [2 x i32]* nonnull align 4 dereferenceable(8) [[VEC:%.*]], float* nonnull align 4 dereferenceable(4) [[T_VAR:%.*]], [2 x %struct.S]* nonnull align 4 dereferenceable(8) [[S_ARR:%.*]], %struct.S* nonnull align 4 dereferenceable(4) [[VAR:%.*]], %struct.S* nonnull align 4 dereferenceable(4) [[VAR1:%.*]], float* nonnull align 4 dereferenceable(4) [[T_VAR1:%.*]]) #[[ATTR1]] { 593 // CHECK1-NEXT: entry: 594 // CHECK1-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 595 // CHECK1-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 596 // CHECK1-NEXT: [[VEC_ADDR:%.*]] = alloca [2 x i32]*, align 8 597 // CHECK1-NEXT: [[T_VAR_ADDR:%.*]] = alloca float*, align 8 598 // CHECK1-NEXT: [[S_ARR_ADDR:%.*]] = alloca [2 x %struct.S]*, align 8 599 // CHECK1-NEXT: [[VAR_ADDR:%.*]] = alloca %struct.S*, align 8 600 // CHECK1-NEXT: [[VAR1_ADDR:%.*]] = alloca %struct.S*, align 8 601 // CHECK1-NEXT: [[T_VAR1_ADDR:%.*]] = alloca float*, align 8 602 // CHECK1-NEXT: [[T_VAR2:%.*]] = alloca float, align 4 603 // CHECK1-NEXT: [[VAR3:%.*]] = alloca [[STRUCT_S:%.*]], align 4 604 // CHECK1-NEXT: [[VAR14:%.*]] = alloca [[STRUCT_S]], align 4 605 // CHECK1-NEXT: [[T_VAR15:%.*]] = alloca float, align 4 606 // CHECK1-NEXT: [[DOTOMP_REDUCTION_RED_LIST:%.*]] = alloca [4 x i8*], align 8 607 // CHECK1-NEXT: [[REF_TMP:%.*]] = alloca [[STRUCT_S]], align 4 608 // CHECK1-NEXT: [[ATOMIC_TEMP:%.*]] = alloca float, align 4 609 // CHECK1-NEXT: [[TMP:%.*]] = alloca float, align 4 610 // CHECK1-NEXT: [[REF_TMP13:%.*]] = alloca [[STRUCT_S]], align 4 611 // CHECK1-NEXT: [[ATOMIC_TEMP23:%.*]] = alloca float, align 4 612 // CHECK1-NEXT: [[_TMP24:%.*]] = alloca float, align 4 613 // CHECK1-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8 614 // CHECK1-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8 615 // CHECK1-NEXT: store [2 x i32]* [[VEC]], [2 x i32]** [[VEC_ADDR]], align 8 616 // CHECK1-NEXT: store float* [[T_VAR]], float** [[T_VAR_ADDR]], align 8 617 // CHECK1-NEXT: store [2 x %struct.S]* [[S_ARR]], [2 x %struct.S]** [[S_ARR_ADDR]], align 8 618 // CHECK1-NEXT: store %struct.S* [[VAR]], %struct.S** [[VAR_ADDR]], align 8 619 // CHECK1-NEXT: store %struct.S* [[VAR1]], %struct.S** [[VAR1_ADDR]], align 8 620 // CHECK1-NEXT: store float* [[T_VAR1]], float** [[T_VAR1_ADDR]], align 8 621 // CHECK1-NEXT: [[TMP0:%.*]] = load [2 x i32]*, [2 x i32]** [[VEC_ADDR]], align 8 622 // CHECK1-NEXT: [[TMP1:%.*]] = load float*, float** [[T_VAR_ADDR]], align 8 623 // CHECK1-NEXT: [[TMP2:%.*]] = load [2 x %struct.S]*, [2 x %struct.S]** [[S_ARR_ADDR]], align 8 624 // CHECK1-NEXT: [[TMP3:%.*]] = load %struct.S*, %struct.S** [[VAR_ADDR]], align 8 625 // CHECK1-NEXT: [[TMP4:%.*]] = load %struct.S*, %struct.S** [[VAR1_ADDR]], align 8 626 // CHECK1-NEXT: [[TMP5:%.*]] = load float*, float** [[T_VAR1_ADDR]], align 8 627 // CHECK1-NEXT: store float 0.000000e+00, float* [[T_VAR2]], align 4 628 // CHECK1-NEXT: call void @_ZN1SIfEC1Ev(%struct.S* nonnull align 4 dereferenceable(4) [[VAR3]]) 629 // CHECK1-NEXT: call void @_ZN1SIfEC1Ev(%struct.S* nonnull align 4 dereferenceable(4) [[VAR14]]) 630 // CHECK1-NEXT: store float 0x47EFFFFFE0000000, float* [[T_VAR15]], align 4 631 // CHECK1-NEXT: [[TMP6:%.*]] = load float, float* [[T_VAR2]], align 4 632 // CHECK1-NEXT: [[CONV:%.*]] = fptosi float [[TMP6]] to i32 633 // CHECK1-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [2 x i32], [2 x i32]* [[TMP0]], i64 0, i64 0 634 // CHECK1-NEXT: store i32 [[CONV]], i32* [[ARRAYIDX]], align 4 635 // CHECK1-NEXT: [[ARRAYIDX6:%.*]] = getelementptr inbounds [2 x %struct.S], [2 x %struct.S]* [[TMP2]], i64 0, i64 0 636 // CHECK1-NEXT: [[TMP7:%.*]] = bitcast %struct.S* [[ARRAYIDX6]] to i8* 637 // CHECK1-NEXT: [[TMP8:%.*]] = bitcast %struct.S* [[VAR3]] to i8* 638 // CHECK1-NEXT: call void @llvm.memcpy.p0i8.p0i8.i64(i8* align 4 [[TMP7]], i8* align 4 [[TMP8]], i64 4, i1 false) 639 // CHECK1-NEXT: [[TMP9:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOMP_REDUCTION_RED_LIST]], i64 0, i64 0 640 // CHECK1-NEXT: [[TMP10:%.*]] = bitcast float* [[T_VAR2]] to i8* 641 // CHECK1-NEXT: store i8* [[TMP10]], i8** [[TMP9]], align 8 642 // CHECK1-NEXT: [[TMP11:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOMP_REDUCTION_RED_LIST]], i64 0, i64 1 643 // CHECK1-NEXT: [[TMP12:%.*]] = bitcast %struct.S* [[VAR3]] to i8* 644 // CHECK1-NEXT: store i8* [[TMP12]], i8** [[TMP11]], align 8 645 // CHECK1-NEXT: [[TMP13:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOMP_REDUCTION_RED_LIST]], i64 0, i64 2 646 // CHECK1-NEXT: [[TMP14:%.*]] = bitcast %struct.S* [[VAR14]] to i8* 647 // CHECK1-NEXT: store i8* [[TMP14]], i8** [[TMP13]], align 8 648 // CHECK1-NEXT: [[TMP15:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOMP_REDUCTION_RED_LIST]], i64 0, i64 3 649 // CHECK1-NEXT: [[TMP16:%.*]] = bitcast float* [[T_VAR15]] to i8* 650 // CHECK1-NEXT: store i8* [[TMP16]], i8** [[TMP15]], align 8 651 // CHECK1-NEXT: [[TMP17:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 652 // CHECK1-NEXT: [[TMP18:%.*]] = load i32, i32* [[TMP17]], align 4 653 // CHECK1-NEXT: [[TMP19:%.*]] = bitcast [4 x i8*]* [[DOTOMP_REDUCTION_RED_LIST]] to i8* 654 // CHECK1-NEXT: [[TMP20:%.*]] = call i32 @__kmpc_reduce_nowait(%struct.ident_t* @[[GLOB1]], i32 [[TMP18]], i32 4, i64 32, i8* [[TMP19]], void (i8*, i8*)* @.omp.reduction.reduction_func.2, [8 x i32]* @.gomp_critical_user_.reduction.var) 655 // CHECK1-NEXT: switch i32 [[TMP20]], label [[DOTOMP_REDUCTION_DEFAULT:%.*]] [ 656 // CHECK1-NEXT: i32 1, label [[DOTOMP_REDUCTION_CASE1:%.*]] 657 // CHECK1-NEXT: i32 2, label [[DOTOMP_REDUCTION_CASE2:%.*]] 658 // CHECK1-NEXT: ] 659 // CHECK1: .omp.reduction.case1: 660 // CHECK1-NEXT: [[TMP21:%.*]] = load float, float* [[TMP1]], align 4 661 // CHECK1-NEXT: [[TMP22:%.*]] = load float, float* [[T_VAR2]], align 4 662 // CHECK1-NEXT: [[ADD:%.*]] = fadd float [[TMP21]], [[TMP22]] 663 // CHECK1-NEXT: store float [[ADD]], float* [[TMP1]], align 4 664 // CHECK1-NEXT: [[CALL:%.*]] = call nonnull align 4 dereferenceable(4) %struct.S* @_ZN1SIfEanERKS0_(%struct.S* nonnull align 4 dereferenceable(4) [[TMP3]], %struct.S* nonnull align 4 dereferenceable(4) [[VAR3]]) 665 // CHECK1-NEXT: [[TMP23:%.*]] = bitcast %struct.S* [[TMP3]] to i8* 666 // CHECK1-NEXT: [[TMP24:%.*]] = bitcast %struct.S* [[CALL]] to i8* 667 // CHECK1-NEXT: call void @llvm.memcpy.p0i8.p0i8.i64(i8* align 4 [[TMP23]], i8* align 4 [[TMP24]], i64 4, i1 false) 668 // CHECK1-NEXT: [[CALL7:%.*]] = call float @_ZN1SIfEcvfEv(%struct.S* nonnull align 4 dereferenceable(4) [[TMP4]]) 669 // CHECK1-NEXT: [[TOBOOL:%.*]] = fcmp une float [[CALL7]], 0.000000e+00 670 // CHECK1-NEXT: br i1 [[TOBOOL]], label [[LAND_RHS:%.*]], label [[LAND_END:%.*]] 671 // CHECK1: land.rhs: 672 // CHECK1-NEXT: [[CALL8:%.*]] = call float @_ZN1SIfEcvfEv(%struct.S* nonnull align 4 dereferenceable(4) [[VAR14]]) 673 // CHECK1-NEXT: [[TOBOOL9:%.*]] = fcmp une float [[CALL8]], 0.000000e+00 674 // CHECK1-NEXT: br label [[LAND_END]] 675 // CHECK1: land.end: 676 // CHECK1-NEXT: [[TMP25:%.*]] = phi i1 [ false, [[DOTOMP_REDUCTION_CASE1]] ], [ [[TOBOOL9]], [[LAND_RHS]] ] 677 // CHECK1-NEXT: [[CONV10:%.*]] = uitofp i1 [[TMP25]] to float 678 // CHECK1-NEXT: call void @_ZN1SIfEC1Ef(%struct.S* nonnull align 4 dereferenceable(4) [[REF_TMP]], float [[CONV10]]) 679 // CHECK1-NEXT: [[TMP26:%.*]] = bitcast %struct.S* [[TMP4]] to i8* 680 // CHECK1-NEXT: [[TMP27:%.*]] = bitcast %struct.S* [[REF_TMP]] to i8* 681 // CHECK1-NEXT: call void @llvm.memcpy.p0i8.p0i8.i64(i8* align 4 [[TMP26]], i8* align 4 [[TMP27]], i64 4, i1 false) 682 // CHECK1-NEXT: call void @_ZN1SIfED1Ev(%struct.S* nonnull align 4 dereferenceable(4) [[REF_TMP]]) #[[ATTR5]] 683 // CHECK1-NEXT: [[TMP28:%.*]] = load float, float* [[TMP5]], align 4 684 // CHECK1-NEXT: [[TMP29:%.*]] = load float, float* [[T_VAR15]], align 4 685 // CHECK1-NEXT: [[CMP:%.*]] = fcmp olt float [[TMP28]], [[TMP29]] 686 // CHECK1-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] 687 // CHECK1: cond.true: 688 // CHECK1-NEXT: [[TMP30:%.*]] = load float, float* [[TMP5]], align 4 689 // CHECK1-NEXT: br label [[COND_END:%.*]] 690 // CHECK1: cond.false: 691 // CHECK1-NEXT: [[TMP31:%.*]] = load float, float* [[T_VAR15]], align 4 692 // CHECK1-NEXT: br label [[COND_END]] 693 // CHECK1: cond.end: 694 // CHECK1-NEXT: [[COND:%.*]] = phi float [ [[TMP30]], [[COND_TRUE]] ], [ [[TMP31]], [[COND_FALSE]] ] 695 // CHECK1-NEXT: store float [[COND]], float* [[TMP5]], align 4 696 // CHECK1-NEXT: call void @__kmpc_end_reduce_nowait(%struct.ident_t* @[[GLOB1]], i32 [[TMP18]], [8 x i32]* @.gomp_critical_user_.reduction.var) 697 // CHECK1-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]] 698 // CHECK1: .omp.reduction.case2: 699 // CHECK1-NEXT: [[TMP32:%.*]] = load float, float* [[T_VAR2]], align 4 700 // CHECK1-NEXT: [[TMP33:%.*]] = bitcast float* [[TMP1]] to i32* 701 // CHECK1-NEXT: [[ATOMIC_LOAD:%.*]] = load atomic i32, i32* [[TMP33]] monotonic, align 4 702 // CHECK1-NEXT: br label [[ATOMIC_CONT:%.*]] 703 // CHECK1: atomic_cont: 704 // CHECK1-NEXT: [[TMP34:%.*]] = phi i32 [ [[ATOMIC_LOAD]], [[DOTOMP_REDUCTION_CASE2]] ], [ [[TMP42:%.*]], [[ATOMIC_CONT]] ] 705 // CHECK1-NEXT: [[TMP35:%.*]] = bitcast float* [[ATOMIC_TEMP]] to i32* 706 // CHECK1-NEXT: [[TMP36:%.*]] = bitcast i32 [[TMP34]] to float 707 // CHECK1-NEXT: store float [[TMP36]], float* [[TMP]], align 4 708 // CHECK1-NEXT: [[TMP37:%.*]] = load float, float* [[TMP]], align 4 709 // CHECK1-NEXT: [[TMP38:%.*]] = load float, float* [[T_VAR2]], align 4 710 // CHECK1-NEXT: [[ADD11:%.*]] = fadd float [[TMP37]], [[TMP38]] 711 // CHECK1-NEXT: store float [[ADD11]], float* [[ATOMIC_TEMP]], align 4 712 // CHECK1-NEXT: [[TMP39:%.*]] = load i32, i32* [[TMP35]], align 4 713 // CHECK1-NEXT: [[TMP40:%.*]] = bitcast float* [[TMP1]] to i32* 714 // CHECK1-NEXT: [[TMP41:%.*]] = cmpxchg i32* [[TMP40]], i32 [[TMP34]], i32 [[TMP39]] monotonic monotonic, align 4 715 // CHECK1-NEXT: [[TMP42]] = extractvalue { i32, i1 } [[TMP41]], 0 716 // CHECK1-NEXT: [[TMP43:%.*]] = extractvalue { i32, i1 } [[TMP41]], 1 717 // CHECK1-NEXT: br i1 [[TMP43]], label [[ATOMIC_EXIT:%.*]], label [[ATOMIC_CONT]] 718 // CHECK1: atomic_exit: 719 // CHECK1-NEXT: call void @__kmpc_critical(%struct.ident_t* @[[GLOB2]], i32 [[TMP18]], [8 x i32]* @.gomp_critical_user_.atomic_reduction.var) 720 // CHECK1-NEXT: [[CALL12:%.*]] = call nonnull align 4 dereferenceable(4) %struct.S* @_ZN1SIfEanERKS0_(%struct.S* nonnull align 4 dereferenceable(4) [[TMP3]], %struct.S* nonnull align 4 dereferenceable(4) [[VAR3]]) 721 // CHECK1-NEXT: [[TMP44:%.*]] = bitcast %struct.S* [[TMP3]] to i8* 722 // CHECK1-NEXT: [[TMP45:%.*]] = bitcast %struct.S* [[CALL12]] to i8* 723 // CHECK1-NEXT: call void @llvm.memcpy.p0i8.p0i8.i64(i8* align 4 [[TMP44]], i8* align 4 [[TMP45]], i64 4, i1 false) 724 // CHECK1-NEXT: call void @__kmpc_end_critical(%struct.ident_t* @[[GLOB2]], i32 [[TMP18]], [8 x i32]* @.gomp_critical_user_.atomic_reduction.var) 725 // CHECK1-NEXT: call void @__kmpc_critical(%struct.ident_t* @[[GLOB2]], i32 [[TMP18]], [8 x i32]* @.gomp_critical_user_.atomic_reduction.var) 726 // CHECK1-NEXT: [[CALL14:%.*]] = call float @_ZN1SIfEcvfEv(%struct.S* nonnull align 4 dereferenceable(4) [[TMP4]]) 727 // CHECK1-NEXT: [[TOBOOL15:%.*]] = fcmp une float [[CALL14]], 0.000000e+00 728 // CHECK1-NEXT: br i1 [[TOBOOL15]], label [[LAND_RHS16:%.*]], label [[LAND_END19:%.*]] 729 // CHECK1: land.rhs16: 730 // CHECK1-NEXT: [[CALL17:%.*]] = call float @_ZN1SIfEcvfEv(%struct.S* nonnull align 4 dereferenceable(4) [[VAR14]]) 731 // CHECK1-NEXT: [[TOBOOL18:%.*]] = fcmp une float [[CALL17]], 0.000000e+00 732 // CHECK1-NEXT: br label [[LAND_END19]] 733 // CHECK1: land.end19: 734 // CHECK1-NEXT: [[TMP46:%.*]] = phi i1 [ false, [[ATOMIC_EXIT]] ], [ [[TOBOOL18]], [[LAND_RHS16]] ] 735 // CHECK1-NEXT: [[CONV20:%.*]] = uitofp i1 [[TMP46]] to float 736 // CHECK1-NEXT: call void @_ZN1SIfEC1Ef(%struct.S* nonnull align 4 dereferenceable(4) [[REF_TMP13]], float [[CONV20]]) 737 // CHECK1-NEXT: [[TMP47:%.*]] = bitcast %struct.S* [[TMP4]] to i8* 738 // CHECK1-NEXT: [[TMP48:%.*]] = bitcast %struct.S* [[REF_TMP13]] to i8* 739 // CHECK1-NEXT: call void @llvm.memcpy.p0i8.p0i8.i64(i8* align 4 [[TMP47]], i8* align 4 [[TMP48]], i64 4, i1 false) 740 // CHECK1-NEXT: call void @_ZN1SIfED1Ev(%struct.S* nonnull align 4 dereferenceable(4) [[REF_TMP13]]) #[[ATTR5]] 741 // CHECK1-NEXT: call void @__kmpc_end_critical(%struct.ident_t* @[[GLOB2]], i32 [[TMP18]], [8 x i32]* @.gomp_critical_user_.atomic_reduction.var) 742 // CHECK1-NEXT: [[TMP49:%.*]] = load float, float* [[T_VAR15]], align 4 743 // CHECK1-NEXT: [[TMP50:%.*]] = bitcast float* [[TMP5]] to i32* 744 // CHECK1-NEXT: [[ATOMIC_LOAD21:%.*]] = load atomic i32, i32* [[TMP50]] monotonic, align 4 745 // CHECK1-NEXT: br label [[ATOMIC_CONT22:%.*]] 746 // CHECK1: atomic_cont22: 747 // CHECK1-NEXT: [[TMP51:%.*]] = phi i32 [ [[ATOMIC_LOAD21]], [[LAND_END19]] ], [ [[TMP61:%.*]], [[COND_END28:%.*]] ] 748 // CHECK1-NEXT: [[TMP52:%.*]] = bitcast float* [[ATOMIC_TEMP23]] to i32* 749 // CHECK1-NEXT: [[TMP53:%.*]] = bitcast i32 [[TMP51]] to float 750 // CHECK1-NEXT: store float [[TMP53]], float* [[_TMP24]], align 4 751 // CHECK1-NEXT: [[TMP54:%.*]] = load float, float* [[_TMP24]], align 4 752 // CHECK1-NEXT: [[TMP55:%.*]] = load float, float* [[T_VAR15]], align 4 753 // CHECK1-NEXT: [[CMP25:%.*]] = fcmp olt float [[TMP54]], [[TMP55]] 754 // CHECK1-NEXT: br i1 [[CMP25]], label [[COND_TRUE26:%.*]], label [[COND_FALSE27:%.*]] 755 // CHECK1: cond.true26: 756 // CHECK1-NEXT: [[TMP56:%.*]] = load float, float* [[_TMP24]], align 4 757 // CHECK1-NEXT: br label [[COND_END28]] 758 // CHECK1: cond.false27: 759 // CHECK1-NEXT: [[TMP57:%.*]] = load float, float* [[T_VAR15]], align 4 760 // CHECK1-NEXT: br label [[COND_END28]] 761 // CHECK1: cond.end28: 762 // CHECK1-NEXT: [[COND29:%.*]] = phi float [ [[TMP56]], [[COND_TRUE26]] ], [ [[TMP57]], [[COND_FALSE27]] ] 763 // CHECK1-NEXT: store float [[COND29]], float* [[ATOMIC_TEMP23]], align 4 764 // CHECK1-NEXT: [[TMP58:%.*]] = load i32, i32* [[TMP52]], align 4 765 // CHECK1-NEXT: [[TMP59:%.*]] = bitcast float* [[TMP5]] to i32* 766 // CHECK1-NEXT: [[TMP60:%.*]] = cmpxchg i32* [[TMP59]], i32 [[TMP51]], i32 [[TMP58]] monotonic monotonic, align 4 767 // CHECK1-NEXT: [[TMP61]] = extractvalue { i32, i1 } [[TMP60]], 0 768 // CHECK1-NEXT: [[TMP62:%.*]] = extractvalue { i32, i1 } [[TMP60]], 1 769 // CHECK1-NEXT: br i1 [[TMP62]], label [[ATOMIC_EXIT30:%.*]], label [[ATOMIC_CONT22]] 770 // CHECK1: atomic_exit30: 771 // CHECK1-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]] 772 // CHECK1: .omp.reduction.default: 773 // CHECK1-NEXT: call void @_ZN1SIfED1Ev(%struct.S* nonnull align 4 dereferenceable(4) [[VAR14]]) #[[ATTR5]] 774 // CHECK1-NEXT: call void @_ZN1SIfED1Ev(%struct.S* nonnull align 4 dereferenceable(4) [[VAR3]]) #[[ATTR5]] 775 // CHECK1-NEXT: ret void 776 // 777 // 778 // CHECK1-LABEL: define {{[^@]+}}@.omp.reduction.reduction_func.2 779 // CHECK1-SAME: (i8* [[TMP0:%.*]], i8* [[TMP1:%.*]]) #[[ATTR3]] { 780 // CHECK1-NEXT: entry: 781 // CHECK1-NEXT: [[DOTADDR:%.*]] = alloca i8*, align 8 782 // CHECK1-NEXT: [[DOTADDR1:%.*]] = alloca i8*, align 8 783 // CHECK1-NEXT: [[REF_TMP:%.*]] = alloca [[STRUCT_S:%.*]], align 4 784 // CHECK1-NEXT: store i8* [[TMP0]], i8** [[DOTADDR]], align 8 785 // CHECK1-NEXT: store i8* [[TMP1]], i8** [[DOTADDR1]], align 8 786 // CHECK1-NEXT: [[TMP2:%.*]] = load i8*, i8** [[DOTADDR]], align 8 787 // CHECK1-NEXT: [[TMP3:%.*]] = bitcast i8* [[TMP2]] to [4 x i8*]* 788 // CHECK1-NEXT: [[TMP4:%.*]] = load i8*, i8** [[DOTADDR1]], align 8 789 // CHECK1-NEXT: [[TMP5:%.*]] = bitcast i8* [[TMP4]] to [4 x i8*]* 790 // CHECK1-NEXT: [[TMP6:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[TMP5]], i64 0, i64 0 791 // CHECK1-NEXT: [[TMP7:%.*]] = load i8*, i8** [[TMP6]], align 8 792 // CHECK1-NEXT: [[TMP8:%.*]] = bitcast i8* [[TMP7]] to float* 793 // CHECK1-NEXT: [[TMP9:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[TMP3]], i64 0, i64 0 794 // CHECK1-NEXT: [[TMP10:%.*]] = load i8*, i8** [[TMP9]], align 8 795 // CHECK1-NEXT: [[TMP11:%.*]] = bitcast i8* [[TMP10]] to float* 796 // CHECK1-NEXT: [[TMP12:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[TMP5]], i64 0, i64 1 797 // CHECK1-NEXT: [[TMP13:%.*]] = load i8*, i8** [[TMP12]], align 8 798 // CHECK1-NEXT: [[TMP14:%.*]] = bitcast i8* [[TMP13]] to %struct.S* 799 // CHECK1-NEXT: [[TMP15:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[TMP3]], i64 0, i64 1 800 // CHECK1-NEXT: [[TMP16:%.*]] = load i8*, i8** [[TMP15]], align 8 801 // CHECK1-NEXT: [[TMP17:%.*]] = bitcast i8* [[TMP16]] to %struct.S* 802 // CHECK1-NEXT: [[TMP18:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[TMP5]], i64 0, i64 2 803 // CHECK1-NEXT: [[TMP19:%.*]] = load i8*, i8** [[TMP18]], align 8 804 // CHECK1-NEXT: [[TMP20:%.*]] = bitcast i8* [[TMP19]] to %struct.S* 805 // CHECK1-NEXT: [[TMP21:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[TMP3]], i64 0, i64 2 806 // CHECK1-NEXT: [[TMP22:%.*]] = load i8*, i8** [[TMP21]], align 8 807 // CHECK1-NEXT: [[TMP23:%.*]] = bitcast i8* [[TMP22]] to %struct.S* 808 // CHECK1-NEXT: [[TMP24:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[TMP5]], i64 0, i64 3 809 // CHECK1-NEXT: [[TMP25:%.*]] = load i8*, i8** [[TMP24]], align 8 810 // CHECK1-NEXT: [[TMP26:%.*]] = bitcast i8* [[TMP25]] to float* 811 // CHECK1-NEXT: [[TMP27:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[TMP3]], i64 0, i64 3 812 // CHECK1-NEXT: [[TMP28:%.*]] = load i8*, i8** [[TMP27]], align 8 813 // CHECK1-NEXT: [[TMP29:%.*]] = bitcast i8* [[TMP28]] to float* 814 // CHECK1-NEXT: [[TMP30:%.*]] = load float, float* [[TMP11]], align 4 815 // CHECK1-NEXT: [[TMP31:%.*]] = load float, float* [[TMP8]], align 4 816 // CHECK1-NEXT: [[ADD:%.*]] = fadd float [[TMP30]], [[TMP31]] 817 // CHECK1-NEXT: store float [[ADD]], float* [[TMP11]], align 4 818 // CHECK1-NEXT: [[CALL:%.*]] = call nonnull align 4 dereferenceable(4) %struct.S* @_ZN1SIfEanERKS0_(%struct.S* nonnull align 4 dereferenceable(4) [[TMP17]], %struct.S* nonnull align 4 dereferenceable(4) [[TMP14]]) 819 // CHECK1-NEXT: [[TMP32:%.*]] = bitcast %struct.S* [[TMP17]] to i8* 820 // CHECK1-NEXT: [[TMP33:%.*]] = bitcast %struct.S* [[CALL]] to i8* 821 // CHECK1-NEXT: call void @llvm.memcpy.p0i8.p0i8.i64(i8* align 4 [[TMP32]], i8* align 4 [[TMP33]], i64 4, i1 false) 822 // CHECK1-NEXT: [[CALL2:%.*]] = call float @_ZN1SIfEcvfEv(%struct.S* nonnull align 4 dereferenceable(4) [[TMP23]]) 823 // CHECK1-NEXT: [[TOBOOL:%.*]] = fcmp une float [[CALL2]], 0.000000e+00 824 // CHECK1-NEXT: br i1 [[TOBOOL]], label [[LAND_RHS:%.*]], label [[LAND_END:%.*]] 825 // CHECK1: land.rhs: 826 // CHECK1-NEXT: [[CALL3:%.*]] = call float @_ZN1SIfEcvfEv(%struct.S* nonnull align 4 dereferenceable(4) [[TMP20]]) 827 // CHECK1-NEXT: [[TOBOOL4:%.*]] = fcmp une float [[CALL3]], 0.000000e+00 828 // CHECK1-NEXT: br label [[LAND_END]] 829 // CHECK1: land.end: 830 // CHECK1-NEXT: [[TMP34:%.*]] = phi i1 [ false, [[ENTRY:%.*]] ], [ [[TOBOOL4]], [[LAND_RHS]] ] 831 // CHECK1-NEXT: [[CONV:%.*]] = uitofp i1 [[TMP34]] to float 832 // CHECK1-NEXT: call void @_ZN1SIfEC1Ef(%struct.S* nonnull align 4 dereferenceable(4) [[REF_TMP]], float [[CONV]]) 833 // CHECK1-NEXT: [[TMP35:%.*]] = bitcast %struct.S* [[TMP23]] to i8* 834 // CHECK1-NEXT: [[TMP36:%.*]] = bitcast %struct.S* [[REF_TMP]] to i8* 835 // CHECK1-NEXT: call void @llvm.memcpy.p0i8.p0i8.i64(i8* align 4 [[TMP35]], i8* align 4 [[TMP36]], i64 4, i1 false) 836 // CHECK1-NEXT: call void @_ZN1SIfED1Ev(%struct.S* nonnull align 4 dereferenceable(4) [[REF_TMP]]) #[[ATTR5]] 837 // CHECK1-NEXT: [[TMP37:%.*]] = load float, float* [[TMP29]], align 4 838 // CHECK1-NEXT: [[TMP38:%.*]] = load float, float* [[TMP26]], align 4 839 // CHECK1-NEXT: [[CMP:%.*]] = fcmp olt float [[TMP37]], [[TMP38]] 840 // CHECK1-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] 841 // CHECK1: cond.true: 842 // CHECK1-NEXT: [[TMP39:%.*]] = load float, float* [[TMP29]], align 4 843 // CHECK1-NEXT: br label [[COND_END:%.*]] 844 // CHECK1: cond.false: 845 // CHECK1-NEXT: [[TMP40:%.*]] = load float, float* [[TMP26]], align 4 846 // CHECK1-NEXT: br label [[COND_END]] 847 // CHECK1: cond.end: 848 // CHECK1-NEXT: [[COND:%.*]] = phi float [ [[TMP39]], [[COND_TRUE]] ], [ [[TMP40]], [[COND_FALSE]] ] 849 // CHECK1-NEXT: store float [[COND]], float* [[TMP29]], align 4 850 // CHECK1-NEXT: ret void 851 // 852 // 853 // CHECK1-LABEL: define {{[^@]+}}@_ZN1SIfEanERKS0_ 854 // CHECK1-SAME: (%struct.S* nonnull align 4 dereferenceable(4) [[THIS:%.*]], %struct.S* nonnull align 4 dereferenceable(4) [[TMP0:%.*]]) #[[ATTR0]] align 2 { 855 // CHECK1-NEXT: entry: 856 // CHECK1-NEXT: [[THIS_ADDR:%.*]] = alloca %struct.S*, align 8 857 // CHECK1-NEXT: [[DOTADDR:%.*]] = alloca %struct.S*, align 8 858 // CHECK1-NEXT: store %struct.S* [[THIS]], %struct.S** [[THIS_ADDR]], align 8 859 // CHECK1-NEXT: store %struct.S* [[TMP0]], %struct.S** [[DOTADDR]], align 8 860 // CHECK1-NEXT: [[THIS1:%.*]] = load %struct.S*, %struct.S** [[THIS_ADDR]], align 8 861 // CHECK1-NEXT: ret %struct.S* [[THIS1]] 862 // 863 // 864 // CHECK1-LABEL: define {{[^@]+}}@_ZN1SIfEcvfEv 865 // CHECK1-SAME: (%struct.S* nonnull align 4 dereferenceable(4) [[THIS:%.*]]) #[[ATTR0]] align 2 { 866 // CHECK1-NEXT: entry: 867 // CHECK1-NEXT: [[THIS_ADDR:%.*]] = alloca %struct.S*, align 8 868 // CHECK1-NEXT: store %struct.S* [[THIS]], %struct.S** [[THIS_ADDR]], align 8 869 // CHECK1-NEXT: [[THIS1:%.*]] = load %struct.S*, %struct.S** [[THIS_ADDR]], align 8 870 // CHECK1-NEXT: ret float 0.000000e+00 871 // 872 // 873 // CHECK1-LABEL: define {{[^@]+}}@_ZN1SIfED1Ev 874 // CHECK1-SAME: (%struct.S* nonnull align 4 dereferenceable(4) [[THIS:%.*]]) unnamed_addr #[[ATTR7]] align 2 { 875 // CHECK1-NEXT: entry: 876 // CHECK1-NEXT: [[THIS_ADDR:%.*]] = alloca %struct.S*, align 8 877 // CHECK1-NEXT: store %struct.S* [[THIS]], %struct.S** [[THIS_ADDR]], align 8 878 // CHECK1-NEXT: [[THIS1:%.*]] = load %struct.S*, %struct.S** [[THIS_ADDR]], align 8 879 // CHECK1-NEXT: call void @_ZN1SIfED2Ev(%struct.S* nonnull align 4 dereferenceable(4) [[THIS1]]) #[[ATTR5]] 880 // CHECK1-NEXT: ret void 881 // 882 // 883 // CHECK1-LABEL: define {{[^@]+}}@.omp_outlined..3 884 // CHECK1-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], [2 x i32]* nonnull align 4 dereferenceable(8) [[VEC:%.*]], float* nonnull align 4 dereferenceable(4) [[T_VAR:%.*]], [2 x %struct.S]* nonnull align 4 dereferenceable(8) [[S_ARR:%.*]], %struct.S* nonnull align 4 dereferenceable(4) [[VAR:%.*]], %struct.S* nonnull align 4 dereferenceable(4) [[VAR1:%.*]], float* nonnull align 4 dereferenceable(4) [[T_VAR1:%.*]]) #[[ATTR1]] { 885 // CHECK1-NEXT: entry: 886 // CHECK1-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 887 // CHECK1-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 888 // CHECK1-NEXT: [[VEC_ADDR:%.*]] = alloca [2 x i32]*, align 8 889 // CHECK1-NEXT: [[T_VAR_ADDR:%.*]] = alloca float*, align 8 890 // CHECK1-NEXT: [[S_ARR_ADDR:%.*]] = alloca [2 x %struct.S]*, align 8 891 // CHECK1-NEXT: [[VAR_ADDR:%.*]] = alloca %struct.S*, align 8 892 // CHECK1-NEXT: [[VAR1_ADDR:%.*]] = alloca %struct.S*, align 8 893 // CHECK1-NEXT: [[T_VAR1_ADDR:%.*]] = alloca float*, align 8 894 // CHECK1-NEXT: [[T_VAR2:%.*]] = alloca float, align 4 895 // CHECK1-NEXT: [[VAR3:%.*]] = alloca [[STRUCT_S:%.*]], align 4 896 // CHECK1-NEXT: [[VAR14:%.*]] = alloca [[STRUCT_S]], align 4 897 // CHECK1-NEXT: [[T_VAR15:%.*]] = alloca float, align 4 898 // CHECK1-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8 899 // CHECK1-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8 900 // CHECK1-NEXT: store [2 x i32]* [[VEC]], [2 x i32]** [[VEC_ADDR]], align 8 901 // CHECK1-NEXT: store float* [[T_VAR]], float** [[T_VAR_ADDR]], align 8 902 // CHECK1-NEXT: store [2 x %struct.S]* [[S_ARR]], [2 x %struct.S]** [[S_ARR_ADDR]], align 8 903 // CHECK1-NEXT: store %struct.S* [[VAR]], %struct.S** [[VAR_ADDR]], align 8 904 // CHECK1-NEXT: store %struct.S* [[VAR1]], %struct.S** [[VAR1_ADDR]], align 8 905 // CHECK1-NEXT: store float* [[T_VAR1]], float** [[T_VAR1_ADDR]], align 8 906 // CHECK1-NEXT: [[TMP0:%.*]] = load [2 x i32]*, [2 x i32]** [[VEC_ADDR]], align 8 907 // CHECK1-NEXT: [[TMP1:%.*]] = load float*, float** [[T_VAR_ADDR]], align 8 908 // CHECK1-NEXT: [[TMP2:%.*]] = load [2 x %struct.S]*, [2 x %struct.S]** [[S_ARR_ADDR]], align 8 909 // CHECK1-NEXT: [[TMP3:%.*]] = load %struct.S*, %struct.S** [[VAR_ADDR]], align 8 910 // CHECK1-NEXT: [[TMP4:%.*]] = load %struct.S*, %struct.S** [[VAR1_ADDR]], align 8 911 // CHECK1-NEXT: [[TMP5:%.*]] = load float*, float** [[T_VAR1_ADDR]], align 8 912 // CHECK1-NEXT: store float 0.000000e+00, float* [[T_VAR2]], align 4 913 // CHECK1-NEXT: call void @_ZN1SIfEC1Ev(%struct.S* nonnull align 4 dereferenceable(4) [[VAR3]]) 914 // CHECK1-NEXT: call void @_ZN1SIfEC1Ev(%struct.S* nonnull align 4 dereferenceable(4) [[VAR14]]) 915 // CHECK1-NEXT: store float 0x47EFFFFFE0000000, float* [[T_VAR15]], align 4 916 // CHECK1-NEXT: br label [[WHILE_COND:%.*]] 917 // CHECK1: while.cond: 918 // CHECK1-NEXT: br label [[WHILE_BODY:%.*]] 919 // CHECK1: while.body: 920 // CHECK1-NEXT: [[TMP6:%.*]] = load float, float* [[T_VAR2]], align 4 921 // CHECK1-NEXT: [[CONV:%.*]] = fptosi float [[TMP6]] to i32 922 // CHECK1-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [2 x i32], [2 x i32]* [[TMP0]], i64 0, i64 0 923 // CHECK1-NEXT: store i32 [[CONV]], i32* [[ARRAYIDX]], align 4 924 // CHECK1-NEXT: [[ARRAYIDX6:%.*]] = getelementptr inbounds [2 x %struct.S], [2 x %struct.S]* [[TMP2]], i64 0, i64 0 925 // CHECK1-NEXT: [[TMP7:%.*]] = bitcast %struct.S* [[ARRAYIDX6]] to i8* 926 // CHECK1-NEXT: [[TMP8:%.*]] = bitcast %struct.S* [[VAR3]] to i8* 927 // CHECK1-NEXT: call void @llvm.memcpy.p0i8.p0i8.i64(i8* align 4 [[TMP7]], i8* align 4 [[TMP8]], i64 4, i1 false) 928 // CHECK1-NEXT: br label [[WHILE_COND]], !llvm.loop [[LOOP4:![0-9]+]] 929 // 930 // 931 // CHECK1-LABEL: define {{[^@]+}}@.omp_outlined..4 932 // CHECK1-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], { float, float }* nonnull align 4 dereferenceable(8) [[CF:%.*]]) #[[ATTR1]] { 933 // CHECK1-NEXT: entry: 934 // CHECK1-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 935 // CHECK1-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 936 // CHECK1-NEXT: [[CF_ADDR:%.*]] = alloca { float, float }*, align 8 937 // CHECK1-NEXT: [[CF1:%.*]] = alloca { float, float }, align 4 938 // CHECK1-NEXT: [[DOTOMP_REDUCTION_RED_LIST:%.*]] = alloca [1 x i8*], align 8 939 // CHECK1-NEXT: [[ATOMIC_TEMP:%.*]] = alloca { float, float }, align 4 940 // CHECK1-NEXT: [[ATOMIC_TEMP10:%.*]] = alloca { float, float }, align 4 941 // CHECK1-NEXT: [[TMP:%.*]] = alloca { float, float }, align 4 942 // CHECK1-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8 943 // CHECK1-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8 944 // CHECK1-NEXT: store { float, float }* [[CF]], { float, float }** [[CF_ADDR]], align 8 945 // CHECK1-NEXT: [[TMP0:%.*]] = load { float, float }*, { float, float }** [[CF_ADDR]], align 8 946 // CHECK1-NEXT: [[CF1_REALP:%.*]] = getelementptr inbounds { float, float }, { float, float }* [[CF1]], i32 0, i32 0 947 // CHECK1-NEXT: [[CF1_IMAGP:%.*]] = getelementptr inbounds { float, float }, { float, float }* [[CF1]], i32 0, i32 1 948 // CHECK1-NEXT: store float 0.000000e+00, float* [[CF1_REALP]], align 4 949 // CHECK1-NEXT: store float 0.000000e+00, float* [[CF1_IMAGP]], align 4 950 // CHECK1-NEXT: [[TMP1:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOMP_REDUCTION_RED_LIST]], i64 0, i64 0 951 // CHECK1-NEXT: [[TMP2:%.*]] = bitcast { float, float }* [[CF1]] to i8* 952 // CHECK1-NEXT: store i8* [[TMP2]], i8** [[TMP1]], align 8 953 // CHECK1-NEXT: [[TMP3:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 954 // CHECK1-NEXT: [[TMP4:%.*]] = load i32, i32* [[TMP3]], align 4 955 // CHECK1-NEXT: [[TMP5:%.*]] = bitcast [1 x i8*]* [[DOTOMP_REDUCTION_RED_LIST]] to i8* 956 // CHECK1-NEXT: [[TMP6:%.*]] = call i32 @__kmpc_reduce_nowait(%struct.ident_t* @[[GLOB1]], i32 [[TMP4]], i32 1, i64 8, i8* [[TMP5]], void (i8*, i8*)* @.omp.reduction.reduction_func.5, [8 x i32]* @.gomp_critical_user_.reduction.var) 957 // CHECK1-NEXT: switch i32 [[TMP6]], label [[DOTOMP_REDUCTION_DEFAULT:%.*]] [ 958 // CHECK1-NEXT: i32 1, label [[DOTOMP_REDUCTION_CASE1:%.*]] 959 // CHECK1-NEXT: i32 2, label [[DOTOMP_REDUCTION_CASE2:%.*]] 960 // CHECK1-NEXT: ] 961 // CHECK1: .omp.reduction.case1: 962 // CHECK1-NEXT: [[DOTREALP:%.*]] = getelementptr inbounds { float, float }, { float, float }* [[TMP0]], i32 0, i32 0 963 // CHECK1-NEXT: [[DOTREAL:%.*]] = load float, float* [[DOTREALP]], align 4 964 // CHECK1-NEXT: [[DOTIMAGP:%.*]] = getelementptr inbounds { float, float }, { float, float }* [[TMP0]], i32 0, i32 1 965 // CHECK1-NEXT: [[DOTIMAG:%.*]] = load float, float* [[DOTIMAGP]], align 4 966 // CHECK1-NEXT: [[CF1_REALP2:%.*]] = getelementptr inbounds { float, float }, { float, float }* [[CF1]], i32 0, i32 0 967 // CHECK1-NEXT: [[CF1_REAL:%.*]] = load float, float* [[CF1_REALP2]], align 4 968 // CHECK1-NEXT: [[CF1_IMAGP3:%.*]] = getelementptr inbounds { float, float }, { float, float }* [[CF1]], i32 0, i32 1 969 // CHECK1-NEXT: [[CF1_IMAG:%.*]] = load float, float* [[CF1_IMAGP3]], align 4 970 // CHECK1-NEXT: [[ADD_R:%.*]] = fadd float [[DOTREAL]], [[CF1_REAL]] 971 // CHECK1-NEXT: [[ADD_I:%.*]] = fadd float [[DOTIMAG]], [[CF1_IMAG]] 972 // CHECK1-NEXT: [[DOTREALP4:%.*]] = getelementptr inbounds { float, float }, { float, float }* [[TMP0]], i32 0, i32 0 973 // CHECK1-NEXT: [[DOTIMAGP5:%.*]] = getelementptr inbounds { float, float }, { float, float }* [[TMP0]], i32 0, i32 1 974 // CHECK1-NEXT: store float [[ADD_R]], float* [[DOTREALP4]], align 4 975 // CHECK1-NEXT: store float [[ADD_I]], float* [[DOTIMAGP5]], align 4 976 // CHECK1-NEXT: call void @__kmpc_end_reduce_nowait(%struct.ident_t* @[[GLOB1]], i32 [[TMP4]], [8 x i32]* @.gomp_critical_user_.reduction.var) 977 // CHECK1-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]] 978 // CHECK1: .omp.reduction.case2: 979 // CHECK1-NEXT: [[CF1_REALP6:%.*]] = getelementptr inbounds { float, float }, { float, float }* [[CF1]], i32 0, i32 0 980 // CHECK1-NEXT: [[CF1_REAL7:%.*]] = load float, float* [[CF1_REALP6]], align 4 981 // CHECK1-NEXT: [[CF1_IMAGP8:%.*]] = getelementptr inbounds { float, float }, { float, float }* [[CF1]], i32 0, i32 1 982 // CHECK1-NEXT: [[CF1_IMAG9:%.*]] = load float, float* [[CF1_IMAGP8]], align 4 983 // CHECK1-NEXT: [[TMP7:%.*]] = bitcast { float, float }* [[TMP0]] to i8* 984 // CHECK1-NEXT: [[TMP8:%.*]] = bitcast { float, float }* [[ATOMIC_TEMP]] to i8* 985 // CHECK1-NEXT: call void @__atomic_load(i64 8, i8* [[TMP7]], i8* [[TMP8]], i32 0) 986 // CHECK1-NEXT: br label [[ATOMIC_CONT:%.*]] 987 // CHECK1: atomic_cont: 988 // CHECK1-NEXT: [[ATOMIC_TEMP_REALP:%.*]] = getelementptr inbounds { float, float }, { float, float }* [[ATOMIC_TEMP]], i32 0, i32 0 989 // CHECK1-NEXT: [[ATOMIC_TEMP_REAL:%.*]] = load float, float* [[ATOMIC_TEMP_REALP]], align 4 990 // CHECK1-NEXT: [[ATOMIC_TEMP_IMAGP:%.*]] = getelementptr inbounds { float, float }, { float, float }* [[ATOMIC_TEMP]], i32 0, i32 1 991 // CHECK1-NEXT: [[ATOMIC_TEMP_IMAG:%.*]] = load float, float* [[ATOMIC_TEMP_IMAGP]], align 4 992 // CHECK1-NEXT: [[TMP_REALP:%.*]] = getelementptr inbounds { float, float }, { float, float }* [[TMP]], i32 0, i32 0 993 // CHECK1-NEXT: [[TMP_IMAGP:%.*]] = getelementptr inbounds { float, float }, { float, float }* [[TMP]], i32 0, i32 1 994 // CHECK1-NEXT: store float [[ATOMIC_TEMP_REAL]], float* [[TMP_REALP]], align 4 995 // CHECK1-NEXT: store float [[ATOMIC_TEMP_IMAG]], float* [[TMP_IMAGP]], align 4 996 // CHECK1-NEXT: [[TMP_REALP11:%.*]] = getelementptr inbounds { float, float }, { float, float }* [[TMP]], i32 0, i32 0 997 // CHECK1-NEXT: [[TMP_REAL:%.*]] = load float, float* [[TMP_REALP11]], align 4 998 // CHECK1-NEXT: [[TMP_IMAGP12:%.*]] = getelementptr inbounds { float, float }, { float, float }* [[TMP]], i32 0, i32 1 999 // CHECK1-NEXT: [[TMP_IMAG:%.*]] = load float, float* [[TMP_IMAGP12]], align 4 1000 // CHECK1-NEXT: [[CF1_REALP13:%.*]] = getelementptr inbounds { float, float }, { float, float }* [[CF1]], i32 0, i32 0 1001 // CHECK1-NEXT: [[CF1_REAL14:%.*]] = load float, float* [[CF1_REALP13]], align 4 1002 // CHECK1-NEXT: [[CF1_IMAGP15:%.*]] = getelementptr inbounds { float, float }, { float, float }* [[CF1]], i32 0, i32 1 1003 // CHECK1-NEXT: [[CF1_IMAG16:%.*]] = load float, float* [[CF1_IMAGP15]], align 4 1004 // CHECK1-NEXT: [[ADD_R17:%.*]] = fadd float [[TMP_REAL]], [[CF1_REAL14]] 1005 // CHECK1-NEXT: [[ADD_I18:%.*]] = fadd float [[TMP_IMAG]], [[CF1_IMAG16]] 1006 // CHECK1-NEXT: [[ATOMIC_TEMP10_REALP:%.*]] = getelementptr inbounds { float, float }, { float, float }* [[ATOMIC_TEMP10]], i32 0, i32 0 1007 // CHECK1-NEXT: [[ATOMIC_TEMP10_IMAGP:%.*]] = getelementptr inbounds { float, float }, { float, float }* [[ATOMIC_TEMP10]], i32 0, i32 1 1008 // CHECK1-NEXT: store float [[ADD_R17]], float* [[ATOMIC_TEMP10_REALP]], align 4 1009 // CHECK1-NEXT: store float [[ADD_I18]], float* [[ATOMIC_TEMP10_IMAGP]], align 4 1010 // CHECK1-NEXT: [[TMP9:%.*]] = bitcast { float, float }* [[TMP0]] to i8* 1011 // CHECK1-NEXT: [[TMP10:%.*]] = bitcast { float, float }* [[ATOMIC_TEMP]] to i8* 1012 // CHECK1-NEXT: [[TMP11:%.*]] = bitcast { float, float }* [[ATOMIC_TEMP10]] to i8* 1013 // CHECK1-NEXT: [[CALL:%.*]] = call zeroext i1 @__atomic_compare_exchange(i64 8, i8* [[TMP9]], i8* [[TMP10]], i8* [[TMP11]], i32 0, i32 0) 1014 // CHECK1-NEXT: br i1 [[CALL]], label [[ATOMIC_EXIT:%.*]], label [[ATOMIC_CONT]] 1015 // CHECK1: atomic_exit: 1016 // CHECK1-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]] 1017 // CHECK1: .omp.reduction.default: 1018 // CHECK1-NEXT: ret void 1019 // 1020 // 1021 // CHECK1-LABEL: define {{[^@]+}}@.omp.reduction.reduction_func.5 1022 // CHECK1-SAME: (i8* [[TMP0:%.*]], i8* [[TMP1:%.*]]) #[[ATTR3]] { 1023 // CHECK1-NEXT: entry: 1024 // CHECK1-NEXT: [[DOTADDR:%.*]] = alloca i8*, align 8 1025 // CHECK1-NEXT: [[DOTADDR1:%.*]] = alloca i8*, align 8 1026 // CHECK1-NEXT: store i8* [[TMP0]], i8** [[DOTADDR]], align 8 1027 // CHECK1-NEXT: store i8* [[TMP1]], i8** [[DOTADDR1]], align 8 1028 // CHECK1-NEXT: [[TMP2:%.*]] = load i8*, i8** [[DOTADDR]], align 8 1029 // CHECK1-NEXT: [[TMP3:%.*]] = bitcast i8* [[TMP2]] to [1 x i8*]* 1030 // CHECK1-NEXT: [[TMP4:%.*]] = load i8*, i8** [[DOTADDR1]], align 8 1031 // CHECK1-NEXT: [[TMP5:%.*]] = bitcast i8* [[TMP4]] to [1 x i8*]* 1032 // CHECK1-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[TMP5]], i64 0, i64 0 1033 // CHECK1-NEXT: [[TMP7:%.*]] = load i8*, i8** [[TMP6]], align 8 1034 // CHECK1-NEXT: [[TMP8:%.*]] = bitcast i8* [[TMP7]] to { float, float }* 1035 // CHECK1-NEXT: [[TMP9:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[TMP3]], i64 0, i64 0 1036 // CHECK1-NEXT: [[TMP10:%.*]] = load i8*, i8** [[TMP9]], align 8 1037 // CHECK1-NEXT: [[TMP11:%.*]] = bitcast i8* [[TMP10]] to { float, float }* 1038 // CHECK1-NEXT: [[DOTREALP:%.*]] = getelementptr inbounds { float, float }, { float, float }* [[TMP11]], i32 0, i32 0 1039 // CHECK1-NEXT: [[DOTREAL:%.*]] = load float, float* [[DOTREALP]], align 4 1040 // CHECK1-NEXT: [[DOTIMAGP:%.*]] = getelementptr inbounds { float, float }, { float, float }* [[TMP11]], i32 0, i32 1 1041 // CHECK1-NEXT: [[DOTIMAG:%.*]] = load float, float* [[DOTIMAGP]], align 4 1042 // CHECK1-NEXT: [[DOTREALP2:%.*]] = getelementptr inbounds { float, float }, { float, float }* [[TMP8]], i32 0, i32 0 1043 // CHECK1-NEXT: [[DOTREAL3:%.*]] = load float, float* [[DOTREALP2]], align 4 1044 // CHECK1-NEXT: [[DOTIMAGP4:%.*]] = getelementptr inbounds { float, float }, { float, float }* [[TMP8]], i32 0, i32 1 1045 // CHECK1-NEXT: [[DOTIMAG5:%.*]] = load float, float* [[DOTIMAGP4]], align 4 1046 // CHECK1-NEXT: [[ADD_R:%.*]] = fadd float [[DOTREAL]], [[DOTREAL3]] 1047 // CHECK1-NEXT: [[ADD_I:%.*]] = fadd float [[DOTIMAG]], [[DOTIMAG5]] 1048 // CHECK1-NEXT: [[DOTREALP6:%.*]] = getelementptr inbounds { float, float }, { float, float }* [[TMP11]], i32 0, i32 0 1049 // CHECK1-NEXT: [[DOTIMAGP7:%.*]] = getelementptr inbounds { float, float }, { float, float }* [[TMP11]], i32 0, i32 1 1050 // CHECK1-NEXT: store float [[ADD_R]], float* [[DOTREALP6]], align 4 1051 // CHECK1-NEXT: store float [[ADD_I]], float* [[DOTIMAGP7]], align 4 1052 // CHECK1-NEXT: ret void 1053 // 1054 // 1055 // CHECK1-LABEL: define {{[^@]+}}@_Z5tmainIiET_v 1056 // CHECK1-SAME: () #[[ATTR0]] { 1057 // CHECK1-NEXT: entry: 1058 // CHECK1-NEXT: [[RETVAL:%.*]] = alloca i32, align 4 1059 // CHECK1-NEXT: [[T:%.*]] = alloca i32, align 4 1060 // CHECK1-NEXT: [[TEST:%.*]] = alloca [[STRUCT_S_0:%.*]], align 4 1061 // CHECK1-NEXT: [[SST:%.*]] = alloca [[STRUCT_SST:%.*]], align 4 1062 // CHECK1-NEXT: [[T_VAR:%.*]] = alloca i32, align 128 1063 // CHECK1-NEXT: [[T_VAR1:%.*]] = alloca i32, align 128 1064 // CHECK1-NEXT: [[VEC:%.*]] = alloca [2 x i32], align 4 1065 // CHECK1-NEXT: [[S_ARR:%.*]] = alloca [2 x %struct.S.0], align 4 1066 // CHECK1-NEXT: [[VAR:%.*]] = alloca [[STRUCT_S_0]], align 128 1067 // CHECK1-NEXT: [[VAR1:%.*]] = alloca [[STRUCT_S_0]], align 128 1068 // CHECK1-NEXT: call void @_ZN1SIiEC1Ev(%struct.S.0* nonnull align 4 dereferenceable(4) [[TEST]]) 1069 // CHECK1-NEXT: call void @_ZN3SSTIiEC1Ev(%struct.SST* nonnull align 4 dereferenceable(4) [[SST]]) 1070 // CHECK1-NEXT: store i32 0, i32* [[T_VAR]], align 128 1071 // CHECK1-NEXT: [[TMP0:%.*]] = bitcast [2 x i32]* [[VEC]] to i8* 1072 // CHECK1-NEXT: call void @llvm.memcpy.p0i8.p0i8.i64(i8* align 4 [[TMP0]], i8* align 4 bitcast ([2 x i32]* @__const._Z5tmainIiET_v.vec to i8*), i64 8, i1 false) 1073 // CHECK1-NEXT: [[ARRAYINIT_BEGIN:%.*]] = getelementptr inbounds [2 x %struct.S.0], [2 x %struct.S.0]* [[S_ARR]], i64 0, i64 0 1074 // CHECK1-NEXT: call void @_ZN1SIiEC1Ei(%struct.S.0* nonnull align 4 dereferenceable(4) [[ARRAYINIT_BEGIN]], i32 1) 1075 // CHECK1-NEXT: [[ARRAYINIT_ELEMENT:%.*]] = getelementptr inbounds [[STRUCT_S_0]], %struct.S.0* [[ARRAYINIT_BEGIN]], i64 1 1076 // CHECK1-NEXT: call void @_ZN1SIiEC1Ei(%struct.S.0* nonnull align 4 dereferenceable(4) [[ARRAYINIT_ELEMENT]], i32 2) 1077 // CHECK1-NEXT: call void @_ZN1SIiEC1Ei(%struct.S.0* nonnull align 4 dereferenceable(4) [[VAR]], i32 3) 1078 // CHECK1-NEXT: call void @_ZN1SIiEC1Ev(%struct.S.0* nonnull align 4 dereferenceable(4) [[VAR1]]) 1079 // CHECK1-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB2]], i32 6, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, [2 x i32]*, i32*, [2 x %struct.S.0]*, %struct.S.0*, %struct.S.0*, i32*)* @.omp_outlined..8 to void (i32*, i32*, ...)*), [2 x i32]* [[VEC]], i32* [[T_VAR]], [2 x %struct.S.0]* [[S_ARR]], %struct.S.0* [[VAR]], %struct.S.0* [[VAR1]], i32* [[T_VAR1]]) 1080 // CHECK1-NEXT: store i32 0, i32* [[RETVAL]], align 4 1081 // CHECK1-NEXT: call void @_ZN1SIiED1Ev(%struct.S.0* nonnull align 4 dereferenceable(4) [[VAR1]]) #[[ATTR5]] 1082 // CHECK1-NEXT: call void @_ZN1SIiED1Ev(%struct.S.0* nonnull align 4 dereferenceable(4) [[VAR]]) #[[ATTR5]] 1083 // CHECK1-NEXT: [[ARRAY_BEGIN:%.*]] = getelementptr inbounds [2 x %struct.S.0], [2 x %struct.S.0]* [[S_ARR]], i32 0, i32 0 1084 // CHECK1-NEXT: [[TMP1:%.*]] = getelementptr inbounds [[STRUCT_S_0]], %struct.S.0* [[ARRAY_BEGIN]], i64 2 1085 // CHECK1-NEXT: br label [[ARRAYDESTROY_BODY:%.*]] 1086 // CHECK1: arraydestroy.body: 1087 // CHECK1-NEXT: [[ARRAYDESTROY_ELEMENTPAST:%.*]] = phi %struct.S.0* [ [[TMP1]], [[ENTRY:%.*]] ], [ [[ARRAYDESTROY_ELEMENT:%.*]], [[ARRAYDESTROY_BODY]] ] 1088 // CHECK1-NEXT: [[ARRAYDESTROY_ELEMENT]] = getelementptr inbounds [[STRUCT_S_0]], %struct.S.0* [[ARRAYDESTROY_ELEMENTPAST]], i64 -1 1089 // CHECK1-NEXT: call void @_ZN1SIiED1Ev(%struct.S.0* nonnull align 4 dereferenceable(4) [[ARRAYDESTROY_ELEMENT]]) #[[ATTR5]] 1090 // CHECK1-NEXT: [[ARRAYDESTROY_DONE:%.*]] = icmp eq %struct.S.0* [[ARRAYDESTROY_ELEMENT]], [[ARRAY_BEGIN]] 1091 // CHECK1-NEXT: br i1 [[ARRAYDESTROY_DONE]], label [[ARRAYDESTROY_DONE1:%.*]], label [[ARRAYDESTROY_BODY]] 1092 // CHECK1: arraydestroy.done1: 1093 // CHECK1-NEXT: call void @_ZN1SIiED1Ev(%struct.S.0* nonnull align 4 dereferenceable(4) [[TEST]]) #[[ATTR5]] 1094 // CHECK1-NEXT: [[TMP2:%.*]] = load i32, i32* [[RETVAL]], align 4 1095 // CHECK1-NEXT: ret i32 [[TMP2]] 1096 // 1097 // 1098 // CHECK1-LABEL: define {{[^@]+}}@_ZN2SSC2ERi 1099 // CHECK1-SAME: (%struct.SS* nonnull align 8 dereferenceable(16) [[THIS:%.*]], i32* nonnull align 4 dereferenceable(4) [[D:%.*]]) unnamed_addr #[[ATTR7]] align 2 { 1100 // CHECK1-NEXT: entry: 1101 // CHECK1-NEXT: [[THIS_ADDR:%.*]] = alloca %struct.SS*, align 8 1102 // CHECK1-NEXT: [[D_ADDR:%.*]] = alloca i32*, align 8 1103 // CHECK1-NEXT: [[A2:%.*]] = alloca i32*, align 8 1104 // CHECK1-NEXT: [[B4:%.*]] = alloca i32, align 4 1105 // CHECK1-NEXT: [[C5:%.*]] = alloca i32*, align 8 1106 // CHECK1-NEXT: store %struct.SS* [[THIS]], %struct.SS** [[THIS_ADDR]], align 8 1107 // CHECK1-NEXT: store i32* [[D]], i32** [[D_ADDR]], align 8 1108 // CHECK1-NEXT: [[THIS1:%.*]] = load %struct.SS*, %struct.SS** [[THIS_ADDR]], align 8 1109 // CHECK1-NEXT: [[A:%.*]] = getelementptr inbounds [[STRUCT_SS:%.*]], %struct.SS* [[THIS1]], i32 0, i32 0 1110 // CHECK1-NEXT: store i32 0, i32* [[A]], align 8 1111 // CHECK1-NEXT: [[B:%.*]] = getelementptr inbounds [[STRUCT_SS]], %struct.SS* [[THIS1]], i32 0, i32 1 1112 // CHECK1-NEXT: [[BF_LOAD:%.*]] = load i8, i8* [[B]], align 4 1113 // CHECK1-NEXT: [[BF_CLEAR:%.*]] = and i8 [[BF_LOAD]], -16 1114 // CHECK1-NEXT: store i8 [[BF_CLEAR]], i8* [[B]], align 4 1115 // CHECK1-NEXT: [[C:%.*]] = getelementptr inbounds [[STRUCT_SS]], %struct.SS* [[THIS1]], i32 0, i32 2 1116 // CHECK1-NEXT: [[TMP0:%.*]] = load i32*, i32** [[D_ADDR]], align 8 1117 // CHECK1-NEXT: store i32* [[TMP0]], i32** [[C]], align 8 1118 // CHECK1-NEXT: [[A3:%.*]] = getelementptr inbounds [[STRUCT_SS]], %struct.SS* [[THIS1]], i32 0, i32 0 1119 // CHECK1-NEXT: store i32* [[A3]], i32** [[A2]], align 8 1120 // CHECK1-NEXT: [[C6:%.*]] = getelementptr inbounds [[STRUCT_SS]], %struct.SS* [[THIS1]], i32 0, i32 2 1121 // CHECK1-NEXT: [[TMP1:%.*]] = load i32*, i32** [[C6]], align 8 1122 // CHECK1-NEXT: store i32* [[TMP1]], i32** [[C5]], align 8 1123 // CHECK1-NEXT: [[TMP2:%.*]] = load i32*, i32** [[A2]], align 8 1124 // CHECK1-NEXT: [[TMP3:%.*]] = load i32*, i32** [[C5]], align 8 1125 // CHECK1-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB2]], i32 4, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, %struct.SS*, i32*, i32*, i32*)* @.omp_outlined..6 to void (i32*, i32*, ...)*), %struct.SS* [[THIS1]], i32* [[TMP2]], i32* [[B4]], i32* [[TMP3]]) 1126 // CHECK1-NEXT: [[TMP4:%.*]] = load i32, i32* [[B4]], align 4 1127 // CHECK1-NEXT: [[B7:%.*]] = getelementptr inbounds [[STRUCT_SS]], %struct.SS* [[THIS1]], i32 0, i32 1 1128 // CHECK1-NEXT: [[TMP5:%.*]] = trunc i32 [[TMP4]] to i8 1129 // CHECK1-NEXT: [[BF_LOAD8:%.*]] = load i8, i8* [[B7]], align 4 1130 // CHECK1-NEXT: [[BF_VALUE:%.*]] = and i8 [[TMP5]], 15 1131 // CHECK1-NEXT: [[BF_CLEAR9:%.*]] = and i8 [[BF_LOAD8]], -16 1132 // CHECK1-NEXT: [[BF_SET:%.*]] = or i8 [[BF_CLEAR9]], [[BF_VALUE]] 1133 // CHECK1-NEXT: store i8 [[BF_SET]], i8* [[B7]], align 4 1134 // CHECK1-NEXT: ret void 1135 // 1136 // 1137 // CHECK1-LABEL: define {{[^@]+}}@.omp_outlined..6 1138 // CHECK1-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], %struct.SS* [[THIS:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]], i32* nonnull align 4 dereferenceable(4) [[B:%.*]], i32* nonnull align 4 dereferenceable(4) [[C:%.*]]) #[[ATTR1]] { 1139 // CHECK1-NEXT: entry: 1140 // CHECK1-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 1141 // CHECK1-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 1142 // CHECK1-NEXT: [[THIS_ADDR:%.*]] = alloca %struct.SS*, align 8 1143 // CHECK1-NEXT: [[A_ADDR:%.*]] = alloca i32*, align 8 1144 // CHECK1-NEXT: [[B_ADDR:%.*]] = alloca i32*, align 8 1145 // CHECK1-NEXT: [[C_ADDR:%.*]] = alloca i32*, align 8 1146 // CHECK1-NEXT: [[TMP:%.*]] = alloca i32*, align 8 1147 // CHECK1-NEXT: [[_TMP1:%.*]] = alloca i32*, align 8 1148 // CHECK1-NEXT: [[A2:%.*]] = alloca i32, align 4 1149 // CHECK1-NEXT: [[_TMP3:%.*]] = alloca i32*, align 8 1150 // CHECK1-NEXT: [[B4:%.*]] = alloca i32, align 4 1151 // CHECK1-NEXT: [[C5:%.*]] = alloca i32, align 4 1152 // CHECK1-NEXT: [[_TMP6:%.*]] = alloca i32*, align 8 1153 // CHECK1-NEXT: [[DOTOMP_REDUCTION_RED_LIST:%.*]] = alloca [3 x i8*], align 8 1154 // CHECK1-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8 1155 // CHECK1-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8 1156 // CHECK1-NEXT: store %struct.SS* [[THIS]], %struct.SS** [[THIS_ADDR]], align 8 1157 // CHECK1-NEXT: store i32* [[A]], i32** [[A_ADDR]], align 8 1158 // CHECK1-NEXT: store i32* [[B]], i32** [[B_ADDR]], align 8 1159 // CHECK1-NEXT: store i32* [[C]], i32** [[C_ADDR]], align 8 1160 // CHECK1-NEXT: [[TMP0:%.*]] = load %struct.SS*, %struct.SS** [[THIS_ADDR]], align 8 1161 // CHECK1-NEXT: [[TMP1:%.*]] = load i32*, i32** [[A_ADDR]], align 8 1162 // CHECK1-NEXT: [[TMP2:%.*]] = load i32*, i32** [[B_ADDR]], align 8 1163 // CHECK1-NEXT: [[TMP3:%.*]] = load i32*, i32** [[C_ADDR]], align 8 1164 // CHECK1-NEXT: store i32* [[TMP1]], i32** [[TMP]], align 8 1165 // CHECK1-NEXT: store i32* [[TMP3]], i32** [[_TMP1]], align 8 1166 // CHECK1-NEXT: [[TMP4:%.*]] = load i32*, i32** [[TMP]], align 8 1167 // CHECK1-NEXT: store i32 0, i32* [[A2]], align 4 1168 // CHECK1-NEXT: store i32* [[A2]], i32** [[_TMP3]], align 8 1169 // CHECK1-NEXT: store i32 0, i32* [[B4]], align 4 1170 // CHECK1-NEXT: [[TMP5:%.*]] = load i32*, i32** [[_TMP1]], align 8 1171 // CHECK1-NEXT: store i32 0, i32* [[C5]], align 4 1172 // CHECK1-NEXT: store i32* [[C5]], i32** [[_TMP6]], align 8 1173 // CHECK1-NEXT: [[TMP6:%.*]] = load i32*, i32** [[_TMP3]], align 8 1174 // CHECK1-NEXT: [[TMP7:%.*]] = load i32, i32* [[TMP6]], align 4 1175 // CHECK1-NEXT: [[INC:%.*]] = add nsw i32 [[TMP7]], 1 1176 // CHECK1-NEXT: store i32 [[INC]], i32* [[TMP6]], align 4 1177 // CHECK1-NEXT: [[TMP8:%.*]] = load i32, i32* [[B4]], align 4 1178 // CHECK1-NEXT: [[DEC:%.*]] = add nsw i32 [[TMP8]], -1 1179 // CHECK1-NEXT: store i32 [[DEC]], i32* [[B4]], align 4 1180 // CHECK1-NEXT: [[TMP9:%.*]] = load i32*, i32** [[_TMP6]], align 8 1181 // CHECK1-NEXT: [[TMP10:%.*]] = load i32, i32* [[TMP9]], align 4 1182 // CHECK1-NEXT: [[DIV:%.*]] = sdiv i32 [[TMP10]], 1 1183 // CHECK1-NEXT: store i32 [[DIV]], i32* [[TMP9]], align 4 1184 // CHECK1-NEXT: [[TMP11:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOMP_REDUCTION_RED_LIST]], i64 0, i64 0 1185 // CHECK1-NEXT: [[TMP12:%.*]] = bitcast i32* [[A2]] to i8* 1186 // CHECK1-NEXT: store i8* [[TMP12]], i8** [[TMP11]], align 8 1187 // CHECK1-NEXT: [[TMP13:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOMP_REDUCTION_RED_LIST]], i64 0, i64 1 1188 // CHECK1-NEXT: [[TMP14:%.*]] = bitcast i32* [[B4]] to i8* 1189 // CHECK1-NEXT: store i8* [[TMP14]], i8** [[TMP13]], align 8 1190 // CHECK1-NEXT: [[TMP15:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOMP_REDUCTION_RED_LIST]], i64 0, i64 2 1191 // CHECK1-NEXT: [[TMP16:%.*]] = bitcast i32* [[C5]] to i8* 1192 // CHECK1-NEXT: store i8* [[TMP16]], i8** [[TMP15]], align 8 1193 // CHECK1-NEXT: [[TMP17:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 1194 // CHECK1-NEXT: [[TMP18:%.*]] = load i32, i32* [[TMP17]], align 4 1195 // CHECK1-NEXT: [[TMP19:%.*]] = bitcast [3 x i8*]* [[DOTOMP_REDUCTION_RED_LIST]] to i8* 1196 // CHECK1-NEXT: [[TMP20:%.*]] = call i32 @__kmpc_reduce_nowait(%struct.ident_t* @[[GLOB1]], i32 [[TMP18]], i32 3, i64 24, i8* [[TMP19]], void (i8*, i8*)* @.omp.reduction.reduction_func.7, [8 x i32]* @.gomp_critical_user_.reduction.var) 1197 // CHECK1-NEXT: switch i32 [[TMP20]], label [[DOTOMP_REDUCTION_DEFAULT:%.*]] [ 1198 // CHECK1-NEXT: i32 1, label [[DOTOMP_REDUCTION_CASE1:%.*]] 1199 // CHECK1-NEXT: i32 2, label [[DOTOMP_REDUCTION_CASE2:%.*]] 1200 // CHECK1-NEXT: ] 1201 // CHECK1: .omp.reduction.case1: 1202 // CHECK1-NEXT: [[TMP21:%.*]] = load i32, i32* [[TMP4]], align 4 1203 // CHECK1-NEXT: [[TMP22:%.*]] = load i32, i32* [[A2]], align 4 1204 // CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP21]], [[TMP22]] 1205 // CHECK1-NEXT: store i32 [[ADD]], i32* [[TMP4]], align 4 1206 // CHECK1-NEXT: [[TMP23:%.*]] = load i32, i32* [[TMP2]], align 4 1207 // CHECK1-NEXT: [[TMP24:%.*]] = load i32, i32* [[B4]], align 4 1208 // CHECK1-NEXT: [[ADD7:%.*]] = add nsw i32 [[TMP23]], [[TMP24]] 1209 // CHECK1-NEXT: store i32 [[ADD7]], i32* [[TMP2]], align 4 1210 // CHECK1-NEXT: [[TMP25:%.*]] = load i32, i32* [[TMP5]], align 4 1211 // CHECK1-NEXT: [[TMP26:%.*]] = load i32, i32* [[C5]], align 4 1212 // CHECK1-NEXT: [[ADD8:%.*]] = add nsw i32 [[TMP25]], [[TMP26]] 1213 // CHECK1-NEXT: store i32 [[ADD8]], i32* [[TMP5]], align 4 1214 // CHECK1-NEXT: call void @__kmpc_end_reduce_nowait(%struct.ident_t* @[[GLOB1]], i32 [[TMP18]], [8 x i32]* @.gomp_critical_user_.reduction.var) 1215 // CHECK1-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]] 1216 // CHECK1: .omp.reduction.case2: 1217 // CHECK1-NEXT: [[TMP27:%.*]] = load i32, i32* [[A2]], align 4 1218 // CHECK1-NEXT: [[TMP28:%.*]] = atomicrmw add i32* [[TMP4]], i32 [[TMP27]] monotonic, align 4 1219 // CHECK1-NEXT: [[TMP29:%.*]] = load i32, i32* [[B4]], align 4 1220 // CHECK1-NEXT: [[TMP30:%.*]] = atomicrmw add i32* [[TMP2]], i32 [[TMP29]] monotonic, align 4 1221 // CHECK1-NEXT: [[TMP31:%.*]] = load i32, i32* [[C5]], align 4 1222 // CHECK1-NEXT: [[TMP32:%.*]] = atomicrmw add i32* [[TMP5]], i32 [[TMP31]] monotonic, align 4 1223 // CHECK1-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]] 1224 // CHECK1: .omp.reduction.default: 1225 // CHECK1-NEXT: ret void 1226 // 1227 // 1228 // CHECK1-LABEL: define {{[^@]+}}@.omp.reduction.reduction_func.7 1229 // CHECK1-SAME: (i8* [[TMP0:%.*]], i8* [[TMP1:%.*]]) #[[ATTR3]] { 1230 // CHECK1-NEXT: entry: 1231 // CHECK1-NEXT: [[DOTADDR:%.*]] = alloca i8*, align 8 1232 // CHECK1-NEXT: [[DOTADDR1:%.*]] = alloca i8*, align 8 1233 // CHECK1-NEXT: store i8* [[TMP0]], i8** [[DOTADDR]], align 8 1234 // CHECK1-NEXT: store i8* [[TMP1]], i8** [[DOTADDR1]], align 8 1235 // CHECK1-NEXT: [[TMP2:%.*]] = load i8*, i8** [[DOTADDR]], align 8 1236 // CHECK1-NEXT: [[TMP3:%.*]] = bitcast i8* [[TMP2]] to [3 x i8*]* 1237 // CHECK1-NEXT: [[TMP4:%.*]] = load i8*, i8** [[DOTADDR1]], align 8 1238 // CHECK1-NEXT: [[TMP5:%.*]] = bitcast i8* [[TMP4]] to [3 x i8*]* 1239 // CHECK1-NEXT: [[TMP6:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[TMP5]], i64 0, i64 0 1240 // CHECK1-NEXT: [[TMP7:%.*]] = load i8*, i8** [[TMP6]], align 8 1241 // CHECK1-NEXT: [[TMP8:%.*]] = bitcast i8* [[TMP7]] to i32* 1242 // CHECK1-NEXT: [[TMP9:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[TMP3]], i64 0, i64 0 1243 // CHECK1-NEXT: [[TMP10:%.*]] = load i8*, i8** [[TMP9]], align 8 1244 // CHECK1-NEXT: [[TMP11:%.*]] = bitcast i8* [[TMP10]] to i32* 1245 // CHECK1-NEXT: [[TMP12:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[TMP5]], i64 0, i64 1 1246 // CHECK1-NEXT: [[TMP13:%.*]] = load i8*, i8** [[TMP12]], align 8 1247 // CHECK1-NEXT: [[TMP14:%.*]] = bitcast i8* [[TMP13]] to i32* 1248 // CHECK1-NEXT: [[TMP15:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[TMP3]], i64 0, i64 1 1249 // CHECK1-NEXT: [[TMP16:%.*]] = load i8*, i8** [[TMP15]], align 8 1250 // CHECK1-NEXT: [[TMP17:%.*]] = bitcast i8* [[TMP16]] to i32* 1251 // CHECK1-NEXT: [[TMP18:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[TMP5]], i64 0, i64 2 1252 // CHECK1-NEXT: [[TMP19:%.*]] = load i8*, i8** [[TMP18]], align 8 1253 // CHECK1-NEXT: [[TMP20:%.*]] = bitcast i8* [[TMP19]] to i32* 1254 // CHECK1-NEXT: [[TMP21:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[TMP3]], i64 0, i64 2 1255 // CHECK1-NEXT: [[TMP22:%.*]] = load i8*, i8** [[TMP21]], align 8 1256 // CHECK1-NEXT: [[TMP23:%.*]] = bitcast i8* [[TMP22]] to i32* 1257 // CHECK1-NEXT: [[TMP24:%.*]] = load i32, i32* [[TMP11]], align 4 1258 // CHECK1-NEXT: [[TMP25:%.*]] = load i32, i32* [[TMP8]], align 4 1259 // CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP24]], [[TMP25]] 1260 // CHECK1-NEXT: store i32 [[ADD]], i32* [[TMP11]], align 4 1261 // CHECK1-NEXT: [[TMP26:%.*]] = load i32, i32* [[TMP17]], align 4 1262 // CHECK1-NEXT: [[TMP27:%.*]] = load i32, i32* [[TMP14]], align 4 1263 // CHECK1-NEXT: [[ADD2:%.*]] = add nsw i32 [[TMP26]], [[TMP27]] 1264 // CHECK1-NEXT: store i32 [[ADD2]], i32* [[TMP17]], align 4 1265 // CHECK1-NEXT: [[TMP28:%.*]] = load i32, i32* [[TMP23]], align 4 1266 // CHECK1-NEXT: [[TMP29:%.*]] = load i32, i32* [[TMP20]], align 4 1267 // CHECK1-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP28]], [[TMP29]] 1268 // CHECK1-NEXT: store i32 [[ADD3]], i32* [[TMP23]], align 4 1269 // CHECK1-NEXT: ret void 1270 // 1271 // 1272 // CHECK1-LABEL: define {{[^@]+}}@_ZN1SIfEC2Ev 1273 // CHECK1-SAME: (%struct.S* nonnull align 4 dereferenceable(4) [[THIS:%.*]]) unnamed_addr #[[ATTR7]] align 2 { 1274 // CHECK1-NEXT: entry: 1275 // CHECK1-NEXT: [[THIS_ADDR:%.*]] = alloca %struct.S*, align 8 1276 // CHECK1-NEXT: store %struct.S* [[THIS]], %struct.S** [[THIS_ADDR]], align 8 1277 // CHECK1-NEXT: [[THIS1:%.*]] = load %struct.S*, %struct.S** [[THIS_ADDR]], align 8 1278 // CHECK1-NEXT: [[F:%.*]] = getelementptr inbounds [[STRUCT_S:%.*]], %struct.S* [[THIS1]], i32 0, i32 0 1279 // CHECK1-NEXT: [[TMP0:%.*]] = load volatile i32, i32* @g, align 128 1280 // CHECK1-NEXT: [[CONV:%.*]] = sitofp i32 [[TMP0]] to float 1281 // CHECK1-NEXT: store float [[CONV]], float* [[F]], align 4 1282 // CHECK1-NEXT: ret void 1283 // 1284 // 1285 // CHECK1-LABEL: define {{[^@]+}}@_ZN1SIfEC2Ef 1286 // CHECK1-SAME: (%struct.S* nonnull align 4 dereferenceable(4) [[THIS:%.*]], float [[A:%.*]]) unnamed_addr #[[ATTR7]] align 2 { 1287 // CHECK1-NEXT: entry: 1288 // CHECK1-NEXT: [[THIS_ADDR:%.*]] = alloca %struct.S*, align 8 1289 // CHECK1-NEXT: [[A_ADDR:%.*]] = alloca float, align 4 1290 // CHECK1-NEXT: store %struct.S* [[THIS]], %struct.S** [[THIS_ADDR]], align 8 1291 // CHECK1-NEXT: store float [[A]], float* [[A_ADDR]], align 4 1292 // CHECK1-NEXT: [[THIS1:%.*]] = load %struct.S*, %struct.S** [[THIS_ADDR]], align 8 1293 // CHECK1-NEXT: [[F:%.*]] = getelementptr inbounds [[STRUCT_S:%.*]], %struct.S* [[THIS1]], i32 0, i32 0 1294 // CHECK1-NEXT: [[TMP0:%.*]] = load float, float* [[A_ADDR]], align 4 1295 // CHECK1-NEXT: [[TMP1:%.*]] = load volatile i32, i32* @g, align 128 1296 // CHECK1-NEXT: [[CONV:%.*]] = sitofp i32 [[TMP1]] to float 1297 // CHECK1-NEXT: [[ADD:%.*]] = fadd float [[TMP0]], [[CONV]] 1298 // CHECK1-NEXT: store float [[ADD]], float* [[F]], align 4 1299 // CHECK1-NEXT: ret void 1300 // 1301 // 1302 // CHECK1-LABEL: define {{[^@]+}}@_ZN1SIfED2Ev 1303 // CHECK1-SAME: (%struct.S* nonnull align 4 dereferenceable(4) [[THIS:%.*]]) unnamed_addr #[[ATTR7]] align 2 { 1304 // CHECK1-NEXT: entry: 1305 // CHECK1-NEXT: [[THIS_ADDR:%.*]] = alloca %struct.S*, align 8 1306 // CHECK1-NEXT: store %struct.S* [[THIS]], %struct.S** [[THIS_ADDR]], align 8 1307 // CHECK1-NEXT: [[THIS1:%.*]] = load %struct.S*, %struct.S** [[THIS_ADDR]], align 8 1308 // CHECK1-NEXT: ret void 1309 // 1310 // 1311 // CHECK1-LABEL: define {{[^@]+}}@_ZN1SIiEC1Ev 1312 // CHECK1-SAME: (%struct.S.0* nonnull align 4 dereferenceable(4) [[THIS:%.*]]) unnamed_addr #[[ATTR7]] align 2 { 1313 // CHECK1-NEXT: entry: 1314 // CHECK1-NEXT: [[THIS_ADDR:%.*]] = alloca %struct.S.0*, align 8 1315 // CHECK1-NEXT: store %struct.S.0* [[THIS]], %struct.S.0** [[THIS_ADDR]], align 8 1316 // CHECK1-NEXT: [[THIS1:%.*]] = load %struct.S.0*, %struct.S.0** [[THIS_ADDR]], align 8 1317 // CHECK1-NEXT: call void @_ZN1SIiEC2Ev(%struct.S.0* nonnull align 4 dereferenceable(4) [[THIS1]]) 1318 // CHECK1-NEXT: ret void 1319 // 1320 // 1321 // CHECK1-LABEL: define {{[^@]+}}@_ZN3SSTIiEC1Ev 1322 // CHECK1-SAME: (%struct.SST* nonnull align 4 dereferenceable(4) [[THIS:%.*]]) unnamed_addr #[[ATTR7]] align 2 { 1323 // CHECK1-NEXT: entry: 1324 // CHECK1-NEXT: [[THIS_ADDR:%.*]] = alloca %struct.SST*, align 8 1325 // CHECK1-NEXT: store %struct.SST* [[THIS]], %struct.SST** [[THIS_ADDR]], align 8 1326 // CHECK1-NEXT: [[THIS1:%.*]] = load %struct.SST*, %struct.SST** [[THIS_ADDR]], align 8 1327 // CHECK1-NEXT: call void @_ZN3SSTIiEC2Ev(%struct.SST* nonnull align 4 dereferenceable(4) [[THIS1]]) 1328 // CHECK1-NEXT: ret void 1329 // 1330 // 1331 // CHECK1-LABEL: define {{[^@]+}}@_ZN1SIiEC1Ei 1332 // CHECK1-SAME: (%struct.S.0* nonnull align 4 dereferenceable(4) [[THIS:%.*]], i32 [[A:%.*]]) unnamed_addr #[[ATTR7]] align 2 { 1333 // CHECK1-NEXT: entry: 1334 // CHECK1-NEXT: [[THIS_ADDR:%.*]] = alloca %struct.S.0*, align 8 1335 // CHECK1-NEXT: [[A_ADDR:%.*]] = alloca i32, align 4 1336 // CHECK1-NEXT: store %struct.S.0* [[THIS]], %struct.S.0** [[THIS_ADDR]], align 8 1337 // CHECK1-NEXT: store i32 [[A]], i32* [[A_ADDR]], align 4 1338 // CHECK1-NEXT: [[THIS1:%.*]] = load %struct.S.0*, %struct.S.0** [[THIS_ADDR]], align 8 1339 // CHECK1-NEXT: [[TMP0:%.*]] = load i32, i32* [[A_ADDR]], align 4 1340 // CHECK1-NEXT: call void @_ZN1SIiEC2Ei(%struct.S.0* nonnull align 4 dereferenceable(4) [[THIS1]], i32 [[TMP0]]) 1341 // CHECK1-NEXT: ret void 1342 // 1343 // 1344 // CHECK1-LABEL: define {{[^@]+}}@.omp_outlined..8 1345 // CHECK1-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], [2 x i32]* nonnull align 4 dereferenceable(8) [[VEC:%.*]], i32* nonnull align 4 dereferenceable(4) [[T_VAR:%.*]], [2 x %struct.S.0]* nonnull align 4 dereferenceable(8) [[S_ARR:%.*]], %struct.S.0* nonnull align 4 dereferenceable(4) [[VAR:%.*]], %struct.S.0* nonnull align 4 dereferenceable(4) [[VAR1:%.*]], i32* nonnull align 4 dereferenceable(4) [[T_VAR1:%.*]]) #[[ATTR1]] { 1346 // CHECK1-NEXT: entry: 1347 // CHECK1-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 1348 // CHECK1-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 1349 // CHECK1-NEXT: [[VEC_ADDR:%.*]] = alloca [2 x i32]*, align 8 1350 // CHECK1-NEXT: [[T_VAR_ADDR:%.*]] = alloca i32*, align 8 1351 // CHECK1-NEXT: [[S_ARR_ADDR:%.*]] = alloca [2 x %struct.S.0]*, align 8 1352 // CHECK1-NEXT: [[VAR_ADDR:%.*]] = alloca %struct.S.0*, align 8 1353 // CHECK1-NEXT: [[VAR1_ADDR:%.*]] = alloca %struct.S.0*, align 8 1354 // CHECK1-NEXT: [[T_VAR1_ADDR:%.*]] = alloca i32*, align 8 1355 // CHECK1-NEXT: [[T_VAR2:%.*]] = alloca i32, align 128 1356 // CHECK1-NEXT: [[VAR3:%.*]] = alloca [[STRUCT_S_0:%.*]], align 128 1357 // CHECK1-NEXT: [[VAR14:%.*]] = alloca [[STRUCT_S_0]], align 128 1358 // CHECK1-NEXT: [[T_VAR15:%.*]] = alloca i32, align 128 1359 // CHECK1-NEXT: [[DOTOMP_REDUCTION_RED_LIST:%.*]] = alloca [4 x i8*], align 8 1360 // CHECK1-NEXT: [[REF_TMP:%.*]] = alloca [[STRUCT_S_0]], align 4 1361 // CHECK1-NEXT: [[REF_TMP11:%.*]] = alloca [[STRUCT_S_0]], align 4 1362 // CHECK1-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8 1363 // CHECK1-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8 1364 // CHECK1-NEXT: store [2 x i32]* [[VEC]], [2 x i32]** [[VEC_ADDR]], align 8 1365 // CHECK1-NEXT: store i32* [[T_VAR]], i32** [[T_VAR_ADDR]], align 8 1366 // CHECK1-NEXT: store [2 x %struct.S.0]* [[S_ARR]], [2 x %struct.S.0]** [[S_ARR_ADDR]], align 8 1367 // CHECK1-NEXT: store %struct.S.0* [[VAR]], %struct.S.0** [[VAR_ADDR]], align 8 1368 // CHECK1-NEXT: store %struct.S.0* [[VAR1]], %struct.S.0** [[VAR1_ADDR]], align 8 1369 // CHECK1-NEXT: store i32* [[T_VAR1]], i32** [[T_VAR1_ADDR]], align 8 1370 // CHECK1-NEXT: [[TMP0:%.*]] = load [2 x i32]*, [2 x i32]** [[VEC_ADDR]], align 8 1371 // CHECK1-NEXT: [[TMP1:%.*]] = load i32*, i32** [[T_VAR_ADDR]], align 8 1372 // CHECK1-NEXT: [[TMP2:%.*]] = load [2 x %struct.S.0]*, [2 x %struct.S.0]** [[S_ARR_ADDR]], align 8 1373 // CHECK1-NEXT: [[TMP3:%.*]] = load %struct.S.0*, %struct.S.0** [[VAR_ADDR]], align 8 1374 // CHECK1-NEXT: [[TMP4:%.*]] = load %struct.S.0*, %struct.S.0** [[VAR1_ADDR]], align 8 1375 // CHECK1-NEXT: [[TMP5:%.*]] = load i32*, i32** [[T_VAR1_ADDR]], align 8 1376 // CHECK1-NEXT: store i32 0, i32* [[T_VAR2]], align 128 1377 // CHECK1-NEXT: call void @_ZN1SIiEC1Ev(%struct.S.0* nonnull align 4 dereferenceable(4) [[VAR3]]) 1378 // CHECK1-NEXT: call void @_ZN1SIiEC1Ev(%struct.S.0* nonnull align 4 dereferenceable(4) [[VAR14]]) 1379 // CHECK1-NEXT: store i32 2147483647, i32* [[T_VAR15]], align 128 1380 // CHECK1-NEXT: [[TMP6:%.*]] = load i32, i32* [[T_VAR2]], align 128 1381 // CHECK1-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [2 x i32], [2 x i32]* [[TMP0]], i64 0, i64 0 1382 // CHECK1-NEXT: store i32 [[TMP6]], i32* [[ARRAYIDX]], align 4 1383 // CHECK1-NEXT: [[ARRAYIDX6:%.*]] = getelementptr inbounds [2 x %struct.S.0], [2 x %struct.S.0]* [[TMP2]], i64 0, i64 0 1384 // CHECK1-NEXT: [[TMP7:%.*]] = bitcast %struct.S.0* [[ARRAYIDX6]] to i8* 1385 // CHECK1-NEXT: [[TMP8:%.*]] = bitcast %struct.S.0* [[VAR3]] to i8* 1386 // CHECK1-NEXT: call void @llvm.memcpy.p0i8.p0i8.i64(i8* align 4 [[TMP7]], i8* align 128 [[TMP8]], i64 4, i1 false) 1387 // CHECK1-NEXT: [[TMP9:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOMP_REDUCTION_RED_LIST]], i64 0, i64 0 1388 // CHECK1-NEXT: [[TMP10:%.*]] = bitcast i32* [[T_VAR2]] to i8* 1389 // CHECK1-NEXT: store i8* [[TMP10]], i8** [[TMP9]], align 8 1390 // CHECK1-NEXT: [[TMP11:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOMP_REDUCTION_RED_LIST]], i64 0, i64 1 1391 // CHECK1-NEXT: [[TMP12:%.*]] = bitcast %struct.S.0* [[VAR3]] to i8* 1392 // CHECK1-NEXT: store i8* [[TMP12]], i8** [[TMP11]], align 8 1393 // CHECK1-NEXT: [[TMP13:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOMP_REDUCTION_RED_LIST]], i64 0, i64 2 1394 // CHECK1-NEXT: [[TMP14:%.*]] = bitcast %struct.S.0* [[VAR14]] to i8* 1395 // CHECK1-NEXT: store i8* [[TMP14]], i8** [[TMP13]], align 8 1396 // CHECK1-NEXT: [[TMP15:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOMP_REDUCTION_RED_LIST]], i64 0, i64 3 1397 // CHECK1-NEXT: [[TMP16:%.*]] = bitcast i32* [[T_VAR15]] to i8* 1398 // CHECK1-NEXT: store i8* [[TMP16]], i8** [[TMP15]], align 8 1399 // CHECK1-NEXT: [[TMP17:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 1400 // CHECK1-NEXT: [[TMP18:%.*]] = load i32, i32* [[TMP17]], align 4 1401 // CHECK1-NEXT: [[TMP19:%.*]] = bitcast [4 x i8*]* [[DOTOMP_REDUCTION_RED_LIST]] to i8* 1402 // CHECK1-NEXT: [[TMP20:%.*]] = call i32 @__kmpc_reduce_nowait(%struct.ident_t* @[[GLOB1]], i32 [[TMP18]], i32 4, i64 32, i8* [[TMP19]], void (i8*, i8*)* @.omp.reduction.reduction_func.9, [8 x i32]* @.gomp_critical_user_.reduction.var) 1403 // CHECK1-NEXT: switch i32 [[TMP20]], label [[DOTOMP_REDUCTION_DEFAULT:%.*]] [ 1404 // CHECK1-NEXT: i32 1, label [[DOTOMP_REDUCTION_CASE1:%.*]] 1405 // CHECK1-NEXT: i32 2, label [[DOTOMP_REDUCTION_CASE2:%.*]] 1406 // CHECK1-NEXT: ] 1407 // CHECK1: .omp.reduction.case1: 1408 // CHECK1-NEXT: [[TMP21:%.*]] = load i32, i32* [[TMP1]], align 128 1409 // CHECK1-NEXT: [[TMP22:%.*]] = load i32, i32* [[T_VAR2]], align 128 1410 // CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP21]], [[TMP22]] 1411 // CHECK1-NEXT: store i32 [[ADD]], i32* [[TMP1]], align 128 1412 // CHECK1-NEXT: [[CALL:%.*]] = call nonnull align 4 dereferenceable(4) %struct.S.0* @_ZN1SIiEanERKS0_(%struct.S.0* nonnull align 4 dereferenceable(4) [[TMP3]], %struct.S.0* nonnull align 4 dereferenceable(4) [[VAR3]]) 1413 // CHECK1-NEXT: [[TMP23:%.*]] = bitcast %struct.S.0* [[TMP3]] to i8* 1414 // CHECK1-NEXT: [[TMP24:%.*]] = bitcast %struct.S.0* [[CALL]] to i8* 1415 // CHECK1-NEXT: call void @llvm.memcpy.p0i8.p0i8.i64(i8* align 128 [[TMP23]], i8* align 4 [[TMP24]], i64 4, i1 false) 1416 // CHECK1-NEXT: [[CALL7:%.*]] = call i32 @_ZN1SIiEcviEv(%struct.S.0* nonnull align 4 dereferenceable(4) [[TMP4]]) 1417 // CHECK1-NEXT: [[TOBOOL:%.*]] = icmp ne i32 [[CALL7]], 0 1418 // CHECK1-NEXT: br i1 [[TOBOOL]], label [[LAND_RHS:%.*]], label [[LAND_END:%.*]] 1419 // CHECK1: land.rhs: 1420 // CHECK1-NEXT: [[CALL8:%.*]] = call i32 @_ZN1SIiEcviEv(%struct.S.0* nonnull align 4 dereferenceable(4) [[VAR14]]) 1421 // CHECK1-NEXT: [[TOBOOL9:%.*]] = icmp ne i32 [[CALL8]], 0 1422 // CHECK1-NEXT: br label [[LAND_END]] 1423 // CHECK1: land.end: 1424 // CHECK1-NEXT: [[TMP25:%.*]] = phi i1 [ false, [[DOTOMP_REDUCTION_CASE1]] ], [ [[TOBOOL9]], [[LAND_RHS]] ] 1425 // CHECK1-NEXT: [[CONV:%.*]] = zext i1 [[TMP25]] to i32 1426 // CHECK1-NEXT: call void @_ZN1SIiEC1Ei(%struct.S.0* nonnull align 4 dereferenceable(4) [[REF_TMP]], i32 [[CONV]]) 1427 // CHECK1-NEXT: [[TMP26:%.*]] = bitcast %struct.S.0* [[TMP4]] to i8* 1428 // CHECK1-NEXT: [[TMP27:%.*]] = bitcast %struct.S.0* [[REF_TMP]] to i8* 1429 // CHECK1-NEXT: call void @llvm.memcpy.p0i8.p0i8.i64(i8* align 128 [[TMP26]], i8* align 4 [[TMP27]], i64 4, i1 false) 1430 // CHECK1-NEXT: call void @_ZN1SIiED1Ev(%struct.S.0* nonnull align 4 dereferenceable(4) [[REF_TMP]]) #[[ATTR5]] 1431 // CHECK1-NEXT: [[TMP28:%.*]] = load i32, i32* [[TMP5]], align 128 1432 // CHECK1-NEXT: [[TMP29:%.*]] = load i32, i32* [[T_VAR15]], align 128 1433 // CHECK1-NEXT: [[CMP:%.*]] = icmp slt i32 [[TMP28]], [[TMP29]] 1434 // CHECK1-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] 1435 // CHECK1: cond.true: 1436 // CHECK1-NEXT: [[TMP30:%.*]] = load i32, i32* [[TMP5]], align 128 1437 // CHECK1-NEXT: br label [[COND_END:%.*]] 1438 // CHECK1: cond.false: 1439 // CHECK1-NEXT: [[TMP31:%.*]] = load i32, i32* [[T_VAR15]], align 128 1440 // CHECK1-NEXT: br label [[COND_END]] 1441 // CHECK1: cond.end: 1442 // CHECK1-NEXT: [[COND:%.*]] = phi i32 [ [[TMP30]], [[COND_TRUE]] ], [ [[TMP31]], [[COND_FALSE]] ] 1443 // CHECK1-NEXT: store i32 [[COND]], i32* [[TMP5]], align 128 1444 // CHECK1-NEXT: call void @__kmpc_end_reduce_nowait(%struct.ident_t* @[[GLOB1]], i32 [[TMP18]], [8 x i32]* @.gomp_critical_user_.reduction.var) 1445 // CHECK1-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]] 1446 // CHECK1: .omp.reduction.case2: 1447 // CHECK1-NEXT: [[TMP32:%.*]] = load i32, i32* [[T_VAR2]], align 128 1448 // CHECK1-NEXT: [[TMP33:%.*]] = atomicrmw add i32* [[TMP1]], i32 [[TMP32]] monotonic, align 4 1449 // CHECK1-NEXT: call void @__kmpc_critical(%struct.ident_t* @[[GLOB2]], i32 [[TMP18]], [8 x i32]* @.gomp_critical_user_.atomic_reduction.var) 1450 // CHECK1-NEXT: [[CALL10:%.*]] = call nonnull align 4 dereferenceable(4) %struct.S.0* @_ZN1SIiEanERKS0_(%struct.S.0* nonnull align 4 dereferenceable(4) [[TMP3]], %struct.S.0* nonnull align 4 dereferenceable(4) [[VAR3]]) 1451 // CHECK1-NEXT: [[TMP34:%.*]] = bitcast %struct.S.0* [[TMP3]] to i8* 1452 // CHECK1-NEXT: [[TMP35:%.*]] = bitcast %struct.S.0* [[CALL10]] to i8* 1453 // CHECK1-NEXT: call void @llvm.memcpy.p0i8.p0i8.i64(i8* align 128 [[TMP34]], i8* align 4 [[TMP35]], i64 4, i1 false) 1454 // CHECK1-NEXT: call void @__kmpc_end_critical(%struct.ident_t* @[[GLOB2]], i32 [[TMP18]], [8 x i32]* @.gomp_critical_user_.atomic_reduction.var) 1455 // CHECK1-NEXT: call void @__kmpc_critical(%struct.ident_t* @[[GLOB2]], i32 [[TMP18]], [8 x i32]* @.gomp_critical_user_.atomic_reduction.var) 1456 // CHECK1-NEXT: [[CALL12:%.*]] = call i32 @_ZN1SIiEcviEv(%struct.S.0* nonnull align 4 dereferenceable(4) [[TMP4]]) 1457 // CHECK1-NEXT: [[TOBOOL13:%.*]] = icmp ne i32 [[CALL12]], 0 1458 // CHECK1-NEXT: br i1 [[TOBOOL13]], label [[LAND_RHS14:%.*]], label [[LAND_END17:%.*]] 1459 // CHECK1: land.rhs14: 1460 // CHECK1-NEXT: [[CALL15:%.*]] = call i32 @_ZN1SIiEcviEv(%struct.S.0* nonnull align 4 dereferenceable(4) [[VAR14]]) 1461 // CHECK1-NEXT: [[TOBOOL16:%.*]] = icmp ne i32 [[CALL15]], 0 1462 // CHECK1-NEXT: br label [[LAND_END17]] 1463 // CHECK1: land.end17: 1464 // CHECK1-NEXT: [[TMP36:%.*]] = phi i1 [ false, [[DOTOMP_REDUCTION_CASE2]] ], [ [[TOBOOL16]], [[LAND_RHS14]] ] 1465 // CHECK1-NEXT: [[CONV18:%.*]] = zext i1 [[TMP36]] to i32 1466 // CHECK1-NEXT: call void @_ZN1SIiEC1Ei(%struct.S.0* nonnull align 4 dereferenceable(4) [[REF_TMP11]], i32 [[CONV18]]) 1467 // CHECK1-NEXT: [[TMP37:%.*]] = bitcast %struct.S.0* [[TMP4]] to i8* 1468 // CHECK1-NEXT: [[TMP38:%.*]] = bitcast %struct.S.0* [[REF_TMP11]] to i8* 1469 // CHECK1-NEXT: call void @llvm.memcpy.p0i8.p0i8.i64(i8* align 128 [[TMP37]], i8* align 4 [[TMP38]], i64 4, i1 false) 1470 // CHECK1-NEXT: call void @_ZN1SIiED1Ev(%struct.S.0* nonnull align 4 dereferenceable(4) [[REF_TMP11]]) #[[ATTR5]] 1471 // CHECK1-NEXT: call void @__kmpc_end_critical(%struct.ident_t* @[[GLOB2]], i32 [[TMP18]], [8 x i32]* @.gomp_critical_user_.atomic_reduction.var) 1472 // CHECK1-NEXT: [[TMP39:%.*]] = load i32, i32* [[T_VAR15]], align 128 1473 // CHECK1-NEXT: [[TMP40:%.*]] = atomicrmw min i32* [[TMP5]], i32 [[TMP39]] monotonic, align 4 1474 // CHECK1-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]] 1475 // CHECK1: .omp.reduction.default: 1476 // CHECK1-NEXT: call void @_ZN1SIiED1Ev(%struct.S.0* nonnull align 4 dereferenceable(4) [[VAR14]]) #[[ATTR5]] 1477 // CHECK1-NEXT: call void @_ZN1SIiED1Ev(%struct.S.0* nonnull align 4 dereferenceable(4) [[VAR3]]) #[[ATTR5]] 1478 // CHECK1-NEXT: ret void 1479 // 1480 // 1481 // CHECK1-LABEL: define {{[^@]+}}@.omp.reduction.reduction_func.9 1482 // CHECK1-SAME: (i8* [[TMP0:%.*]], i8* [[TMP1:%.*]]) #[[ATTR3]] { 1483 // CHECK1-NEXT: entry: 1484 // CHECK1-NEXT: [[DOTADDR:%.*]] = alloca i8*, align 8 1485 // CHECK1-NEXT: [[DOTADDR1:%.*]] = alloca i8*, align 8 1486 // CHECK1-NEXT: [[REF_TMP:%.*]] = alloca [[STRUCT_S_0:%.*]], align 4 1487 // CHECK1-NEXT: store i8* [[TMP0]], i8** [[DOTADDR]], align 8 1488 // CHECK1-NEXT: store i8* [[TMP1]], i8** [[DOTADDR1]], align 8 1489 // CHECK1-NEXT: [[TMP2:%.*]] = load i8*, i8** [[DOTADDR]], align 8 1490 // CHECK1-NEXT: [[TMP3:%.*]] = bitcast i8* [[TMP2]] to [4 x i8*]* 1491 // CHECK1-NEXT: [[TMP4:%.*]] = load i8*, i8** [[DOTADDR1]], align 8 1492 // CHECK1-NEXT: [[TMP5:%.*]] = bitcast i8* [[TMP4]] to [4 x i8*]* 1493 // CHECK1-NEXT: [[TMP6:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[TMP5]], i64 0, i64 0 1494 // CHECK1-NEXT: [[TMP7:%.*]] = load i8*, i8** [[TMP6]], align 8 1495 // CHECK1-NEXT: [[TMP8:%.*]] = bitcast i8* [[TMP7]] to i32* 1496 // CHECK1-NEXT: [[TMP9:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[TMP3]], i64 0, i64 0 1497 // CHECK1-NEXT: [[TMP10:%.*]] = load i8*, i8** [[TMP9]], align 8 1498 // CHECK1-NEXT: [[TMP11:%.*]] = bitcast i8* [[TMP10]] to i32* 1499 // CHECK1-NEXT: [[TMP12:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[TMP5]], i64 0, i64 1 1500 // CHECK1-NEXT: [[TMP13:%.*]] = load i8*, i8** [[TMP12]], align 8 1501 // CHECK1-NEXT: [[TMP14:%.*]] = bitcast i8* [[TMP13]] to %struct.S.0* 1502 // CHECK1-NEXT: [[TMP15:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[TMP3]], i64 0, i64 1 1503 // CHECK1-NEXT: [[TMP16:%.*]] = load i8*, i8** [[TMP15]], align 8 1504 // CHECK1-NEXT: [[TMP17:%.*]] = bitcast i8* [[TMP16]] to %struct.S.0* 1505 // CHECK1-NEXT: [[TMP18:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[TMP5]], i64 0, i64 2 1506 // CHECK1-NEXT: [[TMP19:%.*]] = load i8*, i8** [[TMP18]], align 8 1507 // CHECK1-NEXT: [[TMP20:%.*]] = bitcast i8* [[TMP19]] to %struct.S.0* 1508 // CHECK1-NEXT: [[TMP21:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[TMP3]], i64 0, i64 2 1509 // CHECK1-NEXT: [[TMP22:%.*]] = load i8*, i8** [[TMP21]], align 8 1510 // CHECK1-NEXT: [[TMP23:%.*]] = bitcast i8* [[TMP22]] to %struct.S.0* 1511 // CHECK1-NEXT: [[TMP24:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[TMP5]], i64 0, i64 3 1512 // CHECK1-NEXT: [[TMP25:%.*]] = load i8*, i8** [[TMP24]], align 8 1513 // CHECK1-NEXT: [[TMP26:%.*]] = bitcast i8* [[TMP25]] to i32* 1514 // CHECK1-NEXT: [[TMP27:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[TMP3]], i64 0, i64 3 1515 // CHECK1-NEXT: [[TMP28:%.*]] = load i8*, i8** [[TMP27]], align 8 1516 // CHECK1-NEXT: [[TMP29:%.*]] = bitcast i8* [[TMP28]] to i32* 1517 // CHECK1-NEXT: [[TMP30:%.*]] = load i32, i32* [[TMP11]], align 128 1518 // CHECK1-NEXT: [[TMP31:%.*]] = load i32, i32* [[TMP8]], align 128 1519 // CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP30]], [[TMP31]] 1520 // CHECK1-NEXT: store i32 [[ADD]], i32* [[TMP11]], align 128 1521 // CHECK1-NEXT: [[CALL:%.*]] = call nonnull align 4 dereferenceable(4) %struct.S.0* @_ZN1SIiEanERKS0_(%struct.S.0* nonnull align 4 dereferenceable(4) [[TMP17]], %struct.S.0* nonnull align 4 dereferenceable(4) [[TMP14]]) 1522 // CHECK1-NEXT: [[TMP32:%.*]] = bitcast %struct.S.0* [[TMP17]] to i8* 1523 // CHECK1-NEXT: [[TMP33:%.*]] = bitcast %struct.S.0* [[CALL]] to i8* 1524 // CHECK1-NEXT: call void @llvm.memcpy.p0i8.p0i8.i64(i8* align 128 [[TMP32]], i8* align 4 [[TMP33]], i64 4, i1 false) 1525 // CHECK1-NEXT: [[CALL2:%.*]] = call i32 @_ZN1SIiEcviEv(%struct.S.0* nonnull align 4 dereferenceable(4) [[TMP23]]) 1526 // CHECK1-NEXT: [[TOBOOL:%.*]] = icmp ne i32 [[CALL2]], 0 1527 // CHECK1-NEXT: br i1 [[TOBOOL]], label [[LAND_RHS:%.*]], label [[LAND_END:%.*]] 1528 // CHECK1: land.rhs: 1529 // CHECK1-NEXT: [[CALL3:%.*]] = call i32 @_ZN1SIiEcviEv(%struct.S.0* nonnull align 4 dereferenceable(4) [[TMP20]]) 1530 // CHECK1-NEXT: [[TOBOOL4:%.*]] = icmp ne i32 [[CALL3]], 0 1531 // CHECK1-NEXT: br label [[LAND_END]] 1532 // CHECK1: land.end: 1533 // CHECK1-NEXT: [[TMP34:%.*]] = phi i1 [ false, [[ENTRY:%.*]] ], [ [[TOBOOL4]], [[LAND_RHS]] ] 1534 // CHECK1-NEXT: [[CONV:%.*]] = zext i1 [[TMP34]] to i32 1535 // CHECK1-NEXT: call void @_ZN1SIiEC1Ei(%struct.S.0* nonnull align 4 dereferenceable(4) [[REF_TMP]], i32 [[CONV]]) 1536 // CHECK1-NEXT: [[TMP35:%.*]] = bitcast %struct.S.0* [[TMP23]] to i8* 1537 // CHECK1-NEXT: [[TMP36:%.*]] = bitcast %struct.S.0* [[REF_TMP]] to i8* 1538 // CHECK1-NEXT: call void @llvm.memcpy.p0i8.p0i8.i64(i8* align 128 [[TMP35]], i8* align 4 [[TMP36]], i64 4, i1 false) 1539 // CHECK1-NEXT: call void @_ZN1SIiED1Ev(%struct.S.0* nonnull align 4 dereferenceable(4) [[REF_TMP]]) #[[ATTR5]] 1540 // CHECK1-NEXT: [[TMP37:%.*]] = load i32, i32* [[TMP29]], align 128 1541 // CHECK1-NEXT: [[TMP38:%.*]] = load i32, i32* [[TMP26]], align 128 1542 // CHECK1-NEXT: [[CMP:%.*]] = icmp slt i32 [[TMP37]], [[TMP38]] 1543 // CHECK1-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] 1544 // CHECK1: cond.true: 1545 // CHECK1-NEXT: [[TMP39:%.*]] = load i32, i32* [[TMP29]], align 128 1546 // CHECK1-NEXT: br label [[COND_END:%.*]] 1547 // CHECK1: cond.false: 1548 // CHECK1-NEXT: [[TMP40:%.*]] = load i32, i32* [[TMP26]], align 128 1549 // CHECK1-NEXT: br label [[COND_END]] 1550 // CHECK1: cond.end: 1551 // CHECK1-NEXT: [[COND:%.*]] = phi i32 [ [[TMP39]], [[COND_TRUE]] ], [ [[TMP40]], [[COND_FALSE]] ] 1552 // CHECK1-NEXT: store i32 [[COND]], i32* [[TMP29]], align 128 1553 // CHECK1-NEXT: ret void 1554 // 1555 // 1556 // CHECK1-LABEL: define {{[^@]+}}@_ZN1SIiEanERKS0_ 1557 // CHECK1-SAME: (%struct.S.0* nonnull align 4 dereferenceable(4) [[THIS:%.*]], %struct.S.0* nonnull align 4 dereferenceable(4) [[TMP0:%.*]]) #[[ATTR0]] align 2 { 1558 // CHECK1-NEXT: entry: 1559 // CHECK1-NEXT: [[THIS_ADDR:%.*]] = alloca %struct.S.0*, align 8 1560 // CHECK1-NEXT: [[DOTADDR:%.*]] = alloca %struct.S.0*, align 8 1561 // CHECK1-NEXT: store %struct.S.0* [[THIS]], %struct.S.0** [[THIS_ADDR]], align 8 1562 // CHECK1-NEXT: store %struct.S.0* [[TMP0]], %struct.S.0** [[DOTADDR]], align 8 1563 // CHECK1-NEXT: [[THIS1:%.*]] = load %struct.S.0*, %struct.S.0** [[THIS_ADDR]], align 8 1564 // CHECK1-NEXT: ret %struct.S.0* [[THIS1]] 1565 // 1566 // 1567 // CHECK1-LABEL: define {{[^@]+}}@_ZN1SIiEcviEv 1568 // CHECK1-SAME: (%struct.S.0* nonnull align 4 dereferenceable(4) [[THIS:%.*]]) #[[ATTR0]] align 2 { 1569 // CHECK1-NEXT: entry: 1570 // CHECK1-NEXT: [[THIS_ADDR:%.*]] = alloca %struct.S.0*, align 8 1571 // CHECK1-NEXT: store %struct.S.0* [[THIS]], %struct.S.0** [[THIS_ADDR]], align 8 1572 // CHECK1-NEXT: [[THIS1:%.*]] = load %struct.S.0*, %struct.S.0** [[THIS_ADDR]], align 8 1573 // CHECK1-NEXT: ret i32 0 1574 // 1575 // 1576 // CHECK1-LABEL: define {{[^@]+}}@_ZN1SIiED1Ev 1577 // CHECK1-SAME: (%struct.S.0* nonnull align 4 dereferenceable(4) [[THIS:%.*]]) unnamed_addr #[[ATTR7]] align 2 { 1578 // CHECK1-NEXT: entry: 1579 // CHECK1-NEXT: [[THIS_ADDR:%.*]] = alloca %struct.S.0*, align 8 1580 // CHECK1-NEXT: store %struct.S.0* [[THIS]], %struct.S.0** [[THIS_ADDR]], align 8 1581 // CHECK1-NEXT: [[THIS1:%.*]] = load %struct.S.0*, %struct.S.0** [[THIS_ADDR]], align 8 1582 // CHECK1-NEXT: call void @_ZN1SIiED2Ev(%struct.S.0* nonnull align 4 dereferenceable(4) [[THIS1]]) #[[ATTR5]] 1583 // CHECK1-NEXT: ret void 1584 // 1585 // 1586 // CHECK1-LABEL: define {{[^@]+}}@_ZN1SIiEC2Ev 1587 // CHECK1-SAME: (%struct.S.0* nonnull align 4 dereferenceable(4) [[THIS:%.*]]) unnamed_addr #[[ATTR7]] align 2 { 1588 // CHECK1-NEXT: entry: 1589 // CHECK1-NEXT: [[THIS_ADDR:%.*]] = alloca %struct.S.0*, align 8 1590 // CHECK1-NEXT: store %struct.S.0* [[THIS]], %struct.S.0** [[THIS_ADDR]], align 8 1591 // CHECK1-NEXT: [[THIS1:%.*]] = load %struct.S.0*, %struct.S.0** [[THIS_ADDR]], align 8 1592 // CHECK1-NEXT: [[F:%.*]] = getelementptr inbounds [[STRUCT_S_0:%.*]], %struct.S.0* [[THIS1]], i32 0, i32 0 1593 // CHECK1-NEXT: [[TMP0:%.*]] = load volatile i32, i32* @g, align 128 1594 // CHECK1-NEXT: store i32 [[TMP0]], i32* [[F]], align 4 1595 // CHECK1-NEXT: ret void 1596 // 1597 // 1598 // CHECK1-LABEL: define {{[^@]+}}@_ZN3SSTIiEC2Ev 1599 // CHECK1-SAME: (%struct.SST* nonnull align 4 dereferenceable(4) [[THIS:%.*]]) unnamed_addr #[[ATTR7]] align 2 { 1600 // CHECK1-NEXT: entry: 1601 // CHECK1-NEXT: [[THIS_ADDR:%.*]] = alloca %struct.SST*, align 8 1602 // CHECK1-NEXT: [[A2:%.*]] = alloca i32*, align 8 1603 // CHECK1-NEXT: store %struct.SST* [[THIS]], %struct.SST** [[THIS_ADDR]], align 8 1604 // CHECK1-NEXT: [[THIS1:%.*]] = load %struct.SST*, %struct.SST** [[THIS_ADDR]], align 8 1605 // CHECK1-NEXT: [[A:%.*]] = getelementptr inbounds [[STRUCT_SST:%.*]], %struct.SST* [[THIS1]], i32 0, i32 0 1606 // CHECK1-NEXT: store i32 0, i32* [[A]], align 4 1607 // CHECK1-NEXT: [[A3:%.*]] = getelementptr inbounds [[STRUCT_SST]], %struct.SST* [[THIS1]], i32 0, i32 0 1608 // CHECK1-NEXT: store i32* [[A3]], i32** [[A2]], align 8 1609 // CHECK1-NEXT: [[TMP0:%.*]] = load i32*, i32** [[A2]], align 8 1610 // CHECK1-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB2]], i32 2, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, %struct.SST*, i32*)* @.omp_outlined..10 to void (i32*, i32*, ...)*), %struct.SST* [[THIS1]], i32* [[TMP0]]) 1611 // CHECK1-NEXT: ret void 1612 // 1613 // 1614 // CHECK1-LABEL: define {{[^@]+}}@.omp_outlined..10 1615 // CHECK1-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], %struct.SST* [[THIS:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR1]] { 1616 // CHECK1-NEXT: entry: 1617 // CHECK1-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 1618 // CHECK1-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 1619 // CHECK1-NEXT: [[THIS_ADDR:%.*]] = alloca %struct.SST*, align 8 1620 // CHECK1-NEXT: [[A_ADDR:%.*]] = alloca i32*, align 8 1621 // CHECK1-NEXT: [[TMP:%.*]] = alloca i32*, align 8 1622 // CHECK1-NEXT: [[A1:%.*]] = alloca i32, align 4 1623 // CHECK1-NEXT: [[_TMP2:%.*]] = alloca i32*, align 8 1624 // CHECK1-NEXT: [[DOTOMP_REDUCTION_RED_LIST:%.*]] = alloca [1 x i8*], align 8 1625 // CHECK1-NEXT: [[ATOMIC_TEMP:%.*]] = alloca i32, align 4 1626 // CHECK1-NEXT: [[_TMP3:%.*]] = alloca i32, align 4 1627 // CHECK1-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8 1628 // CHECK1-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8 1629 // CHECK1-NEXT: store %struct.SST* [[THIS]], %struct.SST** [[THIS_ADDR]], align 8 1630 // CHECK1-NEXT: store i32* [[A]], i32** [[A_ADDR]], align 8 1631 // CHECK1-NEXT: [[TMP0:%.*]] = load %struct.SST*, %struct.SST** [[THIS_ADDR]], align 8 1632 // CHECK1-NEXT: [[TMP1:%.*]] = load i32*, i32** [[A_ADDR]], align 8 1633 // CHECK1-NEXT: store i32* [[TMP1]], i32** [[TMP]], align 8 1634 // CHECK1-NEXT: [[TMP2:%.*]] = load i32*, i32** [[TMP]], align 8 1635 // CHECK1-NEXT: store i32 1, i32* [[A1]], align 4 1636 // CHECK1-NEXT: store i32* [[A1]], i32** [[_TMP2]], align 8 1637 // CHECK1-NEXT: [[TMP3:%.*]] = load i32*, i32** [[_TMP2]], align 8 1638 // CHECK1-NEXT: [[TMP4:%.*]] = load i32, i32* [[TMP3]], align 4 1639 // CHECK1-NEXT: [[INC:%.*]] = add nsw i32 [[TMP4]], 1 1640 // CHECK1-NEXT: store i32 [[INC]], i32* [[TMP3]], align 4 1641 // CHECK1-NEXT: [[TMP5:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOMP_REDUCTION_RED_LIST]], i64 0, i64 0 1642 // CHECK1-NEXT: [[TMP6:%.*]] = bitcast i32* [[A1]] to i8* 1643 // CHECK1-NEXT: store i8* [[TMP6]], i8** [[TMP5]], align 8 1644 // CHECK1-NEXT: [[TMP7:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 1645 // CHECK1-NEXT: [[TMP8:%.*]] = load i32, i32* [[TMP7]], align 4 1646 // CHECK1-NEXT: [[TMP9:%.*]] = bitcast [1 x i8*]* [[DOTOMP_REDUCTION_RED_LIST]] to i8* 1647 // CHECK1-NEXT: [[TMP10:%.*]] = call i32 @__kmpc_reduce_nowait(%struct.ident_t* @[[GLOB1]], i32 [[TMP8]], i32 1, i64 8, i8* [[TMP9]], void (i8*, i8*)* @.omp.reduction.reduction_func.11, [8 x i32]* @.gomp_critical_user_.reduction.var) 1648 // CHECK1-NEXT: switch i32 [[TMP10]], label [[DOTOMP_REDUCTION_DEFAULT:%.*]] [ 1649 // CHECK1-NEXT: i32 1, label [[DOTOMP_REDUCTION_CASE1:%.*]] 1650 // CHECK1-NEXT: i32 2, label [[DOTOMP_REDUCTION_CASE2:%.*]] 1651 // CHECK1-NEXT: ] 1652 // CHECK1: .omp.reduction.case1: 1653 // CHECK1-NEXT: [[TMP11:%.*]] = load i32, i32* [[TMP2]], align 4 1654 // CHECK1-NEXT: [[TMP12:%.*]] = load i32, i32* [[A1]], align 4 1655 // CHECK1-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP11]], [[TMP12]] 1656 // CHECK1-NEXT: store i32 [[MUL]], i32* [[TMP2]], align 4 1657 // CHECK1-NEXT: call void @__kmpc_end_reduce_nowait(%struct.ident_t* @[[GLOB1]], i32 [[TMP8]], [8 x i32]* @.gomp_critical_user_.reduction.var) 1658 // CHECK1-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]] 1659 // CHECK1: .omp.reduction.case2: 1660 // CHECK1-NEXT: [[TMP13:%.*]] = load i32, i32* [[A1]], align 4 1661 // CHECK1-NEXT: [[ATOMIC_LOAD:%.*]] = load atomic i32, i32* [[TMP2]] monotonic, align 4 1662 // CHECK1-NEXT: br label [[ATOMIC_CONT:%.*]] 1663 // CHECK1: atomic_cont: 1664 // CHECK1-NEXT: [[TMP14:%.*]] = phi i32 [ [[ATOMIC_LOAD]], [[DOTOMP_REDUCTION_CASE2]] ], [ [[TMP19:%.*]], [[ATOMIC_CONT]] ] 1665 // CHECK1-NEXT: store i32 [[TMP14]], i32* [[_TMP3]], align 4 1666 // CHECK1-NEXT: [[TMP15:%.*]] = load i32, i32* [[_TMP3]], align 4 1667 // CHECK1-NEXT: [[TMP16:%.*]] = load i32, i32* [[A1]], align 4 1668 // CHECK1-NEXT: [[MUL4:%.*]] = mul nsw i32 [[TMP15]], [[TMP16]] 1669 // CHECK1-NEXT: store i32 [[MUL4]], i32* [[ATOMIC_TEMP]], align 4 1670 // CHECK1-NEXT: [[TMP17:%.*]] = load i32, i32* [[ATOMIC_TEMP]], align 4 1671 // CHECK1-NEXT: [[TMP18:%.*]] = cmpxchg i32* [[TMP2]], i32 [[TMP14]], i32 [[TMP17]] monotonic monotonic, align 4 1672 // CHECK1-NEXT: [[TMP19]] = extractvalue { i32, i1 } [[TMP18]], 0 1673 // CHECK1-NEXT: [[TMP20:%.*]] = extractvalue { i32, i1 } [[TMP18]], 1 1674 // CHECK1-NEXT: br i1 [[TMP20]], label [[ATOMIC_EXIT:%.*]], label [[ATOMIC_CONT]] 1675 // CHECK1: atomic_exit: 1676 // CHECK1-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]] 1677 // CHECK1: .omp.reduction.default: 1678 // CHECK1-NEXT: ret void 1679 // 1680 // 1681 // CHECK1-LABEL: define {{[^@]+}}@.omp.reduction.reduction_func.11 1682 // CHECK1-SAME: (i8* [[TMP0:%.*]], i8* [[TMP1:%.*]]) #[[ATTR3]] { 1683 // CHECK1-NEXT: entry: 1684 // CHECK1-NEXT: [[DOTADDR:%.*]] = alloca i8*, align 8 1685 // CHECK1-NEXT: [[DOTADDR1:%.*]] = alloca i8*, align 8 1686 // CHECK1-NEXT: store i8* [[TMP0]], i8** [[DOTADDR]], align 8 1687 // CHECK1-NEXT: store i8* [[TMP1]], i8** [[DOTADDR1]], align 8 1688 // CHECK1-NEXT: [[TMP2:%.*]] = load i8*, i8** [[DOTADDR]], align 8 1689 // CHECK1-NEXT: [[TMP3:%.*]] = bitcast i8* [[TMP2]] to [1 x i8*]* 1690 // CHECK1-NEXT: [[TMP4:%.*]] = load i8*, i8** [[DOTADDR1]], align 8 1691 // CHECK1-NEXT: [[TMP5:%.*]] = bitcast i8* [[TMP4]] to [1 x i8*]* 1692 // CHECK1-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[TMP5]], i64 0, i64 0 1693 // CHECK1-NEXT: [[TMP7:%.*]] = load i8*, i8** [[TMP6]], align 8 1694 // CHECK1-NEXT: [[TMP8:%.*]] = bitcast i8* [[TMP7]] to i32* 1695 // CHECK1-NEXT: [[TMP9:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[TMP3]], i64 0, i64 0 1696 // CHECK1-NEXT: [[TMP10:%.*]] = load i8*, i8** [[TMP9]], align 8 1697 // CHECK1-NEXT: [[TMP11:%.*]] = bitcast i8* [[TMP10]] to i32* 1698 // CHECK1-NEXT: [[TMP12:%.*]] = load i32, i32* [[TMP11]], align 4 1699 // CHECK1-NEXT: [[TMP13:%.*]] = load i32, i32* [[TMP8]], align 4 1700 // CHECK1-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP12]], [[TMP13]] 1701 // CHECK1-NEXT: store i32 [[MUL]], i32* [[TMP11]], align 4 1702 // CHECK1-NEXT: ret void 1703 // 1704 // 1705 // CHECK1-LABEL: define {{[^@]+}}@_ZN1SIiEC2Ei 1706 // CHECK1-SAME: (%struct.S.0* nonnull align 4 dereferenceable(4) [[THIS:%.*]], i32 [[A:%.*]]) unnamed_addr #[[ATTR7]] align 2 { 1707 // CHECK1-NEXT: entry: 1708 // CHECK1-NEXT: [[THIS_ADDR:%.*]] = alloca %struct.S.0*, align 8 1709 // CHECK1-NEXT: [[A_ADDR:%.*]] = alloca i32, align 4 1710 // CHECK1-NEXT: store %struct.S.0* [[THIS]], %struct.S.0** [[THIS_ADDR]], align 8 1711 // CHECK1-NEXT: store i32 [[A]], i32* [[A_ADDR]], align 4 1712 // CHECK1-NEXT: [[THIS1:%.*]] = load %struct.S.0*, %struct.S.0** [[THIS_ADDR]], align 8 1713 // CHECK1-NEXT: [[F:%.*]] = getelementptr inbounds [[STRUCT_S_0:%.*]], %struct.S.0* [[THIS1]], i32 0, i32 0 1714 // CHECK1-NEXT: [[TMP0:%.*]] = load i32, i32* [[A_ADDR]], align 4 1715 // CHECK1-NEXT: [[TMP1:%.*]] = load volatile i32, i32* @g, align 128 1716 // CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP0]], [[TMP1]] 1717 // CHECK1-NEXT: store i32 [[ADD]], i32* [[F]], align 4 1718 // CHECK1-NEXT: ret void 1719 // 1720 // 1721 // CHECK1-LABEL: define {{[^@]+}}@_ZN1SIiED2Ev 1722 // CHECK1-SAME: (%struct.S.0* nonnull align 4 dereferenceable(4) [[THIS:%.*]]) unnamed_addr #[[ATTR7]] align 2 { 1723 // CHECK1-NEXT: entry: 1724 // CHECK1-NEXT: [[THIS_ADDR:%.*]] = alloca %struct.S.0*, align 8 1725 // CHECK1-NEXT: store %struct.S.0* [[THIS]], %struct.S.0** [[THIS_ADDR]], align 8 1726 // CHECK1-NEXT: [[THIS1:%.*]] = load %struct.S.0*, %struct.S.0** [[THIS_ADDR]], align 8 1727 // CHECK1-NEXT: ret void 1728 // 1729 // 1730 // CHECK2-LABEL: define {{[^@]+}}@_Z14foo_array_sectPs 1731 // CHECK2-SAME: (i16* [[X:%.*]]) #[[ATTR0:[0-9]+]] { 1732 // CHECK2-NEXT: entry: 1733 // CHECK2-NEXT: [[X_ADDR:%.*]] = alloca i16*, align 8 1734 // CHECK2-NEXT: store i16* [[X]], i16** [[X_ADDR]], align 8 1735 // CHECK2-NEXT: [[TMP0:%.*]] = load i16*, i16** [[X_ADDR]], align 8 1736 // CHECK2-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB2:[0-9]+]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i16*)* @.omp_outlined. to void (i32*, i32*, ...)*), i16* [[TMP0]]) 1737 // CHECK2-NEXT: ret void 1738 // 1739 // 1740 // CHECK2-LABEL: define {{[^@]+}}@.omp_outlined. 1741 // CHECK2-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i16* [[X:%.*]]) #[[ATTR1:[0-9]+]] { 1742 // CHECK2-NEXT: entry: 1743 // CHECK2-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 1744 // CHECK2-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 1745 // CHECK2-NEXT: [[X_ADDR:%.*]] = alloca i16*, align 8 1746 // CHECK2-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 8 1747 // CHECK2-NEXT: [[__VLA_EXPR0:%.*]] = alloca i64, align 8 1748 // CHECK2-NEXT: [[TMP:%.*]] = alloca i16*, align 8 1749 // CHECK2-NEXT: [[DOTOMP_REDUCTION_RED_LIST:%.*]] = alloca [2 x i8*], align 8 1750 // CHECK2-NEXT: [[ATOMIC_TEMP:%.*]] = alloca i16, align 2 1751 // CHECK2-NEXT: [[_TMP13:%.*]] = alloca i16, align 2 1752 // CHECK2-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8 1753 // CHECK2-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8 1754 // CHECK2-NEXT: store i16* [[X]], i16** [[X_ADDR]], align 8 1755 // CHECK2-NEXT: [[TMP0:%.*]] = load i16*, i16** [[X_ADDR]], align 8 1756 // CHECK2-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i16, i16* [[TMP0]], i64 0 1757 // CHECK2-NEXT: [[TMP1:%.*]] = load i16*, i16** [[X_ADDR]], align 8 1758 // CHECK2-NEXT: [[ARRAYIDX1:%.*]] = getelementptr inbounds i16, i16* [[TMP1]], i64 0 1759 // CHECK2-NEXT: [[TMP2:%.*]] = ptrtoint i16* [[ARRAYIDX1]] to i64 1760 // CHECK2-NEXT: [[TMP3:%.*]] = ptrtoint i16* [[ARRAYIDX]] to i64 1761 // CHECK2-NEXT: [[TMP4:%.*]] = sub i64 [[TMP2]], [[TMP3]] 1762 // CHECK2-NEXT: [[TMP5:%.*]] = sdiv exact i64 [[TMP4]], ptrtoint (i16* getelementptr (i16, i16* null, i32 1) to i64) 1763 // CHECK2-NEXT: [[TMP6:%.*]] = add nuw i64 [[TMP5]], 1 1764 // CHECK2-NEXT: [[TMP7:%.*]] = mul nuw i64 [[TMP6]], ptrtoint (i16* getelementptr (i16, i16* null, i32 1) to i64) 1765 // CHECK2-NEXT: [[TMP8:%.*]] = call i8* @llvm.stacksave() 1766 // CHECK2-NEXT: store i8* [[TMP8]], i8** [[SAVED_STACK]], align 8 1767 // CHECK2-NEXT: [[VLA:%.*]] = alloca i16, i64 [[TMP6]], align 16 1768 // CHECK2-NEXT: store i64 [[TMP6]], i64* [[__VLA_EXPR0]], align 8 1769 // CHECK2-NEXT: [[TMP9:%.*]] = getelementptr i16, i16* [[VLA]], i64 [[TMP6]] 1770 // CHECK2-NEXT: [[OMP_ARRAYINIT_ISEMPTY:%.*]] = icmp eq i16* [[VLA]], [[TMP9]] 1771 // CHECK2-NEXT: br i1 [[OMP_ARRAYINIT_ISEMPTY]], label [[OMP_ARRAYINIT_DONE:%.*]], label [[OMP_ARRAYINIT_BODY:%.*]] 1772 // CHECK2: omp.arrayinit.body: 1773 // CHECK2-NEXT: [[OMP_ARRAYCPY_DESTELEMENTPAST:%.*]] = phi i16* [ [[VLA]], [[ENTRY:%.*]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT:%.*]], [[OMP_ARRAYINIT_BODY]] ] 1774 // CHECK2-NEXT: store i16 0, i16* [[OMP_ARRAYCPY_DESTELEMENTPAST]], align 2 1775 // CHECK2-NEXT: [[OMP_ARRAYCPY_DEST_ELEMENT]] = getelementptr i16, i16* [[OMP_ARRAYCPY_DESTELEMENTPAST]], i32 1 1776 // CHECK2-NEXT: [[OMP_ARRAYCPY_DONE:%.*]] = icmp eq i16* [[OMP_ARRAYCPY_DEST_ELEMENT]], [[TMP9]] 1777 // CHECK2-NEXT: br i1 [[OMP_ARRAYCPY_DONE]], label [[OMP_ARRAYINIT_DONE]], label [[OMP_ARRAYINIT_BODY]] 1778 // CHECK2: omp.arrayinit.done: 1779 // CHECK2-NEXT: [[TMP10:%.*]] = load i16*, i16** [[X_ADDR]], align 8 1780 // CHECK2-NEXT: [[TMP11:%.*]] = ptrtoint i16* [[TMP10]] to i64 1781 // CHECK2-NEXT: [[TMP12:%.*]] = ptrtoint i16* [[ARRAYIDX]] to i64 1782 // CHECK2-NEXT: [[TMP13:%.*]] = sub i64 [[TMP11]], [[TMP12]] 1783 // CHECK2-NEXT: [[TMP14:%.*]] = sdiv exact i64 [[TMP13]], ptrtoint (i16* getelementptr (i16, i16* null, i32 1) to i64) 1784 // CHECK2-NEXT: [[TMP15:%.*]] = getelementptr i16, i16* [[VLA]], i64 [[TMP14]] 1785 // CHECK2-NEXT: store i16* [[TMP15]], i16** [[TMP]], align 8 1786 // CHECK2-NEXT: [[TMP16:%.*]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[DOTOMP_REDUCTION_RED_LIST]], i64 0, i64 0 1787 // CHECK2-NEXT: [[TMP17:%.*]] = bitcast i16* [[VLA]] to i8* 1788 // CHECK2-NEXT: store i8* [[TMP17]], i8** [[TMP16]], align 8 1789 // CHECK2-NEXT: [[TMP18:%.*]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[DOTOMP_REDUCTION_RED_LIST]], i64 0, i64 1 1790 // CHECK2-NEXT: [[TMP19:%.*]] = inttoptr i64 [[TMP6]] to i8* 1791 // CHECK2-NEXT: store i8* [[TMP19]], i8** [[TMP18]], align 8 1792 // CHECK2-NEXT: [[TMP20:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 1793 // CHECK2-NEXT: [[TMP21:%.*]] = load i32, i32* [[TMP20]], align 4 1794 // CHECK2-NEXT: [[TMP22:%.*]] = bitcast [2 x i8*]* [[DOTOMP_REDUCTION_RED_LIST]] to i8* 1795 // CHECK2-NEXT: [[TMP23:%.*]] = call i32 @__kmpc_reduce_nowait(%struct.ident_t* @[[GLOB1:[0-9]+]], i32 [[TMP21]], i32 1, i64 16, i8* [[TMP22]], void (i8*, i8*)* @.omp.reduction.reduction_func, [8 x i32]* @.gomp_critical_user_.reduction.var) 1796 // CHECK2-NEXT: switch i32 [[TMP23]], label [[DOTOMP_REDUCTION_DEFAULT:%.*]] [ 1797 // CHECK2-NEXT: i32 1, label [[DOTOMP_REDUCTION_CASE1:%.*]] 1798 // CHECK2-NEXT: i32 2, label [[DOTOMP_REDUCTION_CASE2:%.*]] 1799 // CHECK2-NEXT: ] 1800 // CHECK2: .omp.reduction.case1: 1801 // CHECK2-NEXT: [[TMP24:%.*]] = getelementptr i16, i16* [[ARRAYIDX]], i64 [[TMP6]] 1802 // CHECK2-NEXT: [[OMP_ARRAYCPY_ISEMPTY:%.*]] = icmp eq i16* [[ARRAYIDX]], [[TMP24]] 1803 // CHECK2-NEXT: br i1 [[OMP_ARRAYCPY_ISEMPTY]], label [[OMP_ARRAYCPY_DONE7:%.*]], label [[OMP_ARRAYCPY_BODY:%.*]] 1804 // CHECK2: omp.arraycpy.body: 1805 // CHECK2-NEXT: [[OMP_ARRAYCPY_SRCELEMENTPAST:%.*]] = phi i16* [ [[VLA]], [[DOTOMP_REDUCTION_CASE1]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ] 1806 // CHECK2-NEXT: [[OMP_ARRAYCPY_DESTELEMENTPAST2:%.*]] = phi i16* [ [[ARRAYIDX]], [[DOTOMP_REDUCTION_CASE1]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT5:%.*]], [[OMP_ARRAYCPY_BODY]] ] 1807 // CHECK2-NEXT: [[TMP25:%.*]] = load i16, i16* [[OMP_ARRAYCPY_DESTELEMENTPAST2]], align 2 1808 // CHECK2-NEXT: [[CONV:%.*]] = sext i16 [[TMP25]] to i32 1809 // CHECK2-NEXT: [[TMP26:%.*]] = load i16, i16* [[OMP_ARRAYCPY_SRCELEMENTPAST]], align 2 1810 // CHECK2-NEXT: [[CONV3:%.*]] = sext i16 [[TMP26]] to i32 1811 // CHECK2-NEXT: [[ADD:%.*]] = add nsw i32 [[CONV]], [[CONV3]] 1812 // CHECK2-NEXT: [[CONV4:%.*]] = trunc i32 [[ADD]] to i16 1813 // CHECK2-NEXT: store i16 [[CONV4]], i16* [[OMP_ARRAYCPY_DESTELEMENTPAST2]], align 2 1814 // CHECK2-NEXT: [[OMP_ARRAYCPY_DEST_ELEMENT5]] = getelementptr i16, i16* [[OMP_ARRAYCPY_DESTELEMENTPAST2]], i32 1 1815 // CHECK2-NEXT: [[OMP_ARRAYCPY_SRC_ELEMENT]] = getelementptr i16, i16* [[OMP_ARRAYCPY_SRCELEMENTPAST]], i32 1 1816 // CHECK2-NEXT: [[OMP_ARRAYCPY_DONE6:%.*]] = icmp eq i16* [[OMP_ARRAYCPY_DEST_ELEMENT5]], [[TMP24]] 1817 // CHECK2-NEXT: br i1 [[OMP_ARRAYCPY_DONE6]], label [[OMP_ARRAYCPY_DONE7]], label [[OMP_ARRAYCPY_BODY]] 1818 // CHECK2: omp.arraycpy.done7: 1819 // CHECK2-NEXT: call void @__kmpc_end_reduce_nowait(%struct.ident_t* @[[GLOB1]], i32 [[TMP21]], [8 x i32]* @.gomp_critical_user_.reduction.var) 1820 // CHECK2-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]] 1821 // CHECK2: .omp.reduction.case2: 1822 // CHECK2-NEXT: [[TMP27:%.*]] = getelementptr i16, i16* [[ARRAYIDX]], i64 [[TMP6]] 1823 // CHECK2-NEXT: [[OMP_ARRAYCPY_ISEMPTY8:%.*]] = icmp eq i16* [[ARRAYIDX]], [[TMP27]] 1824 // CHECK2-NEXT: br i1 [[OMP_ARRAYCPY_ISEMPTY8]], label [[OMP_ARRAYCPY_DONE21:%.*]], label [[OMP_ARRAYCPY_BODY9:%.*]] 1825 // CHECK2: omp.arraycpy.body9: 1826 // CHECK2-NEXT: [[OMP_ARRAYCPY_SRCELEMENTPAST10:%.*]] = phi i16* [ [[VLA]], [[DOTOMP_REDUCTION_CASE2]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT19:%.*]], [[ATOMIC_EXIT:%.*]] ] 1827 // CHECK2-NEXT: [[OMP_ARRAYCPY_DESTELEMENTPAST11:%.*]] = phi i16* [ [[ARRAYIDX]], [[DOTOMP_REDUCTION_CASE2]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT18:%.*]], [[ATOMIC_EXIT]] ] 1828 // CHECK2-NEXT: [[TMP28:%.*]] = load i16, i16* [[OMP_ARRAYCPY_SRCELEMENTPAST10]], align 2 1829 // CHECK2-NEXT: [[CONV12:%.*]] = sext i16 [[TMP28]] to i32 1830 // CHECK2-NEXT: [[ATOMIC_LOAD:%.*]] = load atomic i16, i16* [[OMP_ARRAYCPY_DESTELEMENTPAST11]] monotonic, align 2 1831 // CHECK2-NEXT: br label [[ATOMIC_CONT:%.*]] 1832 // CHECK2: atomic_cont: 1833 // CHECK2-NEXT: [[TMP29:%.*]] = phi i16 [ [[ATOMIC_LOAD]], [[OMP_ARRAYCPY_BODY9]] ], [ [[TMP34:%.*]], [[ATOMIC_CONT]] ] 1834 // CHECK2-NEXT: store i16 [[TMP29]], i16* [[_TMP13]], align 2 1835 // CHECK2-NEXT: [[TMP30:%.*]] = load i16, i16* [[_TMP13]], align 2 1836 // CHECK2-NEXT: [[CONV14:%.*]] = sext i16 [[TMP30]] to i32 1837 // CHECK2-NEXT: [[TMP31:%.*]] = load i16, i16* [[OMP_ARRAYCPY_SRCELEMENTPAST10]], align 2 1838 // CHECK2-NEXT: [[CONV15:%.*]] = sext i16 [[TMP31]] to i32 1839 // CHECK2-NEXT: [[ADD16:%.*]] = add nsw i32 [[CONV14]], [[CONV15]] 1840 // CHECK2-NEXT: [[CONV17:%.*]] = trunc i32 [[ADD16]] to i16 1841 // CHECK2-NEXT: store i16 [[CONV17]], i16* [[ATOMIC_TEMP]], align 2 1842 // CHECK2-NEXT: [[TMP32:%.*]] = load i16, i16* [[ATOMIC_TEMP]], align 2 1843 // CHECK2-NEXT: [[TMP33:%.*]] = cmpxchg i16* [[OMP_ARRAYCPY_DESTELEMENTPAST11]], i16 [[TMP29]], i16 [[TMP32]] monotonic monotonic, align 2 1844 // CHECK2-NEXT: [[TMP34]] = extractvalue { i16, i1 } [[TMP33]], 0 1845 // CHECK2-NEXT: [[TMP35:%.*]] = extractvalue { i16, i1 } [[TMP33]], 1 1846 // CHECK2-NEXT: br i1 [[TMP35]], label [[ATOMIC_EXIT]], label [[ATOMIC_CONT]] 1847 // CHECK2: atomic_exit: 1848 // CHECK2-NEXT: [[OMP_ARRAYCPY_DEST_ELEMENT18]] = getelementptr i16, i16* [[OMP_ARRAYCPY_DESTELEMENTPAST11]], i32 1 1849 // CHECK2-NEXT: [[OMP_ARRAYCPY_SRC_ELEMENT19]] = getelementptr i16, i16* [[OMP_ARRAYCPY_SRCELEMENTPAST10]], i32 1 1850 // CHECK2-NEXT: [[OMP_ARRAYCPY_DONE20:%.*]] = icmp eq i16* [[OMP_ARRAYCPY_DEST_ELEMENT18]], [[TMP27]] 1851 // CHECK2-NEXT: br i1 [[OMP_ARRAYCPY_DONE20]], label [[OMP_ARRAYCPY_DONE21]], label [[OMP_ARRAYCPY_BODY9]] 1852 // CHECK2: omp.arraycpy.done21: 1853 // CHECK2-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]] 1854 // CHECK2: .omp.reduction.default: 1855 // CHECK2-NEXT: [[TMP36:%.*]] = load i8*, i8** [[SAVED_STACK]], align 8 1856 // CHECK2-NEXT: call void @llvm.stackrestore(i8* [[TMP36]]) 1857 // CHECK2-NEXT: ret void 1858 // 1859 // 1860 // CHECK2-LABEL: define {{[^@]+}}@.omp.reduction.reduction_func 1861 // CHECK2-SAME: (i8* [[TMP0:%.*]], i8* [[TMP1:%.*]]) #[[ATTR3:[0-9]+]] { 1862 // CHECK2-NEXT: entry: 1863 // CHECK2-NEXT: [[DOTADDR:%.*]] = alloca i8*, align 8 1864 // CHECK2-NEXT: [[DOTADDR1:%.*]] = alloca i8*, align 8 1865 // CHECK2-NEXT: store i8* [[TMP0]], i8** [[DOTADDR]], align 8 1866 // CHECK2-NEXT: store i8* [[TMP1]], i8** [[DOTADDR1]], align 8 1867 // CHECK2-NEXT: [[TMP2:%.*]] = load i8*, i8** [[DOTADDR]], align 8 1868 // CHECK2-NEXT: [[TMP3:%.*]] = bitcast i8* [[TMP2]] to [2 x i8*]* 1869 // CHECK2-NEXT: [[TMP4:%.*]] = load i8*, i8** [[DOTADDR1]], align 8 1870 // CHECK2-NEXT: [[TMP5:%.*]] = bitcast i8* [[TMP4]] to [2 x i8*]* 1871 // CHECK2-NEXT: [[TMP6:%.*]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[TMP5]], i64 0, i64 0 1872 // CHECK2-NEXT: [[TMP7:%.*]] = load i8*, i8** [[TMP6]], align 8 1873 // CHECK2-NEXT: [[TMP8:%.*]] = bitcast i8* [[TMP7]] to i16* 1874 // CHECK2-NEXT: [[TMP9:%.*]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[TMP3]], i64 0, i64 0 1875 // CHECK2-NEXT: [[TMP10:%.*]] = load i8*, i8** [[TMP9]], align 8 1876 // CHECK2-NEXT: [[TMP11:%.*]] = bitcast i8* [[TMP10]] to i16* 1877 // CHECK2-NEXT: [[TMP12:%.*]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[TMP3]], i64 0, i64 1 1878 // CHECK2-NEXT: [[TMP13:%.*]] = load i8*, i8** [[TMP12]], align 8 1879 // CHECK2-NEXT: [[TMP14:%.*]] = ptrtoint i8* [[TMP13]] to i64 1880 // CHECK2-NEXT: [[TMP15:%.*]] = getelementptr i16, i16* [[TMP11]], i64 [[TMP14]] 1881 // CHECK2-NEXT: [[OMP_ARRAYCPY_ISEMPTY:%.*]] = icmp eq i16* [[TMP11]], [[TMP15]] 1882 // CHECK2-NEXT: br i1 [[OMP_ARRAYCPY_ISEMPTY]], label [[OMP_ARRAYCPY_DONE4:%.*]], label [[OMP_ARRAYCPY_BODY:%.*]] 1883 // CHECK2: omp.arraycpy.body: 1884 // CHECK2-NEXT: [[OMP_ARRAYCPY_SRCELEMENTPAST:%.*]] = phi i16* [ [[TMP8]], [[ENTRY:%.*]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ] 1885 // CHECK2-NEXT: [[OMP_ARRAYCPY_DESTELEMENTPAST:%.*]] = phi i16* [ [[TMP11]], [[ENTRY]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ] 1886 // CHECK2-NEXT: [[TMP16:%.*]] = load i16, i16* [[OMP_ARRAYCPY_DESTELEMENTPAST]], align 2 1887 // CHECK2-NEXT: [[CONV:%.*]] = sext i16 [[TMP16]] to i32 1888 // CHECK2-NEXT: [[TMP17:%.*]] = load i16, i16* [[OMP_ARRAYCPY_SRCELEMENTPAST]], align 2 1889 // CHECK2-NEXT: [[CONV2:%.*]] = sext i16 [[TMP17]] to i32 1890 // CHECK2-NEXT: [[ADD:%.*]] = add nsw i32 [[CONV]], [[CONV2]] 1891 // CHECK2-NEXT: [[CONV3:%.*]] = trunc i32 [[ADD]] to i16 1892 // CHECK2-NEXT: store i16 [[CONV3]], i16* [[OMP_ARRAYCPY_DESTELEMENTPAST]], align 2 1893 // CHECK2-NEXT: [[OMP_ARRAYCPY_DEST_ELEMENT]] = getelementptr i16, i16* [[OMP_ARRAYCPY_DESTELEMENTPAST]], i32 1 1894 // CHECK2-NEXT: [[OMP_ARRAYCPY_SRC_ELEMENT]] = getelementptr i16, i16* [[OMP_ARRAYCPY_SRCELEMENTPAST]], i32 1 1895 // CHECK2-NEXT: [[OMP_ARRAYCPY_DONE:%.*]] = icmp eq i16* [[OMP_ARRAYCPY_DEST_ELEMENT]], [[TMP15]] 1896 // CHECK2-NEXT: br i1 [[OMP_ARRAYCPY_DONE]], label [[OMP_ARRAYCPY_DONE4]], label [[OMP_ARRAYCPY_BODY]] 1897 // CHECK2: omp.arraycpy.done4: 1898 // CHECK2-NEXT: ret void 1899 // 1900 // 1901 // CHECK2-LABEL: define {{[^@]+}}@main 1902 // CHECK2-SAME: () #[[ATTR6:[0-9]+]] { 1903 // CHECK2-NEXT: entry: 1904 // CHECK2-NEXT: [[RETVAL:%.*]] = alloca i32, align 4 1905 // CHECK2-NEXT: [[SS:%.*]] = alloca [[STRUCT_SS:%.*]], align 8 1906 // CHECK2-NEXT: [[TEST:%.*]] = alloca [[STRUCT_S:%.*]], align 4 1907 // CHECK2-NEXT: [[T_VAR:%.*]] = alloca float, align 4 1908 // CHECK2-NEXT: [[T_VAR1:%.*]] = alloca float, align 4 1909 // CHECK2-NEXT: [[VEC:%.*]] = alloca [2 x i32], align 4 1910 // CHECK2-NEXT: [[S_ARR:%.*]] = alloca [2 x %struct.S], align 4 1911 // CHECK2-NEXT: [[VAR:%.*]] = alloca [[STRUCT_S]], align 4 1912 // CHECK2-NEXT: [[VAR1:%.*]] = alloca [[STRUCT_S]], align 4 1913 // CHECK2-NEXT: [[CF:%.*]] = alloca { float, float }, align 4 1914 // CHECK2-NEXT: store i32 0, i32* [[RETVAL]], align 4 1915 // CHECK2-NEXT: call void @_ZN2SSC1ERi(%struct.SS* nonnull align 8 dereferenceable(16) [[SS]], i32* nonnull align 4 dereferenceable(4) @sivar) 1916 // CHECK2-NEXT: call void @_ZN1SIfEC1Ev(%struct.S* nonnull align 4 dereferenceable(4) [[TEST]]) 1917 // CHECK2-NEXT: store float 0.000000e+00, float* [[T_VAR]], align 4 1918 // CHECK2-NEXT: [[TMP0:%.*]] = bitcast [2 x i32]* [[VEC]] to i8* 1919 // CHECK2-NEXT: call void @llvm.memcpy.p0i8.p0i8.i64(i8* align 4 [[TMP0]], i8* align 4 bitcast ([2 x i32]* @__const.main.vec to i8*), i64 8, i1 false) 1920 // CHECK2-NEXT: [[ARRAYINIT_BEGIN:%.*]] = getelementptr inbounds [2 x %struct.S], [2 x %struct.S]* [[S_ARR]], i64 0, i64 0 1921 // CHECK2-NEXT: call void @_ZN1SIfEC1Ef(%struct.S* nonnull align 4 dereferenceable(4) [[ARRAYINIT_BEGIN]], float 1.000000e+00) 1922 // CHECK2-NEXT: [[ARRAYINIT_ELEMENT:%.*]] = getelementptr inbounds [[STRUCT_S]], %struct.S* [[ARRAYINIT_BEGIN]], i64 1 1923 // CHECK2-NEXT: call void @_ZN1SIfEC1Ef(%struct.S* nonnull align 4 dereferenceable(4) [[ARRAYINIT_ELEMENT]], float 2.000000e+00) 1924 // CHECK2-NEXT: call void @_ZN1SIfEC1Ef(%struct.S* nonnull align 4 dereferenceable(4) [[VAR]], float 3.000000e+00) 1925 // CHECK2-NEXT: call void @_ZN1SIfEC1Ev(%struct.S* nonnull align 4 dereferenceable(4) [[VAR1]]) 1926 // CHECK2-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB2]], i32 6, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, [2 x i32]*, float*, [2 x %struct.S]*, %struct.S*, %struct.S*, float*)* @.omp_outlined..1 to void (i32*, i32*, ...)*), [2 x i32]* [[VEC]], float* [[T_VAR]], [2 x %struct.S]* [[S_ARR]], %struct.S* [[VAR]], %struct.S* [[VAR1]], float* [[T_VAR1]]) 1927 // CHECK2-NEXT: [[CALL:%.*]] = call float @_ZN1SIfEcvfEv(%struct.S* nonnull align 4 dereferenceable(4) [[VAR1]]) 1928 // CHECK2-NEXT: [[TOBOOL:%.*]] = fcmp une float [[CALL]], 0.000000e+00 1929 // CHECK2-NEXT: br i1 [[TOBOOL]], label [[IF_THEN:%.*]], label [[IF_END:%.*]] 1930 // CHECK2: if.then: 1931 // CHECK2-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB2]], i32 6, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, [2 x i32]*, float*, [2 x %struct.S]*, %struct.S*, %struct.S*, float*)* @.omp_outlined..3 to void (i32*, i32*, ...)*), [2 x i32]* [[VEC]], float* [[T_VAR]], [2 x %struct.S]* [[S_ARR]], %struct.S* [[VAR]], %struct.S* [[VAR1]], float* [[T_VAR1]]) 1932 // CHECK2-NEXT: br label [[IF_END]] 1933 // CHECK2: if.end: 1934 // CHECK2-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB2]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, { float, float }*)* @.omp_outlined..4 to void (i32*, i32*, ...)*), { float, float }* [[CF]]) 1935 // CHECK2-NEXT: [[CALL1:%.*]] = call i32 @_Z5tmainIiET_v() 1936 // CHECK2-NEXT: store i32 [[CALL1]], i32* [[RETVAL]], align 4 1937 // CHECK2-NEXT: call void @_ZN1SIfED1Ev(%struct.S* nonnull align 4 dereferenceable(4) [[VAR1]]) #[[ATTR5:[0-9]+]] 1938 // CHECK2-NEXT: call void @_ZN1SIfED1Ev(%struct.S* nonnull align 4 dereferenceable(4) [[VAR]]) #[[ATTR5]] 1939 // CHECK2-NEXT: [[ARRAY_BEGIN:%.*]] = getelementptr inbounds [2 x %struct.S], [2 x %struct.S]* [[S_ARR]], i32 0, i32 0 1940 // CHECK2-NEXT: [[TMP1:%.*]] = getelementptr inbounds [[STRUCT_S]], %struct.S* [[ARRAY_BEGIN]], i64 2 1941 // CHECK2-NEXT: br label [[ARRAYDESTROY_BODY:%.*]] 1942 // CHECK2: arraydestroy.body: 1943 // CHECK2-NEXT: [[ARRAYDESTROY_ELEMENTPAST:%.*]] = phi %struct.S* [ [[TMP1]], [[IF_END]] ], [ [[ARRAYDESTROY_ELEMENT:%.*]], [[ARRAYDESTROY_BODY]] ] 1944 // CHECK2-NEXT: [[ARRAYDESTROY_ELEMENT]] = getelementptr inbounds [[STRUCT_S]], %struct.S* [[ARRAYDESTROY_ELEMENTPAST]], i64 -1 1945 // CHECK2-NEXT: call void @_ZN1SIfED1Ev(%struct.S* nonnull align 4 dereferenceable(4) [[ARRAYDESTROY_ELEMENT]]) #[[ATTR5]] 1946 // CHECK2-NEXT: [[ARRAYDESTROY_DONE:%.*]] = icmp eq %struct.S* [[ARRAYDESTROY_ELEMENT]], [[ARRAY_BEGIN]] 1947 // CHECK2-NEXT: br i1 [[ARRAYDESTROY_DONE]], label [[ARRAYDESTROY_DONE2:%.*]], label [[ARRAYDESTROY_BODY]] 1948 // CHECK2: arraydestroy.done2: 1949 // CHECK2-NEXT: call void @_ZN1SIfED1Ev(%struct.S* nonnull align 4 dereferenceable(4) [[TEST]]) #[[ATTR5]] 1950 // CHECK2-NEXT: [[TMP2:%.*]] = load i32, i32* [[RETVAL]], align 4 1951 // CHECK2-NEXT: ret i32 [[TMP2]] 1952 // 1953 // 1954 // CHECK2-LABEL: define {{[^@]+}}@_ZN2SSC1ERi 1955 // CHECK2-SAME: (%struct.SS* nonnull align 8 dereferenceable(16) [[THIS:%.*]], i32* nonnull align 4 dereferenceable(4) [[D:%.*]]) unnamed_addr #[[ATTR7:[0-9]+]] align 2 { 1956 // CHECK2-NEXT: entry: 1957 // CHECK2-NEXT: [[THIS_ADDR:%.*]] = alloca %struct.SS*, align 8 1958 // CHECK2-NEXT: [[D_ADDR:%.*]] = alloca i32*, align 8 1959 // CHECK2-NEXT: store %struct.SS* [[THIS]], %struct.SS** [[THIS_ADDR]], align 8 1960 // CHECK2-NEXT: store i32* [[D]], i32** [[D_ADDR]], align 8 1961 // CHECK2-NEXT: [[THIS1:%.*]] = load %struct.SS*, %struct.SS** [[THIS_ADDR]], align 8 1962 // CHECK2-NEXT: [[TMP0:%.*]] = load i32*, i32** [[D_ADDR]], align 8 1963 // CHECK2-NEXT: call void @_ZN2SSC2ERi(%struct.SS* nonnull align 8 dereferenceable(16) [[THIS1]], i32* nonnull align 4 dereferenceable(4) [[TMP0]]) 1964 // CHECK2-NEXT: ret void 1965 // 1966 // 1967 // CHECK2-LABEL: define {{[^@]+}}@_ZN1SIfEC1Ev 1968 // CHECK2-SAME: (%struct.S* nonnull align 4 dereferenceable(4) [[THIS:%.*]]) unnamed_addr #[[ATTR7]] align 2 { 1969 // CHECK2-NEXT: entry: 1970 // CHECK2-NEXT: [[THIS_ADDR:%.*]] = alloca %struct.S*, align 8 1971 // CHECK2-NEXT: store %struct.S* [[THIS]], %struct.S** [[THIS_ADDR]], align 8 1972 // CHECK2-NEXT: [[THIS1:%.*]] = load %struct.S*, %struct.S** [[THIS_ADDR]], align 8 1973 // CHECK2-NEXT: call void @_ZN1SIfEC2Ev(%struct.S* nonnull align 4 dereferenceable(4) [[THIS1]]) 1974 // CHECK2-NEXT: ret void 1975 // 1976 // 1977 // CHECK2-LABEL: define {{[^@]+}}@_ZN1SIfEC1Ef 1978 // CHECK2-SAME: (%struct.S* nonnull align 4 dereferenceable(4) [[THIS:%.*]], float [[A:%.*]]) unnamed_addr #[[ATTR7]] align 2 { 1979 // CHECK2-NEXT: entry: 1980 // CHECK2-NEXT: [[THIS_ADDR:%.*]] = alloca %struct.S*, align 8 1981 // CHECK2-NEXT: [[A_ADDR:%.*]] = alloca float, align 4 1982 // CHECK2-NEXT: store %struct.S* [[THIS]], %struct.S** [[THIS_ADDR]], align 8 1983 // CHECK2-NEXT: store float [[A]], float* [[A_ADDR]], align 4 1984 // CHECK2-NEXT: [[THIS1:%.*]] = load %struct.S*, %struct.S** [[THIS_ADDR]], align 8 1985 // CHECK2-NEXT: [[TMP0:%.*]] = load float, float* [[A_ADDR]], align 4 1986 // CHECK2-NEXT: call void @_ZN1SIfEC2Ef(%struct.S* nonnull align 4 dereferenceable(4) [[THIS1]], float [[TMP0]]) 1987 // CHECK2-NEXT: ret void 1988 // 1989 // 1990 // CHECK2-LABEL: define {{[^@]+}}@.omp_outlined..1 1991 // CHECK2-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], [2 x i32]* nonnull align 4 dereferenceable(8) [[VEC:%.*]], float* nonnull align 4 dereferenceable(4) [[T_VAR:%.*]], [2 x %struct.S]* nonnull align 4 dereferenceable(8) [[S_ARR:%.*]], %struct.S* nonnull align 4 dereferenceable(4) [[VAR:%.*]], %struct.S* nonnull align 4 dereferenceable(4) [[VAR1:%.*]], float* nonnull align 4 dereferenceable(4) [[T_VAR1:%.*]]) #[[ATTR1]] { 1992 // CHECK2-NEXT: entry: 1993 // CHECK2-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 1994 // CHECK2-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 1995 // CHECK2-NEXT: [[VEC_ADDR:%.*]] = alloca [2 x i32]*, align 8 1996 // CHECK2-NEXT: [[T_VAR_ADDR:%.*]] = alloca float*, align 8 1997 // CHECK2-NEXT: [[S_ARR_ADDR:%.*]] = alloca [2 x %struct.S]*, align 8 1998 // CHECK2-NEXT: [[VAR_ADDR:%.*]] = alloca %struct.S*, align 8 1999 // CHECK2-NEXT: [[VAR1_ADDR:%.*]] = alloca %struct.S*, align 8 2000 // CHECK2-NEXT: [[T_VAR1_ADDR:%.*]] = alloca float*, align 8 2001 // CHECK2-NEXT: [[T_VAR2:%.*]] = alloca float, align 4 2002 // CHECK2-NEXT: [[VAR3:%.*]] = alloca [[STRUCT_S:%.*]], align 4 2003 // CHECK2-NEXT: [[VAR14:%.*]] = alloca [[STRUCT_S]], align 4 2004 // CHECK2-NEXT: [[T_VAR15:%.*]] = alloca float, align 4 2005 // CHECK2-NEXT: [[DOTOMP_REDUCTION_RED_LIST:%.*]] = alloca [4 x i8*], align 8 2006 // CHECK2-NEXT: [[REF_TMP:%.*]] = alloca [[STRUCT_S]], align 4 2007 // CHECK2-NEXT: [[ATOMIC_TEMP:%.*]] = alloca float, align 4 2008 // CHECK2-NEXT: [[TMP:%.*]] = alloca float, align 4 2009 // CHECK2-NEXT: [[REF_TMP13:%.*]] = alloca [[STRUCT_S]], align 4 2010 // CHECK2-NEXT: [[ATOMIC_TEMP23:%.*]] = alloca float, align 4 2011 // CHECK2-NEXT: [[_TMP24:%.*]] = alloca float, align 4 2012 // CHECK2-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8 2013 // CHECK2-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8 2014 // CHECK2-NEXT: store [2 x i32]* [[VEC]], [2 x i32]** [[VEC_ADDR]], align 8 2015 // CHECK2-NEXT: store float* [[T_VAR]], float** [[T_VAR_ADDR]], align 8 2016 // CHECK2-NEXT: store [2 x %struct.S]* [[S_ARR]], [2 x %struct.S]** [[S_ARR_ADDR]], align 8 2017 // CHECK2-NEXT: store %struct.S* [[VAR]], %struct.S** [[VAR_ADDR]], align 8 2018 // CHECK2-NEXT: store %struct.S* [[VAR1]], %struct.S** [[VAR1_ADDR]], align 8 2019 // CHECK2-NEXT: store float* [[T_VAR1]], float** [[T_VAR1_ADDR]], align 8 2020 // CHECK2-NEXT: [[TMP0:%.*]] = load [2 x i32]*, [2 x i32]** [[VEC_ADDR]], align 8 2021 // CHECK2-NEXT: [[TMP1:%.*]] = load float*, float** [[T_VAR_ADDR]], align 8 2022 // CHECK2-NEXT: [[TMP2:%.*]] = load [2 x %struct.S]*, [2 x %struct.S]** [[S_ARR_ADDR]], align 8 2023 // CHECK2-NEXT: [[TMP3:%.*]] = load %struct.S*, %struct.S** [[VAR_ADDR]], align 8 2024 // CHECK2-NEXT: [[TMP4:%.*]] = load %struct.S*, %struct.S** [[VAR1_ADDR]], align 8 2025 // CHECK2-NEXT: [[TMP5:%.*]] = load float*, float** [[T_VAR1_ADDR]], align 8 2026 // CHECK2-NEXT: store float 0.000000e+00, float* [[T_VAR2]], align 4 2027 // CHECK2-NEXT: call void @_ZN1SIfEC1Ev(%struct.S* nonnull align 4 dereferenceable(4) [[VAR3]]) 2028 // CHECK2-NEXT: call void @_ZN1SIfEC1Ev(%struct.S* nonnull align 4 dereferenceable(4) [[VAR14]]) 2029 // CHECK2-NEXT: store float 0x47EFFFFFE0000000, float* [[T_VAR15]], align 4 2030 // CHECK2-NEXT: [[TMP6:%.*]] = load float, float* [[T_VAR2]], align 4 2031 // CHECK2-NEXT: [[CONV:%.*]] = fptosi float [[TMP6]] to i32 2032 // CHECK2-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [2 x i32], [2 x i32]* [[TMP0]], i64 0, i64 0 2033 // CHECK2-NEXT: store i32 [[CONV]], i32* [[ARRAYIDX]], align 4 2034 // CHECK2-NEXT: [[ARRAYIDX6:%.*]] = getelementptr inbounds [2 x %struct.S], [2 x %struct.S]* [[TMP2]], i64 0, i64 0 2035 // CHECK2-NEXT: [[TMP7:%.*]] = bitcast %struct.S* [[ARRAYIDX6]] to i8* 2036 // CHECK2-NEXT: [[TMP8:%.*]] = bitcast %struct.S* [[VAR3]] to i8* 2037 // CHECK2-NEXT: call void @llvm.memcpy.p0i8.p0i8.i64(i8* align 4 [[TMP7]], i8* align 4 [[TMP8]], i64 4, i1 false) 2038 // CHECK2-NEXT: [[TMP9:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOMP_REDUCTION_RED_LIST]], i64 0, i64 0 2039 // CHECK2-NEXT: [[TMP10:%.*]] = bitcast float* [[T_VAR2]] to i8* 2040 // CHECK2-NEXT: store i8* [[TMP10]], i8** [[TMP9]], align 8 2041 // CHECK2-NEXT: [[TMP11:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOMP_REDUCTION_RED_LIST]], i64 0, i64 1 2042 // CHECK2-NEXT: [[TMP12:%.*]] = bitcast %struct.S* [[VAR3]] to i8* 2043 // CHECK2-NEXT: store i8* [[TMP12]], i8** [[TMP11]], align 8 2044 // CHECK2-NEXT: [[TMP13:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOMP_REDUCTION_RED_LIST]], i64 0, i64 2 2045 // CHECK2-NEXT: [[TMP14:%.*]] = bitcast %struct.S* [[VAR14]] to i8* 2046 // CHECK2-NEXT: store i8* [[TMP14]], i8** [[TMP13]], align 8 2047 // CHECK2-NEXT: [[TMP15:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOMP_REDUCTION_RED_LIST]], i64 0, i64 3 2048 // CHECK2-NEXT: [[TMP16:%.*]] = bitcast float* [[T_VAR15]] to i8* 2049 // CHECK2-NEXT: store i8* [[TMP16]], i8** [[TMP15]], align 8 2050 // CHECK2-NEXT: [[TMP17:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 2051 // CHECK2-NEXT: [[TMP18:%.*]] = load i32, i32* [[TMP17]], align 4 2052 // CHECK2-NEXT: [[TMP19:%.*]] = bitcast [4 x i8*]* [[DOTOMP_REDUCTION_RED_LIST]] to i8* 2053 // CHECK2-NEXT: [[TMP20:%.*]] = call i32 @__kmpc_reduce_nowait(%struct.ident_t* @[[GLOB1]], i32 [[TMP18]], i32 4, i64 32, i8* [[TMP19]], void (i8*, i8*)* @.omp.reduction.reduction_func.2, [8 x i32]* @.gomp_critical_user_.reduction.var) 2054 // CHECK2-NEXT: switch i32 [[TMP20]], label [[DOTOMP_REDUCTION_DEFAULT:%.*]] [ 2055 // CHECK2-NEXT: i32 1, label [[DOTOMP_REDUCTION_CASE1:%.*]] 2056 // CHECK2-NEXT: i32 2, label [[DOTOMP_REDUCTION_CASE2:%.*]] 2057 // CHECK2-NEXT: ] 2058 // CHECK2: .omp.reduction.case1: 2059 // CHECK2-NEXT: [[TMP21:%.*]] = load float, float* [[TMP1]], align 4 2060 // CHECK2-NEXT: [[TMP22:%.*]] = load float, float* [[T_VAR2]], align 4 2061 // CHECK2-NEXT: [[ADD:%.*]] = fadd float [[TMP21]], [[TMP22]] 2062 // CHECK2-NEXT: store float [[ADD]], float* [[TMP1]], align 4 2063 // CHECK2-NEXT: [[CALL:%.*]] = call nonnull align 4 dereferenceable(4) %struct.S* @_ZN1SIfEanERKS0_(%struct.S* nonnull align 4 dereferenceable(4) [[TMP3]], %struct.S* nonnull align 4 dereferenceable(4) [[VAR3]]) 2064 // CHECK2-NEXT: [[TMP23:%.*]] = bitcast %struct.S* [[TMP3]] to i8* 2065 // CHECK2-NEXT: [[TMP24:%.*]] = bitcast %struct.S* [[CALL]] to i8* 2066 // CHECK2-NEXT: call void @llvm.memcpy.p0i8.p0i8.i64(i8* align 4 [[TMP23]], i8* align 4 [[TMP24]], i64 4, i1 false) 2067 // CHECK2-NEXT: [[CALL7:%.*]] = call float @_ZN1SIfEcvfEv(%struct.S* nonnull align 4 dereferenceable(4) [[TMP4]]) 2068 // CHECK2-NEXT: [[TOBOOL:%.*]] = fcmp une float [[CALL7]], 0.000000e+00 2069 // CHECK2-NEXT: br i1 [[TOBOOL]], label [[LAND_RHS:%.*]], label [[LAND_END:%.*]] 2070 // CHECK2: land.rhs: 2071 // CHECK2-NEXT: [[CALL8:%.*]] = call float @_ZN1SIfEcvfEv(%struct.S* nonnull align 4 dereferenceable(4) [[VAR14]]) 2072 // CHECK2-NEXT: [[TOBOOL9:%.*]] = fcmp une float [[CALL8]], 0.000000e+00 2073 // CHECK2-NEXT: br label [[LAND_END]] 2074 // CHECK2: land.end: 2075 // CHECK2-NEXT: [[TMP25:%.*]] = phi i1 [ false, [[DOTOMP_REDUCTION_CASE1]] ], [ [[TOBOOL9]], [[LAND_RHS]] ] 2076 // CHECK2-NEXT: [[CONV10:%.*]] = uitofp i1 [[TMP25]] to float 2077 // CHECK2-NEXT: call void @_ZN1SIfEC1Ef(%struct.S* nonnull align 4 dereferenceable(4) [[REF_TMP]], float [[CONV10]]) 2078 // CHECK2-NEXT: [[TMP26:%.*]] = bitcast %struct.S* [[TMP4]] to i8* 2079 // CHECK2-NEXT: [[TMP27:%.*]] = bitcast %struct.S* [[REF_TMP]] to i8* 2080 // CHECK2-NEXT: call void @llvm.memcpy.p0i8.p0i8.i64(i8* align 4 [[TMP26]], i8* align 4 [[TMP27]], i64 4, i1 false) 2081 // CHECK2-NEXT: call void @_ZN1SIfED1Ev(%struct.S* nonnull align 4 dereferenceable(4) [[REF_TMP]]) #[[ATTR5]] 2082 // CHECK2-NEXT: [[TMP28:%.*]] = load float, float* [[TMP5]], align 4 2083 // CHECK2-NEXT: [[TMP29:%.*]] = load float, float* [[T_VAR15]], align 4 2084 // CHECK2-NEXT: [[CMP:%.*]] = fcmp olt float [[TMP28]], [[TMP29]] 2085 // CHECK2-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] 2086 // CHECK2: cond.true: 2087 // CHECK2-NEXT: [[TMP30:%.*]] = load float, float* [[TMP5]], align 4 2088 // CHECK2-NEXT: br label [[COND_END:%.*]] 2089 // CHECK2: cond.false: 2090 // CHECK2-NEXT: [[TMP31:%.*]] = load float, float* [[T_VAR15]], align 4 2091 // CHECK2-NEXT: br label [[COND_END]] 2092 // CHECK2: cond.end: 2093 // CHECK2-NEXT: [[COND:%.*]] = phi float [ [[TMP30]], [[COND_TRUE]] ], [ [[TMP31]], [[COND_FALSE]] ] 2094 // CHECK2-NEXT: store float [[COND]], float* [[TMP5]], align 4 2095 // CHECK2-NEXT: call void @__kmpc_end_reduce_nowait(%struct.ident_t* @[[GLOB1]], i32 [[TMP18]], [8 x i32]* @.gomp_critical_user_.reduction.var) 2096 // CHECK2-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]] 2097 // CHECK2: .omp.reduction.case2: 2098 // CHECK2-NEXT: [[TMP32:%.*]] = load float, float* [[T_VAR2]], align 4 2099 // CHECK2-NEXT: [[TMP33:%.*]] = bitcast float* [[TMP1]] to i32* 2100 // CHECK2-NEXT: [[ATOMIC_LOAD:%.*]] = load atomic i32, i32* [[TMP33]] monotonic, align 4 2101 // CHECK2-NEXT: br label [[ATOMIC_CONT:%.*]] 2102 // CHECK2: atomic_cont: 2103 // CHECK2-NEXT: [[TMP34:%.*]] = phi i32 [ [[ATOMIC_LOAD]], [[DOTOMP_REDUCTION_CASE2]] ], [ [[TMP42:%.*]], [[ATOMIC_CONT]] ] 2104 // CHECK2-NEXT: [[TMP35:%.*]] = bitcast float* [[ATOMIC_TEMP]] to i32* 2105 // CHECK2-NEXT: [[TMP36:%.*]] = bitcast i32 [[TMP34]] to float 2106 // CHECK2-NEXT: store float [[TMP36]], float* [[TMP]], align 4 2107 // CHECK2-NEXT: [[TMP37:%.*]] = load float, float* [[TMP]], align 4 2108 // CHECK2-NEXT: [[TMP38:%.*]] = load float, float* [[T_VAR2]], align 4 2109 // CHECK2-NEXT: [[ADD11:%.*]] = fadd float [[TMP37]], [[TMP38]] 2110 // CHECK2-NEXT: store float [[ADD11]], float* [[ATOMIC_TEMP]], align 4 2111 // CHECK2-NEXT: [[TMP39:%.*]] = load i32, i32* [[TMP35]], align 4 2112 // CHECK2-NEXT: [[TMP40:%.*]] = bitcast float* [[TMP1]] to i32* 2113 // CHECK2-NEXT: [[TMP41:%.*]] = cmpxchg i32* [[TMP40]], i32 [[TMP34]], i32 [[TMP39]] monotonic monotonic, align 4 2114 // CHECK2-NEXT: [[TMP42]] = extractvalue { i32, i1 } [[TMP41]], 0 2115 // CHECK2-NEXT: [[TMP43:%.*]] = extractvalue { i32, i1 } [[TMP41]], 1 2116 // CHECK2-NEXT: br i1 [[TMP43]], label [[ATOMIC_EXIT:%.*]], label [[ATOMIC_CONT]] 2117 // CHECK2: atomic_exit: 2118 // CHECK2-NEXT: call void @__kmpc_critical(%struct.ident_t* @[[GLOB2]], i32 [[TMP18]], [8 x i32]* @.gomp_critical_user_.atomic_reduction.var) 2119 // CHECK2-NEXT: [[CALL12:%.*]] = call nonnull align 4 dereferenceable(4) %struct.S* @_ZN1SIfEanERKS0_(%struct.S* nonnull align 4 dereferenceable(4) [[TMP3]], %struct.S* nonnull align 4 dereferenceable(4) [[VAR3]]) 2120 // CHECK2-NEXT: [[TMP44:%.*]] = bitcast %struct.S* [[TMP3]] to i8* 2121 // CHECK2-NEXT: [[TMP45:%.*]] = bitcast %struct.S* [[CALL12]] to i8* 2122 // CHECK2-NEXT: call void @llvm.memcpy.p0i8.p0i8.i64(i8* align 4 [[TMP44]], i8* align 4 [[TMP45]], i64 4, i1 false) 2123 // CHECK2-NEXT: call void @__kmpc_end_critical(%struct.ident_t* @[[GLOB2]], i32 [[TMP18]], [8 x i32]* @.gomp_critical_user_.atomic_reduction.var) 2124 // CHECK2-NEXT: call void @__kmpc_critical(%struct.ident_t* @[[GLOB2]], i32 [[TMP18]], [8 x i32]* @.gomp_critical_user_.atomic_reduction.var) 2125 // CHECK2-NEXT: [[CALL14:%.*]] = call float @_ZN1SIfEcvfEv(%struct.S* nonnull align 4 dereferenceable(4) [[TMP4]]) 2126 // CHECK2-NEXT: [[TOBOOL15:%.*]] = fcmp une float [[CALL14]], 0.000000e+00 2127 // CHECK2-NEXT: br i1 [[TOBOOL15]], label [[LAND_RHS16:%.*]], label [[LAND_END19:%.*]] 2128 // CHECK2: land.rhs16: 2129 // CHECK2-NEXT: [[CALL17:%.*]] = call float @_ZN1SIfEcvfEv(%struct.S* nonnull align 4 dereferenceable(4) [[VAR14]]) 2130 // CHECK2-NEXT: [[TOBOOL18:%.*]] = fcmp une float [[CALL17]], 0.000000e+00 2131 // CHECK2-NEXT: br label [[LAND_END19]] 2132 // CHECK2: land.end19: 2133 // CHECK2-NEXT: [[TMP46:%.*]] = phi i1 [ false, [[ATOMIC_EXIT]] ], [ [[TOBOOL18]], [[LAND_RHS16]] ] 2134 // CHECK2-NEXT: [[CONV20:%.*]] = uitofp i1 [[TMP46]] to float 2135 // CHECK2-NEXT: call void @_ZN1SIfEC1Ef(%struct.S* nonnull align 4 dereferenceable(4) [[REF_TMP13]], float [[CONV20]]) 2136 // CHECK2-NEXT: [[TMP47:%.*]] = bitcast %struct.S* [[TMP4]] to i8* 2137 // CHECK2-NEXT: [[TMP48:%.*]] = bitcast %struct.S* [[REF_TMP13]] to i8* 2138 // CHECK2-NEXT: call void @llvm.memcpy.p0i8.p0i8.i64(i8* align 4 [[TMP47]], i8* align 4 [[TMP48]], i64 4, i1 false) 2139 // CHECK2-NEXT: call void @_ZN1SIfED1Ev(%struct.S* nonnull align 4 dereferenceable(4) [[REF_TMP13]]) #[[ATTR5]] 2140 // CHECK2-NEXT: call void @__kmpc_end_critical(%struct.ident_t* @[[GLOB2]], i32 [[TMP18]], [8 x i32]* @.gomp_critical_user_.atomic_reduction.var) 2141 // CHECK2-NEXT: [[TMP49:%.*]] = load float, float* [[T_VAR15]], align 4 2142 // CHECK2-NEXT: [[TMP50:%.*]] = bitcast float* [[TMP5]] to i32* 2143 // CHECK2-NEXT: [[ATOMIC_LOAD21:%.*]] = load atomic i32, i32* [[TMP50]] monotonic, align 4 2144 // CHECK2-NEXT: br label [[ATOMIC_CONT22:%.*]] 2145 // CHECK2: atomic_cont22: 2146 // CHECK2-NEXT: [[TMP51:%.*]] = phi i32 [ [[ATOMIC_LOAD21]], [[LAND_END19]] ], [ [[TMP61:%.*]], [[COND_END28:%.*]] ] 2147 // CHECK2-NEXT: [[TMP52:%.*]] = bitcast float* [[ATOMIC_TEMP23]] to i32* 2148 // CHECK2-NEXT: [[TMP53:%.*]] = bitcast i32 [[TMP51]] to float 2149 // CHECK2-NEXT: store float [[TMP53]], float* [[_TMP24]], align 4 2150 // CHECK2-NEXT: [[TMP54:%.*]] = load float, float* [[_TMP24]], align 4 2151 // CHECK2-NEXT: [[TMP55:%.*]] = load float, float* [[T_VAR15]], align 4 2152 // CHECK2-NEXT: [[CMP25:%.*]] = fcmp olt float [[TMP54]], [[TMP55]] 2153 // CHECK2-NEXT: br i1 [[CMP25]], label [[COND_TRUE26:%.*]], label [[COND_FALSE27:%.*]] 2154 // CHECK2: cond.true26: 2155 // CHECK2-NEXT: [[TMP56:%.*]] = load float, float* [[_TMP24]], align 4 2156 // CHECK2-NEXT: br label [[COND_END28]] 2157 // CHECK2: cond.false27: 2158 // CHECK2-NEXT: [[TMP57:%.*]] = load float, float* [[T_VAR15]], align 4 2159 // CHECK2-NEXT: br label [[COND_END28]] 2160 // CHECK2: cond.end28: 2161 // CHECK2-NEXT: [[COND29:%.*]] = phi float [ [[TMP56]], [[COND_TRUE26]] ], [ [[TMP57]], [[COND_FALSE27]] ] 2162 // CHECK2-NEXT: store float [[COND29]], float* [[ATOMIC_TEMP23]], align 4 2163 // CHECK2-NEXT: [[TMP58:%.*]] = load i32, i32* [[TMP52]], align 4 2164 // CHECK2-NEXT: [[TMP59:%.*]] = bitcast float* [[TMP5]] to i32* 2165 // CHECK2-NEXT: [[TMP60:%.*]] = cmpxchg i32* [[TMP59]], i32 [[TMP51]], i32 [[TMP58]] monotonic monotonic, align 4 2166 // CHECK2-NEXT: [[TMP61]] = extractvalue { i32, i1 } [[TMP60]], 0 2167 // CHECK2-NEXT: [[TMP62:%.*]] = extractvalue { i32, i1 } [[TMP60]], 1 2168 // CHECK2-NEXT: br i1 [[TMP62]], label [[ATOMIC_EXIT30:%.*]], label [[ATOMIC_CONT22]] 2169 // CHECK2: atomic_exit30: 2170 // CHECK2-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]] 2171 // CHECK2: .omp.reduction.default: 2172 // CHECK2-NEXT: call void @_ZN1SIfED1Ev(%struct.S* nonnull align 4 dereferenceable(4) [[VAR14]]) #[[ATTR5]] 2173 // CHECK2-NEXT: call void @_ZN1SIfED1Ev(%struct.S* nonnull align 4 dereferenceable(4) [[VAR3]]) #[[ATTR5]] 2174 // CHECK2-NEXT: ret void 2175 // 2176 // 2177 // CHECK2-LABEL: define {{[^@]+}}@.omp.reduction.reduction_func.2 2178 // CHECK2-SAME: (i8* [[TMP0:%.*]], i8* [[TMP1:%.*]]) #[[ATTR3]] { 2179 // CHECK2-NEXT: entry: 2180 // CHECK2-NEXT: [[DOTADDR:%.*]] = alloca i8*, align 8 2181 // CHECK2-NEXT: [[DOTADDR1:%.*]] = alloca i8*, align 8 2182 // CHECK2-NEXT: [[REF_TMP:%.*]] = alloca [[STRUCT_S:%.*]], align 4 2183 // CHECK2-NEXT: store i8* [[TMP0]], i8** [[DOTADDR]], align 8 2184 // CHECK2-NEXT: store i8* [[TMP1]], i8** [[DOTADDR1]], align 8 2185 // CHECK2-NEXT: [[TMP2:%.*]] = load i8*, i8** [[DOTADDR]], align 8 2186 // CHECK2-NEXT: [[TMP3:%.*]] = bitcast i8* [[TMP2]] to [4 x i8*]* 2187 // CHECK2-NEXT: [[TMP4:%.*]] = load i8*, i8** [[DOTADDR1]], align 8 2188 // CHECK2-NEXT: [[TMP5:%.*]] = bitcast i8* [[TMP4]] to [4 x i8*]* 2189 // CHECK2-NEXT: [[TMP6:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[TMP5]], i64 0, i64 0 2190 // CHECK2-NEXT: [[TMP7:%.*]] = load i8*, i8** [[TMP6]], align 8 2191 // CHECK2-NEXT: [[TMP8:%.*]] = bitcast i8* [[TMP7]] to float* 2192 // CHECK2-NEXT: [[TMP9:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[TMP3]], i64 0, i64 0 2193 // CHECK2-NEXT: [[TMP10:%.*]] = load i8*, i8** [[TMP9]], align 8 2194 // CHECK2-NEXT: [[TMP11:%.*]] = bitcast i8* [[TMP10]] to float* 2195 // CHECK2-NEXT: [[TMP12:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[TMP5]], i64 0, i64 1 2196 // CHECK2-NEXT: [[TMP13:%.*]] = load i8*, i8** [[TMP12]], align 8 2197 // CHECK2-NEXT: [[TMP14:%.*]] = bitcast i8* [[TMP13]] to %struct.S* 2198 // CHECK2-NEXT: [[TMP15:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[TMP3]], i64 0, i64 1 2199 // CHECK2-NEXT: [[TMP16:%.*]] = load i8*, i8** [[TMP15]], align 8 2200 // CHECK2-NEXT: [[TMP17:%.*]] = bitcast i8* [[TMP16]] to %struct.S* 2201 // CHECK2-NEXT: [[TMP18:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[TMP5]], i64 0, i64 2 2202 // CHECK2-NEXT: [[TMP19:%.*]] = load i8*, i8** [[TMP18]], align 8 2203 // CHECK2-NEXT: [[TMP20:%.*]] = bitcast i8* [[TMP19]] to %struct.S* 2204 // CHECK2-NEXT: [[TMP21:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[TMP3]], i64 0, i64 2 2205 // CHECK2-NEXT: [[TMP22:%.*]] = load i8*, i8** [[TMP21]], align 8 2206 // CHECK2-NEXT: [[TMP23:%.*]] = bitcast i8* [[TMP22]] to %struct.S* 2207 // CHECK2-NEXT: [[TMP24:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[TMP5]], i64 0, i64 3 2208 // CHECK2-NEXT: [[TMP25:%.*]] = load i8*, i8** [[TMP24]], align 8 2209 // CHECK2-NEXT: [[TMP26:%.*]] = bitcast i8* [[TMP25]] to float* 2210 // CHECK2-NEXT: [[TMP27:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[TMP3]], i64 0, i64 3 2211 // CHECK2-NEXT: [[TMP28:%.*]] = load i8*, i8** [[TMP27]], align 8 2212 // CHECK2-NEXT: [[TMP29:%.*]] = bitcast i8* [[TMP28]] to float* 2213 // CHECK2-NEXT: [[TMP30:%.*]] = load float, float* [[TMP11]], align 4 2214 // CHECK2-NEXT: [[TMP31:%.*]] = load float, float* [[TMP8]], align 4 2215 // CHECK2-NEXT: [[ADD:%.*]] = fadd float [[TMP30]], [[TMP31]] 2216 // CHECK2-NEXT: store float [[ADD]], float* [[TMP11]], align 4 2217 // CHECK2-NEXT: [[CALL:%.*]] = call nonnull align 4 dereferenceable(4) %struct.S* @_ZN1SIfEanERKS0_(%struct.S* nonnull align 4 dereferenceable(4) [[TMP17]], %struct.S* nonnull align 4 dereferenceable(4) [[TMP14]]) 2218 // CHECK2-NEXT: [[TMP32:%.*]] = bitcast %struct.S* [[TMP17]] to i8* 2219 // CHECK2-NEXT: [[TMP33:%.*]] = bitcast %struct.S* [[CALL]] to i8* 2220 // CHECK2-NEXT: call void @llvm.memcpy.p0i8.p0i8.i64(i8* align 4 [[TMP32]], i8* align 4 [[TMP33]], i64 4, i1 false) 2221 // CHECK2-NEXT: [[CALL2:%.*]] = call float @_ZN1SIfEcvfEv(%struct.S* nonnull align 4 dereferenceable(4) [[TMP23]]) 2222 // CHECK2-NEXT: [[TOBOOL:%.*]] = fcmp une float [[CALL2]], 0.000000e+00 2223 // CHECK2-NEXT: br i1 [[TOBOOL]], label [[LAND_RHS:%.*]], label [[LAND_END:%.*]] 2224 // CHECK2: land.rhs: 2225 // CHECK2-NEXT: [[CALL3:%.*]] = call float @_ZN1SIfEcvfEv(%struct.S* nonnull align 4 dereferenceable(4) [[TMP20]]) 2226 // CHECK2-NEXT: [[TOBOOL4:%.*]] = fcmp une float [[CALL3]], 0.000000e+00 2227 // CHECK2-NEXT: br label [[LAND_END]] 2228 // CHECK2: land.end: 2229 // CHECK2-NEXT: [[TMP34:%.*]] = phi i1 [ false, [[ENTRY:%.*]] ], [ [[TOBOOL4]], [[LAND_RHS]] ] 2230 // CHECK2-NEXT: [[CONV:%.*]] = uitofp i1 [[TMP34]] to float 2231 // CHECK2-NEXT: call void @_ZN1SIfEC1Ef(%struct.S* nonnull align 4 dereferenceable(4) [[REF_TMP]], float [[CONV]]) 2232 // CHECK2-NEXT: [[TMP35:%.*]] = bitcast %struct.S* [[TMP23]] to i8* 2233 // CHECK2-NEXT: [[TMP36:%.*]] = bitcast %struct.S* [[REF_TMP]] to i8* 2234 // CHECK2-NEXT: call void @llvm.memcpy.p0i8.p0i8.i64(i8* align 4 [[TMP35]], i8* align 4 [[TMP36]], i64 4, i1 false) 2235 // CHECK2-NEXT: call void @_ZN1SIfED1Ev(%struct.S* nonnull align 4 dereferenceable(4) [[REF_TMP]]) #[[ATTR5]] 2236 // CHECK2-NEXT: [[TMP37:%.*]] = load float, float* [[TMP29]], align 4 2237 // CHECK2-NEXT: [[TMP38:%.*]] = load float, float* [[TMP26]], align 4 2238 // CHECK2-NEXT: [[CMP:%.*]] = fcmp olt float [[TMP37]], [[TMP38]] 2239 // CHECK2-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] 2240 // CHECK2: cond.true: 2241 // CHECK2-NEXT: [[TMP39:%.*]] = load float, float* [[TMP29]], align 4 2242 // CHECK2-NEXT: br label [[COND_END:%.*]] 2243 // CHECK2: cond.false: 2244 // CHECK2-NEXT: [[TMP40:%.*]] = load float, float* [[TMP26]], align 4 2245 // CHECK2-NEXT: br label [[COND_END]] 2246 // CHECK2: cond.end: 2247 // CHECK2-NEXT: [[COND:%.*]] = phi float [ [[TMP39]], [[COND_TRUE]] ], [ [[TMP40]], [[COND_FALSE]] ] 2248 // CHECK2-NEXT: store float [[COND]], float* [[TMP29]], align 4 2249 // CHECK2-NEXT: ret void 2250 // 2251 // 2252 // CHECK2-LABEL: define {{[^@]+}}@_ZN1SIfEanERKS0_ 2253 // CHECK2-SAME: (%struct.S* nonnull align 4 dereferenceable(4) [[THIS:%.*]], %struct.S* nonnull align 4 dereferenceable(4) [[TMP0:%.*]]) #[[ATTR0]] align 2 { 2254 // CHECK2-NEXT: entry: 2255 // CHECK2-NEXT: [[THIS_ADDR:%.*]] = alloca %struct.S*, align 8 2256 // CHECK2-NEXT: [[DOTADDR:%.*]] = alloca %struct.S*, align 8 2257 // CHECK2-NEXT: store %struct.S* [[THIS]], %struct.S** [[THIS_ADDR]], align 8 2258 // CHECK2-NEXT: store %struct.S* [[TMP0]], %struct.S** [[DOTADDR]], align 8 2259 // CHECK2-NEXT: [[THIS1:%.*]] = load %struct.S*, %struct.S** [[THIS_ADDR]], align 8 2260 // CHECK2-NEXT: ret %struct.S* [[THIS1]] 2261 // 2262 // 2263 // CHECK2-LABEL: define {{[^@]+}}@_ZN1SIfEcvfEv 2264 // CHECK2-SAME: (%struct.S* nonnull align 4 dereferenceable(4) [[THIS:%.*]]) #[[ATTR0]] align 2 { 2265 // CHECK2-NEXT: entry: 2266 // CHECK2-NEXT: [[THIS_ADDR:%.*]] = alloca %struct.S*, align 8 2267 // CHECK2-NEXT: store %struct.S* [[THIS]], %struct.S** [[THIS_ADDR]], align 8 2268 // CHECK2-NEXT: [[THIS1:%.*]] = load %struct.S*, %struct.S** [[THIS_ADDR]], align 8 2269 // CHECK2-NEXT: ret float 0.000000e+00 2270 // 2271 // 2272 // CHECK2-LABEL: define {{[^@]+}}@_ZN1SIfED1Ev 2273 // CHECK2-SAME: (%struct.S* nonnull align 4 dereferenceable(4) [[THIS:%.*]]) unnamed_addr #[[ATTR7]] align 2 { 2274 // CHECK2-NEXT: entry: 2275 // CHECK2-NEXT: [[THIS_ADDR:%.*]] = alloca %struct.S*, align 8 2276 // CHECK2-NEXT: store %struct.S* [[THIS]], %struct.S** [[THIS_ADDR]], align 8 2277 // CHECK2-NEXT: [[THIS1:%.*]] = load %struct.S*, %struct.S** [[THIS_ADDR]], align 8 2278 // CHECK2-NEXT: call void @_ZN1SIfED2Ev(%struct.S* nonnull align 4 dereferenceable(4) [[THIS1]]) #[[ATTR5]] 2279 // CHECK2-NEXT: ret void 2280 // 2281 // 2282 // CHECK2-LABEL: define {{[^@]+}}@.omp_outlined..3 2283 // CHECK2-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], [2 x i32]* nonnull align 4 dereferenceable(8) [[VEC:%.*]], float* nonnull align 4 dereferenceable(4) [[T_VAR:%.*]], [2 x %struct.S]* nonnull align 4 dereferenceable(8) [[S_ARR:%.*]], %struct.S* nonnull align 4 dereferenceable(4) [[VAR:%.*]], %struct.S* nonnull align 4 dereferenceable(4) [[VAR1:%.*]], float* nonnull align 4 dereferenceable(4) [[T_VAR1:%.*]]) #[[ATTR1]] { 2284 // CHECK2-NEXT: entry: 2285 // CHECK2-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 2286 // CHECK2-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 2287 // CHECK2-NEXT: [[VEC_ADDR:%.*]] = alloca [2 x i32]*, align 8 2288 // CHECK2-NEXT: [[T_VAR_ADDR:%.*]] = alloca float*, align 8 2289 // CHECK2-NEXT: [[S_ARR_ADDR:%.*]] = alloca [2 x %struct.S]*, align 8 2290 // CHECK2-NEXT: [[VAR_ADDR:%.*]] = alloca %struct.S*, align 8 2291 // CHECK2-NEXT: [[VAR1_ADDR:%.*]] = alloca %struct.S*, align 8 2292 // CHECK2-NEXT: [[T_VAR1_ADDR:%.*]] = alloca float*, align 8 2293 // CHECK2-NEXT: [[T_VAR2:%.*]] = alloca float, align 4 2294 // CHECK2-NEXT: [[VAR3:%.*]] = alloca [[STRUCT_S:%.*]], align 4 2295 // CHECK2-NEXT: [[VAR14:%.*]] = alloca [[STRUCT_S]], align 4 2296 // CHECK2-NEXT: [[T_VAR15:%.*]] = alloca float, align 4 2297 // CHECK2-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8 2298 // CHECK2-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8 2299 // CHECK2-NEXT: store [2 x i32]* [[VEC]], [2 x i32]** [[VEC_ADDR]], align 8 2300 // CHECK2-NEXT: store float* [[T_VAR]], float** [[T_VAR_ADDR]], align 8 2301 // CHECK2-NEXT: store [2 x %struct.S]* [[S_ARR]], [2 x %struct.S]** [[S_ARR_ADDR]], align 8 2302 // CHECK2-NEXT: store %struct.S* [[VAR]], %struct.S** [[VAR_ADDR]], align 8 2303 // CHECK2-NEXT: store %struct.S* [[VAR1]], %struct.S** [[VAR1_ADDR]], align 8 2304 // CHECK2-NEXT: store float* [[T_VAR1]], float** [[T_VAR1_ADDR]], align 8 2305 // CHECK2-NEXT: [[TMP0:%.*]] = load [2 x i32]*, [2 x i32]** [[VEC_ADDR]], align 8 2306 // CHECK2-NEXT: [[TMP1:%.*]] = load float*, float** [[T_VAR_ADDR]], align 8 2307 // CHECK2-NEXT: [[TMP2:%.*]] = load [2 x %struct.S]*, [2 x %struct.S]** [[S_ARR_ADDR]], align 8 2308 // CHECK2-NEXT: [[TMP3:%.*]] = load %struct.S*, %struct.S** [[VAR_ADDR]], align 8 2309 // CHECK2-NEXT: [[TMP4:%.*]] = load %struct.S*, %struct.S** [[VAR1_ADDR]], align 8 2310 // CHECK2-NEXT: [[TMP5:%.*]] = load float*, float** [[T_VAR1_ADDR]], align 8 2311 // CHECK2-NEXT: store float 0.000000e+00, float* [[T_VAR2]], align 4 2312 // CHECK2-NEXT: call void @_ZN1SIfEC1Ev(%struct.S* nonnull align 4 dereferenceable(4) [[VAR3]]) 2313 // CHECK2-NEXT: call void @_ZN1SIfEC1Ev(%struct.S* nonnull align 4 dereferenceable(4) [[VAR14]]) 2314 // CHECK2-NEXT: store float 0x47EFFFFFE0000000, float* [[T_VAR15]], align 4 2315 // CHECK2-NEXT: br label [[WHILE_COND:%.*]] 2316 // CHECK2: while.cond: 2317 // CHECK2-NEXT: br label [[WHILE_BODY:%.*]] 2318 // CHECK2: while.body: 2319 // CHECK2-NEXT: [[TMP6:%.*]] = load float, float* [[T_VAR2]], align 4 2320 // CHECK2-NEXT: [[CONV:%.*]] = fptosi float [[TMP6]] to i32 2321 // CHECK2-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [2 x i32], [2 x i32]* [[TMP0]], i64 0, i64 0 2322 // CHECK2-NEXT: store i32 [[CONV]], i32* [[ARRAYIDX]], align 4 2323 // CHECK2-NEXT: [[ARRAYIDX6:%.*]] = getelementptr inbounds [2 x %struct.S], [2 x %struct.S]* [[TMP2]], i64 0, i64 0 2324 // CHECK2-NEXT: [[TMP7:%.*]] = bitcast %struct.S* [[ARRAYIDX6]] to i8* 2325 // CHECK2-NEXT: [[TMP8:%.*]] = bitcast %struct.S* [[VAR3]] to i8* 2326 // CHECK2-NEXT: call void @llvm.memcpy.p0i8.p0i8.i64(i8* align 4 [[TMP7]], i8* align 4 [[TMP8]], i64 4, i1 false) 2327 // CHECK2-NEXT: br label [[WHILE_COND]], !llvm.loop [[LOOP4:![0-9]+]] 2328 // 2329 // 2330 // CHECK2-LABEL: define {{[^@]+}}@.omp_outlined..4 2331 // CHECK2-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], { float, float }* nonnull align 4 dereferenceable(8) [[CF:%.*]]) #[[ATTR1]] { 2332 // CHECK2-NEXT: entry: 2333 // CHECK2-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 2334 // CHECK2-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 2335 // CHECK2-NEXT: [[CF_ADDR:%.*]] = alloca { float, float }*, align 8 2336 // CHECK2-NEXT: [[CF1:%.*]] = alloca { float, float }, align 4 2337 // CHECK2-NEXT: [[DOTOMP_REDUCTION_RED_LIST:%.*]] = alloca [1 x i8*], align 8 2338 // CHECK2-NEXT: [[ATOMIC_TEMP:%.*]] = alloca { float, float }, align 4 2339 // CHECK2-NEXT: [[ATOMIC_TEMP10:%.*]] = alloca { float, float }, align 4 2340 // CHECK2-NEXT: [[TMP:%.*]] = alloca { float, float }, align 4 2341 // CHECK2-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8 2342 // CHECK2-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8 2343 // CHECK2-NEXT: store { float, float }* [[CF]], { float, float }** [[CF_ADDR]], align 8 2344 // CHECK2-NEXT: [[TMP0:%.*]] = load { float, float }*, { float, float }** [[CF_ADDR]], align 8 2345 // CHECK2-NEXT: [[CF1_REALP:%.*]] = getelementptr inbounds { float, float }, { float, float }* [[CF1]], i32 0, i32 0 2346 // CHECK2-NEXT: [[CF1_IMAGP:%.*]] = getelementptr inbounds { float, float }, { float, float }* [[CF1]], i32 0, i32 1 2347 // CHECK2-NEXT: store float 0.000000e+00, float* [[CF1_REALP]], align 4 2348 // CHECK2-NEXT: store float 0.000000e+00, float* [[CF1_IMAGP]], align 4 2349 // CHECK2-NEXT: [[TMP1:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOMP_REDUCTION_RED_LIST]], i64 0, i64 0 2350 // CHECK2-NEXT: [[TMP2:%.*]] = bitcast { float, float }* [[CF1]] to i8* 2351 // CHECK2-NEXT: store i8* [[TMP2]], i8** [[TMP1]], align 8 2352 // CHECK2-NEXT: [[TMP3:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 2353 // CHECK2-NEXT: [[TMP4:%.*]] = load i32, i32* [[TMP3]], align 4 2354 // CHECK2-NEXT: [[TMP5:%.*]] = bitcast [1 x i8*]* [[DOTOMP_REDUCTION_RED_LIST]] to i8* 2355 // CHECK2-NEXT: [[TMP6:%.*]] = call i32 @__kmpc_reduce_nowait(%struct.ident_t* @[[GLOB1]], i32 [[TMP4]], i32 1, i64 8, i8* [[TMP5]], void (i8*, i8*)* @.omp.reduction.reduction_func.5, [8 x i32]* @.gomp_critical_user_.reduction.var) 2356 // CHECK2-NEXT: switch i32 [[TMP6]], label [[DOTOMP_REDUCTION_DEFAULT:%.*]] [ 2357 // CHECK2-NEXT: i32 1, label [[DOTOMP_REDUCTION_CASE1:%.*]] 2358 // CHECK2-NEXT: i32 2, label [[DOTOMP_REDUCTION_CASE2:%.*]] 2359 // CHECK2-NEXT: ] 2360 // CHECK2: .omp.reduction.case1: 2361 // CHECK2-NEXT: [[DOTREALP:%.*]] = getelementptr inbounds { float, float }, { float, float }* [[TMP0]], i32 0, i32 0 2362 // CHECK2-NEXT: [[DOTREAL:%.*]] = load float, float* [[DOTREALP]], align 4 2363 // CHECK2-NEXT: [[DOTIMAGP:%.*]] = getelementptr inbounds { float, float }, { float, float }* [[TMP0]], i32 0, i32 1 2364 // CHECK2-NEXT: [[DOTIMAG:%.*]] = load float, float* [[DOTIMAGP]], align 4 2365 // CHECK2-NEXT: [[CF1_REALP2:%.*]] = getelementptr inbounds { float, float }, { float, float }* [[CF1]], i32 0, i32 0 2366 // CHECK2-NEXT: [[CF1_REAL:%.*]] = load float, float* [[CF1_REALP2]], align 4 2367 // CHECK2-NEXT: [[CF1_IMAGP3:%.*]] = getelementptr inbounds { float, float }, { float, float }* [[CF1]], i32 0, i32 1 2368 // CHECK2-NEXT: [[CF1_IMAG:%.*]] = load float, float* [[CF1_IMAGP3]], align 4 2369 // CHECK2-NEXT: [[ADD_R:%.*]] = fadd float [[DOTREAL]], [[CF1_REAL]] 2370 // CHECK2-NEXT: [[ADD_I:%.*]] = fadd float [[DOTIMAG]], [[CF1_IMAG]] 2371 // CHECK2-NEXT: [[DOTREALP4:%.*]] = getelementptr inbounds { float, float }, { float, float }* [[TMP0]], i32 0, i32 0 2372 // CHECK2-NEXT: [[DOTIMAGP5:%.*]] = getelementptr inbounds { float, float }, { float, float }* [[TMP0]], i32 0, i32 1 2373 // CHECK2-NEXT: store float [[ADD_R]], float* [[DOTREALP4]], align 4 2374 // CHECK2-NEXT: store float [[ADD_I]], float* [[DOTIMAGP5]], align 4 2375 // CHECK2-NEXT: call void @__kmpc_end_reduce_nowait(%struct.ident_t* @[[GLOB1]], i32 [[TMP4]], [8 x i32]* @.gomp_critical_user_.reduction.var) 2376 // CHECK2-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]] 2377 // CHECK2: .omp.reduction.case2: 2378 // CHECK2-NEXT: [[CF1_REALP6:%.*]] = getelementptr inbounds { float, float }, { float, float }* [[CF1]], i32 0, i32 0 2379 // CHECK2-NEXT: [[CF1_REAL7:%.*]] = load float, float* [[CF1_REALP6]], align 4 2380 // CHECK2-NEXT: [[CF1_IMAGP8:%.*]] = getelementptr inbounds { float, float }, { float, float }* [[CF1]], i32 0, i32 1 2381 // CHECK2-NEXT: [[CF1_IMAG9:%.*]] = load float, float* [[CF1_IMAGP8]], align 4 2382 // CHECK2-NEXT: [[TMP7:%.*]] = bitcast { float, float }* [[TMP0]] to i8* 2383 // CHECK2-NEXT: [[TMP8:%.*]] = bitcast { float, float }* [[ATOMIC_TEMP]] to i8* 2384 // CHECK2-NEXT: call void @__atomic_load(i64 8, i8* [[TMP7]], i8* [[TMP8]], i32 0) 2385 // CHECK2-NEXT: br label [[ATOMIC_CONT:%.*]] 2386 // CHECK2: atomic_cont: 2387 // CHECK2-NEXT: [[ATOMIC_TEMP_REALP:%.*]] = getelementptr inbounds { float, float }, { float, float }* [[ATOMIC_TEMP]], i32 0, i32 0 2388 // CHECK2-NEXT: [[ATOMIC_TEMP_REAL:%.*]] = load float, float* [[ATOMIC_TEMP_REALP]], align 4 2389 // CHECK2-NEXT: [[ATOMIC_TEMP_IMAGP:%.*]] = getelementptr inbounds { float, float }, { float, float }* [[ATOMIC_TEMP]], i32 0, i32 1 2390 // CHECK2-NEXT: [[ATOMIC_TEMP_IMAG:%.*]] = load float, float* [[ATOMIC_TEMP_IMAGP]], align 4 2391 // CHECK2-NEXT: [[TMP_REALP:%.*]] = getelementptr inbounds { float, float }, { float, float }* [[TMP]], i32 0, i32 0 2392 // CHECK2-NEXT: [[TMP_IMAGP:%.*]] = getelementptr inbounds { float, float }, { float, float }* [[TMP]], i32 0, i32 1 2393 // CHECK2-NEXT: store float [[ATOMIC_TEMP_REAL]], float* [[TMP_REALP]], align 4 2394 // CHECK2-NEXT: store float [[ATOMIC_TEMP_IMAG]], float* [[TMP_IMAGP]], align 4 2395 // CHECK2-NEXT: [[TMP_REALP11:%.*]] = getelementptr inbounds { float, float }, { float, float }* [[TMP]], i32 0, i32 0 2396 // CHECK2-NEXT: [[TMP_REAL:%.*]] = load float, float* [[TMP_REALP11]], align 4 2397 // CHECK2-NEXT: [[TMP_IMAGP12:%.*]] = getelementptr inbounds { float, float }, { float, float }* [[TMP]], i32 0, i32 1 2398 // CHECK2-NEXT: [[TMP_IMAG:%.*]] = load float, float* [[TMP_IMAGP12]], align 4 2399 // CHECK2-NEXT: [[CF1_REALP13:%.*]] = getelementptr inbounds { float, float }, { float, float }* [[CF1]], i32 0, i32 0 2400 // CHECK2-NEXT: [[CF1_REAL14:%.*]] = load float, float* [[CF1_REALP13]], align 4 2401 // CHECK2-NEXT: [[CF1_IMAGP15:%.*]] = getelementptr inbounds { float, float }, { float, float }* [[CF1]], i32 0, i32 1 2402 // CHECK2-NEXT: [[CF1_IMAG16:%.*]] = load float, float* [[CF1_IMAGP15]], align 4 2403 // CHECK2-NEXT: [[ADD_R17:%.*]] = fadd float [[TMP_REAL]], [[CF1_REAL14]] 2404 // CHECK2-NEXT: [[ADD_I18:%.*]] = fadd float [[TMP_IMAG]], [[CF1_IMAG16]] 2405 // CHECK2-NEXT: [[ATOMIC_TEMP10_REALP:%.*]] = getelementptr inbounds { float, float }, { float, float }* [[ATOMIC_TEMP10]], i32 0, i32 0 2406 // CHECK2-NEXT: [[ATOMIC_TEMP10_IMAGP:%.*]] = getelementptr inbounds { float, float }, { float, float }* [[ATOMIC_TEMP10]], i32 0, i32 1 2407 // CHECK2-NEXT: store float [[ADD_R17]], float* [[ATOMIC_TEMP10_REALP]], align 4 2408 // CHECK2-NEXT: store float [[ADD_I18]], float* [[ATOMIC_TEMP10_IMAGP]], align 4 2409 // CHECK2-NEXT: [[TMP9:%.*]] = bitcast { float, float }* [[TMP0]] to i8* 2410 // CHECK2-NEXT: [[TMP10:%.*]] = bitcast { float, float }* [[ATOMIC_TEMP]] to i8* 2411 // CHECK2-NEXT: [[TMP11:%.*]] = bitcast { float, float }* [[ATOMIC_TEMP10]] to i8* 2412 // CHECK2-NEXT: [[CALL:%.*]] = call zeroext i1 @__atomic_compare_exchange(i64 8, i8* [[TMP9]], i8* [[TMP10]], i8* [[TMP11]], i32 0, i32 0) 2413 // CHECK2-NEXT: br i1 [[CALL]], label [[ATOMIC_EXIT:%.*]], label [[ATOMIC_CONT]] 2414 // CHECK2: atomic_exit: 2415 // CHECK2-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]] 2416 // CHECK2: .omp.reduction.default: 2417 // CHECK2-NEXT: ret void 2418 // 2419 // 2420 // CHECK2-LABEL: define {{[^@]+}}@.omp.reduction.reduction_func.5 2421 // CHECK2-SAME: (i8* [[TMP0:%.*]], i8* [[TMP1:%.*]]) #[[ATTR3]] { 2422 // CHECK2-NEXT: entry: 2423 // CHECK2-NEXT: [[DOTADDR:%.*]] = alloca i8*, align 8 2424 // CHECK2-NEXT: [[DOTADDR1:%.*]] = alloca i8*, align 8 2425 // CHECK2-NEXT: store i8* [[TMP0]], i8** [[DOTADDR]], align 8 2426 // CHECK2-NEXT: store i8* [[TMP1]], i8** [[DOTADDR1]], align 8 2427 // CHECK2-NEXT: [[TMP2:%.*]] = load i8*, i8** [[DOTADDR]], align 8 2428 // CHECK2-NEXT: [[TMP3:%.*]] = bitcast i8* [[TMP2]] to [1 x i8*]* 2429 // CHECK2-NEXT: [[TMP4:%.*]] = load i8*, i8** [[DOTADDR1]], align 8 2430 // CHECK2-NEXT: [[TMP5:%.*]] = bitcast i8* [[TMP4]] to [1 x i8*]* 2431 // CHECK2-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[TMP5]], i64 0, i64 0 2432 // CHECK2-NEXT: [[TMP7:%.*]] = load i8*, i8** [[TMP6]], align 8 2433 // CHECK2-NEXT: [[TMP8:%.*]] = bitcast i8* [[TMP7]] to { float, float }* 2434 // CHECK2-NEXT: [[TMP9:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[TMP3]], i64 0, i64 0 2435 // CHECK2-NEXT: [[TMP10:%.*]] = load i8*, i8** [[TMP9]], align 8 2436 // CHECK2-NEXT: [[TMP11:%.*]] = bitcast i8* [[TMP10]] to { float, float }* 2437 // CHECK2-NEXT: [[DOTREALP:%.*]] = getelementptr inbounds { float, float }, { float, float }* [[TMP11]], i32 0, i32 0 2438 // CHECK2-NEXT: [[DOTREAL:%.*]] = load float, float* [[DOTREALP]], align 4 2439 // CHECK2-NEXT: [[DOTIMAGP:%.*]] = getelementptr inbounds { float, float }, { float, float }* [[TMP11]], i32 0, i32 1 2440 // CHECK2-NEXT: [[DOTIMAG:%.*]] = load float, float* [[DOTIMAGP]], align 4 2441 // CHECK2-NEXT: [[DOTREALP2:%.*]] = getelementptr inbounds { float, float }, { float, float }* [[TMP8]], i32 0, i32 0 2442 // CHECK2-NEXT: [[DOTREAL3:%.*]] = load float, float* [[DOTREALP2]], align 4 2443 // CHECK2-NEXT: [[DOTIMAGP4:%.*]] = getelementptr inbounds { float, float }, { float, float }* [[TMP8]], i32 0, i32 1 2444 // CHECK2-NEXT: [[DOTIMAG5:%.*]] = load float, float* [[DOTIMAGP4]], align 4 2445 // CHECK2-NEXT: [[ADD_R:%.*]] = fadd float [[DOTREAL]], [[DOTREAL3]] 2446 // CHECK2-NEXT: [[ADD_I:%.*]] = fadd float [[DOTIMAG]], [[DOTIMAG5]] 2447 // CHECK2-NEXT: [[DOTREALP6:%.*]] = getelementptr inbounds { float, float }, { float, float }* [[TMP11]], i32 0, i32 0 2448 // CHECK2-NEXT: [[DOTIMAGP7:%.*]] = getelementptr inbounds { float, float }, { float, float }* [[TMP11]], i32 0, i32 1 2449 // CHECK2-NEXT: store float [[ADD_R]], float* [[DOTREALP6]], align 4 2450 // CHECK2-NEXT: store float [[ADD_I]], float* [[DOTIMAGP7]], align 4 2451 // CHECK2-NEXT: ret void 2452 // 2453 // 2454 // CHECK2-LABEL: define {{[^@]+}}@_Z5tmainIiET_v 2455 // CHECK2-SAME: () #[[ATTR0]] { 2456 // CHECK2-NEXT: entry: 2457 // CHECK2-NEXT: [[RETVAL:%.*]] = alloca i32, align 4 2458 // CHECK2-NEXT: [[T:%.*]] = alloca i32, align 4 2459 // CHECK2-NEXT: [[TEST:%.*]] = alloca [[STRUCT_S_0:%.*]], align 4 2460 // CHECK2-NEXT: [[SST:%.*]] = alloca [[STRUCT_SST:%.*]], align 4 2461 // CHECK2-NEXT: [[T_VAR:%.*]] = alloca i32, align 128 2462 // CHECK2-NEXT: [[T_VAR1:%.*]] = alloca i32, align 128 2463 // CHECK2-NEXT: [[VEC:%.*]] = alloca [2 x i32], align 4 2464 // CHECK2-NEXT: [[S_ARR:%.*]] = alloca [2 x %struct.S.0], align 4 2465 // CHECK2-NEXT: [[VAR:%.*]] = alloca [[STRUCT_S_0]], align 128 2466 // CHECK2-NEXT: [[VAR1:%.*]] = alloca [[STRUCT_S_0]], align 128 2467 // CHECK2-NEXT: call void @_ZN1SIiEC1Ev(%struct.S.0* nonnull align 4 dereferenceable(4) [[TEST]]) 2468 // CHECK2-NEXT: call void @_ZN3SSTIiEC1Ev(%struct.SST* nonnull align 4 dereferenceable(4) [[SST]]) 2469 // CHECK2-NEXT: store i32 0, i32* [[T_VAR]], align 128 2470 // CHECK2-NEXT: [[TMP0:%.*]] = bitcast [2 x i32]* [[VEC]] to i8* 2471 // CHECK2-NEXT: call void @llvm.memcpy.p0i8.p0i8.i64(i8* align 4 [[TMP0]], i8* align 4 bitcast ([2 x i32]* @__const._Z5tmainIiET_v.vec to i8*), i64 8, i1 false) 2472 // CHECK2-NEXT: [[ARRAYINIT_BEGIN:%.*]] = getelementptr inbounds [2 x %struct.S.0], [2 x %struct.S.0]* [[S_ARR]], i64 0, i64 0 2473 // CHECK2-NEXT: call void @_ZN1SIiEC1Ei(%struct.S.0* nonnull align 4 dereferenceable(4) [[ARRAYINIT_BEGIN]], i32 1) 2474 // CHECK2-NEXT: [[ARRAYINIT_ELEMENT:%.*]] = getelementptr inbounds [[STRUCT_S_0]], %struct.S.0* [[ARRAYINIT_BEGIN]], i64 1 2475 // CHECK2-NEXT: call void @_ZN1SIiEC1Ei(%struct.S.0* nonnull align 4 dereferenceable(4) [[ARRAYINIT_ELEMENT]], i32 2) 2476 // CHECK2-NEXT: call void @_ZN1SIiEC1Ei(%struct.S.0* nonnull align 4 dereferenceable(4) [[VAR]], i32 3) 2477 // CHECK2-NEXT: call void @_ZN1SIiEC1Ev(%struct.S.0* nonnull align 4 dereferenceable(4) [[VAR1]]) 2478 // CHECK2-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB2]], i32 6, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, [2 x i32]*, i32*, [2 x %struct.S.0]*, %struct.S.0*, %struct.S.0*, i32*)* @.omp_outlined..8 to void (i32*, i32*, ...)*), [2 x i32]* [[VEC]], i32* [[T_VAR]], [2 x %struct.S.0]* [[S_ARR]], %struct.S.0* [[VAR]], %struct.S.0* [[VAR1]], i32* [[T_VAR1]]) 2479 // CHECK2-NEXT: store i32 0, i32* [[RETVAL]], align 4 2480 // CHECK2-NEXT: call void @_ZN1SIiED1Ev(%struct.S.0* nonnull align 4 dereferenceable(4) [[VAR1]]) #[[ATTR5]] 2481 // CHECK2-NEXT: call void @_ZN1SIiED1Ev(%struct.S.0* nonnull align 4 dereferenceable(4) [[VAR]]) #[[ATTR5]] 2482 // CHECK2-NEXT: [[ARRAY_BEGIN:%.*]] = getelementptr inbounds [2 x %struct.S.0], [2 x %struct.S.0]* [[S_ARR]], i32 0, i32 0 2483 // CHECK2-NEXT: [[TMP1:%.*]] = getelementptr inbounds [[STRUCT_S_0]], %struct.S.0* [[ARRAY_BEGIN]], i64 2 2484 // CHECK2-NEXT: br label [[ARRAYDESTROY_BODY:%.*]] 2485 // CHECK2: arraydestroy.body: 2486 // CHECK2-NEXT: [[ARRAYDESTROY_ELEMENTPAST:%.*]] = phi %struct.S.0* [ [[TMP1]], [[ENTRY:%.*]] ], [ [[ARRAYDESTROY_ELEMENT:%.*]], [[ARRAYDESTROY_BODY]] ] 2487 // CHECK2-NEXT: [[ARRAYDESTROY_ELEMENT]] = getelementptr inbounds [[STRUCT_S_0]], %struct.S.0* [[ARRAYDESTROY_ELEMENTPAST]], i64 -1 2488 // CHECK2-NEXT: call void @_ZN1SIiED1Ev(%struct.S.0* nonnull align 4 dereferenceable(4) [[ARRAYDESTROY_ELEMENT]]) #[[ATTR5]] 2489 // CHECK2-NEXT: [[ARRAYDESTROY_DONE:%.*]] = icmp eq %struct.S.0* [[ARRAYDESTROY_ELEMENT]], [[ARRAY_BEGIN]] 2490 // CHECK2-NEXT: br i1 [[ARRAYDESTROY_DONE]], label [[ARRAYDESTROY_DONE1:%.*]], label [[ARRAYDESTROY_BODY]] 2491 // CHECK2: arraydestroy.done1: 2492 // CHECK2-NEXT: call void @_ZN1SIiED1Ev(%struct.S.0* nonnull align 4 dereferenceable(4) [[TEST]]) #[[ATTR5]] 2493 // CHECK2-NEXT: [[TMP2:%.*]] = load i32, i32* [[RETVAL]], align 4 2494 // CHECK2-NEXT: ret i32 [[TMP2]] 2495 // 2496 // 2497 // CHECK2-LABEL: define {{[^@]+}}@_ZN2SSC2ERi 2498 // CHECK2-SAME: (%struct.SS* nonnull align 8 dereferenceable(16) [[THIS:%.*]], i32* nonnull align 4 dereferenceable(4) [[D:%.*]]) unnamed_addr #[[ATTR7]] align 2 { 2499 // CHECK2-NEXT: entry: 2500 // CHECK2-NEXT: [[THIS_ADDR:%.*]] = alloca %struct.SS*, align 8 2501 // CHECK2-NEXT: [[D_ADDR:%.*]] = alloca i32*, align 8 2502 // CHECK2-NEXT: [[A2:%.*]] = alloca i32*, align 8 2503 // CHECK2-NEXT: [[B4:%.*]] = alloca i32, align 4 2504 // CHECK2-NEXT: [[C5:%.*]] = alloca i32*, align 8 2505 // CHECK2-NEXT: store %struct.SS* [[THIS]], %struct.SS** [[THIS_ADDR]], align 8 2506 // CHECK2-NEXT: store i32* [[D]], i32** [[D_ADDR]], align 8 2507 // CHECK2-NEXT: [[THIS1:%.*]] = load %struct.SS*, %struct.SS** [[THIS_ADDR]], align 8 2508 // CHECK2-NEXT: [[A:%.*]] = getelementptr inbounds [[STRUCT_SS:%.*]], %struct.SS* [[THIS1]], i32 0, i32 0 2509 // CHECK2-NEXT: store i32 0, i32* [[A]], align 8 2510 // CHECK2-NEXT: [[B:%.*]] = getelementptr inbounds [[STRUCT_SS]], %struct.SS* [[THIS1]], i32 0, i32 1 2511 // CHECK2-NEXT: [[BF_LOAD:%.*]] = load i8, i8* [[B]], align 4 2512 // CHECK2-NEXT: [[BF_CLEAR:%.*]] = and i8 [[BF_LOAD]], -16 2513 // CHECK2-NEXT: store i8 [[BF_CLEAR]], i8* [[B]], align 4 2514 // CHECK2-NEXT: [[C:%.*]] = getelementptr inbounds [[STRUCT_SS]], %struct.SS* [[THIS1]], i32 0, i32 2 2515 // CHECK2-NEXT: [[TMP0:%.*]] = load i32*, i32** [[D_ADDR]], align 8 2516 // CHECK2-NEXT: store i32* [[TMP0]], i32** [[C]], align 8 2517 // CHECK2-NEXT: [[A3:%.*]] = getelementptr inbounds [[STRUCT_SS]], %struct.SS* [[THIS1]], i32 0, i32 0 2518 // CHECK2-NEXT: store i32* [[A3]], i32** [[A2]], align 8 2519 // CHECK2-NEXT: [[C6:%.*]] = getelementptr inbounds [[STRUCT_SS]], %struct.SS* [[THIS1]], i32 0, i32 2 2520 // CHECK2-NEXT: [[TMP1:%.*]] = load i32*, i32** [[C6]], align 8 2521 // CHECK2-NEXT: store i32* [[TMP1]], i32** [[C5]], align 8 2522 // CHECK2-NEXT: [[TMP2:%.*]] = load i32*, i32** [[A2]], align 8 2523 // CHECK2-NEXT: [[TMP3:%.*]] = load i32*, i32** [[C5]], align 8 2524 // CHECK2-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB2]], i32 4, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, %struct.SS*, i32*, i32*, i32*)* @.omp_outlined..6 to void (i32*, i32*, ...)*), %struct.SS* [[THIS1]], i32* [[TMP2]], i32* [[B4]], i32* [[TMP3]]) 2525 // CHECK2-NEXT: [[TMP4:%.*]] = load i32, i32* [[B4]], align 4 2526 // CHECK2-NEXT: [[B7:%.*]] = getelementptr inbounds [[STRUCT_SS]], %struct.SS* [[THIS1]], i32 0, i32 1 2527 // CHECK2-NEXT: [[TMP5:%.*]] = trunc i32 [[TMP4]] to i8 2528 // CHECK2-NEXT: [[BF_LOAD8:%.*]] = load i8, i8* [[B7]], align 4 2529 // CHECK2-NEXT: [[BF_VALUE:%.*]] = and i8 [[TMP5]], 15 2530 // CHECK2-NEXT: [[BF_CLEAR9:%.*]] = and i8 [[BF_LOAD8]], -16 2531 // CHECK2-NEXT: [[BF_SET:%.*]] = or i8 [[BF_CLEAR9]], [[BF_VALUE]] 2532 // CHECK2-NEXT: store i8 [[BF_SET]], i8* [[B7]], align 4 2533 // CHECK2-NEXT: ret void 2534 // 2535 // 2536 // CHECK2-LABEL: define {{[^@]+}}@.omp_outlined..6 2537 // CHECK2-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], %struct.SS* [[THIS:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]], i32* nonnull align 4 dereferenceable(4) [[B:%.*]], i32* nonnull align 4 dereferenceable(4) [[C:%.*]]) #[[ATTR1]] { 2538 // CHECK2-NEXT: entry: 2539 // CHECK2-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 2540 // CHECK2-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 2541 // CHECK2-NEXT: [[THIS_ADDR:%.*]] = alloca %struct.SS*, align 8 2542 // CHECK2-NEXT: [[A_ADDR:%.*]] = alloca i32*, align 8 2543 // CHECK2-NEXT: [[B_ADDR:%.*]] = alloca i32*, align 8 2544 // CHECK2-NEXT: [[C_ADDR:%.*]] = alloca i32*, align 8 2545 // CHECK2-NEXT: [[TMP:%.*]] = alloca i32*, align 8 2546 // CHECK2-NEXT: [[_TMP1:%.*]] = alloca i32*, align 8 2547 // CHECK2-NEXT: [[A2:%.*]] = alloca i32, align 4 2548 // CHECK2-NEXT: [[_TMP3:%.*]] = alloca i32*, align 8 2549 // CHECK2-NEXT: [[B4:%.*]] = alloca i32, align 4 2550 // CHECK2-NEXT: [[C5:%.*]] = alloca i32, align 4 2551 // CHECK2-NEXT: [[_TMP6:%.*]] = alloca i32*, align 8 2552 // CHECK2-NEXT: [[DOTOMP_REDUCTION_RED_LIST:%.*]] = alloca [3 x i8*], align 8 2553 // CHECK2-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8 2554 // CHECK2-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8 2555 // CHECK2-NEXT: store %struct.SS* [[THIS]], %struct.SS** [[THIS_ADDR]], align 8 2556 // CHECK2-NEXT: store i32* [[A]], i32** [[A_ADDR]], align 8 2557 // CHECK2-NEXT: store i32* [[B]], i32** [[B_ADDR]], align 8 2558 // CHECK2-NEXT: store i32* [[C]], i32** [[C_ADDR]], align 8 2559 // CHECK2-NEXT: [[TMP0:%.*]] = load %struct.SS*, %struct.SS** [[THIS_ADDR]], align 8 2560 // CHECK2-NEXT: [[TMP1:%.*]] = load i32*, i32** [[A_ADDR]], align 8 2561 // CHECK2-NEXT: [[TMP2:%.*]] = load i32*, i32** [[B_ADDR]], align 8 2562 // CHECK2-NEXT: [[TMP3:%.*]] = load i32*, i32** [[C_ADDR]], align 8 2563 // CHECK2-NEXT: store i32* [[TMP1]], i32** [[TMP]], align 8 2564 // CHECK2-NEXT: store i32* [[TMP3]], i32** [[_TMP1]], align 8 2565 // CHECK2-NEXT: [[TMP4:%.*]] = load i32*, i32** [[TMP]], align 8 2566 // CHECK2-NEXT: store i32 0, i32* [[A2]], align 4 2567 // CHECK2-NEXT: store i32* [[A2]], i32** [[_TMP3]], align 8 2568 // CHECK2-NEXT: store i32 0, i32* [[B4]], align 4 2569 // CHECK2-NEXT: [[TMP5:%.*]] = load i32*, i32** [[_TMP1]], align 8 2570 // CHECK2-NEXT: store i32 0, i32* [[C5]], align 4 2571 // CHECK2-NEXT: store i32* [[C5]], i32** [[_TMP6]], align 8 2572 // CHECK2-NEXT: [[TMP6:%.*]] = load i32*, i32** [[_TMP3]], align 8 2573 // CHECK2-NEXT: [[TMP7:%.*]] = load i32, i32* [[TMP6]], align 4 2574 // CHECK2-NEXT: [[INC:%.*]] = add nsw i32 [[TMP7]], 1 2575 // CHECK2-NEXT: store i32 [[INC]], i32* [[TMP6]], align 4 2576 // CHECK2-NEXT: [[TMP8:%.*]] = load i32, i32* [[B4]], align 4 2577 // CHECK2-NEXT: [[DEC:%.*]] = add nsw i32 [[TMP8]], -1 2578 // CHECK2-NEXT: store i32 [[DEC]], i32* [[B4]], align 4 2579 // CHECK2-NEXT: [[TMP9:%.*]] = load i32*, i32** [[_TMP6]], align 8 2580 // CHECK2-NEXT: [[TMP10:%.*]] = load i32, i32* [[TMP9]], align 4 2581 // CHECK2-NEXT: [[DIV:%.*]] = sdiv i32 [[TMP10]], 1 2582 // CHECK2-NEXT: store i32 [[DIV]], i32* [[TMP9]], align 4 2583 // CHECK2-NEXT: [[TMP11:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOMP_REDUCTION_RED_LIST]], i64 0, i64 0 2584 // CHECK2-NEXT: [[TMP12:%.*]] = bitcast i32* [[A2]] to i8* 2585 // CHECK2-NEXT: store i8* [[TMP12]], i8** [[TMP11]], align 8 2586 // CHECK2-NEXT: [[TMP13:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOMP_REDUCTION_RED_LIST]], i64 0, i64 1 2587 // CHECK2-NEXT: [[TMP14:%.*]] = bitcast i32* [[B4]] to i8* 2588 // CHECK2-NEXT: store i8* [[TMP14]], i8** [[TMP13]], align 8 2589 // CHECK2-NEXT: [[TMP15:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOMP_REDUCTION_RED_LIST]], i64 0, i64 2 2590 // CHECK2-NEXT: [[TMP16:%.*]] = bitcast i32* [[C5]] to i8* 2591 // CHECK2-NEXT: store i8* [[TMP16]], i8** [[TMP15]], align 8 2592 // CHECK2-NEXT: [[TMP17:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 2593 // CHECK2-NEXT: [[TMP18:%.*]] = load i32, i32* [[TMP17]], align 4 2594 // CHECK2-NEXT: [[TMP19:%.*]] = bitcast [3 x i8*]* [[DOTOMP_REDUCTION_RED_LIST]] to i8* 2595 // CHECK2-NEXT: [[TMP20:%.*]] = call i32 @__kmpc_reduce_nowait(%struct.ident_t* @[[GLOB1]], i32 [[TMP18]], i32 3, i64 24, i8* [[TMP19]], void (i8*, i8*)* @.omp.reduction.reduction_func.7, [8 x i32]* @.gomp_critical_user_.reduction.var) 2596 // CHECK2-NEXT: switch i32 [[TMP20]], label [[DOTOMP_REDUCTION_DEFAULT:%.*]] [ 2597 // CHECK2-NEXT: i32 1, label [[DOTOMP_REDUCTION_CASE1:%.*]] 2598 // CHECK2-NEXT: i32 2, label [[DOTOMP_REDUCTION_CASE2:%.*]] 2599 // CHECK2-NEXT: ] 2600 // CHECK2: .omp.reduction.case1: 2601 // CHECK2-NEXT: [[TMP21:%.*]] = load i32, i32* [[TMP4]], align 4 2602 // CHECK2-NEXT: [[TMP22:%.*]] = load i32, i32* [[A2]], align 4 2603 // CHECK2-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP21]], [[TMP22]] 2604 // CHECK2-NEXT: store i32 [[ADD]], i32* [[TMP4]], align 4 2605 // CHECK2-NEXT: [[TMP23:%.*]] = load i32, i32* [[TMP2]], align 4 2606 // CHECK2-NEXT: [[TMP24:%.*]] = load i32, i32* [[B4]], align 4 2607 // CHECK2-NEXT: [[ADD7:%.*]] = add nsw i32 [[TMP23]], [[TMP24]] 2608 // CHECK2-NEXT: store i32 [[ADD7]], i32* [[TMP2]], align 4 2609 // CHECK2-NEXT: [[TMP25:%.*]] = load i32, i32* [[TMP5]], align 4 2610 // CHECK2-NEXT: [[TMP26:%.*]] = load i32, i32* [[C5]], align 4 2611 // CHECK2-NEXT: [[ADD8:%.*]] = add nsw i32 [[TMP25]], [[TMP26]] 2612 // CHECK2-NEXT: store i32 [[ADD8]], i32* [[TMP5]], align 4 2613 // CHECK2-NEXT: call void @__kmpc_end_reduce_nowait(%struct.ident_t* @[[GLOB1]], i32 [[TMP18]], [8 x i32]* @.gomp_critical_user_.reduction.var) 2614 // CHECK2-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]] 2615 // CHECK2: .omp.reduction.case2: 2616 // CHECK2-NEXT: [[TMP27:%.*]] = load i32, i32* [[A2]], align 4 2617 // CHECK2-NEXT: [[TMP28:%.*]] = atomicrmw add i32* [[TMP4]], i32 [[TMP27]] monotonic, align 4 2618 // CHECK2-NEXT: [[TMP29:%.*]] = load i32, i32* [[B4]], align 4 2619 // CHECK2-NEXT: [[TMP30:%.*]] = atomicrmw add i32* [[TMP2]], i32 [[TMP29]] monotonic, align 4 2620 // CHECK2-NEXT: [[TMP31:%.*]] = load i32, i32* [[C5]], align 4 2621 // CHECK2-NEXT: [[TMP32:%.*]] = atomicrmw add i32* [[TMP5]], i32 [[TMP31]] monotonic, align 4 2622 // CHECK2-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]] 2623 // CHECK2: .omp.reduction.default: 2624 // CHECK2-NEXT: ret void 2625 // 2626 // 2627 // CHECK2-LABEL: define {{[^@]+}}@.omp.reduction.reduction_func.7 2628 // CHECK2-SAME: (i8* [[TMP0:%.*]], i8* [[TMP1:%.*]]) #[[ATTR3]] { 2629 // CHECK2-NEXT: entry: 2630 // CHECK2-NEXT: [[DOTADDR:%.*]] = alloca i8*, align 8 2631 // CHECK2-NEXT: [[DOTADDR1:%.*]] = alloca i8*, align 8 2632 // CHECK2-NEXT: store i8* [[TMP0]], i8** [[DOTADDR]], align 8 2633 // CHECK2-NEXT: store i8* [[TMP1]], i8** [[DOTADDR1]], align 8 2634 // CHECK2-NEXT: [[TMP2:%.*]] = load i8*, i8** [[DOTADDR]], align 8 2635 // CHECK2-NEXT: [[TMP3:%.*]] = bitcast i8* [[TMP2]] to [3 x i8*]* 2636 // CHECK2-NEXT: [[TMP4:%.*]] = load i8*, i8** [[DOTADDR1]], align 8 2637 // CHECK2-NEXT: [[TMP5:%.*]] = bitcast i8* [[TMP4]] to [3 x i8*]* 2638 // CHECK2-NEXT: [[TMP6:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[TMP5]], i64 0, i64 0 2639 // CHECK2-NEXT: [[TMP7:%.*]] = load i8*, i8** [[TMP6]], align 8 2640 // CHECK2-NEXT: [[TMP8:%.*]] = bitcast i8* [[TMP7]] to i32* 2641 // CHECK2-NEXT: [[TMP9:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[TMP3]], i64 0, i64 0 2642 // CHECK2-NEXT: [[TMP10:%.*]] = load i8*, i8** [[TMP9]], align 8 2643 // CHECK2-NEXT: [[TMP11:%.*]] = bitcast i8* [[TMP10]] to i32* 2644 // CHECK2-NEXT: [[TMP12:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[TMP5]], i64 0, i64 1 2645 // CHECK2-NEXT: [[TMP13:%.*]] = load i8*, i8** [[TMP12]], align 8 2646 // CHECK2-NEXT: [[TMP14:%.*]] = bitcast i8* [[TMP13]] to i32* 2647 // CHECK2-NEXT: [[TMP15:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[TMP3]], i64 0, i64 1 2648 // CHECK2-NEXT: [[TMP16:%.*]] = load i8*, i8** [[TMP15]], align 8 2649 // CHECK2-NEXT: [[TMP17:%.*]] = bitcast i8* [[TMP16]] to i32* 2650 // CHECK2-NEXT: [[TMP18:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[TMP5]], i64 0, i64 2 2651 // CHECK2-NEXT: [[TMP19:%.*]] = load i8*, i8** [[TMP18]], align 8 2652 // CHECK2-NEXT: [[TMP20:%.*]] = bitcast i8* [[TMP19]] to i32* 2653 // CHECK2-NEXT: [[TMP21:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[TMP3]], i64 0, i64 2 2654 // CHECK2-NEXT: [[TMP22:%.*]] = load i8*, i8** [[TMP21]], align 8 2655 // CHECK2-NEXT: [[TMP23:%.*]] = bitcast i8* [[TMP22]] to i32* 2656 // CHECK2-NEXT: [[TMP24:%.*]] = load i32, i32* [[TMP11]], align 4 2657 // CHECK2-NEXT: [[TMP25:%.*]] = load i32, i32* [[TMP8]], align 4 2658 // CHECK2-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP24]], [[TMP25]] 2659 // CHECK2-NEXT: store i32 [[ADD]], i32* [[TMP11]], align 4 2660 // CHECK2-NEXT: [[TMP26:%.*]] = load i32, i32* [[TMP17]], align 4 2661 // CHECK2-NEXT: [[TMP27:%.*]] = load i32, i32* [[TMP14]], align 4 2662 // CHECK2-NEXT: [[ADD2:%.*]] = add nsw i32 [[TMP26]], [[TMP27]] 2663 // CHECK2-NEXT: store i32 [[ADD2]], i32* [[TMP17]], align 4 2664 // CHECK2-NEXT: [[TMP28:%.*]] = load i32, i32* [[TMP23]], align 4 2665 // CHECK2-NEXT: [[TMP29:%.*]] = load i32, i32* [[TMP20]], align 4 2666 // CHECK2-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP28]], [[TMP29]] 2667 // CHECK2-NEXT: store i32 [[ADD3]], i32* [[TMP23]], align 4 2668 // CHECK2-NEXT: ret void 2669 // 2670 // 2671 // CHECK2-LABEL: define {{[^@]+}}@_ZN1SIfEC2Ev 2672 // CHECK2-SAME: (%struct.S* nonnull align 4 dereferenceable(4) [[THIS:%.*]]) unnamed_addr #[[ATTR7]] align 2 { 2673 // CHECK2-NEXT: entry: 2674 // CHECK2-NEXT: [[THIS_ADDR:%.*]] = alloca %struct.S*, align 8 2675 // CHECK2-NEXT: store %struct.S* [[THIS]], %struct.S** [[THIS_ADDR]], align 8 2676 // CHECK2-NEXT: [[THIS1:%.*]] = load %struct.S*, %struct.S** [[THIS_ADDR]], align 8 2677 // CHECK2-NEXT: [[F:%.*]] = getelementptr inbounds [[STRUCT_S:%.*]], %struct.S* [[THIS1]], i32 0, i32 0 2678 // CHECK2-NEXT: [[TMP0:%.*]] = load volatile i32, i32* @g, align 128 2679 // CHECK2-NEXT: [[CONV:%.*]] = sitofp i32 [[TMP0]] to float 2680 // CHECK2-NEXT: store float [[CONV]], float* [[F]], align 4 2681 // CHECK2-NEXT: ret void 2682 // 2683 // 2684 // CHECK2-LABEL: define {{[^@]+}}@_ZN1SIfEC2Ef 2685 // CHECK2-SAME: (%struct.S* nonnull align 4 dereferenceable(4) [[THIS:%.*]], float [[A:%.*]]) unnamed_addr #[[ATTR7]] align 2 { 2686 // CHECK2-NEXT: entry: 2687 // CHECK2-NEXT: [[THIS_ADDR:%.*]] = alloca %struct.S*, align 8 2688 // CHECK2-NEXT: [[A_ADDR:%.*]] = alloca float, align 4 2689 // CHECK2-NEXT: store %struct.S* [[THIS]], %struct.S** [[THIS_ADDR]], align 8 2690 // CHECK2-NEXT: store float [[A]], float* [[A_ADDR]], align 4 2691 // CHECK2-NEXT: [[THIS1:%.*]] = load %struct.S*, %struct.S** [[THIS_ADDR]], align 8 2692 // CHECK2-NEXT: [[F:%.*]] = getelementptr inbounds [[STRUCT_S:%.*]], %struct.S* [[THIS1]], i32 0, i32 0 2693 // CHECK2-NEXT: [[TMP0:%.*]] = load float, float* [[A_ADDR]], align 4 2694 // CHECK2-NEXT: [[TMP1:%.*]] = load volatile i32, i32* @g, align 128 2695 // CHECK2-NEXT: [[CONV:%.*]] = sitofp i32 [[TMP1]] to float 2696 // CHECK2-NEXT: [[ADD:%.*]] = fadd float [[TMP0]], [[CONV]] 2697 // CHECK2-NEXT: store float [[ADD]], float* [[F]], align 4 2698 // CHECK2-NEXT: ret void 2699 // 2700 // 2701 // CHECK2-LABEL: define {{[^@]+}}@_ZN1SIfED2Ev 2702 // CHECK2-SAME: (%struct.S* nonnull align 4 dereferenceable(4) [[THIS:%.*]]) unnamed_addr #[[ATTR7]] align 2 { 2703 // CHECK2-NEXT: entry: 2704 // CHECK2-NEXT: [[THIS_ADDR:%.*]] = alloca %struct.S*, align 8 2705 // CHECK2-NEXT: store %struct.S* [[THIS]], %struct.S** [[THIS_ADDR]], align 8 2706 // CHECK2-NEXT: [[THIS1:%.*]] = load %struct.S*, %struct.S** [[THIS_ADDR]], align 8 2707 // CHECK2-NEXT: ret void 2708 // 2709 // 2710 // CHECK2-LABEL: define {{[^@]+}}@_ZN1SIiEC1Ev 2711 // CHECK2-SAME: (%struct.S.0* nonnull align 4 dereferenceable(4) [[THIS:%.*]]) unnamed_addr #[[ATTR7]] align 2 { 2712 // CHECK2-NEXT: entry: 2713 // CHECK2-NEXT: [[THIS_ADDR:%.*]] = alloca %struct.S.0*, align 8 2714 // CHECK2-NEXT: store %struct.S.0* [[THIS]], %struct.S.0** [[THIS_ADDR]], align 8 2715 // CHECK2-NEXT: [[THIS1:%.*]] = load %struct.S.0*, %struct.S.0** [[THIS_ADDR]], align 8 2716 // CHECK2-NEXT: call void @_ZN1SIiEC2Ev(%struct.S.0* nonnull align 4 dereferenceable(4) [[THIS1]]) 2717 // CHECK2-NEXT: ret void 2718 // 2719 // 2720 // CHECK2-LABEL: define {{[^@]+}}@_ZN3SSTIiEC1Ev 2721 // CHECK2-SAME: (%struct.SST* nonnull align 4 dereferenceable(4) [[THIS:%.*]]) unnamed_addr #[[ATTR7]] align 2 { 2722 // CHECK2-NEXT: entry: 2723 // CHECK2-NEXT: [[THIS_ADDR:%.*]] = alloca %struct.SST*, align 8 2724 // CHECK2-NEXT: store %struct.SST* [[THIS]], %struct.SST** [[THIS_ADDR]], align 8 2725 // CHECK2-NEXT: [[THIS1:%.*]] = load %struct.SST*, %struct.SST** [[THIS_ADDR]], align 8 2726 // CHECK2-NEXT: call void @_ZN3SSTIiEC2Ev(%struct.SST* nonnull align 4 dereferenceable(4) [[THIS1]]) 2727 // CHECK2-NEXT: ret void 2728 // 2729 // 2730 // CHECK2-LABEL: define {{[^@]+}}@_ZN1SIiEC1Ei 2731 // CHECK2-SAME: (%struct.S.0* nonnull align 4 dereferenceable(4) [[THIS:%.*]], i32 [[A:%.*]]) unnamed_addr #[[ATTR7]] align 2 { 2732 // CHECK2-NEXT: entry: 2733 // CHECK2-NEXT: [[THIS_ADDR:%.*]] = alloca %struct.S.0*, align 8 2734 // CHECK2-NEXT: [[A_ADDR:%.*]] = alloca i32, align 4 2735 // CHECK2-NEXT: store %struct.S.0* [[THIS]], %struct.S.0** [[THIS_ADDR]], align 8 2736 // CHECK2-NEXT: store i32 [[A]], i32* [[A_ADDR]], align 4 2737 // CHECK2-NEXT: [[THIS1:%.*]] = load %struct.S.0*, %struct.S.0** [[THIS_ADDR]], align 8 2738 // CHECK2-NEXT: [[TMP0:%.*]] = load i32, i32* [[A_ADDR]], align 4 2739 // CHECK2-NEXT: call void @_ZN1SIiEC2Ei(%struct.S.0* nonnull align 4 dereferenceable(4) [[THIS1]], i32 [[TMP0]]) 2740 // CHECK2-NEXT: ret void 2741 // 2742 // 2743 // CHECK2-LABEL: define {{[^@]+}}@.omp_outlined..8 2744 // CHECK2-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], [2 x i32]* nonnull align 4 dereferenceable(8) [[VEC:%.*]], i32* nonnull align 4 dereferenceable(4) [[T_VAR:%.*]], [2 x %struct.S.0]* nonnull align 4 dereferenceable(8) [[S_ARR:%.*]], %struct.S.0* nonnull align 4 dereferenceable(4) [[VAR:%.*]], %struct.S.0* nonnull align 4 dereferenceable(4) [[VAR1:%.*]], i32* nonnull align 4 dereferenceable(4) [[T_VAR1:%.*]]) #[[ATTR1]] { 2745 // CHECK2-NEXT: entry: 2746 // CHECK2-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 2747 // CHECK2-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 2748 // CHECK2-NEXT: [[VEC_ADDR:%.*]] = alloca [2 x i32]*, align 8 2749 // CHECK2-NEXT: [[T_VAR_ADDR:%.*]] = alloca i32*, align 8 2750 // CHECK2-NEXT: [[S_ARR_ADDR:%.*]] = alloca [2 x %struct.S.0]*, align 8 2751 // CHECK2-NEXT: [[VAR_ADDR:%.*]] = alloca %struct.S.0*, align 8 2752 // CHECK2-NEXT: [[VAR1_ADDR:%.*]] = alloca %struct.S.0*, align 8 2753 // CHECK2-NEXT: [[T_VAR1_ADDR:%.*]] = alloca i32*, align 8 2754 // CHECK2-NEXT: [[T_VAR2:%.*]] = alloca i32, align 128 2755 // CHECK2-NEXT: [[VAR3:%.*]] = alloca [[STRUCT_S_0:%.*]], align 128 2756 // CHECK2-NEXT: [[VAR14:%.*]] = alloca [[STRUCT_S_0]], align 128 2757 // CHECK2-NEXT: [[T_VAR15:%.*]] = alloca i32, align 128 2758 // CHECK2-NEXT: [[DOTOMP_REDUCTION_RED_LIST:%.*]] = alloca [4 x i8*], align 8 2759 // CHECK2-NEXT: [[REF_TMP:%.*]] = alloca [[STRUCT_S_0]], align 4 2760 // CHECK2-NEXT: [[REF_TMP11:%.*]] = alloca [[STRUCT_S_0]], align 4 2761 // CHECK2-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8 2762 // CHECK2-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8 2763 // CHECK2-NEXT: store [2 x i32]* [[VEC]], [2 x i32]** [[VEC_ADDR]], align 8 2764 // CHECK2-NEXT: store i32* [[T_VAR]], i32** [[T_VAR_ADDR]], align 8 2765 // CHECK2-NEXT: store [2 x %struct.S.0]* [[S_ARR]], [2 x %struct.S.0]** [[S_ARR_ADDR]], align 8 2766 // CHECK2-NEXT: store %struct.S.0* [[VAR]], %struct.S.0** [[VAR_ADDR]], align 8 2767 // CHECK2-NEXT: store %struct.S.0* [[VAR1]], %struct.S.0** [[VAR1_ADDR]], align 8 2768 // CHECK2-NEXT: store i32* [[T_VAR1]], i32** [[T_VAR1_ADDR]], align 8 2769 // CHECK2-NEXT: [[TMP0:%.*]] = load [2 x i32]*, [2 x i32]** [[VEC_ADDR]], align 8 2770 // CHECK2-NEXT: [[TMP1:%.*]] = load i32*, i32** [[T_VAR_ADDR]], align 8 2771 // CHECK2-NEXT: [[TMP2:%.*]] = load [2 x %struct.S.0]*, [2 x %struct.S.0]** [[S_ARR_ADDR]], align 8 2772 // CHECK2-NEXT: [[TMP3:%.*]] = load %struct.S.0*, %struct.S.0** [[VAR_ADDR]], align 8 2773 // CHECK2-NEXT: [[TMP4:%.*]] = load %struct.S.0*, %struct.S.0** [[VAR1_ADDR]], align 8 2774 // CHECK2-NEXT: [[TMP5:%.*]] = load i32*, i32** [[T_VAR1_ADDR]], align 8 2775 // CHECK2-NEXT: store i32 0, i32* [[T_VAR2]], align 128 2776 // CHECK2-NEXT: call void @_ZN1SIiEC1Ev(%struct.S.0* nonnull align 4 dereferenceable(4) [[VAR3]]) 2777 // CHECK2-NEXT: call void @_ZN1SIiEC1Ev(%struct.S.0* nonnull align 4 dereferenceable(4) [[VAR14]]) 2778 // CHECK2-NEXT: store i32 2147483647, i32* [[T_VAR15]], align 128 2779 // CHECK2-NEXT: [[TMP6:%.*]] = load i32, i32* [[T_VAR2]], align 128 2780 // CHECK2-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [2 x i32], [2 x i32]* [[TMP0]], i64 0, i64 0 2781 // CHECK2-NEXT: store i32 [[TMP6]], i32* [[ARRAYIDX]], align 4 2782 // CHECK2-NEXT: [[ARRAYIDX6:%.*]] = getelementptr inbounds [2 x %struct.S.0], [2 x %struct.S.0]* [[TMP2]], i64 0, i64 0 2783 // CHECK2-NEXT: [[TMP7:%.*]] = bitcast %struct.S.0* [[ARRAYIDX6]] to i8* 2784 // CHECK2-NEXT: [[TMP8:%.*]] = bitcast %struct.S.0* [[VAR3]] to i8* 2785 // CHECK2-NEXT: call void @llvm.memcpy.p0i8.p0i8.i64(i8* align 4 [[TMP7]], i8* align 128 [[TMP8]], i64 4, i1 false) 2786 // CHECK2-NEXT: [[TMP9:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOMP_REDUCTION_RED_LIST]], i64 0, i64 0 2787 // CHECK2-NEXT: [[TMP10:%.*]] = bitcast i32* [[T_VAR2]] to i8* 2788 // CHECK2-NEXT: store i8* [[TMP10]], i8** [[TMP9]], align 8 2789 // CHECK2-NEXT: [[TMP11:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOMP_REDUCTION_RED_LIST]], i64 0, i64 1 2790 // CHECK2-NEXT: [[TMP12:%.*]] = bitcast %struct.S.0* [[VAR3]] to i8* 2791 // CHECK2-NEXT: store i8* [[TMP12]], i8** [[TMP11]], align 8 2792 // CHECK2-NEXT: [[TMP13:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOMP_REDUCTION_RED_LIST]], i64 0, i64 2 2793 // CHECK2-NEXT: [[TMP14:%.*]] = bitcast %struct.S.0* [[VAR14]] to i8* 2794 // CHECK2-NEXT: store i8* [[TMP14]], i8** [[TMP13]], align 8 2795 // CHECK2-NEXT: [[TMP15:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOMP_REDUCTION_RED_LIST]], i64 0, i64 3 2796 // CHECK2-NEXT: [[TMP16:%.*]] = bitcast i32* [[T_VAR15]] to i8* 2797 // CHECK2-NEXT: store i8* [[TMP16]], i8** [[TMP15]], align 8 2798 // CHECK2-NEXT: [[TMP17:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 2799 // CHECK2-NEXT: [[TMP18:%.*]] = load i32, i32* [[TMP17]], align 4 2800 // CHECK2-NEXT: [[TMP19:%.*]] = bitcast [4 x i8*]* [[DOTOMP_REDUCTION_RED_LIST]] to i8* 2801 // CHECK2-NEXT: [[TMP20:%.*]] = call i32 @__kmpc_reduce_nowait(%struct.ident_t* @[[GLOB1]], i32 [[TMP18]], i32 4, i64 32, i8* [[TMP19]], void (i8*, i8*)* @.omp.reduction.reduction_func.9, [8 x i32]* @.gomp_critical_user_.reduction.var) 2802 // CHECK2-NEXT: switch i32 [[TMP20]], label [[DOTOMP_REDUCTION_DEFAULT:%.*]] [ 2803 // CHECK2-NEXT: i32 1, label [[DOTOMP_REDUCTION_CASE1:%.*]] 2804 // CHECK2-NEXT: i32 2, label [[DOTOMP_REDUCTION_CASE2:%.*]] 2805 // CHECK2-NEXT: ] 2806 // CHECK2: .omp.reduction.case1: 2807 // CHECK2-NEXT: [[TMP21:%.*]] = load i32, i32* [[TMP1]], align 128 2808 // CHECK2-NEXT: [[TMP22:%.*]] = load i32, i32* [[T_VAR2]], align 128 2809 // CHECK2-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP21]], [[TMP22]] 2810 // CHECK2-NEXT: store i32 [[ADD]], i32* [[TMP1]], align 128 2811 // CHECK2-NEXT: [[CALL:%.*]] = call nonnull align 4 dereferenceable(4) %struct.S.0* @_ZN1SIiEanERKS0_(%struct.S.0* nonnull align 4 dereferenceable(4) [[TMP3]], %struct.S.0* nonnull align 4 dereferenceable(4) [[VAR3]]) 2812 // CHECK2-NEXT: [[TMP23:%.*]] = bitcast %struct.S.0* [[TMP3]] to i8* 2813 // CHECK2-NEXT: [[TMP24:%.*]] = bitcast %struct.S.0* [[CALL]] to i8* 2814 // CHECK2-NEXT: call void @llvm.memcpy.p0i8.p0i8.i64(i8* align 128 [[TMP23]], i8* align 4 [[TMP24]], i64 4, i1 false) 2815 // CHECK2-NEXT: [[CALL7:%.*]] = call i32 @_ZN1SIiEcviEv(%struct.S.0* nonnull align 4 dereferenceable(4) [[TMP4]]) 2816 // CHECK2-NEXT: [[TOBOOL:%.*]] = icmp ne i32 [[CALL7]], 0 2817 // CHECK2-NEXT: br i1 [[TOBOOL]], label [[LAND_RHS:%.*]], label [[LAND_END:%.*]] 2818 // CHECK2: land.rhs: 2819 // CHECK2-NEXT: [[CALL8:%.*]] = call i32 @_ZN1SIiEcviEv(%struct.S.0* nonnull align 4 dereferenceable(4) [[VAR14]]) 2820 // CHECK2-NEXT: [[TOBOOL9:%.*]] = icmp ne i32 [[CALL8]], 0 2821 // CHECK2-NEXT: br label [[LAND_END]] 2822 // CHECK2: land.end: 2823 // CHECK2-NEXT: [[TMP25:%.*]] = phi i1 [ false, [[DOTOMP_REDUCTION_CASE1]] ], [ [[TOBOOL9]], [[LAND_RHS]] ] 2824 // CHECK2-NEXT: [[CONV:%.*]] = zext i1 [[TMP25]] to i32 2825 // CHECK2-NEXT: call void @_ZN1SIiEC1Ei(%struct.S.0* nonnull align 4 dereferenceable(4) [[REF_TMP]], i32 [[CONV]]) 2826 // CHECK2-NEXT: [[TMP26:%.*]] = bitcast %struct.S.0* [[TMP4]] to i8* 2827 // CHECK2-NEXT: [[TMP27:%.*]] = bitcast %struct.S.0* [[REF_TMP]] to i8* 2828 // CHECK2-NEXT: call void @llvm.memcpy.p0i8.p0i8.i64(i8* align 128 [[TMP26]], i8* align 4 [[TMP27]], i64 4, i1 false) 2829 // CHECK2-NEXT: call void @_ZN1SIiED1Ev(%struct.S.0* nonnull align 4 dereferenceable(4) [[REF_TMP]]) #[[ATTR5]] 2830 // CHECK2-NEXT: [[TMP28:%.*]] = load i32, i32* [[TMP5]], align 128 2831 // CHECK2-NEXT: [[TMP29:%.*]] = load i32, i32* [[T_VAR15]], align 128 2832 // CHECK2-NEXT: [[CMP:%.*]] = icmp slt i32 [[TMP28]], [[TMP29]] 2833 // CHECK2-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] 2834 // CHECK2: cond.true: 2835 // CHECK2-NEXT: [[TMP30:%.*]] = load i32, i32* [[TMP5]], align 128 2836 // CHECK2-NEXT: br label [[COND_END:%.*]] 2837 // CHECK2: cond.false: 2838 // CHECK2-NEXT: [[TMP31:%.*]] = load i32, i32* [[T_VAR15]], align 128 2839 // CHECK2-NEXT: br label [[COND_END]] 2840 // CHECK2: cond.end: 2841 // CHECK2-NEXT: [[COND:%.*]] = phi i32 [ [[TMP30]], [[COND_TRUE]] ], [ [[TMP31]], [[COND_FALSE]] ] 2842 // CHECK2-NEXT: store i32 [[COND]], i32* [[TMP5]], align 128 2843 // CHECK2-NEXT: call void @__kmpc_end_reduce_nowait(%struct.ident_t* @[[GLOB1]], i32 [[TMP18]], [8 x i32]* @.gomp_critical_user_.reduction.var) 2844 // CHECK2-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]] 2845 // CHECK2: .omp.reduction.case2: 2846 // CHECK2-NEXT: [[TMP32:%.*]] = load i32, i32* [[T_VAR2]], align 128 2847 // CHECK2-NEXT: [[TMP33:%.*]] = atomicrmw add i32* [[TMP1]], i32 [[TMP32]] monotonic, align 4 2848 // CHECK2-NEXT: call void @__kmpc_critical(%struct.ident_t* @[[GLOB2]], i32 [[TMP18]], [8 x i32]* @.gomp_critical_user_.atomic_reduction.var) 2849 // CHECK2-NEXT: [[CALL10:%.*]] = call nonnull align 4 dereferenceable(4) %struct.S.0* @_ZN1SIiEanERKS0_(%struct.S.0* nonnull align 4 dereferenceable(4) [[TMP3]], %struct.S.0* nonnull align 4 dereferenceable(4) [[VAR3]]) 2850 // CHECK2-NEXT: [[TMP34:%.*]] = bitcast %struct.S.0* [[TMP3]] to i8* 2851 // CHECK2-NEXT: [[TMP35:%.*]] = bitcast %struct.S.0* [[CALL10]] to i8* 2852 // CHECK2-NEXT: call void @llvm.memcpy.p0i8.p0i8.i64(i8* align 128 [[TMP34]], i8* align 4 [[TMP35]], i64 4, i1 false) 2853 // CHECK2-NEXT: call void @__kmpc_end_critical(%struct.ident_t* @[[GLOB2]], i32 [[TMP18]], [8 x i32]* @.gomp_critical_user_.atomic_reduction.var) 2854 // CHECK2-NEXT: call void @__kmpc_critical(%struct.ident_t* @[[GLOB2]], i32 [[TMP18]], [8 x i32]* @.gomp_critical_user_.atomic_reduction.var) 2855 // CHECK2-NEXT: [[CALL12:%.*]] = call i32 @_ZN1SIiEcviEv(%struct.S.0* nonnull align 4 dereferenceable(4) [[TMP4]]) 2856 // CHECK2-NEXT: [[TOBOOL13:%.*]] = icmp ne i32 [[CALL12]], 0 2857 // CHECK2-NEXT: br i1 [[TOBOOL13]], label [[LAND_RHS14:%.*]], label [[LAND_END17:%.*]] 2858 // CHECK2: land.rhs14: 2859 // CHECK2-NEXT: [[CALL15:%.*]] = call i32 @_ZN1SIiEcviEv(%struct.S.0* nonnull align 4 dereferenceable(4) [[VAR14]]) 2860 // CHECK2-NEXT: [[TOBOOL16:%.*]] = icmp ne i32 [[CALL15]], 0 2861 // CHECK2-NEXT: br label [[LAND_END17]] 2862 // CHECK2: land.end17: 2863 // CHECK2-NEXT: [[TMP36:%.*]] = phi i1 [ false, [[DOTOMP_REDUCTION_CASE2]] ], [ [[TOBOOL16]], [[LAND_RHS14]] ] 2864 // CHECK2-NEXT: [[CONV18:%.*]] = zext i1 [[TMP36]] to i32 2865 // CHECK2-NEXT: call void @_ZN1SIiEC1Ei(%struct.S.0* nonnull align 4 dereferenceable(4) [[REF_TMP11]], i32 [[CONV18]]) 2866 // CHECK2-NEXT: [[TMP37:%.*]] = bitcast %struct.S.0* [[TMP4]] to i8* 2867 // CHECK2-NEXT: [[TMP38:%.*]] = bitcast %struct.S.0* [[REF_TMP11]] to i8* 2868 // CHECK2-NEXT: call void @llvm.memcpy.p0i8.p0i8.i64(i8* align 128 [[TMP37]], i8* align 4 [[TMP38]], i64 4, i1 false) 2869 // CHECK2-NEXT: call void @_ZN1SIiED1Ev(%struct.S.0* nonnull align 4 dereferenceable(4) [[REF_TMP11]]) #[[ATTR5]] 2870 // CHECK2-NEXT: call void @__kmpc_end_critical(%struct.ident_t* @[[GLOB2]], i32 [[TMP18]], [8 x i32]* @.gomp_critical_user_.atomic_reduction.var) 2871 // CHECK2-NEXT: [[TMP39:%.*]] = load i32, i32* [[T_VAR15]], align 128 2872 // CHECK2-NEXT: [[TMP40:%.*]] = atomicrmw min i32* [[TMP5]], i32 [[TMP39]] monotonic, align 4 2873 // CHECK2-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]] 2874 // CHECK2: .omp.reduction.default: 2875 // CHECK2-NEXT: call void @_ZN1SIiED1Ev(%struct.S.0* nonnull align 4 dereferenceable(4) [[VAR14]]) #[[ATTR5]] 2876 // CHECK2-NEXT: call void @_ZN1SIiED1Ev(%struct.S.0* nonnull align 4 dereferenceable(4) [[VAR3]]) #[[ATTR5]] 2877 // CHECK2-NEXT: ret void 2878 // 2879 // 2880 // CHECK2-LABEL: define {{[^@]+}}@.omp.reduction.reduction_func.9 2881 // CHECK2-SAME: (i8* [[TMP0:%.*]], i8* [[TMP1:%.*]]) #[[ATTR3]] { 2882 // CHECK2-NEXT: entry: 2883 // CHECK2-NEXT: [[DOTADDR:%.*]] = alloca i8*, align 8 2884 // CHECK2-NEXT: [[DOTADDR1:%.*]] = alloca i8*, align 8 2885 // CHECK2-NEXT: [[REF_TMP:%.*]] = alloca [[STRUCT_S_0:%.*]], align 4 2886 // CHECK2-NEXT: store i8* [[TMP0]], i8** [[DOTADDR]], align 8 2887 // CHECK2-NEXT: store i8* [[TMP1]], i8** [[DOTADDR1]], align 8 2888 // CHECK2-NEXT: [[TMP2:%.*]] = load i8*, i8** [[DOTADDR]], align 8 2889 // CHECK2-NEXT: [[TMP3:%.*]] = bitcast i8* [[TMP2]] to [4 x i8*]* 2890 // CHECK2-NEXT: [[TMP4:%.*]] = load i8*, i8** [[DOTADDR1]], align 8 2891 // CHECK2-NEXT: [[TMP5:%.*]] = bitcast i8* [[TMP4]] to [4 x i8*]* 2892 // CHECK2-NEXT: [[TMP6:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[TMP5]], i64 0, i64 0 2893 // CHECK2-NEXT: [[TMP7:%.*]] = load i8*, i8** [[TMP6]], align 8 2894 // CHECK2-NEXT: [[TMP8:%.*]] = bitcast i8* [[TMP7]] to i32* 2895 // CHECK2-NEXT: [[TMP9:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[TMP3]], i64 0, i64 0 2896 // CHECK2-NEXT: [[TMP10:%.*]] = load i8*, i8** [[TMP9]], align 8 2897 // CHECK2-NEXT: [[TMP11:%.*]] = bitcast i8* [[TMP10]] to i32* 2898 // CHECK2-NEXT: [[TMP12:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[TMP5]], i64 0, i64 1 2899 // CHECK2-NEXT: [[TMP13:%.*]] = load i8*, i8** [[TMP12]], align 8 2900 // CHECK2-NEXT: [[TMP14:%.*]] = bitcast i8* [[TMP13]] to %struct.S.0* 2901 // CHECK2-NEXT: [[TMP15:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[TMP3]], i64 0, i64 1 2902 // CHECK2-NEXT: [[TMP16:%.*]] = load i8*, i8** [[TMP15]], align 8 2903 // CHECK2-NEXT: [[TMP17:%.*]] = bitcast i8* [[TMP16]] to %struct.S.0* 2904 // CHECK2-NEXT: [[TMP18:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[TMP5]], i64 0, i64 2 2905 // CHECK2-NEXT: [[TMP19:%.*]] = load i8*, i8** [[TMP18]], align 8 2906 // CHECK2-NEXT: [[TMP20:%.*]] = bitcast i8* [[TMP19]] to %struct.S.0* 2907 // CHECK2-NEXT: [[TMP21:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[TMP3]], i64 0, i64 2 2908 // CHECK2-NEXT: [[TMP22:%.*]] = load i8*, i8** [[TMP21]], align 8 2909 // CHECK2-NEXT: [[TMP23:%.*]] = bitcast i8* [[TMP22]] to %struct.S.0* 2910 // CHECK2-NEXT: [[TMP24:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[TMP5]], i64 0, i64 3 2911 // CHECK2-NEXT: [[TMP25:%.*]] = load i8*, i8** [[TMP24]], align 8 2912 // CHECK2-NEXT: [[TMP26:%.*]] = bitcast i8* [[TMP25]] to i32* 2913 // CHECK2-NEXT: [[TMP27:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[TMP3]], i64 0, i64 3 2914 // CHECK2-NEXT: [[TMP28:%.*]] = load i8*, i8** [[TMP27]], align 8 2915 // CHECK2-NEXT: [[TMP29:%.*]] = bitcast i8* [[TMP28]] to i32* 2916 // CHECK2-NEXT: [[TMP30:%.*]] = load i32, i32* [[TMP11]], align 128 2917 // CHECK2-NEXT: [[TMP31:%.*]] = load i32, i32* [[TMP8]], align 128 2918 // CHECK2-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP30]], [[TMP31]] 2919 // CHECK2-NEXT: store i32 [[ADD]], i32* [[TMP11]], align 128 2920 // CHECK2-NEXT: [[CALL:%.*]] = call nonnull align 4 dereferenceable(4) %struct.S.0* @_ZN1SIiEanERKS0_(%struct.S.0* nonnull align 4 dereferenceable(4) [[TMP17]], %struct.S.0* nonnull align 4 dereferenceable(4) [[TMP14]]) 2921 // CHECK2-NEXT: [[TMP32:%.*]] = bitcast %struct.S.0* [[TMP17]] to i8* 2922 // CHECK2-NEXT: [[TMP33:%.*]] = bitcast %struct.S.0* [[CALL]] to i8* 2923 // CHECK2-NEXT: call void @llvm.memcpy.p0i8.p0i8.i64(i8* align 128 [[TMP32]], i8* align 4 [[TMP33]], i64 4, i1 false) 2924 // CHECK2-NEXT: [[CALL2:%.*]] = call i32 @_ZN1SIiEcviEv(%struct.S.0* nonnull align 4 dereferenceable(4) [[TMP23]]) 2925 // CHECK2-NEXT: [[TOBOOL:%.*]] = icmp ne i32 [[CALL2]], 0 2926 // CHECK2-NEXT: br i1 [[TOBOOL]], label [[LAND_RHS:%.*]], label [[LAND_END:%.*]] 2927 // CHECK2: land.rhs: 2928 // CHECK2-NEXT: [[CALL3:%.*]] = call i32 @_ZN1SIiEcviEv(%struct.S.0* nonnull align 4 dereferenceable(4) [[TMP20]]) 2929 // CHECK2-NEXT: [[TOBOOL4:%.*]] = icmp ne i32 [[CALL3]], 0 2930 // CHECK2-NEXT: br label [[LAND_END]] 2931 // CHECK2: land.end: 2932 // CHECK2-NEXT: [[TMP34:%.*]] = phi i1 [ false, [[ENTRY:%.*]] ], [ [[TOBOOL4]], [[LAND_RHS]] ] 2933 // CHECK2-NEXT: [[CONV:%.*]] = zext i1 [[TMP34]] to i32 2934 // CHECK2-NEXT: call void @_ZN1SIiEC1Ei(%struct.S.0* nonnull align 4 dereferenceable(4) [[REF_TMP]], i32 [[CONV]]) 2935 // CHECK2-NEXT: [[TMP35:%.*]] = bitcast %struct.S.0* [[TMP23]] to i8* 2936 // CHECK2-NEXT: [[TMP36:%.*]] = bitcast %struct.S.0* [[REF_TMP]] to i8* 2937 // CHECK2-NEXT: call void @llvm.memcpy.p0i8.p0i8.i64(i8* align 128 [[TMP35]], i8* align 4 [[TMP36]], i64 4, i1 false) 2938 // CHECK2-NEXT: call void @_ZN1SIiED1Ev(%struct.S.0* nonnull align 4 dereferenceable(4) [[REF_TMP]]) #[[ATTR5]] 2939 // CHECK2-NEXT: [[TMP37:%.*]] = load i32, i32* [[TMP29]], align 128 2940 // CHECK2-NEXT: [[TMP38:%.*]] = load i32, i32* [[TMP26]], align 128 2941 // CHECK2-NEXT: [[CMP:%.*]] = icmp slt i32 [[TMP37]], [[TMP38]] 2942 // CHECK2-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] 2943 // CHECK2: cond.true: 2944 // CHECK2-NEXT: [[TMP39:%.*]] = load i32, i32* [[TMP29]], align 128 2945 // CHECK2-NEXT: br label [[COND_END:%.*]] 2946 // CHECK2: cond.false: 2947 // CHECK2-NEXT: [[TMP40:%.*]] = load i32, i32* [[TMP26]], align 128 2948 // CHECK2-NEXT: br label [[COND_END]] 2949 // CHECK2: cond.end: 2950 // CHECK2-NEXT: [[COND:%.*]] = phi i32 [ [[TMP39]], [[COND_TRUE]] ], [ [[TMP40]], [[COND_FALSE]] ] 2951 // CHECK2-NEXT: store i32 [[COND]], i32* [[TMP29]], align 128 2952 // CHECK2-NEXT: ret void 2953 // 2954 // 2955 // CHECK2-LABEL: define {{[^@]+}}@_ZN1SIiEanERKS0_ 2956 // CHECK2-SAME: (%struct.S.0* nonnull align 4 dereferenceable(4) [[THIS:%.*]], %struct.S.0* nonnull align 4 dereferenceable(4) [[TMP0:%.*]]) #[[ATTR0]] align 2 { 2957 // CHECK2-NEXT: entry: 2958 // CHECK2-NEXT: [[THIS_ADDR:%.*]] = alloca %struct.S.0*, align 8 2959 // CHECK2-NEXT: [[DOTADDR:%.*]] = alloca %struct.S.0*, align 8 2960 // CHECK2-NEXT: store %struct.S.0* [[THIS]], %struct.S.0** [[THIS_ADDR]], align 8 2961 // CHECK2-NEXT: store %struct.S.0* [[TMP0]], %struct.S.0** [[DOTADDR]], align 8 2962 // CHECK2-NEXT: [[THIS1:%.*]] = load %struct.S.0*, %struct.S.0** [[THIS_ADDR]], align 8 2963 // CHECK2-NEXT: ret %struct.S.0* [[THIS1]] 2964 // 2965 // 2966 // CHECK2-LABEL: define {{[^@]+}}@_ZN1SIiEcviEv 2967 // CHECK2-SAME: (%struct.S.0* nonnull align 4 dereferenceable(4) [[THIS:%.*]]) #[[ATTR0]] align 2 { 2968 // CHECK2-NEXT: entry: 2969 // CHECK2-NEXT: [[THIS_ADDR:%.*]] = alloca %struct.S.0*, align 8 2970 // CHECK2-NEXT: store %struct.S.0* [[THIS]], %struct.S.0** [[THIS_ADDR]], align 8 2971 // CHECK2-NEXT: [[THIS1:%.*]] = load %struct.S.0*, %struct.S.0** [[THIS_ADDR]], align 8 2972 // CHECK2-NEXT: ret i32 0 2973 // 2974 // 2975 // CHECK2-LABEL: define {{[^@]+}}@_ZN1SIiED1Ev 2976 // CHECK2-SAME: (%struct.S.0* nonnull align 4 dereferenceable(4) [[THIS:%.*]]) unnamed_addr #[[ATTR7]] align 2 { 2977 // CHECK2-NEXT: entry: 2978 // CHECK2-NEXT: [[THIS_ADDR:%.*]] = alloca %struct.S.0*, align 8 2979 // CHECK2-NEXT: store %struct.S.0* [[THIS]], %struct.S.0** [[THIS_ADDR]], align 8 2980 // CHECK2-NEXT: [[THIS1:%.*]] = load %struct.S.0*, %struct.S.0** [[THIS_ADDR]], align 8 2981 // CHECK2-NEXT: call void @_ZN1SIiED2Ev(%struct.S.0* nonnull align 4 dereferenceable(4) [[THIS1]]) #[[ATTR5]] 2982 // CHECK2-NEXT: ret void 2983 // 2984 // 2985 // CHECK2-LABEL: define {{[^@]+}}@_ZN1SIiEC2Ev 2986 // CHECK2-SAME: (%struct.S.0* nonnull align 4 dereferenceable(4) [[THIS:%.*]]) unnamed_addr #[[ATTR7]] align 2 { 2987 // CHECK2-NEXT: entry: 2988 // CHECK2-NEXT: [[THIS_ADDR:%.*]] = alloca %struct.S.0*, align 8 2989 // CHECK2-NEXT: store %struct.S.0* [[THIS]], %struct.S.0** [[THIS_ADDR]], align 8 2990 // CHECK2-NEXT: [[THIS1:%.*]] = load %struct.S.0*, %struct.S.0** [[THIS_ADDR]], align 8 2991 // CHECK2-NEXT: [[F:%.*]] = getelementptr inbounds [[STRUCT_S_0:%.*]], %struct.S.0* [[THIS1]], i32 0, i32 0 2992 // CHECK2-NEXT: [[TMP0:%.*]] = load volatile i32, i32* @g, align 128 2993 // CHECK2-NEXT: store i32 [[TMP0]], i32* [[F]], align 4 2994 // CHECK2-NEXT: ret void 2995 // 2996 // 2997 // CHECK2-LABEL: define {{[^@]+}}@_ZN3SSTIiEC2Ev 2998 // CHECK2-SAME: (%struct.SST* nonnull align 4 dereferenceable(4) [[THIS:%.*]]) unnamed_addr #[[ATTR7]] align 2 { 2999 // CHECK2-NEXT: entry: 3000 // CHECK2-NEXT: [[THIS_ADDR:%.*]] = alloca %struct.SST*, align 8 3001 // CHECK2-NEXT: [[A2:%.*]] = alloca i32*, align 8 3002 // CHECK2-NEXT: store %struct.SST* [[THIS]], %struct.SST** [[THIS_ADDR]], align 8 3003 // CHECK2-NEXT: [[THIS1:%.*]] = load %struct.SST*, %struct.SST** [[THIS_ADDR]], align 8 3004 // CHECK2-NEXT: [[A:%.*]] = getelementptr inbounds [[STRUCT_SST:%.*]], %struct.SST* [[THIS1]], i32 0, i32 0 3005 // CHECK2-NEXT: store i32 0, i32* [[A]], align 4 3006 // CHECK2-NEXT: [[A3:%.*]] = getelementptr inbounds [[STRUCT_SST]], %struct.SST* [[THIS1]], i32 0, i32 0 3007 // CHECK2-NEXT: store i32* [[A3]], i32** [[A2]], align 8 3008 // CHECK2-NEXT: [[TMP0:%.*]] = load i32*, i32** [[A2]], align 8 3009 // CHECK2-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB2]], i32 2, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, %struct.SST*, i32*)* @.omp_outlined..10 to void (i32*, i32*, ...)*), %struct.SST* [[THIS1]], i32* [[TMP0]]) 3010 // CHECK2-NEXT: ret void 3011 // 3012 // 3013 // CHECK2-LABEL: define {{[^@]+}}@.omp_outlined..10 3014 // CHECK2-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], %struct.SST* [[THIS:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR1]] { 3015 // CHECK2-NEXT: entry: 3016 // CHECK2-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 3017 // CHECK2-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 3018 // CHECK2-NEXT: [[THIS_ADDR:%.*]] = alloca %struct.SST*, align 8 3019 // CHECK2-NEXT: [[A_ADDR:%.*]] = alloca i32*, align 8 3020 // CHECK2-NEXT: [[TMP:%.*]] = alloca i32*, align 8 3021 // CHECK2-NEXT: [[A1:%.*]] = alloca i32, align 4 3022 // CHECK2-NEXT: [[_TMP2:%.*]] = alloca i32*, align 8 3023 // CHECK2-NEXT: [[DOTOMP_REDUCTION_RED_LIST:%.*]] = alloca [1 x i8*], align 8 3024 // CHECK2-NEXT: [[ATOMIC_TEMP:%.*]] = alloca i32, align 4 3025 // CHECK2-NEXT: [[_TMP3:%.*]] = alloca i32, align 4 3026 // CHECK2-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8 3027 // CHECK2-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8 3028 // CHECK2-NEXT: store %struct.SST* [[THIS]], %struct.SST** [[THIS_ADDR]], align 8 3029 // CHECK2-NEXT: store i32* [[A]], i32** [[A_ADDR]], align 8 3030 // CHECK2-NEXT: [[TMP0:%.*]] = load %struct.SST*, %struct.SST** [[THIS_ADDR]], align 8 3031 // CHECK2-NEXT: [[TMP1:%.*]] = load i32*, i32** [[A_ADDR]], align 8 3032 // CHECK2-NEXT: store i32* [[TMP1]], i32** [[TMP]], align 8 3033 // CHECK2-NEXT: [[TMP2:%.*]] = load i32*, i32** [[TMP]], align 8 3034 // CHECK2-NEXT: store i32 1, i32* [[A1]], align 4 3035 // CHECK2-NEXT: store i32* [[A1]], i32** [[_TMP2]], align 8 3036 // CHECK2-NEXT: [[TMP3:%.*]] = load i32*, i32** [[_TMP2]], align 8 3037 // CHECK2-NEXT: [[TMP4:%.*]] = load i32, i32* [[TMP3]], align 4 3038 // CHECK2-NEXT: [[INC:%.*]] = add nsw i32 [[TMP4]], 1 3039 // CHECK2-NEXT: store i32 [[INC]], i32* [[TMP3]], align 4 3040 // CHECK2-NEXT: [[TMP5:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOMP_REDUCTION_RED_LIST]], i64 0, i64 0 3041 // CHECK2-NEXT: [[TMP6:%.*]] = bitcast i32* [[A1]] to i8* 3042 // CHECK2-NEXT: store i8* [[TMP6]], i8** [[TMP5]], align 8 3043 // CHECK2-NEXT: [[TMP7:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 3044 // CHECK2-NEXT: [[TMP8:%.*]] = load i32, i32* [[TMP7]], align 4 3045 // CHECK2-NEXT: [[TMP9:%.*]] = bitcast [1 x i8*]* [[DOTOMP_REDUCTION_RED_LIST]] to i8* 3046 // CHECK2-NEXT: [[TMP10:%.*]] = call i32 @__kmpc_reduce_nowait(%struct.ident_t* @[[GLOB1]], i32 [[TMP8]], i32 1, i64 8, i8* [[TMP9]], void (i8*, i8*)* @.omp.reduction.reduction_func.11, [8 x i32]* @.gomp_critical_user_.reduction.var) 3047 // CHECK2-NEXT: switch i32 [[TMP10]], label [[DOTOMP_REDUCTION_DEFAULT:%.*]] [ 3048 // CHECK2-NEXT: i32 1, label [[DOTOMP_REDUCTION_CASE1:%.*]] 3049 // CHECK2-NEXT: i32 2, label [[DOTOMP_REDUCTION_CASE2:%.*]] 3050 // CHECK2-NEXT: ] 3051 // CHECK2: .omp.reduction.case1: 3052 // CHECK2-NEXT: [[TMP11:%.*]] = load i32, i32* [[TMP2]], align 4 3053 // CHECK2-NEXT: [[TMP12:%.*]] = load i32, i32* [[A1]], align 4 3054 // CHECK2-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP11]], [[TMP12]] 3055 // CHECK2-NEXT: store i32 [[MUL]], i32* [[TMP2]], align 4 3056 // CHECK2-NEXT: call void @__kmpc_end_reduce_nowait(%struct.ident_t* @[[GLOB1]], i32 [[TMP8]], [8 x i32]* @.gomp_critical_user_.reduction.var) 3057 // CHECK2-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]] 3058 // CHECK2: .omp.reduction.case2: 3059 // CHECK2-NEXT: [[TMP13:%.*]] = load i32, i32* [[A1]], align 4 3060 // CHECK2-NEXT: [[ATOMIC_LOAD:%.*]] = load atomic i32, i32* [[TMP2]] monotonic, align 4 3061 // CHECK2-NEXT: br label [[ATOMIC_CONT:%.*]] 3062 // CHECK2: atomic_cont: 3063 // CHECK2-NEXT: [[TMP14:%.*]] = phi i32 [ [[ATOMIC_LOAD]], [[DOTOMP_REDUCTION_CASE2]] ], [ [[TMP19:%.*]], [[ATOMIC_CONT]] ] 3064 // CHECK2-NEXT: store i32 [[TMP14]], i32* [[_TMP3]], align 4 3065 // CHECK2-NEXT: [[TMP15:%.*]] = load i32, i32* [[_TMP3]], align 4 3066 // CHECK2-NEXT: [[TMP16:%.*]] = load i32, i32* [[A1]], align 4 3067 // CHECK2-NEXT: [[MUL4:%.*]] = mul nsw i32 [[TMP15]], [[TMP16]] 3068 // CHECK2-NEXT: store i32 [[MUL4]], i32* [[ATOMIC_TEMP]], align 4 3069 // CHECK2-NEXT: [[TMP17:%.*]] = load i32, i32* [[ATOMIC_TEMP]], align 4 3070 // CHECK2-NEXT: [[TMP18:%.*]] = cmpxchg i32* [[TMP2]], i32 [[TMP14]], i32 [[TMP17]] monotonic monotonic, align 4 3071 // CHECK2-NEXT: [[TMP19]] = extractvalue { i32, i1 } [[TMP18]], 0 3072 // CHECK2-NEXT: [[TMP20:%.*]] = extractvalue { i32, i1 } [[TMP18]], 1 3073 // CHECK2-NEXT: br i1 [[TMP20]], label [[ATOMIC_EXIT:%.*]], label [[ATOMIC_CONT]] 3074 // CHECK2: atomic_exit: 3075 // CHECK2-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]] 3076 // CHECK2: .omp.reduction.default: 3077 // CHECK2-NEXT: ret void 3078 // 3079 // 3080 // CHECK2-LABEL: define {{[^@]+}}@.omp.reduction.reduction_func.11 3081 // CHECK2-SAME: (i8* [[TMP0:%.*]], i8* [[TMP1:%.*]]) #[[ATTR3]] { 3082 // CHECK2-NEXT: entry: 3083 // CHECK2-NEXT: [[DOTADDR:%.*]] = alloca i8*, align 8 3084 // CHECK2-NEXT: [[DOTADDR1:%.*]] = alloca i8*, align 8 3085 // CHECK2-NEXT: store i8* [[TMP0]], i8** [[DOTADDR]], align 8 3086 // CHECK2-NEXT: store i8* [[TMP1]], i8** [[DOTADDR1]], align 8 3087 // CHECK2-NEXT: [[TMP2:%.*]] = load i8*, i8** [[DOTADDR]], align 8 3088 // CHECK2-NEXT: [[TMP3:%.*]] = bitcast i8* [[TMP2]] to [1 x i8*]* 3089 // CHECK2-NEXT: [[TMP4:%.*]] = load i8*, i8** [[DOTADDR1]], align 8 3090 // CHECK2-NEXT: [[TMP5:%.*]] = bitcast i8* [[TMP4]] to [1 x i8*]* 3091 // CHECK2-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[TMP5]], i64 0, i64 0 3092 // CHECK2-NEXT: [[TMP7:%.*]] = load i8*, i8** [[TMP6]], align 8 3093 // CHECK2-NEXT: [[TMP8:%.*]] = bitcast i8* [[TMP7]] to i32* 3094 // CHECK2-NEXT: [[TMP9:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[TMP3]], i64 0, i64 0 3095 // CHECK2-NEXT: [[TMP10:%.*]] = load i8*, i8** [[TMP9]], align 8 3096 // CHECK2-NEXT: [[TMP11:%.*]] = bitcast i8* [[TMP10]] to i32* 3097 // CHECK2-NEXT: [[TMP12:%.*]] = load i32, i32* [[TMP11]], align 4 3098 // CHECK2-NEXT: [[TMP13:%.*]] = load i32, i32* [[TMP8]], align 4 3099 // CHECK2-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP12]], [[TMP13]] 3100 // CHECK2-NEXT: store i32 [[MUL]], i32* [[TMP11]], align 4 3101 // CHECK2-NEXT: ret void 3102 // 3103 // 3104 // CHECK2-LABEL: define {{[^@]+}}@_ZN1SIiEC2Ei 3105 // CHECK2-SAME: (%struct.S.0* nonnull align 4 dereferenceable(4) [[THIS:%.*]], i32 [[A:%.*]]) unnamed_addr #[[ATTR7]] align 2 { 3106 // CHECK2-NEXT: entry: 3107 // CHECK2-NEXT: [[THIS_ADDR:%.*]] = alloca %struct.S.0*, align 8 3108 // CHECK2-NEXT: [[A_ADDR:%.*]] = alloca i32, align 4 3109 // CHECK2-NEXT: store %struct.S.0* [[THIS]], %struct.S.0** [[THIS_ADDR]], align 8 3110 // CHECK2-NEXT: store i32 [[A]], i32* [[A_ADDR]], align 4 3111 // CHECK2-NEXT: [[THIS1:%.*]] = load %struct.S.0*, %struct.S.0** [[THIS_ADDR]], align 8 3112 // CHECK2-NEXT: [[F:%.*]] = getelementptr inbounds [[STRUCT_S_0:%.*]], %struct.S.0* [[THIS1]], i32 0, i32 0 3113 // CHECK2-NEXT: [[TMP0:%.*]] = load i32, i32* [[A_ADDR]], align 4 3114 // CHECK2-NEXT: [[TMP1:%.*]] = load volatile i32, i32* @g, align 128 3115 // CHECK2-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP0]], [[TMP1]] 3116 // CHECK2-NEXT: store i32 [[ADD]], i32* [[F]], align 4 3117 // CHECK2-NEXT: ret void 3118 // 3119 // 3120 // CHECK2-LABEL: define {{[^@]+}}@_ZN1SIiED2Ev 3121 // CHECK2-SAME: (%struct.S.0* nonnull align 4 dereferenceable(4) [[THIS:%.*]]) unnamed_addr #[[ATTR7]] align 2 { 3122 // CHECK2-NEXT: entry: 3123 // CHECK2-NEXT: [[THIS_ADDR:%.*]] = alloca %struct.S.0*, align 8 3124 // CHECK2-NEXT: store %struct.S.0* [[THIS]], %struct.S.0** [[THIS_ADDR]], align 8 3125 // CHECK2-NEXT: [[THIS1:%.*]] = load %struct.S.0*, %struct.S.0** [[THIS_ADDR]], align 8 3126 // CHECK2-NEXT: ret void 3127 // 3128 // 3129 // CHECK3-LABEL: define {{[^@]+}}@_Z14foo_array_sectPs 3130 // CHECK3-SAME: (i16* [[X:%.*]]) #[[ATTR0:[0-9]+]] { 3131 // CHECK3-NEXT: entry: 3132 // CHECK3-NEXT: [[X_ADDR:%.*]] = alloca i16*, align 8 3133 // CHECK3-NEXT: store i16* [[X]], i16** [[X_ADDR]], align 8 3134 // CHECK3-NEXT: [[TMP0:%.*]] = load i16*, i16** [[X_ADDR]], align 8 3135 // CHECK3-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB2:[0-9]+]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i16*)* @.omp_outlined. to void (i32*, i32*, ...)*), i16* [[TMP0]]) 3136 // CHECK3-NEXT: ret void 3137 // 3138 // 3139 // CHECK3-LABEL: define {{[^@]+}}@.omp_outlined. 3140 // CHECK3-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i16* [[X:%.*]]) #[[ATTR1:[0-9]+]] { 3141 // CHECK3-NEXT: entry: 3142 // CHECK3-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 3143 // CHECK3-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 3144 // CHECK3-NEXT: [[X_ADDR:%.*]] = alloca i16*, align 8 3145 // CHECK3-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 8 3146 // CHECK3-NEXT: [[__VLA_EXPR0:%.*]] = alloca i64, align 8 3147 // CHECK3-NEXT: [[TMP:%.*]] = alloca i16*, align 8 3148 // CHECK3-NEXT: [[DOTOMP_REDUCTION_RED_LIST:%.*]] = alloca [2 x i8*], align 8 3149 // CHECK3-NEXT: [[ATOMIC_TEMP:%.*]] = alloca i16, align 2 3150 // CHECK3-NEXT: [[_TMP13:%.*]] = alloca i16, align 2 3151 // CHECK3-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8 3152 // CHECK3-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8 3153 // CHECK3-NEXT: store i16* [[X]], i16** [[X_ADDR]], align 8 3154 // CHECK3-NEXT: [[TMP0:%.*]] = load i16*, i16** [[X_ADDR]], align 8 3155 // CHECK3-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i16, i16* [[TMP0]], i64 0 3156 // CHECK3-NEXT: [[TMP1:%.*]] = load i16*, i16** [[X_ADDR]], align 8 3157 // CHECK3-NEXT: [[ARRAYIDX1:%.*]] = getelementptr inbounds i16, i16* [[TMP1]], i64 0 3158 // CHECK3-NEXT: [[TMP2:%.*]] = ptrtoint i16* [[ARRAYIDX1]] to i64 3159 // CHECK3-NEXT: [[TMP3:%.*]] = ptrtoint i16* [[ARRAYIDX]] to i64 3160 // CHECK3-NEXT: [[TMP4:%.*]] = sub i64 [[TMP2]], [[TMP3]] 3161 // CHECK3-NEXT: [[TMP5:%.*]] = sdiv exact i64 [[TMP4]], ptrtoint (i16* getelementptr (i16, i16* null, i32 1) to i64) 3162 // CHECK3-NEXT: [[TMP6:%.*]] = add nuw i64 [[TMP5]], 1 3163 // CHECK3-NEXT: [[TMP7:%.*]] = mul nuw i64 [[TMP6]], ptrtoint (i16* getelementptr (i16, i16* null, i32 1) to i64) 3164 // CHECK3-NEXT: [[TMP8:%.*]] = call i8* @llvm.stacksave() 3165 // CHECK3-NEXT: store i8* [[TMP8]], i8** [[SAVED_STACK]], align 8 3166 // CHECK3-NEXT: [[VLA:%.*]] = alloca i16, i64 [[TMP6]], align 16 3167 // CHECK3-NEXT: store i64 [[TMP6]], i64* [[__VLA_EXPR0]], align 8 3168 // CHECK3-NEXT: [[TMP9:%.*]] = getelementptr i16, i16* [[VLA]], i64 [[TMP6]] 3169 // CHECK3-NEXT: [[OMP_ARRAYINIT_ISEMPTY:%.*]] = icmp eq i16* [[VLA]], [[TMP9]] 3170 // CHECK3-NEXT: br i1 [[OMP_ARRAYINIT_ISEMPTY]], label [[OMP_ARRAYINIT_DONE:%.*]], label [[OMP_ARRAYINIT_BODY:%.*]] 3171 // CHECK3: omp.arrayinit.body: 3172 // CHECK3-NEXT: [[OMP_ARRAYCPY_DESTELEMENTPAST:%.*]] = phi i16* [ [[VLA]], [[ENTRY:%.*]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT:%.*]], [[OMP_ARRAYINIT_BODY]] ] 3173 // CHECK3-NEXT: store i16 0, i16* [[OMP_ARRAYCPY_DESTELEMENTPAST]], align 2 3174 // CHECK3-NEXT: [[OMP_ARRAYCPY_DEST_ELEMENT]] = getelementptr i16, i16* [[OMP_ARRAYCPY_DESTELEMENTPAST]], i32 1 3175 // CHECK3-NEXT: [[OMP_ARRAYCPY_DONE:%.*]] = icmp eq i16* [[OMP_ARRAYCPY_DEST_ELEMENT]], [[TMP9]] 3176 // CHECK3-NEXT: br i1 [[OMP_ARRAYCPY_DONE]], label [[OMP_ARRAYINIT_DONE]], label [[OMP_ARRAYINIT_BODY]] 3177 // CHECK3: omp.arrayinit.done: 3178 // CHECK3-NEXT: [[TMP10:%.*]] = load i16*, i16** [[X_ADDR]], align 8 3179 // CHECK3-NEXT: [[TMP11:%.*]] = ptrtoint i16* [[TMP10]] to i64 3180 // CHECK3-NEXT: [[TMP12:%.*]] = ptrtoint i16* [[ARRAYIDX]] to i64 3181 // CHECK3-NEXT: [[TMP13:%.*]] = sub i64 [[TMP11]], [[TMP12]] 3182 // CHECK3-NEXT: [[TMP14:%.*]] = sdiv exact i64 [[TMP13]], ptrtoint (i16* getelementptr (i16, i16* null, i32 1) to i64) 3183 // CHECK3-NEXT: [[TMP15:%.*]] = getelementptr i16, i16* [[VLA]], i64 [[TMP14]] 3184 // CHECK3-NEXT: store i16* [[TMP15]], i16** [[TMP]], align 8 3185 // CHECK3-NEXT: [[TMP16:%.*]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[DOTOMP_REDUCTION_RED_LIST]], i64 0, i64 0 3186 // CHECK3-NEXT: [[TMP17:%.*]] = bitcast i16* [[VLA]] to i8* 3187 // CHECK3-NEXT: store i8* [[TMP17]], i8** [[TMP16]], align 8 3188 // CHECK3-NEXT: [[TMP18:%.*]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[DOTOMP_REDUCTION_RED_LIST]], i64 0, i64 1 3189 // CHECK3-NEXT: [[TMP19:%.*]] = inttoptr i64 [[TMP6]] to i8* 3190 // CHECK3-NEXT: store i8* [[TMP19]], i8** [[TMP18]], align 8 3191 // CHECK3-NEXT: [[TMP20:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 3192 // CHECK3-NEXT: [[TMP21:%.*]] = load i32, i32* [[TMP20]], align 4 3193 // CHECK3-NEXT: [[TMP22:%.*]] = bitcast [2 x i8*]* [[DOTOMP_REDUCTION_RED_LIST]] to i8* 3194 // CHECK3-NEXT: [[TMP23:%.*]] = call i32 @__kmpc_reduce_nowait(%struct.ident_t* @[[GLOB1:[0-9]+]], i32 [[TMP21]], i32 1, i64 16, i8* [[TMP22]], void (i8*, i8*)* @.omp.reduction.reduction_func, [8 x i32]* @.gomp_critical_user_.reduction.var) 3195 // CHECK3-NEXT: switch i32 [[TMP23]], label [[DOTOMP_REDUCTION_DEFAULT:%.*]] [ 3196 // CHECK3-NEXT: i32 1, label [[DOTOMP_REDUCTION_CASE1:%.*]] 3197 // CHECK3-NEXT: i32 2, label [[DOTOMP_REDUCTION_CASE2:%.*]] 3198 // CHECK3-NEXT: ] 3199 // CHECK3: .omp.reduction.case1: 3200 // CHECK3-NEXT: [[TMP24:%.*]] = getelementptr i16, i16* [[ARRAYIDX]], i64 [[TMP6]] 3201 // CHECK3-NEXT: [[OMP_ARRAYCPY_ISEMPTY:%.*]] = icmp eq i16* [[ARRAYIDX]], [[TMP24]] 3202 // CHECK3-NEXT: br i1 [[OMP_ARRAYCPY_ISEMPTY]], label [[OMP_ARRAYCPY_DONE7:%.*]], label [[OMP_ARRAYCPY_BODY:%.*]] 3203 // CHECK3: omp.arraycpy.body: 3204 // CHECK3-NEXT: [[OMP_ARRAYCPY_SRCELEMENTPAST:%.*]] = phi i16* [ [[VLA]], [[DOTOMP_REDUCTION_CASE1]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ] 3205 // CHECK3-NEXT: [[OMP_ARRAYCPY_DESTELEMENTPAST2:%.*]] = phi i16* [ [[ARRAYIDX]], [[DOTOMP_REDUCTION_CASE1]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT5:%.*]], [[OMP_ARRAYCPY_BODY]] ] 3206 // CHECK3-NEXT: [[TMP25:%.*]] = load i16, i16* [[OMP_ARRAYCPY_DESTELEMENTPAST2]], align 2 3207 // CHECK3-NEXT: [[CONV:%.*]] = sext i16 [[TMP25]] to i32 3208 // CHECK3-NEXT: [[TMP26:%.*]] = load i16, i16* [[OMP_ARRAYCPY_SRCELEMENTPAST]], align 2 3209 // CHECK3-NEXT: [[CONV3:%.*]] = sext i16 [[TMP26]] to i32 3210 // CHECK3-NEXT: [[ADD:%.*]] = add nsw i32 [[CONV]], [[CONV3]] 3211 // CHECK3-NEXT: [[CONV4:%.*]] = trunc i32 [[ADD]] to i16 3212 // CHECK3-NEXT: store i16 [[CONV4]], i16* [[OMP_ARRAYCPY_DESTELEMENTPAST2]], align 2 3213 // CHECK3-NEXT: [[OMP_ARRAYCPY_DEST_ELEMENT5]] = getelementptr i16, i16* [[OMP_ARRAYCPY_DESTELEMENTPAST2]], i32 1 3214 // CHECK3-NEXT: [[OMP_ARRAYCPY_SRC_ELEMENT]] = getelementptr i16, i16* [[OMP_ARRAYCPY_SRCELEMENTPAST]], i32 1 3215 // CHECK3-NEXT: [[OMP_ARRAYCPY_DONE6:%.*]] = icmp eq i16* [[OMP_ARRAYCPY_DEST_ELEMENT5]], [[TMP24]] 3216 // CHECK3-NEXT: br i1 [[OMP_ARRAYCPY_DONE6]], label [[OMP_ARRAYCPY_DONE7]], label [[OMP_ARRAYCPY_BODY]] 3217 // CHECK3: omp.arraycpy.done7: 3218 // CHECK3-NEXT: call void @__kmpc_end_reduce_nowait(%struct.ident_t* @[[GLOB1]], i32 [[TMP21]], [8 x i32]* @.gomp_critical_user_.reduction.var) 3219 // CHECK3-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]] 3220 // CHECK3: .omp.reduction.case2: 3221 // CHECK3-NEXT: [[TMP27:%.*]] = getelementptr i16, i16* [[ARRAYIDX]], i64 [[TMP6]] 3222 // CHECK3-NEXT: [[OMP_ARRAYCPY_ISEMPTY8:%.*]] = icmp eq i16* [[ARRAYIDX]], [[TMP27]] 3223 // CHECK3-NEXT: br i1 [[OMP_ARRAYCPY_ISEMPTY8]], label [[OMP_ARRAYCPY_DONE21:%.*]], label [[OMP_ARRAYCPY_BODY9:%.*]] 3224 // CHECK3: omp.arraycpy.body9: 3225 // CHECK3-NEXT: [[OMP_ARRAYCPY_SRCELEMENTPAST10:%.*]] = phi i16* [ [[VLA]], [[DOTOMP_REDUCTION_CASE2]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT19:%.*]], [[ATOMIC_EXIT:%.*]] ] 3226 // CHECK3-NEXT: [[OMP_ARRAYCPY_DESTELEMENTPAST11:%.*]] = phi i16* [ [[ARRAYIDX]], [[DOTOMP_REDUCTION_CASE2]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT18:%.*]], [[ATOMIC_EXIT]] ] 3227 // CHECK3-NEXT: [[TMP28:%.*]] = load i16, i16* [[OMP_ARRAYCPY_SRCELEMENTPAST10]], align 2 3228 // CHECK3-NEXT: [[CONV12:%.*]] = sext i16 [[TMP28]] to i32 3229 // CHECK3-NEXT: [[ATOMIC_LOAD:%.*]] = load atomic i16, i16* [[OMP_ARRAYCPY_DESTELEMENTPAST11]] monotonic, align 2 3230 // CHECK3-NEXT: br label [[ATOMIC_CONT:%.*]] 3231 // CHECK3: atomic_cont: 3232 // CHECK3-NEXT: [[TMP29:%.*]] = phi i16 [ [[ATOMIC_LOAD]], [[OMP_ARRAYCPY_BODY9]] ], [ [[TMP34:%.*]], [[ATOMIC_CONT]] ] 3233 // CHECK3-NEXT: store i16 [[TMP29]], i16* [[_TMP13]], align 2 3234 // CHECK3-NEXT: [[TMP30:%.*]] = load i16, i16* [[_TMP13]], align 2 3235 // CHECK3-NEXT: [[CONV14:%.*]] = sext i16 [[TMP30]] to i32 3236 // CHECK3-NEXT: [[TMP31:%.*]] = load i16, i16* [[OMP_ARRAYCPY_SRCELEMENTPAST10]], align 2 3237 // CHECK3-NEXT: [[CONV15:%.*]] = sext i16 [[TMP31]] to i32 3238 // CHECK3-NEXT: [[ADD16:%.*]] = add nsw i32 [[CONV14]], [[CONV15]] 3239 // CHECK3-NEXT: [[CONV17:%.*]] = trunc i32 [[ADD16]] to i16 3240 // CHECK3-NEXT: store i16 [[CONV17]], i16* [[ATOMIC_TEMP]], align 2 3241 // CHECK3-NEXT: [[TMP32:%.*]] = load i16, i16* [[ATOMIC_TEMP]], align 2 3242 // CHECK3-NEXT: [[TMP33:%.*]] = cmpxchg i16* [[OMP_ARRAYCPY_DESTELEMENTPAST11]], i16 [[TMP29]], i16 [[TMP32]] monotonic monotonic, align 2 3243 // CHECK3-NEXT: [[TMP34]] = extractvalue { i16, i1 } [[TMP33]], 0 3244 // CHECK3-NEXT: [[TMP35:%.*]] = extractvalue { i16, i1 } [[TMP33]], 1 3245 // CHECK3-NEXT: br i1 [[TMP35]], label [[ATOMIC_EXIT]], label [[ATOMIC_CONT]] 3246 // CHECK3: atomic_exit: 3247 // CHECK3-NEXT: [[OMP_ARRAYCPY_DEST_ELEMENT18]] = getelementptr i16, i16* [[OMP_ARRAYCPY_DESTELEMENTPAST11]], i32 1 3248 // CHECK3-NEXT: [[OMP_ARRAYCPY_SRC_ELEMENT19]] = getelementptr i16, i16* [[OMP_ARRAYCPY_SRCELEMENTPAST10]], i32 1 3249 // CHECK3-NEXT: [[OMP_ARRAYCPY_DONE20:%.*]] = icmp eq i16* [[OMP_ARRAYCPY_DEST_ELEMENT18]], [[TMP27]] 3250 // CHECK3-NEXT: br i1 [[OMP_ARRAYCPY_DONE20]], label [[OMP_ARRAYCPY_DONE21]], label [[OMP_ARRAYCPY_BODY9]] 3251 // CHECK3: omp.arraycpy.done21: 3252 // CHECK3-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]] 3253 // CHECK3: .omp.reduction.default: 3254 // CHECK3-NEXT: [[TMP36:%.*]] = load i8*, i8** [[SAVED_STACK]], align 8 3255 // CHECK3-NEXT: call void @llvm.stackrestore(i8* [[TMP36]]) 3256 // CHECK3-NEXT: ret void 3257 // 3258 // 3259 // CHECK3-LABEL: define {{[^@]+}}@.omp.reduction.reduction_func 3260 // CHECK3-SAME: (i8* [[TMP0:%.*]], i8* [[TMP1:%.*]]) #[[ATTR3:[0-9]+]] { 3261 // CHECK3-NEXT: entry: 3262 // CHECK3-NEXT: [[DOTADDR:%.*]] = alloca i8*, align 8 3263 // CHECK3-NEXT: [[DOTADDR1:%.*]] = alloca i8*, align 8 3264 // CHECK3-NEXT: store i8* [[TMP0]], i8** [[DOTADDR]], align 8 3265 // CHECK3-NEXT: store i8* [[TMP1]], i8** [[DOTADDR1]], align 8 3266 // CHECK3-NEXT: [[TMP2:%.*]] = load i8*, i8** [[DOTADDR]], align 8 3267 // CHECK3-NEXT: [[TMP3:%.*]] = bitcast i8* [[TMP2]] to [2 x i8*]* 3268 // CHECK3-NEXT: [[TMP4:%.*]] = load i8*, i8** [[DOTADDR1]], align 8 3269 // CHECK3-NEXT: [[TMP5:%.*]] = bitcast i8* [[TMP4]] to [2 x i8*]* 3270 // CHECK3-NEXT: [[TMP6:%.*]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[TMP5]], i64 0, i64 0 3271 // CHECK3-NEXT: [[TMP7:%.*]] = load i8*, i8** [[TMP6]], align 8 3272 // CHECK3-NEXT: [[TMP8:%.*]] = bitcast i8* [[TMP7]] to i16* 3273 // CHECK3-NEXT: [[TMP9:%.*]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[TMP3]], i64 0, i64 0 3274 // CHECK3-NEXT: [[TMP10:%.*]] = load i8*, i8** [[TMP9]], align 8 3275 // CHECK3-NEXT: [[TMP11:%.*]] = bitcast i8* [[TMP10]] to i16* 3276 // CHECK3-NEXT: [[TMP12:%.*]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[TMP3]], i64 0, i64 1 3277 // CHECK3-NEXT: [[TMP13:%.*]] = load i8*, i8** [[TMP12]], align 8 3278 // CHECK3-NEXT: [[TMP14:%.*]] = ptrtoint i8* [[TMP13]] to i64 3279 // CHECK3-NEXT: [[TMP15:%.*]] = getelementptr i16, i16* [[TMP11]], i64 [[TMP14]] 3280 // CHECK3-NEXT: [[OMP_ARRAYCPY_ISEMPTY:%.*]] = icmp eq i16* [[TMP11]], [[TMP15]] 3281 // CHECK3-NEXT: br i1 [[OMP_ARRAYCPY_ISEMPTY]], label [[OMP_ARRAYCPY_DONE4:%.*]], label [[OMP_ARRAYCPY_BODY:%.*]] 3282 // CHECK3: omp.arraycpy.body: 3283 // CHECK3-NEXT: [[OMP_ARRAYCPY_SRCELEMENTPAST:%.*]] = phi i16* [ [[TMP8]], [[ENTRY:%.*]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ] 3284 // CHECK3-NEXT: [[OMP_ARRAYCPY_DESTELEMENTPAST:%.*]] = phi i16* [ [[TMP11]], [[ENTRY]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ] 3285 // CHECK3-NEXT: [[TMP16:%.*]] = load i16, i16* [[OMP_ARRAYCPY_DESTELEMENTPAST]], align 2 3286 // CHECK3-NEXT: [[CONV:%.*]] = sext i16 [[TMP16]] to i32 3287 // CHECK3-NEXT: [[TMP17:%.*]] = load i16, i16* [[OMP_ARRAYCPY_SRCELEMENTPAST]], align 2 3288 // CHECK3-NEXT: [[CONV2:%.*]] = sext i16 [[TMP17]] to i32 3289 // CHECK3-NEXT: [[ADD:%.*]] = add nsw i32 [[CONV]], [[CONV2]] 3290 // CHECK3-NEXT: [[CONV3:%.*]] = trunc i32 [[ADD]] to i16 3291 // CHECK3-NEXT: store i16 [[CONV3]], i16* [[OMP_ARRAYCPY_DESTELEMENTPAST]], align 2 3292 // CHECK3-NEXT: [[OMP_ARRAYCPY_DEST_ELEMENT]] = getelementptr i16, i16* [[OMP_ARRAYCPY_DESTELEMENTPAST]], i32 1 3293 // CHECK3-NEXT: [[OMP_ARRAYCPY_SRC_ELEMENT]] = getelementptr i16, i16* [[OMP_ARRAYCPY_SRCELEMENTPAST]], i32 1 3294 // CHECK3-NEXT: [[OMP_ARRAYCPY_DONE:%.*]] = icmp eq i16* [[OMP_ARRAYCPY_DEST_ELEMENT]], [[TMP15]] 3295 // CHECK3-NEXT: br i1 [[OMP_ARRAYCPY_DONE]], label [[OMP_ARRAYCPY_DONE4]], label [[OMP_ARRAYCPY_BODY]] 3296 // CHECK3: omp.arraycpy.done4: 3297 // CHECK3-NEXT: ret void 3298 // 3299 // 3300 // CHECK3-LABEL: define {{[^@]+}}@main 3301 // CHECK3-SAME: () #[[ATTR6:[0-9]+]] { 3302 // CHECK3-NEXT: entry: 3303 // CHECK3-NEXT: [[RETVAL:%.*]] = alloca i32, align 4 3304 // CHECK3-NEXT: [[SS:%.*]] = alloca [[STRUCT_SS:%.*]], align 8 3305 // CHECK3-NEXT: [[REF_TMP:%.*]] = alloca [[CLASS_ANON:%.*]], align 1 3306 // CHECK3-NEXT: store i32 0, i32* [[RETVAL]], align 4 3307 // CHECK3-NEXT: call void @_ZN2SSC1ERi(%struct.SS* nonnull align 8 dereferenceable(16) [[SS]], i32* nonnull align 4 dereferenceable(4) @sivar) 3308 // CHECK3-NEXT: call void @"_ZZ4mainENK3$_0clEv"(%class.anon* nonnull align 1 dereferenceable(1) [[REF_TMP]]) 3309 // CHECK3-NEXT: ret i32 0 3310 // 3311 // 3312 // CHECK3-LABEL: define {{[^@]+}}@_ZN2SSC1ERi 3313 // CHECK3-SAME: (%struct.SS* nonnull align 8 dereferenceable(16) [[THIS:%.*]], i32* nonnull align 4 dereferenceable(4) [[D:%.*]]) unnamed_addr #[[ATTR7:[0-9]+]] align 2 { 3314 // CHECK3-NEXT: entry: 3315 // CHECK3-NEXT: [[THIS_ADDR:%.*]] = alloca %struct.SS*, align 8 3316 // CHECK3-NEXT: [[D_ADDR:%.*]] = alloca i32*, align 8 3317 // CHECK3-NEXT: store %struct.SS* [[THIS]], %struct.SS** [[THIS_ADDR]], align 8 3318 // CHECK3-NEXT: store i32* [[D]], i32** [[D_ADDR]], align 8 3319 // CHECK3-NEXT: [[THIS1:%.*]] = load %struct.SS*, %struct.SS** [[THIS_ADDR]], align 8 3320 // CHECK3-NEXT: [[TMP0:%.*]] = load i32*, i32** [[D_ADDR]], align 8 3321 // CHECK3-NEXT: call void @_ZN2SSC2ERi(%struct.SS* nonnull align 8 dereferenceable(16) [[THIS1]], i32* nonnull align 4 dereferenceable(4) [[TMP0]]) 3322 // CHECK3-NEXT: ret void 3323 // 3324 // 3325 // CHECK3-LABEL: define {{[^@]+}}@_ZN2SSC2ERi 3326 // CHECK3-SAME: (%struct.SS* nonnull align 8 dereferenceable(16) [[THIS:%.*]], i32* nonnull align 4 dereferenceable(4) [[D:%.*]]) unnamed_addr #[[ATTR7]] align 2 { 3327 // CHECK3-NEXT: entry: 3328 // CHECK3-NEXT: [[THIS_ADDR:%.*]] = alloca %struct.SS*, align 8 3329 // CHECK3-NEXT: [[D_ADDR:%.*]] = alloca i32*, align 8 3330 // CHECK3-NEXT: [[A2:%.*]] = alloca i32*, align 8 3331 // CHECK3-NEXT: [[B4:%.*]] = alloca i32, align 4 3332 // CHECK3-NEXT: [[C5:%.*]] = alloca i32*, align 8 3333 // CHECK3-NEXT: store %struct.SS* [[THIS]], %struct.SS** [[THIS_ADDR]], align 8 3334 // CHECK3-NEXT: store i32* [[D]], i32** [[D_ADDR]], align 8 3335 // CHECK3-NEXT: [[THIS1:%.*]] = load %struct.SS*, %struct.SS** [[THIS_ADDR]], align 8 3336 // CHECK3-NEXT: [[A:%.*]] = getelementptr inbounds [[STRUCT_SS:%.*]], %struct.SS* [[THIS1]], i32 0, i32 0 3337 // CHECK3-NEXT: store i32 0, i32* [[A]], align 8 3338 // CHECK3-NEXT: [[B:%.*]] = getelementptr inbounds [[STRUCT_SS]], %struct.SS* [[THIS1]], i32 0, i32 1 3339 // CHECK3-NEXT: [[BF_LOAD:%.*]] = load i8, i8* [[B]], align 4 3340 // CHECK3-NEXT: [[BF_CLEAR:%.*]] = and i8 [[BF_LOAD]], -16 3341 // CHECK3-NEXT: store i8 [[BF_CLEAR]], i8* [[B]], align 4 3342 // CHECK3-NEXT: [[C:%.*]] = getelementptr inbounds [[STRUCT_SS]], %struct.SS* [[THIS1]], i32 0, i32 2 3343 // CHECK3-NEXT: [[TMP0:%.*]] = load i32*, i32** [[D_ADDR]], align 8 3344 // CHECK3-NEXT: store i32* [[TMP0]], i32** [[C]], align 8 3345 // CHECK3-NEXT: [[A3:%.*]] = getelementptr inbounds [[STRUCT_SS]], %struct.SS* [[THIS1]], i32 0, i32 0 3346 // CHECK3-NEXT: store i32* [[A3]], i32** [[A2]], align 8 3347 // CHECK3-NEXT: [[C6:%.*]] = getelementptr inbounds [[STRUCT_SS]], %struct.SS* [[THIS1]], i32 0, i32 2 3348 // CHECK3-NEXT: [[TMP1:%.*]] = load i32*, i32** [[C6]], align 8 3349 // CHECK3-NEXT: store i32* [[TMP1]], i32** [[C5]], align 8 3350 // CHECK3-NEXT: [[TMP2:%.*]] = load i32*, i32** [[A2]], align 8 3351 // CHECK3-NEXT: [[TMP3:%.*]] = load i32*, i32** [[C5]], align 8 3352 // CHECK3-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB2]], i32 4, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, %struct.SS*, i32*, i32*, i32*)* @.omp_outlined..1 to void (i32*, i32*, ...)*), %struct.SS* [[THIS1]], i32* [[TMP2]], i32* [[B4]], i32* [[TMP3]]) 3353 // CHECK3-NEXT: [[TMP4:%.*]] = load i32, i32* [[B4]], align 4 3354 // CHECK3-NEXT: [[B7:%.*]] = getelementptr inbounds [[STRUCT_SS]], %struct.SS* [[THIS1]], i32 0, i32 1 3355 // CHECK3-NEXT: [[TMP5:%.*]] = trunc i32 [[TMP4]] to i8 3356 // CHECK3-NEXT: [[BF_LOAD8:%.*]] = load i8, i8* [[B7]], align 4 3357 // CHECK3-NEXT: [[BF_VALUE:%.*]] = and i8 [[TMP5]], 15 3358 // CHECK3-NEXT: [[BF_CLEAR9:%.*]] = and i8 [[BF_LOAD8]], -16 3359 // CHECK3-NEXT: [[BF_SET:%.*]] = or i8 [[BF_CLEAR9]], [[BF_VALUE]] 3360 // CHECK3-NEXT: store i8 [[BF_SET]], i8* [[B7]], align 4 3361 // CHECK3-NEXT: ret void 3362 // 3363 // 3364 // CHECK3-LABEL: define {{[^@]+}}@.omp_outlined..1 3365 // CHECK3-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], %struct.SS* [[THIS:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]], i32* nonnull align 4 dereferenceable(4) [[B:%.*]], i32* nonnull align 4 dereferenceable(4) [[C:%.*]]) #[[ATTR1]] { 3366 // CHECK3-NEXT: entry: 3367 // CHECK3-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 3368 // CHECK3-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 3369 // CHECK3-NEXT: [[THIS_ADDR:%.*]] = alloca %struct.SS*, align 8 3370 // CHECK3-NEXT: [[A_ADDR:%.*]] = alloca i32*, align 8 3371 // CHECK3-NEXT: [[B_ADDR:%.*]] = alloca i32*, align 8 3372 // CHECK3-NEXT: [[C_ADDR:%.*]] = alloca i32*, align 8 3373 // CHECK3-NEXT: [[TMP:%.*]] = alloca i32*, align 8 3374 // CHECK3-NEXT: [[_TMP1:%.*]] = alloca i32*, align 8 3375 // CHECK3-NEXT: [[A2:%.*]] = alloca i32, align 4 3376 // CHECK3-NEXT: [[_TMP3:%.*]] = alloca i32*, align 8 3377 // CHECK3-NEXT: [[B4:%.*]] = alloca i32, align 4 3378 // CHECK3-NEXT: [[C5:%.*]] = alloca i32, align 4 3379 // CHECK3-NEXT: [[_TMP6:%.*]] = alloca i32*, align 8 3380 // CHECK3-NEXT: [[REF_TMP:%.*]] = alloca [[CLASS_ANON_0:%.*]], align 8 3381 // CHECK3-NEXT: [[DOTOMP_REDUCTION_RED_LIST:%.*]] = alloca [3 x i8*], align 8 3382 // CHECK3-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8 3383 // CHECK3-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8 3384 // CHECK3-NEXT: store %struct.SS* [[THIS]], %struct.SS** [[THIS_ADDR]], align 8 3385 // CHECK3-NEXT: store i32* [[A]], i32** [[A_ADDR]], align 8 3386 // CHECK3-NEXT: store i32* [[B]], i32** [[B_ADDR]], align 8 3387 // CHECK3-NEXT: store i32* [[C]], i32** [[C_ADDR]], align 8 3388 // CHECK3-NEXT: [[TMP0:%.*]] = load %struct.SS*, %struct.SS** [[THIS_ADDR]], align 8 3389 // CHECK3-NEXT: [[TMP1:%.*]] = load i32*, i32** [[A_ADDR]], align 8 3390 // CHECK3-NEXT: [[TMP2:%.*]] = load i32*, i32** [[B_ADDR]], align 8 3391 // CHECK3-NEXT: [[TMP3:%.*]] = load i32*, i32** [[C_ADDR]], align 8 3392 // CHECK3-NEXT: store i32* [[TMP1]], i32** [[TMP]], align 8 3393 // CHECK3-NEXT: store i32* [[TMP3]], i32** [[_TMP1]], align 8 3394 // CHECK3-NEXT: [[TMP4:%.*]] = load i32*, i32** [[TMP]], align 8 3395 // CHECK3-NEXT: store i32 0, i32* [[A2]], align 4 3396 // CHECK3-NEXT: store i32* [[A2]], i32** [[_TMP3]], align 8 3397 // CHECK3-NEXT: store i32 0, i32* [[B4]], align 4 3398 // CHECK3-NEXT: [[TMP5:%.*]] = load i32*, i32** [[_TMP1]], align 8 3399 // CHECK3-NEXT: store i32 0, i32* [[C5]], align 4 3400 // CHECK3-NEXT: store i32* [[C5]], i32** [[_TMP6]], align 8 3401 // CHECK3-NEXT: [[TMP6:%.*]] = getelementptr inbounds [[CLASS_ANON_0]], %class.anon.0* [[REF_TMP]], i32 0, i32 0 3402 // CHECK3-NEXT: store %struct.SS* [[TMP0]], %struct.SS** [[TMP6]], align 8 3403 // CHECK3-NEXT: [[TMP7:%.*]] = getelementptr inbounds [[CLASS_ANON_0]], %class.anon.0* [[REF_TMP]], i32 0, i32 1 3404 // CHECK3-NEXT: [[TMP8:%.*]] = load i32*, i32** [[_TMP3]], align 8 3405 // CHECK3-NEXT: store i32* [[TMP8]], i32** [[TMP7]], align 8 3406 // CHECK3-NEXT: [[TMP9:%.*]] = getelementptr inbounds [[CLASS_ANON_0]], %class.anon.0* [[REF_TMP]], i32 0, i32 2 3407 // CHECK3-NEXT: store i32* [[B4]], i32** [[TMP9]], align 8 3408 // CHECK3-NEXT: [[TMP10:%.*]] = getelementptr inbounds [[CLASS_ANON_0]], %class.anon.0* [[REF_TMP]], i32 0, i32 3 3409 // CHECK3-NEXT: [[TMP11:%.*]] = load i32*, i32** [[_TMP6]], align 8 3410 // CHECK3-NEXT: store i32* [[TMP11]], i32** [[TMP10]], align 8 3411 // CHECK3-NEXT: call void @_ZZN2SSC1ERiENKUlvE_clEv(%class.anon.0* nonnull align 8 dereferenceable(32) [[REF_TMP]]) 3412 // CHECK3-NEXT: [[TMP12:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOMP_REDUCTION_RED_LIST]], i64 0, i64 0 3413 // CHECK3-NEXT: [[TMP13:%.*]] = bitcast i32* [[A2]] to i8* 3414 // CHECK3-NEXT: store i8* [[TMP13]], i8** [[TMP12]], align 8 3415 // CHECK3-NEXT: [[TMP14:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOMP_REDUCTION_RED_LIST]], i64 0, i64 1 3416 // CHECK3-NEXT: [[TMP15:%.*]] = bitcast i32* [[B4]] to i8* 3417 // CHECK3-NEXT: store i8* [[TMP15]], i8** [[TMP14]], align 8 3418 // CHECK3-NEXT: [[TMP16:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOMP_REDUCTION_RED_LIST]], i64 0, i64 2 3419 // CHECK3-NEXT: [[TMP17:%.*]] = bitcast i32* [[C5]] to i8* 3420 // CHECK3-NEXT: store i8* [[TMP17]], i8** [[TMP16]], align 8 3421 // CHECK3-NEXT: [[TMP18:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 3422 // CHECK3-NEXT: [[TMP19:%.*]] = load i32, i32* [[TMP18]], align 4 3423 // CHECK3-NEXT: [[TMP20:%.*]] = bitcast [3 x i8*]* [[DOTOMP_REDUCTION_RED_LIST]] to i8* 3424 // CHECK3-NEXT: [[TMP21:%.*]] = call i32 @__kmpc_reduce_nowait(%struct.ident_t* @[[GLOB1]], i32 [[TMP19]], i32 3, i64 24, i8* [[TMP20]], void (i8*, i8*)* @.omp.reduction.reduction_func.2, [8 x i32]* @.gomp_critical_user_.reduction.var) 3425 // CHECK3-NEXT: switch i32 [[TMP21]], label [[DOTOMP_REDUCTION_DEFAULT:%.*]] [ 3426 // CHECK3-NEXT: i32 1, label [[DOTOMP_REDUCTION_CASE1:%.*]] 3427 // CHECK3-NEXT: i32 2, label [[DOTOMP_REDUCTION_CASE2:%.*]] 3428 // CHECK3-NEXT: ] 3429 // CHECK3: .omp.reduction.case1: 3430 // CHECK3-NEXT: [[TMP22:%.*]] = load i32, i32* [[TMP4]], align 4 3431 // CHECK3-NEXT: [[TMP23:%.*]] = load i32, i32* [[A2]], align 4 3432 // CHECK3-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP22]], [[TMP23]] 3433 // CHECK3-NEXT: store i32 [[ADD]], i32* [[TMP4]], align 4 3434 // CHECK3-NEXT: [[TMP24:%.*]] = load i32, i32* [[TMP2]], align 4 3435 // CHECK3-NEXT: [[TMP25:%.*]] = load i32, i32* [[B4]], align 4 3436 // CHECK3-NEXT: [[ADD7:%.*]] = add nsw i32 [[TMP24]], [[TMP25]] 3437 // CHECK3-NEXT: store i32 [[ADD7]], i32* [[TMP2]], align 4 3438 // CHECK3-NEXT: [[TMP26:%.*]] = load i32, i32* [[TMP5]], align 4 3439 // CHECK3-NEXT: [[TMP27:%.*]] = load i32, i32* [[C5]], align 4 3440 // CHECK3-NEXT: [[ADD8:%.*]] = add nsw i32 [[TMP26]], [[TMP27]] 3441 // CHECK3-NEXT: store i32 [[ADD8]], i32* [[TMP5]], align 4 3442 // CHECK3-NEXT: call void @__kmpc_end_reduce_nowait(%struct.ident_t* @[[GLOB1]], i32 [[TMP19]], [8 x i32]* @.gomp_critical_user_.reduction.var) 3443 // CHECK3-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]] 3444 // CHECK3: .omp.reduction.case2: 3445 // CHECK3-NEXT: [[TMP28:%.*]] = load i32, i32* [[A2]], align 4 3446 // CHECK3-NEXT: [[TMP29:%.*]] = atomicrmw add i32* [[TMP4]], i32 [[TMP28]] monotonic, align 4 3447 // CHECK3-NEXT: [[TMP30:%.*]] = load i32, i32* [[B4]], align 4 3448 // CHECK3-NEXT: [[TMP31:%.*]] = atomicrmw add i32* [[TMP2]], i32 [[TMP30]] monotonic, align 4 3449 // CHECK3-NEXT: [[TMP32:%.*]] = load i32, i32* [[C5]], align 4 3450 // CHECK3-NEXT: [[TMP33:%.*]] = atomicrmw add i32* [[TMP5]], i32 [[TMP32]] monotonic, align 4 3451 // CHECK3-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]] 3452 // CHECK3: .omp.reduction.default: 3453 // CHECK3-NEXT: ret void 3454 // 3455 // 3456 // CHECK3-LABEL: define {{[^@]+}}@_ZZN2SSC1ERiENKUlvE_clEv 3457 // CHECK3-SAME: (%class.anon.0* nonnull align 8 dereferenceable(32) [[THIS:%.*]]) #[[ATTR0]] align 2 { 3458 // CHECK3-NEXT: entry: 3459 // CHECK3-NEXT: [[THIS_ADDR:%.*]] = alloca %class.anon.0*, align 8 3460 // CHECK3-NEXT: store %class.anon.0* [[THIS]], %class.anon.0** [[THIS_ADDR]], align 8 3461 // CHECK3-NEXT: [[THIS1:%.*]] = load %class.anon.0*, %class.anon.0** [[THIS_ADDR]], align 8 3462 // CHECK3-NEXT: [[TMP0:%.*]] = getelementptr inbounds [[CLASS_ANON_0:%.*]], %class.anon.0* [[THIS1]], i32 0, i32 0 3463 // CHECK3-NEXT: [[TMP1:%.*]] = load %struct.SS*, %struct.SS** [[TMP0]], align 8 3464 // CHECK3-NEXT: [[TMP2:%.*]] = getelementptr inbounds [[CLASS_ANON_0]], %class.anon.0* [[THIS1]], i32 0, i32 1 3465 // CHECK3-NEXT: [[TMP3:%.*]] = load i32*, i32** [[TMP2]], align 8 3466 // CHECK3-NEXT: [[TMP4:%.*]] = load i32, i32* [[TMP3]], align 4 3467 // CHECK3-NEXT: [[INC:%.*]] = add nsw i32 [[TMP4]], 1 3468 // CHECK3-NEXT: store i32 [[INC]], i32* [[TMP3]], align 4 3469 // CHECK3-NEXT: [[TMP5:%.*]] = getelementptr inbounds [[CLASS_ANON_0]], %class.anon.0* [[THIS1]], i32 0, i32 2 3470 // CHECK3-NEXT: [[TMP6:%.*]] = load i32*, i32** [[TMP5]], align 8 3471 // CHECK3-NEXT: [[TMP7:%.*]] = load i32, i32* [[TMP6]], align 4 3472 // CHECK3-NEXT: [[DEC:%.*]] = add nsw i32 [[TMP7]], -1 3473 // CHECK3-NEXT: store i32 [[DEC]], i32* [[TMP6]], align 4 3474 // CHECK3-NEXT: [[TMP8:%.*]] = getelementptr inbounds [[CLASS_ANON_0]], %class.anon.0* [[THIS1]], i32 0, i32 3 3475 // CHECK3-NEXT: [[TMP9:%.*]] = load i32*, i32** [[TMP8]], align 8 3476 // CHECK3-NEXT: [[TMP10:%.*]] = load i32, i32* [[TMP9]], align 4 3477 // CHECK3-NEXT: [[DIV:%.*]] = sdiv i32 [[TMP10]], 1 3478 // CHECK3-NEXT: store i32 [[DIV]], i32* [[TMP9]], align 4 3479 // CHECK3-NEXT: [[TMP11:%.*]] = getelementptr inbounds [[CLASS_ANON_0]], %class.anon.0* [[THIS1]], i32 0, i32 1 3480 // CHECK3-NEXT: [[TMP12:%.*]] = load i32*, i32** [[TMP11]], align 8 3481 // CHECK3-NEXT: [[TMP13:%.*]] = getelementptr inbounds [[CLASS_ANON_0]], %class.anon.0* [[THIS1]], i32 0, i32 2 3482 // CHECK3-NEXT: [[TMP14:%.*]] = load i32*, i32** [[TMP13]], align 8 3483 // CHECK3-NEXT: [[TMP15:%.*]] = getelementptr inbounds [[CLASS_ANON_0]], %class.anon.0* [[THIS1]], i32 0, i32 3 3484 // CHECK3-NEXT: [[TMP16:%.*]] = load i32*, i32** [[TMP15]], align 8 3485 // CHECK3-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB2]], i32 4, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, %struct.SS*, i32*, i32*, i32*)* @.omp_outlined..3 to void (i32*, i32*, ...)*), %struct.SS* [[TMP1]], i32* [[TMP12]], i32* [[TMP14]], i32* [[TMP16]]) 3486 // CHECK3-NEXT: ret void 3487 // 3488 // 3489 // CHECK3-LABEL: define {{[^@]+}}@.omp.reduction.reduction_func.2 3490 // CHECK3-SAME: (i8* [[TMP0:%.*]], i8* [[TMP1:%.*]]) #[[ATTR3]] { 3491 // CHECK3-NEXT: entry: 3492 // CHECK3-NEXT: [[DOTADDR:%.*]] = alloca i8*, align 8 3493 // CHECK3-NEXT: [[DOTADDR1:%.*]] = alloca i8*, align 8 3494 // CHECK3-NEXT: store i8* [[TMP0]], i8** [[DOTADDR]], align 8 3495 // CHECK3-NEXT: store i8* [[TMP1]], i8** [[DOTADDR1]], align 8 3496 // CHECK3-NEXT: [[TMP2:%.*]] = load i8*, i8** [[DOTADDR]], align 8 3497 // CHECK3-NEXT: [[TMP3:%.*]] = bitcast i8* [[TMP2]] to [3 x i8*]* 3498 // CHECK3-NEXT: [[TMP4:%.*]] = load i8*, i8** [[DOTADDR1]], align 8 3499 // CHECK3-NEXT: [[TMP5:%.*]] = bitcast i8* [[TMP4]] to [3 x i8*]* 3500 // CHECK3-NEXT: [[TMP6:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[TMP5]], i64 0, i64 0 3501 // CHECK3-NEXT: [[TMP7:%.*]] = load i8*, i8** [[TMP6]], align 8 3502 // CHECK3-NEXT: [[TMP8:%.*]] = bitcast i8* [[TMP7]] to i32* 3503 // CHECK3-NEXT: [[TMP9:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[TMP3]], i64 0, i64 0 3504 // CHECK3-NEXT: [[TMP10:%.*]] = load i8*, i8** [[TMP9]], align 8 3505 // CHECK3-NEXT: [[TMP11:%.*]] = bitcast i8* [[TMP10]] to i32* 3506 // CHECK3-NEXT: [[TMP12:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[TMP5]], i64 0, i64 1 3507 // CHECK3-NEXT: [[TMP13:%.*]] = load i8*, i8** [[TMP12]], align 8 3508 // CHECK3-NEXT: [[TMP14:%.*]] = bitcast i8* [[TMP13]] to i32* 3509 // CHECK3-NEXT: [[TMP15:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[TMP3]], i64 0, i64 1 3510 // CHECK3-NEXT: [[TMP16:%.*]] = load i8*, i8** [[TMP15]], align 8 3511 // CHECK3-NEXT: [[TMP17:%.*]] = bitcast i8* [[TMP16]] to i32* 3512 // CHECK3-NEXT: [[TMP18:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[TMP5]], i64 0, i64 2 3513 // CHECK3-NEXT: [[TMP19:%.*]] = load i8*, i8** [[TMP18]], align 8 3514 // CHECK3-NEXT: [[TMP20:%.*]] = bitcast i8* [[TMP19]] to i32* 3515 // CHECK3-NEXT: [[TMP21:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[TMP3]], i64 0, i64 2 3516 // CHECK3-NEXT: [[TMP22:%.*]] = load i8*, i8** [[TMP21]], align 8 3517 // CHECK3-NEXT: [[TMP23:%.*]] = bitcast i8* [[TMP22]] to i32* 3518 // CHECK3-NEXT: [[TMP24:%.*]] = load i32, i32* [[TMP11]], align 4 3519 // CHECK3-NEXT: [[TMP25:%.*]] = load i32, i32* [[TMP8]], align 4 3520 // CHECK3-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP24]], [[TMP25]] 3521 // CHECK3-NEXT: store i32 [[ADD]], i32* [[TMP11]], align 4 3522 // CHECK3-NEXT: [[TMP26:%.*]] = load i32, i32* [[TMP17]], align 4 3523 // CHECK3-NEXT: [[TMP27:%.*]] = load i32, i32* [[TMP14]], align 4 3524 // CHECK3-NEXT: [[ADD2:%.*]] = add nsw i32 [[TMP26]], [[TMP27]] 3525 // CHECK3-NEXT: store i32 [[ADD2]], i32* [[TMP17]], align 4 3526 // CHECK3-NEXT: [[TMP28:%.*]] = load i32, i32* [[TMP23]], align 4 3527 // CHECK3-NEXT: [[TMP29:%.*]] = load i32, i32* [[TMP20]], align 4 3528 // CHECK3-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP28]], [[TMP29]] 3529 // CHECK3-NEXT: store i32 [[ADD3]], i32* [[TMP23]], align 4 3530 // CHECK3-NEXT: ret void 3531 // 3532 // 3533 // CHECK3-LABEL: define {{[^@]+}}@.omp_outlined..3 3534 // CHECK3-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], %struct.SS* [[THIS:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]], i32* nonnull align 4 dereferenceable(4) [[B:%.*]], i32* nonnull align 4 dereferenceable(4) [[C:%.*]]) #[[ATTR1]] { 3535 // CHECK3-NEXT: entry: 3536 // CHECK3-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 3537 // CHECK3-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 3538 // CHECK3-NEXT: [[THIS_ADDR:%.*]] = alloca %struct.SS*, align 8 3539 // CHECK3-NEXT: [[A_ADDR:%.*]] = alloca i32*, align 8 3540 // CHECK3-NEXT: [[B_ADDR:%.*]] = alloca i32*, align 8 3541 // CHECK3-NEXT: [[C_ADDR:%.*]] = alloca i32*, align 8 3542 // CHECK3-NEXT: [[TMP:%.*]] = alloca i32*, align 8 3543 // CHECK3-NEXT: [[_TMP1:%.*]] = alloca i32*, align 8 3544 // CHECK3-NEXT: [[A2:%.*]] = alloca i32, align 4 3545 // CHECK3-NEXT: [[_TMP3:%.*]] = alloca i32*, align 8 3546 // CHECK3-NEXT: [[B4:%.*]] = alloca i32, align 4 3547 // CHECK3-NEXT: [[C5:%.*]] = alloca i32, align 4 3548 // CHECK3-NEXT: [[_TMP6:%.*]] = alloca i32*, align 8 3549 // CHECK3-NEXT: [[DOTOMP_REDUCTION_RED_LIST:%.*]] = alloca [3 x i8*], align 8 3550 // CHECK3-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8 3551 // CHECK3-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8 3552 // CHECK3-NEXT: store %struct.SS* [[THIS]], %struct.SS** [[THIS_ADDR]], align 8 3553 // CHECK3-NEXT: store i32* [[A]], i32** [[A_ADDR]], align 8 3554 // CHECK3-NEXT: store i32* [[B]], i32** [[B_ADDR]], align 8 3555 // CHECK3-NEXT: store i32* [[C]], i32** [[C_ADDR]], align 8 3556 // CHECK3-NEXT: [[TMP0:%.*]] = load %struct.SS*, %struct.SS** [[THIS_ADDR]], align 8 3557 // CHECK3-NEXT: [[TMP1:%.*]] = load i32*, i32** [[A_ADDR]], align 8 3558 // CHECK3-NEXT: [[TMP2:%.*]] = load i32*, i32** [[B_ADDR]], align 8 3559 // CHECK3-NEXT: [[TMP3:%.*]] = load i32*, i32** [[C_ADDR]], align 8 3560 // CHECK3-NEXT: store i32* [[TMP1]], i32** [[TMP]], align 8 3561 // CHECK3-NEXT: store i32* [[TMP3]], i32** [[_TMP1]], align 8 3562 // CHECK3-NEXT: [[TMP4:%.*]] = load i32*, i32** [[TMP]], align 8 3563 // CHECK3-NEXT: store i32 -1, i32* [[A2]], align 4 3564 // CHECK3-NEXT: store i32* [[A2]], i32** [[_TMP3]], align 8 3565 // CHECK3-NEXT: store i32 -1, i32* [[B4]], align 4 3566 // CHECK3-NEXT: [[TMP5:%.*]] = load i32*, i32** [[_TMP1]], align 8 3567 // CHECK3-NEXT: store i32 -1, i32* [[C5]], align 4 3568 // CHECK3-NEXT: store i32* [[C5]], i32** [[_TMP6]], align 8 3569 // CHECK3-NEXT: [[TMP6:%.*]] = load i32*, i32** [[_TMP3]], align 8 3570 // CHECK3-NEXT: [[TMP7:%.*]] = load i32, i32* [[TMP6]], align 4 3571 // CHECK3-NEXT: [[INC:%.*]] = add nsw i32 [[TMP7]], 1 3572 // CHECK3-NEXT: store i32 [[INC]], i32* [[TMP6]], align 4 3573 // CHECK3-NEXT: [[TMP8:%.*]] = load i32, i32* [[B4]], align 4 3574 // CHECK3-NEXT: [[DEC:%.*]] = add nsw i32 [[TMP8]], -1 3575 // CHECK3-NEXT: store i32 [[DEC]], i32* [[B4]], align 4 3576 // CHECK3-NEXT: [[TMP9:%.*]] = load i32*, i32** [[_TMP6]], align 8 3577 // CHECK3-NEXT: [[TMP10:%.*]] = load i32, i32* [[TMP9]], align 4 3578 // CHECK3-NEXT: [[DIV:%.*]] = sdiv i32 [[TMP10]], 1 3579 // CHECK3-NEXT: store i32 [[DIV]], i32* [[TMP9]], align 4 3580 // CHECK3-NEXT: [[TMP11:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOMP_REDUCTION_RED_LIST]], i64 0, i64 0 3581 // CHECK3-NEXT: [[TMP12:%.*]] = bitcast i32* [[A2]] to i8* 3582 // CHECK3-NEXT: store i8* [[TMP12]], i8** [[TMP11]], align 8 3583 // CHECK3-NEXT: [[TMP13:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOMP_REDUCTION_RED_LIST]], i64 0, i64 1 3584 // CHECK3-NEXT: [[TMP14:%.*]] = bitcast i32* [[B4]] to i8* 3585 // CHECK3-NEXT: store i8* [[TMP14]], i8** [[TMP13]], align 8 3586 // CHECK3-NEXT: [[TMP15:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOMP_REDUCTION_RED_LIST]], i64 0, i64 2 3587 // CHECK3-NEXT: [[TMP16:%.*]] = bitcast i32* [[C5]] to i8* 3588 // CHECK3-NEXT: store i8* [[TMP16]], i8** [[TMP15]], align 8 3589 // CHECK3-NEXT: [[TMP17:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 3590 // CHECK3-NEXT: [[TMP18:%.*]] = load i32, i32* [[TMP17]], align 4 3591 // CHECK3-NEXT: [[TMP19:%.*]] = bitcast [3 x i8*]* [[DOTOMP_REDUCTION_RED_LIST]] to i8* 3592 // CHECK3-NEXT: [[TMP20:%.*]] = call i32 @__kmpc_reduce_nowait(%struct.ident_t* @[[GLOB1]], i32 [[TMP18]], i32 3, i64 24, i8* [[TMP19]], void (i8*, i8*)* @.omp.reduction.reduction_func.4, [8 x i32]* @.gomp_critical_user_.reduction.var) 3593 // CHECK3-NEXT: switch i32 [[TMP20]], label [[DOTOMP_REDUCTION_DEFAULT:%.*]] [ 3594 // CHECK3-NEXT: i32 1, label [[DOTOMP_REDUCTION_CASE1:%.*]] 3595 // CHECK3-NEXT: i32 2, label [[DOTOMP_REDUCTION_CASE2:%.*]] 3596 // CHECK3-NEXT: ] 3597 // CHECK3: .omp.reduction.case1: 3598 // CHECK3-NEXT: [[TMP21:%.*]] = load i32, i32* [[TMP4]], align 4 3599 // CHECK3-NEXT: [[TMP22:%.*]] = load i32, i32* [[A2]], align 4 3600 // CHECK3-NEXT: [[AND:%.*]] = and i32 [[TMP21]], [[TMP22]] 3601 // CHECK3-NEXT: store i32 [[AND]], i32* [[TMP4]], align 4 3602 // CHECK3-NEXT: [[TMP23:%.*]] = load i32, i32* [[TMP2]], align 4 3603 // CHECK3-NEXT: [[TMP24:%.*]] = load i32, i32* [[B4]], align 4 3604 // CHECK3-NEXT: [[AND7:%.*]] = and i32 [[TMP23]], [[TMP24]] 3605 // CHECK3-NEXT: store i32 [[AND7]], i32* [[TMP2]], align 4 3606 // CHECK3-NEXT: [[TMP25:%.*]] = load i32, i32* [[TMP5]], align 4 3607 // CHECK3-NEXT: [[TMP26:%.*]] = load i32, i32* [[C5]], align 4 3608 // CHECK3-NEXT: [[AND8:%.*]] = and i32 [[TMP25]], [[TMP26]] 3609 // CHECK3-NEXT: store i32 [[AND8]], i32* [[TMP5]], align 4 3610 // CHECK3-NEXT: call void @__kmpc_end_reduce_nowait(%struct.ident_t* @[[GLOB1]], i32 [[TMP18]], [8 x i32]* @.gomp_critical_user_.reduction.var) 3611 // CHECK3-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]] 3612 // CHECK3: .omp.reduction.case2: 3613 // CHECK3-NEXT: [[TMP27:%.*]] = load i32, i32* [[A2]], align 4 3614 // CHECK3-NEXT: [[TMP28:%.*]] = atomicrmw and i32* [[TMP4]], i32 [[TMP27]] monotonic, align 4 3615 // CHECK3-NEXT: [[TMP29:%.*]] = load i32, i32* [[B4]], align 4 3616 // CHECK3-NEXT: [[TMP30:%.*]] = atomicrmw and i32* [[TMP2]], i32 [[TMP29]] monotonic, align 4 3617 // CHECK3-NEXT: [[TMP31:%.*]] = load i32, i32* [[C5]], align 4 3618 // CHECK3-NEXT: [[TMP32:%.*]] = atomicrmw and i32* [[TMP5]], i32 [[TMP31]] monotonic, align 4 3619 // CHECK3-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]] 3620 // CHECK3: .omp.reduction.default: 3621 // CHECK3-NEXT: ret void 3622 // 3623 // 3624 // CHECK3-LABEL: define {{[^@]+}}@.omp.reduction.reduction_func.4 3625 // CHECK3-SAME: (i8* [[TMP0:%.*]], i8* [[TMP1:%.*]]) #[[ATTR3]] { 3626 // CHECK3-NEXT: entry: 3627 // CHECK3-NEXT: [[DOTADDR:%.*]] = alloca i8*, align 8 3628 // CHECK3-NEXT: [[DOTADDR1:%.*]] = alloca i8*, align 8 3629 // CHECK3-NEXT: store i8* [[TMP0]], i8** [[DOTADDR]], align 8 3630 // CHECK3-NEXT: store i8* [[TMP1]], i8** [[DOTADDR1]], align 8 3631 // CHECK3-NEXT: [[TMP2:%.*]] = load i8*, i8** [[DOTADDR]], align 8 3632 // CHECK3-NEXT: [[TMP3:%.*]] = bitcast i8* [[TMP2]] to [3 x i8*]* 3633 // CHECK3-NEXT: [[TMP4:%.*]] = load i8*, i8** [[DOTADDR1]], align 8 3634 // CHECK3-NEXT: [[TMP5:%.*]] = bitcast i8* [[TMP4]] to [3 x i8*]* 3635 // CHECK3-NEXT: [[TMP6:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[TMP5]], i64 0, i64 0 3636 // CHECK3-NEXT: [[TMP7:%.*]] = load i8*, i8** [[TMP6]], align 8 3637 // CHECK3-NEXT: [[TMP8:%.*]] = bitcast i8* [[TMP7]] to i32* 3638 // CHECK3-NEXT: [[TMP9:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[TMP3]], i64 0, i64 0 3639 // CHECK3-NEXT: [[TMP10:%.*]] = load i8*, i8** [[TMP9]], align 8 3640 // CHECK3-NEXT: [[TMP11:%.*]] = bitcast i8* [[TMP10]] to i32* 3641 // CHECK3-NEXT: [[TMP12:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[TMP5]], i64 0, i64 1 3642 // CHECK3-NEXT: [[TMP13:%.*]] = load i8*, i8** [[TMP12]], align 8 3643 // CHECK3-NEXT: [[TMP14:%.*]] = bitcast i8* [[TMP13]] to i32* 3644 // CHECK3-NEXT: [[TMP15:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[TMP3]], i64 0, i64 1 3645 // CHECK3-NEXT: [[TMP16:%.*]] = load i8*, i8** [[TMP15]], align 8 3646 // CHECK3-NEXT: [[TMP17:%.*]] = bitcast i8* [[TMP16]] to i32* 3647 // CHECK3-NEXT: [[TMP18:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[TMP5]], i64 0, i64 2 3648 // CHECK3-NEXT: [[TMP19:%.*]] = load i8*, i8** [[TMP18]], align 8 3649 // CHECK3-NEXT: [[TMP20:%.*]] = bitcast i8* [[TMP19]] to i32* 3650 // CHECK3-NEXT: [[TMP21:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[TMP3]], i64 0, i64 2 3651 // CHECK3-NEXT: [[TMP22:%.*]] = load i8*, i8** [[TMP21]], align 8 3652 // CHECK3-NEXT: [[TMP23:%.*]] = bitcast i8* [[TMP22]] to i32* 3653 // CHECK3-NEXT: [[TMP24:%.*]] = load i32, i32* [[TMP11]], align 4 3654 // CHECK3-NEXT: [[TMP25:%.*]] = load i32, i32* [[TMP8]], align 4 3655 // CHECK3-NEXT: [[AND:%.*]] = and i32 [[TMP24]], [[TMP25]] 3656 // CHECK3-NEXT: store i32 [[AND]], i32* [[TMP11]], align 4 3657 // CHECK3-NEXT: [[TMP26:%.*]] = load i32, i32* [[TMP17]], align 4 3658 // CHECK3-NEXT: [[TMP27:%.*]] = load i32, i32* [[TMP14]], align 4 3659 // CHECK3-NEXT: [[AND2:%.*]] = and i32 [[TMP26]], [[TMP27]] 3660 // CHECK3-NEXT: store i32 [[AND2]], i32* [[TMP17]], align 4 3661 // CHECK3-NEXT: [[TMP28:%.*]] = load i32, i32* [[TMP23]], align 4 3662 // CHECK3-NEXT: [[TMP29:%.*]] = load i32, i32* [[TMP20]], align 4 3663 // CHECK3-NEXT: [[AND3:%.*]] = and i32 [[TMP28]], [[TMP29]] 3664 // CHECK3-NEXT: store i32 [[AND3]], i32* [[TMP23]], align 4 3665 // CHECK3-NEXT: ret void 3666 // 3667 // 3668 // CHECK3-LABEL: define {{[^@]+}}@.omp_outlined..5 3669 // CHECK3-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i32* nonnull align 4 dereferenceable(4) [[G:%.*]]) #[[ATTR1]] { 3670 // CHECK3-NEXT: entry: 3671 // CHECK3-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 3672 // CHECK3-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 3673 // CHECK3-NEXT: [[G_ADDR:%.*]] = alloca i32*, align 8 3674 // CHECK3-NEXT: [[G1:%.*]] = alloca i32, align 128 3675 // CHECK3-NEXT: [[REF_TMP:%.*]] = alloca [[CLASS_ANON_1:%.*]], align 8 3676 // CHECK3-NEXT: [[DOTOMP_REDUCTION_RED_LIST:%.*]] = alloca [1 x i8*], align 8 3677 // CHECK3-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8 3678 // CHECK3-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8 3679 // CHECK3-NEXT: store i32* [[G]], i32** [[G_ADDR]], align 8 3680 // CHECK3-NEXT: [[TMP0:%.*]] = load i32*, i32** [[G_ADDR]], align 8 3681 // CHECK3-NEXT: store i32 0, i32* [[G1]], align 128 3682 // CHECK3-NEXT: store i32 1, i32* [[G1]], align 128 3683 // CHECK3-NEXT: [[TMP1:%.*]] = getelementptr inbounds [[CLASS_ANON_1]], %class.anon.1* [[REF_TMP]], i32 0, i32 0 3684 // CHECK3-NEXT: store i32* [[G1]], i32** [[TMP1]], align 8 3685 // CHECK3-NEXT: call void @"_ZZZ4mainENK3$_0clEvENKUlvE_clEv"(%class.anon.1* nonnull align 8 dereferenceable(8) [[REF_TMP]]) 3686 // CHECK3-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOMP_REDUCTION_RED_LIST]], i64 0, i64 0 3687 // CHECK3-NEXT: [[TMP3:%.*]] = bitcast i32* [[G1]] to i8* 3688 // CHECK3-NEXT: store i8* [[TMP3]], i8** [[TMP2]], align 8 3689 // CHECK3-NEXT: [[TMP4:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 3690 // CHECK3-NEXT: [[TMP5:%.*]] = load i32, i32* [[TMP4]], align 4 3691 // CHECK3-NEXT: [[TMP6:%.*]] = bitcast [1 x i8*]* [[DOTOMP_REDUCTION_RED_LIST]] to i8* 3692 // CHECK3-NEXT: [[TMP7:%.*]] = call i32 @__kmpc_reduce_nowait(%struct.ident_t* @[[GLOB1]], i32 [[TMP5]], i32 1, i64 8, i8* [[TMP6]], void (i8*, i8*)* @.omp.reduction.reduction_func.6, [8 x i32]* @.gomp_critical_user_.reduction.var) 3693 // CHECK3-NEXT: switch i32 [[TMP7]], label [[DOTOMP_REDUCTION_DEFAULT:%.*]] [ 3694 // CHECK3-NEXT: i32 1, label [[DOTOMP_REDUCTION_CASE1:%.*]] 3695 // CHECK3-NEXT: i32 2, label [[DOTOMP_REDUCTION_CASE2:%.*]] 3696 // CHECK3-NEXT: ] 3697 // CHECK3: .omp.reduction.case1: 3698 // CHECK3-NEXT: [[TMP8:%.*]] = load i32, i32* [[TMP0]], align 128 3699 // CHECK3-NEXT: [[TMP9:%.*]] = load i32, i32* [[G1]], align 128 3700 // CHECK3-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP8]], [[TMP9]] 3701 // CHECK3-NEXT: store i32 [[ADD]], i32* [[TMP0]], align 128 3702 // CHECK3-NEXT: call void @__kmpc_end_reduce_nowait(%struct.ident_t* @[[GLOB1]], i32 [[TMP5]], [8 x i32]* @.gomp_critical_user_.reduction.var) 3703 // CHECK3-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]] 3704 // CHECK3: .omp.reduction.case2: 3705 // CHECK3-NEXT: [[TMP10:%.*]] = load i32, i32* [[G1]], align 128 3706 // CHECK3-NEXT: [[TMP11:%.*]] = atomicrmw add i32* [[TMP0]], i32 [[TMP10]] monotonic, align 4 3707 // CHECK3-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]] 3708 // CHECK3: .omp.reduction.default: 3709 // CHECK3-NEXT: ret void 3710 // 3711 // 3712 // CHECK3-LABEL: define {{[^@]+}}@.omp.reduction.reduction_func.6 3713 // CHECK3-SAME: (i8* [[TMP0:%.*]], i8* [[TMP1:%.*]]) #[[ATTR3]] { 3714 // CHECK3-NEXT: entry: 3715 // CHECK3-NEXT: [[DOTADDR:%.*]] = alloca i8*, align 8 3716 // CHECK3-NEXT: [[DOTADDR1:%.*]] = alloca i8*, align 8 3717 // CHECK3-NEXT: store i8* [[TMP0]], i8** [[DOTADDR]], align 8 3718 // CHECK3-NEXT: store i8* [[TMP1]], i8** [[DOTADDR1]], align 8 3719 // CHECK3-NEXT: [[TMP2:%.*]] = load i8*, i8** [[DOTADDR]], align 8 3720 // CHECK3-NEXT: [[TMP3:%.*]] = bitcast i8* [[TMP2]] to [1 x i8*]* 3721 // CHECK3-NEXT: [[TMP4:%.*]] = load i8*, i8** [[DOTADDR1]], align 8 3722 // CHECK3-NEXT: [[TMP5:%.*]] = bitcast i8* [[TMP4]] to [1 x i8*]* 3723 // CHECK3-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[TMP5]], i64 0, i64 0 3724 // CHECK3-NEXT: [[TMP7:%.*]] = load i8*, i8** [[TMP6]], align 8 3725 // CHECK3-NEXT: [[TMP8:%.*]] = bitcast i8* [[TMP7]] to i32* 3726 // CHECK3-NEXT: [[TMP9:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[TMP3]], i64 0, i64 0 3727 // CHECK3-NEXT: [[TMP10:%.*]] = load i8*, i8** [[TMP9]], align 8 3728 // CHECK3-NEXT: [[TMP11:%.*]] = bitcast i8* [[TMP10]] to i32* 3729 // CHECK3-NEXT: [[TMP12:%.*]] = load i32, i32* [[TMP11]], align 128 3730 // CHECK3-NEXT: [[TMP13:%.*]] = load i32, i32* [[TMP8]], align 128 3731 // CHECK3-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP12]], [[TMP13]] 3732 // CHECK3-NEXT: store i32 [[ADD]], i32* [[TMP11]], align 128 3733 // CHECK3-NEXT: ret void 3734 // 3735 // 3736 // CHECK4-LABEL: define {{[^@]+}}@_Z14foo_array_sectPs 3737 // CHECK4-SAME: (i16* [[X:%.*]]) #[[ATTR1:[0-9]+]] { 3738 // CHECK4-NEXT: entry: 3739 // CHECK4-NEXT: [[X_ADDR:%.*]] = alloca i16*, align 8 3740 // CHECK4-NEXT: store i16* [[X]], i16** [[X_ADDR]], align 8 3741 // CHECK4-NEXT: [[TMP0:%.*]] = load i16*, i16** [[X_ADDR]], align 8 3742 // CHECK4-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB2:[0-9]+]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i16*)* @.omp_outlined. to void (i32*, i32*, ...)*), i16* [[TMP0]]) 3743 // CHECK4-NEXT: ret void 3744 // 3745 // 3746 // CHECK4-LABEL: define {{[^@]+}}@.omp_outlined. 3747 // CHECK4-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i16* [[X:%.*]]) #[[ATTR2:[0-9]+]] { 3748 // CHECK4-NEXT: entry: 3749 // CHECK4-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 3750 // CHECK4-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 3751 // CHECK4-NEXT: [[X_ADDR:%.*]] = alloca i16*, align 8 3752 // CHECK4-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 8 3753 // CHECK4-NEXT: [[__VLA_EXPR0:%.*]] = alloca i64, align 8 3754 // CHECK4-NEXT: [[TMP:%.*]] = alloca i16*, align 8 3755 // CHECK4-NEXT: [[DOTOMP_REDUCTION_RED_LIST:%.*]] = alloca [2 x i8*], align 8 3756 // CHECK4-NEXT: [[ATOMIC_TEMP:%.*]] = alloca i16, align 2 3757 // CHECK4-NEXT: [[_TMP13:%.*]] = alloca i16, align 2 3758 // CHECK4-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8 3759 // CHECK4-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8 3760 // CHECK4-NEXT: store i16* [[X]], i16** [[X_ADDR]], align 8 3761 // CHECK4-NEXT: [[TMP0:%.*]] = load i16*, i16** [[X_ADDR]], align 8 3762 // CHECK4-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i16, i16* [[TMP0]], i64 0 3763 // CHECK4-NEXT: [[TMP1:%.*]] = load i16*, i16** [[X_ADDR]], align 8 3764 // CHECK4-NEXT: [[ARRAYIDX1:%.*]] = getelementptr inbounds i16, i16* [[TMP1]], i64 0 3765 // CHECK4-NEXT: [[TMP2:%.*]] = ptrtoint i16* [[ARRAYIDX1]] to i64 3766 // CHECK4-NEXT: [[TMP3:%.*]] = ptrtoint i16* [[ARRAYIDX]] to i64 3767 // CHECK4-NEXT: [[TMP4:%.*]] = sub i64 [[TMP2]], [[TMP3]] 3768 // CHECK4-NEXT: [[TMP5:%.*]] = sdiv exact i64 [[TMP4]], ptrtoint (i16* getelementptr (i16, i16* null, i32 1) to i64) 3769 // CHECK4-NEXT: [[TMP6:%.*]] = add nuw i64 [[TMP5]], 1 3770 // CHECK4-NEXT: [[TMP7:%.*]] = mul nuw i64 [[TMP6]], ptrtoint (i16* getelementptr (i16, i16* null, i32 1) to i64) 3771 // CHECK4-NEXT: [[TMP8:%.*]] = call i8* @llvm.stacksave() 3772 // CHECK4-NEXT: store i8* [[TMP8]], i8** [[SAVED_STACK]], align 8 3773 // CHECK4-NEXT: [[VLA:%.*]] = alloca i16, i64 [[TMP6]], align 16 3774 // CHECK4-NEXT: store i64 [[TMP6]], i64* [[__VLA_EXPR0]], align 8 3775 // CHECK4-NEXT: [[TMP9:%.*]] = getelementptr i16, i16* [[VLA]], i64 [[TMP6]] 3776 // CHECK4-NEXT: [[OMP_ARRAYINIT_ISEMPTY:%.*]] = icmp eq i16* [[VLA]], [[TMP9]] 3777 // CHECK4-NEXT: br i1 [[OMP_ARRAYINIT_ISEMPTY]], label [[OMP_ARRAYINIT_DONE:%.*]], label [[OMP_ARRAYINIT_BODY:%.*]] 3778 // CHECK4: omp.arrayinit.body: 3779 // CHECK4-NEXT: [[OMP_ARRAYCPY_DESTELEMENTPAST:%.*]] = phi i16* [ [[VLA]], [[ENTRY:%.*]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT:%.*]], [[OMP_ARRAYINIT_BODY]] ] 3780 // CHECK4-NEXT: store i16 0, i16* [[OMP_ARRAYCPY_DESTELEMENTPAST]], align 2 3781 // CHECK4-NEXT: [[OMP_ARRAYCPY_DEST_ELEMENT]] = getelementptr i16, i16* [[OMP_ARRAYCPY_DESTELEMENTPAST]], i32 1 3782 // CHECK4-NEXT: [[OMP_ARRAYCPY_DONE:%.*]] = icmp eq i16* [[OMP_ARRAYCPY_DEST_ELEMENT]], [[TMP9]] 3783 // CHECK4-NEXT: br i1 [[OMP_ARRAYCPY_DONE]], label [[OMP_ARRAYINIT_DONE]], label [[OMP_ARRAYINIT_BODY]] 3784 // CHECK4: omp.arrayinit.done: 3785 // CHECK4-NEXT: [[TMP10:%.*]] = load i16*, i16** [[X_ADDR]], align 8 3786 // CHECK4-NEXT: [[TMP11:%.*]] = ptrtoint i16* [[TMP10]] to i64 3787 // CHECK4-NEXT: [[TMP12:%.*]] = ptrtoint i16* [[ARRAYIDX]] to i64 3788 // CHECK4-NEXT: [[TMP13:%.*]] = sub i64 [[TMP11]], [[TMP12]] 3789 // CHECK4-NEXT: [[TMP14:%.*]] = sdiv exact i64 [[TMP13]], ptrtoint (i16* getelementptr (i16, i16* null, i32 1) to i64) 3790 // CHECK4-NEXT: [[TMP15:%.*]] = getelementptr i16, i16* [[VLA]], i64 [[TMP14]] 3791 // CHECK4-NEXT: store i16* [[TMP15]], i16** [[TMP]], align 8 3792 // CHECK4-NEXT: [[TMP16:%.*]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[DOTOMP_REDUCTION_RED_LIST]], i64 0, i64 0 3793 // CHECK4-NEXT: [[TMP17:%.*]] = bitcast i16* [[VLA]] to i8* 3794 // CHECK4-NEXT: store i8* [[TMP17]], i8** [[TMP16]], align 8 3795 // CHECK4-NEXT: [[TMP18:%.*]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[DOTOMP_REDUCTION_RED_LIST]], i64 0, i64 1 3796 // CHECK4-NEXT: [[TMP19:%.*]] = inttoptr i64 [[TMP6]] to i8* 3797 // CHECK4-NEXT: store i8* [[TMP19]], i8** [[TMP18]], align 8 3798 // CHECK4-NEXT: [[TMP20:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 3799 // CHECK4-NEXT: [[TMP21:%.*]] = load i32, i32* [[TMP20]], align 4 3800 // CHECK4-NEXT: [[TMP22:%.*]] = bitcast [2 x i8*]* [[DOTOMP_REDUCTION_RED_LIST]] to i8* 3801 // CHECK4-NEXT: [[TMP23:%.*]] = call i32 @__kmpc_reduce_nowait(%struct.ident_t* @[[GLOB1:[0-9]+]], i32 [[TMP21]], i32 1, i64 16, i8* [[TMP22]], void (i8*, i8*)* @.omp.reduction.reduction_func, [8 x i32]* @.gomp_critical_user_.reduction.var) 3802 // CHECK4-NEXT: switch i32 [[TMP23]], label [[DOTOMP_REDUCTION_DEFAULT:%.*]] [ 3803 // CHECK4-NEXT: i32 1, label [[DOTOMP_REDUCTION_CASE1:%.*]] 3804 // CHECK4-NEXT: i32 2, label [[DOTOMP_REDUCTION_CASE2:%.*]] 3805 // CHECK4-NEXT: ] 3806 // CHECK4: .omp.reduction.case1: 3807 // CHECK4-NEXT: [[TMP24:%.*]] = getelementptr i16, i16* [[ARRAYIDX]], i64 [[TMP6]] 3808 // CHECK4-NEXT: [[OMP_ARRAYCPY_ISEMPTY:%.*]] = icmp eq i16* [[ARRAYIDX]], [[TMP24]] 3809 // CHECK4-NEXT: br i1 [[OMP_ARRAYCPY_ISEMPTY]], label [[OMP_ARRAYCPY_DONE7:%.*]], label [[OMP_ARRAYCPY_BODY:%.*]] 3810 // CHECK4: omp.arraycpy.body: 3811 // CHECK4-NEXT: [[OMP_ARRAYCPY_SRCELEMENTPAST:%.*]] = phi i16* [ [[VLA]], [[DOTOMP_REDUCTION_CASE1]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ] 3812 // CHECK4-NEXT: [[OMP_ARRAYCPY_DESTELEMENTPAST2:%.*]] = phi i16* [ [[ARRAYIDX]], [[DOTOMP_REDUCTION_CASE1]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT5:%.*]], [[OMP_ARRAYCPY_BODY]] ] 3813 // CHECK4-NEXT: [[TMP25:%.*]] = load i16, i16* [[OMP_ARRAYCPY_DESTELEMENTPAST2]], align 2 3814 // CHECK4-NEXT: [[CONV:%.*]] = sext i16 [[TMP25]] to i32 3815 // CHECK4-NEXT: [[TMP26:%.*]] = load i16, i16* [[OMP_ARRAYCPY_SRCELEMENTPAST]], align 2 3816 // CHECK4-NEXT: [[CONV3:%.*]] = sext i16 [[TMP26]] to i32 3817 // CHECK4-NEXT: [[ADD:%.*]] = add nsw i32 [[CONV]], [[CONV3]] 3818 // CHECK4-NEXT: [[CONV4:%.*]] = trunc i32 [[ADD]] to i16 3819 // CHECK4-NEXT: store i16 [[CONV4]], i16* [[OMP_ARRAYCPY_DESTELEMENTPAST2]], align 2 3820 // CHECK4-NEXT: [[OMP_ARRAYCPY_DEST_ELEMENT5]] = getelementptr i16, i16* [[OMP_ARRAYCPY_DESTELEMENTPAST2]], i32 1 3821 // CHECK4-NEXT: [[OMP_ARRAYCPY_SRC_ELEMENT]] = getelementptr i16, i16* [[OMP_ARRAYCPY_SRCELEMENTPAST]], i32 1 3822 // CHECK4-NEXT: [[OMP_ARRAYCPY_DONE6:%.*]] = icmp eq i16* [[OMP_ARRAYCPY_DEST_ELEMENT5]], [[TMP24]] 3823 // CHECK4-NEXT: br i1 [[OMP_ARRAYCPY_DONE6]], label [[OMP_ARRAYCPY_DONE7]], label [[OMP_ARRAYCPY_BODY]] 3824 // CHECK4: omp.arraycpy.done7: 3825 // CHECK4-NEXT: call void @__kmpc_end_reduce_nowait(%struct.ident_t* @[[GLOB1]], i32 [[TMP21]], [8 x i32]* @.gomp_critical_user_.reduction.var) 3826 // CHECK4-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]] 3827 // CHECK4: .omp.reduction.case2: 3828 // CHECK4-NEXT: [[TMP27:%.*]] = getelementptr i16, i16* [[ARRAYIDX]], i64 [[TMP6]] 3829 // CHECK4-NEXT: [[OMP_ARRAYCPY_ISEMPTY8:%.*]] = icmp eq i16* [[ARRAYIDX]], [[TMP27]] 3830 // CHECK4-NEXT: br i1 [[OMP_ARRAYCPY_ISEMPTY8]], label [[OMP_ARRAYCPY_DONE21:%.*]], label [[OMP_ARRAYCPY_BODY9:%.*]] 3831 // CHECK4: omp.arraycpy.body9: 3832 // CHECK4-NEXT: [[OMP_ARRAYCPY_SRCELEMENTPAST10:%.*]] = phi i16* [ [[VLA]], [[DOTOMP_REDUCTION_CASE2]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT19:%.*]], [[ATOMIC_EXIT:%.*]] ] 3833 // CHECK4-NEXT: [[OMP_ARRAYCPY_DESTELEMENTPAST11:%.*]] = phi i16* [ [[ARRAYIDX]], [[DOTOMP_REDUCTION_CASE2]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT18:%.*]], [[ATOMIC_EXIT]] ] 3834 // CHECK4-NEXT: [[TMP28:%.*]] = load i16, i16* [[OMP_ARRAYCPY_SRCELEMENTPAST10]], align 2 3835 // CHECK4-NEXT: [[CONV12:%.*]] = sext i16 [[TMP28]] to i32 3836 // CHECK4-NEXT: [[ATOMIC_LOAD:%.*]] = load atomic i16, i16* [[OMP_ARRAYCPY_DESTELEMENTPAST11]] monotonic, align 2 3837 // CHECK4-NEXT: br label [[ATOMIC_CONT:%.*]] 3838 // CHECK4: atomic_cont: 3839 // CHECK4-NEXT: [[TMP29:%.*]] = phi i16 [ [[ATOMIC_LOAD]], [[OMP_ARRAYCPY_BODY9]] ], [ [[TMP34:%.*]], [[ATOMIC_CONT]] ] 3840 // CHECK4-NEXT: store i16 [[TMP29]], i16* [[_TMP13]], align 2 3841 // CHECK4-NEXT: [[TMP30:%.*]] = load i16, i16* [[_TMP13]], align 2 3842 // CHECK4-NEXT: [[CONV14:%.*]] = sext i16 [[TMP30]] to i32 3843 // CHECK4-NEXT: [[TMP31:%.*]] = load i16, i16* [[OMP_ARRAYCPY_SRCELEMENTPAST10]], align 2 3844 // CHECK4-NEXT: [[CONV15:%.*]] = sext i16 [[TMP31]] to i32 3845 // CHECK4-NEXT: [[ADD16:%.*]] = add nsw i32 [[CONV14]], [[CONV15]] 3846 // CHECK4-NEXT: [[CONV17:%.*]] = trunc i32 [[ADD16]] to i16 3847 // CHECK4-NEXT: store i16 [[CONV17]], i16* [[ATOMIC_TEMP]], align 2 3848 // CHECK4-NEXT: [[TMP32:%.*]] = load i16, i16* [[ATOMIC_TEMP]], align 2 3849 // CHECK4-NEXT: [[TMP33:%.*]] = cmpxchg i16* [[OMP_ARRAYCPY_DESTELEMENTPAST11]], i16 [[TMP29]], i16 [[TMP32]] monotonic monotonic, align 2 3850 // CHECK4-NEXT: [[TMP34]] = extractvalue { i16, i1 } [[TMP33]], 0 3851 // CHECK4-NEXT: [[TMP35:%.*]] = extractvalue { i16, i1 } [[TMP33]], 1 3852 // CHECK4-NEXT: br i1 [[TMP35]], label [[ATOMIC_EXIT]], label [[ATOMIC_CONT]] 3853 // CHECK4: atomic_exit: 3854 // CHECK4-NEXT: [[OMP_ARRAYCPY_DEST_ELEMENT18]] = getelementptr i16, i16* [[OMP_ARRAYCPY_DESTELEMENTPAST11]], i32 1 3855 // CHECK4-NEXT: [[OMP_ARRAYCPY_SRC_ELEMENT19]] = getelementptr i16, i16* [[OMP_ARRAYCPY_SRCELEMENTPAST10]], i32 1 3856 // CHECK4-NEXT: [[OMP_ARRAYCPY_DONE20:%.*]] = icmp eq i16* [[OMP_ARRAYCPY_DEST_ELEMENT18]], [[TMP27]] 3857 // CHECK4-NEXT: br i1 [[OMP_ARRAYCPY_DONE20]], label [[OMP_ARRAYCPY_DONE21]], label [[OMP_ARRAYCPY_BODY9]] 3858 // CHECK4: omp.arraycpy.done21: 3859 // CHECK4-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]] 3860 // CHECK4: .omp.reduction.default: 3861 // CHECK4-NEXT: [[TMP36:%.*]] = load i8*, i8** [[SAVED_STACK]], align 8 3862 // CHECK4-NEXT: call void @llvm.stackrestore(i8* [[TMP36]]) 3863 // CHECK4-NEXT: ret void 3864 // 3865 // 3866 // CHECK4-LABEL: define {{[^@]+}}@.omp.reduction.reduction_func 3867 // CHECK4-SAME: (i8* [[TMP0:%.*]], i8* [[TMP1:%.*]]) #[[ATTR4:[0-9]+]] { 3868 // CHECK4-NEXT: entry: 3869 // CHECK4-NEXT: [[DOTADDR:%.*]] = alloca i8*, align 8 3870 // CHECK4-NEXT: [[DOTADDR1:%.*]] = alloca i8*, align 8 3871 // CHECK4-NEXT: store i8* [[TMP0]], i8** [[DOTADDR]], align 8 3872 // CHECK4-NEXT: store i8* [[TMP1]], i8** [[DOTADDR1]], align 8 3873 // CHECK4-NEXT: [[TMP2:%.*]] = load i8*, i8** [[DOTADDR]], align 8 3874 // CHECK4-NEXT: [[TMP3:%.*]] = bitcast i8* [[TMP2]] to [2 x i8*]* 3875 // CHECK4-NEXT: [[TMP4:%.*]] = load i8*, i8** [[DOTADDR1]], align 8 3876 // CHECK4-NEXT: [[TMP5:%.*]] = bitcast i8* [[TMP4]] to [2 x i8*]* 3877 // CHECK4-NEXT: [[TMP6:%.*]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[TMP5]], i64 0, i64 0 3878 // CHECK4-NEXT: [[TMP7:%.*]] = load i8*, i8** [[TMP6]], align 8 3879 // CHECK4-NEXT: [[TMP8:%.*]] = bitcast i8* [[TMP7]] to i16* 3880 // CHECK4-NEXT: [[TMP9:%.*]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[TMP3]], i64 0, i64 0 3881 // CHECK4-NEXT: [[TMP10:%.*]] = load i8*, i8** [[TMP9]], align 8 3882 // CHECK4-NEXT: [[TMP11:%.*]] = bitcast i8* [[TMP10]] to i16* 3883 // CHECK4-NEXT: [[TMP12:%.*]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[TMP3]], i64 0, i64 1 3884 // CHECK4-NEXT: [[TMP13:%.*]] = load i8*, i8** [[TMP12]], align 8 3885 // CHECK4-NEXT: [[TMP14:%.*]] = ptrtoint i8* [[TMP13]] to i64 3886 // CHECK4-NEXT: [[TMP15:%.*]] = getelementptr i16, i16* [[TMP11]], i64 [[TMP14]] 3887 // CHECK4-NEXT: [[OMP_ARRAYCPY_ISEMPTY:%.*]] = icmp eq i16* [[TMP11]], [[TMP15]] 3888 // CHECK4-NEXT: br i1 [[OMP_ARRAYCPY_ISEMPTY]], label [[OMP_ARRAYCPY_DONE4:%.*]], label [[OMP_ARRAYCPY_BODY:%.*]] 3889 // CHECK4: omp.arraycpy.body: 3890 // CHECK4-NEXT: [[OMP_ARRAYCPY_SRCELEMENTPAST:%.*]] = phi i16* [ [[TMP8]], [[ENTRY:%.*]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ] 3891 // CHECK4-NEXT: [[OMP_ARRAYCPY_DESTELEMENTPAST:%.*]] = phi i16* [ [[TMP11]], [[ENTRY]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ] 3892 // CHECK4-NEXT: [[TMP16:%.*]] = load i16, i16* [[OMP_ARRAYCPY_DESTELEMENTPAST]], align 2 3893 // CHECK4-NEXT: [[CONV:%.*]] = sext i16 [[TMP16]] to i32 3894 // CHECK4-NEXT: [[TMP17:%.*]] = load i16, i16* [[OMP_ARRAYCPY_SRCELEMENTPAST]], align 2 3895 // CHECK4-NEXT: [[CONV2:%.*]] = sext i16 [[TMP17]] to i32 3896 // CHECK4-NEXT: [[ADD:%.*]] = add nsw i32 [[CONV]], [[CONV2]] 3897 // CHECK4-NEXT: [[CONV3:%.*]] = trunc i32 [[ADD]] to i16 3898 // CHECK4-NEXT: store i16 [[CONV3]], i16* [[OMP_ARRAYCPY_DESTELEMENTPAST]], align 2 3899 // CHECK4-NEXT: [[OMP_ARRAYCPY_DEST_ELEMENT]] = getelementptr i16, i16* [[OMP_ARRAYCPY_DESTELEMENTPAST]], i32 1 3900 // CHECK4-NEXT: [[OMP_ARRAYCPY_SRC_ELEMENT]] = getelementptr i16, i16* [[OMP_ARRAYCPY_SRCELEMENTPAST]], i32 1 3901 // CHECK4-NEXT: [[OMP_ARRAYCPY_DONE:%.*]] = icmp eq i16* [[OMP_ARRAYCPY_DEST_ELEMENT]], [[TMP15]] 3902 // CHECK4-NEXT: br i1 [[OMP_ARRAYCPY_DONE]], label [[OMP_ARRAYCPY_DONE4]], label [[OMP_ARRAYCPY_BODY]] 3903 // CHECK4: omp.arraycpy.done4: 3904 // CHECK4-NEXT: ret void 3905 // 3906 // 3907 // CHECK4-LABEL: define {{[^@]+}}@main 3908 // CHECK4-SAME: () #[[ATTR7:[0-9]+]] { 3909 // CHECK4-NEXT: entry: 3910 // CHECK4-NEXT: [[RETVAL:%.*]] = alloca i32, align 4 3911 // CHECK4-NEXT: [[SS:%.*]] = alloca [[STRUCT_SS:%.*]], align 8 3912 // CHECK4-NEXT: store i32 0, i32* [[RETVAL]], align 4 3913 // CHECK4-NEXT: call void @_ZN2SSC1ERi(%struct.SS* nonnull align 8 dereferenceable(16) [[SS]], i32* nonnull align 4 dereferenceable(4) @sivar) 3914 // CHECK4-NEXT: [[TMP0:%.*]] = load i8*, i8** getelementptr inbounds ([[STRUCT___BLOCK_LITERAL_GENERIC:%.*]], %struct.__block_literal_generic* bitcast ({ i8**, i32, i32, i8*, %struct.__block_descriptor* }* @__block_literal_global to %struct.__block_literal_generic*), i32 0, i32 3), align 8 3915 // CHECK4-NEXT: [[TMP1:%.*]] = bitcast i8* [[TMP0]] to void (i8*)* 3916 // CHECK4-NEXT: call void [[TMP1]](i8* bitcast ({ i8**, i32, i32, i8*, %struct.__block_descriptor* }* @__block_literal_global to i8*)) 3917 // CHECK4-NEXT: ret i32 0 3918 // 3919 // 3920 // CHECK4-LABEL: define {{[^@]+}}@_ZN2SSC1ERi 3921 // CHECK4-SAME: (%struct.SS* nonnull align 8 dereferenceable(16) [[THIS:%.*]], i32* nonnull align 4 dereferenceable(4) [[D:%.*]]) unnamed_addr #[[ATTR8:[0-9]+]] align 2 { 3922 // CHECK4-NEXT: entry: 3923 // CHECK4-NEXT: [[THIS_ADDR:%.*]] = alloca %struct.SS*, align 8 3924 // CHECK4-NEXT: [[D_ADDR:%.*]] = alloca i32*, align 8 3925 // CHECK4-NEXT: store %struct.SS* [[THIS]], %struct.SS** [[THIS_ADDR]], align 8 3926 // CHECK4-NEXT: store i32* [[D]], i32** [[D_ADDR]], align 8 3927 // CHECK4-NEXT: [[THIS1:%.*]] = load %struct.SS*, %struct.SS** [[THIS_ADDR]], align 8 3928 // CHECK4-NEXT: [[TMP0:%.*]] = load i32*, i32** [[D_ADDR]], align 8 3929 // CHECK4-NEXT: call void @_ZN2SSC2ERi(%struct.SS* nonnull align 8 dereferenceable(16) [[THIS1]], i32* nonnull align 4 dereferenceable(4) [[TMP0]]) 3930 // CHECK4-NEXT: ret void 3931 // 3932 // 3933 // CHECK4-LABEL: define {{[^@]+}}@__main_block_invoke 3934 // CHECK4-SAME: (i8* [[DOTBLOCK_DESCRIPTOR:%.*]]) #[[ATTR8]] { 3935 // CHECK4-NEXT: entry: 3936 // CHECK4-NEXT: [[DOTBLOCK_DESCRIPTOR_ADDR:%.*]] = alloca i8*, align 8 3937 // CHECK4-NEXT: [[BLOCK_ADDR:%.*]] = alloca <{ i8*, i32, i32, i8*, %struct.__block_descriptor* }>*, align 8 3938 // CHECK4-NEXT: store i8* [[DOTBLOCK_DESCRIPTOR]], i8** [[DOTBLOCK_DESCRIPTOR_ADDR]], align 8 3939 // CHECK4-NEXT: [[BLOCK:%.*]] = bitcast i8* [[DOTBLOCK_DESCRIPTOR]] to <{ i8*, i32, i32, i8*, %struct.__block_descriptor* }>* 3940 // CHECK4-NEXT: store <{ i8*, i32, i32, i8*, %struct.__block_descriptor* }>* [[BLOCK]], <{ i8*, i32, i32, i8*, %struct.__block_descriptor* }>** [[BLOCK_ADDR]], align 8 3941 // CHECK4-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB2]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32*)* @.omp_outlined..1 to void (i32*, i32*, ...)*), i32* @g) 3942 // CHECK4-NEXT: ret void 3943 // 3944 // 3945 // CHECK4-LABEL: define {{[^@]+}}@.omp_outlined..1 3946 // CHECK4-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i32* nonnull align 4 dereferenceable(4) [[G:%.*]]) #[[ATTR2]] { 3947 // CHECK4-NEXT: entry: 3948 // CHECK4-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 3949 // CHECK4-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 3950 // CHECK4-NEXT: [[G_ADDR:%.*]] = alloca i32*, align 8 3951 // CHECK4-NEXT: [[G1:%.*]] = alloca i32, align 128 3952 // CHECK4-NEXT: [[BLOCK:%.*]] = alloca <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, [96 x i8], i32 }>, align 128 3953 // CHECK4-NEXT: [[DOTOMP_REDUCTION_RED_LIST:%.*]] = alloca [1 x i8*], align 8 3954 // CHECK4-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8 3955 // CHECK4-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8 3956 // CHECK4-NEXT: store i32* [[G]], i32** [[G_ADDR]], align 8 3957 // CHECK4-NEXT: [[TMP0:%.*]] = load i32*, i32** [[G_ADDR]], align 8 3958 // CHECK4-NEXT: store i32 0, i32* [[G1]], align 128 3959 // CHECK4-NEXT: store i32 1, i32* [[G1]], align 128 3960 // CHECK4-NEXT: [[BLOCK_ISA:%.*]] = getelementptr inbounds <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, [96 x i8], i32 }>, <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, [96 x i8], i32 }>* [[BLOCK]], i32 0, i32 0 3961 // CHECK4-NEXT: store i8* bitcast (i8** @_NSConcreteStackBlock to i8*), i8** [[BLOCK_ISA]], align 128 3962 // CHECK4-NEXT: [[BLOCK_FLAGS:%.*]] = getelementptr inbounds <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, [96 x i8], i32 }>, <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, [96 x i8], i32 }>* [[BLOCK]], i32 0, i32 1 3963 // CHECK4-NEXT: store i32 1073741824, i32* [[BLOCK_FLAGS]], align 8 3964 // CHECK4-NEXT: [[BLOCK_RESERVED:%.*]] = getelementptr inbounds <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, [96 x i8], i32 }>, <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, [96 x i8], i32 }>* [[BLOCK]], i32 0, i32 2 3965 // CHECK4-NEXT: store i32 0, i32* [[BLOCK_RESERVED]], align 4 3966 // CHECK4-NEXT: [[BLOCK_INVOKE:%.*]] = getelementptr inbounds <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, [96 x i8], i32 }>, <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, [96 x i8], i32 }>* [[BLOCK]], i32 0, i32 3 3967 // CHECK4-NEXT: store i8* bitcast (void (i8*)* @g_block_invoke to i8*), i8** [[BLOCK_INVOKE]], align 16 3968 // CHECK4-NEXT: [[BLOCK_DESCRIPTOR:%.*]] = getelementptr inbounds <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, [96 x i8], i32 }>, <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, [96 x i8], i32 }>* [[BLOCK]], i32 0, i32 4 3969 // CHECK4-NEXT: store %struct.__block_descriptor* bitcast ({ i64, i64, i8*, i8* }* @__block_descriptor_tmp.2 to %struct.__block_descriptor*), %struct.__block_descriptor** [[BLOCK_DESCRIPTOR]], align 8 3970 // CHECK4-NEXT: [[BLOCK_CAPTURED:%.*]] = getelementptr inbounds <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, [96 x i8], i32 }>, <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, [96 x i8], i32 }>* [[BLOCK]], i32 0, i32 6 3971 // CHECK4-NEXT: [[TMP1:%.*]] = load volatile i32, i32* [[G1]], align 128 3972 // CHECK4-NEXT: store volatile i32 [[TMP1]], i32* [[BLOCK_CAPTURED]], align 128 3973 // CHECK4-NEXT: [[TMP2:%.*]] = bitcast <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, [96 x i8], i32 }>* [[BLOCK]] to void ()* 3974 // CHECK4-NEXT: [[BLOCK_LITERAL:%.*]] = bitcast void ()* [[TMP2]] to %struct.__block_literal_generic* 3975 // CHECK4-NEXT: [[TMP3:%.*]] = getelementptr inbounds [[STRUCT___BLOCK_LITERAL_GENERIC:%.*]], %struct.__block_literal_generic* [[BLOCK_LITERAL]], i32 0, i32 3 3976 // CHECK4-NEXT: [[TMP4:%.*]] = bitcast %struct.__block_literal_generic* [[BLOCK_LITERAL]] to i8* 3977 // CHECK4-NEXT: [[TMP5:%.*]] = load i8*, i8** [[TMP3]], align 8 3978 // CHECK4-NEXT: [[TMP6:%.*]] = bitcast i8* [[TMP5]] to void (i8*)* 3979 // CHECK4-NEXT: call void [[TMP6]](i8* [[TMP4]]) 3980 // CHECK4-NEXT: [[TMP7:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOMP_REDUCTION_RED_LIST]], i64 0, i64 0 3981 // CHECK4-NEXT: [[TMP8:%.*]] = bitcast i32* [[G1]] to i8* 3982 // CHECK4-NEXT: store i8* [[TMP8]], i8** [[TMP7]], align 8 3983 // CHECK4-NEXT: [[TMP9:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 3984 // CHECK4-NEXT: [[TMP10:%.*]] = load i32, i32* [[TMP9]], align 4 3985 // CHECK4-NEXT: [[TMP11:%.*]] = bitcast [1 x i8*]* [[DOTOMP_REDUCTION_RED_LIST]] to i8* 3986 // CHECK4-NEXT: [[TMP12:%.*]] = call i32 @__kmpc_reduce_nowait(%struct.ident_t* @[[GLOB1]], i32 [[TMP10]], i32 1, i64 8, i8* [[TMP11]], void (i8*, i8*)* @.omp.reduction.reduction_func.3, [8 x i32]* @.gomp_critical_user_.reduction.var) 3987 // CHECK4-NEXT: switch i32 [[TMP12]], label [[DOTOMP_REDUCTION_DEFAULT:%.*]] [ 3988 // CHECK4-NEXT: i32 1, label [[DOTOMP_REDUCTION_CASE1:%.*]] 3989 // CHECK4-NEXT: i32 2, label [[DOTOMP_REDUCTION_CASE2:%.*]] 3990 // CHECK4-NEXT: ] 3991 // CHECK4: .omp.reduction.case1: 3992 // CHECK4-NEXT: [[TMP13:%.*]] = load i32, i32* [[TMP0]], align 128 3993 // CHECK4-NEXT: [[TMP14:%.*]] = load i32, i32* [[G1]], align 128 3994 // CHECK4-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP13]], [[TMP14]] 3995 // CHECK4-NEXT: store i32 [[ADD]], i32* [[TMP0]], align 128 3996 // CHECK4-NEXT: call void @__kmpc_end_reduce_nowait(%struct.ident_t* @[[GLOB1]], i32 [[TMP10]], [8 x i32]* @.gomp_critical_user_.reduction.var) 3997 // CHECK4-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]] 3998 // CHECK4: .omp.reduction.case2: 3999 // CHECK4-NEXT: [[TMP15:%.*]] = load i32, i32* [[G1]], align 128 4000 // CHECK4-NEXT: [[TMP16:%.*]] = atomicrmw add i32* [[TMP0]], i32 [[TMP15]] monotonic, align 4 4001 // CHECK4-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]] 4002 // CHECK4: .omp.reduction.default: 4003 // CHECK4-NEXT: ret void 4004 // 4005 // 4006 // CHECK4-LABEL: define {{[^@]+}}@g_block_invoke 4007 // CHECK4-SAME: (i8* [[DOTBLOCK_DESCRIPTOR:%.*]]) #[[ATTR8]] { 4008 // CHECK4-NEXT: entry: 4009 // CHECK4-NEXT: [[DOTBLOCK_DESCRIPTOR_ADDR:%.*]] = alloca i8*, align 8 4010 // CHECK4-NEXT: [[BLOCK_ADDR:%.*]] = alloca <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, [96 x i8], i32 }>*, align 8 4011 // CHECK4-NEXT: store i8* [[DOTBLOCK_DESCRIPTOR]], i8** [[DOTBLOCK_DESCRIPTOR_ADDR]], align 8 4012 // CHECK4-NEXT: [[BLOCK:%.*]] = bitcast i8* [[DOTBLOCK_DESCRIPTOR]] to <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, [96 x i8], i32 }>* 4013 // CHECK4-NEXT: store <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, [96 x i8], i32 }>* [[BLOCK]], <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, [96 x i8], i32 }>** [[BLOCK_ADDR]], align 8 4014 // CHECK4-NEXT: [[BLOCK_CAPTURE_ADDR:%.*]] = getelementptr inbounds <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, [96 x i8], i32 }>, <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, [96 x i8], i32 }>* [[BLOCK]], i32 0, i32 6 4015 // CHECK4-NEXT: store i32 2, i32* [[BLOCK_CAPTURE_ADDR]], align 128 4016 // CHECK4-NEXT: ret void 4017 // 4018 // 4019 // CHECK4-LABEL: define {{[^@]+}}@.omp.reduction.reduction_func.3 4020 // CHECK4-SAME: (i8* [[TMP0:%.*]], i8* [[TMP1:%.*]]) #[[ATTR4]] { 4021 // CHECK4-NEXT: entry: 4022 // CHECK4-NEXT: [[DOTADDR:%.*]] = alloca i8*, align 8 4023 // CHECK4-NEXT: [[DOTADDR1:%.*]] = alloca i8*, align 8 4024 // CHECK4-NEXT: store i8* [[TMP0]], i8** [[DOTADDR]], align 8 4025 // CHECK4-NEXT: store i8* [[TMP1]], i8** [[DOTADDR1]], align 8 4026 // CHECK4-NEXT: [[TMP2:%.*]] = load i8*, i8** [[DOTADDR]], align 8 4027 // CHECK4-NEXT: [[TMP3:%.*]] = bitcast i8* [[TMP2]] to [1 x i8*]* 4028 // CHECK4-NEXT: [[TMP4:%.*]] = load i8*, i8** [[DOTADDR1]], align 8 4029 // CHECK4-NEXT: [[TMP5:%.*]] = bitcast i8* [[TMP4]] to [1 x i8*]* 4030 // CHECK4-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[TMP5]], i64 0, i64 0 4031 // CHECK4-NEXT: [[TMP7:%.*]] = load i8*, i8** [[TMP6]], align 8 4032 // CHECK4-NEXT: [[TMP8:%.*]] = bitcast i8* [[TMP7]] to i32* 4033 // CHECK4-NEXT: [[TMP9:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[TMP3]], i64 0, i64 0 4034 // CHECK4-NEXT: [[TMP10:%.*]] = load i8*, i8** [[TMP9]], align 8 4035 // CHECK4-NEXT: [[TMP11:%.*]] = bitcast i8* [[TMP10]] to i32* 4036 // CHECK4-NEXT: [[TMP12:%.*]] = load i32, i32* [[TMP11]], align 128 4037 // CHECK4-NEXT: [[TMP13:%.*]] = load i32, i32* [[TMP8]], align 128 4038 // CHECK4-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP12]], [[TMP13]] 4039 // CHECK4-NEXT: store i32 [[ADD]], i32* [[TMP11]], align 128 4040 // CHECK4-NEXT: ret void 4041 // 4042 // 4043 // CHECK4-LABEL: define {{[^@]+}}@_ZN2SSC2ERi 4044 // CHECK4-SAME: (%struct.SS* nonnull align 8 dereferenceable(16) [[THIS:%.*]], i32* nonnull align 4 dereferenceable(4) [[D:%.*]]) unnamed_addr #[[ATTR8]] align 2 { 4045 // CHECK4-NEXT: entry: 4046 // CHECK4-NEXT: [[THIS_ADDR:%.*]] = alloca %struct.SS*, align 8 4047 // CHECK4-NEXT: [[D_ADDR:%.*]] = alloca i32*, align 8 4048 // CHECK4-NEXT: [[A2:%.*]] = alloca i32*, align 8 4049 // CHECK4-NEXT: [[B4:%.*]] = alloca i32, align 4 4050 // CHECK4-NEXT: [[C5:%.*]] = alloca i32*, align 8 4051 // CHECK4-NEXT: store %struct.SS* [[THIS]], %struct.SS** [[THIS_ADDR]], align 8 4052 // CHECK4-NEXT: store i32* [[D]], i32** [[D_ADDR]], align 8 4053 // CHECK4-NEXT: [[THIS1:%.*]] = load %struct.SS*, %struct.SS** [[THIS_ADDR]], align 8 4054 // CHECK4-NEXT: [[A:%.*]] = getelementptr inbounds [[STRUCT_SS:%.*]], %struct.SS* [[THIS1]], i32 0, i32 0 4055 // CHECK4-NEXT: store i32 0, i32* [[A]], align 8 4056 // CHECK4-NEXT: [[B:%.*]] = getelementptr inbounds [[STRUCT_SS]], %struct.SS* [[THIS1]], i32 0, i32 1 4057 // CHECK4-NEXT: [[BF_LOAD:%.*]] = load i8, i8* [[B]], align 4 4058 // CHECK4-NEXT: [[BF_CLEAR:%.*]] = and i8 [[BF_LOAD]], -16 4059 // CHECK4-NEXT: store i8 [[BF_CLEAR]], i8* [[B]], align 4 4060 // CHECK4-NEXT: [[C:%.*]] = getelementptr inbounds [[STRUCT_SS]], %struct.SS* [[THIS1]], i32 0, i32 2 4061 // CHECK4-NEXT: [[TMP0:%.*]] = load i32*, i32** [[D_ADDR]], align 8 4062 // CHECK4-NEXT: store i32* [[TMP0]], i32** [[C]], align 8 4063 // CHECK4-NEXT: [[A3:%.*]] = getelementptr inbounds [[STRUCT_SS]], %struct.SS* [[THIS1]], i32 0, i32 0 4064 // CHECK4-NEXT: store i32* [[A3]], i32** [[A2]], align 8 4065 // CHECK4-NEXT: [[C6:%.*]] = getelementptr inbounds [[STRUCT_SS]], %struct.SS* [[THIS1]], i32 0, i32 2 4066 // CHECK4-NEXT: [[TMP1:%.*]] = load i32*, i32** [[C6]], align 8 4067 // CHECK4-NEXT: store i32* [[TMP1]], i32** [[C5]], align 8 4068 // CHECK4-NEXT: [[TMP2:%.*]] = load i32*, i32** [[A2]], align 8 4069 // CHECK4-NEXT: [[TMP3:%.*]] = load i32*, i32** [[C5]], align 8 4070 // CHECK4-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB2]], i32 4, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, %struct.SS*, i32*, i32*, i32*)* @.omp_outlined..4 to void (i32*, i32*, ...)*), %struct.SS* [[THIS1]], i32* [[TMP2]], i32* [[B4]], i32* [[TMP3]]) 4071 // CHECK4-NEXT: [[TMP4:%.*]] = load i32, i32* [[B4]], align 4 4072 // CHECK4-NEXT: [[B7:%.*]] = getelementptr inbounds [[STRUCT_SS]], %struct.SS* [[THIS1]], i32 0, i32 1 4073 // CHECK4-NEXT: [[TMP5:%.*]] = trunc i32 [[TMP4]] to i8 4074 // CHECK4-NEXT: [[BF_LOAD8:%.*]] = load i8, i8* [[B7]], align 4 4075 // CHECK4-NEXT: [[BF_VALUE:%.*]] = and i8 [[TMP5]], 15 4076 // CHECK4-NEXT: [[BF_CLEAR9:%.*]] = and i8 [[BF_LOAD8]], -16 4077 // CHECK4-NEXT: [[BF_SET:%.*]] = or i8 [[BF_CLEAR9]], [[BF_VALUE]] 4078 // CHECK4-NEXT: store i8 [[BF_SET]], i8* [[B7]], align 4 4079 // CHECK4-NEXT: ret void 4080 // 4081 // 4082 // CHECK4-LABEL: define {{[^@]+}}@.omp_outlined..4 4083 // CHECK4-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], %struct.SS* [[THIS:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]], i32* nonnull align 4 dereferenceable(4) [[B:%.*]], i32* nonnull align 4 dereferenceable(4) [[C:%.*]]) #[[ATTR2]] { 4084 // CHECK4-NEXT: entry: 4085 // CHECK4-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 4086 // CHECK4-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 4087 // CHECK4-NEXT: [[THIS_ADDR:%.*]] = alloca %struct.SS*, align 8 4088 // CHECK4-NEXT: [[A_ADDR:%.*]] = alloca i32*, align 8 4089 // CHECK4-NEXT: [[B_ADDR:%.*]] = alloca i32*, align 8 4090 // CHECK4-NEXT: [[C_ADDR:%.*]] = alloca i32*, align 8 4091 // CHECK4-NEXT: [[TMP:%.*]] = alloca i32*, align 8 4092 // CHECK4-NEXT: [[_TMP1:%.*]] = alloca i32*, align 8 4093 // CHECK4-NEXT: [[A2:%.*]] = alloca i32, align 4 4094 // CHECK4-NEXT: [[_TMP3:%.*]] = alloca i32*, align 8 4095 // CHECK4-NEXT: [[B4:%.*]] = alloca i32, align 4 4096 // CHECK4-NEXT: [[C5:%.*]] = alloca i32, align 4 4097 // CHECK4-NEXT: [[_TMP6:%.*]] = alloca i32*, align 8 4098 // CHECK4-NEXT: [[BLOCK:%.*]] = alloca <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, %struct.SS*, i32*, i32*, i32 }>, align 8 4099 // CHECK4-NEXT: [[DOTOMP_REDUCTION_RED_LIST:%.*]] = alloca [3 x i8*], align 8 4100 // CHECK4-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8 4101 // CHECK4-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8 4102 // CHECK4-NEXT: store %struct.SS* [[THIS]], %struct.SS** [[THIS_ADDR]], align 8 4103 // CHECK4-NEXT: store i32* [[A]], i32** [[A_ADDR]], align 8 4104 // CHECK4-NEXT: store i32* [[B]], i32** [[B_ADDR]], align 8 4105 // CHECK4-NEXT: store i32* [[C]], i32** [[C_ADDR]], align 8 4106 // CHECK4-NEXT: [[TMP0:%.*]] = load %struct.SS*, %struct.SS** [[THIS_ADDR]], align 8 4107 // CHECK4-NEXT: [[TMP1:%.*]] = load i32*, i32** [[A_ADDR]], align 8 4108 // CHECK4-NEXT: [[TMP2:%.*]] = load i32*, i32** [[B_ADDR]], align 8 4109 // CHECK4-NEXT: [[TMP3:%.*]] = load i32*, i32** [[C_ADDR]], align 8 4110 // CHECK4-NEXT: store i32* [[TMP1]], i32** [[TMP]], align 8 4111 // CHECK4-NEXT: store i32* [[TMP3]], i32** [[_TMP1]], align 8 4112 // CHECK4-NEXT: [[TMP4:%.*]] = load i32*, i32** [[TMP]], align 8 4113 // CHECK4-NEXT: store i32 0, i32* [[A2]], align 4 4114 // CHECK4-NEXT: store i32* [[A2]], i32** [[_TMP3]], align 8 4115 // CHECK4-NEXT: store i32 0, i32* [[B4]], align 4 4116 // CHECK4-NEXT: [[TMP5:%.*]] = load i32*, i32** [[_TMP1]], align 8 4117 // CHECK4-NEXT: store i32 0, i32* [[C5]], align 4 4118 // CHECK4-NEXT: store i32* [[C5]], i32** [[_TMP6]], align 8 4119 // CHECK4-NEXT: [[BLOCK_ISA:%.*]] = getelementptr inbounds <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, %struct.SS*, i32*, i32*, i32 }>, <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, %struct.SS*, i32*, i32*, i32 }>* [[BLOCK]], i32 0, i32 0 4120 // CHECK4-NEXT: store i8* bitcast (i8** @_NSConcreteStackBlock to i8*), i8** [[BLOCK_ISA]], align 8 4121 // CHECK4-NEXT: [[BLOCK_FLAGS:%.*]] = getelementptr inbounds <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, %struct.SS*, i32*, i32*, i32 }>, <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, %struct.SS*, i32*, i32*, i32 }>* [[BLOCK]], i32 0, i32 1 4122 // CHECK4-NEXT: store i32 1073741824, i32* [[BLOCK_FLAGS]], align 8 4123 // CHECK4-NEXT: [[BLOCK_RESERVED:%.*]] = getelementptr inbounds <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, %struct.SS*, i32*, i32*, i32 }>, <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, %struct.SS*, i32*, i32*, i32 }>* [[BLOCK]], i32 0, i32 2 4124 // CHECK4-NEXT: store i32 0, i32* [[BLOCK_RESERVED]], align 4 4125 // CHECK4-NEXT: [[BLOCK_INVOKE:%.*]] = getelementptr inbounds <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, %struct.SS*, i32*, i32*, i32 }>, <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, %struct.SS*, i32*, i32*, i32 }>* [[BLOCK]], i32 0, i32 3 4126 // CHECK4-NEXT: store i8* bitcast (void (i8*)* @g_block_invoke_2 to i8*), i8** [[BLOCK_INVOKE]], align 8 4127 // CHECK4-NEXT: [[BLOCK_DESCRIPTOR:%.*]] = getelementptr inbounds <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, %struct.SS*, i32*, i32*, i32 }>, <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, %struct.SS*, i32*, i32*, i32 }>* [[BLOCK]], i32 0, i32 4 4128 // CHECK4-NEXT: store %struct.__block_descriptor* bitcast ({ i64, i64, i8*, i8* }* @__block_descriptor_tmp.7 to %struct.__block_descriptor*), %struct.__block_descriptor** [[BLOCK_DESCRIPTOR]], align 8 4129 // CHECK4-NEXT: [[BLOCK_CAPTURED_THIS_ADDR:%.*]] = getelementptr inbounds <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, %struct.SS*, i32*, i32*, i32 }>, <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, %struct.SS*, i32*, i32*, i32 }>* [[BLOCK]], i32 0, i32 5 4130 // CHECK4-NEXT: store %struct.SS* [[TMP0]], %struct.SS** [[BLOCK_CAPTURED_THIS_ADDR]], align 8 4131 // CHECK4-NEXT: [[BLOCK_CAPTURED:%.*]] = getelementptr inbounds <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, %struct.SS*, i32*, i32*, i32 }>, <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, %struct.SS*, i32*, i32*, i32 }>* [[BLOCK]], i32 0, i32 6 4132 // CHECK4-NEXT: [[TMP6:%.*]] = load i32*, i32** [[_TMP3]], align 8 4133 // CHECK4-NEXT: store i32* [[TMP6]], i32** [[BLOCK_CAPTURED]], align 8 4134 // CHECK4-NEXT: [[BLOCK_CAPTURED7:%.*]] = getelementptr inbounds <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, %struct.SS*, i32*, i32*, i32 }>, <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, %struct.SS*, i32*, i32*, i32 }>* [[BLOCK]], i32 0, i32 8 4135 // CHECK4-NEXT: [[TMP7:%.*]] = load i32, i32* [[B4]], align 4 4136 // CHECK4-NEXT: store i32 [[TMP7]], i32* [[BLOCK_CAPTURED7]], align 8 4137 // CHECK4-NEXT: [[BLOCK_CAPTURED8:%.*]] = getelementptr inbounds <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, %struct.SS*, i32*, i32*, i32 }>, <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, %struct.SS*, i32*, i32*, i32 }>* [[BLOCK]], i32 0, i32 7 4138 // CHECK4-NEXT: [[TMP8:%.*]] = load i32*, i32** [[_TMP6]], align 8 4139 // CHECK4-NEXT: store i32* [[TMP8]], i32** [[BLOCK_CAPTURED8]], align 8 4140 // CHECK4-NEXT: [[TMP9:%.*]] = bitcast <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, %struct.SS*, i32*, i32*, i32 }>* [[BLOCK]] to void ()* 4141 // CHECK4-NEXT: [[BLOCK_LITERAL:%.*]] = bitcast void ()* [[TMP9]] to %struct.__block_literal_generic* 4142 // CHECK4-NEXT: [[TMP10:%.*]] = getelementptr inbounds [[STRUCT___BLOCK_LITERAL_GENERIC:%.*]], %struct.__block_literal_generic* [[BLOCK_LITERAL]], i32 0, i32 3 4143 // CHECK4-NEXT: [[TMP11:%.*]] = bitcast %struct.__block_literal_generic* [[BLOCK_LITERAL]] to i8* 4144 // CHECK4-NEXT: [[TMP12:%.*]] = load i8*, i8** [[TMP10]], align 8 4145 // CHECK4-NEXT: [[TMP13:%.*]] = bitcast i8* [[TMP12]] to void (i8*)* 4146 // CHECK4-NEXT: call void [[TMP13]](i8* [[TMP11]]) 4147 // CHECK4-NEXT: [[TMP14:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOMP_REDUCTION_RED_LIST]], i64 0, i64 0 4148 // CHECK4-NEXT: [[TMP15:%.*]] = bitcast i32* [[A2]] to i8* 4149 // CHECK4-NEXT: store i8* [[TMP15]], i8** [[TMP14]], align 8 4150 // CHECK4-NEXT: [[TMP16:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOMP_REDUCTION_RED_LIST]], i64 0, i64 1 4151 // CHECK4-NEXT: [[TMP17:%.*]] = bitcast i32* [[B4]] to i8* 4152 // CHECK4-NEXT: store i8* [[TMP17]], i8** [[TMP16]], align 8 4153 // CHECK4-NEXT: [[TMP18:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOMP_REDUCTION_RED_LIST]], i64 0, i64 2 4154 // CHECK4-NEXT: [[TMP19:%.*]] = bitcast i32* [[C5]] to i8* 4155 // CHECK4-NEXT: store i8* [[TMP19]], i8** [[TMP18]], align 8 4156 // CHECK4-NEXT: [[TMP20:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 4157 // CHECK4-NEXT: [[TMP21:%.*]] = load i32, i32* [[TMP20]], align 4 4158 // CHECK4-NEXT: [[TMP22:%.*]] = bitcast [3 x i8*]* [[DOTOMP_REDUCTION_RED_LIST]] to i8* 4159 // CHECK4-NEXT: [[TMP23:%.*]] = call i32 @__kmpc_reduce_nowait(%struct.ident_t* @[[GLOB1]], i32 [[TMP21]], i32 3, i64 24, i8* [[TMP22]], void (i8*, i8*)* @.omp.reduction.reduction_func.8, [8 x i32]* @.gomp_critical_user_.reduction.var) 4160 // CHECK4-NEXT: switch i32 [[TMP23]], label [[DOTOMP_REDUCTION_DEFAULT:%.*]] [ 4161 // CHECK4-NEXT: i32 1, label [[DOTOMP_REDUCTION_CASE1:%.*]] 4162 // CHECK4-NEXT: i32 2, label [[DOTOMP_REDUCTION_CASE2:%.*]] 4163 // CHECK4-NEXT: ] 4164 // CHECK4: .omp.reduction.case1: 4165 // CHECK4-NEXT: [[TMP24:%.*]] = load i32, i32* [[TMP4]], align 4 4166 // CHECK4-NEXT: [[TMP25:%.*]] = load i32, i32* [[A2]], align 4 4167 // CHECK4-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP24]], [[TMP25]] 4168 // CHECK4-NEXT: store i32 [[ADD]], i32* [[TMP4]], align 4 4169 // CHECK4-NEXT: [[TMP26:%.*]] = load i32, i32* [[TMP2]], align 4 4170 // CHECK4-NEXT: [[TMP27:%.*]] = load i32, i32* [[B4]], align 4 4171 // CHECK4-NEXT: [[ADD9:%.*]] = add nsw i32 [[TMP26]], [[TMP27]] 4172 // CHECK4-NEXT: store i32 [[ADD9]], i32* [[TMP2]], align 4 4173 // CHECK4-NEXT: [[TMP28:%.*]] = load i32, i32* [[TMP5]], align 4 4174 // CHECK4-NEXT: [[TMP29:%.*]] = load i32, i32* [[C5]], align 4 4175 // CHECK4-NEXT: [[ADD10:%.*]] = add nsw i32 [[TMP28]], [[TMP29]] 4176 // CHECK4-NEXT: store i32 [[ADD10]], i32* [[TMP5]], align 4 4177 // CHECK4-NEXT: call void @__kmpc_end_reduce_nowait(%struct.ident_t* @[[GLOB1]], i32 [[TMP21]], [8 x i32]* @.gomp_critical_user_.reduction.var) 4178 // CHECK4-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]] 4179 // CHECK4: .omp.reduction.case2: 4180 // CHECK4-NEXT: [[TMP30:%.*]] = load i32, i32* [[A2]], align 4 4181 // CHECK4-NEXT: [[TMP31:%.*]] = atomicrmw add i32* [[TMP4]], i32 [[TMP30]] monotonic, align 4 4182 // CHECK4-NEXT: [[TMP32:%.*]] = load i32, i32* [[B4]], align 4 4183 // CHECK4-NEXT: [[TMP33:%.*]] = atomicrmw add i32* [[TMP2]], i32 [[TMP32]] monotonic, align 4 4184 // CHECK4-NEXT: [[TMP34:%.*]] = load i32, i32* [[C5]], align 4 4185 // CHECK4-NEXT: [[TMP35:%.*]] = atomicrmw add i32* [[TMP5]], i32 [[TMP34]] monotonic, align 4 4186 // CHECK4-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]] 4187 // CHECK4: .omp.reduction.default: 4188 // CHECK4-NEXT: ret void 4189 // 4190 // 4191 // CHECK4-LABEL: define {{[^@]+}}@g_block_invoke_2 4192 // CHECK4-SAME: (i8* [[DOTBLOCK_DESCRIPTOR:%.*]]) #[[ATTR8]] { 4193 // CHECK4-NEXT: entry: 4194 // CHECK4-NEXT: [[DOTBLOCK_DESCRIPTOR_ADDR:%.*]] = alloca i8*, align 8 4195 // CHECK4-NEXT: [[BLOCK_ADDR:%.*]] = alloca <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, %struct.SS*, i32*, i32*, i32 }>*, align 8 4196 // CHECK4-NEXT: store i8* [[DOTBLOCK_DESCRIPTOR]], i8** [[DOTBLOCK_DESCRIPTOR_ADDR]], align 8 4197 // CHECK4-NEXT: [[BLOCK:%.*]] = bitcast i8* [[DOTBLOCK_DESCRIPTOR]] to <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, %struct.SS*, i32*, i32*, i32 }>* 4198 // CHECK4-NEXT: store <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, %struct.SS*, i32*, i32*, i32 }>* [[BLOCK]], <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, %struct.SS*, i32*, i32*, i32 }>** [[BLOCK_ADDR]], align 8 4199 // CHECK4-NEXT: [[BLOCK_CAPTURED_THIS:%.*]] = getelementptr inbounds <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, %struct.SS*, i32*, i32*, i32 }>, <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, %struct.SS*, i32*, i32*, i32 }>* [[BLOCK]], i32 0, i32 5 4200 // CHECK4-NEXT: [[THIS:%.*]] = load %struct.SS*, %struct.SS** [[BLOCK_CAPTURED_THIS]], align 8 4201 // CHECK4-NEXT: [[BLOCK_CAPTURE_ADDR:%.*]] = getelementptr inbounds <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, %struct.SS*, i32*, i32*, i32 }>, <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, %struct.SS*, i32*, i32*, i32 }>* [[BLOCK]], i32 0, i32 6 4202 // CHECK4-NEXT: [[TMP0:%.*]] = load i32*, i32** [[BLOCK_CAPTURE_ADDR]], align 8 4203 // CHECK4-NEXT: [[TMP1:%.*]] = load i32, i32* [[TMP0]], align 4 4204 // CHECK4-NEXT: [[INC:%.*]] = add nsw i32 [[TMP1]], 1 4205 // CHECK4-NEXT: store i32 [[INC]], i32* [[TMP0]], align 4 4206 // CHECK4-NEXT: [[BLOCK_CAPTURE_ADDR1:%.*]] = getelementptr inbounds <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, %struct.SS*, i32*, i32*, i32 }>, <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, %struct.SS*, i32*, i32*, i32 }>* [[BLOCK]], i32 0, i32 8 4207 // CHECK4-NEXT: [[TMP2:%.*]] = load i32, i32* [[BLOCK_CAPTURE_ADDR1]], align 8 4208 // CHECK4-NEXT: [[DEC:%.*]] = add nsw i32 [[TMP2]], -1 4209 // CHECK4-NEXT: store i32 [[DEC]], i32* [[BLOCK_CAPTURE_ADDR1]], align 8 4210 // CHECK4-NEXT: [[BLOCK_CAPTURE_ADDR2:%.*]] = getelementptr inbounds <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, %struct.SS*, i32*, i32*, i32 }>, <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, %struct.SS*, i32*, i32*, i32 }>* [[BLOCK]], i32 0, i32 7 4211 // CHECK4-NEXT: [[TMP3:%.*]] = load i32*, i32** [[BLOCK_CAPTURE_ADDR2]], align 8 4212 // CHECK4-NEXT: [[TMP4:%.*]] = load i32, i32* [[TMP3]], align 4 4213 // CHECK4-NEXT: [[DIV:%.*]] = sdiv i32 [[TMP4]], 1 4214 // CHECK4-NEXT: store i32 [[DIV]], i32* [[TMP3]], align 4 4215 // CHECK4-NEXT: [[BLOCK_CAPTURE_ADDR3:%.*]] = getelementptr inbounds <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, %struct.SS*, i32*, i32*, i32 }>, <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, %struct.SS*, i32*, i32*, i32 }>* [[BLOCK]], i32 0, i32 6 4216 // CHECK4-NEXT: [[TMP5:%.*]] = load i32*, i32** [[BLOCK_CAPTURE_ADDR3]], align 8 4217 // CHECK4-NEXT: [[BLOCK_CAPTURE_ADDR4:%.*]] = getelementptr inbounds <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, %struct.SS*, i32*, i32*, i32 }>, <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, %struct.SS*, i32*, i32*, i32 }>* [[BLOCK]], i32 0, i32 8 4218 // CHECK4-NEXT: [[BLOCK_CAPTURE_ADDR5:%.*]] = getelementptr inbounds <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, %struct.SS*, i32*, i32*, i32 }>, <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, %struct.SS*, i32*, i32*, i32 }>* [[BLOCK]], i32 0, i32 7 4219 // CHECK4-NEXT: [[TMP6:%.*]] = load i32*, i32** [[BLOCK_CAPTURE_ADDR5]], align 8 4220 // CHECK4-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB2]], i32 4, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, %struct.SS*, i32*, i32*, i32*)* @.omp_outlined..5 to void (i32*, i32*, ...)*), %struct.SS* [[THIS]], i32* [[TMP5]], i32* [[BLOCK_CAPTURE_ADDR4]], i32* [[TMP6]]) 4221 // CHECK4-NEXT: ret void 4222 // 4223 // 4224 // CHECK4-LABEL: define {{[^@]+}}@.omp_outlined..5 4225 // CHECK4-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], %struct.SS* [[THIS:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]], i32* nonnull align 4 dereferenceable(4) [[B:%.*]], i32* nonnull align 4 dereferenceable(4) [[C:%.*]]) #[[ATTR2]] { 4226 // CHECK4-NEXT: entry: 4227 // CHECK4-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 4228 // CHECK4-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 4229 // CHECK4-NEXT: [[THIS_ADDR:%.*]] = alloca %struct.SS*, align 8 4230 // CHECK4-NEXT: [[A_ADDR:%.*]] = alloca i32*, align 8 4231 // CHECK4-NEXT: [[B_ADDR:%.*]] = alloca i32*, align 8 4232 // CHECK4-NEXT: [[C_ADDR:%.*]] = alloca i32*, align 8 4233 // CHECK4-NEXT: [[TMP:%.*]] = alloca i32*, align 8 4234 // CHECK4-NEXT: [[_TMP1:%.*]] = alloca i32*, align 8 4235 // CHECK4-NEXT: [[A2:%.*]] = alloca i32, align 4 4236 // CHECK4-NEXT: [[_TMP3:%.*]] = alloca i32*, align 8 4237 // CHECK4-NEXT: [[B4:%.*]] = alloca i32, align 4 4238 // CHECK4-NEXT: [[C5:%.*]] = alloca i32, align 4 4239 // CHECK4-NEXT: [[_TMP6:%.*]] = alloca i32*, align 8 4240 // CHECK4-NEXT: [[DOTOMP_REDUCTION_RED_LIST:%.*]] = alloca [3 x i8*], align 8 4241 // CHECK4-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8 4242 // CHECK4-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8 4243 // CHECK4-NEXT: store %struct.SS* [[THIS]], %struct.SS** [[THIS_ADDR]], align 8 4244 // CHECK4-NEXT: store i32* [[A]], i32** [[A_ADDR]], align 8 4245 // CHECK4-NEXT: store i32* [[B]], i32** [[B_ADDR]], align 8 4246 // CHECK4-NEXT: store i32* [[C]], i32** [[C_ADDR]], align 8 4247 // CHECK4-NEXT: [[TMP0:%.*]] = load %struct.SS*, %struct.SS** [[THIS_ADDR]], align 8 4248 // CHECK4-NEXT: [[TMP1:%.*]] = load i32*, i32** [[A_ADDR]], align 8 4249 // CHECK4-NEXT: [[TMP2:%.*]] = load i32*, i32** [[B_ADDR]], align 8 4250 // CHECK4-NEXT: [[TMP3:%.*]] = load i32*, i32** [[C_ADDR]], align 8 4251 // CHECK4-NEXT: store i32* [[TMP1]], i32** [[TMP]], align 8 4252 // CHECK4-NEXT: store i32* [[TMP3]], i32** [[_TMP1]], align 8 4253 // CHECK4-NEXT: [[TMP4:%.*]] = load i32*, i32** [[TMP]], align 8 4254 // CHECK4-NEXT: store i32 0, i32* [[A2]], align 4 4255 // CHECK4-NEXT: store i32* [[A2]], i32** [[_TMP3]], align 8 4256 // CHECK4-NEXT: store i32 0, i32* [[B4]], align 4 4257 // CHECK4-NEXT: [[TMP5:%.*]] = load i32*, i32** [[_TMP1]], align 8 4258 // CHECK4-NEXT: store i32 0, i32* [[C5]], align 4 4259 // CHECK4-NEXT: store i32* [[C5]], i32** [[_TMP6]], align 8 4260 // CHECK4-NEXT: [[TMP6:%.*]] = load i32*, i32** [[_TMP3]], align 8 4261 // CHECK4-NEXT: [[TMP7:%.*]] = load i32, i32* [[TMP6]], align 4 4262 // CHECK4-NEXT: [[INC:%.*]] = add nsw i32 [[TMP7]], 1 4263 // CHECK4-NEXT: store i32 [[INC]], i32* [[TMP6]], align 4 4264 // CHECK4-NEXT: [[TMP8:%.*]] = load i32, i32* [[B4]], align 4 4265 // CHECK4-NEXT: [[DEC:%.*]] = add nsw i32 [[TMP8]], -1 4266 // CHECK4-NEXT: store i32 [[DEC]], i32* [[B4]], align 4 4267 // CHECK4-NEXT: [[TMP9:%.*]] = load i32*, i32** [[_TMP6]], align 8 4268 // CHECK4-NEXT: [[TMP10:%.*]] = load i32, i32* [[TMP9]], align 4 4269 // CHECK4-NEXT: [[DIV:%.*]] = sdiv i32 [[TMP10]], 1 4270 // CHECK4-NEXT: store i32 [[DIV]], i32* [[TMP9]], align 4 4271 // CHECK4-NEXT: [[TMP11:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOMP_REDUCTION_RED_LIST]], i64 0, i64 0 4272 // CHECK4-NEXT: [[TMP12:%.*]] = bitcast i32* [[A2]] to i8* 4273 // CHECK4-NEXT: store i8* [[TMP12]], i8** [[TMP11]], align 8 4274 // CHECK4-NEXT: [[TMP13:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOMP_REDUCTION_RED_LIST]], i64 0, i64 1 4275 // CHECK4-NEXT: [[TMP14:%.*]] = bitcast i32* [[B4]] to i8* 4276 // CHECK4-NEXT: store i8* [[TMP14]], i8** [[TMP13]], align 8 4277 // CHECK4-NEXT: [[TMP15:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOMP_REDUCTION_RED_LIST]], i64 0, i64 2 4278 // CHECK4-NEXT: [[TMP16:%.*]] = bitcast i32* [[C5]] to i8* 4279 // CHECK4-NEXT: store i8* [[TMP16]], i8** [[TMP15]], align 8 4280 // CHECK4-NEXT: [[TMP17:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 4281 // CHECK4-NEXT: [[TMP18:%.*]] = load i32, i32* [[TMP17]], align 4 4282 // CHECK4-NEXT: [[TMP19:%.*]] = bitcast [3 x i8*]* [[DOTOMP_REDUCTION_RED_LIST]] to i8* 4283 // CHECK4-NEXT: [[TMP20:%.*]] = call i32 @__kmpc_reduce_nowait(%struct.ident_t* @[[GLOB1]], i32 [[TMP18]], i32 3, i64 24, i8* [[TMP19]], void (i8*, i8*)* @.omp.reduction.reduction_func.6, [8 x i32]* @.gomp_critical_user_.reduction.var) 4284 // CHECK4-NEXT: switch i32 [[TMP20]], label [[DOTOMP_REDUCTION_DEFAULT:%.*]] [ 4285 // CHECK4-NEXT: i32 1, label [[DOTOMP_REDUCTION_CASE1:%.*]] 4286 // CHECK4-NEXT: i32 2, label [[DOTOMP_REDUCTION_CASE2:%.*]] 4287 // CHECK4-NEXT: ] 4288 // CHECK4: .omp.reduction.case1: 4289 // CHECK4-NEXT: [[TMP21:%.*]] = load i32, i32* [[TMP4]], align 4 4290 // CHECK4-NEXT: [[TMP22:%.*]] = load i32, i32* [[A2]], align 4 4291 // CHECK4-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP21]], [[TMP22]] 4292 // CHECK4-NEXT: store i32 [[ADD]], i32* [[TMP4]], align 4 4293 // CHECK4-NEXT: [[TMP23:%.*]] = load i32, i32* [[TMP2]], align 4 4294 // CHECK4-NEXT: [[TMP24:%.*]] = load i32, i32* [[B4]], align 4 4295 // CHECK4-NEXT: [[ADD7:%.*]] = add nsw i32 [[TMP23]], [[TMP24]] 4296 // CHECK4-NEXT: store i32 [[ADD7]], i32* [[TMP2]], align 4 4297 // CHECK4-NEXT: [[TMP25:%.*]] = load i32, i32* [[TMP5]], align 4 4298 // CHECK4-NEXT: [[TMP26:%.*]] = load i32, i32* [[C5]], align 4 4299 // CHECK4-NEXT: [[ADD8:%.*]] = add nsw i32 [[TMP25]], [[TMP26]] 4300 // CHECK4-NEXT: store i32 [[ADD8]], i32* [[TMP5]], align 4 4301 // CHECK4-NEXT: call void @__kmpc_end_reduce_nowait(%struct.ident_t* @[[GLOB1]], i32 [[TMP18]], [8 x i32]* @.gomp_critical_user_.reduction.var) 4302 // CHECK4-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]] 4303 // CHECK4: .omp.reduction.case2: 4304 // CHECK4-NEXT: [[TMP27:%.*]] = load i32, i32* [[A2]], align 4 4305 // CHECK4-NEXT: [[TMP28:%.*]] = atomicrmw add i32* [[TMP4]], i32 [[TMP27]] monotonic, align 4 4306 // CHECK4-NEXT: [[TMP29:%.*]] = load i32, i32* [[B4]], align 4 4307 // CHECK4-NEXT: [[TMP30:%.*]] = atomicrmw add i32* [[TMP2]], i32 [[TMP29]] monotonic, align 4 4308 // CHECK4-NEXT: [[TMP31:%.*]] = load i32, i32* [[C5]], align 4 4309 // CHECK4-NEXT: [[TMP32:%.*]] = atomicrmw add i32* [[TMP5]], i32 [[TMP31]] monotonic, align 4 4310 // CHECK4-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]] 4311 // CHECK4: .omp.reduction.default: 4312 // CHECK4-NEXT: ret void 4313 // 4314 // 4315 // CHECK4-LABEL: define {{[^@]+}}@.omp.reduction.reduction_func.6 4316 // CHECK4-SAME: (i8* [[TMP0:%.*]], i8* [[TMP1:%.*]]) #[[ATTR4]] { 4317 // CHECK4-NEXT: entry: 4318 // CHECK4-NEXT: [[DOTADDR:%.*]] = alloca i8*, align 8 4319 // CHECK4-NEXT: [[DOTADDR1:%.*]] = alloca i8*, align 8 4320 // CHECK4-NEXT: store i8* [[TMP0]], i8** [[DOTADDR]], align 8 4321 // CHECK4-NEXT: store i8* [[TMP1]], i8** [[DOTADDR1]], align 8 4322 // CHECK4-NEXT: [[TMP2:%.*]] = load i8*, i8** [[DOTADDR]], align 8 4323 // CHECK4-NEXT: [[TMP3:%.*]] = bitcast i8* [[TMP2]] to [3 x i8*]* 4324 // CHECK4-NEXT: [[TMP4:%.*]] = load i8*, i8** [[DOTADDR1]], align 8 4325 // CHECK4-NEXT: [[TMP5:%.*]] = bitcast i8* [[TMP4]] to [3 x i8*]* 4326 // CHECK4-NEXT: [[TMP6:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[TMP5]], i64 0, i64 0 4327 // CHECK4-NEXT: [[TMP7:%.*]] = load i8*, i8** [[TMP6]], align 8 4328 // CHECK4-NEXT: [[TMP8:%.*]] = bitcast i8* [[TMP7]] to i32* 4329 // CHECK4-NEXT: [[TMP9:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[TMP3]], i64 0, i64 0 4330 // CHECK4-NEXT: [[TMP10:%.*]] = load i8*, i8** [[TMP9]], align 8 4331 // CHECK4-NEXT: [[TMP11:%.*]] = bitcast i8* [[TMP10]] to i32* 4332 // CHECK4-NEXT: [[TMP12:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[TMP5]], i64 0, i64 1 4333 // CHECK4-NEXT: [[TMP13:%.*]] = load i8*, i8** [[TMP12]], align 8 4334 // CHECK4-NEXT: [[TMP14:%.*]] = bitcast i8* [[TMP13]] to i32* 4335 // CHECK4-NEXT: [[TMP15:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[TMP3]], i64 0, i64 1 4336 // CHECK4-NEXT: [[TMP16:%.*]] = load i8*, i8** [[TMP15]], align 8 4337 // CHECK4-NEXT: [[TMP17:%.*]] = bitcast i8* [[TMP16]] to i32* 4338 // CHECK4-NEXT: [[TMP18:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[TMP5]], i64 0, i64 2 4339 // CHECK4-NEXT: [[TMP19:%.*]] = load i8*, i8** [[TMP18]], align 8 4340 // CHECK4-NEXT: [[TMP20:%.*]] = bitcast i8* [[TMP19]] to i32* 4341 // CHECK4-NEXT: [[TMP21:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[TMP3]], i64 0, i64 2 4342 // CHECK4-NEXT: [[TMP22:%.*]] = load i8*, i8** [[TMP21]], align 8 4343 // CHECK4-NEXT: [[TMP23:%.*]] = bitcast i8* [[TMP22]] to i32* 4344 // CHECK4-NEXT: [[TMP24:%.*]] = load i32, i32* [[TMP11]], align 4 4345 // CHECK4-NEXT: [[TMP25:%.*]] = load i32, i32* [[TMP8]], align 4 4346 // CHECK4-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP24]], [[TMP25]] 4347 // CHECK4-NEXT: store i32 [[ADD]], i32* [[TMP11]], align 4 4348 // CHECK4-NEXT: [[TMP26:%.*]] = load i32, i32* [[TMP17]], align 4 4349 // CHECK4-NEXT: [[TMP27:%.*]] = load i32, i32* [[TMP14]], align 4 4350 // CHECK4-NEXT: [[ADD2:%.*]] = add nsw i32 [[TMP26]], [[TMP27]] 4351 // CHECK4-NEXT: store i32 [[ADD2]], i32* [[TMP17]], align 4 4352 // CHECK4-NEXT: [[TMP28:%.*]] = load i32, i32* [[TMP23]], align 4 4353 // CHECK4-NEXT: [[TMP29:%.*]] = load i32, i32* [[TMP20]], align 4 4354 // CHECK4-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP28]], [[TMP29]] 4355 // CHECK4-NEXT: store i32 [[ADD3]], i32* [[TMP23]], align 4 4356 // CHECK4-NEXT: ret void 4357 // 4358 // 4359 // CHECK4-LABEL: define {{[^@]+}}@.omp.reduction.reduction_func.8 4360 // CHECK4-SAME: (i8* [[TMP0:%.*]], i8* [[TMP1:%.*]]) #[[ATTR4]] { 4361 // CHECK4-NEXT: entry: 4362 // CHECK4-NEXT: [[DOTADDR:%.*]] = alloca i8*, align 8 4363 // CHECK4-NEXT: [[DOTADDR1:%.*]] = alloca i8*, align 8 4364 // CHECK4-NEXT: store i8* [[TMP0]], i8** [[DOTADDR]], align 8 4365 // CHECK4-NEXT: store i8* [[TMP1]], i8** [[DOTADDR1]], align 8 4366 // CHECK4-NEXT: [[TMP2:%.*]] = load i8*, i8** [[DOTADDR]], align 8 4367 // CHECK4-NEXT: [[TMP3:%.*]] = bitcast i8* [[TMP2]] to [3 x i8*]* 4368 // CHECK4-NEXT: [[TMP4:%.*]] = load i8*, i8** [[DOTADDR1]], align 8 4369 // CHECK4-NEXT: [[TMP5:%.*]] = bitcast i8* [[TMP4]] to [3 x i8*]* 4370 // CHECK4-NEXT: [[TMP6:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[TMP5]], i64 0, i64 0 4371 // CHECK4-NEXT: [[TMP7:%.*]] = load i8*, i8** [[TMP6]], align 8 4372 // CHECK4-NEXT: [[TMP8:%.*]] = bitcast i8* [[TMP7]] to i32* 4373 // CHECK4-NEXT: [[TMP9:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[TMP3]], i64 0, i64 0 4374 // CHECK4-NEXT: [[TMP10:%.*]] = load i8*, i8** [[TMP9]], align 8 4375 // CHECK4-NEXT: [[TMP11:%.*]] = bitcast i8* [[TMP10]] to i32* 4376 // CHECK4-NEXT: [[TMP12:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[TMP5]], i64 0, i64 1 4377 // CHECK4-NEXT: [[TMP13:%.*]] = load i8*, i8** [[TMP12]], align 8 4378 // CHECK4-NEXT: [[TMP14:%.*]] = bitcast i8* [[TMP13]] to i32* 4379 // CHECK4-NEXT: [[TMP15:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[TMP3]], i64 0, i64 1 4380 // CHECK4-NEXT: [[TMP16:%.*]] = load i8*, i8** [[TMP15]], align 8 4381 // CHECK4-NEXT: [[TMP17:%.*]] = bitcast i8* [[TMP16]] to i32* 4382 // CHECK4-NEXT: [[TMP18:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[TMP5]], i64 0, i64 2 4383 // CHECK4-NEXT: [[TMP19:%.*]] = load i8*, i8** [[TMP18]], align 8 4384 // CHECK4-NEXT: [[TMP20:%.*]] = bitcast i8* [[TMP19]] to i32* 4385 // CHECK4-NEXT: [[TMP21:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[TMP3]], i64 0, i64 2 4386 // CHECK4-NEXT: [[TMP22:%.*]] = load i8*, i8** [[TMP21]], align 8 4387 // CHECK4-NEXT: [[TMP23:%.*]] = bitcast i8* [[TMP22]] to i32* 4388 // CHECK4-NEXT: [[TMP24:%.*]] = load i32, i32* [[TMP11]], align 4 4389 // CHECK4-NEXT: [[TMP25:%.*]] = load i32, i32* [[TMP8]], align 4 4390 // CHECK4-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP24]], [[TMP25]] 4391 // CHECK4-NEXT: store i32 [[ADD]], i32* [[TMP11]], align 4 4392 // CHECK4-NEXT: [[TMP26:%.*]] = load i32, i32* [[TMP17]], align 4 4393 // CHECK4-NEXT: [[TMP27:%.*]] = load i32, i32* [[TMP14]], align 4 4394 // CHECK4-NEXT: [[ADD2:%.*]] = add nsw i32 [[TMP26]], [[TMP27]] 4395 // CHECK4-NEXT: store i32 [[ADD2]], i32* [[TMP17]], align 4 4396 // CHECK4-NEXT: [[TMP28:%.*]] = load i32, i32* [[TMP23]], align 4 4397 // CHECK4-NEXT: [[TMP29:%.*]] = load i32, i32* [[TMP20]], align 4 4398 // CHECK4-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP28]], [[TMP29]] 4399 // CHECK4-NEXT: store i32 [[ADD3]], i32* [[TMP23]], align 4 4400 // CHECK4-NEXT: ret void 4401 // 4402 //