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 // Test host code gen 3 // RUN: %clang_cc1 -no-opaque-pointers -DLAMBDA -verify -fopenmp -fopenmp-version=45 -x c++ -std=c++11 -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -emit-llvm %s -o - | FileCheck %s --check-prefix=CHECK1 4 // RUN: %clang_cc1 -no-opaque-pointers -DLAMBDA -fopenmp -fopenmp-version=45 -x c++ -std=c++11 -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -emit-pch -o %t %s 5 // RUN: %clang_cc1 -no-opaque-pointers -DLAMBDA -fopenmp -fopenmp-version=45 -x c++ -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck %s --check-prefix=CHECK1 6 // RUN: %clang_cc1 -no-opaque-pointers -DLAMBDA -verify -fopenmp -fopenmp-version=45 -x c++ -std=c++11 -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -emit-llvm %s -o - | FileCheck %s --check-prefix=CHECK3 7 // RUN: %clang_cc1 -no-opaque-pointers -DLAMBDA -fopenmp -fopenmp-version=45 -x c++ -std=c++11 -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -emit-pch -o %t %s 8 // RUN: %clang_cc1 -no-opaque-pointers -DLAMBDA -fopenmp -fopenmp-version=45 -x c++ -std=c++11 -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck %s --check-prefix=CHECK3 9 10 // RUN: %clang_cc1 -no-opaque-pointers -DLAMBDA -verify -fopenmp-simd -fopenmp-version=45 -x c++ -std=c++11 -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -emit-llvm %s -o - | FileCheck %s --check-prefix=CHECK5 11 // RUN: %clang_cc1 -no-opaque-pointers -DLAMBDA -fopenmp-simd -fopenmp-version=45 -x c++ -std=c++11 -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -emit-pch -o %t %s 12 // RUN: %clang_cc1 -no-opaque-pointers -DLAMBDA -fopenmp-simd -fopenmp-version=45 -x c++ -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck %s --check-prefix=CHECK5 13 // RUN: %clang_cc1 -no-opaque-pointers -DLAMBDA -verify -fopenmp-simd -fopenmp-version=45 -x c++ -std=c++11 -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -emit-llvm %s -o - | FileCheck %s --check-prefix=CHECK7 14 // RUN: %clang_cc1 -no-opaque-pointers -DLAMBDA -fopenmp-simd -fopenmp-version=45 -x c++ -std=c++11 -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -emit-pch -o %t %s 15 // RUN: %clang_cc1 -no-opaque-pointers -DLAMBDA -fopenmp-simd -fopenmp-version=45 -x c++ -std=c++11 -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck %s --check-prefix=CHECK7 16 17 // RUN: %clang_cc1 -no-opaque-pointers -verify -fopenmp -fopenmp-version=45 -x c++ -std=c++11 -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -emit-llvm %s -o - | FileCheck %s --check-prefix=CHECK9 18 // RUN: %clang_cc1 -no-opaque-pointers -fopenmp -fopenmp-version=45 -x c++ -std=c++11 -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -emit-pch -o %t %s 19 // RUN: %clang_cc1 -no-opaque-pointers -fopenmp -fopenmp-version=45 -x c++ -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck %s --check-prefix=CHECK9 20 // RUN: %clang_cc1 -no-opaque-pointers -verify -fopenmp -fopenmp-version=45 -x c++ -std=c++11 -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -emit-llvm %s -o - | FileCheck %s --check-prefix=CHECK11 21 // RUN: %clang_cc1 -no-opaque-pointers -fopenmp -fopenmp-version=45 -x c++ -std=c++11 -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -emit-pch -o %t %s 22 // RUN: %clang_cc1 -no-opaque-pointers -fopenmp -fopenmp-version=45 -x c++ -std=c++11 -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck %s --check-prefix=CHECK11 23 24 // RUN: %clang_cc1 -no-opaque-pointers -verify -fopenmp-simd -fopenmp-version=45 -x c++ -std=c++11 -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -emit-llvm %s -o - | FileCheck %s --check-prefix=CHECK13 25 // RUN: %clang_cc1 -no-opaque-pointers -fopenmp-simd -fopenmp-version=45 -x c++ -std=c++11 -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -emit-pch -o %t %s 26 // RUN: %clang_cc1 -no-opaque-pointers -fopenmp-simd -fopenmp-version=45 -x c++ -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck %s --check-prefix=CHECK13 27 // RUN: %clang_cc1 -no-opaque-pointers -verify -fopenmp-simd -fopenmp-version=45 -x c++ -std=c++11 -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -emit-llvm %s -o - | FileCheck %s --check-prefix=CHECK15 28 // RUN: %clang_cc1 -no-opaque-pointers -fopenmp-simd -fopenmp-version=45 -x c++ -std=c++11 -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -emit-pch -o %t %s 29 // RUN: %clang_cc1 -no-opaque-pointers -fopenmp-simd -fopenmp-version=45 -x c++ -std=c++11 -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck %s --check-prefix=CHECK15 30 // expected-no-diagnostics 31 #ifndef HEADER 32 #define HEADER 33 34 35 template <typename T> 36 T tmain() { 37 T *a, *b, *c; 38 int n = 10000; 39 int ch = 100; 40 41 // no schedule clauses 42 #pragma omp target 43 #pragma omp teams 44 #pragma omp distribute parallel for simd 45 for (int i = 0; i < n; ++i) { 46 a[i] = b[i] + c[i]; 47 } 48 49 // dist_schedule: static no chunk 50 #pragma omp target 51 #pragma omp teams 52 #pragma omp distribute parallel for simd dist_schedule(static) 53 for (int i = 0; i < n; ++i) { 54 a[i] = b[i] + c[i]; 55 } 56 57 // dist_schedule: static chunk 58 #pragma omp target 59 #pragma omp teams 60 #pragma omp distribute parallel for simd dist_schedule(static, ch) 61 for (int i = 0; i < n; ++i) { 62 a[i] = b[i] + c[i]; 63 } 64 65 // schedule: static no chunk 66 #pragma omp target 67 #pragma omp teams 68 #pragma omp distribute parallel for simd schedule(static) 69 for (int i = 0; i < n; ++i) { 70 a[i] = b[i] + c[i]; 71 } 72 73 // schedule: static chunk 74 #pragma omp target 75 #pragma omp teams 76 #pragma omp distribute parallel for simd schedule(static, ch) 77 for (int i = 0; i < n; ++i) { 78 a[i] = b[i] + c[i]; 79 } 80 81 // schedule: dynamic no chunk 82 #pragma omp target 83 #pragma omp teams 84 #pragma omp distribute parallel for simd schedule(dynamic) 85 for (int i = 0; i < n; ++i) { 86 a[i] = b[i] + c[i]; 87 } 88 89 // schedule: dynamic chunk 90 #pragma omp target 91 #pragma omp teams 92 #pragma omp distribute parallel for simd schedule(dynamic, ch) 93 for (int i = 0; i < n; ++i) { 94 a[i] = b[i] + c[i]; 95 } 96 97 return T(); 98 } 99 100 int main() { 101 double *a, *b, *c; 102 int n = 10000; 103 int ch = 100; 104 105 #ifdef LAMBDA 106 [&]() { 107 108 109 110 111 112 113 114 115 // no schedule clauses 116 #pragma omp target 117 #pragma omp teams 118 119 #pragma omp distribute parallel for simd 120 for (int i = 0; i < n; ++i) { 121 a[i] = b[i] + c[i]; 122 123 124 // check EUB for distribute 125 126 // initialize omp.iv 127 128 // check exit condition 129 130 // check that PrevLB and PrevUB are passed to the 'for' 131 // check that distlb and distub are properly passed to fork_call 132 133 // increment by stride (distInc - 'parallel for' executes the whole chunk) and latch 134 135 136 // implementation of 'parallel for' 137 138 139 // initialize lb and ub to PrevLB and PrevUB 140 141 // PrevEUB is only used when 'for' has a chunked schedule, otherwise EUB is used 142 // In this case we use EUB 143 144 // initialize omp.iv 145 146 // check exit condition 147 148 // check that PrevLB and PrevUB are passed to the 'for' 149 150 // check stride 1 for 'for' in 'distribute parallel for simd' 151 152 153 [&]() { 154 a[i] = b[i] + c[i]; 155 }(); 156 } 157 158 // dist_schedule: static no chunk (same sa default - no dist_schedule) 159 #pragma omp target 160 #pragma omp teams 161 162 #pragma omp distribute parallel for simd dist_schedule(static) 163 for (int i = 0; i < n; ++i) { 164 a[i] = b[i] + c[i]; 165 166 167 // check EUB for distribute 168 169 // initialize omp.iv 170 171 // check exit condition 172 173 // check that PrevLB and PrevUB are passed to the 'for' 174 // check that distlb and distub are properly passed to fork_call 175 176 // increment by stride (distInc - 'parallel for' executes the whole chunk) and latch 177 178 179 // implementation of 'parallel for' 180 181 182 // initialize lb and ub to PrevLB and PrevUB 183 184 // PrevEUB is only used when 'for' has a chunked schedule, otherwise EUB is used 185 // In this case we use EUB 186 187 // initialize omp.iv 188 189 // check exit condition 190 191 // check that PrevLB and PrevUB are passed to the 'for' 192 193 // check stride 1 for 'for' in 'distribute parallel for simd' 194 195 [&]() { 196 a[i] = b[i] + c[i]; 197 }(); 198 } 199 200 // dist_schedule: static chunk 201 #pragma omp target 202 #pragma omp teams 203 204 #pragma omp distribute parallel for simd dist_schedule(static, ch) 205 for (int i = 0; i < n; ++i) { 206 a[i] = b[i] + c[i]; 207 208 209 // check EUB for distribute 210 211 // initialize omp.iv 212 213 // check exit condition 214 215 // check that PrevLB and PrevUB are passed to the 'for' 216 // check that distlb and distub are properly passed to fork_call 217 218 // check DistInc 219 220 // Update UB 221 222 // Store LB in IV 223 224 225 // loop exit 226 227 // skip implementation of 'parallel for': using default scheduling and was tested above 228 [&]() { 229 a[i] = b[i] + c[i]; 230 }(); 231 } 232 233 // schedule: static no chunk 234 #pragma omp target 235 #pragma omp teams 236 237 #pragma omp distribute parallel for simd schedule(static) 238 for (int i = 0; i < n; ++i) { 239 a[i] = b[i] + c[i]; 240 241 // skip rest of implementation of 'distribute' as it is tested above for default dist_schedule case 242 243 // 'parallel for' implementation is the same as the case without schedule clase (static no chunk is the default) 244 245 246 // initialize lb and ub to PrevLB and PrevUB 247 248 // PrevEUB is only used when 'for' has a chunked schedule, otherwise EUB is used 249 // In this case we use EUB 250 251 // initialize omp.iv 252 253 // check exit condition 254 255 // check that PrevLB and PrevUB are passed to the 'for' 256 257 // check stride 1 for 'for' in 'distribute parallel for simd' 258 259 260 [&]() { 261 a[i] = b[i] + c[i]; 262 }(); 263 } 264 265 // schedule: static chunk 266 #pragma omp target 267 #pragma omp teams 268 269 #pragma omp distribute parallel for simd schedule(static, ch) 270 for (int i = 0; i < n; ++i) { 271 a[i] = b[i] + c[i]; 272 // skip rest of implementation of 'distribute' as it is tested above for default dist_schedule case 273 274 // 'parallel for' implementation using outer and inner loops and PrevEUB 275 276 // initialize lb and ub to PrevLB and PrevUB 277 278 // check PrevEUB (using PrevUB instead of NumIt as upper bound) 279 280 // initialize omp.iv (IV = LB) 281 282 // outer loop: while (IV < UB) { 283 284 285 286 // skip body branch 287 288 // IV = IV + 1 and inner loop latch 289 290 // check NextLB and NextUB 291 292 293 [&]() { 294 a[i] = b[i] + c[i]; 295 }(); 296 } 297 298 // schedule: dynamic no chunk 299 #pragma omp target 300 #pragma omp teams 301 302 #pragma omp distribute parallel for simd schedule(dynamic) 303 for (int i = 0; i < n; ++i) { 304 a[i] = b[i] + c[i]; 305 // skip rest of implementation of 'distribute' as it is tested above for default dist_schedule case 306 307 // 'parallel for' implementation using outer and inner loops and PrevEUB 308 309 // initialize lb and ub to PrevLB and PrevUB 310 311 312 // initialize omp.iv (IV = LB) 313 314 315 // skip body branch 316 317 // IV = IV + 1 and inner loop latch 318 319 // check NextLB and NextUB 320 321 322 [&]() { 323 a[i] = b[i] + c[i]; 324 }(); 325 } 326 327 // schedule: dynamic chunk 328 #pragma omp target 329 #pragma omp teams 330 331 #pragma omp distribute parallel for simd schedule(dynamic, ch) 332 for (int i = 0; i < n; ++i) { 333 a[i] = b[i] + c[i]; 334 // skip rest of implementation of 'distribute' as it is tested above for default dist_schedule case 335 336 // 'parallel for' implementation using outer and inner loops and PrevEUB 337 338 // initialize lb and ub to PrevLB and PrevUB 339 340 341 // initialize omp.iv (IV = LB) 342 343 344 // skip body branch 345 346 // IV = IV + 1 and inner loop latch 347 348 // check NextLB and NextUB 349 350 351 [&]() { 352 a[i] = b[i] + c[i]; 353 }(); 354 } 355 }(); 356 return 0; 357 #else 358 359 360 361 362 363 364 365 366 367 // no schedule clauses 368 #pragma omp target 369 #pragma omp teams 370 371 #pragma omp distribute parallel for simd 372 for (int i = 0; i < n; ++i) { 373 a[i] = b[i] + c[i]; 374 375 376 // check EUB for distribute 377 378 // initialize omp.iv 379 380 // check exit condition 381 382 // check that PrevLB and PrevUB are passed to the 'for' 383 // check that distlb and distub are properly passed to fork_call 384 385 // increment by stride (distInc - 'parallel for' executes the whole chunk) and latch 386 387 388 // implementation of 'parallel for' 389 390 391 // initialize lb and ub to PrevLB and PrevUB 392 393 // PrevEUB is only used when 'for' has a chunked schedule, otherwise EUB is used 394 // In this case we use EUB 395 396 // initialize omp.iv 397 398 // check exit condition 399 400 // check that PrevLB and PrevUB are passed to the 'for' 401 402 // check stride 1 for 'for' in 'distribute parallel for simd' 403 404 } 405 406 // dist_schedule: static no chunk 407 #pragma omp target 408 #pragma omp teams 409 410 #pragma omp distribute parallel for simd dist_schedule(static) 411 for (int i = 0; i < n; ++i) { 412 a[i] = b[i] + c[i]; 413 414 415 // check EUB for distribute 416 417 // initialize omp.iv 418 419 // check exit condition 420 421 // check that PrevLB and PrevUB are passed to the 'for' 422 // check that distlb and distub are properly passed to fork_call 423 424 // increment by stride (distInc - 'parallel for' executes the whole chunk) and latch 425 426 427 // implementation of 'parallel for' 428 429 430 // initialize lb and ub to PrevLB and PrevUB 431 432 // PrevEUB is only used when 'for' has a chunked schedule, otherwise EUB is used 433 // In this case we use EUB 434 435 // initialize omp.iv 436 437 // check exit condition 438 439 // check that PrevLB and PrevUB are passed to the 'for' 440 441 // check stride 1 for 'for' in 'distribute parallel for simd' 442 443 } 444 445 // dist_schedule: static chunk 446 #pragma omp target 447 #pragma omp teams 448 449 #pragma omp distribute parallel for simd dist_schedule(static, ch) 450 for (int i = 0; i < n; ++i) { 451 a[i] = b[i] + c[i]; 452 453 // unlike the previous tests, in this one we have a outer and inner loop for 'distribute' 454 455 // check EUB for distribute 456 457 // initialize omp.iv 458 459 // check exit condition 460 461 // check that PrevLB and PrevUB are passed to the 'for' 462 // check that distlb and distub are properly passed to fork_call 463 464 // check DistInc 465 466 // Update UB 467 468 // Store LB in IV 469 470 471 // loop exit 472 473 // skip implementation of 'parallel for': using default scheduling and was tested above 474 } 475 476 // schedule: static no chunk 477 #pragma omp target 478 #pragma omp teams 479 480 #pragma omp distribute parallel for simd schedule(static) 481 for (int i = 0; i < n; ++i) { 482 a[i] = b[i] + c[i]; 483 484 // skip rest of implementation of 'distribute' as it is tested above for default dist_schedule case 485 486 // 'parallel for' implementation is the same as the case without schedule clase (static no chunk is the default) 487 488 489 // initialize lb and ub to PrevLB and PrevUB 490 491 // PrevEUB is only used when 'for' has a chunked schedule, otherwise EUB is used 492 // In this case we use EUB 493 494 // initialize omp.iv 495 496 // check exit condition 497 498 // check that PrevLB and PrevUB are passed to the 'for' 499 500 // check stride 1 for 'for' in 'distribute parallel for simd' 501 502 } 503 504 // schedule: static chunk 505 #pragma omp target 506 #pragma omp teams 507 508 #pragma omp distribute parallel for simd schedule(static, ch) 509 for (int i = 0; i < n; ++i) { 510 a[i] = b[i] + c[i]; 511 // skip rest of implementation of 'distribute' as it is tested above for default dist_schedule case 512 513 // 'parallel for' implementation using outer and inner loops and PrevEUB 514 515 // initialize lb and ub to PrevLB and PrevUB 516 517 // check PrevEUB (using PrevUB instead of NumIt as upper bound) 518 519 // initialize omp.iv (IV = LB) 520 521 // outer loop: while (IV < UB) { 522 523 524 525 // skip body branch 526 527 // IV = IV + 1 and inner loop latch 528 529 // check NextLB and NextUB 530 531 532 } 533 534 // schedule: dynamic no chunk 535 #pragma omp target 536 #pragma omp teams 537 538 #pragma omp distribute parallel for simd schedule(dynamic) 539 for (int i = 0; i < n; ++i) { 540 a[i] = b[i] + c[i]; 541 // skip rest of implementation of 'distribute' as it is tested above for default dist_schedule case 542 543 // 'parallel for' implementation using outer and inner loops and PrevEUB 544 545 // initialize lb and ub to PrevLB and PrevUB 546 547 548 // initialize omp.iv (IV = LB) 549 550 551 // skip body branch 552 553 // IV = IV + 1 and inner loop latch 554 555 // check NextLB and NextUB 556 557 558 } 559 560 // schedule: dynamic chunk 561 #pragma omp target 562 #pragma omp teams 563 564 #pragma omp distribute parallel for simd schedule(dynamic, ch) 565 for (int i = 0; i < n; ++i) { 566 a[i] = b[i] + c[i]; 567 // skip rest of implementation of 'distribute' as it is tested above for default dist_schedule case 568 569 // 'parallel for' implementation using outer and inner loops and PrevEUB 570 571 // initialize lb and ub to PrevLB and PrevUB 572 573 574 // initialize omp.iv (IV = LB) 575 576 577 // skip body branch 578 579 // IV = IV + 1 and inner loop latch 580 581 // check NextLB and NextUB 582 583 584 } 585 586 return tmain<int>(); 587 #endif 588 } 589 590 // check code 591 592 593 594 595 596 597 598 599 600 601 602 // check EUB for distribute 603 604 // initialize omp.iv 605 606 // check exit condition 607 608 // check that PrevLB and PrevUB are passed to the 'for' 609 // check that distlb and distub are properly passed to fork_call 610 611 // increment by stride (distInc - 'parallel for' executes the whole chunk) and latch 612 613 614 // implementation of 'parallel for' 615 616 617 // initialize lb and ub to PrevLB and PrevUB 618 619 // PrevEUB is only used when 'for' has a chunked schedule, otherwise EUB is used 620 // In this case we use EUB 621 622 // initialize omp.iv 623 624 // check exit condition 625 626 // check that PrevLB and PrevUB are passed to the 'for' 627 628 // check stride 1 for 'for' in 'distribute parallel for simd' 629 630 631 632 633 634 // check EUB for distribute 635 636 // initialize omp.iv 637 638 // check exit condition 639 640 // check that PrevLB and PrevUB are passed to the 'for' 641 // check that distlb and distub are properly passed to fork_call 642 643 // increment by stride (distInc - 'parallel for' executes the whole chunk) and latch 644 645 646 // implementation of 'parallel for' 647 648 649 // initialize lb and ub to PrevLB and PrevUB 650 651 // PrevEUB is only used when 'for' has a chunked schedule, otherwise EUB is used 652 // In this case we use EUB 653 654 // initialize omp.iv 655 656 // check exit condition 657 658 // check that PrevLB and PrevUB are passed to the 'for' 659 660 // check stride 1 for 'for' in 'distribute parallel for simd' 661 662 663 664 665 // unlike the previous tests, in this one we have a outer and inner loop for 'distribute' 666 667 // check EUB for distribute 668 669 // initialize omp.iv 670 671 // check exit condition 672 673 // check that PrevLB and PrevUB are passed to the 'for' 674 // check that distlb and distub are properly passed to fork_call 675 676 // check DistInc 677 678 // Update UB 679 680 // Store LB in IV 681 682 683 // loop exit 684 685 // skip implementation of 'parallel for': using default scheduling and was tested above 686 687 688 689 // skip rest of implementation of 'distribute' as it is tested above for default dist_schedule case 690 691 // 'parallel for' implementation is the same as the case without schedule clase (static no chunk is the default) 692 693 694 // initialize lb and ub to PrevLB and PrevUB 695 696 // PrevEUB is only used when 'for' has a chunked schedule, otherwise EUB is used 697 // In this case we use EUB 698 699 // initialize omp.iv 700 701 // check exit condition 702 703 // check that PrevLB and PrevUB are passed to the 'for' 704 705 // check stride 1 for 'for' in 'distribute parallel for simd' 706 707 708 709 // skip rest of implementation of 'distribute' as it is tested above for default dist_schedule case 710 711 // 'parallel for' implementation using outer and inner loops and PrevEUB 712 713 // initialize lb and ub to PrevLB and PrevUB 714 715 // check PrevEUB (using PrevUB instead of NumIt as upper bound) 716 717 // initialize omp.iv (IV = LB) 718 719 // outer loop: while (IV < UB) { 720 721 722 723 // skip body branch 724 725 // IV = IV + 1 and inner loop latch 726 727 // check NextLB and NextUB 728 729 730 731 732 // skip rest of implementation of 'distribute' as it is tested above for default dist_schedule case 733 734 // 'parallel for' implementation using outer and inner loops and PrevEUB 735 736 // initialize lb and ub to PrevLB and PrevUB 737 738 739 // initialize omp.iv (IV = LB) 740 741 742 // skip body branch 743 744 // IV = IV + 1 and inner loop latch 745 746 // check NextLB and NextUB 747 748 749 750 751 // skip rest of implementation of 'distribute' as it is tested above for default dist_schedule case 752 753 // 'parallel for' implementation using outer and inner loops and PrevEUB 754 755 // initialize lb and ub to PrevLB and PrevUB 756 757 758 // initialize omp.iv (IV = LB) 759 760 761 // skip body branch 762 763 // IV = IV + 1 and inner loop latch 764 765 // check NextLB and NextUB 766 767 768 769 #endif 770 // CHECK1-LABEL: define {{[^@]+}}@main 771 // CHECK1-SAME: () #[[ATTR0:[0-9]+]] { 772 // CHECK1-NEXT: entry: 773 // CHECK1-NEXT: [[RETVAL:%.*]] = alloca i32, align 4 774 // CHECK1-NEXT: [[A:%.*]] = alloca double*, align 8 775 // CHECK1-NEXT: [[B:%.*]] = alloca double*, align 8 776 // CHECK1-NEXT: [[C:%.*]] = alloca double*, align 8 777 // CHECK1-NEXT: [[N:%.*]] = alloca i32, align 4 778 // CHECK1-NEXT: [[CH:%.*]] = alloca i32, align 4 779 // CHECK1-NEXT: [[REF_TMP:%.*]] = alloca [[CLASS_ANON:%.*]], align 8 780 // CHECK1-NEXT: store i32 0, i32* [[RETVAL]], align 4 781 // CHECK1-NEXT: store i32 10000, i32* [[N]], align 4 782 // CHECK1-NEXT: store i32 100, i32* [[CH]], align 4 783 // CHECK1-NEXT: [[TMP0:%.*]] = getelementptr inbounds [[CLASS_ANON]], %class.anon* [[REF_TMP]], i32 0, i32 0 784 // CHECK1-NEXT: store i32* [[N]], i32** [[TMP0]], align 8 785 // CHECK1-NEXT: [[TMP1:%.*]] = getelementptr inbounds [[CLASS_ANON]], %class.anon* [[REF_TMP]], i32 0, i32 1 786 // CHECK1-NEXT: store double** [[A]], double*** [[TMP1]], align 8 787 // CHECK1-NEXT: [[TMP2:%.*]] = getelementptr inbounds [[CLASS_ANON]], %class.anon* [[REF_TMP]], i32 0, i32 2 788 // CHECK1-NEXT: store double** [[B]], double*** [[TMP2]], align 8 789 // CHECK1-NEXT: [[TMP3:%.*]] = getelementptr inbounds [[CLASS_ANON]], %class.anon* [[REF_TMP]], i32 0, i32 3 790 // CHECK1-NEXT: store double** [[C]], double*** [[TMP3]], align 8 791 // CHECK1-NEXT: [[TMP4:%.*]] = getelementptr inbounds [[CLASS_ANON]], %class.anon* [[REF_TMP]], i32 0, i32 4 792 // CHECK1-NEXT: store i32* [[CH]], i32** [[TMP4]], align 8 793 // CHECK1-NEXT: call void @"_ZZ4mainENK3$_0clEv"(%class.anon* noundef nonnull align 8 dereferenceable(40) [[REF_TMP]]) 794 // CHECK1-NEXT: ret i32 0 795 // 796 // 797 // CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l116 798 // CHECK1-SAME: (i64 noundef [[N:%.*]], double* noundef [[A:%.*]], double* noundef [[B:%.*]], double* noundef [[C:%.*]]) #[[ATTR2:[0-9]+]] { 799 // CHECK1-NEXT: entry: 800 // CHECK1-NEXT: [[N_ADDR:%.*]] = alloca i64, align 8 801 // CHECK1-NEXT: [[A_ADDR:%.*]] = alloca double*, align 8 802 // CHECK1-NEXT: [[B_ADDR:%.*]] = alloca double*, align 8 803 // CHECK1-NEXT: [[C_ADDR:%.*]] = alloca double*, align 8 804 // CHECK1-NEXT: store i64 [[N]], i64* [[N_ADDR]], align 8 805 // CHECK1-NEXT: store double* [[A]], double** [[A_ADDR]], align 8 806 // CHECK1-NEXT: store double* [[B]], double** [[B_ADDR]], align 8 807 // CHECK1-NEXT: store double* [[C]], double** [[C_ADDR]], align 8 808 // CHECK1-NEXT: [[CONV:%.*]] = bitcast i64* [[N_ADDR]] to i32* 809 // CHECK1-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3:[0-9]+]], i32 4, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32*, double**, double**, double**)* @.omp_outlined. to void (i32*, i32*, ...)*), i32* [[CONV]], double** [[A_ADDR]], double** [[B_ADDR]], double** [[C_ADDR]]) 810 // CHECK1-NEXT: ret void 811 // 812 // 813 // CHECK1-LABEL: define {{[^@]+}}@.omp_outlined. 814 // CHECK1-SAME: (i32* noalias noundef [[DOTGLOBAL_TID_:%.*]], i32* noalias noundef [[DOTBOUND_TID_:%.*]], i32* noundef nonnull align 4 dereferenceable(4) [[N:%.*]], double** noundef nonnull align 8 dereferenceable(8) [[A:%.*]], double** noundef nonnull align 8 dereferenceable(8) [[B:%.*]], double** noundef nonnull align 8 dereferenceable(8) [[C:%.*]]) #[[ATTR2]] { 815 // CHECK1-NEXT: entry: 816 // CHECK1-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 817 // CHECK1-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 818 // CHECK1-NEXT: [[N_ADDR:%.*]] = alloca i32*, align 8 819 // CHECK1-NEXT: [[A_ADDR:%.*]] = alloca double**, align 8 820 // CHECK1-NEXT: [[B_ADDR:%.*]] = alloca double**, align 8 821 // CHECK1-NEXT: [[C_ADDR:%.*]] = alloca double**, align 8 822 // CHECK1-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 823 // CHECK1-NEXT: [[TMP:%.*]] = alloca i32, align 4 824 // CHECK1-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 825 // CHECK1-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 826 // CHECK1-NEXT: [[I:%.*]] = alloca i32, align 4 827 // CHECK1-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4 828 // CHECK1-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4 829 // CHECK1-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 830 // CHECK1-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 831 // CHECK1-NEXT: [[I3:%.*]] = alloca i32, align 4 832 // CHECK1-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8 833 // CHECK1-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8 834 // CHECK1-NEXT: store i32* [[N]], i32** [[N_ADDR]], align 8 835 // CHECK1-NEXT: store double** [[A]], double*** [[A_ADDR]], align 8 836 // CHECK1-NEXT: store double** [[B]], double*** [[B_ADDR]], align 8 837 // CHECK1-NEXT: store double** [[C]], double*** [[C_ADDR]], align 8 838 // CHECK1-NEXT: [[TMP0:%.*]] = load i32*, i32** [[N_ADDR]], align 8 839 // CHECK1-NEXT: [[TMP1:%.*]] = load double**, double*** [[A_ADDR]], align 8 840 // CHECK1-NEXT: [[TMP2:%.*]] = load double**, double*** [[B_ADDR]], align 8 841 // CHECK1-NEXT: [[TMP3:%.*]] = load double**, double*** [[C_ADDR]], align 8 842 // CHECK1-NEXT: [[TMP4:%.*]] = load i32, i32* [[TMP0]], align 4 843 // CHECK1-NEXT: store i32 [[TMP4]], i32* [[DOTCAPTURE_EXPR_]], align 4 844 // CHECK1-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 845 // CHECK1-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP5]], 0 846 // CHECK1-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 847 // CHECK1-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1 848 // CHECK1-NEXT: store i32 [[SUB2]], i32* [[DOTCAPTURE_EXPR_1]], align 4 849 // CHECK1-NEXT: store i32 0, i32* [[I]], align 4 850 // CHECK1-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 851 // CHECK1-NEXT: [[CMP:%.*]] = icmp slt i32 0, [[TMP6]] 852 // CHECK1-NEXT: br i1 [[CMP]], label [[OMP_PRECOND_THEN:%.*]], label [[OMP_PRECOND_END:%.*]] 853 // CHECK1: omp.precond.then: 854 // CHECK1-NEXT: store i32 0, i32* [[DOTOMP_COMB_LB]], align 4 855 // CHECK1-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 856 // CHECK1-NEXT: store i32 [[TMP7]], i32* [[DOTOMP_COMB_UB]], align 4 857 // CHECK1-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4 858 // CHECK1-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4 859 // CHECK1-NEXT: [[TMP8:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 860 // CHECK1-NEXT: [[TMP9:%.*]] = load i32, i32* [[TMP8]], align 4 861 // CHECK1-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1:[0-9]+]], i32 [[TMP9]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_COMB_LB]], i32* [[DOTOMP_COMB_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1) 862 // CHECK1-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 863 // CHECK1-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 864 // CHECK1-NEXT: [[CMP4:%.*]] = icmp sgt i32 [[TMP10]], [[TMP11]] 865 // CHECK1-NEXT: br i1 [[CMP4]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] 866 // CHECK1: cond.true: 867 // CHECK1-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 868 // CHECK1-NEXT: br label [[COND_END:%.*]] 869 // CHECK1: cond.false: 870 // CHECK1-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 871 // CHECK1-NEXT: br label [[COND_END]] 872 // CHECK1: cond.end: 873 // CHECK1-NEXT: [[COND:%.*]] = phi i32 [ [[TMP12]], [[COND_TRUE]] ], [ [[TMP13]], [[COND_FALSE]] ] 874 // CHECK1-NEXT: store i32 [[COND]], i32* [[DOTOMP_COMB_UB]], align 4 875 // CHECK1-NEXT: [[TMP14:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4 876 // CHECK1-NEXT: store i32 [[TMP14]], i32* [[DOTOMP_IV]], align 4 877 // CHECK1-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] 878 // CHECK1: omp.inner.for.cond: 879 // CHECK1-NEXT: [[TMP15:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !10 880 // CHECK1-NEXT: [[TMP16:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4, !llvm.access.group !10 881 // CHECK1-NEXT: [[CMP5:%.*]] = icmp sle i32 [[TMP15]], [[TMP16]] 882 // CHECK1-NEXT: br i1 [[CMP5]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] 883 // CHECK1: omp.inner.for.body: 884 // CHECK1-NEXT: [[TMP17:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4, !llvm.access.group !10 885 // CHECK1-NEXT: [[TMP18:%.*]] = zext i32 [[TMP17]] to i64 886 // CHECK1-NEXT: [[TMP19:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4, !llvm.access.group !10 887 // CHECK1-NEXT: [[TMP20:%.*]] = zext i32 [[TMP19]] to i64 888 // CHECK1-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 6, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64, i32*, double**, double**, double**)* @.omp_outlined..1 to void (i32*, i32*, ...)*), i64 [[TMP18]], i64 [[TMP20]], i32* [[TMP0]], double** [[TMP1]], double** [[TMP2]], double** [[TMP3]]), !llvm.access.group !10 889 // CHECK1-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] 890 // CHECK1: omp.inner.for.inc: 891 // CHECK1-NEXT: [[TMP21:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !10 892 // CHECK1-NEXT: [[TMP22:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4, !llvm.access.group !10 893 // CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP21]], [[TMP22]] 894 // CHECK1-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4, !llvm.access.group !10 895 // CHECK1-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP11:![0-9]+]] 896 // CHECK1: omp.inner.for.end: 897 // CHECK1-NEXT: br label [[OMP_LOOP_EXIT:%.*]] 898 // CHECK1: omp.loop.exit: 899 // CHECK1-NEXT: [[TMP23:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 900 // CHECK1-NEXT: [[TMP24:%.*]] = load i32, i32* [[TMP23]], align 4 901 // CHECK1-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP24]]) 902 // CHECK1-NEXT: [[TMP25:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4 903 // CHECK1-NEXT: [[TMP26:%.*]] = icmp ne i32 [[TMP25]], 0 904 // CHECK1-NEXT: br i1 [[TMP26]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]] 905 // CHECK1: .omp.final.then: 906 // CHECK1-NEXT: [[TMP27:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 907 // CHECK1-NEXT: [[SUB6:%.*]] = sub nsw i32 [[TMP27]], 0 908 // CHECK1-NEXT: [[DIV7:%.*]] = sdiv i32 [[SUB6]], 1 909 // CHECK1-NEXT: [[MUL:%.*]] = mul nsw i32 [[DIV7]], 1 910 // CHECK1-NEXT: [[ADD8:%.*]] = add nsw i32 0, [[MUL]] 911 // CHECK1-NEXT: store i32 [[ADD8]], i32* [[I3]], align 4 912 // CHECK1-NEXT: br label [[DOTOMP_FINAL_DONE]] 913 // CHECK1: .omp.final.done: 914 // CHECK1-NEXT: br label [[OMP_PRECOND_END]] 915 // CHECK1: omp.precond.end: 916 // CHECK1-NEXT: ret void 917 // 918 // 919 // CHECK1-LABEL: define {{[^@]+}}@.omp_outlined..1 920 // CHECK1-SAME: (i32* noalias noundef [[DOTGLOBAL_TID_:%.*]], i32* noalias noundef [[DOTBOUND_TID_:%.*]], i64 noundef [[DOTPREVIOUS_LB_:%.*]], i64 noundef [[DOTPREVIOUS_UB_:%.*]], i32* noundef nonnull align 4 dereferenceable(4) [[N:%.*]], double** noundef nonnull align 8 dereferenceable(8) [[A:%.*]], double** noundef nonnull align 8 dereferenceable(8) [[B:%.*]], double** noundef nonnull align 8 dereferenceable(8) [[C:%.*]]) #[[ATTR2]] { 921 // CHECK1-NEXT: entry: 922 // CHECK1-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 923 // CHECK1-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 924 // CHECK1-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8 925 // CHECK1-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8 926 // CHECK1-NEXT: [[N_ADDR:%.*]] = alloca i32*, align 8 927 // CHECK1-NEXT: [[A_ADDR:%.*]] = alloca double**, align 8 928 // CHECK1-NEXT: [[B_ADDR:%.*]] = alloca double**, align 8 929 // CHECK1-NEXT: [[C_ADDR:%.*]] = alloca double**, align 8 930 // CHECK1-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 931 // CHECK1-NEXT: [[TMP:%.*]] = alloca i32, align 4 932 // CHECK1-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 933 // CHECK1-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 934 // CHECK1-NEXT: [[I:%.*]] = alloca i32, align 4 935 // CHECK1-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 936 // CHECK1-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4 937 // CHECK1-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 938 // CHECK1-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 939 // CHECK1-NEXT: [[I4:%.*]] = alloca i32, align 4 940 // CHECK1-NEXT: [[REF_TMP:%.*]] = alloca [[CLASS_ANON_0:%.*]], align 8 941 // CHECK1-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8 942 // CHECK1-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8 943 // CHECK1-NEXT: store i64 [[DOTPREVIOUS_LB_]], i64* [[DOTPREVIOUS_LB__ADDR]], align 8 944 // CHECK1-NEXT: store i64 [[DOTPREVIOUS_UB_]], i64* [[DOTPREVIOUS_UB__ADDR]], align 8 945 // CHECK1-NEXT: store i32* [[N]], i32** [[N_ADDR]], align 8 946 // CHECK1-NEXT: store double** [[A]], double*** [[A_ADDR]], align 8 947 // CHECK1-NEXT: store double** [[B]], double*** [[B_ADDR]], align 8 948 // CHECK1-NEXT: store double** [[C]], double*** [[C_ADDR]], align 8 949 // CHECK1-NEXT: [[TMP0:%.*]] = load i32*, i32** [[N_ADDR]], align 8 950 // CHECK1-NEXT: [[TMP1:%.*]] = load double**, double*** [[A_ADDR]], align 8 951 // CHECK1-NEXT: [[TMP2:%.*]] = load double**, double*** [[B_ADDR]], align 8 952 // CHECK1-NEXT: [[TMP3:%.*]] = load double**, double*** [[C_ADDR]], align 8 953 // CHECK1-NEXT: [[TMP4:%.*]] = load i32, i32* [[TMP0]], align 4 954 // CHECK1-NEXT: store i32 [[TMP4]], i32* [[DOTCAPTURE_EXPR_]], align 4 955 // CHECK1-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 956 // CHECK1-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP5]], 0 957 // CHECK1-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 958 // CHECK1-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1 959 // CHECK1-NEXT: store i32 [[SUB2]], i32* [[DOTCAPTURE_EXPR_1]], align 4 960 // CHECK1-NEXT: store i32 0, i32* [[I]], align 4 961 // CHECK1-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 962 // CHECK1-NEXT: [[CMP:%.*]] = icmp slt i32 0, [[TMP6]] 963 // CHECK1-NEXT: br i1 [[CMP]], label [[OMP_PRECOND_THEN:%.*]], label [[OMP_PRECOND_END:%.*]] 964 // CHECK1: omp.precond.then: 965 // CHECK1-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4 966 // CHECK1-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 967 // CHECK1-NEXT: store i32 [[TMP7]], i32* [[DOTOMP_UB]], align 4 968 // CHECK1-NEXT: [[TMP8:%.*]] = load i64, i64* [[DOTPREVIOUS_LB__ADDR]], align 8 969 // CHECK1-NEXT: [[CONV:%.*]] = trunc i64 [[TMP8]] to i32 970 // CHECK1-NEXT: [[TMP9:%.*]] = load i64, i64* [[DOTPREVIOUS_UB__ADDR]], align 8 971 // CHECK1-NEXT: [[CONV3:%.*]] = trunc i64 [[TMP9]] to i32 972 // CHECK1-NEXT: store i32 [[CONV]], i32* [[DOTOMP_LB]], align 4 973 // CHECK1-NEXT: store i32 [[CONV3]], i32* [[DOTOMP_UB]], align 4 974 // CHECK1-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4 975 // CHECK1-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4 976 // CHECK1-NEXT: [[TMP10:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 977 // CHECK1-NEXT: [[TMP11:%.*]] = load i32, i32* [[TMP10]], align 4 978 // CHECK1-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB2:[0-9]+]], i32 [[TMP11]], i32 34, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1) 979 // CHECK1-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 980 // CHECK1-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 981 // CHECK1-NEXT: [[CMP5:%.*]] = icmp sgt i32 [[TMP12]], [[TMP13]] 982 // CHECK1-NEXT: br i1 [[CMP5]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] 983 // CHECK1: cond.true: 984 // CHECK1-NEXT: [[TMP14:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 985 // CHECK1-NEXT: br label [[COND_END:%.*]] 986 // CHECK1: cond.false: 987 // CHECK1-NEXT: [[TMP15:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 988 // CHECK1-NEXT: br label [[COND_END]] 989 // CHECK1: cond.end: 990 // CHECK1-NEXT: [[COND:%.*]] = phi i32 [ [[TMP14]], [[COND_TRUE]] ], [ [[TMP15]], [[COND_FALSE]] ] 991 // CHECK1-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4 992 // CHECK1-NEXT: [[TMP16:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4 993 // CHECK1-NEXT: store i32 [[TMP16]], i32* [[DOTOMP_IV]], align 4 994 // CHECK1-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] 995 // CHECK1: omp.inner.for.cond: 996 // CHECK1-NEXT: [[TMP17:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !14 997 // CHECK1-NEXT: [[TMP18:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4, !llvm.access.group !14 998 // CHECK1-NEXT: [[CMP6:%.*]] = icmp sle i32 [[TMP17]], [[TMP18]] 999 // CHECK1-NEXT: br i1 [[CMP6]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] 1000 // CHECK1: omp.inner.for.body: 1001 // CHECK1-NEXT: [[TMP19:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !14 1002 // CHECK1-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP19]], 1 1003 // CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] 1004 // CHECK1-NEXT: store i32 [[ADD]], i32* [[I4]], align 4, !llvm.access.group !14 1005 // CHECK1-NEXT: [[TMP20:%.*]] = load double*, double** [[TMP2]], align 8, !llvm.access.group !14 1006 // CHECK1-NEXT: [[TMP21:%.*]] = load i32, i32* [[I4]], align 4, !llvm.access.group !14 1007 // CHECK1-NEXT: [[IDXPROM:%.*]] = sext i32 [[TMP21]] to i64 1008 // CHECK1-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds double, double* [[TMP20]], i64 [[IDXPROM]] 1009 // CHECK1-NEXT: [[TMP22:%.*]] = load double, double* [[ARRAYIDX]], align 8, !llvm.access.group !14 1010 // CHECK1-NEXT: [[TMP23:%.*]] = load double*, double** [[TMP3]], align 8, !llvm.access.group !14 1011 // CHECK1-NEXT: [[TMP24:%.*]] = load i32, i32* [[I4]], align 4, !llvm.access.group !14 1012 // CHECK1-NEXT: [[IDXPROM7:%.*]] = sext i32 [[TMP24]] to i64 1013 // CHECK1-NEXT: [[ARRAYIDX8:%.*]] = getelementptr inbounds double, double* [[TMP23]], i64 [[IDXPROM7]] 1014 // CHECK1-NEXT: [[TMP25:%.*]] = load double, double* [[ARRAYIDX8]], align 8, !llvm.access.group !14 1015 // CHECK1-NEXT: [[ADD9:%.*]] = fadd double [[TMP22]], [[TMP25]] 1016 // CHECK1-NEXT: [[TMP26:%.*]] = load double*, double** [[TMP1]], align 8, !llvm.access.group !14 1017 // CHECK1-NEXT: [[TMP27:%.*]] = load i32, i32* [[I4]], align 4, !llvm.access.group !14 1018 // CHECK1-NEXT: [[IDXPROM10:%.*]] = sext i32 [[TMP27]] to i64 1019 // CHECK1-NEXT: [[ARRAYIDX11:%.*]] = getelementptr inbounds double, double* [[TMP26]], i64 [[IDXPROM10]] 1020 // CHECK1-NEXT: store double [[ADD9]], double* [[ARRAYIDX11]], align 8, !llvm.access.group !14 1021 // CHECK1-NEXT: [[TMP28:%.*]] = getelementptr inbounds [[CLASS_ANON_0]], %class.anon.0* [[REF_TMP]], i32 0, i32 0 1022 // CHECK1-NEXT: store double** [[TMP1]], double*** [[TMP28]], align 8, !llvm.access.group !14 1023 // CHECK1-NEXT: [[TMP29:%.*]] = getelementptr inbounds [[CLASS_ANON_0]], %class.anon.0* [[REF_TMP]], i32 0, i32 1 1024 // CHECK1-NEXT: store i32* [[I4]], i32** [[TMP29]], align 8, !llvm.access.group !14 1025 // CHECK1-NEXT: [[TMP30:%.*]] = getelementptr inbounds [[CLASS_ANON_0]], %class.anon.0* [[REF_TMP]], i32 0, i32 2 1026 // CHECK1-NEXT: store double** [[TMP2]], double*** [[TMP30]], align 8, !llvm.access.group !14 1027 // CHECK1-NEXT: [[TMP31:%.*]] = getelementptr inbounds [[CLASS_ANON_0]], %class.anon.0* [[REF_TMP]], i32 0, i32 3 1028 // CHECK1-NEXT: store double** [[TMP3]], double*** [[TMP31]], align 8, !llvm.access.group !14 1029 // CHECK1-NEXT: call void @"_ZZZ4mainENK3$_0clEvENKUlvE_clEv"(%class.anon.0* noundef nonnull align 8 dereferenceable(32) [[REF_TMP]]), !llvm.access.group !14 1030 // CHECK1-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] 1031 // CHECK1: omp.body.continue: 1032 // CHECK1-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] 1033 // CHECK1: omp.inner.for.inc: 1034 // CHECK1-NEXT: [[TMP32:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !14 1035 // CHECK1-NEXT: [[ADD12:%.*]] = add nsw i32 [[TMP32]], 1 1036 // CHECK1-NEXT: store i32 [[ADD12]], i32* [[DOTOMP_IV]], align 4, !llvm.access.group !14 1037 // CHECK1-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP15:![0-9]+]] 1038 // CHECK1: omp.inner.for.end: 1039 // CHECK1-NEXT: br label [[OMP_LOOP_EXIT:%.*]] 1040 // CHECK1: omp.loop.exit: 1041 // CHECK1-NEXT: [[TMP33:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 1042 // CHECK1-NEXT: [[TMP34:%.*]] = load i32, i32* [[TMP33]], align 4 1043 // CHECK1-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP34]]) 1044 // CHECK1-NEXT: [[TMP35:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4 1045 // CHECK1-NEXT: [[TMP36:%.*]] = icmp ne i32 [[TMP35]], 0 1046 // CHECK1-NEXT: br i1 [[TMP36]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]] 1047 // CHECK1: .omp.final.then: 1048 // CHECK1-NEXT: [[TMP37:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 1049 // CHECK1-NEXT: [[SUB13:%.*]] = sub nsw i32 [[TMP37]], 0 1050 // CHECK1-NEXT: [[DIV14:%.*]] = sdiv i32 [[SUB13]], 1 1051 // CHECK1-NEXT: [[MUL15:%.*]] = mul nsw i32 [[DIV14]], 1 1052 // CHECK1-NEXT: [[ADD16:%.*]] = add nsw i32 0, [[MUL15]] 1053 // CHECK1-NEXT: store i32 [[ADD16]], i32* [[I4]], align 4 1054 // CHECK1-NEXT: br label [[DOTOMP_FINAL_DONE]] 1055 // CHECK1: .omp.final.done: 1056 // CHECK1-NEXT: br label [[OMP_PRECOND_END]] 1057 // CHECK1: omp.precond.end: 1058 // CHECK1-NEXT: ret void 1059 // 1060 // 1061 // CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l159 1062 // CHECK1-SAME: (i64 noundef [[N:%.*]], double* noundef [[A:%.*]], double* noundef [[B:%.*]], double* noundef [[C:%.*]]) #[[ATTR2]] { 1063 // CHECK1-NEXT: entry: 1064 // CHECK1-NEXT: [[N_ADDR:%.*]] = alloca i64, align 8 1065 // CHECK1-NEXT: [[A_ADDR:%.*]] = alloca double*, align 8 1066 // CHECK1-NEXT: [[B_ADDR:%.*]] = alloca double*, align 8 1067 // CHECK1-NEXT: [[C_ADDR:%.*]] = alloca double*, align 8 1068 // CHECK1-NEXT: store i64 [[N]], i64* [[N_ADDR]], align 8 1069 // CHECK1-NEXT: store double* [[A]], double** [[A_ADDR]], align 8 1070 // CHECK1-NEXT: store double* [[B]], double** [[B_ADDR]], align 8 1071 // CHECK1-NEXT: store double* [[C]], double** [[C_ADDR]], align 8 1072 // CHECK1-NEXT: [[CONV:%.*]] = bitcast i64* [[N_ADDR]] to i32* 1073 // CHECK1-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 4, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32*, double**, double**, double**)* @.omp_outlined..2 to void (i32*, i32*, ...)*), i32* [[CONV]], double** [[A_ADDR]], double** [[B_ADDR]], double** [[C_ADDR]]) 1074 // CHECK1-NEXT: ret void 1075 // 1076 // 1077 // CHECK1-LABEL: define {{[^@]+}}@.omp_outlined..2 1078 // CHECK1-SAME: (i32* noalias noundef [[DOTGLOBAL_TID_:%.*]], i32* noalias noundef [[DOTBOUND_TID_:%.*]], i32* noundef nonnull align 4 dereferenceable(4) [[N:%.*]], double** noundef nonnull align 8 dereferenceable(8) [[A:%.*]], double** noundef nonnull align 8 dereferenceable(8) [[B:%.*]], double** noundef nonnull align 8 dereferenceable(8) [[C:%.*]]) #[[ATTR2]] { 1079 // CHECK1-NEXT: entry: 1080 // CHECK1-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 1081 // CHECK1-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 1082 // CHECK1-NEXT: [[N_ADDR:%.*]] = alloca i32*, align 8 1083 // CHECK1-NEXT: [[A_ADDR:%.*]] = alloca double**, align 8 1084 // CHECK1-NEXT: [[B_ADDR:%.*]] = alloca double**, align 8 1085 // CHECK1-NEXT: [[C_ADDR:%.*]] = alloca double**, align 8 1086 // CHECK1-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 1087 // CHECK1-NEXT: [[TMP:%.*]] = alloca i32, align 4 1088 // CHECK1-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 1089 // CHECK1-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 1090 // CHECK1-NEXT: [[I:%.*]] = alloca i32, align 4 1091 // CHECK1-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4 1092 // CHECK1-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4 1093 // CHECK1-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 1094 // CHECK1-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 1095 // CHECK1-NEXT: [[I3:%.*]] = alloca i32, align 4 1096 // CHECK1-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8 1097 // CHECK1-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8 1098 // CHECK1-NEXT: store i32* [[N]], i32** [[N_ADDR]], align 8 1099 // CHECK1-NEXT: store double** [[A]], double*** [[A_ADDR]], align 8 1100 // CHECK1-NEXT: store double** [[B]], double*** [[B_ADDR]], align 8 1101 // CHECK1-NEXT: store double** [[C]], double*** [[C_ADDR]], align 8 1102 // CHECK1-NEXT: [[TMP0:%.*]] = load i32*, i32** [[N_ADDR]], align 8 1103 // CHECK1-NEXT: [[TMP1:%.*]] = load double**, double*** [[A_ADDR]], align 8 1104 // CHECK1-NEXT: [[TMP2:%.*]] = load double**, double*** [[B_ADDR]], align 8 1105 // CHECK1-NEXT: [[TMP3:%.*]] = load double**, double*** [[C_ADDR]], align 8 1106 // CHECK1-NEXT: [[TMP4:%.*]] = load i32, i32* [[TMP0]], align 4 1107 // CHECK1-NEXT: store i32 [[TMP4]], i32* [[DOTCAPTURE_EXPR_]], align 4 1108 // CHECK1-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 1109 // CHECK1-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP5]], 0 1110 // CHECK1-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 1111 // CHECK1-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1 1112 // CHECK1-NEXT: store i32 [[SUB2]], i32* [[DOTCAPTURE_EXPR_1]], align 4 1113 // CHECK1-NEXT: store i32 0, i32* [[I]], align 4 1114 // CHECK1-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 1115 // CHECK1-NEXT: [[CMP:%.*]] = icmp slt i32 0, [[TMP6]] 1116 // CHECK1-NEXT: br i1 [[CMP]], label [[OMP_PRECOND_THEN:%.*]], label [[OMP_PRECOND_END:%.*]] 1117 // CHECK1: omp.precond.then: 1118 // CHECK1-NEXT: store i32 0, i32* [[DOTOMP_COMB_LB]], align 4 1119 // CHECK1-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 1120 // CHECK1-NEXT: store i32 [[TMP7]], i32* [[DOTOMP_COMB_UB]], align 4 1121 // CHECK1-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4 1122 // CHECK1-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4 1123 // CHECK1-NEXT: [[TMP8:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 1124 // CHECK1-NEXT: [[TMP9:%.*]] = load i32, i32* [[TMP8]], align 4 1125 // CHECK1-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1]], i32 [[TMP9]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_COMB_LB]], i32* [[DOTOMP_COMB_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1) 1126 // CHECK1-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 1127 // CHECK1-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 1128 // CHECK1-NEXT: [[CMP4:%.*]] = icmp sgt i32 [[TMP10]], [[TMP11]] 1129 // CHECK1-NEXT: br i1 [[CMP4]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] 1130 // CHECK1: cond.true: 1131 // CHECK1-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 1132 // CHECK1-NEXT: br label [[COND_END:%.*]] 1133 // CHECK1: cond.false: 1134 // CHECK1-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 1135 // CHECK1-NEXT: br label [[COND_END]] 1136 // CHECK1: cond.end: 1137 // CHECK1-NEXT: [[COND:%.*]] = phi i32 [ [[TMP12]], [[COND_TRUE]] ], [ [[TMP13]], [[COND_FALSE]] ] 1138 // CHECK1-NEXT: store i32 [[COND]], i32* [[DOTOMP_COMB_UB]], align 4 1139 // CHECK1-NEXT: [[TMP14:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4 1140 // CHECK1-NEXT: store i32 [[TMP14]], i32* [[DOTOMP_IV]], align 4 1141 // CHECK1-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] 1142 // CHECK1: omp.inner.for.cond: 1143 // CHECK1-NEXT: [[TMP15:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !19 1144 // CHECK1-NEXT: [[TMP16:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4, !llvm.access.group !19 1145 // CHECK1-NEXT: [[CMP5:%.*]] = icmp sle i32 [[TMP15]], [[TMP16]] 1146 // CHECK1-NEXT: br i1 [[CMP5]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] 1147 // CHECK1: omp.inner.for.body: 1148 // CHECK1-NEXT: [[TMP17:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4, !llvm.access.group !19 1149 // CHECK1-NEXT: [[TMP18:%.*]] = zext i32 [[TMP17]] to i64 1150 // CHECK1-NEXT: [[TMP19:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4, !llvm.access.group !19 1151 // CHECK1-NEXT: [[TMP20:%.*]] = zext i32 [[TMP19]] to i64 1152 // CHECK1-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 6, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64, i32*, double**, double**, double**)* @.omp_outlined..3 to void (i32*, i32*, ...)*), i64 [[TMP18]], i64 [[TMP20]], i32* [[TMP0]], double** [[TMP1]], double** [[TMP2]], double** [[TMP3]]), !llvm.access.group !19 1153 // CHECK1-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] 1154 // CHECK1: omp.inner.for.inc: 1155 // CHECK1-NEXT: [[TMP21:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !19 1156 // CHECK1-NEXT: [[TMP22:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4, !llvm.access.group !19 1157 // CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP21]], [[TMP22]] 1158 // CHECK1-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4, !llvm.access.group !19 1159 // CHECK1-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP20:![0-9]+]] 1160 // CHECK1: omp.inner.for.end: 1161 // CHECK1-NEXT: br label [[OMP_LOOP_EXIT:%.*]] 1162 // CHECK1: omp.loop.exit: 1163 // CHECK1-NEXT: [[TMP23:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 1164 // CHECK1-NEXT: [[TMP24:%.*]] = load i32, i32* [[TMP23]], align 4 1165 // CHECK1-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP24]]) 1166 // CHECK1-NEXT: [[TMP25:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4 1167 // CHECK1-NEXT: [[TMP26:%.*]] = icmp ne i32 [[TMP25]], 0 1168 // CHECK1-NEXT: br i1 [[TMP26]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]] 1169 // CHECK1: .omp.final.then: 1170 // CHECK1-NEXT: [[TMP27:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 1171 // CHECK1-NEXT: [[SUB6:%.*]] = sub nsw i32 [[TMP27]], 0 1172 // CHECK1-NEXT: [[DIV7:%.*]] = sdiv i32 [[SUB6]], 1 1173 // CHECK1-NEXT: [[MUL:%.*]] = mul nsw i32 [[DIV7]], 1 1174 // CHECK1-NEXT: [[ADD8:%.*]] = add nsw i32 0, [[MUL]] 1175 // CHECK1-NEXT: store i32 [[ADD8]], i32* [[I3]], align 4 1176 // CHECK1-NEXT: br label [[DOTOMP_FINAL_DONE]] 1177 // CHECK1: .omp.final.done: 1178 // CHECK1-NEXT: br label [[OMP_PRECOND_END]] 1179 // CHECK1: omp.precond.end: 1180 // CHECK1-NEXT: ret void 1181 // 1182 // 1183 // CHECK1-LABEL: define {{[^@]+}}@.omp_outlined..3 1184 // CHECK1-SAME: (i32* noalias noundef [[DOTGLOBAL_TID_:%.*]], i32* noalias noundef [[DOTBOUND_TID_:%.*]], i64 noundef [[DOTPREVIOUS_LB_:%.*]], i64 noundef [[DOTPREVIOUS_UB_:%.*]], i32* noundef nonnull align 4 dereferenceable(4) [[N:%.*]], double** noundef nonnull align 8 dereferenceable(8) [[A:%.*]], double** noundef nonnull align 8 dereferenceable(8) [[B:%.*]], double** noundef nonnull align 8 dereferenceable(8) [[C:%.*]]) #[[ATTR2]] { 1185 // CHECK1-NEXT: entry: 1186 // CHECK1-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 1187 // CHECK1-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 1188 // CHECK1-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8 1189 // CHECK1-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8 1190 // CHECK1-NEXT: [[N_ADDR:%.*]] = alloca i32*, align 8 1191 // CHECK1-NEXT: [[A_ADDR:%.*]] = alloca double**, align 8 1192 // CHECK1-NEXT: [[B_ADDR:%.*]] = alloca double**, align 8 1193 // CHECK1-NEXT: [[C_ADDR:%.*]] = alloca double**, align 8 1194 // CHECK1-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 1195 // CHECK1-NEXT: [[TMP:%.*]] = alloca i32, align 4 1196 // CHECK1-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 1197 // CHECK1-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 1198 // CHECK1-NEXT: [[I:%.*]] = alloca i32, align 4 1199 // CHECK1-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 1200 // CHECK1-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4 1201 // CHECK1-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 1202 // CHECK1-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 1203 // CHECK1-NEXT: [[I4:%.*]] = alloca i32, align 4 1204 // CHECK1-NEXT: [[REF_TMP:%.*]] = alloca [[CLASS_ANON_1:%.*]], align 8 1205 // CHECK1-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8 1206 // CHECK1-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8 1207 // CHECK1-NEXT: store i64 [[DOTPREVIOUS_LB_]], i64* [[DOTPREVIOUS_LB__ADDR]], align 8 1208 // CHECK1-NEXT: store i64 [[DOTPREVIOUS_UB_]], i64* [[DOTPREVIOUS_UB__ADDR]], align 8 1209 // CHECK1-NEXT: store i32* [[N]], i32** [[N_ADDR]], align 8 1210 // CHECK1-NEXT: store double** [[A]], double*** [[A_ADDR]], align 8 1211 // CHECK1-NEXT: store double** [[B]], double*** [[B_ADDR]], align 8 1212 // CHECK1-NEXT: store double** [[C]], double*** [[C_ADDR]], align 8 1213 // CHECK1-NEXT: [[TMP0:%.*]] = load i32*, i32** [[N_ADDR]], align 8 1214 // CHECK1-NEXT: [[TMP1:%.*]] = load double**, double*** [[A_ADDR]], align 8 1215 // CHECK1-NEXT: [[TMP2:%.*]] = load double**, double*** [[B_ADDR]], align 8 1216 // CHECK1-NEXT: [[TMP3:%.*]] = load double**, double*** [[C_ADDR]], align 8 1217 // CHECK1-NEXT: [[TMP4:%.*]] = load i32, i32* [[TMP0]], align 4 1218 // CHECK1-NEXT: store i32 [[TMP4]], i32* [[DOTCAPTURE_EXPR_]], align 4 1219 // CHECK1-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 1220 // CHECK1-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP5]], 0 1221 // CHECK1-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 1222 // CHECK1-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1 1223 // CHECK1-NEXT: store i32 [[SUB2]], i32* [[DOTCAPTURE_EXPR_1]], align 4 1224 // CHECK1-NEXT: store i32 0, i32* [[I]], align 4 1225 // CHECK1-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 1226 // CHECK1-NEXT: [[CMP:%.*]] = icmp slt i32 0, [[TMP6]] 1227 // CHECK1-NEXT: br i1 [[CMP]], label [[OMP_PRECOND_THEN:%.*]], label [[OMP_PRECOND_END:%.*]] 1228 // CHECK1: omp.precond.then: 1229 // CHECK1-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4 1230 // CHECK1-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 1231 // CHECK1-NEXT: store i32 [[TMP7]], i32* [[DOTOMP_UB]], align 4 1232 // CHECK1-NEXT: [[TMP8:%.*]] = load i64, i64* [[DOTPREVIOUS_LB__ADDR]], align 8 1233 // CHECK1-NEXT: [[CONV:%.*]] = trunc i64 [[TMP8]] to i32 1234 // CHECK1-NEXT: [[TMP9:%.*]] = load i64, i64* [[DOTPREVIOUS_UB__ADDR]], align 8 1235 // CHECK1-NEXT: [[CONV3:%.*]] = trunc i64 [[TMP9]] to i32 1236 // CHECK1-NEXT: store i32 [[CONV]], i32* [[DOTOMP_LB]], align 4 1237 // CHECK1-NEXT: store i32 [[CONV3]], i32* [[DOTOMP_UB]], align 4 1238 // CHECK1-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4 1239 // CHECK1-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4 1240 // CHECK1-NEXT: [[TMP10:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 1241 // CHECK1-NEXT: [[TMP11:%.*]] = load i32, i32* [[TMP10]], align 4 1242 // CHECK1-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB2]], i32 [[TMP11]], i32 34, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1) 1243 // CHECK1-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 1244 // CHECK1-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 1245 // CHECK1-NEXT: [[CMP5:%.*]] = icmp sgt i32 [[TMP12]], [[TMP13]] 1246 // CHECK1-NEXT: br i1 [[CMP5]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] 1247 // CHECK1: cond.true: 1248 // CHECK1-NEXT: [[TMP14:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 1249 // CHECK1-NEXT: br label [[COND_END:%.*]] 1250 // CHECK1: cond.false: 1251 // CHECK1-NEXT: [[TMP15:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 1252 // CHECK1-NEXT: br label [[COND_END]] 1253 // CHECK1: cond.end: 1254 // CHECK1-NEXT: [[COND:%.*]] = phi i32 [ [[TMP14]], [[COND_TRUE]] ], [ [[TMP15]], [[COND_FALSE]] ] 1255 // CHECK1-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4 1256 // CHECK1-NEXT: [[TMP16:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4 1257 // CHECK1-NEXT: store i32 [[TMP16]], i32* [[DOTOMP_IV]], align 4 1258 // CHECK1-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] 1259 // CHECK1: omp.inner.for.cond: 1260 // CHECK1-NEXT: [[TMP17:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !22 1261 // CHECK1-NEXT: [[TMP18:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4, !llvm.access.group !22 1262 // CHECK1-NEXT: [[CMP6:%.*]] = icmp sle i32 [[TMP17]], [[TMP18]] 1263 // CHECK1-NEXT: br i1 [[CMP6]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] 1264 // CHECK1: omp.inner.for.body: 1265 // CHECK1-NEXT: [[TMP19:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !22 1266 // CHECK1-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP19]], 1 1267 // CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] 1268 // CHECK1-NEXT: store i32 [[ADD]], i32* [[I4]], align 4, !llvm.access.group !22 1269 // CHECK1-NEXT: [[TMP20:%.*]] = load double*, double** [[TMP2]], align 8, !llvm.access.group !22 1270 // CHECK1-NEXT: [[TMP21:%.*]] = load i32, i32* [[I4]], align 4, !llvm.access.group !22 1271 // CHECK1-NEXT: [[IDXPROM:%.*]] = sext i32 [[TMP21]] to i64 1272 // CHECK1-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds double, double* [[TMP20]], i64 [[IDXPROM]] 1273 // CHECK1-NEXT: [[TMP22:%.*]] = load double, double* [[ARRAYIDX]], align 8, !llvm.access.group !22 1274 // CHECK1-NEXT: [[TMP23:%.*]] = load double*, double** [[TMP3]], align 8, !llvm.access.group !22 1275 // CHECK1-NEXT: [[TMP24:%.*]] = load i32, i32* [[I4]], align 4, !llvm.access.group !22 1276 // CHECK1-NEXT: [[IDXPROM7:%.*]] = sext i32 [[TMP24]] to i64 1277 // CHECK1-NEXT: [[ARRAYIDX8:%.*]] = getelementptr inbounds double, double* [[TMP23]], i64 [[IDXPROM7]] 1278 // CHECK1-NEXT: [[TMP25:%.*]] = load double, double* [[ARRAYIDX8]], align 8, !llvm.access.group !22 1279 // CHECK1-NEXT: [[ADD9:%.*]] = fadd double [[TMP22]], [[TMP25]] 1280 // CHECK1-NEXT: [[TMP26:%.*]] = load double*, double** [[TMP1]], align 8, !llvm.access.group !22 1281 // CHECK1-NEXT: [[TMP27:%.*]] = load i32, i32* [[I4]], align 4, !llvm.access.group !22 1282 // CHECK1-NEXT: [[IDXPROM10:%.*]] = sext i32 [[TMP27]] to i64 1283 // CHECK1-NEXT: [[ARRAYIDX11:%.*]] = getelementptr inbounds double, double* [[TMP26]], i64 [[IDXPROM10]] 1284 // CHECK1-NEXT: store double [[ADD9]], double* [[ARRAYIDX11]], align 8, !llvm.access.group !22 1285 // CHECK1-NEXT: [[TMP28:%.*]] = getelementptr inbounds [[CLASS_ANON_1]], %class.anon.1* [[REF_TMP]], i32 0, i32 0 1286 // CHECK1-NEXT: store double** [[TMP1]], double*** [[TMP28]], align 8, !llvm.access.group !22 1287 // CHECK1-NEXT: [[TMP29:%.*]] = getelementptr inbounds [[CLASS_ANON_1]], %class.anon.1* [[REF_TMP]], i32 0, i32 1 1288 // CHECK1-NEXT: store i32* [[I4]], i32** [[TMP29]], align 8, !llvm.access.group !22 1289 // CHECK1-NEXT: [[TMP30:%.*]] = getelementptr inbounds [[CLASS_ANON_1]], %class.anon.1* [[REF_TMP]], i32 0, i32 2 1290 // CHECK1-NEXT: store double** [[TMP2]], double*** [[TMP30]], align 8, !llvm.access.group !22 1291 // CHECK1-NEXT: [[TMP31:%.*]] = getelementptr inbounds [[CLASS_ANON_1]], %class.anon.1* [[REF_TMP]], i32 0, i32 3 1292 // CHECK1-NEXT: store double** [[TMP3]], double*** [[TMP31]], align 8, !llvm.access.group !22 1293 // CHECK1-NEXT: call void @"_ZZZ4mainENK3$_0clEvENKUlvE0_clEv"(%class.anon.1* noundef nonnull align 8 dereferenceable(32) [[REF_TMP]]), !llvm.access.group !22 1294 // CHECK1-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] 1295 // CHECK1: omp.body.continue: 1296 // CHECK1-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] 1297 // CHECK1: omp.inner.for.inc: 1298 // CHECK1-NEXT: [[TMP32:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !22 1299 // CHECK1-NEXT: [[ADD12:%.*]] = add nsw i32 [[TMP32]], 1 1300 // CHECK1-NEXT: store i32 [[ADD12]], i32* [[DOTOMP_IV]], align 4, !llvm.access.group !22 1301 // CHECK1-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP23:![0-9]+]] 1302 // CHECK1: omp.inner.for.end: 1303 // CHECK1-NEXT: br label [[OMP_LOOP_EXIT:%.*]] 1304 // CHECK1: omp.loop.exit: 1305 // CHECK1-NEXT: [[TMP33:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 1306 // CHECK1-NEXT: [[TMP34:%.*]] = load i32, i32* [[TMP33]], align 4 1307 // CHECK1-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP34]]) 1308 // CHECK1-NEXT: [[TMP35:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4 1309 // CHECK1-NEXT: [[TMP36:%.*]] = icmp ne i32 [[TMP35]], 0 1310 // CHECK1-NEXT: br i1 [[TMP36]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]] 1311 // CHECK1: .omp.final.then: 1312 // CHECK1-NEXT: [[TMP37:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 1313 // CHECK1-NEXT: [[SUB13:%.*]] = sub nsw i32 [[TMP37]], 0 1314 // CHECK1-NEXT: [[DIV14:%.*]] = sdiv i32 [[SUB13]], 1 1315 // CHECK1-NEXT: [[MUL15:%.*]] = mul nsw i32 [[DIV14]], 1 1316 // CHECK1-NEXT: [[ADD16:%.*]] = add nsw i32 0, [[MUL15]] 1317 // CHECK1-NEXT: store i32 [[ADD16]], i32* [[I4]], align 4 1318 // CHECK1-NEXT: br label [[DOTOMP_FINAL_DONE]] 1319 // CHECK1: .omp.final.done: 1320 // CHECK1-NEXT: br label [[OMP_PRECOND_END]] 1321 // CHECK1: omp.precond.end: 1322 // CHECK1-NEXT: ret void 1323 // 1324 // 1325 // CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l201 1326 // CHECK1-SAME: (i64 noundef [[CH:%.*]], i64 noundef [[N:%.*]], double* noundef [[A:%.*]], double* noundef [[B:%.*]], double* noundef [[C:%.*]]) #[[ATTR2]] { 1327 // CHECK1-NEXT: entry: 1328 // CHECK1-NEXT: [[CH_ADDR:%.*]] = alloca i64, align 8 1329 // CHECK1-NEXT: [[N_ADDR:%.*]] = alloca i64, align 8 1330 // CHECK1-NEXT: [[A_ADDR:%.*]] = alloca double*, align 8 1331 // CHECK1-NEXT: [[B_ADDR:%.*]] = alloca double*, align 8 1332 // CHECK1-NEXT: [[C_ADDR:%.*]] = alloca double*, align 8 1333 // CHECK1-NEXT: store i64 [[CH]], i64* [[CH_ADDR]], align 8 1334 // CHECK1-NEXT: store i64 [[N]], i64* [[N_ADDR]], align 8 1335 // CHECK1-NEXT: store double* [[A]], double** [[A_ADDR]], align 8 1336 // CHECK1-NEXT: store double* [[B]], double** [[B_ADDR]], align 8 1337 // CHECK1-NEXT: store double* [[C]], double** [[C_ADDR]], align 8 1338 // CHECK1-NEXT: [[CONV:%.*]] = bitcast i64* [[CH_ADDR]] to i32* 1339 // CHECK1-NEXT: [[CONV1:%.*]] = bitcast i64* [[N_ADDR]] to i32* 1340 // CHECK1-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 5, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32*, i32*, double**, double**, double**)* @.omp_outlined..6 to void (i32*, i32*, ...)*), i32* [[CONV]], i32* [[CONV1]], double** [[A_ADDR]], double** [[B_ADDR]], double** [[C_ADDR]]) 1341 // CHECK1-NEXT: ret void 1342 // 1343 // 1344 // CHECK1-LABEL: define {{[^@]+}}@.omp_outlined..6 1345 // CHECK1-SAME: (i32* noalias noundef [[DOTGLOBAL_TID_:%.*]], i32* noalias noundef [[DOTBOUND_TID_:%.*]], i32* noundef nonnull align 4 dereferenceable(4) [[CH:%.*]], i32* noundef nonnull align 4 dereferenceable(4) [[N:%.*]], double** noundef nonnull align 8 dereferenceable(8) [[A:%.*]], double** noundef nonnull align 8 dereferenceable(8) [[B:%.*]], double** noundef nonnull align 8 dereferenceable(8) [[C:%.*]]) #[[ATTR2]] { 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: [[CH_ADDR:%.*]] = alloca i32*, align 8 1350 // CHECK1-NEXT: [[N_ADDR:%.*]] = alloca i32*, align 8 1351 // CHECK1-NEXT: [[A_ADDR:%.*]] = alloca double**, align 8 1352 // CHECK1-NEXT: [[B_ADDR:%.*]] = alloca double**, align 8 1353 // CHECK1-NEXT: [[C_ADDR:%.*]] = alloca double**, align 8 1354 // CHECK1-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 1355 // CHECK1-NEXT: [[TMP:%.*]] = alloca i32, align 4 1356 // CHECK1-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 1357 // CHECK1-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 1358 // CHECK1-NEXT: [[I:%.*]] = alloca i32, align 4 1359 // CHECK1-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4 1360 // CHECK1-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4 1361 // CHECK1-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 1362 // CHECK1-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 1363 // CHECK1-NEXT: [[I3:%.*]] = alloca i32, align 4 1364 // CHECK1-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8 1365 // CHECK1-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8 1366 // CHECK1-NEXT: store i32* [[CH]], i32** [[CH_ADDR]], align 8 1367 // CHECK1-NEXT: store i32* [[N]], i32** [[N_ADDR]], align 8 1368 // CHECK1-NEXT: store double** [[A]], double*** [[A_ADDR]], align 8 1369 // CHECK1-NEXT: store double** [[B]], double*** [[B_ADDR]], align 8 1370 // CHECK1-NEXT: store double** [[C]], double*** [[C_ADDR]], align 8 1371 // CHECK1-NEXT: [[TMP0:%.*]] = load i32*, i32** [[CH_ADDR]], align 8 1372 // CHECK1-NEXT: [[TMP1:%.*]] = load i32*, i32** [[N_ADDR]], align 8 1373 // CHECK1-NEXT: [[TMP2:%.*]] = load double**, double*** [[A_ADDR]], align 8 1374 // CHECK1-NEXT: [[TMP3:%.*]] = load double**, double*** [[B_ADDR]], align 8 1375 // CHECK1-NEXT: [[TMP4:%.*]] = load double**, double*** [[C_ADDR]], align 8 1376 // CHECK1-NEXT: [[TMP5:%.*]] = load i32, i32* [[TMP1]], align 4 1377 // CHECK1-NEXT: store i32 [[TMP5]], i32* [[DOTCAPTURE_EXPR_]], align 4 1378 // CHECK1-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 1379 // CHECK1-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP6]], 0 1380 // CHECK1-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 1381 // CHECK1-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1 1382 // CHECK1-NEXT: store i32 [[SUB2]], i32* [[DOTCAPTURE_EXPR_1]], align 4 1383 // CHECK1-NEXT: store i32 0, i32* [[I]], align 4 1384 // CHECK1-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 1385 // CHECK1-NEXT: [[CMP:%.*]] = icmp slt i32 0, [[TMP7]] 1386 // CHECK1-NEXT: br i1 [[CMP]], label [[OMP_PRECOND_THEN:%.*]], label [[OMP_PRECOND_END:%.*]] 1387 // CHECK1: omp.precond.then: 1388 // CHECK1-NEXT: store i32 0, i32* [[DOTOMP_COMB_LB]], align 4 1389 // CHECK1-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 1390 // CHECK1-NEXT: store i32 [[TMP8]], i32* [[DOTOMP_COMB_UB]], align 4 1391 // CHECK1-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4 1392 // CHECK1-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4 1393 // CHECK1-NEXT: [[TMP9:%.*]] = load i32, i32* [[TMP0]], align 4 1394 // CHECK1-NEXT: [[TMP10:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 1395 // CHECK1-NEXT: [[TMP11:%.*]] = load i32, i32* [[TMP10]], align 4 1396 // CHECK1-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1]], i32 [[TMP11]], i32 91, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_COMB_LB]], i32* [[DOTOMP_COMB_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 [[TMP9]]) 1397 // CHECK1-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 1398 // CHECK1-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 1399 // CHECK1-NEXT: [[CMP4:%.*]] = icmp sgt i32 [[TMP12]], [[TMP13]] 1400 // CHECK1-NEXT: br i1 [[CMP4]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] 1401 // CHECK1: cond.true: 1402 // CHECK1-NEXT: [[TMP14:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 1403 // CHECK1-NEXT: br label [[COND_END:%.*]] 1404 // CHECK1: cond.false: 1405 // CHECK1-NEXT: [[TMP15:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 1406 // CHECK1-NEXT: br label [[COND_END]] 1407 // CHECK1: cond.end: 1408 // CHECK1-NEXT: [[COND:%.*]] = phi i32 [ [[TMP14]], [[COND_TRUE]] ], [ [[TMP15]], [[COND_FALSE]] ] 1409 // CHECK1-NEXT: store i32 [[COND]], i32* [[DOTOMP_COMB_UB]], align 4 1410 // CHECK1-NEXT: [[TMP16:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4 1411 // CHECK1-NEXT: store i32 [[TMP16]], i32* [[DOTOMP_IV]], align 4 1412 // CHECK1-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] 1413 // CHECK1: omp.inner.for.cond: 1414 // CHECK1-NEXT: [[TMP17:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !25 1415 // CHECK1-NEXT: [[TMP18:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4, !llvm.access.group !25 1416 // CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP18]], 1 1417 // CHECK1-NEXT: [[CMP5:%.*]] = icmp slt i32 [[TMP17]], [[ADD]] 1418 // CHECK1-NEXT: br i1 [[CMP5]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] 1419 // CHECK1: omp.inner.for.body: 1420 // CHECK1-NEXT: [[TMP19:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4, !llvm.access.group !25 1421 // CHECK1-NEXT: [[TMP20:%.*]] = zext i32 [[TMP19]] to i64 1422 // CHECK1-NEXT: [[TMP21:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4, !llvm.access.group !25 1423 // CHECK1-NEXT: [[TMP22:%.*]] = zext i32 [[TMP21]] to i64 1424 // CHECK1-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 6, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64, i32*, double**, double**, double**)* @.omp_outlined..7 to void (i32*, i32*, ...)*), i64 [[TMP20]], i64 [[TMP22]], i32* [[TMP1]], double** [[TMP2]], double** [[TMP3]], double** [[TMP4]]), !llvm.access.group !25 1425 // CHECK1-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] 1426 // CHECK1: omp.inner.for.inc: 1427 // CHECK1-NEXT: [[TMP23:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !25 1428 // CHECK1-NEXT: [[TMP24:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4, !llvm.access.group !25 1429 // CHECK1-NEXT: [[ADD6:%.*]] = add nsw i32 [[TMP23]], [[TMP24]] 1430 // CHECK1-NEXT: store i32 [[ADD6]], i32* [[DOTOMP_IV]], align 4, !llvm.access.group !25 1431 // CHECK1-NEXT: [[TMP25:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4, !llvm.access.group !25 1432 // CHECK1-NEXT: [[TMP26:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4, !llvm.access.group !25 1433 // CHECK1-NEXT: [[ADD7:%.*]] = add nsw i32 [[TMP25]], [[TMP26]] 1434 // CHECK1-NEXT: store i32 [[ADD7]], i32* [[DOTOMP_COMB_LB]], align 4, !llvm.access.group !25 1435 // CHECK1-NEXT: [[TMP27:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4, !llvm.access.group !25 1436 // CHECK1-NEXT: [[TMP28:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4, !llvm.access.group !25 1437 // CHECK1-NEXT: [[ADD8:%.*]] = add nsw i32 [[TMP27]], [[TMP28]] 1438 // CHECK1-NEXT: store i32 [[ADD8]], i32* [[DOTOMP_COMB_UB]], align 4, !llvm.access.group !25 1439 // CHECK1-NEXT: [[TMP29:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4, !llvm.access.group !25 1440 // CHECK1-NEXT: [[TMP30:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4, !llvm.access.group !25 1441 // CHECK1-NEXT: [[CMP9:%.*]] = icmp sgt i32 [[TMP29]], [[TMP30]] 1442 // CHECK1-NEXT: br i1 [[CMP9]], label [[COND_TRUE10:%.*]], label [[COND_FALSE11:%.*]] 1443 // CHECK1: cond.true10: 1444 // CHECK1-NEXT: [[TMP31:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4, !llvm.access.group !25 1445 // CHECK1-NEXT: br label [[COND_END12:%.*]] 1446 // CHECK1: cond.false11: 1447 // CHECK1-NEXT: [[TMP32:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4, !llvm.access.group !25 1448 // CHECK1-NEXT: br label [[COND_END12]] 1449 // CHECK1: cond.end12: 1450 // CHECK1-NEXT: [[COND13:%.*]] = phi i32 [ [[TMP31]], [[COND_TRUE10]] ], [ [[TMP32]], [[COND_FALSE11]] ] 1451 // CHECK1-NEXT: store i32 [[COND13]], i32* [[DOTOMP_COMB_UB]], align 4, !llvm.access.group !25 1452 // CHECK1-NEXT: [[TMP33:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4, !llvm.access.group !25 1453 // CHECK1-NEXT: store i32 [[TMP33]], i32* [[DOTOMP_IV]], align 4, !llvm.access.group !25 1454 // CHECK1-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP26:![0-9]+]] 1455 // CHECK1: omp.inner.for.end: 1456 // CHECK1-NEXT: br label [[OMP_LOOP_EXIT:%.*]] 1457 // CHECK1: omp.loop.exit: 1458 // CHECK1-NEXT: [[TMP34:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 1459 // CHECK1-NEXT: [[TMP35:%.*]] = load i32, i32* [[TMP34]], align 4 1460 // CHECK1-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP35]]) 1461 // CHECK1-NEXT: [[TMP36:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4 1462 // CHECK1-NEXT: [[TMP37:%.*]] = icmp ne i32 [[TMP36]], 0 1463 // CHECK1-NEXT: br i1 [[TMP37]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]] 1464 // CHECK1: .omp.final.then: 1465 // CHECK1-NEXT: [[TMP38:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 1466 // CHECK1-NEXT: [[SUB14:%.*]] = sub nsw i32 [[TMP38]], 0 1467 // CHECK1-NEXT: [[DIV15:%.*]] = sdiv i32 [[SUB14]], 1 1468 // CHECK1-NEXT: [[MUL:%.*]] = mul nsw i32 [[DIV15]], 1 1469 // CHECK1-NEXT: [[ADD16:%.*]] = add nsw i32 0, [[MUL]] 1470 // CHECK1-NEXT: store i32 [[ADD16]], i32* [[I3]], align 4 1471 // CHECK1-NEXT: br label [[DOTOMP_FINAL_DONE]] 1472 // CHECK1: .omp.final.done: 1473 // CHECK1-NEXT: br label [[OMP_PRECOND_END]] 1474 // CHECK1: omp.precond.end: 1475 // CHECK1-NEXT: ret void 1476 // 1477 // 1478 // CHECK1-LABEL: define {{[^@]+}}@.omp_outlined..7 1479 // CHECK1-SAME: (i32* noalias noundef [[DOTGLOBAL_TID_:%.*]], i32* noalias noundef [[DOTBOUND_TID_:%.*]], i64 noundef [[DOTPREVIOUS_LB_:%.*]], i64 noundef [[DOTPREVIOUS_UB_:%.*]], i32* noundef nonnull align 4 dereferenceable(4) [[N:%.*]], double** noundef nonnull align 8 dereferenceable(8) [[A:%.*]], double** noundef nonnull align 8 dereferenceable(8) [[B:%.*]], double** noundef nonnull align 8 dereferenceable(8) [[C:%.*]]) #[[ATTR2]] { 1480 // CHECK1-NEXT: entry: 1481 // CHECK1-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 1482 // CHECK1-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 1483 // CHECK1-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8 1484 // CHECK1-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8 1485 // CHECK1-NEXT: [[N_ADDR:%.*]] = alloca i32*, align 8 1486 // CHECK1-NEXT: [[A_ADDR:%.*]] = alloca double**, align 8 1487 // CHECK1-NEXT: [[B_ADDR:%.*]] = alloca double**, align 8 1488 // CHECK1-NEXT: [[C_ADDR:%.*]] = alloca double**, align 8 1489 // CHECK1-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 1490 // CHECK1-NEXT: [[TMP:%.*]] = alloca i32, align 4 1491 // CHECK1-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 1492 // CHECK1-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 1493 // CHECK1-NEXT: [[I:%.*]] = alloca i32, align 4 1494 // CHECK1-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 1495 // CHECK1-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4 1496 // CHECK1-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 1497 // CHECK1-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 1498 // CHECK1-NEXT: [[I4:%.*]] = alloca i32, align 4 1499 // CHECK1-NEXT: [[REF_TMP:%.*]] = alloca [[CLASS_ANON_2:%.*]], align 8 1500 // CHECK1-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8 1501 // CHECK1-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8 1502 // CHECK1-NEXT: store i64 [[DOTPREVIOUS_LB_]], i64* [[DOTPREVIOUS_LB__ADDR]], align 8 1503 // CHECK1-NEXT: store i64 [[DOTPREVIOUS_UB_]], i64* [[DOTPREVIOUS_UB__ADDR]], align 8 1504 // CHECK1-NEXT: store i32* [[N]], i32** [[N_ADDR]], align 8 1505 // CHECK1-NEXT: store double** [[A]], double*** [[A_ADDR]], align 8 1506 // CHECK1-NEXT: store double** [[B]], double*** [[B_ADDR]], align 8 1507 // CHECK1-NEXT: store double** [[C]], double*** [[C_ADDR]], align 8 1508 // CHECK1-NEXT: [[TMP0:%.*]] = load i32*, i32** [[N_ADDR]], align 8 1509 // CHECK1-NEXT: [[TMP1:%.*]] = load double**, double*** [[A_ADDR]], align 8 1510 // CHECK1-NEXT: [[TMP2:%.*]] = load double**, double*** [[B_ADDR]], align 8 1511 // CHECK1-NEXT: [[TMP3:%.*]] = load double**, double*** [[C_ADDR]], align 8 1512 // CHECK1-NEXT: [[TMP4:%.*]] = load i32, i32* [[TMP0]], align 4 1513 // CHECK1-NEXT: store i32 [[TMP4]], i32* [[DOTCAPTURE_EXPR_]], align 4 1514 // CHECK1-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 1515 // CHECK1-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP5]], 0 1516 // CHECK1-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 1517 // CHECK1-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1 1518 // CHECK1-NEXT: store i32 [[SUB2]], i32* [[DOTCAPTURE_EXPR_1]], align 4 1519 // CHECK1-NEXT: store i32 0, i32* [[I]], align 4 1520 // CHECK1-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 1521 // CHECK1-NEXT: [[CMP:%.*]] = icmp slt i32 0, [[TMP6]] 1522 // CHECK1-NEXT: br i1 [[CMP]], label [[OMP_PRECOND_THEN:%.*]], label [[OMP_PRECOND_END:%.*]] 1523 // CHECK1: omp.precond.then: 1524 // CHECK1-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4 1525 // CHECK1-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 1526 // CHECK1-NEXT: store i32 [[TMP7]], i32* [[DOTOMP_UB]], align 4 1527 // CHECK1-NEXT: [[TMP8:%.*]] = load i64, i64* [[DOTPREVIOUS_LB__ADDR]], align 8 1528 // CHECK1-NEXT: [[CONV:%.*]] = trunc i64 [[TMP8]] to i32 1529 // CHECK1-NEXT: [[TMP9:%.*]] = load i64, i64* [[DOTPREVIOUS_UB__ADDR]], align 8 1530 // CHECK1-NEXT: [[CONV3:%.*]] = trunc i64 [[TMP9]] to i32 1531 // CHECK1-NEXT: store i32 [[CONV]], i32* [[DOTOMP_LB]], align 4 1532 // CHECK1-NEXT: store i32 [[CONV3]], i32* [[DOTOMP_UB]], align 4 1533 // CHECK1-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4 1534 // CHECK1-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4 1535 // CHECK1-NEXT: [[TMP10:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 1536 // CHECK1-NEXT: [[TMP11:%.*]] = load i32, i32* [[TMP10]], align 4 1537 // CHECK1-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB2]], i32 [[TMP11]], i32 34, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1) 1538 // CHECK1-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 1539 // CHECK1-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 1540 // CHECK1-NEXT: [[CMP5:%.*]] = icmp sgt i32 [[TMP12]], [[TMP13]] 1541 // CHECK1-NEXT: br i1 [[CMP5]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] 1542 // CHECK1: cond.true: 1543 // CHECK1-NEXT: [[TMP14:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 1544 // CHECK1-NEXT: br label [[COND_END:%.*]] 1545 // CHECK1: cond.false: 1546 // CHECK1-NEXT: [[TMP15:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 1547 // CHECK1-NEXT: br label [[COND_END]] 1548 // CHECK1: cond.end: 1549 // CHECK1-NEXT: [[COND:%.*]] = phi i32 [ [[TMP14]], [[COND_TRUE]] ], [ [[TMP15]], [[COND_FALSE]] ] 1550 // CHECK1-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4 1551 // CHECK1-NEXT: [[TMP16:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4 1552 // CHECK1-NEXT: store i32 [[TMP16]], i32* [[DOTOMP_IV]], align 4 1553 // CHECK1-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] 1554 // CHECK1: omp.inner.for.cond: 1555 // CHECK1-NEXT: [[TMP17:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !28 1556 // CHECK1-NEXT: [[TMP18:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4, !llvm.access.group !28 1557 // CHECK1-NEXT: [[CMP6:%.*]] = icmp sle i32 [[TMP17]], [[TMP18]] 1558 // CHECK1-NEXT: br i1 [[CMP6]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] 1559 // CHECK1: omp.inner.for.body: 1560 // CHECK1-NEXT: [[TMP19:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !28 1561 // CHECK1-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP19]], 1 1562 // CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] 1563 // CHECK1-NEXT: store i32 [[ADD]], i32* [[I4]], align 4, !llvm.access.group !28 1564 // CHECK1-NEXT: [[TMP20:%.*]] = load double*, double** [[TMP2]], align 8, !llvm.access.group !28 1565 // CHECK1-NEXT: [[TMP21:%.*]] = load i32, i32* [[I4]], align 4, !llvm.access.group !28 1566 // CHECK1-NEXT: [[IDXPROM:%.*]] = sext i32 [[TMP21]] to i64 1567 // CHECK1-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds double, double* [[TMP20]], i64 [[IDXPROM]] 1568 // CHECK1-NEXT: [[TMP22:%.*]] = load double, double* [[ARRAYIDX]], align 8, !llvm.access.group !28 1569 // CHECK1-NEXT: [[TMP23:%.*]] = load double*, double** [[TMP3]], align 8, !llvm.access.group !28 1570 // CHECK1-NEXT: [[TMP24:%.*]] = load i32, i32* [[I4]], align 4, !llvm.access.group !28 1571 // CHECK1-NEXT: [[IDXPROM7:%.*]] = sext i32 [[TMP24]] to i64 1572 // CHECK1-NEXT: [[ARRAYIDX8:%.*]] = getelementptr inbounds double, double* [[TMP23]], i64 [[IDXPROM7]] 1573 // CHECK1-NEXT: [[TMP25:%.*]] = load double, double* [[ARRAYIDX8]], align 8, !llvm.access.group !28 1574 // CHECK1-NEXT: [[ADD9:%.*]] = fadd double [[TMP22]], [[TMP25]] 1575 // CHECK1-NEXT: [[TMP26:%.*]] = load double*, double** [[TMP1]], align 8, !llvm.access.group !28 1576 // CHECK1-NEXT: [[TMP27:%.*]] = load i32, i32* [[I4]], align 4, !llvm.access.group !28 1577 // CHECK1-NEXT: [[IDXPROM10:%.*]] = sext i32 [[TMP27]] to i64 1578 // CHECK1-NEXT: [[ARRAYIDX11:%.*]] = getelementptr inbounds double, double* [[TMP26]], i64 [[IDXPROM10]] 1579 // CHECK1-NEXT: store double [[ADD9]], double* [[ARRAYIDX11]], align 8, !llvm.access.group !28 1580 // CHECK1-NEXT: [[TMP28:%.*]] = getelementptr inbounds [[CLASS_ANON_2]], %class.anon.2* [[REF_TMP]], i32 0, i32 0 1581 // CHECK1-NEXT: store double** [[TMP1]], double*** [[TMP28]], align 8, !llvm.access.group !28 1582 // CHECK1-NEXT: [[TMP29:%.*]] = getelementptr inbounds [[CLASS_ANON_2]], %class.anon.2* [[REF_TMP]], i32 0, i32 1 1583 // CHECK1-NEXT: store i32* [[I4]], i32** [[TMP29]], align 8, !llvm.access.group !28 1584 // CHECK1-NEXT: [[TMP30:%.*]] = getelementptr inbounds [[CLASS_ANON_2]], %class.anon.2* [[REF_TMP]], i32 0, i32 2 1585 // CHECK1-NEXT: store double** [[TMP2]], double*** [[TMP30]], align 8, !llvm.access.group !28 1586 // CHECK1-NEXT: [[TMP31:%.*]] = getelementptr inbounds [[CLASS_ANON_2]], %class.anon.2* [[REF_TMP]], i32 0, i32 3 1587 // CHECK1-NEXT: store double** [[TMP3]], double*** [[TMP31]], align 8, !llvm.access.group !28 1588 // CHECK1-NEXT: call void @"_ZZZ4mainENK3$_0clEvENKUlvE1_clEv"(%class.anon.2* noundef nonnull align 8 dereferenceable(32) [[REF_TMP]]), !llvm.access.group !28 1589 // CHECK1-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] 1590 // CHECK1: omp.body.continue: 1591 // CHECK1-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] 1592 // CHECK1: omp.inner.for.inc: 1593 // CHECK1-NEXT: [[TMP32:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !28 1594 // CHECK1-NEXT: [[ADD12:%.*]] = add nsw i32 [[TMP32]], 1 1595 // CHECK1-NEXT: store i32 [[ADD12]], i32* [[DOTOMP_IV]], align 4, !llvm.access.group !28 1596 // CHECK1-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP29:![0-9]+]] 1597 // CHECK1: omp.inner.for.end: 1598 // CHECK1-NEXT: br label [[OMP_LOOP_EXIT:%.*]] 1599 // CHECK1: omp.loop.exit: 1600 // CHECK1-NEXT: [[TMP33:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 1601 // CHECK1-NEXT: [[TMP34:%.*]] = load i32, i32* [[TMP33]], align 4 1602 // CHECK1-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP34]]) 1603 // CHECK1-NEXT: [[TMP35:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4 1604 // CHECK1-NEXT: [[TMP36:%.*]] = icmp ne i32 [[TMP35]], 0 1605 // CHECK1-NEXT: br i1 [[TMP36]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]] 1606 // CHECK1: .omp.final.then: 1607 // CHECK1-NEXT: [[TMP37:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 1608 // CHECK1-NEXT: [[SUB13:%.*]] = sub nsw i32 [[TMP37]], 0 1609 // CHECK1-NEXT: [[DIV14:%.*]] = sdiv i32 [[SUB13]], 1 1610 // CHECK1-NEXT: [[MUL15:%.*]] = mul nsw i32 [[DIV14]], 1 1611 // CHECK1-NEXT: [[ADD16:%.*]] = add nsw i32 0, [[MUL15]] 1612 // CHECK1-NEXT: store i32 [[ADD16]], i32* [[I4]], align 4 1613 // CHECK1-NEXT: br label [[DOTOMP_FINAL_DONE]] 1614 // CHECK1: .omp.final.done: 1615 // CHECK1-NEXT: br label [[OMP_PRECOND_END]] 1616 // CHECK1: omp.precond.end: 1617 // CHECK1-NEXT: ret void 1618 // 1619 // 1620 // CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l234 1621 // CHECK1-SAME: (i64 noundef [[N:%.*]], double* noundef [[A:%.*]], double* noundef [[B:%.*]], double* noundef [[C:%.*]]) #[[ATTR2]] { 1622 // CHECK1-NEXT: entry: 1623 // CHECK1-NEXT: [[N_ADDR:%.*]] = alloca i64, align 8 1624 // CHECK1-NEXT: [[A_ADDR:%.*]] = alloca double*, align 8 1625 // CHECK1-NEXT: [[B_ADDR:%.*]] = alloca double*, align 8 1626 // CHECK1-NEXT: [[C_ADDR:%.*]] = alloca double*, align 8 1627 // CHECK1-NEXT: store i64 [[N]], i64* [[N_ADDR]], align 8 1628 // CHECK1-NEXT: store double* [[A]], double** [[A_ADDR]], align 8 1629 // CHECK1-NEXT: store double* [[B]], double** [[B_ADDR]], align 8 1630 // CHECK1-NEXT: store double* [[C]], double** [[C_ADDR]], align 8 1631 // CHECK1-NEXT: [[CONV:%.*]] = bitcast i64* [[N_ADDR]] to i32* 1632 // CHECK1-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 4, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32*, double**, double**, double**)* @.omp_outlined..10 to void (i32*, i32*, ...)*), i32* [[CONV]], double** [[A_ADDR]], double** [[B_ADDR]], double** [[C_ADDR]]) 1633 // CHECK1-NEXT: ret void 1634 // 1635 // 1636 // CHECK1-LABEL: define {{[^@]+}}@.omp_outlined..10 1637 // CHECK1-SAME: (i32* noalias noundef [[DOTGLOBAL_TID_:%.*]], i32* noalias noundef [[DOTBOUND_TID_:%.*]], i32* noundef nonnull align 4 dereferenceable(4) [[N:%.*]], double** noundef nonnull align 8 dereferenceable(8) [[A:%.*]], double** noundef nonnull align 8 dereferenceable(8) [[B:%.*]], double** noundef nonnull align 8 dereferenceable(8) [[C:%.*]]) #[[ATTR2]] { 1638 // CHECK1-NEXT: entry: 1639 // CHECK1-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 1640 // CHECK1-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 1641 // CHECK1-NEXT: [[N_ADDR:%.*]] = alloca i32*, align 8 1642 // CHECK1-NEXT: [[A_ADDR:%.*]] = alloca double**, align 8 1643 // CHECK1-NEXT: [[B_ADDR:%.*]] = alloca double**, align 8 1644 // CHECK1-NEXT: [[C_ADDR:%.*]] = alloca double**, align 8 1645 // CHECK1-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 1646 // CHECK1-NEXT: [[TMP:%.*]] = alloca i32, align 4 1647 // CHECK1-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 1648 // CHECK1-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 1649 // CHECK1-NEXT: [[I:%.*]] = alloca i32, align 4 1650 // CHECK1-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4 1651 // CHECK1-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4 1652 // CHECK1-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 1653 // CHECK1-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 1654 // CHECK1-NEXT: [[I3:%.*]] = alloca i32, align 4 1655 // CHECK1-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8 1656 // CHECK1-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8 1657 // CHECK1-NEXT: store i32* [[N]], i32** [[N_ADDR]], align 8 1658 // CHECK1-NEXT: store double** [[A]], double*** [[A_ADDR]], align 8 1659 // CHECK1-NEXT: store double** [[B]], double*** [[B_ADDR]], align 8 1660 // CHECK1-NEXT: store double** [[C]], double*** [[C_ADDR]], align 8 1661 // CHECK1-NEXT: [[TMP0:%.*]] = load i32*, i32** [[N_ADDR]], align 8 1662 // CHECK1-NEXT: [[TMP1:%.*]] = load double**, double*** [[A_ADDR]], align 8 1663 // CHECK1-NEXT: [[TMP2:%.*]] = load double**, double*** [[B_ADDR]], align 8 1664 // CHECK1-NEXT: [[TMP3:%.*]] = load double**, double*** [[C_ADDR]], align 8 1665 // CHECK1-NEXT: [[TMP4:%.*]] = load i32, i32* [[TMP0]], align 4 1666 // CHECK1-NEXT: store i32 [[TMP4]], i32* [[DOTCAPTURE_EXPR_]], align 4 1667 // CHECK1-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 1668 // CHECK1-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP5]], 0 1669 // CHECK1-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 1670 // CHECK1-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1 1671 // CHECK1-NEXT: store i32 [[SUB2]], i32* [[DOTCAPTURE_EXPR_1]], align 4 1672 // CHECK1-NEXT: store i32 0, i32* [[I]], align 4 1673 // CHECK1-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 1674 // CHECK1-NEXT: [[CMP:%.*]] = icmp slt i32 0, [[TMP6]] 1675 // CHECK1-NEXT: br i1 [[CMP]], label [[OMP_PRECOND_THEN:%.*]], label [[OMP_PRECOND_END:%.*]] 1676 // CHECK1: omp.precond.then: 1677 // CHECK1-NEXT: store i32 0, i32* [[DOTOMP_COMB_LB]], align 4 1678 // CHECK1-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 1679 // CHECK1-NEXT: store i32 [[TMP7]], i32* [[DOTOMP_COMB_UB]], align 4 1680 // CHECK1-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4 1681 // CHECK1-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4 1682 // CHECK1-NEXT: [[TMP8:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 1683 // CHECK1-NEXT: [[TMP9:%.*]] = load i32, i32* [[TMP8]], align 4 1684 // CHECK1-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1]], i32 [[TMP9]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_COMB_LB]], i32* [[DOTOMP_COMB_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1) 1685 // CHECK1-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 1686 // CHECK1-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 1687 // CHECK1-NEXT: [[CMP4:%.*]] = icmp sgt i32 [[TMP10]], [[TMP11]] 1688 // CHECK1-NEXT: br i1 [[CMP4]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] 1689 // CHECK1: cond.true: 1690 // CHECK1-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 1691 // CHECK1-NEXT: br label [[COND_END:%.*]] 1692 // CHECK1: cond.false: 1693 // CHECK1-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 1694 // CHECK1-NEXT: br label [[COND_END]] 1695 // CHECK1: cond.end: 1696 // CHECK1-NEXT: [[COND:%.*]] = phi i32 [ [[TMP12]], [[COND_TRUE]] ], [ [[TMP13]], [[COND_FALSE]] ] 1697 // CHECK1-NEXT: store i32 [[COND]], i32* [[DOTOMP_COMB_UB]], align 4 1698 // CHECK1-NEXT: [[TMP14:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4 1699 // CHECK1-NEXT: store i32 [[TMP14]], i32* [[DOTOMP_IV]], align 4 1700 // CHECK1-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] 1701 // CHECK1: omp.inner.for.cond: 1702 // CHECK1-NEXT: [[TMP15:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !31 1703 // CHECK1-NEXT: [[TMP16:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4, !llvm.access.group !31 1704 // CHECK1-NEXT: [[CMP5:%.*]] = icmp sle i32 [[TMP15]], [[TMP16]] 1705 // CHECK1-NEXT: br i1 [[CMP5]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] 1706 // CHECK1: omp.inner.for.body: 1707 // CHECK1-NEXT: [[TMP17:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4, !llvm.access.group !31 1708 // CHECK1-NEXT: [[TMP18:%.*]] = zext i32 [[TMP17]] to i64 1709 // CHECK1-NEXT: [[TMP19:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4, !llvm.access.group !31 1710 // CHECK1-NEXT: [[TMP20:%.*]] = zext i32 [[TMP19]] to i64 1711 // CHECK1-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 6, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64, i32*, double**, double**, double**)* @.omp_outlined..11 to void (i32*, i32*, ...)*), i64 [[TMP18]], i64 [[TMP20]], i32* [[TMP0]], double** [[TMP1]], double** [[TMP2]], double** [[TMP3]]), !llvm.access.group !31 1712 // CHECK1-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] 1713 // CHECK1: omp.inner.for.inc: 1714 // CHECK1-NEXT: [[TMP21:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !31 1715 // CHECK1-NEXT: [[TMP22:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4, !llvm.access.group !31 1716 // CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP21]], [[TMP22]] 1717 // CHECK1-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4, !llvm.access.group !31 1718 // CHECK1-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP32:![0-9]+]] 1719 // CHECK1: omp.inner.for.end: 1720 // CHECK1-NEXT: br label [[OMP_LOOP_EXIT:%.*]] 1721 // CHECK1: omp.loop.exit: 1722 // CHECK1-NEXT: [[TMP23:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 1723 // CHECK1-NEXT: [[TMP24:%.*]] = load i32, i32* [[TMP23]], align 4 1724 // CHECK1-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP24]]) 1725 // CHECK1-NEXT: [[TMP25:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4 1726 // CHECK1-NEXT: [[TMP26:%.*]] = icmp ne i32 [[TMP25]], 0 1727 // CHECK1-NEXT: br i1 [[TMP26]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]] 1728 // CHECK1: .omp.final.then: 1729 // CHECK1-NEXT: [[TMP27:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 1730 // CHECK1-NEXT: [[SUB6:%.*]] = sub nsw i32 [[TMP27]], 0 1731 // CHECK1-NEXT: [[DIV7:%.*]] = sdiv i32 [[SUB6]], 1 1732 // CHECK1-NEXT: [[MUL:%.*]] = mul nsw i32 [[DIV7]], 1 1733 // CHECK1-NEXT: [[ADD8:%.*]] = add nsw i32 0, [[MUL]] 1734 // CHECK1-NEXT: store i32 [[ADD8]], i32* [[I3]], align 4 1735 // CHECK1-NEXT: br label [[DOTOMP_FINAL_DONE]] 1736 // CHECK1: .omp.final.done: 1737 // CHECK1-NEXT: br label [[OMP_PRECOND_END]] 1738 // CHECK1: omp.precond.end: 1739 // CHECK1-NEXT: ret void 1740 // 1741 // 1742 // CHECK1-LABEL: define {{[^@]+}}@.omp_outlined..11 1743 // CHECK1-SAME: (i32* noalias noundef [[DOTGLOBAL_TID_:%.*]], i32* noalias noundef [[DOTBOUND_TID_:%.*]], i64 noundef [[DOTPREVIOUS_LB_:%.*]], i64 noundef [[DOTPREVIOUS_UB_:%.*]], i32* noundef nonnull align 4 dereferenceable(4) [[N:%.*]], double** noundef nonnull align 8 dereferenceable(8) [[A:%.*]], double** noundef nonnull align 8 dereferenceable(8) [[B:%.*]], double** noundef nonnull align 8 dereferenceable(8) [[C:%.*]]) #[[ATTR2]] { 1744 // CHECK1-NEXT: entry: 1745 // CHECK1-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 1746 // CHECK1-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 1747 // CHECK1-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8 1748 // CHECK1-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8 1749 // CHECK1-NEXT: [[N_ADDR:%.*]] = alloca i32*, align 8 1750 // CHECK1-NEXT: [[A_ADDR:%.*]] = alloca double**, align 8 1751 // CHECK1-NEXT: [[B_ADDR:%.*]] = alloca double**, align 8 1752 // CHECK1-NEXT: [[C_ADDR:%.*]] = alloca double**, align 8 1753 // CHECK1-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 1754 // CHECK1-NEXT: [[TMP:%.*]] = alloca i32, align 4 1755 // CHECK1-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 1756 // CHECK1-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 1757 // CHECK1-NEXT: [[I:%.*]] = alloca i32, align 4 1758 // CHECK1-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 1759 // CHECK1-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4 1760 // CHECK1-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 1761 // CHECK1-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 1762 // CHECK1-NEXT: [[I4:%.*]] = alloca i32, align 4 1763 // CHECK1-NEXT: [[REF_TMP:%.*]] = alloca [[CLASS_ANON_3:%.*]], align 8 1764 // CHECK1-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8 1765 // CHECK1-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8 1766 // CHECK1-NEXT: store i64 [[DOTPREVIOUS_LB_]], i64* [[DOTPREVIOUS_LB__ADDR]], align 8 1767 // CHECK1-NEXT: store i64 [[DOTPREVIOUS_UB_]], i64* [[DOTPREVIOUS_UB__ADDR]], align 8 1768 // CHECK1-NEXT: store i32* [[N]], i32** [[N_ADDR]], align 8 1769 // CHECK1-NEXT: store double** [[A]], double*** [[A_ADDR]], align 8 1770 // CHECK1-NEXT: store double** [[B]], double*** [[B_ADDR]], align 8 1771 // CHECK1-NEXT: store double** [[C]], double*** [[C_ADDR]], align 8 1772 // CHECK1-NEXT: [[TMP0:%.*]] = load i32*, i32** [[N_ADDR]], align 8 1773 // CHECK1-NEXT: [[TMP1:%.*]] = load double**, double*** [[A_ADDR]], align 8 1774 // CHECK1-NEXT: [[TMP2:%.*]] = load double**, double*** [[B_ADDR]], align 8 1775 // CHECK1-NEXT: [[TMP3:%.*]] = load double**, double*** [[C_ADDR]], align 8 1776 // CHECK1-NEXT: [[TMP4:%.*]] = load i32, i32* [[TMP0]], align 4 1777 // CHECK1-NEXT: store i32 [[TMP4]], i32* [[DOTCAPTURE_EXPR_]], align 4 1778 // CHECK1-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 1779 // CHECK1-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP5]], 0 1780 // CHECK1-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 1781 // CHECK1-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1 1782 // CHECK1-NEXT: store i32 [[SUB2]], i32* [[DOTCAPTURE_EXPR_1]], align 4 1783 // CHECK1-NEXT: store i32 0, i32* [[I]], align 4 1784 // CHECK1-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 1785 // CHECK1-NEXT: [[CMP:%.*]] = icmp slt i32 0, [[TMP6]] 1786 // CHECK1-NEXT: br i1 [[CMP]], label [[OMP_PRECOND_THEN:%.*]], label [[OMP_PRECOND_END:%.*]] 1787 // CHECK1: omp.precond.then: 1788 // CHECK1-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4 1789 // CHECK1-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 1790 // CHECK1-NEXT: store i32 [[TMP7]], i32* [[DOTOMP_UB]], align 4 1791 // CHECK1-NEXT: [[TMP8:%.*]] = load i64, i64* [[DOTPREVIOUS_LB__ADDR]], align 8 1792 // CHECK1-NEXT: [[CONV:%.*]] = trunc i64 [[TMP8]] to i32 1793 // CHECK1-NEXT: [[TMP9:%.*]] = load i64, i64* [[DOTPREVIOUS_UB__ADDR]], align 8 1794 // CHECK1-NEXT: [[CONV3:%.*]] = trunc i64 [[TMP9]] to i32 1795 // CHECK1-NEXT: store i32 [[CONV]], i32* [[DOTOMP_LB]], align 4 1796 // CHECK1-NEXT: store i32 [[CONV3]], i32* [[DOTOMP_UB]], align 4 1797 // CHECK1-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4 1798 // CHECK1-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4 1799 // CHECK1-NEXT: [[TMP10:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 1800 // CHECK1-NEXT: [[TMP11:%.*]] = load i32, i32* [[TMP10]], align 4 1801 // CHECK1-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB2]], i32 [[TMP11]], i32 34, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1) 1802 // CHECK1-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 1803 // CHECK1-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 1804 // CHECK1-NEXT: [[CMP5:%.*]] = icmp sgt i32 [[TMP12]], [[TMP13]] 1805 // CHECK1-NEXT: br i1 [[CMP5]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] 1806 // CHECK1: cond.true: 1807 // CHECK1-NEXT: [[TMP14:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 1808 // CHECK1-NEXT: br label [[COND_END:%.*]] 1809 // CHECK1: cond.false: 1810 // CHECK1-NEXT: [[TMP15:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 1811 // CHECK1-NEXT: br label [[COND_END]] 1812 // CHECK1: cond.end: 1813 // CHECK1-NEXT: [[COND:%.*]] = phi i32 [ [[TMP14]], [[COND_TRUE]] ], [ [[TMP15]], [[COND_FALSE]] ] 1814 // CHECK1-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4 1815 // CHECK1-NEXT: [[TMP16:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4 1816 // CHECK1-NEXT: store i32 [[TMP16]], i32* [[DOTOMP_IV]], align 4 1817 // CHECK1-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] 1818 // CHECK1: omp.inner.for.cond: 1819 // CHECK1-NEXT: [[TMP17:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !34 1820 // CHECK1-NEXT: [[TMP18:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4, !llvm.access.group !34 1821 // CHECK1-NEXT: [[CMP6:%.*]] = icmp sle i32 [[TMP17]], [[TMP18]] 1822 // CHECK1-NEXT: br i1 [[CMP6]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] 1823 // CHECK1: omp.inner.for.body: 1824 // CHECK1-NEXT: [[TMP19:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !34 1825 // CHECK1-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP19]], 1 1826 // CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] 1827 // CHECK1-NEXT: store i32 [[ADD]], i32* [[I4]], align 4, !llvm.access.group !34 1828 // CHECK1-NEXT: [[TMP20:%.*]] = load double*, double** [[TMP2]], align 8, !llvm.access.group !34 1829 // CHECK1-NEXT: [[TMP21:%.*]] = load i32, i32* [[I4]], align 4, !llvm.access.group !34 1830 // CHECK1-NEXT: [[IDXPROM:%.*]] = sext i32 [[TMP21]] to i64 1831 // CHECK1-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds double, double* [[TMP20]], i64 [[IDXPROM]] 1832 // CHECK1-NEXT: [[TMP22:%.*]] = load double, double* [[ARRAYIDX]], align 8, !llvm.access.group !34 1833 // CHECK1-NEXT: [[TMP23:%.*]] = load double*, double** [[TMP3]], align 8, !llvm.access.group !34 1834 // CHECK1-NEXT: [[TMP24:%.*]] = load i32, i32* [[I4]], align 4, !llvm.access.group !34 1835 // CHECK1-NEXT: [[IDXPROM7:%.*]] = sext i32 [[TMP24]] to i64 1836 // CHECK1-NEXT: [[ARRAYIDX8:%.*]] = getelementptr inbounds double, double* [[TMP23]], i64 [[IDXPROM7]] 1837 // CHECK1-NEXT: [[TMP25:%.*]] = load double, double* [[ARRAYIDX8]], align 8, !llvm.access.group !34 1838 // CHECK1-NEXT: [[ADD9:%.*]] = fadd double [[TMP22]], [[TMP25]] 1839 // CHECK1-NEXT: [[TMP26:%.*]] = load double*, double** [[TMP1]], align 8, !llvm.access.group !34 1840 // CHECK1-NEXT: [[TMP27:%.*]] = load i32, i32* [[I4]], align 4, !llvm.access.group !34 1841 // CHECK1-NEXT: [[IDXPROM10:%.*]] = sext i32 [[TMP27]] to i64 1842 // CHECK1-NEXT: [[ARRAYIDX11:%.*]] = getelementptr inbounds double, double* [[TMP26]], i64 [[IDXPROM10]] 1843 // CHECK1-NEXT: store double [[ADD9]], double* [[ARRAYIDX11]], align 8, !llvm.access.group !34 1844 // CHECK1-NEXT: [[TMP28:%.*]] = getelementptr inbounds [[CLASS_ANON_3]], %class.anon.3* [[REF_TMP]], i32 0, i32 0 1845 // CHECK1-NEXT: store double** [[TMP1]], double*** [[TMP28]], align 8, !llvm.access.group !34 1846 // CHECK1-NEXT: [[TMP29:%.*]] = getelementptr inbounds [[CLASS_ANON_3]], %class.anon.3* [[REF_TMP]], i32 0, i32 1 1847 // CHECK1-NEXT: store i32* [[I4]], i32** [[TMP29]], align 8, !llvm.access.group !34 1848 // CHECK1-NEXT: [[TMP30:%.*]] = getelementptr inbounds [[CLASS_ANON_3]], %class.anon.3* [[REF_TMP]], i32 0, i32 2 1849 // CHECK1-NEXT: store double** [[TMP2]], double*** [[TMP30]], align 8, !llvm.access.group !34 1850 // CHECK1-NEXT: [[TMP31:%.*]] = getelementptr inbounds [[CLASS_ANON_3]], %class.anon.3* [[REF_TMP]], i32 0, i32 3 1851 // CHECK1-NEXT: store double** [[TMP3]], double*** [[TMP31]], align 8, !llvm.access.group !34 1852 // CHECK1-NEXT: call void @"_ZZZ4mainENK3$_0clEvENKUlvE2_clEv"(%class.anon.3* noundef nonnull align 8 dereferenceable(32) [[REF_TMP]]), !llvm.access.group !34 1853 // CHECK1-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] 1854 // CHECK1: omp.body.continue: 1855 // CHECK1-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] 1856 // CHECK1: omp.inner.for.inc: 1857 // CHECK1-NEXT: [[TMP32:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !34 1858 // CHECK1-NEXT: [[ADD12:%.*]] = add nsw i32 [[TMP32]], 1 1859 // CHECK1-NEXT: store i32 [[ADD12]], i32* [[DOTOMP_IV]], align 4, !llvm.access.group !34 1860 // CHECK1-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP35:![0-9]+]] 1861 // CHECK1: omp.inner.for.end: 1862 // CHECK1-NEXT: br label [[OMP_LOOP_EXIT:%.*]] 1863 // CHECK1: omp.loop.exit: 1864 // CHECK1-NEXT: [[TMP33:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 1865 // CHECK1-NEXT: [[TMP34:%.*]] = load i32, i32* [[TMP33]], align 4 1866 // CHECK1-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP34]]) 1867 // CHECK1-NEXT: [[TMP35:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4 1868 // CHECK1-NEXT: [[TMP36:%.*]] = icmp ne i32 [[TMP35]], 0 1869 // CHECK1-NEXT: br i1 [[TMP36]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]] 1870 // CHECK1: .omp.final.then: 1871 // CHECK1-NEXT: [[TMP37:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 1872 // CHECK1-NEXT: [[SUB13:%.*]] = sub nsw i32 [[TMP37]], 0 1873 // CHECK1-NEXT: [[DIV14:%.*]] = sdiv i32 [[SUB13]], 1 1874 // CHECK1-NEXT: [[MUL15:%.*]] = mul nsw i32 [[DIV14]], 1 1875 // CHECK1-NEXT: [[ADD16:%.*]] = add nsw i32 0, [[MUL15]] 1876 // CHECK1-NEXT: store i32 [[ADD16]], i32* [[I4]], align 4 1877 // CHECK1-NEXT: br label [[DOTOMP_FINAL_DONE]] 1878 // CHECK1: .omp.final.done: 1879 // CHECK1-NEXT: br label [[OMP_PRECOND_END]] 1880 // CHECK1: omp.precond.end: 1881 // CHECK1-NEXT: ret void 1882 // 1883 // 1884 // CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l266 1885 // CHECK1-SAME: (i64 noundef [[CH:%.*]], i64 noundef [[N:%.*]], double* noundef [[A:%.*]], double* noundef [[B:%.*]], double* noundef [[C:%.*]]) #[[ATTR2]] { 1886 // CHECK1-NEXT: entry: 1887 // CHECK1-NEXT: [[CH_ADDR:%.*]] = alloca i64, align 8 1888 // CHECK1-NEXT: [[N_ADDR:%.*]] = alloca i64, align 8 1889 // CHECK1-NEXT: [[A_ADDR:%.*]] = alloca double*, align 8 1890 // CHECK1-NEXT: [[B_ADDR:%.*]] = alloca double*, align 8 1891 // CHECK1-NEXT: [[C_ADDR:%.*]] = alloca double*, align 8 1892 // CHECK1-NEXT: store i64 [[CH]], i64* [[CH_ADDR]], align 8 1893 // CHECK1-NEXT: store i64 [[N]], i64* [[N_ADDR]], align 8 1894 // CHECK1-NEXT: store double* [[A]], double** [[A_ADDR]], align 8 1895 // CHECK1-NEXT: store double* [[B]], double** [[B_ADDR]], align 8 1896 // CHECK1-NEXT: store double* [[C]], double** [[C_ADDR]], align 8 1897 // CHECK1-NEXT: [[CONV:%.*]] = bitcast i64* [[CH_ADDR]] to i32* 1898 // CHECK1-NEXT: [[CONV1:%.*]] = bitcast i64* [[N_ADDR]] to i32* 1899 // CHECK1-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 5, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32*, i32*, double**, double**, double**)* @.omp_outlined..14 to void (i32*, i32*, ...)*), i32* [[CONV]], i32* [[CONV1]], double** [[A_ADDR]], double** [[B_ADDR]], double** [[C_ADDR]]) 1900 // CHECK1-NEXT: ret void 1901 // 1902 // 1903 // CHECK1-LABEL: define {{[^@]+}}@.omp_outlined..14 1904 // CHECK1-SAME: (i32* noalias noundef [[DOTGLOBAL_TID_:%.*]], i32* noalias noundef [[DOTBOUND_TID_:%.*]], i32* noundef nonnull align 4 dereferenceable(4) [[CH:%.*]], i32* noundef nonnull align 4 dereferenceable(4) [[N:%.*]], double** noundef nonnull align 8 dereferenceable(8) [[A:%.*]], double** noundef nonnull align 8 dereferenceable(8) [[B:%.*]], double** noundef nonnull align 8 dereferenceable(8) [[C:%.*]]) #[[ATTR2]] { 1905 // CHECK1-NEXT: entry: 1906 // CHECK1-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 1907 // CHECK1-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 1908 // CHECK1-NEXT: [[CH_ADDR:%.*]] = alloca i32*, align 8 1909 // CHECK1-NEXT: [[N_ADDR:%.*]] = alloca i32*, align 8 1910 // CHECK1-NEXT: [[A_ADDR:%.*]] = alloca double**, align 8 1911 // CHECK1-NEXT: [[B_ADDR:%.*]] = alloca double**, align 8 1912 // CHECK1-NEXT: [[C_ADDR:%.*]] = alloca double**, align 8 1913 // CHECK1-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 1914 // CHECK1-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 1915 // CHECK1-NEXT: [[TMP:%.*]] = alloca i32, align 4 1916 // CHECK1-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 1917 // CHECK1-NEXT: [[DOTCAPTURE_EXPR_2:%.*]] = alloca i32, align 4 1918 // CHECK1-NEXT: [[I:%.*]] = alloca i32, align 4 1919 // CHECK1-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4 1920 // CHECK1-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4 1921 // CHECK1-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 1922 // CHECK1-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 1923 // CHECK1-NEXT: [[I4:%.*]] = alloca i32, align 4 1924 // CHECK1-NEXT: [[DOTCAPTURE_EXPR__CASTED:%.*]] = alloca i64, align 8 1925 // CHECK1-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8 1926 // CHECK1-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8 1927 // CHECK1-NEXT: store i32* [[CH]], i32** [[CH_ADDR]], align 8 1928 // CHECK1-NEXT: store i32* [[N]], i32** [[N_ADDR]], align 8 1929 // CHECK1-NEXT: store double** [[A]], double*** [[A_ADDR]], align 8 1930 // CHECK1-NEXT: store double** [[B]], double*** [[B_ADDR]], align 8 1931 // CHECK1-NEXT: store double** [[C]], double*** [[C_ADDR]], align 8 1932 // CHECK1-NEXT: [[TMP0:%.*]] = load i32*, i32** [[CH_ADDR]], align 8 1933 // CHECK1-NEXT: [[TMP1:%.*]] = load i32*, i32** [[N_ADDR]], align 8 1934 // CHECK1-NEXT: [[TMP2:%.*]] = load double**, double*** [[A_ADDR]], align 8 1935 // CHECK1-NEXT: [[TMP3:%.*]] = load double**, double*** [[B_ADDR]], align 8 1936 // CHECK1-NEXT: [[TMP4:%.*]] = load double**, double*** [[C_ADDR]], align 8 1937 // CHECK1-NEXT: [[TMP5:%.*]] = load i32, i32* [[TMP0]], align 4 1938 // CHECK1-NEXT: store i32 [[TMP5]], i32* [[DOTCAPTURE_EXPR_]], align 4 1939 // CHECK1-NEXT: [[TMP6:%.*]] = load i32, i32* [[TMP1]], align 4 1940 // CHECK1-NEXT: store i32 [[TMP6]], i32* [[DOTCAPTURE_EXPR_1]], align 4 1941 // CHECK1-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 1942 // CHECK1-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP7]], 0 1943 // CHECK1-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 1944 // CHECK1-NEXT: [[SUB3:%.*]] = sub nsw i32 [[DIV]], 1 1945 // CHECK1-NEXT: store i32 [[SUB3]], i32* [[DOTCAPTURE_EXPR_2]], align 4 1946 // CHECK1-NEXT: store i32 0, i32* [[I]], align 4 1947 // CHECK1-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 1948 // CHECK1-NEXT: [[CMP:%.*]] = icmp slt i32 0, [[TMP8]] 1949 // CHECK1-NEXT: br i1 [[CMP]], label [[OMP_PRECOND_THEN:%.*]], label [[OMP_PRECOND_END:%.*]] 1950 // CHECK1: omp.precond.then: 1951 // CHECK1-NEXT: store i32 0, i32* [[DOTOMP_COMB_LB]], align 4 1952 // CHECK1-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_2]], align 4 1953 // CHECK1-NEXT: store i32 [[TMP9]], i32* [[DOTOMP_COMB_UB]], align 4 1954 // CHECK1-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4 1955 // CHECK1-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4 1956 // CHECK1-NEXT: [[TMP10:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 1957 // CHECK1-NEXT: [[TMP11:%.*]] = load i32, i32* [[TMP10]], align 4 1958 // CHECK1-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1]], i32 [[TMP11]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_COMB_LB]], i32* [[DOTOMP_COMB_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1) 1959 // CHECK1-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 1960 // CHECK1-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_2]], align 4 1961 // CHECK1-NEXT: [[CMP5:%.*]] = icmp sgt i32 [[TMP12]], [[TMP13]] 1962 // CHECK1-NEXT: br i1 [[CMP5]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] 1963 // CHECK1: cond.true: 1964 // CHECK1-NEXT: [[TMP14:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_2]], align 4 1965 // CHECK1-NEXT: br label [[COND_END:%.*]] 1966 // CHECK1: cond.false: 1967 // CHECK1-NEXT: [[TMP15:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 1968 // CHECK1-NEXT: br label [[COND_END]] 1969 // CHECK1: cond.end: 1970 // CHECK1-NEXT: [[COND:%.*]] = phi i32 [ [[TMP14]], [[COND_TRUE]] ], [ [[TMP15]], [[COND_FALSE]] ] 1971 // CHECK1-NEXT: store i32 [[COND]], i32* [[DOTOMP_COMB_UB]], align 4 1972 // CHECK1-NEXT: [[TMP16:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4 1973 // CHECK1-NEXT: store i32 [[TMP16]], i32* [[DOTOMP_IV]], align 4 1974 // CHECK1-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] 1975 // CHECK1: omp.inner.for.cond: 1976 // CHECK1-NEXT: [[TMP17:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !37 1977 // CHECK1-NEXT: [[TMP18:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4, !llvm.access.group !37 1978 // CHECK1-NEXT: [[CMP6:%.*]] = icmp sle i32 [[TMP17]], [[TMP18]] 1979 // CHECK1-NEXT: br i1 [[CMP6]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] 1980 // CHECK1: omp.inner.for.body: 1981 // CHECK1-NEXT: [[TMP19:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4, !llvm.access.group !37 1982 // CHECK1-NEXT: [[TMP20:%.*]] = zext i32 [[TMP19]] to i64 1983 // CHECK1-NEXT: [[TMP21:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4, !llvm.access.group !37 1984 // CHECK1-NEXT: [[TMP22:%.*]] = zext i32 [[TMP21]] to i64 1985 // CHECK1-NEXT: [[TMP23:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4, !llvm.access.group !37 1986 // CHECK1-NEXT: [[CONV:%.*]] = bitcast i64* [[DOTCAPTURE_EXPR__CASTED]] to i32* 1987 // CHECK1-NEXT: store i32 [[TMP23]], i32* [[CONV]], align 4, !llvm.access.group !37 1988 // CHECK1-NEXT: [[TMP24:%.*]] = load i64, i64* [[DOTCAPTURE_EXPR__CASTED]], align 8, !llvm.access.group !37 1989 // CHECK1-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 7, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64, i32*, double**, double**, double**, i64)* @.omp_outlined..15 to void (i32*, i32*, ...)*), i64 [[TMP20]], i64 [[TMP22]], i32* [[TMP1]], double** [[TMP2]], double** [[TMP3]], double** [[TMP4]], i64 [[TMP24]]), !llvm.access.group !37 1990 // CHECK1-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] 1991 // CHECK1: omp.inner.for.inc: 1992 // CHECK1-NEXT: [[TMP25:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !37 1993 // CHECK1-NEXT: [[TMP26:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4, !llvm.access.group !37 1994 // CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP25]], [[TMP26]] 1995 // CHECK1-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4, !llvm.access.group !37 1996 // CHECK1-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP38:![0-9]+]] 1997 // CHECK1: omp.inner.for.end: 1998 // CHECK1-NEXT: br label [[OMP_LOOP_EXIT:%.*]] 1999 // CHECK1: omp.loop.exit: 2000 // CHECK1-NEXT: [[TMP27:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 2001 // CHECK1-NEXT: [[TMP28:%.*]] = load i32, i32* [[TMP27]], align 4 2002 // CHECK1-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP28]]) 2003 // CHECK1-NEXT: [[TMP29:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4 2004 // CHECK1-NEXT: [[TMP30:%.*]] = icmp ne i32 [[TMP29]], 0 2005 // CHECK1-NEXT: br i1 [[TMP30]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]] 2006 // CHECK1: .omp.final.then: 2007 // CHECK1-NEXT: [[TMP31:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 2008 // CHECK1-NEXT: [[SUB7:%.*]] = sub nsw i32 [[TMP31]], 0 2009 // CHECK1-NEXT: [[DIV8:%.*]] = sdiv i32 [[SUB7]], 1 2010 // CHECK1-NEXT: [[MUL:%.*]] = mul nsw i32 [[DIV8]], 1 2011 // CHECK1-NEXT: [[ADD9:%.*]] = add nsw i32 0, [[MUL]] 2012 // CHECK1-NEXT: store i32 [[ADD9]], i32* [[I4]], align 4 2013 // CHECK1-NEXT: br label [[DOTOMP_FINAL_DONE]] 2014 // CHECK1: .omp.final.done: 2015 // CHECK1-NEXT: br label [[OMP_PRECOND_END]] 2016 // CHECK1: omp.precond.end: 2017 // CHECK1-NEXT: ret void 2018 // 2019 // 2020 // CHECK1-LABEL: define {{[^@]+}}@.omp_outlined..15 2021 // CHECK1-SAME: (i32* noalias noundef [[DOTGLOBAL_TID_:%.*]], i32* noalias noundef [[DOTBOUND_TID_:%.*]], i64 noundef [[DOTPREVIOUS_LB_:%.*]], i64 noundef [[DOTPREVIOUS_UB_:%.*]], i32* noundef nonnull align 4 dereferenceable(4) [[N:%.*]], double** noundef nonnull align 8 dereferenceable(8) [[A:%.*]], double** noundef nonnull align 8 dereferenceable(8) [[B:%.*]], double** noundef nonnull align 8 dereferenceable(8) [[C:%.*]], i64 noundef [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR2]] { 2022 // CHECK1-NEXT: entry: 2023 // CHECK1-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 2024 // CHECK1-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 2025 // CHECK1-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8 2026 // CHECK1-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8 2027 // CHECK1-NEXT: [[N_ADDR:%.*]] = alloca i32*, align 8 2028 // CHECK1-NEXT: [[A_ADDR:%.*]] = alloca double**, align 8 2029 // CHECK1-NEXT: [[B_ADDR:%.*]] = alloca double**, align 8 2030 // CHECK1-NEXT: [[C_ADDR:%.*]] = alloca double**, align 8 2031 // CHECK1-NEXT: [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i64, align 8 2032 // CHECK1-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 2033 // CHECK1-NEXT: [[TMP:%.*]] = alloca i32, align 4 2034 // CHECK1-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 2035 // CHECK1-NEXT: [[DOTCAPTURE_EXPR_2:%.*]] = alloca i32, align 4 2036 // CHECK1-NEXT: [[I:%.*]] = alloca i32, align 4 2037 // CHECK1-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 2038 // CHECK1-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4 2039 // CHECK1-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 2040 // CHECK1-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 2041 // CHECK1-NEXT: [[I6:%.*]] = alloca i32, align 4 2042 // CHECK1-NEXT: [[REF_TMP:%.*]] = alloca [[CLASS_ANON_4:%.*]], align 8 2043 // CHECK1-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8 2044 // CHECK1-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8 2045 // CHECK1-NEXT: store i64 [[DOTPREVIOUS_LB_]], i64* [[DOTPREVIOUS_LB__ADDR]], align 8 2046 // CHECK1-NEXT: store i64 [[DOTPREVIOUS_UB_]], i64* [[DOTPREVIOUS_UB__ADDR]], align 8 2047 // CHECK1-NEXT: store i32* [[N]], i32** [[N_ADDR]], align 8 2048 // CHECK1-NEXT: store double** [[A]], double*** [[A_ADDR]], align 8 2049 // CHECK1-NEXT: store double** [[B]], double*** [[B_ADDR]], align 8 2050 // CHECK1-NEXT: store double** [[C]], double*** [[C_ADDR]], align 8 2051 // CHECK1-NEXT: store i64 [[DOTCAPTURE_EXPR_]], i64* [[DOTCAPTURE_EXPR__ADDR]], align 8 2052 // CHECK1-NEXT: [[TMP0:%.*]] = load i32*, i32** [[N_ADDR]], align 8 2053 // CHECK1-NEXT: [[TMP1:%.*]] = load double**, double*** [[A_ADDR]], align 8 2054 // CHECK1-NEXT: [[TMP2:%.*]] = load double**, double*** [[B_ADDR]], align 8 2055 // CHECK1-NEXT: [[TMP3:%.*]] = load double**, double*** [[C_ADDR]], align 8 2056 // CHECK1-NEXT: [[CONV:%.*]] = bitcast i64* [[DOTCAPTURE_EXPR__ADDR]] to i32* 2057 // CHECK1-NEXT: [[TMP4:%.*]] = load i32, i32* [[TMP0]], align 4 2058 // CHECK1-NEXT: store i32 [[TMP4]], i32* [[DOTCAPTURE_EXPR_1]], align 4 2059 // CHECK1-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 2060 // CHECK1-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP5]], 0 2061 // CHECK1-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 2062 // CHECK1-NEXT: [[SUB3:%.*]] = sub nsw i32 [[DIV]], 1 2063 // CHECK1-NEXT: store i32 [[SUB3]], i32* [[DOTCAPTURE_EXPR_2]], align 4 2064 // CHECK1-NEXT: store i32 0, i32* [[I]], align 4 2065 // CHECK1-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 2066 // CHECK1-NEXT: [[CMP:%.*]] = icmp slt i32 0, [[TMP6]] 2067 // CHECK1-NEXT: br i1 [[CMP]], label [[OMP_PRECOND_THEN:%.*]], label [[OMP_PRECOND_END:%.*]] 2068 // CHECK1: omp.precond.then: 2069 // CHECK1-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4 2070 // CHECK1-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_2]], align 4 2071 // CHECK1-NEXT: store i32 [[TMP7]], i32* [[DOTOMP_UB]], align 4 2072 // CHECK1-NEXT: [[TMP8:%.*]] = load i64, i64* [[DOTPREVIOUS_LB__ADDR]], align 8 2073 // CHECK1-NEXT: [[CONV4:%.*]] = trunc i64 [[TMP8]] to i32 2074 // CHECK1-NEXT: [[TMP9:%.*]] = load i64, i64* [[DOTPREVIOUS_UB__ADDR]], align 8 2075 // CHECK1-NEXT: [[CONV5:%.*]] = trunc i64 [[TMP9]] to i32 2076 // CHECK1-NEXT: store i32 [[CONV4]], i32* [[DOTOMP_LB]], align 4 2077 // CHECK1-NEXT: store i32 [[CONV5]], i32* [[DOTOMP_UB]], align 4 2078 // CHECK1-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4 2079 // CHECK1-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4 2080 // CHECK1-NEXT: [[TMP10:%.*]] = load i32, i32* [[CONV]], align 4 2081 // CHECK1-NEXT: [[TMP11:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 2082 // CHECK1-NEXT: [[TMP12:%.*]] = load i32, i32* [[TMP11]], align 4 2083 // CHECK1-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB2]], i32 [[TMP12]], i32 33, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 [[TMP10]]) 2084 // CHECK1-NEXT: br label [[OMP_DISPATCH_COND:%.*]] 2085 // CHECK1: omp.dispatch.cond: 2086 // CHECK1-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 2087 // CHECK1-NEXT: [[TMP14:%.*]] = load i64, i64* [[DOTPREVIOUS_UB__ADDR]], align 8 2088 // CHECK1-NEXT: [[CONV7:%.*]] = trunc i64 [[TMP14]] to i32 2089 // CHECK1-NEXT: [[CMP8:%.*]] = icmp sgt i32 [[TMP13]], [[CONV7]] 2090 // CHECK1-NEXT: br i1 [[CMP8]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] 2091 // CHECK1: cond.true: 2092 // CHECK1-NEXT: [[TMP15:%.*]] = load i64, i64* [[DOTPREVIOUS_UB__ADDR]], align 8 2093 // CHECK1-NEXT: [[CONV9:%.*]] = trunc i64 [[TMP15]] to i32 2094 // CHECK1-NEXT: br label [[COND_END:%.*]] 2095 // CHECK1: cond.false: 2096 // CHECK1-NEXT: [[TMP16:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 2097 // CHECK1-NEXT: br label [[COND_END]] 2098 // CHECK1: cond.end: 2099 // CHECK1-NEXT: [[COND:%.*]] = phi i32 [ [[CONV9]], [[COND_TRUE]] ], [ [[TMP16]], [[COND_FALSE]] ] 2100 // CHECK1-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4 2101 // CHECK1-NEXT: [[TMP17:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4 2102 // CHECK1-NEXT: store i32 [[TMP17]], i32* [[DOTOMP_IV]], align 4 2103 // CHECK1-NEXT: [[TMP18:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 2104 // CHECK1-NEXT: [[TMP19:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 2105 // CHECK1-NEXT: [[CMP10:%.*]] = icmp sle i32 [[TMP18]], [[TMP19]] 2106 // CHECK1-NEXT: br i1 [[CMP10]], label [[OMP_DISPATCH_BODY:%.*]], label [[OMP_DISPATCH_END:%.*]] 2107 // CHECK1: omp.dispatch.body: 2108 // CHECK1-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] 2109 // CHECK1: omp.inner.for.cond: 2110 // CHECK1-NEXT: [[TMP20:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !40 2111 // CHECK1-NEXT: [[TMP21:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4, !llvm.access.group !40 2112 // CHECK1-NEXT: [[CMP11:%.*]] = icmp sle i32 [[TMP20]], [[TMP21]] 2113 // CHECK1-NEXT: br i1 [[CMP11]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] 2114 // CHECK1: omp.inner.for.body: 2115 // CHECK1-NEXT: [[TMP22:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !40 2116 // CHECK1-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP22]], 1 2117 // CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] 2118 // CHECK1-NEXT: store i32 [[ADD]], i32* [[I6]], align 4, !llvm.access.group !40 2119 // CHECK1-NEXT: [[TMP23:%.*]] = load double*, double** [[TMP2]], align 8, !llvm.access.group !40 2120 // CHECK1-NEXT: [[TMP24:%.*]] = load i32, i32* [[I6]], align 4, !llvm.access.group !40 2121 // CHECK1-NEXT: [[IDXPROM:%.*]] = sext i32 [[TMP24]] to i64 2122 // CHECK1-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds double, double* [[TMP23]], i64 [[IDXPROM]] 2123 // CHECK1-NEXT: [[TMP25:%.*]] = load double, double* [[ARRAYIDX]], align 8, !llvm.access.group !40 2124 // CHECK1-NEXT: [[TMP26:%.*]] = load double*, double** [[TMP3]], align 8, !llvm.access.group !40 2125 // CHECK1-NEXT: [[TMP27:%.*]] = load i32, i32* [[I6]], align 4, !llvm.access.group !40 2126 // CHECK1-NEXT: [[IDXPROM12:%.*]] = sext i32 [[TMP27]] to i64 2127 // CHECK1-NEXT: [[ARRAYIDX13:%.*]] = getelementptr inbounds double, double* [[TMP26]], i64 [[IDXPROM12]] 2128 // CHECK1-NEXT: [[TMP28:%.*]] = load double, double* [[ARRAYIDX13]], align 8, !llvm.access.group !40 2129 // CHECK1-NEXT: [[ADD14:%.*]] = fadd double [[TMP25]], [[TMP28]] 2130 // CHECK1-NEXT: [[TMP29:%.*]] = load double*, double** [[TMP1]], align 8, !llvm.access.group !40 2131 // CHECK1-NEXT: [[TMP30:%.*]] = load i32, i32* [[I6]], align 4, !llvm.access.group !40 2132 // CHECK1-NEXT: [[IDXPROM15:%.*]] = sext i32 [[TMP30]] to i64 2133 // CHECK1-NEXT: [[ARRAYIDX16:%.*]] = getelementptr inbounds double, double* [[TMP29]], i64 [[IDXPROM15]] 2134 // CHECK1-NEXT: store double [[ADD14]], double* [[ARRAYIDX16]], align 8, !llvm.access.group !40 2135 // CHECK1-NEXT: [[TMP31:%.*]] = getelementptr inbounds [[CLASS_ANON_4]], %class.anon.4* [[REF_TMP]], i32 0, i32 0 2136 // CHECK1-NEXT: store double** [[TMP1]], double*** [[TMP31]], align 8, !llvm.access.group !40 2137 // CHECK1-NEXT: [[TMP32:%.*]] = getelementptr inbounds [[CLASS_ANON_4]], %class.anon.4* [[REF_TMP]], i32 0, i32 1 2138 // CHECK1-NEXT: store i32* [[I6]], i32** [[TMP32]], align 8, !llvm.access.group !40 2139 // CHECK1-NEXT: [[TMP33:%.*]] = getelementptr inbounds [[CLASS_ANON_4]], %class.anon.4* [[REF_TMP]], i32 0, i32 2 2140 // CHECK1-NEXT: store double** [[TMP2]], double*** [[TMP33]], align 8, !llvm.access.group !40 2141 // CHECK1-NEXT: [[TMP34:%.*]] = getelementptr inbounds [[CLASS_ANON_4]], %class.anon.4* [[REF_TMP]], i32 0, i32 3 2142 // CHECK1-NEXT: store double** [[TMP3]], double*** [[TMP34]], align 8, !llvm.access.group !40 2143 // CHECK1-NEXT: call void @"_ZZZ4mainENK3$_0clEvENKUlvE3_clEv"(%class.anon.4* noundef nonnull align 8 dereferenceable(32) [[REF_TMP]]), !llvm.access.group !40 2144 // CHECK1-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] 2145 // CHECK1: omp.body.continue: 2146 // CHECK1-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] 2147 // CHECK1: omp.inner.for.inc: 2148 // CHECK1-NEXT: [[TMP35:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !40 2149 // CHECK1-NEXT: [[ADD17:%.*]] = add nsw i32 [[TMP35]], 1 2150 // CHECK1-NEXT: store i32 [[ADD17]], i32* [[DOTOMP_IV]], align 4, !llvm.access.group !40 2151 // CHECK1-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP41:![0-9]+]] 2152 // CHECK1: omp.inner.for.end: 2153 // CHECK1-NEXT: br label [[OMP_DISPATCH_INC:%.*]] 2154 // CHECK1: omp.dispatch.inc: 2155 // CHECK1-NEXT: [[TMP36:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4 2156 // CHECK1-NEXT: [[TMP37:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4 2157 // CHECK1-NEXT: [[ADD18:%.*]] = add nsw i32 [[TMP36]], [[TMP37]] 2158 // CHECK1-NEXT: store i32 [[ADD18]], i32* [[DOTOMP_LB]], align 4 2159 // CHECK1-NEXT: [[TMP38:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 2160 // CHECK1-NEXT: [[TMP39:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4 2161 // CHECK1-NEXT: [[ADD19:%.*]] = add nsw i32 [[TMP38]], [[TMP39]] 2162 // CHECK1-NEXT: store i32 [[ADD19]], i32* [[DOTOMP_UB]], align 4 2163 // CHECK1-NEXT: br label [[OMP_DISPATCH_COND]] 2164 // CHECK1: omp.dispatch.end: 2165 // CHECK1-NEXT: [[TMP40:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 2166 // CHECK1-NEXT: [[TMP41:%.*]] = load i32, i32* [[TMP40]], align 4 2167 // CHECK1-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP41]]) 2168 // CHECK1-NEXT: [[TMP42:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4 2169 // CHECK1-NEXT: [[TMP43:%.*]] = icmp ne i32 [[TMP42]], 0 2170 // CHECK1-NEXT: br i1 [[TMP43]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]] 2171 // CHECK1: .omp.final.then: 2172 // CHECK1-NEXT: [[TMP44:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 2173 // CHECK1-NEXT: [[SUB20:%.*]] = sub nsw i32 [[TMP44]], 0 2174 // CHECK1-NEXT: [[DIV21:%.*]] = sdiv i32 [[SUB20]], 1 2175 // CHECK1-NEXT: [[MUL22:%.*]] = mul nsw i32 [[DIV21]], 1 2176 // CHECK1-NEXT: [[ADD23:%.*]] = add nsw i32 0, [[MUL22]] 2177 // CHECK1-NEXT: store i32 [[ADD23]], i32* [[I6]], align 4 2178 // CHECK1-NEXT: br label [[DOTOMP_FINAL_DONE]] 2179 // CHECK1: .omp.final.done: 2180 // CHECK1-NEXT: br label [[OMP_PRECOND_END]] 2181 // CHECK1: omp.precond.end: 2182 // CHECK1-NEXT: ret void 2183 // 2184 // 2185 // CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l299 2186 // CHECK1-SAME: (i64 noundef [[N:%.*]], double* noundef [[A:%.*]], double* noundef [[B:%.*]], double* noundef [[C:%.*]]) #[[ATTR2]] { 2187 // CHECK1-NEXT: entry: 2188 // CHECK1-NEXT: [[N_ADDR:%.*]] = alloca i64, align 8 2189 // CHECK1-NEXT: [[A_ADDR:%.*]] = alloca double*, align 8 2190 // CHECK1-NEXT: [[B_ADDR:%.*]] = alloca double*, align 8 2191 // CHECK1-NEXT: [[C_ADDR:%.*]] = alloca double*, align 8 2192 // CHECK1-NEXT: store i64 [[N]], i64* [[N_ADDR]], align 8 2193 // CHECK1-NEXT: store double* [[A]], double** [[A_ADDR]], align 8 2194 // CHECK1-NEXT: store double* [[B]], double** [[B_ADDR]], align 8 2195 // CHECK1-NEXT: store double* [[C]], double** [[C_ADDR]], align 8 2196 // CHECK1-NEXT: [[CONV:%.*]] = bitcast i64* [[N_ADDR]] to i32* 2197 // CHECK1-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 4, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32*, double**, double**, double**)* @.omp_outlined..18 to void (i32*, i32*, ...)*), i32* [[CONV]], double** [[A_ADDR]], double** [[B_ADDR]], double** [[C_ADDR]]) 2198 // CHECK1-NEXT: ret void 2199 // 2200 // 2201 // CHECK1-LABEL: define {{[^@]+}}@.omp_outlined..18 2202 // CHECK1-SAME: (i32* noalias noundef [[DOTGLOBAL_TID_:%.*]], i32* noalias noundef [[DOTBOUND_TID_:%.*]], i32* noundef nonnull align 4 dereferenceable(4) [[N:%.*]], double** noundef nonnull align 8 dereferenceable(8) [[A:%.*]], double** noundef nonnull align 8 dereferenceable(8) [[B:%.*]], double** noundef nonnull align 8 dereferenceable(8) [[C:%.*]]) #[[ATTR2]] { 2203 // CHECK1-NEXT: entry: 2204 // CHECK1-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 2205 // CHECK1-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 2206 // CHECK1-NEXT: [[N_ADDR:%.*]] = alloca i32*, align 8 2207 // CHECK1-NEXT: [[A_ADDR:%.*]] = alloca double**, align 8 2208 // CHECK1-NEXT: [[B_ADDR:%.*]] = alloca double**, align 8 2209 // CHECK1-NEXT: [[C_ADDR:%.*]] = alloca double**, align 8 2210 // CHECK1-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 2211 // CHECK1-NEXT: [[TMP:%.*]] = alloca i32, align 4 2212 // CHECK1-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 2213 // CHECK1-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 2214 // CHECK1-NEXT: [[I:%.*]] = alloca i32, align 4 2215 // CHECK1-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4 2216 // CHECK1-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4 2217 // CHECK1-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 2218 // CHECK1-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 2219 // CHECK1-NEXT: [[I3:%.*]] = alloca i32, align 4 2220 // CHECK1-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8 2221 // CHECK1-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8 2222 // CHECK1-NEXT: store i32* [[N]], i32** [[N_ADDR]], align 8 2223 // CHECK1-NEXT: store double** [[A]], double*** [[A_ADDR]], align 8 2224 // CHECK1-NEXT: store double** [[B]], double*** [[B_ADDR]], align 8 2225 // CHECK1-NEXT: store double** [[C]], double*** [[C_ADDR]], align 8 2226 // CHECK1-NEXT: [[TMP0:%.*]] = load i32*, i32** [[N_ADDR]], align 8 2227 // CHECK1-NEXT: [[TMP1:%.*]] = load double**, double*** [[A_ADDR]], align 8 2228 // CHECK1-NEXT: [[TMP2:%.*]] = load double**, double*** [[B_ADDR]], align 8 2229 // CHECK1-NEXT: [[TMP3:%.*]] = load double**, double*** [[C_ADDR]], align 8 2230 // CHECK1-NEXT: [[TMP4:%.*]] = load i32, i32* [[TMP0]], align 4 2231 // CHECK1-NEXT: store i32 [[TMP4]], i32* [[DOTCAPTURE_EXPR_]], align 4 2232 // CHECK1-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 2233 // CHECK1-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP5]], 0 2234 // CHECK1-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 2235 // CHECK1-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1 2236 // CHECK1-NEXT: store i32 [[SUB2]], i32* [[DOTCAPTURE_EXPR_1]], align 4 2237 // CHECK1-NEXT: store i32 0, i32* [[I]], align 4 2238 // CHECK1-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 2239 // CHECK1-NEXT: [[CMP:%.*]] = icmp slt i32 0, [[TMP6]] 2240 // CHECK1-NEXT: br i1 [[CMP]], label [[OMP_PRECOND_THEN:%.*]], label [[OMP_PRECOND_END:%.*]] 2241 // CHECK1: omp.precond.then: 2242 // CHECK1-NEXT: store i32 0, i32* [[DOTOMP_COMB_LB]], align 4 2243 // CHECK1-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 2244 // CHECK1-NEXT: store i32 [[TMP7]], i32* [[DOTOMP_COMB_UB]], align 4 2245 // CHECK1-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4 2246 // CHECK1-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4 2247 // CHECK1-NEXT: [[TMP8:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 2248 // CHECK1-NEXT: [[TMP9:%.*]] = load i32, i32* [[TMP8]], align 4 2249 // CHECK1-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1]], i32 [[TMP9]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_COMB_LB]], i32* [[DOTOMP_COMB_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1) 2250 // CHECK1-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 2251 // CHECK1-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 2252 // CHECK1-NEXT: [[CMP4:%.*]] = icmp sgt i32 [[TMP10]], [[TMP11]] 2253 // CHECK1-NEXT: br i1 [[CMP4]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] 2254 // CHECK1: cond.true: 2255 // CHECK1-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 2256 // CHECK1-NEXT: br label [[COND_END:%.*]] 2257 // CHECK1: cond.false: 2258 // CHECK1-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 2259 // CHECK1-NEXT: br label [[COND_END]] 2260 // CHECK1: cond.end: 2261 // CHECK1-NEXT: [[COND:%.*]] = phi i32 [ [[TMP12]], [[COND_TRUE]] ], [ [[TMP13]], [[COND_FALSE]] ] 2262 // CHECK1-NEXT: store i32 [[COND]], i32* [[DOTOMP_COMB_UB]], align 4 2263 // CHECK1-NEXT: [[TMP14:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4 2264 // CHECK1-NEXT: store i32 [[TMP14]], i32* [[DOTOMP_IV]], align 4 2265 // CHECK1-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] 2266 // CHECK1: omp.inner.for.cond: 2267 // CHECK1-NEXT: [[TMP15:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !43 2268 // CHECK1-NEXT: [[TMP16:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4, !llvm.access.group !43 2269 // CHECK1-NEXT: [[CMP5:%.*]] = icmp sle i32 [[TMP15]], [[TMP16]] 2270 // CHECK1-NEXT: br i1 [[CMP5]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] 2271 // CHECK1: omp.inner.for.body: 2272 // CHECK1-NEXT: [[TMP17:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4, !llvm.access.group !43 2273 // CHECK1-NEXT: [[TMP18:%.*]] = zext i32 [[TMP17]] to i64 2274 // CHECK1-NEXT: [[TMP19:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4, !llvm.access.group !43 2275 // CHECK1-NEXT: [[TMP20:%.*]] = zext i32 [[TMP19]] to i64 2276 // CHECK1-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 6, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64, i32*, double**, double**, double**)* @.omp_outlined..19 to void (i32*, i32*, ...)*), i64 [[TMP18]], i64 [[TMP20]], i32* [[TMP0]], double** [[TMP1]], double** [[TMP2]], double** [[TMP3]]), !llvm.access.group !43 2277 // CHECK1-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] 2278 // CHECK1: omp.inner.for.inc: 2279 // CHECK1-NEXT: [[TMP21:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !43 2280 // CHECK1-NEXT: [[TMP22:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4, !llvm.access.group !43 2281 // CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP21]], [[TMP22]] 2282 // CHECK1-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4, !llvm.access.group !43 2283 // CHECK1-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP44:![0-9]+]] 2284 // CHECK1: omp.inner.for.end: 2285 // CHECK1-NEXT: br label [[OMP_LOOP_EXIT:%.*]] 2286 // CHECK1: omp.loop.exit: 2287 // CHECK1-NEXT: [[TMP23:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 2288 // CHECK1-NEXT: [[TMP24:%.*]] = load i32, i32* [[TMP23]], align 4 2289 // CHECK1-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP24]]) 2290 // CHECK1-NEXT: [[TMP25:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4 2291 // CHECK1-NEXT: [[TMP26:%.*]] = icmp ne i32 [[TMP25]], 0 2292 // CHECK1-NEXT: br i1 [[TMP26]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]] 2293 // CHECK1: .omp.final.then: 2294 // CHECK1-NEXT: [[TMP27:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 2295 // CHECK1-NEXT: [[SUB6:%.*]] = sub nsw i32 [[TMP27]], 0 2296 // CHECK1-NEXT: [[DIV7:%.*]] = sdiv i32 [[SUB6]], 1 2297 // CHECK1-NEXT: [[MUL:%.*]] = mul nsw i32 [[DIV7]], 1 2298 // CHECK1-NEXT: [[ADD8:%.*]] = add nsw i32 0, [[MUL]] 2299 // CHECK1-NEXT: store i32 [[ADD8]], i32* [[I3]], align 4 2300 // CHECK1-NEXT: br label [[DOTOMP_FINAL_DONE]] 2301 // CHECK1: .omp.final.done: 2302 // CHECK1-NEXT: br label [[OMP_PRECOND_END]] 2303 // CHECK1: omp.precond.end: 2304 // CHECK1-NEXT: ret void 2305 // 2306 // 2307 // CHECK1-LABEL: define {{[^@]+}}@.omp_outlined..19 2308 // CHECK1-SAME: (i32* noalias noundef [[DOTGLOBAL_TID_:%.*]], i32* noalias noundef [[DOTBOUND_TID_:%.*]], i64 noundef [[DOTPREVIOUS_LB_:%.*]], i64 noundef [[DOTPREVIOUS_UB_:%.*]], i32* noundef nonnull align 4 dereferenceable(4) [[N:%.*]], double** noundef nonnull align 8 dereferenceable(8) [[A:%.*]], double** noundef nonnull align 8 dereferenceable(8) [[B:%.*]], double** noundef nonnull align 8 dereferenceable(8) [[C:%.*]]) #[[ATTR2]] { 2309 // CHECK1-NEXT: entry: 2310 // CHECK1-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 2311 // CHECK1-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 2312 // CHECK1-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8 2313 // CHECK1-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8 2314 // CHECK1-NEXT: [[N_ADDR:%.*]] = alloca i32*, align 8 2315 // CHECK1-NEXT: [[A_ADDR:%.*]] = alloca double**, align 8 2316 // CHECK1-NEXT: [[B_ADDR:%.*]] = alloca double**, align 8 2317 // CHECK1-NEXT: [[C_ADDR:%.*]] = alloca double**, align 8 2318 // CHECK1-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 2319 // CHECK1-NEXT: [[TMP:%.*]] = alloca i32, align 4 2320 // CHECK1-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 2321 // CHECK1-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 2322 // CHECK1-NEXT: [[I:%.*]] = alloca i32, align 4 2323 // CHECK1-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 2324 // CHECK1-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4 2325 // CHECK1-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 2326 // CHECK1-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 2327 // CHECK1-NEXT: [[I4:%.*]] = alloca i32, align 4 2328 // CHECK1-NEXT: [[REF_TMP:%.*]] = alloca [[CLASS_ANON_5:%.*]], align 8 2329 // CHECK1-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8 2330 // CHECK1-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8 2331 // CHECK1-NEXT: store i64 [[DOTPREVIOUS_LB_]], i64* [[DOTPREVIOUS_LB__ADDR]], align 8 2332 // CHECK1-NEXT: store i64 [[DOTPREVIOUS_UB_]], i64* [[DOTPREVIOUS_UB__ADDR]], align 8 2333 // CHECK1-NEXT: store i32* [[N]], i32** [[N_ADDR]], align 8 2334 // CHECK1-NEXT: store double** [[A]], double*** [[A_ADDR]], align 8 2335 // CHECK1-NEXT: store double** [[B]], double*** [[B_ADDR]], align 8 2336 // CHECK1-NEXT: store double** [[C]], double*** [[C_ADDR]], align 8 2337 // CHECK1-NEXT: [[TMP0:%.*]] = load i32*, i32** [[N_ADDR]], align 8 2338 // CHECK1-NEXT: [[TMP1:%.*]] = load double**, double*** [[A_ADDR]], align 8 2339 // CHECK1-NEXT: [[TMP2:%.*]] = load double**, double*** [[B_ADDR]], align 8 2340 // CHECK1-NEXT: [[TMP3:%.*]] = load double**, double*** [[C_ADDR]], align 8 2341 // CHECK1-NEXT: [[TMP4:%.*]] = load i32, i32* [[TMP0]], align 4 2342 // CHECK1-NEXT: store i32 [[TMP4]], i32* [[DOTCAPTURE_EXPR_]], align 4 2343 // CHECK1-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 2344 // CHECK1-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP5]], 0 2345 // CHECK1-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 2346 // CHECK1-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1 2347 // CHECK1-NEXT: store i32 [[SUB2]], i32* [[DOTCAPTURE_EXPR_1]], align 4 2348 // CHECK1-NEXT: store i32 0, i32* [[I]], align 4 2349 // CHECK1-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 2350 // CHECK1-NEXT: [[CMP:%.*]] = icmp slt i32 0, [[TMP6]] 2351 // CHECK1-NEXT: br i1 [[CMP]], label [[OMP_PRECOND_THEN:%.*]], label [[OMP_PRECOND_END:%.*]] 2352 // CHECK1: omp.precond.then: 2353 // CHECK1-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4 2354 // CHECK1-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 2355 // CHECK1-NEXT: store i32 [[TMP7]], i32* [[DOTOMP_UB]], align 4 2356 // CHECK1-NEXT: [[TMP8:%.*]] = load i64, i64* [[DOTPREVIOUS_LB__ADDR]], align 8 2357 // CHECK1-NEXT: [[CONV:%.*]] = trunc i64 [[TMP8]] to i32 2358 // CHECK1-NEXT: [[TMP9:%.*]] = load i64, i64* [[DOTPREVIOUS_UB__ADDR]], align 8 2359 // CHECK1-NEXT: [[CONV3:%.*]] = trunc i64 [[TMP9]] to i32 2360 // CHECK1-NEXT: store i32 [[CONV]], i32* [[DOTOMP_LB]], align 4 2361 // CHECK1-NEXT: store i32 [[CONV3]], i32* [[DOTOMP_UB]], align 4 2362 // CHECK1-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4 2363 // CHECK1-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4 2364 // CHECK1-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4 2365 // CHECK1-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 2366 // CHECK1-NEXT: [[TMP12:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 2367 // CHECK1-NEXT: [[TMP13:%.*]] = load i32, i32* [[TMP12]], align 4 2368 // CHECK1-NEXT: call void @__kmpc_dispatch_init_4(%struct.ident_t* @[[GLOB3]], i32 [[TMP13]], i32 35, i32 [[TMP10]], i32 [[TMP11]], i32 1, i32 1) 2369 // CHECK1-NEXT: br label [[OMP_DISPATCH_COND:%.*]] 2370 // CHECK1: omp.dispatch.cond: 2371 // CHECK1-NEXT: [[TMP14:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 2372 // CHECK1-NEXT: [[TMP15:%.*]] = load i32, i32* [[TMP14]], align 4 2373 // CHECK1-NEXT: [[TMP16:%.*]] = call i32 @__kmpc_dispatch_next_4(%struct.ident_t* @[[GLOB3]], i32 [[TMP15]], i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]]) 2374 // CHECK1-NEXT: [[TOBOOL:%.*]] = icmp ne i32 [[TMP16]], 0 2375 // CHECK1-NEXT: br i1 [[TOBOOL]], label [[OMP_DISPATCH_BODY:%.*]], label [[OMP_DISPATCH_END:%.*]] 2376 // CHECK1: omp.dispatch.body: 2377 // CHECK1-NEXT: [[TMP17:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4 2378 // CHECK1-NEXT: store i32 [[TMP17]], i32* [[DOTOMP_IV]], align 4 2379 // CHECK1-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] 2380 // CHECK1: omp.inner.for.cond: 2381 // CHECK1-NEXT: [[TMP18:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !46 2382 // CHECK1-NEXT: [[TMP19:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4, !llvm.access.group !46 2383 // CHECK1-NEXT: [[CMP5:%.*]] = icmp sle i32 [[TMP18]], [[TMP19]] 2384 // CHECK1-NEXT: br i1 [[CMP5]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] 2385 // CHECK1: omp.inner.for.body: 2386 // CHECK1-NEXT: [[TMP20:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !46 2387 // CHECK1-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP20]], 1 2388 // CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] 2389 // CHECK1-NEXT: store i32 [[ADD]], i32* [[I4]], align 4, !llvm.access.group !46 2390 // CHECK1-NEXT: [[TMP21:%.*]] = load double*, double** [[TMP2]], align 8, !llvm.access.group !46 2391 // CHECK1-NEXT: [[TMP22:%.*]] = load i32, i32* [[I4]], align 4, !llvm.access.group !46 2392 // CHECK1-NEXT: [[IDXPROM:%.*]] = sext i32 [[TMP22]] to i64 2393 // CHECK1-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds double, double* [[TMP21]], i64 [[IDXPROM]] 2394 // CHECK1-NEXT: [[TMP23:%.*]] = load double, double* [[ARRAYIDX]], align 8, !llvm.access.group !46 2395 // CHECK1-NEXT: [[TMP24:%.*]] = load double*, double** [[TMP3]], align 8, !llvm.access.group !46 2396 // CHECK1-NEXT: [[TMP25:%.*]] = load i32, i32* [[I4]], align 4, !llvm.access.group !46 2397 // CHECK1-NEXT: [[IDXPROM6:%.*]] = sext i32 [[TMP25]] to i64 2398 // CHECK1-NEXT: [[ARRAYIDX7:%.*]] = getelementptr inbounds double, double* [[TMP24]], i64 [[IDXPROM6]] 2399 // CHECK1-NEXT: [[TMP26:%.*]] = load double, double* [[ARRAYIDX7]], align 8, !llvm.access.group !46 2400 // CHECK1-NEXT: [[ADD8:%.*]] = fadd double [[TMP23]], [[TMP26]] 2401 // CHECK1-NEXT: [[TMP27:%.*]] = load double*, double** [[TMP1]], align 8, !llvm.access.group !46 2402 // CHECK1-NEXT: [[TMP28:%.*]] = load i32, i32* [[I4]], align 4, !llvm.access.group !46 2403 // CHECK1-NEXT: [[IDXPROM9:%.*]] = sext i32 [[TMP28]] to i64 2404 // CHECK1-NEXT: [[ARRAYIDX10:%.*]] = getelementptr inbounds double, double* [[TMP27]], i64 [[IDXPROM9]] 2405 // CHECK1-NEXT: store double [[ADD8]], double* [[ARRAYIDX10]], align 8, !llvm.access.group !46 2406 // CHECK1-NEXT: [[TMP29:%.*]] = getelementptr inbounds [[CLASS_ANON_5]], %class.anon.5* [[REF_TMP]], i32 0, i32 0 2407 // CHECK1-NEXT: store double** [[TMP1]], double*** [[TMP29]], align 8, !llvm.access.group !46 2408 // CHECK1-NEXT: [[TMP30:%.*]] = getelementptr inbounds [[CLASS_ANON_5]], %class.anon.5* [[REF_TMP]], i32 0, i32 1 2409 // CHECK1-NEXT: store i32* [[I4]], i32** [[TMP30]], align 8, !llvm.access.group !46 2410 // CHECK1-NEXT: [[TMP31:%.*]] = getelementptr inbounds [[CLASS_ANON_5]], %class.anon.5* [[REF_TMP]], i32 0, i32 2 2411 // CHECK1-NEXT: store double** [[TMP2]], double*** [[TMP31]], align 8, !llvm.access.group !46 2412 // CHECK1-NEXT: [[TMP32:%.*]] = getelementptr inbounds [[CLASS_ANON_5]], %class.anon.5* [[REF_TMP]], i32 0, i32 3 2413 // CHECK1-NEXT: store double** [[TMP3]], double*** [[TMP32]], align 8, !llvm.access.group !46 2414 // CHECK1-NEXT: call void @"_ZZZ4mainENK3$_0clEvENKUlvE4_clEv"(%class.anon.5* noundef nonnull align 8 dereferenceable(32) [[REF_TMP]]), !llvm.access.group !46 2415 // CHECK1-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] 2416 // CHECK1: omp.body.continue: 2417 // CHECK1-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] 2418 // CHECK1: omp.inner.for.inc: 2419 // CHECK1-NEXT: [[TMP33:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !46 2420 // CHECK1-NEXT: [[ADD11:%.*]] = add nsw i32 [[TMP33]], 1 2421 // CHECK1-NEXT: store i32 [[ADD11]], i32* [[DOTOMP_IV]], align 4, !llvm.access.group !46 2422 // CHECK1-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP47:![0-9]+]] 2423 // CHECK1: omp.inner.for.end: 2424 // CHECK1-NEXT: br label [[OMP_DISPATCH_INC:%.*]] 2425 // CHECK1: omp.dispatch.inc: 2426 // CHECK1-NEXT: br label [[OMP_DISPATCH_COND]] 2427 // CHECK1: omp.dispatch.end: 2428 // CHECK1-NEXT: [[TMP34:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4 2429 // CHECK1-NEXT: [[TMP35:%.*]] = icmp ne i32 [[TMP34]], 0 2430 // CHECK1-NEXT: br i1 [[TMP35]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]] 2431 // CHECK1: .omp.final.then: 2432 // CHECK1-NEXT: [[TMP36:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 2433 // CHECK1-NEXT: [[SUB12:%.*]] = sub nsw i32 [[TMP36]], 0 2434 // CHECK1-NEXT: [[DIV13:%.*]] = sdiv i32 [[SUB12]], 1 2435 // CHECK1-NEXT: [[MUL14:%.*]] = mul nsw i32 [[DIV13]], 1 2436 // CHECK1-NEXT: [[ADD15:%.*]] = add nsw i32 0, [[MUL14]] 2437 // CHECK1-NEXT: store i32 [[ADD15]], i32* [[I4]], align 4 2438 // CHECK1-NEXT: br label [[DOTOMP_FINAL_DONE]] 2439 // CHECK1: .omp.final.done: 2440 // CHECK1-NEXT: br label [[OMP_PRECOND_END]] 2441 // CHECK1: omp.precond.end: 2442 // CHECK1-NEXT: ret void 2443 // 2444 // 2445 // CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l328 2446 // CHECK1-SAME: (i64 noundef [[CH:%.*]], i64 noundef [[N:%.*]], double* noundef [[A:%.*]], double* noundef [[B:%.*]], double* noundef [[C:%.*]]) #[[ATTR2]] { 2447 // CHECK1-NEXT: entry: 2448 // CHECK1-NEXT: [[CH_ADDR:%.*]] = alloca i64, align 8 2449 // CHECK1-NEXT: [[N_ADDR:%.*]] = alloca i64, align 8 2450 // CHECK1-NEXT: [[A_ADDR:%.*]] = alloca double*, align 8 2451 // CHECK1-NEXT: [[B_ADDR:%.*]] = alloca double*, align 8 2452 // CHECK1-NEXT: [[C_ADDR:%.*]] = alloca double*, align 8 2453 // CHECK1-NEXT: store i64 [[CH]], i64* [[CH_ADDR]], align 8 2454 // CHECK1-NEXT: store i64 [[N]], i64* [[N_ADDR]], align 8 2455 // CHECK1-NEXT: store double* [[A]], double** [[A_ADDR]], align 8 2456 // CHECK1-NEXT: store double* [[B]], double** [[B_ADDR]], align 8 2457 // CHECK1-NEXT: store double* [[C]], double** [[C_ADDR]], align 8 2458 // CHECK1-NEXT: [[CONV:%.*]] = bitcast i64* [[CH_ADDR]] to i32* 2459 // CHECK1-NEXT: [[CONV1:%.*]] = bitcast i64* [[N_ADDR]] to i32* 2460 // CHECK1-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 5, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32*, i32*, double**, double**, double**)* @.omp_outlined..22 to void (i32*, i32*, ...)*), i32* [[CONV]], i32* [[CONV1]], double** [[A_ADDR]], double** [[B_ADDR]], double** [[C_ADDR]]) 2461 // CHECK1-NEXT: ret void 2462 // 2463 // 2464 // CHECK1-LABEL: define {{[^@]+}}@.omp_outlined..22 2465 // CHECK1-SAME: (i32* noalias noundef [[DOTGLOBAL_TID_:%.*]], i32* noalias noundef [[DOTBOUND_TID_:%.*]], i32* noundef nonnull align 4 dereferenceable(4) [[CH:%.*]], i32* noundef nonnull align 4 dereferenceable(4) [[N:%.*]], double** noundef nonnull align 8 dereferenceable(8) [[A:%.*]], double** noundef nonnull align 8 dereferenceable(8) [[B:%.*]], double** noundef nonnull align 8 dereferenceable(8) [[C:%.*]]) #[[ATTR2]] { 2466 // CHECK1-NEXT: entry: 2467 // CHECK1-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 2468 // CHECK1-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 2469 // CHECK1-NEXT: [[CH_ADDR:%.*]] = alloca i32*, align 8 2470 // CHECK1-NEXT: [[N_ADDR:%.*]] = alloca i32*, align 8 2471 // CHECK1-NEXT: [[A_ADDR:%.*]] = alloca double**, align 8 2472 // CHECK1-NEXT: [[B_ADDR:%.*]] = alloca double**, align 8 2473 // CHECK1-NEXT: [[C_ADDR:%.*]] = alloca double**, align 8 2474 // CHECK1-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 2475 // CHECK1-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 2476 // CHECK1-NEXT: [[TMP:%.*]] = alloca i32, align 4 2477 // CHECK1-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 2478 // CHECK1-NEXT: [[DOTCAPTURE_EXPR_2:%.*]] = alloca i32, align 4 2479 // CHECK1-NEXT: [[I:%.*]] = alloca i32, align 4 2480 // CHECK1-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4 2481 // CHECK1-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4 2482 // CHECK1-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 2483 // CHECK1-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 2484 // CHECK1-NEXT: [[I4:%.*]] = alloca i32, align 4 2485 // CHECK1-NEXT: [[DOTCAPTURE_EXPR__CASTED:%.*]] = alloca i64, align 8 2486 // CHECK1-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8 2487 // CHECK1-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8 2488 // CHECK1-NEXT: store i32* [[CH]], i32** [[CH_ADDR]], align 8 2489 // CHECK1-NEXT: store i32* [[N]], i32** [[N_ADDR]], align 8 2490 // CHECK1-NEXT: store double** [[A]], double*** [[A_ADDR]], align 8 2491 // CHECK1-NEXT: store double** [[B]], double*** [[B_ADDR]], align 8 2492 // CHECK1-NEXT: store double** [[C]], double*** [[C_ADDR]], align 8 2493 // CHECK1-NEXT: [[TMP0:%.*]] = load i32*, i32** [[CH_ADDR]], align 8 2494 // CHECK1-NEXT: [[TMP1:%.*]] = load i32*, i32** [[N_ADDR]], align 8 2495 // CHECK1-NEXT: [[TMP2:%.*]] = load double**, double*** [[A_ADDR]], align 8 2496 // CHECK1-NEXT: [[TMP3:%.*]] = load double**, double*** [[B_ADDR]], align 8 2497 // CHECK1-NEXT: [[TMP4:%.*]] = load double**, double*** [[C_ADDR]], align 8 2498 // CHECK1-NEXT: [[TMP5:%.*]] = load i32, i32* [[TMP0]], align 4 2499 // CHECK1-NEXT: store i32 [[TMP5]], i32* [[DOTCAPTURE_EXPR_]], align 4 2500 // CHECK1-NEXT: [[TMP6:%.*]] = load i32, i32* [[TMP1]], align 4 2501 // CHECK1-NEXT: store i32 [[TMP6]], i32* [[DOTCAPTURE_EXPR_1]], align 4 2502 // CHECK1-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 2503 // CHECK1-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP7]], 0 2504 // CHECK1-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 2505 // CHECK1-NEXT: [[SUB3:%.*]] = sub nsw i32 [[DIV]], 1 2506 // CHECK1-NEXT: store i32 [[SUB3]], i32* [[DOTCAPTURE_EXPR_2]], align 4 2507 // CHECK1-NEXT: store i32 0, i32* [[I]], align 4 2508 // CHECK1-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 2509 // CHECK1-NEXT: [[CMP:%.*]] = icmp slt i32 0, [[TMP8]] 2510 // CHECK1-NEXT: br i1 [[CMP]], label [[OMP_PRECOND_THEN:%.*]], label [[OMP_PRECOND_END:%.*]] 2511 // CHECK1: omp.precond.then: 2512 // CHECK1-NEXT: store i32 0, i32* [[DOTOMP_COMB_LB]], align 4 2513 // CHECK1-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_2]], align 4 2514 // CHECK1-NEXT: store i32 [[TMP9]], i32* [[DOTOMP_COMB_UB]], align 4 2515 // CHECK1-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4 2516 // CHECK1-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4 2517 // CHECK1-NEXT: [[TMP10:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 2518 // CHECK1-NEXT: [[TMP11:%.*]] = load i32, i32* [[TMP10]], align 4 2519 // CHECK1-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1]], i32 [[TMP11]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_COMB_LB]], i32* [[DOTOMP_COMB_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1) 2520 // CHECK1-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 2521 // CHECK1-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_2]], align 4 2522 // CHECK1-NEXT: [[CMP5:%.*]] = icmp sgt i32 [[TMP12]], [[TMP13]] 2523 // CHECK1-NEXT: br i1 [[CMP5]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] 2524 // CHECK1: cond.true: 2525 // CHECK1-NEXT: [[TMP14:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_2]], align 4 2526 // CHECK1-NEXT: br label [[COND_END:%.*]] 2527 // CHECK1: cond.false: 2528 // CHECK1-NEXT: [[TMP15:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 2529 // CHECK1-NEXT: br label [[COND_END]] 2530 // CHECK1: cond.end: 2531 // CHECK1-NEXT: [[COND:%.*]] = phi i32 [ [[TMP14]], [[COND_TRUE]] ], [ [[TMP15]], [[COND_FALSE]] ] 2532 // CHECK1-NEXT: store i32 [[COND]], i32* [[DOTOMP_COMB_UB]], align 4 2533 // CHECK1-NEXT: [[TMP16:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4 2534 // CHECK1-NEXT: store i32 [[TMP16]], i32* [[DOTOMP_IV]], align 4 2535 // CHECK1-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] 2536 // CHECK1: omp.inner.for.cond: 2537 // CHECK1-NEXT: [[TMP17:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !49 2538 // CHECK1-NEXT: [[TMP18:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4, !llvm.access.group !49 2539 // CHECK1-NEXT: [[CMP6:%.*]] = icmp sle i32 [[TMP17]], [[TMP18]] 2540 // CHECK1-NEXT: br i1 [[CMP6]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] 2541 // CHECK1: omp.inner.for.body: 2542 // CHECK1-NEXT: [[TMP19:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4, !llvm.access.group !49 2543 // CHECK1-NEXT: [[TMP20:%.*]] = zext i32 [[TMP19]] to i64 2544 // CHECK1-NEXT: [[TMP21:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4, !llvm.access.group !49 2545 // CHECK1-NEXT: [[TMP22:%.*]] = zext i32 [[TMP21]] to i64 2546 // CHECK1-NEXT: [[TMP23:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4, !llvm.access.group !49 2547 // CHECK1-NEXT: [[CONV:%.*]] = bitcast i64* [[DOTCAPTURE_EXPR__CASTED]] to i32* 2548 // CHECK1-NEXT: store i32 [[TMP23]], i32* [[CONV]], align 4, !llvm.access.group !49 2549 // CHECK1-NEXT: [[TMP24:%.*]] = load i64, i64* [[DOTCAPTURE_EXPR__CASTED]], align 8, !llvm.access.group !49 2550 // CHECK1-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 7, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64, i32*, double**, double**, double**, i64)* @.omp_outlined..23 to void (i32*, i32*, ...)*), i64 [[TMP20]], i64 [[TMP22]], i32* [[TMP1]], double** [[TMP2]], double** [[TMP3]], double** [[TMP4]], i64 [[TMP24]]), !llvm.access.group !49 2551 // CHECK1-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] 2552 // CHECK1: omp.inner.for.inc: 2553 // CHECK1-NEXT: [[TMP25:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !49 2554 // CHECK1-NEXT: [[TMP26:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4, !llvm.access.group !49 2555 // CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP25]], [[TMP26]] 2556 // CHECK1-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4, !llvm.access.group !49 2557 // CHECK1-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP50:![0-9]+]] 2558 // CHECK1: omp.inner.for.end: 2559 // CHECK1-NEXT: br label [[OMP_LOOP_EXIT:%.*]] 2560 // CHECK1: omp.loop.exit: 2561 // CHECK1-NEXT: [[TMP27:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 2562 // CHECK1-NEXT: [[TMP28:%.*]] = load i32, i32* [[TMP27]], align 4 2563 // CHECK1-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP28]]) 2564 // CHECK1-NEXT: [[TMP29:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4 2565 // CHECK1-NEXT: [[TMP30:%.*]] = icmp ne i32 [[TMP29]], 0 2566 // CHECK1-NEXT: br i1 [[TMP30]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]] 2567 // CHECK1: .omp.final.then: 2568 // CHECK1-NEXT: [[TMP31:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 2569 // CHECK1-NEXT: [[SUB7:%.*]] = sub nsw i32 [[TMP31]], 0 2570 // CHECK1-NEXT: [[DIV8:%.*]] = sdiv i32 [[SUB7]], 1 2571 // CHECK1-NEXT: [[MUL:%.*]] = mul nsw i32 [[DIV8]], 1 2572 // CHECK1-NEXT: [[ADD9:%.*]] = add nsw i32 0, [[MUL]] 2573 // CHECK1-NEXT: store i32 [[ADD9]], i32* [[I4]], align 4 2574 // CHECK1-NEXT: br label [[DOTOMP_FINAL_DONE]] 2575 // CHECK1: .omp.final.done: 2576 // CHECK1-NEXT: br label [[OMP_PRECOND_END]] 2577 // CHECK1: omp.precond.end: 2578 // CHECK1-NEXT: ret void 2579 // 2580 // 2581 // CHECK1-LABEL: define {{[^@]+}}@.omp_outlined..23 2582 // CHECK1-SAME: (i32* noalias noundef [[DOTGLOBAL_TID_:%.*]], i32* noalias noundef [[DOTBOUND_TID_:%.*]], i64 noundef [[DOTPREVIOUS_LB_:%.*]], i64 noundef [[DOTPREVIOUS_UB_:%.*]], i32* noundef nonnull align 4 dereferenceable(4) [[N:%.*]], double** noundef nonnull align 8 dereferenceable(8) [[A:%.*]], double** noundef nonnull align 8 dereferenceable(8) [[B:%.*]], double** noundef nonnull align 8 dereferenceable(8) [[C:%.*]], i64 noundef [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR2]] { 2583 // CHECK1-NEXT: entry: 2584 // CHECK1-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 2585 // CHECK1-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 2586 // CHECK1-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8 2587 // CHECK1-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8 2588 // CHECK1-NEXT: [[N_ADDR:%.*]] = alloca i32*, align 8 2589 // CHECK1-NEXT: [[A_ADDR:%.*]] = alloca double**, align 8 2590 // CHECK1-NEXT: [[B_ADDR:%.*]] = alloca double**, align 8 2591 // CHECK1-NEXT: [[C_ADDR:%.*]] = alloca double**, align 8 2592 // CHECK1-NEXT: [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i64, align 8 2593 // CHECK1-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 2594 // CHECK1-NEXT: [[TMP:%.*]] = alloca i32, align 4 2595 // CHECK1-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 2596 // CHECK1-NEXT: [[DOTCAPTURE_EXPR_2:%.*]] = alloca i32, align 4 2597 // CHECK1-NEXT: [[I:%.*]] = alloca i32, align 4 2598 // CHECK1-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 2599 // CHECK1-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4 2600 // CHECK1-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 2601 // CHECK1-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 2602 // CHECK1-NEXT: [[I6:%.*]] = alloca i32, align 4 2603 // CHECK1-NEXT: [[REF_TMP:%.*]] = alloca [[CLASS_ANON_6:%.*]], align 8 2604 // CHECK1-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8 2605 // CHECK1-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8 2606 // CHECK1-NEXT: store i64 [[DOTPREVIOUS_LB_]], i64* [[DOTPREVIOUS_LB__ADDR]], align 8 2607 // CHECK1-NEXT: store i64 [[DOTPREVIOUS_UB_]], i64* [[DOTPREVIOUS_UB__ADDR]], align 8 2608 // CHECK1-NEXT: store i32* [[N]], i32** [[N_ADDR]], align 8 2609 // CHECK1-NEXT: store double** [[A]], double*** [[A_ADDR]], align 8 2610 // CHECK1-NEXT: store double** [[B]], double*** [[B_ADDR]], align 8 2611 // CHECK1-NEXT: store double** [[C]], double*** [[C_ADDR]], align 8 2612 // CHECK1-NEXT: store i64 [[DOTCAPTURE_EXPR_]], i64* [[DOTCAPTURE_EXPR__ADDR]], align 8 2613 // CHECK1-NEXT: [[TMP0:%.*]] = load i32*, i32** [[N_ADDR]], align 8 2614 // CHECK1-NEXT: [[TMP1:%.*]] = load double**, double*** [[A_ADDR]], align 8 2615 // CHECK1-NEXT: [[TMP2:%.*]] = load double**, double*** [[B_ADDR]], align 8 2616 // CHECK1-NEXT: [[TMP3:%.*]] = load double**, double*** [[C_ADDR]], align 8 2617 // CHECK1-NEXT: [[CONV:%.*]] = bitcast i64* [[DOTCAPTURE_EXPR__ADDR]] to i32* 2618 // CHECK1-NEXT: [[TMP4:%.*]] = load i32, i32* [[TMP0]], align 4 2619 // CHECK1-NEXT: store i32 [[TMP4]], i32* [[DOTCAPTURE_EXPR_1]], align 4 2620 // CHECK1-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 2621 // CHECK1-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP5]], 0 2622 // CHECK1-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 2623 // CHECK1-NEXT: [[SUB3:%.*]] = sub nsw i32 [[DIV]], 1 2624 // CHECK1-NEXT: store i32 [[SUB3]], i32* [[DOTCAPTURE_EXPR_2]], align 4 2625 // CHECK1-NEXT: store i32 0, i32* [[I]], align 4 2626 // CHECK1-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 2627 // CHECK1-NEXT: [[CMP:%.*]] = icmp slt i32 0, [[TMP6]] 2628 // CHECK1-NEXT: br i1 [[CMP]], label [[OMP_PRECOND_THEN:%.*]], label [[OMP_PRECOND_END:%.*]] 2629 // CHECK1: omp.precond.then: 2630 // CHECK1-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4 2631 // CHECK1-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_2]], align 4 2632 // CHECK1-NEXT: store i32 [[TMP7]], i32* [[DOTOMP_UB]], align 4 2633 // CHECK1-NEXT: [[TMP8:%.*]] = load i64, i64* [[DOTPREVIOUS_LB__ADDR]], align 8 2634 // CHECK1-NEXT: [[CONV4:%.*]] = trunc i64 [[TMP8]] to i32 2635 // CHECK1-NEXT: [[TMP9:%.*]] = load i64, i64* [[DOTPREVIOUS_UB__ADDR]], align 8 2636 // CHECK1-NEXT: [[CONV5:%.*]] = trunc i64 [[TMP9]] to i32 2637 // CHECK1-NEXT: store i32 [[CONV4]], i32* [[DOTOMP_LB]], align 4 2638 // CHECK1-NEXT: store i32 [[CONV5]], i32* [[DOTOMP_UB]], align 4 2639 // CHECK1-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4 2640 // CHECK1-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4 2641 // CHECK1-NEXT: [[TMP10:%.*]] = load i32, i32* [[CONV]], align 4 2642 // CHECK1-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4 2643 // CHECK1-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 2644 // CHECK1-NEXT: [[TMP13:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 2645 // CHECK1-NEXT: [[TMP14:%.*]] = load i32, i32* [[TMP13]], align 4 2646 // CHECK1-NEXT: call void @__kmpc_dispatch_init_4(%struct.ident_t* @[[GLOB3]], i32 [[TMP14]], i32 35, i32 [[TMP11]], i32 [[TMP12]], i32 1, i32 [[TMP10]]) 2647 // CHECK1-NEXT: br label [[OMP_DISPATCH_COND:%.*]] 2648 // CHECK1: omp.dispatch.cond: 2649 // CHECK1-NEXT: [[TMP15:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 2650 // CHECK1-NEXT: [[TMP16:%.*]] = load i32, i32* [[TMP15]], align 4 2651 // CHECK1-NEXT: [[TMP17:%.*]] = call i32 @__kmpc_dispatch_next_4(%struct.ident_t* @[[GLOB3]], i32 [[TMP16]], i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]]) 2652 // CHECK1-NEXT: [[TOBOOL:%.*]] = icmp ne i32 [[TMP17]], 0 2653 // CHECK1-NEXT: br i1 [[TOBOOL]], label [[OMP_DISPATCH_BODY:%.*]], label [[OMP_DISPATCH_END:%.*]] 2654 // CHECK1: omp.dispatch.body: 2655 // CHECK1-NEXT: [[TMP18:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4 2656 // CHECK1-NEXT: store i32 [[TMP18]], i32* [[DOTOMP_IV]], align 4 2657 // CHECK1-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] 2658 // CHECK1: omp.inner.for.cond: 2659 // CHECK1-NEXT: [[TMP19:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !52 2660 // CHECK1-NEXT: [[TMP20:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4, !llvm.access.group !52 2661 // CHECK1-NEXT: [[CMP7:%.*]] = icmp sle i32 [[TMP19]], [[TMP20]] 2662 // CHECK1-NEXT: br i1 [[CMP7]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] 2663 // CHECK1: omp.inner.for.body: 2664 // CHECK1-NEXT: [[TMP21:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !52 2665 // CHECK1-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP21]], 1 2666 // CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] 2667 // CHECK1-NEXT: store i32 [[ADD]], i32* [[I6]], align 4, !llvm.access.group !52 2668 // CHECK1-NEXT: [[TMP22:%.*]] = load double*, double** [[TMP2]], align 8, !llvm.access.group !52 2669 // CHECK1-NEXT: [[TMP23:%.*]] = load i32, i32* [[I6]], align 4, !llvm.access.group !52 2670 // CHECK1-NEXT: [[IDXPROM:%.*]] = sext i32 [[TMP23]] to i64 2671 // CHECK1-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds double, double* [[TMP22]], i64 [[IDXPROM]] 2672 // CHECK1-NEXT: [[TMP24:%.*]] = load double, double* [[ARRAYIDX]], align 8, !llvm.access.group !52 2673 // CHECK1-NEXT: [[TMP25:%.*]] = load double*, double** [[TMP3]], align 8, !llvm.access.group !52 2674 // CHECK1-NEXT: [[TMP26:%.*]] = load i32, i32* [[I6]], align 4, !llvm.access.group !52 2675 // CHECK1-NEXT: [[IDXPROM8:%.*]] = sext i32 [[TMP26]] to i64 2676 // CHECK1-NEXT: [[ARRAYIDX9:%.*]] = getelementptr inbounds double, double* [[TMP25]], i64 [[IDXPROM8]] 2677 // CHECK1-NEXT: [[TMP27:%.*]] = load double, double* [[ARRAYIDX9]], align 8, !llvm.access.group !52 2678 // CHECK1-NEXT: [[ADD10:%.*]] = fadd double [[TMP24]], [[TMP27]] 2679 // CHECK1-NEXT: [[TMP28:%.*]] = load double*, double** [[TMP1]], align 8, !llvm.access.group !52 2680 // CHECK1-NEXT: [[TMP29:%.*]] = load i32, i32* [[I6]], align 4, !llvm.access.group !52 2681 // CHECK1-NEXT: [[IDXPROM11:%.*]] = sext i32 [[TMP29]] to i64 2682 // CHECK1-NEXT: [[ARRAYIDX12:%.*]] = getelementptr inbounds double, double* [[TMP28]], i64 [[IDXPROM11]] 2683 // CHECK1-NEXT: store double [[ADD10]], double* [[ARRAYIDX12]], align 8, !llvm.access.group !52 2684 // CHECK1-NEXT: [[TMP30:%.*]] = getelementptr inbounds [[CLASS_ANON_6]], %class.anon.6* [[REF_TMP]], i32 0, i32 0 2685 // CHECK1-NEXT: store double** [[TMP1]], double*** [[TMP30]], align 8, !llvm.access.group !52 2686 // CHECK1-NEXT: [[TMP31:%.*]] = getelementptr inbounds [[CLASS_ANON_6]], %class.anon.6* [[REF_TMP]], i32 0, i32 1 2687 // CHECK1-NEXT: store i32* [[I6]], i32** [[TMP31]], align 8, !llvm.access.group !52 2688 // CHECK1-NEXT: [[TMP32:%.*]] = getelementptr inbounds [[CLASS_ANON_6]], %class.anon.6* [[REF_TMP]], i32 0, i32 2 2689 // CHECK1-NEXT: store double** [[TMP2]], double*** [[TMP32]], align 8, !llvm.access.group !52 2690 // CHECK1-NEXT: [[TMP33:%.*]] = getelementptr inbounds [[CLASS_ANON_6]], %class.anon.6* [[REF_TMP]], i32 0, i32 3 2691 // CHECK1-NEXT: store double** [[TMP3]], double*** [[TMP33]], align 8, !llvm.access.group !52 2692 // CHECK1-NEXT: call void @"_ZZZ4mainENK3$_0clEvENKUlvE5_clEv"(%class.anon.6* noundef nonnull align 8 dereferenceable(32) [[REF_TMP]]), !llvm.access.group !52 2693 // CHECK1-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] 2694 // CHECK1: omp.body.continue: 2695 // CHECK1-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] 2696 // CHECK1: omp.inner.for.inc: 2697 // CHECK1-NEXT: [[TMP34:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !52 2698 // CHECK1-NEXT: [[ADD13:%.*]] = add nsw i32 [[TMP34]], 1 2699 // CHECK1-NEXT: store i32 [[ADD13]], i32* [[DOTOMP_IV]], align 4, !llvm.access.group !52 2700 // CHECK1-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP53:![0-9]+]] 2701 // CHECK1: omp.inner.for.end: 2702 // CHECK1-NEXT: br label [[OMP_DISPATCH_INC:%.*]] 2703 // CHECK1: omp.dispatch.inc: 2704 // CHECK1-NEXT: br label [[OMP_DISPATCH_COND]] 2705 // CHECK1: omp.dispatch.end: 2706 // CHECK1-NEXT: [[TMP35:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4 2707 // CHECK1-NEXT: [[TMP36:%.*]] = icmp ne i32 [[TMP35]], 0 2708 // CHECK1-NEXT: br i1 [[TMP36]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]] 2709 // CHECK1: .omp.final.then: 2710 // CHECK1-NEXT: [[TMP37:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 2711 // CHECK1-NEXT: [[SUB14:%.*]] = sub nsw i32 [[TMP37]], 0 2712 // CHECK1-NEXT: [[DIV15:%.*]] = sdiv i32 [[SUB14]], 1 2713 // CHECK1-NEXT: [[MUL16:%.*]] = mul nsw i32 [[DIV15]], 1 2714 // CHECK1-NEXT: [[ADD17:%.*]] = add nsw i32 0, [[MUL16]] 2715 // CHECK1-NEXT: store i32 [[ADD17]], i32* [[I6]], align 4 2716 // CHECK1-NEXT: br label [[DOTOMP_FINAL_DONE]] 2717 // CHECK1: .omp.final.done: 2718 // CHECK1-NEXT: br label [[OMP_PRECOND_END]] 2719 // CHECK1: omp.precond.end: 2720 // CHECK1-NEXT: ret void 2721 // 2722 // 2723 // CHECK1-LABEL: define {{[^@]+}}@.omp_offloading.requires_reg 2724 // CHECK1-SAME: () #[[ATTR4:[0-9]+]] { 2725 // CHECK1-NEXT: entry: 2726 // CHECK1-NEXT: call void @__tgt_register_requires(i64 1) 2727 // CHECK1-NEXT: ret void 2728 // 2729 // 2730 // CHECK3-LABEL: define {{[^@]+}}@main 2731 // CHECK3-SAME: () #[[ATTR0:[0-9]+]] { 2732 // CHECK3-NEXT: entry: 2733 // CHECK3-NEXT: [[RETVAL:%.*]] = alloca i32, align 4 2734 // CHECK3-NEXT: [[A:%.*]] = alloca double*, align 4 2735 // CHECK3-NEXT: [[B:%.*]] = alloca double*, align 4 2736 // CHECK3-NEXT: [[C:%.*]] = alloca double*, align 4 2737 // CHECK3-NEXT: [[N:%.*]] = alloca i32, align 4 2738 // CHECK3-NEXT: [[CH:%.*]] = alloca i32, align 4 2739 // CHECK3-NEXT: [[REF_TMP:%.*]] = alloca [[CLASS_ANON:%.*]], align 4 2740 // CHECK3-NEXT: store i32 0, i32* [[RETVAL]], align 4 2741 // CHECK3-NEXT: store i32 10000, i32* [[N]], align 4 2742 // CHECK3-NEXT: store i32 100, i32* [[CH]], align 4 2743 // CHECK3-NEXT: [[TMP0:%.*]] = getelementptr inbounds [[CLASS_ANON]], %class.anon* [[REF_TMP]], i32 0, i32 0 2744 // CHECK3-NEXT: store i32* [[N]], i32** [[TMP0]], align 4 2745 // CHECK3-NEXT: [[TMP1:%.*]] = getelementptr inbounds [[CLASS_ANON]], %class.anon* [[REF_TMP]], i32 0, i32 1 2746 // CHECK3-NEXT: store double** [[A]], double*** [[TMP1]], align 4 2747 // CHECK3-NEXT: [[TMP2:%.*]] = getelementptr inbounds [[CLASS_ANON]], %class.anon* [[REF_TMP]], i32 0, i32 2 2748 // CHECK3-NEXT: store double** [[B]], double*** [[TMP2]], align 4 2749 // CHECK3-NEXT: [[TMP3:%.*]] = getelementptr inbounds [[CLASS_ANON]], %class.anon* [[REF_TMP]], i32 0, i32 3 2750 // CHECK3-NEXT: store double** [[C]], double*** [[TMP3]], align 4 2751 // CHECK3-NEXT: [[TMP4:%.*]] = getelementptr inbounds [[CLASS_ANON]], %class.anon* [[REF_TMP]], i32 0, i32 4 2752 // CHECK3-NEXT: store i32* [[CH]], i32** [[TMP4]], align 4 2753 // CHECK3-NEXT: call void @"_ZZ4mainENK3$_0clEv"(%class.anon* noundef nonnull align 4 dereferenceable(20) [[REF_TMP]]) 2754 // CHECK3-NEXT: ret i32 0 2755 // 2756 // 2757 // CHECK3-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l116 2758 // CHECK3-SAME: (i32 noundef [[N:%.*]], double* noundef [[A:%.*]], double* noundef [[B:%.*]], double* noundef [[C:%.*]]) #[[ATTR2:[0-9]+]] { 2759 // CHECK3-NEXT: entry: 2760 // CHECK3-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4 2761 // CHECK3-NEXT: [[A_ADDR:%.*]] = alloca double*, align 4 2762 // CHECK3-NEXT: [[B_ADDR:%.*]] = alloca double*, align 4 2763 // CHECK3-NEXT: [[C_ADDR:%.*]] = alloca double*, align 4 2764 // CHECK3-NEXT: store i32 [[N]], i32* [[N_ADDR]], align 4 2765 // CHECK3-NEXT: store double* [[A]], double** [[A_ADDR]], align 4 2766 // CHECK3-NEXT: store double* [[B]], double** [[B_ADDR]], align 4 2767 // CHECK3-NEXT: store double* [[C]], double** [[C_ADDR]], align 4 2768 // CHECK3-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3:[0-9]+]], i32 4, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32*, double**, double**, double**)* @.omp_outlined. to void (i32*, i32*, ...)*), i32* [[N_ADDR]], double** [[A_ADDR]], double** [[B_ADDR]], double** [[C_ADDR]]) 2769 // CHECK3-NEXT: ret void 2770 // 2771 // 2772 // CHECK3-LABEL: define {{[^@]+}}@.omp_outlined. 2773 // CHECK3-SAME: (i32* noalias noundef [[DOTGLOBAL_TID_:%.*]], i32* noalias noundef [[DOTBOUND_TID_:%.*]], i32* noundef nonnull align 4 dereferenceable(4) [[N:%.*]], double** noundef nonnull align 4 dereferenceable(4) [[A:%.*]], double** noundef nonnull align 4 dereferenceable(4) [[B:%.*]], double** noundef nonnull align 4 dereferenceable(4) [[C:%.*]]) #[[ATTR2]] { 2774 // CHECK3-NEXT: entry: 2775 // CHECK3-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4 2776 // CHECK3-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4 2777 // CHECK3-NEXT: [[N_ADDR:%.*]] = alloca i32*, align 4 2778 // CHECK3-NEXT: [[A_ADDR:%.*]] = alloca double**, align 4 2779 // CHECK3-NEXT: [[B_ADDR:%.*]] = alloca double**, align 4 2780 // CHECK3-NEXT: [[C_ADDR:%.*]] = alloca double**, align 4 2781 // CHECK3-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 2782 // CHECK3-NEXT: [[TMP:%.*]] = alloca i32, align 4 2783 // CHECK3-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 2784 // CHECK3-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 2785 // CHECK3-NEXT: [[I:%.*]] = alloca i32, align 4 2786 // CHECK3-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4 2787 // CHECK3-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4 2788 // CHECK3-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 2789 // CHECK3-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 2790 // CHECK3-NEXT: [[I3:%.*]] = alloca i32, align 4 2791 // CHECK3-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 4 2792 // CHECK3-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 4 2793 // CHECK3-NEXT: store i32* [[N]], i32** [[N_ADDR]], align 4 2794 // CHECK3-NEXT: store double** [[A]], double*** [[A_ADDR]], align 4 2795 // CHECK3-NEXT: store double** [[B]], double*** [[B_ADDR]], align 4 2796 // CHECK3-NEXT: store double** [[C]], double*** [[C_ADDR]], align 4 2797 // CHECK3-NEXT: [[TMP0:%.*]] = load i32*, i32** [[N_ADDR]], align 4 2798 // CHECK3-NEXT: [[TMP1:%.*]] = load double**, double*** [[A_ADDR]], align 4 2799 // CHECK3-NEXT: [[TMP2:%.*]] = load double**, double*** [[B_ADDR]], align 4 2800 // CHECK3-NEXT: [[TMP3:%.*]] = load double**, double*** [[C_ADDR]], align 4 2801 // CHECK3-NEXT: [[TMP4:%.*]] = load i32, i32* [[TMP0]], align 4 2802 // CHECK3-NEXT: store i32 [[TMP4]], i32* [[DOTCAPTURE_EXPR_]], align 4 2803 // CHECK3-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 2804 // CHECK3-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP5]], 0 2805 // CHECK3-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 2806 // CHECK3-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1 2807 // CHECK3-NEXT: store i32 [[SUB2]], i32* [[DOTCAPTURE_EXPR_1]], align 4 2808 // CHECK3-NEXT: store i32 0, i32* [[I]], align 4 2809 // CHECK3-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 2810 // CHECK3-NEXT: [[CMP:%.*]] = icmp slt i32 0, [[TMP6]] 2811 // CHECK3-NEXT: br i1 [[CMP]], label [[OMP_PRECOND_THEN:%.*]], label [[OMP_PRECOND_END:%.*]] 2812 // CHECK3: omp.precond.then: 2813 // CHECK3-NEXT: store i32 0, i32* [[DOTOMP_COMB_LB]], align 4 2814 // CHECK3-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 2815 // CHECK3-NEXT: store i32 [[TMP7]], i32* [[DOTOMP_COMB_UB]], align 4 2816 // CHECK3-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4 2817 // CHECK3-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4 2818 // CHECK3-NEXT: [[TMP8:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 4 2819 // CHECK3-NEXT: [[TMP9:%.*]] = load i32, i32* [[TMP8]], align 4 2820 // CHECK3-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1:[0-9]+]], i32 [[TMP9]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_COMB_LB]], i32* [[DOTOMP_COMB_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1) 2821 // CHECK3-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 2822 // CHECK3-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 2823 // CHECK3-NEXT: [[CMP4:%.*]] = icmp sgt i32 [[TMP10]], [[TMP11]] 2824 // CHECK3-NEXT: br i1 [[CMP4]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] 2825 // CHECK3: cond.true: 2826 // CHECK3-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 2827 // CHECK3-NEXT: br label [[COND_END:%.*]] 2828 // CHECK3: cond.false: 2829 // CHECK3-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 2830 // CHECK3-NEXT: br label [[COND_END]] 2831 // CHECK3: cond.end: 2832 // CHECK3-NEXT: [[COND:%.*]] = phi i32 [ [[TMP12]], [[COND_TRUE]] ], [ [[TMP13]], [[COND_FALSE]] ] 2833 // CHECK3-NEXT: store i32 [[COND]], i32* [[DOTOMP_COMB_UB]], align 4 2834 // CHECK3-NEXT: [[TMP14:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4 2835 // CHECK3-NEXT: store i32 [[TMP14]], i32* [[DOTOMP_IV]], align 4 2836 // CHECK3-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] 2837 // CHECK3: omp.inner.for.cond: 2838 // CHECK3-NEXT: [[TMP15:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !11 2839 // CHECK3-NEXT: [[TMP16:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4, !llvm.access.group !11 2840 // CHECK3-NEXT: [[CMP5:%.*]] = icmp sle i32 [[TMP15]], [[TMP16]] 2841 // CHECK3-NEXT: br i1 [[CMP5]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] 2842 // CHECK3: omp.inner.for.body: 2843 // CHECK3-NEXT: [[TMP17:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4, !llvm.access.group !11 2844 // CHECK3-NEXT: [[TMP18:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4, !llvm.access.group !11 2845 // CHECK3-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 6, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32, i32, i32*, double**, double**, double**)* @.omp_outlined..1 to void (i32*, i32*, ...)*), i32 [[TMP17]], i32 [[TMP18]], i32* [[TMP0]], double** [[TMP1]], double** [[TMP2]], double** [[TMP3]]), !llvm.access.group !11 2846 // CHECK3-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] 2847 // CHECK3: omp.inner.for.inc: 2848 // CHECK3-NEXT: [[TMP19:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !11 2849 // CHECK3-NEXT: [[TMP20:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4, !llvm.access.group !11 2850 // CHECK3-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP19]], [[TMP20]] 2851 // CHECK3-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4, !llvm.access.group !11 2852 // CHECK3-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP12:![0-9]+]] 2853 // CHECK3: omp.inner.for.end: 2854 // CHECK3-NEXT: br label [[OMP_LOOP_EXIT:%.*]] 2855 // CHECK3: omp.loop.exit: 2856 // CHECK3-NEXT: [[TMP21:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 4 2857 // CHECK3-NEXT: [[TMP22:%.*]] = load i32, i32* [[TMP21]], align 4 2858 // CHECK3-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP22]]) 2859 // CHECK3-NEXT: [[TMP23:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4 2860 // CHECK3-NEXT: [[TMP24:%.*]] = icmp ne i32 [[TMP23]], 0 2861 // CHECK3-NEXT: br i1 [[TMP24]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]] 2862 // CHECK3: .omp.final.then: 2863 // CHECK3-NEXT: [[TMP25:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 2864 // CHECK3-NEXT: [[SUB6:%.*]] = sub nsw i32 [[TMP25]], 0 2865 // CHECK3-NEXT: [[DIV7:%.*]] = sdiv i32 [[SUB6]], 1 2866 // CHECK3-NEXT: [[MUL:%.*]] = mul nsw i32 [[DIV7]], 1 2867 // CHECK3-NEXT: [[ADD8:%.*]] = add nsw i32 0, [[MUL]] 2868 // CHECK3-NEXT: store i32 [[ADD8]], i32* [[I3]], align 4 2869 // CHECK3-NEXT: br label [[DOTOMP_FINAL_DONE]] 2870 // CHECK3: .omp.final.done: 2871 // CHECK3-NEXT: br label [[OMP_PRECOND_END]] 2872 // CHECK3: omp.precond.end: 2873 // CHECK3-NEXT: ret void 2874 // 2875 // 2876 // CHECK3-LABEL: define {{[^@]+}}@.omp_outlined..1 2877 // CHECK3-SAME: (i32* noalias noundef [[DOTGLOBAL_TID_:%.*]], i32* noalias noundef [[DOTBOUND_TID_:%.*]], i32 noundef [[DOTPREVIOUS_LB_:%.*]], i32 noundef [[DOTPREVIOUS_UB_:%.*]], i32* noundef nonnull align 4 dereferenceable(4) [[N:%.*]], double** noundef nonnull align 4 dereferenceable(4) [[A:%.*]], double** noundef nonnull align 4 dereferenceable(4) [[B:%.*]], double** noundef nonnull align 4 dereferenceable(4) [[C:%.*]]) #[[ATTR2]] { 2878 // CHECK3-NEXT: entry: 2879 // CHECK3-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4 2880 // CHECK3-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4 2881 // CHECK3-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i32, align 4 2882 // CHECK3-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i32, align 4 2883 // CHECK3-NEXT: [[N_ADDR:%.*]] = alloca i32*, align 4 2884 // CHECK3-NEXT: [[A_ADDR:%.*]] = alloca double**, align 4 2885 // CHECK3-NEXT: [[B_ADDR:%.*]] = alloca double**, align 4 2886 // CHECK3-NEXT: [[C_ADDR:%.*]] = alloca double**, align 4 2887 // CHECK3-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 2888 // CHECK3-NEXT: [[TMP:%.*]] = alloca i32, align 4 2889 // CHECK3-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 2890 // CHECK3-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 2891 // CHECK3-NEXT: [[I:%.*]] = alloca i32, align 4 2892 // CHECK3-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 2893 // CHECK3-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4 2894 // CHECK3-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 2895 // CHECK3-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 2896 // CHECK3-NEXT: [[I3:%.*]] = alloca i32, align 4 2897 // CHECK3-NEXT: [[REF_TMP:%.*]] = alloca [[CLASS_ANON_0:%.*]], align 4 2898 // CHECK3-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 4 2899 // CHECK3-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 4 2900 // CHECK3-NEXT: store i32 [[DOTPREVIOUS_LB_]], i32* [[DOTPREVIOUS_LB__ADDR]], align 4 2901 // CHECK3-NEXT: store i32 [[DOTPREVIOUS_UB_]], i32* [[DOTPREVIOUS_UB__ADDR]], align 4 2902 // CHECK3-NEXT: store i32* [[N]], i32** [[N_ADDR]], align 4 2903 // CHECK3-NEXT: store double** [[A]], double*** [[A_ADDR]], align 4 2904 // CHECK3-NEXT: store double** [[B]], double*** [[B_ADDR]], align 4 2905 // CHECK3-NEXT: store double** [[C]], double*** [[C_ADDR]], align 4 2906 // CHECK3-NEXT: [[TMP0:%.*]] = load i32*, i32** [[N_ADDR]], align 4 2907 // CHECK3-NEXT: [[TMP1:%.*]] = load double**, double*** [[A_ADDR]], align 4 2908 // CHECK3-NEXT: [[TMP2:%.*]] = load double**, double*** [[B_ADDR]], align 4 2909 // CHECK3-NEXT: [[TMP3:%.*]] = load double**, double*** [[C_ADDR]], align 4 2910 // CHECK3-NEXT: [[TMP4:%.*]] = load i32, i32* [[TMP0]], align 4 2911 // CHECK3-NEXT: store i32 [[TMP4]], i32* [[DOTCAPTURE_EXPR_]], align 4 2912 // CHECK3-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 2913 // CHECK3-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP5]], 0 2914 // CHECK3-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 2915 // CHECK3-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1 2916 // CHECK3-NEXT: store i32 [[SUB2]], i32* [[DOTCAPTURE_EXPR_1]], align 4 2917 // CHECK3-NEXT: store i32 0, i32* [[I]], align 4 2918 // CHECK3-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 2919 // CHECK3-NEXT: [[CMP:%.*]] = icmp slt i32 0, [[TMP6]] 2920 // CHECK3-NEXT: br i1 [[CMP]], label [[OMP_PRECOND_THEN:%.*]], label [[OMP_PRECOND_END:%.*]] 2921 // CHECK3: omp.precond.then: 2922 // CHECK3-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4 2923 // CHECK3-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 2924 // CHECK3-NEXT: store i32 [[TMP7]], i32* [[DOTOMP_UB]], align 4 2925 // CHECK3-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTPREVIOUS_LB__ADDR]], align 4 2926 // CHECK3-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTPREVIOUS_UB__ADDR]], align 4 2927 // CHECK3-NEXT: store i32 [[TMP8]], i32* [[DOTOMP_LB]], align 4 2928 // CHECK3-NEXT: store i32 [[TMP9]], i32* [[DOTOMP_UB]], align 4 2929 // CHECK3-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4 2930 // CHECK3-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4 2931 // CHECK3-NEXT: [[TMP10:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 4 2932 // CHECK3-NEXT: [[TMP11:%.*]] = load i32, i32* [[TMP10]], align 4 2933 // CHECK3-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB2:[0-9]+]], i32 [[TMP11]], i32 34, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1) 2934 // CHECK3-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 2935 // CHECK3-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 2936 // CHECK3-NEXT: [[CMP4:%.*]] = icmp sgt i32 [[TMP12]], [[TMP13]] 2937 // CHECK3-NEXT: br i1 [[CMP4]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] 2938 // CHECK3: cond.true: 2939 // CHECK3-NEXT: [[TMP14:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 2940 // CHECK3-NEXT: br label [[COND_END:%.*]] 2941 // CHECK3: cond.false: 2942 // CHECK3-NEXT: [[TMP15:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 2943 // CHECK3-NEXT: br label [[COND_END]] 2944 // CHECK3: cond.end: 2945 // CHECK3-NEXT: [[COND:%.*]] = phi i32 [ [[TMP14]], [[COND_TRUE]] ], [ [[TMP15]], [[COND_FALSE]] ] 2946 // CHECK3-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4 2947 // CHECK3-NEXT: [[TMP16:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4 2948 // CHECK3-NEXT: store i32 [[TMP16]], i32* [[DOTOMP_IV]], align 4 2949 // CHECK3-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] 2950 // CHECK3: omp.inner.for.cond: 2951 // CHECK3-NEXT: [[TMP17:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !15 2952 // CHECK3-NEXT: [[TMP18:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4, !llvm.access.group !15 2953 // CHECK3-NEXT: [[CMP5:%.*]] = icmp sle i32 [[TMP17]], [[TMP18]] 2954 // CHECK3-NEXT: br i1 [[CMP5]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] 2955 // CHECK3: omp.inner.for.body: 2956 // CHECK3-NEXT: [[TMP19:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !15 2957 // CHECK3-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP19]], 1 2958 // CHECK3-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] 2959 // CHECK3-NEXT: store i32 [[ADD]], i32* [[I3]], align 4, !llvm.access.group !15 2960 // CHECK3-NEXT: [[TMP20:%.*]] = load double*, double** [[TMP2]], align 4, !llvm.access.group !15 2961 // CHECK3-NEXT: [[TMP21:%.*]] = load i32, i32* [[I3]], align 4, !llvm.access.group !15 2962 // CHECK3-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds double, double* [[TMP20]], i32 [[TMP21]] 2963 // CHECK3-NEXT: [[TMP22:%.*]] = load double, double* [[ARRAYIDX]], align 4, !llvm.access.group !15 2964 // CHECK3-NEXT: [[TMP23:%.*]] = load double*, double** [[TMP3]], align 4, !llvm.access.group !15 2965 // CHECK3-NEXT: [[TMP24:%.*]] = load i32, i32* [[I3]], align 4, !llvm.access.group !15 2966 // CHECK3-NEXT: [[ARRAYIDX6:%.*]] = getelementptr inbounds double, double* [[TMP23]], i32 [[TMP24]] 2967 // CHECK3-NEXT: [[TMP25:%.*]] = load double, double* [[ARRAYIDX6]], align 4, !llvm.access.group !15 2968 // CHECK3-NEXT: [[ADD7:%.*]] = fadd double [[TMP22]], [[TMP25]] 2969 // CHECK3-NEXT: [[TMP26:%.*]] = load double*, double** [[TMP1]], align 4, !llvm.access.group !15 2970 // CHECK3-NEXT: [[TMP27:%.*]] = load i32, i32* [[I3]], align 4, !llvm.access.group !15 2971 // CHECK3-NEXT: [[ARRAYIDX8:%.*]] = getelementptr inbounds double, double* [[TMP26]], i32 [[TMP27]] 2972 // CHECK3-NEXT: store double [[ADD7]], double* [[ARRAYIDX8]], align 4, !llvm.access.group !15 2973 // CHECK3-NEXT: [[TMP28:%.*]] = getelementptr inbounds [[CLASS_ANON_0]], %class.anon.0* [[REF_TMP]], i32 0, i32 0 2974 // CHECK3-NEXT: store double** [[TMP1]], double*** [[TMP28]], align 4, !llvm.access.group !15 2975 // CHECK3-NEXT: [[TMP29:%.*]] = getelementptr inbounds [[CLASS_ANON_0]], %class.anon.0* [[REF_TMP]], i32 0, i32 1 2976 // CHECK3-NEXT: store i32* [[I3]], i32** [[TMP29]], align 4, !llvm.access.group !15 2977 // CHECK3-NEXT: [[TMP30:%.*]] = getelementptr inbounds [[CLASS_ANON_0]], %class.anon.0* [[REF_TMP]], i32 0, i32 2 2978 // CHECK3-NEXT: store double** [[TMP2]], double*** [[TMP30]], align 4, !llvm.access.group !15 2979 // CHECK3-NEXT: [[TMP31:%.*]] = getelementptr inbounds [[CLASS_ANON_0]], %class.anon.0* [[REF_TMP]], i32 0, i32 3 2980 // CHECK3-NEXT: store double** [[TMP3]], double*** [[TMP31]], align 4, !llvm.access.group !15 2981 // CHECK3-NEXT: call void @"_ZZZ4mainENK3$_0clEvENKUlvE_clEv"(%class.anon.0* noundef nonnull align 4 dereferenceable(16) [[REF_TMP]]), !llvm.access.group !15 2982 // CHECK3-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] 2983 // CHECK3: omp.body.continue: 2984 // CHECK3-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] 2985 // CHECK3: omp.inner.for.inc: 2986 // CHECK3-NEXT: [[TMP32:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !15 2987 // CHECK3-NEXT: [[ADD9:%.*]] = add nsw i32 [[TMP32]], 1 2988 // CHECK3-NEXT: store i32 [[ADD9]], i32* [[DOTOMP_IV]], align 4, !llvm.access.group !15 2989 // CHECK3-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP16:![0-9]+]] 2990 // CHECK3: omp.inner.for.end: 2991 // CHECK3-NEXT: br label [[OMP_LOOP_EXIT:%.*]] 2992 // CHECK3: omp.loop.exit: 2993 // CHECK3-NEXT: [[TMP33:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 4 2994 // CHECK3-NEXT: [[TMP34:%.*]] = load i32, i32* [[TMP33]], align 4 2995 // CHECK3-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP34]]) 2996 // CHECK3-NEXT: [[TMP35:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4 2997 // CHECK3-NEXT: [[TMP36:%.*]] = icmp ne i32 [[TMP35]], 0 2998 // CHECK3-NEXT: br i1 [[TMP36]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]] 2999 // CHECK3: .omp.final.then: 3000 // CHECK3-NEXT: [[TMP37:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 3001 // CHECK3-NEXT: [[SUB10:%.*]] = sub nsw i32 [[TMP37]], 0 3002 // CHECK3-NEXT: [[DIV11:%.*]] = sdiv i32 [[SUB10]], 1 3003 // CHECK3-NEXT: [[MUL12:%.*]] = mul nsw i32 [[DIV11]], 1 3004 // CHECK3-NEXT: [[ADD13:%.*]] = add nsw i32 0, [[MUL12]] 3005 // CHECK3-NEXT: store i32 [[ADD13]], i32* [[I3]], align 4 3006 // CHECK3-NEXT: br label [[DOTOMP_FINAL_DONE]] 3007 // CHECK3: .omp.final.done: 3008 // CHECK3-NEXT: br label [[OMP_PRECOND_END]] 3009 // CHECK3: omp.precond.end: 3010 // CHECK3-NEXT: ret void 3011 // 3012 // 3013 // CHECK3-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l159 3014 // CHECK3-SAME: (i32 noundef [[N:%.*]], double* noundef [[A:%.*]], double* noundef [[B:%.*]], double* noundef [[C:%.*]]) #[[ATTR2]] { 3015 // CHECK3-NEXT: entry: 3016 // CHECK3-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4 3017 // CHECK3-NEXT: [[A_ADDR:%.*]] = alloca double*, align 4 3018 // CHECK3-NEXT: [[B_ADDR:%.*]] = alloca double*, align 4 3019 // CHECK3-NEXT: [[C_ADDR:%.*]] = alloca double*, align 4 3020 // CHECK3-NEXT: store i32 [[N]], i32* [[N_ADDR]], align 4 3021 // CHECK3-NEXT: store double* [[A]], double** [[A_ADDR]], align 4 3022 // CHECK3-NEXT: store double* [[B]], double** [[B_ADDR]], align 4 3023 // CHECK3-NEXT: store double* [[C]], double** [[C_ADDR]], align 4 3024 // CHECK3-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 4, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32*, double**, double**, double**)* @.omp_outlined..2 to void (i32*, i32*, ...)*), i32* [[N_ADDR]], double** [[A_ADDR]], double** [[B_ADDR]], double** [[C_ADDR]]) 3025 // CHECK3-NEXT: ret void 3026 // 3027 // 3028 // CHECK3-LABEL: define {{[^@]+}}@.omp_outlined..2 3029 // CHECK3-SAME: (i32* noalias noundef [[DOTGLOBAL_TID_:%.*]], i32* noalias noundef [[DOTBOUND_TID_:%.*]], i32* noundef nonnull align 4 dereferenceable(4) [[N:%.*]], double** noundef nonnull align 4 dereferenceable(4) [[A:%.*]], double** noundef nonnull align 4 dereferenceable(4) [[B:%.*]], double** noundef nonnull align 4 dereferenceable(4) [[C:%.*]]) #[[ATTR2]] { 3030 // CHECK3-NEXT: entry: 3031 // CHECK3-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4 3032 // CHECK3-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4 3033 // CHECK3-NEXT: [[N_ADDR:%.*]] = alloca i32*, align 4 3034 // CHECK3-NEXT: [[A_ADDR:%.*]] = alloca double**, align 4 3035 // CHECK3-NEXT: [[B_ADDR:%.*]] = alloca double**, align 4 3036 // CHECK3-NEXT: [[C_ADDR:%.*]] = alloca double**, align 4 3037 // CHECK3-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 3038 // CHECK3-NEXT: [[TMP:%.*]] = alloca i32, align 4 3039 // CHECK3-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 3040 // CHECK3-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 3041 // CHECK3-NEXT: [[I:%.*]] = alloca i32, align 4 3042 // CHECK3-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4 3043 // CHECK3-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4 3044 // CHECK3-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 3045 // CHECK3-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 3046 // CHECK3-NEXT: [[I3:%.*]] = alloca i32, align 4 3047 // CHECK3-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 4 3048 // CHECK3-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 4 3049 // CHECK3-NEXT: store i32* [[N]], i32** [[N_ADDR]], align 4 3050 // CHECK3-NEXT: store double** [[A]], double*** [[A_ADDR]], align 4 3051 // CHECK3-NEXT: store double** [[B]], double*** [[B_ADDR]], align 4 3052 // CHECK3-NEXT: store double** [[C]], double*** [[C_ADDR]], align 4 3053 // CHECK3-NEXT: [[TMP0:%.*]] = load i32*, i32** [[N_ADDR]], align 4 3054 // CHECK3-NEXT: [[TMP1:%.*]] = load double**, double*** [[A_ADDR]], align 4 3055 // CHECK3-NEXT: [[TMP2:%.*]] = load double**, double*** [[B_ADDR]], align 4 3056 // CHECK3-NEXT: [[TMP3:%.*]] = load double**, double*** [[C_ADDR]], align 4 3057 // CHECK3-NEXT: [[TMP4:%.*]] = load i32, i32* [[TMP0]], align 4 3058 // CHECK3-NEXT: store i32 [[TMP4]], i32* [[DOTCAPTURE_EXPR_]], align 4 3059 // CHECK3-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 3060 // CHECK3-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP5]], 0 3061 // CHECK3-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 3062 // CHECK3-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1 3063 // CHECK3-NEXT: store i32 [[SUB2]], i32* [[DOTCAPTURE_EXPR_1]], align 4 3064 // CHECK3-NEXT: store i32 0, i32* [[I]], align 4 3065 // CHECK3-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 3066 // CHECK3-NEXT: [[CMP:%.*]] = icmp slt i32 0, [[TMP6]] 3067 // CHECK3-NEXT: br i1 [[CMP]], label [[OMP_PRECOND_THEN:%.*]], label [[OMP_PRECOND_END:%.*]] 3068 // CHECK3: omp.precond.then: 3069 // CHECK3-NEXT: store i32 0, i32* [[DOTOMP_COMB_LB]], align 4 3070 // CHECK3-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 3071 // CHECK3-NEXT: store i32 [[TMP7]], i32* [[DOTOMP_COMB_UB]], align 4 3072 // CHECK3-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4 3073 // CHECK3-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4 3074 // CHECK3-NEXT: [[TMP8:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 4 3075 // CHECK3-NEXT: [[TMP9:%.*]] = load i32, i32* [[TMP8]], align 4 3076 // CHECK3-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1]], i32 [[TMP9]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_COMB_LB]], i32* [[DOTOMP_COMB_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1) 3077 // CHECK3-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 3078 // CHECK3-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 3079 // CHECK3-NEXT: [[CMP4:%.*]] = icmp sgt i32 [[TMP10]], [[TMP11]] 3080 // CHECK3-NEXT: br i1 [[CMP4]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] 3081 // CHECK3: cond.true: 3082 // CHECK3-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 3083 // CHECK3-NEXT: br label [[COND_END:%.*]] 3084 // CHECK3: cond.false: 3085 // CHECK3-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 3086 // CHECK3-NEXT: br label [[COND_END]] 3087 // CHECK3: cond.end: 3088 // CHECK3-NEXT: [[COND:%.*]] = phi i32 [ [[TMP12]], [[COND_TRUE]] ], [ [[TMP13]], [[COND_FALSE]] ] 3089 // CHECK3-NEXT: store i32 [[COND]], i32* [[DOTOMP_COMB_UB]], align 4 3090 // CHECK3-NEXT: [[TMP14:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4 3091 // CHECK3-NEXT: store i32 [[TMP14]], i32* [[DOTOMP_IV]], align 4 3092 // CHECK3-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] 3093 // CHECK3: omp.inner.for.cond: 3094 // CHECK3-NEXT: [[TMP15:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !20 3095 // CHECK3-NEXT: [[TMP16:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4, !llvm.access.group !20 3096 // CHECK3-NEXT: [[CMP5:%.*]] = icmp sle i32 [[TMP15]], [[TMP16]] 3097 // CHECK3-NEXT: br i1 [[CMP5]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] 3098 // CHECK3: omp.inner.for.body: 3099 // CHECK3-NEXT: [[TMP17:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4, !llvm.access.group !20 3100 // CHECK3-NEXT: [[TMP18:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4, !llvm.access.group !20 3101 // CHECK3-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 6, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32, i32, i32*, double**, double**, double**)* @.omp_outlined..3 to void (i32*, i32*, ...)*), i32 [[TMP17]], i32 [[TMP18]], i32* [[TMP0]], double** [[TMP1]], double** [[TMP2]], double** [[TMP3]]), !llvm.access.group !20 3102 // CHECK3-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] 3103 // CHECK3: omp.inner.for.inc: 3104 // CHECK3-NEXT: [[TMP19:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !20 3105 // CHECK3-NEXT: [[TMP20:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4, !llvm.access.group !20 3106 // CHECK3-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP19]], [[TMP20]] 3107 // CHECK3-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4, !llvm.access.group !20 3108 // CHECK3-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP21:![0-9]+]] 3109 // CHECK3: omp.inner.for.end: 3110 // CHECK3-NEXT: br label [[OMP_LOOP_EXIT:%.*]] 3111 // CHECK3: omp.loop.exit: 3112 // CHECK3-NEXT: [[TMP21:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 4 3113 // CHECK3-NEXT: [[TMP22:%.*]] = load i32, i32* [[TMP21]], align 4 3114 // CHECK3-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP22]]) 3115 // CHECK3-NEXT: [[TMP23:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4 3116 // CHECK3-NEXT: [[TMP24:%.*]] = icmp ne i32 [[TMP23]], 0 3117 // CHECK3-NEXT: br i1 [[TMP24]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]] 3118 // CHECK3: .omp.final.then: 3119 // CHECK3-NEXT: [[TMP25:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 3120 // CHECK3-NEXT: [[SUB6:%.*]] = sub nsw i32 [[TMP25]], 0 3121 // CHECK3-NEXT: [[DIV7:%.*]] = sdiv i32 [[SUB6]], 1 3122 // CHECK3-NEXT: [[MUL:%.*]] = mul nsw i32 [[DIV7]], 1 3123 // CHECK3-NEXT: [[ADD8:%.*]] = add nsw i32 0, [[MUL]] 3124 // CHECK3-NEXT: store i32 [[ADD8]], i32* [[I3]], align 4 3125 // CHECK3-NEXT: br label [[DOTOMP_FINAL_DONE]] 3126 // CHECK3: .omp.final.done: 3127 // CHECK3-NEXT: br label [[OMP_PRECOND_END]] 3128 // CHECK3: omp.precond.end: 3129 // CHECK3-NEXT: ret void 3130 // 3131 // 3132 // CHECK3-LABEL: define {{[^@]+}}@.omp_outlined..3 3133 // CHECK3-SAME: (i32* noalias noundef [[DOTGLOBAL_TID_:%.*]], i32* noalias noundef [[DOTBOUND_TID_:%.*]], i32 noundef [[DOTPREVIOUS_LB_:%.*]], i32 noundef [[DOTPREVIOUS_UB_:%.*]], i32* noundef nonnull align 4 dereferenceable(4) [[N:%.*]], double** noundef nonnull align 4 dereferenceable(4) [[A:%.*]], double** noundef nonnull align 4 dereferenceable(4) [[B:%.*]], double** noundef nonnull align 4 dereferenceable(4) [[C:%.*]]) #[[ATTR2]] { 3134 // CHECK3-NEXT: entry: 3135 // CHECK3-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4 3136 // CHECK3-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4 3137 // CHECK3-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i32, align 4 3138 // CHECK3-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i32, align 4 3139 // CHECK3-NEXT: [[N_ADDR:%.*]] = alloca i32*, align 4 3140 // CHECK3-NEXT: [[A_ADDR:%.*]] = alloca double**, align 4 3141 // CHECK3-NEXT: [[B_ADDR:%.*]] = alloca double**, align 4 3142 // CHECK3-NEXT: [[C_ADDR:%.*]] = alloca double**, align 4 3143 // CHECK3-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 3144 // CHECK3-NEXT: [[TMP:%.*]] = alloca i32, align 4 3145 // CHECK3-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 3146 // CHECK3-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 3147 // CHECK3-NEXT: [[I:%.*]] = alloca i32, align 4 3148 // CHECK3-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 3149 // CHECK3-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4 3150 // CHECK3-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 3151 // CHECK3-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 3152 // CHECK3-NEXT: [[I3:%.*]] = alloca i32, align 4 3153 // CHECK3-NEXT: [[REF_TMP:%.*]] = alloca [[CLASS_ANON_1:%.*]], align 4 3154 // CHECK3-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 4 3155 // CHECK3-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 4 3156 // CHECK3-NEXT: store i32 [[DOTPREVIOUS_LB_]], i32* [[DOTPREVIOUS_LB__ADDR]], align 4 3157 // CHECK3-NEXT: store i32 [[DOTPREVIOUS_UB_]], i32* [[DOTPREVIOUS_UB__ADDR]], align 4 3158 // CHECK3-NEXT: store i32* [[N]], i32** [[N_ADDR]], align 4 3159 // CHECK3-NEXT: store double** [[A]], double*** [[A_ADDR]], align 4 3160 // CHECK3-NEXT: store double** [[B]], double*** [[B_ADDR]], align 4 3161 // CHECK3-NEXT: store double** [[C]], double*** [[C_ADDR]], align 4 3162 // CHECK3-NEXT: [[TMP0:%.*]] = load i32*, i32** [[N_ADDR]], align 4 3163 // CHECK3-NEXT: [[TMP1:%.*]] = load double**, double*** [[A_ADDR]], align 4 3164 // CHECK3-NEXT: [[TMP2:%.*]] = load double**, double*** [[B_ADDR]], align 4 3165 // CHECK3-NEXT: [[TMP3:%.*]] = load double**, double*** [[C_ADDR]], align 4 3166 // CHECK3-NEXT: [[TMP4:%.*]] = load i32, i32* [[TMP0]], align 4 3167 // CHECK3-NEXT: store i32 [[TMP4]], i32* [[DOTCAPTURE_EXPR_]], align 4 3168 // CHECK3-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 3169 // CHECK3-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP5]], 0 3170 // CHECK3-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 3171 // CHECK3-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1 3172 // CHECK3-NEXT: store i32 [[SUB2]], i32* [[DOTCAPTURE_EXPR_1]], align 4 3173 // CHECK3-NEXT: store i32 0, i32* [[I]], align 4 3174 // CHECK3-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 3175 // CHECK3-NEXT: [[CMP:%.*]] = icmp slt i32 0, [[TMP6]] 3176 // CHECK3-NEXT: br i1 [[CMP]], label [[OMP_PRECOND_THEN:%.*]], label [[OMP_PRECOND_END:%.*]] 3177 // CHECK3: omp.precond.then: 3178 // CHECK3-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4 3179 // CHECK3-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 3180 // CHECK3-NEXT: store i32 [[TMP7]], i32* [[DOTOMP_UB]], align 4 3181 // CHECK3-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTPREVIOUS_LB__ADDR]], align 4 3182 // CHECK3-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTPREVIOUS_UB__ADDR]], align 4 3183 // CHECK3-NEXT: store i32 [[TMP8]], i32* [[DOTOMP_LB]], align 4 3184 // CHECK3-NEXT: store i32 [[TMP9]], i32* [[DOTOMP_UB]], align 4 3185 // CHECK3-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4 3186 // CHECK3-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4 3187 // CHECK3-NEXT: [[TMP10:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 4 3188 // CHECK3-NEXT: [[TMP11:%.*]] = load i32, i32* [[TMP10]], align 4 3189 // CHECK3-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB2]], i32 [[TMP11]], i32 34, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1) 3190 // CHECK3-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 3191 // CHECK3-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 3192 // CHECK3-NEXT: [[CMP4:%.*]] = icmp sgt i32 [[TMP12]], [[TMP13]] 3193 // CHECK3-NEXT: br i1 [[CMP4]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] 3194 // CHECK3: cond.true: 3195 // CHECK3-NEXT: [[TMP14:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 3196 // CHECK3-NEXT: br label [[COND_END:%.*]] 3197 // CHECK3: cond.false: 3198 // CHECK3-NEXT: [[TMP15:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 3199 // CHECK3-NEXT: br label [[COND_END]] 3200 // CHECK3: cond.end: 3201 // CHECK3-NEXT: [[COND:%.*]] = phi i32 [ [[TMP14]], [[COND_TRUE]] ], [ [[TMP15]], [[COND_FALSE]] ] 3202 // CHECK3-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4 3203 // CHECK3-NEXT: [[TMP16:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4 3204 // CHECK3-NEXT: store i32 [[TMP16]], i32* [[DOTOMP_IV]], align 4 3205 // CHECK3-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] 3206 // CHECK3: omp.inner.for.cond: 3207 // CHECK3-NEXT: [[TMP17:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !23 3208 // CHECK3-NEXT: [[TMP18:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4, !llvm.access.group !23 3209 // CHECK3-NEXT: [[CMP5:%.*]] = icmp sle i32 [[TMP17]], [[TMP18]] 3210 // CHECK3-NEXT: br i1 [[CMP5]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] 3211 // CHECK3: omp.inner.for.body: 3212 // CHECK3-NEXT: [[TMP19:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !23 3213 // CHECK3-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP19]], 1 3214 // CHECK3-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] 3215 // CHECK3-NEXT: store i32 [[ADD]], i32* [[I3]], align 4, !llvm.access.group !23 3216 // CHECK3-NEXT: [[TMP20:%.*]] = load double*, double** [[TMP2]], align 4, !llvm.access.group !23 3217 // CHECK3-NEXT: [[TMP21:%.*]] = load i32, i32* [[I3]], align 4, !llvm.access.group !23 3218 // CHECK3-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds double, double* [[TMP20]], i32 [[TMP21]] 3219 // CHECK3-NEXT: [[TMP22:%.*]] = load double, double* [[ARRAYIDX]], align 4, !llvm.access.group !23 3220 // CHECK3-NEXT: [[TMP23:%.*]] = load double*, double** [[TMP3]], align 4, !llvm.access.group !23 3221 // CHECK3-NEXT: [[TMP24:%.*]] = load i32, i32* [[I3]], align 4, !llvm.access.group !23 3222 // CHECK3-NEXT: [[ARRAYIDX6:%.*]] = getelementptr inbounds double, double* [[TMP23]], i32 [[TMP24]] 3223 // CHECK3-NEXT: [[TMP25:%.*]] = load double, double* [[ARRAYIDX6]], align 4, !llvm.access.group !23 3224 // CHECK3-NEXT: [[ADD7:%.*]] = fadd double [[TMP22]], [[TMP25]] 3225 // CHECK3-NEXT: [[TMP26:%.*]] = load double*, double** [[TMP1]], align 4, !llvm.access.group !23 3226 // CHECK3-NEXT: [[TMP27:%.*]] = load i32, i32* [[I3]], align 4, !llvm.access.group !23 3227 // CHECK3-NEXT: [[ARRAYIDX8:%.*]] = getelementptr inbounds double, double* [[TMP26]], i32 [[TMP27]] 3228 // CHECK3-NEXT: store double [[ADD7]], double* [[ARRAYIDX8]], align 4, !llvm.access.group !23 3229 // CHECK3-NEXT: [[TMP28:%.*]] = getelementptr inbounds [[CLASS_ANON_1]], %class.anon.1* [[REF_TMP]], i32 0, i32 0 3230 // CHECK3-NEXT: store double** [[TMP1]], double*** [[TMP28]], align 4, !llvm.access.group !23 3231 // CHECK3-NEXT: [[TMP29:%.*]] = getelementptr inbounds [[CLASS_ANON_1]], %class.anon.1* [[REF_TMP]], i32 0, i32 1 3232 // CHECK3-NEXT: store i32* [[I3]], i32** [[TMP29]], align 4, !llvm.access.group !23 3233 // CHECK3-NEXT: [[TMP30:%.*]] = getelementptr inbounds [[CLASS_ANON_1]], %class.anon.1* [[REF_TMP]], i32 0, i32 2 3234 // CHECK3-NEXT: store double** [[TMP2]], double*** [[TMP30]], align 4, !llvm.access.group !23 3235 // CHECK3-NEXT: [[TMP31:%.*]] = getelementptr inbounds [[CLASS_ANON_1]], %class.anon.1* [[REF_TMP]], i32 0, i32 3 3236 // CHECK3-NEXT: store double** [[TMP3]], double*** [[TMP31]], align 4, !llvm.access.group !23 3237 // CHECK3-NEXT: call void @"_ZZZ4mainENK3$_0clEvENKUlvE0_clEv"(%class.anon.1* noundef nonnull align 4 dereferenceable(16) [[REF_TMP]]), !llvm.access.group !23 3238 // CHECK3-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] 3239 // CHECK3: omp.body.continue: 3240 // CHECK3-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] 3241 // CHECK3: omp.inner.for.inc: 3242 // CHECK3-NEXT: [[TMP32:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !23 3243 // CHECK3-NEXT: [[ADD9:%.*]] = add nsw i32 [[TMP32]], 1 3244 // CHECK3-NEXT: store i32 [[ADD9]], i32* [[DOTOMP_IV]], align 4, !llvm.access.group !23 3245 // CHECK3-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP24:![0-9]+]] 3246 // CHECK3: omp.inner.for.end: 3247 // CHECK3-NEXT: br label [[OMP_LOOP_EXIT:%.*]] 3248 // CHECK3: omp.loop.exit: 3249 // CHECK3-NEXT: [[TMP33:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 4 3250 // CHECK3-NEXT: [[TMP34:%.*]] = load i32, i32* [[TMP33]], align 4 3251 // CHECK3-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP34]]) 3252 // CHECK3-NEXT: [[TMP35:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4 3253 // CHECK3-NEXT: [[TMP36:%.*]] = icmp ne i32 [[TMP35]], 0 3254 // CHECK3-NEXT: br i1 [[TMP36]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]] 3255 // CHECK3: .omp.final.then: 3256 // CHECK3-NEXT: [[TMP37:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 3257 // CHECK3-NEXT: [[SUB10:%.*]] = sub nsw i32 [[TMP37]], 0 3258 // CHECK3-NEXT: [[DIV11:%.*]] = sdiv i32 [[SUB10]], 1 3259 // CHECK3-NEXT: [[MUL12:%.*]] = mul nsw i32 [[DIV11]], 1 3260 // CHECK3-NEXT: [[ADD13:%.*]] = add nsw i32 0, [[MUL12]] 3261 // CHECK3-NEXT: store i32 [[ADD13]], i32* [[I3]], align 4 3262 // CHECK3-NEXT: br label [[DOTOMP_FINAL_DONE]] 3263 // CHECK3: .omp.final.done: 3264 // CHECK3-NEXT: br label [[OMP_PRECOND_END]] 3265 // CHECK3: omp.precond.end: 3266 // CHECK3-NEXT: ret void 3267 // 3268 // 3269 // CHECK3-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l201 3270 // CHECK3-SAME: (i32 noundef [[CH:%.*]], i32 noundef [[N:%.*]], double* noundef [[A:%.*]], double* noundef [[B:%.*]], double* noundef [[C:%.*]]) #[[ATTR2]] { 3271 // CHECK3-NEXT: entry: 3272 // CHECK3-NEXT: [[CH_ADDR:%.*]] = alloca i32, align 4 3273 // CHECK3-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4 3274 // CHECK3-NEXT: [[A_ADDR:%.*]] = alloca double*, align 4 3275 // CHECK3-NEXT: [[B_ADDR:%.*]] = alloca double*, align 4 3276 // CHECK3-NEXT: [[C_ADDR:%.*]] = alloca double*, align 4 3277 // CHECK3-NEXT: store i32 [[CH]], i32* [[CH_ADDR]], align 4 3278 // CHECK3-NEXT: store i32 [[N]], i32* [[N_ADDR]], align 4 3279 // CHECK3-NEXT: store double* [[A]], double** [[A_ADDR]], align 4 3280 // CHECK3-NEXT: store double* [[B]], double** [[B_ADDR]], align 4 3281 // CHECK3-NEXT: store double* [[C]], double** [[C_ADDR]], align 4 3282 // CHECK3-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 5, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32*, i32*, double**, double**, double**)* @.omp_outlined..6 to void (i32*, i32*, ...)*), i32* [[CH_ADDR]], i32* [[N_ADDR]], double** [[A_ADDR]], double** [[B_ADDR]], double** [[C_ADDR]]) 3283 // CHECK3-NEXT: ret void 3284 // 3285 // 3286 // CHECK3-LABEL: define {{[^@]+}}@.omp_outlined..6 3287 // CHECK3-SAME: (i32* noalias noundef [[DOTGLOBAL_TID_:%.*]], i32* noalias noundef [[DOTBOUND_TID_:%.*]], i32* noundef nonnull align 4 dereferenceable(4) [[CH:%.*]], i32* noundef nonnull align 4 dereferenceable(4) [[N:%.*]], double** noundef nonnull align 4 dereferenceable(4) [[A:%.*]], double** noundef nonnull align 4 dereferenceable(4) [[B:%.*]], double** noundef nonnull align 4 dereferenceable(4) [[C:%.*]]) #[[ATTR2]] { 3288 // CHECK3-NEXT: entry: 3289 // CHECK3-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4 3290 // CHECK3-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4 3291 // CHECK3-NEXT: [[CH_ADDR:%.*]] = alloca i32*, align 4 3292 // CHECK3-NEXT: [[N_ADDR:%.*]] = alloca i32*, align 4 3293 // CHECK3-NEXT: [[A_ADDR:%.*]] = alloca double**, align 4 3294 // CHECK3-NEXT: [[B_ADDR:%.*]] = alloca double**, align 4 3295 // CHECK3-NEXT: [[C_ADDR:%.*]] = alloca double**, align 4 3296 // CHECK3-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 3297 // CHECK3-NEXT: [[TMP:%.*]] = alloca i32, align 4 3298 // CHECK3-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 3299 // CHECK3-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 3300 // CHECK3-NEXT: [[I:%.*]] = alloca i32, align 4 3301 // CHECK3-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4 3302 // CHECK3-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4 3303 // CHECK3-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 3304 // CHECK3-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 3305 // CHECK3-NEXT: [[I3:%.*]] = alloca i32, align 4 3306 // CHECK3-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 4 3307 // CHECK3-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 4 3308 // CHECK3-NEXT: store i32* [[CH]], i32** [[CH_ADDR]], align 4 3309 // CHECK3-NEXT: store i32* [[N]], i32** [[N_ADDR]], align 4 3310 // CHECK3-NEXT: store double** [[A]], double*** [[A_ADDR]], align 4 3311 // CHECK3-NEXT: store double** [[B]], double*** [[B_ADDR]], align 4 3312 // CHECK3-NEXT: store double** [[C]], double*** [[C_ADDR]], align 4 3313 // CHECK3-NEXT: [[TMP0:%.*]] = load i32*, i32** [[CH_ADDR]], align 4 3314 // CHECK3-NEXT: [[TMP1:%.*]] = load i32*, i32** [[N_ADDR]], align 4 3315 // CHECK3-NEXT: [[TMP2:%.*]] = load double**, double*** [[A_ADDR]], align 4 3316 // CHECK3-NEXT: [[TMP3:%.*]] = load double**, double*** [[B_ADDR]], align 4 3317 // CHECK3-NEXT: [[TMP4:%.*]] = load double**, double*** [[C_ADDR]], align 4 3318 // CHECK3-NEXT: [[TMP5:%.*]] = load i32, i32* [[TMP1]], align 4 3319 // CHECK3-NEXT: store i32 [[TMP5]], i32* [[DOTCAPTURE_EXPR_]], align 4 3320 // CHECK3-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 3321 // CHECK3-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP6]], 0 3322 // CHECK3-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 3323 // CHECK3-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1 3324 // CHECK3-NEXT: store i32 [[SUB2]], i32* [[DOTCAPTURE_EXPR_1]], align 4 3325 // CHECK3-NEXT: store i32 0, i32* [[I]], align 4 3326 // CHECK3-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 3327 // CHECK3-NEXT: [[CMP:%.*]] = icmp slt i32 0, [[TMP7]] 3328 // CHECK3-NEXT: br i1 [[CMP]], label [[OMP_PRECOND_THEN:%.*]], label [[OMP_PRECOND_END:%.*]] 3329 // CHECK3: omp.precond.then: 3330 // CHECK3-NEXT: store i32 0, i32* [[DOTOMP_COMB_LB]], align 4 3331 // CHECK3-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 3332 // CHECK3-NEXT: store i32 [[TMP8]], i32* [[DOTOMP_COMB_UB]], align 4 3333 // CHECK3-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4 3334 // CHECK3-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4 3335 // CHECK3-NEXT: [[TMP9:%.*]] = load i32, i32* [[TMP0]], align 4 3336 // CHECK3-NEXT: [[TMP10:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 4 3337 // CHECK3-NEXT: [[TMP11:%.*]] = load i32, i32* [[TMP10]], align 4 3338 // CHECK3-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1]], i32 [[TMP11]], i32 91, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_COMB_LB]], i32* [[DOTOMP_COMB_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 [[TMP9]]) 3339 // CHECK3-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 3340 // CHECK3-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 3341 // CHECK3-NEXT: [[CMP4:%.*]] = icmp sgt i32 [[TMP12]], [[TMP13]] 3342 // CHECK3-NEXT: br i1 [[CMP4]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] 3343 // CHECK3: cond.true: 3344 // CHECK3-NEXT: [[TMP14:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 3345 // CHECK3-NEXT: br label [[COND_END:%.*]] 3346 // CHECK3: cond.false: 3347 // CHECK3-NEXT: [[TMP15:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 3348 // CHECK3-NEXT: br label [[COND_END]] 3349 // CHECK3: cond.end: 3350 // CHECK3-NEXT: [[COND:%.*]] = phi i32 [ [[TMP14]], [[COND_TRUE]] ], [ [[TMP15]], [[COND_FALSE]] ] 3351 // CHECK3-NEXT: store i32 [[COND]], i32* [[DOTOMP_COMB_UB]], align 4 3352 // CHECK3-NEXT: [[TMP16:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4 3353 // CHECK3-NEXT: store i32 [[TMP16]], i32* [[DOTOMP_IV]], align 4 3354 // CHECK3-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] 3355 // CHECK3: omp.inner.for.cond: 3356 // CHECK3-NEXT: [[TMP17:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !26 3357 // CHECK3-NEXT: [[TMP18:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4, !llvm.access.group !26 3358 // CHECK3-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP18]], 1 3359 // CHECK3-NEXT: [[CMP5:%.*]] = icmp slt i32 [[TMP17]], [[ADD]] 3360 // CHECK3-NEXT: br i1 [[CMP5]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] 3361 // CHECK3: omp.inner.for.body: 3362 // CHECK3-NEXT: [[TMP19:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4, !llvm.access.group !26 3363 // CHECK3-NEXT: [[TMP20:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4, !llvm.access.group !26 3364 // CHECK3-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 6, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32, i32, i32*, double**, double**, double**)* @.omp_outlined..7 to void (i32*, i32*, ...)*), i32 [[TMP19]], i32 [[TMP20]], i32* [[TMP1]], double** [[TMP2]], double** [[TMP3]], double** [[TMP4]]), !llvm.access.group !26 3365 // CHECK3-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] 3366 // CHECK3: omp.inner.for.inc: 3367 // CHECK3-NEXT: [[TMP21:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !26 3368 // CHECK3-NEXT: [[TMP22:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4, !llvm.access.group !26 3369 // CHECK3-NEXT: [[ADD6:%.*]] = add nsw i32 [[TMP21]], [[TMP22]] 3370 // CHECK3-NEXT: store i32 [[ADD6]], i32* [[DOTOMP_IV]], align 4, !llvm.access.group !26 3371 // CHECK3-NEXT: [[TMP23:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4, !llvm.access.group !26 3372 // CHECK3-NEXT: [[TMP24:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4, !llvm.access.group !26 3373 // CHECK3-NEXT: [[ADD7:%.*]] = add nsw i32 [[TMP23]], [[TMP24]] 3374 // CHECK3-NEXT: store i32 [[ADD7]], i32* [[DOTOMP_COMB_LB]], align 4, !llvm.access.group !26 3375 // CHECK3-NEXT: [[TMP25:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4, !llvm.access.group !26 3376 // CHECK3-NEXT: [[TMP26:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4, !llvm.access.group !26 3377 // CHECK3-NEXT: [[ADD8:%.*]] = add nsw i32 [[TMP25]], [[TMP26]] 3378 // CHECK3-NEXT: store i32 [[ADD8]], i32* [[DOTOMP_COMB_UB]], align 4, !llvm.access.group !26 3379 // CHECK3-NEXT: [[TMP27:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4, !llvm.access.group !26 3380 // CHECK3-NEXT: [[TMP28:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4, !llvm.access.group !26 3381 // CHECK3-NEXT: [[CMP9:%.*]] = icmp sgt i32 [[TMP27]], [[TMP28]] 3382 // CHECK3-NEXT: br i1 [[CMP9]], label [[COND_TRUE10:%.*]], label [[COND_FALSE11:%.*]] 3383 // CHECK3: cond.true10: 3384 // CHECK3-NEXT: [[TMP29:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4, !llvm.access.group !26 3385 // CHECK3-NEXT: br label [[COND_END12:%.*]] 3386 // CHECK3: cond.false11: 3387 // CHECK3-NEXT: [[TMP30:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4, !llvm.access.group !26 3388 // CHECK3-NEXT: br label [[COND_END12]] 3389 // CHECK3: cond.end12: 3390 // CHECK3-NEXT: [[COND13:%.*]] = phi i32 [ [[TMP29]], [[COND_TRUE10]] ], [ [[TMP30]], [[COND_FALSE11]] ] 3391 // CHECK3-NEXT: store i32 [[COND13]], i32* [[DOTOMP_COMB_UB]], align 4, !llvm.access.group !26 3392 // CHECK3-NEXT: [[TMP31:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4, !llvm.access.group !26 3393 // CHECK3-NEXT: store i32 [[TMP31]], i32* [[DOTOMP_IV]], align 4, !llvm.access.group !26 3394 // CHECK3-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP27:![0-9]+]] 3395 // CHECK3: omp.inner.for.end: 3396 // CHECK3-NEXT: br label [[OMP_LOOP_EXIT:%.*]] 3397 // CHECK3: omp.loop.exit: 3398 // CHECK3-NEXT: [[TMP32:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 4 3399 // CHECK3-NEXT: [[TMP33:%.*]] = load i32, i32* [[TMP32]], align 4 3400 // CHECK3-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP33]]) 3401 // CHECK3-NEXT: [[TMP34:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4 3402 // CHECK3-NEXT: [[TMP35:%.*]] = icmp ne i32 [[TMP34]], 0 3403 // CHECK3-NEXT: br i1 [[TMP35]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]] 3404 // CHECK3: .omp.final.then: 3405 // CHECK3-NEXT: [[TMP36:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 3406 // CHECK3-NEXT: [[SUB14:%.*]] = sub nsw i32 [[TMP36]], 0 3407 // CHECK3-NEXT: [[DIV15:%.*]] = sdiv i32 [[SUB14]], 1 3408 // CHECK3-NEXT: [[MUL:%.*]] = mul nsw i32 [[DIV15]], 1 3409 // CHECK3-NEXT: [[ADD16:%.*]] = add nsw i32 0, [[MUL]] 3410 // CHECK3-NEXT: store i32 [[ADD16]], i32* [[I3]], align 4 3411 // CHECK3-NEXT: br label [[DOTOMP_FINAL_DONE]] 3412 // CHECK3: .omp.final.done: 3413 // CHECK3-NEXT: br label [[OMP_PRECOND_END]] 3414 // CHECK3: omp.precond.end: 3415 // CHECK3-NEXT: ret void 3416 // 3417 // 3418 // CHECK3-LABEL: define {{[^@]+}}@.omp_outlined..7 3419 // CHECK3-SAME: (i32* noalias noundef [[DOTGLOBAL_TID_:%.*]], i32* noalias noundef [[DOTBOUND_TID_:%.*]], i32 noundef [[DOTPREVIOUS_LB_:%.*]], i32 noundef [[DOTPREVIOUS_UB_:%.*]], i32* noundef nonnull align 4 dereferenceable(4) [[N:%.*]], double** noundef nonnull align 4 dereferenceable(4) [[A:%.*]], double** noundef nonnull align 4 dereferenceable(4) [[B:%.*]], double** noundef nonnull align 4 dereferenceable(4) [[C:%.*]]) #[[ATTR2]] { 3420 // CHECK3-NEXT: entry: 3421 // CHECK3-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4 3422 // CHECK3-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4 3423 // CHECK3-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i32, align 4 3424 // CHECK3-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i32, align 4 3425 // CHECK3-NEXT: [[N_ADDR:%.*]] = alloca i32*, align 4 3426 // CHECK3-NEXT: [[A_ADDR:%.*]] = alloca double**, align 4 3427 // CHECK3-NEXT: [[B_ADDR:%.*]] = alloca double**, align 4 3428 // CHECK3-NEXT: [[C_ADDR:%.*]] = alloca double**, align 4 3429 // CHECK3-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 3430 // CHECK3-NEXT: [[TMP:%.*]] = alloca i32, align 4 3431 // CHECK3-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 3432 // CHECK3-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 3433 // CHECK3-NEXT: [[I:%.*]] = alloca i32, align 4 3434 // CHECK3-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 3435 // CHECK3-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4 3436 // CHECK3-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 3437 // CHECK3-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 3438 // CHECK3-NEXT: [[I3:%.*]] = alloca i32, align 4 3439 // CHECK3-NEXT: [[REF_TMP:%.*]] = alloca [[CLASS_ANON_2:%.*]], align 4 3440 // CHECK3-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 4 3441 // CHECK3-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 4 3442 // CHECK3-NEXT: store i32 [[DOTPREVIOUS_LB_]], i32* [[DOTPREVIOUS_LB__ADDR]], align 4 3443 // CHECK3-NEXT: store i32 [[DOTPREVIOUS_UB_]], i32* [[DOTPREVIOUS_UB__ADDR]], align 4 3444 // CHECK3-NEXT: store i32* [[N]], i32** [[N_ADDR]], align 4 3445 // CHECK3-NEXT: store double** [[A]], double*** [[A_ADDR]], align 4 3446 // CHECK3-NEXT: store double** [[B]], double*** [[B_ADDR]], align 4 3447 // CHECK3-NEXT: store double** [[C]], double*** [[C_ADDR]], align 4 3448 // CHECK3-NEXT: [[TMP0:%.*]] = load i32*, i32** [[N_ADDR]], align 4 3449 // CHECK3-NEXT: [[TMP1:%.*]] = load double**, double*** [[A_ADDR]], align 4 3450 // CHECK3-NEXT: [[TMP2:%.*]] = load double**, double*** [[B_ADDR]], align 4 3451 // CHECK3-NEXT: [[TMP3:%.*]] = load double**, double*** [[C_ADDR]], align 4 3452 // CHECK3-NEXT: [[TMP4:%.*]] = load i32, i32* [[TMP0]], align 4 3453 // CHECK3-NEXT: store i32 [[TMP4]], i32* [[DOTCAPTURE_EXPR_]], align 4 3454 // CHECK3-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 3455 // CHECK3-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP5]], 0 3456 // CHECK3-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 3457 // CHECK3-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1 3458 // CHECK3-NEXT: store i32 [[SUB2]], i32* [[DOTCAPTURE_EXPR_1]], align 4 3459 // CHECK3-NEXT: store i32 0, i32* [[I]], align 4 3460 // CHECK3-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 3461 // CHECK3-NEXT: [[CMP:%.*]] = icmp slt i32 0, [[TMP6]] 3462 // CHECK3-NEXT: br i1 [[CMP]], label [[OMP_PRECOND_THEN:%.*]], label [[OMP_PRECOND_END:%.*]] 3463 // CHECK3: omp.precond.then: 3464 // CHECK3-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4 3465 // CHECK3-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 3466 // CHECK3-NEXT: store i32 [[TMP7]], i32* [[DOTOMP_UB]], align 4 3467 // CHECK3-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTPREVIOUS_LB__ADDR]], align 4 3468 // CHECK3-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTPREVIOUS_UB__ADDR]], align 4 3469 // CHECK3-NEXT: store i32 [[TMP8]], i32* [[DOTOMP_LB]], align 4 3470 // CHECK3-NEXT: store i32 [[TMP9]], i32* [[DOTOMP_UB]], align 4 3471 // CHECK3-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4 3472 // CHECK3-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4 3473 // CHECK3-NEXT: [[TMP10:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 4 3474 // CHECK3-NEXT: [[TMP11:%.*]] = load i32, i32* [[TMP10]], align 4 3475 // CHECK3-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB2]], i32 [[TMP11]], i32 34, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1) 3476 // CHECK3-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 3477 // CHECK3-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 3478 // CHECK3-NEXT: [[CMP4:%.*]] = icmp sgt i32 [[TMP12]], [[TMP13]] 3479 // CHECK3-NEXT: br i1 [[CMP4]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] 3480 // CHECK3: cond.true: 3481 // CHECK3-NEXT: [[TMP14:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 3482 // CHECK3-NEXT: br label [[COND_END:%.*]] 3483 // CHECK3: cond.false: 3484 // CHECK3-NEXT: [[TMP15:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 3485 // CHECK3-NEXT: br label [[COND_END]] 3486 // CHECK3: cond.end: 3487 // CHECK3-NEXT: [[COND:%.*]] = phi i32 [ [[TMP14]], [[COND_TRUE]] ], [ [[TMP15]], [[COND_FALSE]] ] 3488 // CHECK3-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4 3489 // CHECK3-NEXT: [[TMP16:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4 3490 // CHECK3-NEXT: store i32 [[TMP16]], i32* [[DOTOMP_IV]], align 4 3491 // CHECK3-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] 3492 // CHECK3: omp.inner.for.cond: 3493 // CHECK3-NEXT: [[TMP17:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !29 3494 // CHECK3-NEXT: [[TMP18:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4, !llvm.access.group !29 3495 // CHECK3-NEXT: [[CMP5:%.*]] = icmp sle i32 [[TMP17]], [[TMP18]] 3496 // CHECK3-NEXT: br i1 [[CMP5]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] 3497 // CHECK3: omp.inner.for.body: 3498 // CHECK3-NEXT: [[TMP19:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !29 3499 // CHECK3-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP19]], 1 3500 // CHECK3-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] 3501 // CHECK3-NEXT: store i32 [[ADD]], i32* [[I3]], align 4, !llvm.access.group !29 3502 // CHECK3-NEXT: [[TMP20:%.*]] = load double*, double** [[TMP2]], align 4, !llvm.access.group !29 3503 // CHECK3-NEXT: [[TMP21:%.*]] = load i32, i32* [[I3]], align 4, !llvm.access.group !29 3504 // CHECK3-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds double, double* [[TMP20]], i32 [[TMP21]] 3505 // CHECK3-NEXT: [[TMP22:%.*]] = load double, double* [[ARRAYIDX]], align 4, !llvm.access.group !29 3506 // CHECK3-NEXT: [[TMP23:%.*]] = load double*, double** [[TMP3]], align 4, !llvm.access.group !29 3507 // CHECK3-NEXT: [[TMP24:%.*]] = load i32, i32* [[I3]], align 4, !llvm.access.group !29 3508 // CHECK3-NEXT: [[ARRAYIDX6:%.*]] = getelementptr inbounds double, double* [[TMP23]], i32 [[TMP24]] 3509 // CHECK3-NEXT: [[TMP25:%.*]] = load double, double* [[ARRAYIDX6]], align 4, !llvm.access.group !29 3510 // CHECK3-NEXT: [[ADD7:%.*]] = fadd double [[TMP22]], [[TMP25]] 3511 // CHECK3-NEXT: [[TMP26:%.*]] = load double*, double** [[TMP1]], align 4, !llvm.access.group !29 3512 // CHECK3-NEXT: [[TMP27:%.*]] = load i32, i32* [[I3]], align 4, !llvm.access.group !29 3513 // CHECK3-NEXT: [[ARRAYIDX8:%.*]] = getelementptr inbounds double, double* [[TMP26]], i32 [[TMP27]] 3514 // CHECK3-NEXT: store double [[ADD7]], double* [[ARRAYIDX8]], align 4, !llvm.access.group !29 3515 // CHECK3-NEXT: [[TMP28:%.*]] = getelementptr inbounds [[CLASS_ANON_2]], %class.anon.2* [[REF_TMP]], i32 0, i32 0 3516 // CHECK3-NEXT: store double** [[TMP1]], double*** [[TMP28]], align 4, !llvm.access.group !29 3517 // CHECK3-NEXT: [[TMP29:%.*]] = getelementptr inbounds [[CLASS_ANON_2]], %class.anon.2* [[REF_TMP]], i32 0, i32 1 3518 // CHECK3-NEXT: store i32* [[I3]], i32** [[TMP29]], align 4, !llvm.access.group !29 3519 // CHECK3-NEXT: [[TMP30:%.*]] = getelementptr inbounds [[CLASS_ANON_2]], %class.anon.2* [[REF_TMP]], i32 0, i32 2 3520 // CHECK3-NEXT: store double** [[TMP2]], double*** [[TMP30]], align 4, !llvm.access.group !29 3521 // CHECK3-NEXT: [[TMP31:%.*]] = getelementptr inbounds [[CLASS_ANON_2]], %class.anon.2* [[REF_TMP]], i32 0, i32 3 3522 // CHECK3-NEXT: store double** [[TMP3]], double*** [[TMP31]], align 4, !llvm.access.group !29 3523 // CHECK3-NEXT: call void @"_ZZZ4mainENK3$_0clEvENKUlvE1_clEv"(%class.anon.2* noundef nonnull align 4 dereferenceable(16) [[REF_TMP]]), !llvm.access.group !29 3524 // CHECK3-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] 3525 // CHECK3: omp.body.continue: 3526 // CHECK3-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] 3527 // CHECK3: omp.inner.for.inc: 3528 // CHECK3-NEXT: [[TMP32:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !29 3529 // CHECK3-NEXT: [[ADD9:%.*]] = add nsw i32 [[TMP32]], 1 3530 // CHECK3-NEXT: store i32 [[ADD9]], i32* [[DOTOMP_IV]], align 4, !llvm.access.group !29 3531 // CHECK3-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP30:![0-9]+]] 3532 // CHECK3: omp.inner.for.end: 3533 // CHECK3-NEXT: br label [[OMP_LOOP_EXIT:%.*]] 3534 // CHECK3: omp.loop.exit: 3535 // CHECK3-NEXT: [[TMP33:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 4 3536 // CHECK3-NEXT: [[TMP34:%.*]] = load i32, i32* [[TMP33]], align 4 3537 // CHECK3-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP34]]) 3538 // CHECK3-NEXT: [[TMP35:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4 3539 // CHECK3-NEXT: [[TMP36:%.*]] = icmp ne i32 [[TMP35]], 0 3540 // CHECK3-NEXT: br i1 [[TMP36]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]] 3541 // CHECK3: .omp.final.then: 3542 // CHECK3-NEXT: [[TMP37:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 3543 // CHECK3-NEXT: [[SUB10:%.*]] = sub nsw i32 [[TMP37]], 0 3544 // CHECK3-NEXT: [[DIV11:%.*]] = sdiv i32 [[SUB10]], 1 3545 // CHECK3-NEXT: [[MUL12:%.*]] = mul nsw i32 [[DIV11]], 1 3546 // CHECK3-NEXT: [[ADD13:%.*]] = add nsw i32 0, [[MUL12]] 3547 // CHECK3-NEXT: store i32 [[ADD13]], i32* [[I3]], align 4 3548 // CHECK3-NEXT: br label [[DOTOMP_FINAL_DONE]] 3549 // CHECK3: .omp.final.done: 3550 // CHECK3-NEXT: br label [[OMP_PRECOND_END]] 3551 // CHECK3: omp.precond.end: 3552 // CHECK3-NEXT: ret void 3553 // 3554 // 3555 // CHECK3-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l234 3556 // CHECK3-SAME: (i32 noundef [[N:%.*]], double* noundef [[A:%.*]], double* noundef [[B:%.*]], double* noundef [[C:%.*]]) #[[ATTR2]] { 3557 // CHECK3-NEXT: entry: 3558 // CHECK3-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4 3559 // CHECK3-NEXT: [[A_ADDR:%.*]] = alloca double*, align 4 3560 // CHECK3-NEXT: [[B_ADDR:%.*]] = alloca double*, align 4 3561 // CHECK3-NEXT: [[C_ADDR:%.*]] = alloca double*, align 4 3562 // CHECK3-NEXT: store i32 [[N]], i32* [[N_ADDR]], align 4 3563 // CHECK3-NEXT: store double* [[A]], double** [[A_ADDR]], align 4 3564 // CHECK3-NEXT: store double* [[B]], double** [[B_ADDR]], align 4 3565 // CHECK3-NEXT: store double* [[C]], double** [[C_ADDR]], align 4 3566 // CHECK3-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 4, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32*, double**, double**, double**)* @.omp_outlined..10 to void (i32*, i32*, ...)*), i32* [[N_ADDR]], double** [[A_ADDR]], double** [[B_ADDR]], double** [[C_ADDR]]) 3567 // CHECK3-NEXT: ret void 3568 // 3569 // 3570 // CHECK3-LABEL: define {{[^@]+}}@.omp_outlined..10 3571 // CHECK3-SAME: (i32* noalias noundef [[DOTGLOBAL_TID_:%.*]], i32* noalias noundef [[DOTBOUND_TID_:%.*]], i32* noundef nonnull align 4 dereferenceable(4) [[N:%.*]], double** noundef nonnull align 4 dereferenceable(4) [[A:%.*]], double** noundef nonnull align 4 dereferenceable(4) [[B:%.*]], double** noundef nonnull align 4 dereferenceable(4) [[C:%.*]]) #[[ATTR2]] { 3572 // CHECK3-NEXT: entry: 3573 // CHECK3-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4 3574 // CHECK3-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4 3575 // CHECK3-NEXT: [[N_ADDR:%.*]] = alloca i32*, align 4 3576 // CHECK3-NEXT: [[A_ADDR:%.*]] = alloca double**, align 4 3577 // CHECK3-NEXT: [[B_ADDR:%.*]] = alloca double**, align 4 3578 // CHECK3-NEXT: [[C_ADDR:%.*]] = alloca double**, align 4 3579 // CHECK3-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 3580 // CHECK3-NEXT: [[TMP:%.*]] = alloca i32, align 4 3581 // CHECK3-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 3582 // CHECK3-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 3583 // CHECK3-NEXT: [[I:%.*]] = alloca i32, align 4 3584 // CHECK3-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4 3585 // CHECK3-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4 3586 // CHECK3-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 3587 // CHECK3-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 3588 // CHECK3-NEXT: [[I3:%.*]] = alloca i32, align 4 3589 // CHECK3-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 4 3590 // CHECK3-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 4 3591 // CHECK3-NEXT: store i32* [[N]], i32** [[N_ADDR]], align 4 3592 // CHECK3-NEXT: store double** [[A]], double*** [[A_ADDR]], align 4 3593 // CHECK3-NEXT: store double** [[B]], double*** [[B_ADDR]], align 4 3594 // CHECK3-NEXT: store double** [[C]], double*** [[C_ADDR]], align 4 3595 // CHECK3-NEXT: [[TMP0:%.*]] = load i32*, i32** [[N_ADDR]], align 4 3596 // CHECK3-NEXT: [[TMP1:%.*]] = load double**, double*** [[A_ADDR]], align 4 3597 // CHECK3-NEXT: [[TMP2:%.*]] = load double**, double*** [[B_ADDR]], align 4 3598 // CHECK3-NEXT: [[TMP3:%.*]] = load double**, double*** [[C_ADDR]], align 4 3599 // CHECK3-NEXT: [[TMP4:%.*]] = load i32, i32* [[TMP0]], align 4 3600 // CHECK3-NEXT: store i32 [[TMP4]], i32* [[DOTCAPTURE_EXPR_]], align 4 3601 // CHECK3-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 3602 // CHECK3-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP5]], 0 3603 // CHECK3-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 3604 // CHECK3-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1 3605 // CHECK3-NEXT: store i32 [[SUB2]], i32* [[DOTCAPTURE_EXPR_1]], align 4 3606 // CHECK3-NEXT: store i32 0, i32* [[I]], align 4 3607 // CHECK3-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 3608 // CHECK3-NEXT: [[CMP:%.*]] = icmp slt i32 0, [[TMP6]] 3609 // CHECK3-NEXT: br i1 [[CMP]], label [[OMP_PRECOND_THEN:%.*]], label [[OMP_PRECOND_END:%.*]] 3610 // CHECK3: omp.precond.then: 3611 // CHECK3-NEXT: store i32 0, i32* [[DOTOMP_COMB_LB]], align 4 3612 // CHECK3-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 3613 // CHECK3-NEXT: store i32 [[TMP7]], i32* [[DOTOMP_COMB_UB]], align 4 3614 // CHECK3-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4 3615 // CHECK3-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4 3616 // CHECK3-NEXT: [[TMP8:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 4 3617 // CHECK3-NEXT: [[TMP9:%.*]] = load i32, i32* [[TMP8]], align 4 3618 // CHECK3-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1]], i32 [[TMP9]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_COMB_LB]], i32* [[DOTOMP_COMB_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1) 3619 // CHECK3-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 3620 // CHECK3-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 3621 // CHECK3-NEXT: [[CMP4:%.*]] = icmp sgt i32 [[TMP10]], [[TMP11]] 3622 // CHECK3-NEXT: br i1 [[CMP4]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] 3623 // CHECK3: cond.true: 3624 // CHECK3-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 3625 // CHECK3-NEXT: br label [[COND_END:%.*]] 3626 // CHECK3: cond.false: 3627 // CHECK3-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 3628 // CHECK3-NEXT: br label [[COND_END]] 3629 // CHECK3: cond.end: 3630 // CHECK3-NEXT: [[COND:%.*]] = phi i32 [ [[TMP12]], [[COND_TRUE]] ], [ [[TMP13]], [[COND_FALSE]] ] 3631 // CHECK3-NEXT: store i32 [[COND]], i32* [[DOTOMP_COMB_UB]], align 4 3632 // CHECK3-NEXT: [[TMP14:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4 3633 // CHECK3-NEXT: store i32 [[TMP14]], i32* [[DOTOMP_IV]], align 4 3634 // CHECK3-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] 3635 // CHECK3: omp.inner.for.cond: 3636 // CHECK3-NEXT: [[TMP15:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !32 3637 // CHECK3-NEXT: [[TMP16:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4, !llvm.access.group !32 3638 // CHECK3-NEXT: [[CMP5:%.*]] = icmp sle i32 [[TMP15]], [[TMP16]] 3639 // CHECK3-NEXT: br i1 [[CMP5]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] 3640 // CHECK3: omp.inner.for.body: 3641 // CHECK3-NEXT: [[TMP17:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4, !llvm.access.group !32 3642 // CHECK3-NEXT: [[TMP18:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4, !llvm.access.group !32 3643 // CHECK3-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 6, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32, i32, i32*, double**, double**, double**)* @.omp_outlined..11 to void (i32*, i32*, ...)*), i32 [[TMP17]], i32 [[TMP18]], i32* [[TMP0]], double** [[TMP1]], double** [[TMP2]], double** [[TMP3]]), !llvm.access.group !32 3644 // CHECK3-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] 3645 // CHECK3: omp.inner.for.inc: 3646 // CHECK3-NEXT: [[TMP19:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !32 3647 // CHECK3-NEXT: [[TMP20:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4, !llvm.access.group !32 3648 // CHECK3-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP19]], [[TMP20]] 3649 // CHECK3-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4, !llvm.access.group !32 3650 // CHECK3-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP33:![0-9]+]] 3651 // CHECK3: omp.inner.for.end: 3652 // CHECK3-NEXT: br label [[OMP_LOOP_EXIT:%.*]] 3653 // CHECK3: omp.loop.exit: 3654 // CHECK3-NEXT: [[TMP21:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 4 3655 // CHECK3-NEXT: [[TMP22:%.*]] = load i32, i32* [[TMP21]], align 4 3656 // CHECK3-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP22]]) 3657 // CHECK3-NEXT: [[TMP23:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4 3658 // CHECK3-NEXT: [[TMP24:%.*]] = icmp ne i32 [[TMP23]], 0 3659 // CHECK3-NEXT: br i1 [[TMP24]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]] 3660 // CHECK3: .omp.final.then: 3661 // CHECK3-NEXT: [[TMP25:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 3662 // CHECK3-NEXT: [[SUB6:%.*]] = sub nsw i32 [[TMP25]], 0 3663 // CHECK3-NEXT: [[DIV7:%.*]] = sdiv i32 [[SUB6]], 1 3664 // CHECK3-NEXT: [[MUL:%.*]] = mul nsw i32 [[DIV7]], 1 3665 // CHECK3-NEXT: [[ADD8:%.*]] = add nsw i32 0, [[MUL]] 3666 // CHECK3-NEXT: store i32 [[ADD8]], i32* [[I3]], align 4 3667 // CHECK3-NEXT: br label [[DOTOMP_FINAL_DONE]] 3668 // CHECK3: .omp.final.done: 3669 // CHECK3-NEXT: br label [[OMP_PRECOND_END]] 3670 // CHECK3: omp.precond.end: 3671 // CHECK3-NEXT: ret void 3672 // 3673 // 3674 // CHECK3-LABEL: define {{[^@]+}}@.omp_outlined..11 3675 // CHECK3-SAME: (i32* noalias noundef [[DOTGLOBAL_TID_:%.*]], i32* noalias noundef [[DOTBOUND_TID_:%.*]], i32 noundef [[DOTPREVIOUS_LB_:%.*]], i32 noundef [[DOTPREVIOUS_UB_:%.*]], i32* noundef nonnull align 4 dereferenceable(4) [[N:%.*]], double** noundef nonnull align 4 dereferenceable(4) [[A:%.*]], double** noundef nonnull align 4 dereferenceable(4) [[B:%.*]], double** noundef nonnull align 4 dereferenceable(4) [[C:%.*]]) #[[ATTR2]] { 3676 // CHECK3-NEXT: entry: 3677 // CHECK3-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4 3678 // CHECK3-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4 3679 // CHECK3-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i32, align 4 3680 // CHECK3-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i32, align 4 3681 // CHECK3-NEXT: [[N_ADDR:%.*]] = alloca i32*, align 4 3682 // CHECK3-NEXT: [[A_ADDR:%.*]] = alloca double**, align 4 3683 // CHECK3-NEXT: [[B_ADDR:%.*]] = alloca double**, align 4 3684 // CHECK3-NEXT: [[C_ADDR:%.*]] = alloca double**, align 4 3685 // CHECK3-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 3686 // CHECK3-NEXT: [[TMP:%.*]] = alloca i32, align 4 3687 // CHECK3-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 3688 // CHECK3-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 3689 // CHECK3-NEXT: [[I:%.*]] = alloca i32, align 4 3690 // CHECK3-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 3691 // CHECK3-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4 3692 // CHECK3-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 3693 // CHECK3-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 3694 // CHECK3-NEXT: [[I3:%.*]] = alloca i32, align 4 3695 // CHECK3-NEXT: [[REF_TMP:%.*]] = alloca [[CLASS_ANON_3:%.*]], align 4 3696 // CHECK3-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 4 3697 // CHECK3-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 4 3698 // CHECK3-NEXT: store i32 [[DOTPREVIOUS_LB_]], i32* [[DOTPREVIOUS_LB__ADDR]], align 4 3699 // CHECK3-NEXT: store i32 [[DOTPREVIOUS_UB_]], i32* [[DOTPREVIOUS_UB__ADDR]], align 4 3700 // CHECK3-NEXT: store i32* [[N]], i32** [[N_ADDR]], align 4 3701 // CHECK3-NEXT: store double** [[A]], double*** [[A_ADDR]], align 4 3702 // CHECK3-NEXT: store double** [[B]], double*** [[B_ADDR]], align 4 3703 // CHECK3-NEXT: store double** [[C]], double*** [[C_ADDR]], align 4 3704 // CHECK3-NEXT: [[TMP0:%.*]] = load i32*, i32** [[N_ADDR]], align 4 3705 // CHECK3-NEXT: [[TMP1:%.*]] = load double**, double*** [[A_ADDR]], align 4 3706 // CHECK3-NEXT: [[TMP2:%.*]] = load double**, double*** [[B_ADDR]], align 4 3707 // CHECK3-NEXT: [[TMP3:%.*]] = load double**, double*** [[C_ADDR]], align 4 3708 // CHECK3-NEXT: [[TMP4:%.*]] = load i32, i32* [[TMP0]], align 4 3709 // CHECK3-NEXT: store i32 [[TMP4]], i32* [[DOTCAPTURE_EXPR_]], align 4 3710 // CHECK3-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 3711 // CHECK3-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP5]], 0 3712 // CHECK3-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 3713 // CHECK3-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1 3714 // CHECK3-NEXT: store i32 [[SUB2]], i32* [[DOTCAPTURE_EXPR_1]], align 4 3715 // CHECK3-NEXT: store i32 0, i32* [[I]], align 4 3716 // CHECK3-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 3717 // CHECK3-NEXT: [[CMP:%.*]] = icmp slt i32 0, [[TMP6]] 3718 // CHECK3-NEXT: br i1 [[CMP]], label [[OMP_PRECOND_THEN:%.*]], label [[OMP_PRECOND_END:%.*]] 3719 // CHECK3: omp.precond.then: 3720 // CHECK3-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4 3721 // CHECK3-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 3722 // CHECK3-NEXT: store i32 [[TMP7]], i32* [[DOTOMP_UB]], align 4 3723 // CHECK3-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTPREVIOUS_LB__ADDR]], align 4 3724 // CHECK3-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTPREVIOUS_UB__ADDR]], align 4 3725 // CHECK3-NEXT: store i32 [[TMP8]], i32* [[DOTOMP_LB]], align 4 3726 // CHECK3-NEXT: store i32 [[TMP9]], i32* [[DOTOMP_UB]], align 4 3727 // CHECK3-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4 3728 // CHECK3-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4 3729 // CHECK3-NEXT: [[TMP10:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 4 3730 // CHECK3-NEXT: [[TMP11:%.*]] = load i32, i32* [[TMP10]], align 4 3731 // CHECK3-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB2]], i32 [[TMP11]], i32 34, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1) 3732 // CHECK3-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 3733 // CHECK3-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 3734 // CHECK3-NEXT: [[CMP4:%.*]] = icmp sgt i32 [[TMP12]], [[TMP13]] 3735 // CHECK3-NEXT: br i1 [[CMP4]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] 3736 // CHECK3: cond.true: 3737 // CHECK3-NEXT: [[TMP14:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 3738 // CHECK3-NEXT: br label [[COND_END:%.*]] 3739 // CHECK3: cond.false: 3740 // CHECK3-NEXT: [[TMP15:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 3741 // CHECK3-NEXT: br label [[COND_END]] 3742 // CHECK3: cond.end: 3743 // CHECK3-NEXT: [[COND:%.*]] = phi i32 [ [[TMP14]], [[COND_TRUE]] ], [ [[TMP15]], [[COND_FALSE]] ] 3744 // CHECK3-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4 3745 // CHECK3-NEXT: [[TMP16:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4 3746 // CHECK3-NEXT: store i32 [[TMP16]], i32* [[DOTOMP_IV]], align 4 3747 // CHECK3-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] 3748 // CHECK3: omp.inner.for.cond: 3749 // CHECK3-NEXT: [[TMP17:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !35 3750 // CHECK3-NEXT: [[TMP18:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4, !llvm.access.group !35 3751 // CHECK3-NEXT: [[CMP5:%.*]] = icmp sle i32 [[TMP17]], [[TMP18]] 3752 // CHECK3-NEXT: br i1 [[CMP5]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] 3753 // CHECK3: omp.inner.for.body: 3754 // CHECK3-NEXT: [[TMP19:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !35 3755 // CHECK3-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP19]], 1 3756 // CHECK3-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] 3757 // CHECK3-NEXT: store i32 [[ADD]], i32* [[I3]], align 4, !llvm.access.group !35 3758 // CHECK3-NEXT: [[TMP20:%.*]] = load double*, double** [[TMP2]], align 4, !llvm.access.group !35 3759 // CHECK3-NEXT: [[TMP21:%.*]] = load i32, i32* [[I3]], align 4, !llvm.access.group !35 3760 // CHECK3-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds double, double* [[TMP20]], i32 [[TMP21]] 3761 // CHECK3-NEXT: [[TMP22:%.*]] = load double, double* [[ARRAYIDX]], align 4, !llvm.access.group !35 3762 // CHECK3-NEXT: [[TMP23:%.*]] = load double*, double** [[TMP3]], align 4, !llvm.access.group !35 3763 // CHECK3-NEXT: [[TMP24:%.*]] = load i32, i32* [[I3]], align 4, !llvm.access.group !35 3764 // CHECK3-NEXT: [[ARRAYIDX6:%.*]] = getelementptr inbounds double, double* [[TMP23]], i32 [[TMP24]] 3765 // CHECK3-NEXT: [[TMP25:%.*]] = load double, double* [[ARRAYIDX6]], align 4, !llvm.access.group !35 3766 // CHECK3-NEXT: [[ADD7:%.*]] = fadd double [[TMP22]], [[TMP25]] 3767 // CHECK3-NEXT: [[TMP26:%.*]] = load double*, double** [[TMP1]], align 4, !llvm.access.group !35 3768 // CHECK3-NEXT: [[TMP27:%.*]] = load i32, i32* [[I3]], align 4, !llvm.access.group !35 3769 // CHECK3-NEXT: [[ARRAYIDX8:%.*]] = getelementptr inbounds double, double* [[TMP26]], i32 [[TMP27]] 3770 // CHECK3-NEXT: store double [[ADD7]], double* [[ARRAYIDX8]], align 4, !llvm.access.group !35 3771 // CHECK3-NEXT: [[TMP28:%.*]] = getelementptr inbounds [[CLASS_ANON_3]], %class.anon.3* [[REF_TMP]], i32 0, i32 0 3772 // CHECK3-NEXT: store double** [[TMP1]], double*** [[TMP28]], align 4, !llvm.access.group !35 3773 // CHECK3-NEXT: [[TMP29:%.*]] = getelementptr inbounds [[CLASS_ANON_3]], %class.anon.3* [[REF_TMP]], i32 0, i32 1 3774 // CHECK3-NEXT: store i32* [[I3]], i32** [[TMP29]], align 4, !llvm.access.group !35 3775 // CHECK3-NEXT: [[TMP30:%.*]] = getelementptr inbounds [[CLASS_ANON_3]], %class.anon.3* [[REF_TMP]], i32 0, i32 2 3776 // CHECK3-NEXT: store double** [[TMP2]], double*** [[TMP30]], align 4, !llvm.access.group !35 3777 // CHECK3-NEXT: [[TMP31:%.*]] = getelementptr inbounds [[CLASS_ANON_3]], %class.anon.3* [[REF_TMP]], i32 0, i32 3 3778 // CHECK3-NEXT: store double** [[TMP3]], double*** [[TMP31]], align 4, !llvm.access.group !35 3779 // CHECK3-NEXT: call void @"_ZZZ4mainENK3$_0clEvENKUlvE2_clEv"(%class.anon.3* noundef nonnull align 4 dereferenceable(16) [[REF_TMP]]), !llvm.access.group !35 3780 // CHECK3-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] 3781 // CHECK3: omp.body.continue: 3782 // CHECK3-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] 3783 // CHECK3: omp.inner.for.inc: 3784 // CHECK3-NEXT: [[TMP32:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !35 3785 // CHECK3-NEXT: [[ADD9:%.*]] = add nsw i32 [[TMP32]], 1 3786 // CHECK3-NEXT: store i32 [[ADD9]], i32* [[DOTOMP_IV]], align 4, !llvm.access.group !35 3787 // CHECK3-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP36:![0-9]+]] 3788 // CHECK3: omp.inner.for.end: 3789 // CHECK3-NEXT: br label [[OMP_LOOP_EXIT:%.*]] 3790 // CHECK3: omp.loop.exit: 3791 // CHECK3-NEXT: [[TMP33:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 4 3792 // CHECK3-NEXT: [[TMP34:%.*]] = load i32, i32* [[TMP33]], align 4 3793 // CHECK3-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP34]]) 3794 // CHECK3-NEXT: [[TMP35:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4 3795 // CHECK3-NEXT: [[TMP36:%.*]] = icmp ne i32 [[TMP35]], 0 3796 // CHECK3-NEXT: br i1 [[TMP36]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]] 3797 // CHECK3: .omp.final.then: 3798 // CHECK3-NEXT: [[TMP37:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 3799 // CHECK3-NEXT: [[SUB10:%.*]] = sub nsw i32 [[TMP37]], 0 3800 // CHECK3-NEXT: [[DIV11:%.*]] = sdiv i32 [[SUB10]], 1 3801 // CHECK3-NEXT: [[MUL12:%.*]] = mul nsw i32 [[DIV11]], 1 3802 // CHECK3-NEXT: [[ADD13:%.*]] = add nsw i32 0, [[MUL12]] 3803 // CHECK3-NEXT: store i32 [[ADD13]], i32* [[I3]], align 4 3804 // CHECK3-NEXT: br label [[DOTOMP_FINAL_DONE]] 3805 // CHECK3: .omp.final.done: 3806 // CHECK3-NEXT: br label [[OMP_PRECOND_END]] 3807 // CHECK3: omp.precond.end: 3808 // CHECK3-NEXT: ret void 3809 // 3810 // 3811 // CHECK3-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l266 3812 // CHECK3-SAME: (i32 noundef [[CH:%.*]], i32 noundef [[N:%.*]], double* noundef [[A:%.*]], double* noundef [[B:%.*]], double* noundef [[C:%.*]]) #[[ATTR2]] { 3813 // CHECK3-NEXT: entry: 3814 // CHECK3-NEXT: [[CH_ADDR:%.*]] = alloca i32, align 4 3815 // CHECK3-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4 3816 // CHECK3-NEXT: [[A_ADDR:%.*]] = alloca double*, align 4 3817 // CHECK3-NEXT: [[B_ADDR:%.*]] = alloca double*, align 4 3818 // CHECK3-NEXT: [[C_ADDR:%.*]] = alloca double*, align 4 3819 // CHECK3-NEXT: store i32 [[CH]], i32* [[CH_ADDR]], align 4 3820 // CHECK3-NEXT: store i32 [[N]], i32* [[N_ADDR]], align 4 3821 // CHECK3-NEXT: store double* [[A]], double** [[A_ADDR]], align 4 3822 // CHECK3-NEXT: store double* [[B]], double** [[B_ADDR]], align 4 3823 // CHECK3-NEXT: store double* [[C]], double** [[C_ADDR]], align 4 3824 // CHECK3-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 5, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32*, i32*, double**, double**, double**)* @.omp_outlined..14 to void (i32*, i32*, ...)*), i32* [[CH_ADDR]], i32* [[N_ADDR]], double** [[A_ADDR]], double** [[B_ADDR]], double** [[C_ADDR]]) 3825 // CHECK3-NEXT: ret void 3826 // 3827 // 3828 // CHECK3-LABEL: define {{[^@]+}}@.omp_outlined..14 3829 // CHECK3-SAME: (i32* noalias noundef [[DOTGLOBAL_TID_:%.*]], i32* noalias noundef [[DOTBOUND_TID_:%.*]], i32* noundef nonnull align 4 dereferenceable(4) [[CH:%.*]], i32* noundef nonnull align 4 dereferenceable(4) [[N:%.*]], double** noundef nonnull align 4 dereferenceable(4) [[A:%.*]], double** noundef nonnull align 4 dereferenceable(4) [[B:%.*]], double** noundef nonnull align 4 dereferenceable(4) [[C:%.*]]) #[[ATTR2]] { 3830 // CHECK3-NEXT: entry: 3831 // CHECK3-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4 3832 // CHECK3-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4 3833 // CHECK3-NEXT: [[CH_ADDR:%.*]] = alloca i32*, align 4 3834 // CHECK3-NEXT: [[N_ADDR:%.*]] = alloca i32*, align 4 3835 // CHECK3-NEXT: [[A_ADDR:%.*]] = alloca double**, align 4 3836 // CHECK3-NEXT: [[B_ADDR:%.*]] = alloca double**, align 4 3837 // CHECK3-NEXT: [[C_ADDR:%.*]] = alloca double**, align 4 3838 // CHECK3-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 3839 // CHECK3-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 3840 // CHECK3-NEXT: [[TMP:%.*]] = alloca i32, align 4 3841 // CHECK3-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 3842 // CHECK3-NEXT: [[DOTCAPTURE_EXPR_2:%.*]] = alloca i32, align 4 3843 // CHECK3-NEXT: [[I:%.*]] = alloca i32, align 4 3844 // CHECK3-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4 3845 // CHECK3-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4 3846 // CHECK3-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 3847 // CHECK3-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 3848 // CHECK3-NEXT: [[I4:%.*]] = alloca i32, align 4 3849 // CHECK3-NEXT: [[DOTCAPTURE_EXPR__CASTED:%.*]] = alloca i32, align 4 3850 // CHECK3-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 4 3851 // CHECK3-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 4 3852 // CHECK3-NEXT: store i32* [[CH]], i32** [[CH_ADDR]], align 4 3853 // CHECK3-NEXT: store i32* [[N]], i32** [[N_ADDR]], align 4 3854 // CHECK3-NEXT: store double** [[A]], double*** [[A_ADDR]], align 4 3855 // CHECK3-NEXT: store double** [[B]], double*** [[B_ADDR]], align 4 3856 // CHECK3-NEXT: store double** [[C]], double*** [[C_ADDR]], align 4 3857 // CHECK3-NEXT: [[TMP0:%.*]] = load i32*, i32** [[CH_ADDR]], align 4 3858 // CHECK3-NEXT: [[TMP1:%.*]] = load i32*, i32** [[N_ADDR]], align 4 3859 // CHECK3-NEXT: [[TMP2:%.*]] = load double**, double*** [[A_ADDR]], align 4 3860 // CHECK3-NEXT: [[TMP3:%.*]] = load double**, double*** [[B_ADDR]], align 4 3861 // CHECK3-NEXT: [[TMP4:%.*]] = load double**, double*** [[C_ADDR]], align 4 3862 // CHECK3-NEXT: [[TMP5:%.*]] = load i32, i32* [[TMP0]], align 4 3863 // CHECK3-NEXT: store i32 [[TMP5]], i32* [[DOTCAPTURE_EXPR_]], align 4 3864 // CHECK3-NEXT: [[TMP6:%.*]] = load i32, i32* [[TMP1]], align 4 3865 // CHECK3-NEXT: store i32 [[TMP6]], i32* [[DOTCAPTURE_EXPR_1]], align 4 3866 // CHECK3-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 3867 // CHECK3-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP7]], 0 3868 // CHECK3-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 3869 // CHECK3-NEXT: [[SUB3:%.*]] = sub nsw i32 [[DIV]], 1 3870 // CHECK3-NEXT: store i32 [[SUB3]], i32* [[DOTCAPTURE_EXPR_2]], align 4 3871 // CHECK3-NEXT: store i32 0, i32* [[I]], align 4 3872 // CHECK3-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 3873 // CHECK3-NEXT: [[CMP:%.*]] = icmp slt i32 0, [[TMP8]] 3874 // CHECK3-NEXT: br i1 [[CMP]], label [[OMP_PRECOND_THEN:%.*]], label [[OMP_PRECOND_END:%.*]] 3875 // CHECK3: omp.precond.then: 3876 // CHECK3-NEXT: store i32 0, i32* [[DOTOMP_COMB_LB]], align 4 3877 // CHECK3-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_2]], align 4 3878 // CHECK3-NEXT: store i32 [[TMP9]], i32* [[DOTOMP_COMB_UB]], align 4 3879 // CHECK3-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4 3880 // CHECK3-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4 3881 // CHECK3-NEXT: [[TMP10:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 4 3882 // CHECK3-NEXT: [[TMP11:%.*]] = load i32, i32* [[TMP10]], align 4 3883 // CHECK3-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1]], i32 [[TMP11]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_COMB_LB]], i32* [[DOTOMP_COMB_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1) 3884 // CHECK3-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 3885 // CHECK3-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_2]], align 4 3886 // CHECK3-NEXT: [[CMP5:%.*]] = icmp sgt i32 [[TMP12]], [[TMP13]] 3887 // CHECK3-NEXT: br i1 [[CMP5]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] 3888 // CHECK3: cond.true: 3889 // CHECK3-NEXT: [[TMP14:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_2]], align 4 3890 // CHECK3-NEXT: br label [[COND_END:%.*]] 3891 // CHECK3: cond.false: 3892 // CHECK3-NEXT: [[TMP15:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 3893 // CHECK3-NEXT: br label [[COND_END]] 3894 // CHECK3: cond.end: 3895 // CHECK3-NEXT: [[COND:%.*]] = phi i32 [ [[TMP14]], [[COND_TRUE]] ], [ [[TMP15]], [[COND_FALSE]] ] 3896 // CHECK3-NEXT: store i32 [[COND]], i32* [[DOTOMP_COMB_UB]], align 4 3897 // CHECK3-NEXT: [[TMP16:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4 3898 // CHECK3-NEXT: store i32 [[TMP16]], i32* [[DOTOMP_IV]], align 4 3899 // CHECK3-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] 3900 // CHECK3: omp.inner.for.cond: 3901 // CHECK3-NEXT: [[TMP17:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !38 3902 // CHECK3-NEXT: [[TMP18:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4, !llvm.access.group !38 3903 // CHECK3-NEXT: [[CMP6:%.*]] = icmp sle i32 [[TMP17]], [[TMP18]] 3904 // CHECK3-NEXT: br i1 [[CMP6]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] 3905 // CHECK3: omp.inner.for.body: 3906 // CHECK3-NEXT: [[TMP19:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4, !llvm.access.group !38 3907 // CHECK3-NEXT: [[TMP20:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4, !llvm.access.group !38 3908 // CHECK3-NEXT: [[TMP21:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4, !llvm.access.group !38 3909 // CHECK3-NEXT: store i32 [[TMP21]], i32* [[DOTCAPTURE_EXPR__CASTED]], align 4, !llvm.access.group !38 3910 // CHECK3-NEXT: [[TMP22:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR__CASTED]], align 4, !llvm.access.group !38 3911 // CHECK3-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 7, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32, i32, i32*, double**, double**, double**, i32)* @.omp_outlined..15 to void (i32*, i32*, ...)*), i32 [[TMP19]], i32 [[TMP20]], i32* [[TMP1]], double** [[TMP2]], double** [[TMP3]], double** [[TMP4]], i32 [[TMP22]]), !llvm.access.group !38 3912 // CHECK3-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] 3913 // CHECK3: omp.inner.for.inc: 3914 // CHECK3-NEXT: [[TMP23:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !38 3915 // CHECK3-NEXT: [[TMP24:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4, !llvm.access.group !38 3916 // CHECK3-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP23]], [[TMP24]] 3917 // CHECK3-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4, !llvm.access.group !38 3918 // CHECK3-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP39:![0-9]+]] 3919 // CHECK3: omp.inner.for.end: 3920 // CHECK3-NEXT: br label [[OMP_LOOP_EXIT:%.*]] 3921 // CHECK3: omp.loop.exit: 3922 // CHECK3-NEXT: [[TMP25:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 4 3923 // CHECK3-NEXT: [[TMP26:%.*]] = load i32, i32* [[TMP25]], align 4 3924 // CHECK3-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP26]]) 3925 // CHECK3-NEXT: [[TMP27:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4 3926 // CHECK3-NEXT: [[TMP28:%.*]] = icmp ne i32 [[TMP27]], 0 3927 // CHECK3-NEXT: br i1 [[TMP28]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]] 3928 // CHECK3: .omp.final.then: 3929 // CHECK3-NEXT: [[TMP29:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 3930 // CHECK3-NEXT: [[SUB7:%.*]] = sub nsw i32 [[TMP29]], 0 3931 // CHECK3-NEXT: [[DIV8:%.*]] = sdiv i32 [[SUB7]], 1 3932 // CHECK3-NEXT: [[MUL:%.*]] = mul nsw i32 [[DIV8]], 1 3933 // CHECK3-NEXT: [[ADD9:%.*]] = add nsw i32 0, [[MUL]] 3934 // CHECK3-NEXT: store i32 [[ADD9]], i32* [[I4]], align 4 3935 // CHECK3-NEXT: br label [[DOTOMP_FINAL_DONE]] 3936 // CHECK3: .omp.final.done: 3937 // CHECK3-NEXT: br label [[OMP_PRECOND_END]] 3938 // CHECK3: omp.precond.end: 3939 // CHECK3-NEXT: ret void 3940 // 3941 // 3942 // CHECK3-LABEL: define {{[^@]+}}@.omp_outlined..15 3943 // CHECK3-SAME: (i32* noalias noundef [[DOTGLOBAL_TID_:%.*]], i32* noalias noundef [[DOTBOUND_TID_:%.*]], i32 noundef [[DOTPREVIOUS_LB_:%.*]], i32 noundef [[DOTPREVIOUS_UB_:%.*]], i32* noundef nonnull align 4 dereferenceable(4) [[N:%.*]], double** noundef nonnull align 4 dereferenceable(4) [[A:%.*]], double** noundef nonnull align 4 dereferenceable(4) [[B:%.*]], double** noundef nonnull align 4 dereferenceable(4) [[C:%.*]], i32 noundef [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR2]] { 3944 // CHECK3-NEXT: entry: 3945 // CHECK3-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4 3946 // CHECK3-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4 3947 // CHECK3-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i32, align 4 3948 // CHECK3-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i32, align 4 3949 // CHECK3-NEXT: [[N_ADDR:%.*]] = alloca i32*, align 4 3950 // CHECK3-NEXT: [[A_ADDR:%.*]] = alloca double**, align 4 3951 // CHECK3-NEXT: [[B_ADDR:%.*]] = alloca double**, align 4 3952 // CHECK3-NEXT: [[C_ADDR:%.*]] = alloca double**, align 4 3953 // CHECK3-NEXT: [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i32, align 4 3954 // CHECK3-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 3955 // CHECK3-NEXT: [[TMP:%.*]] = alloca i32, align 4 3956 // CHECK3-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 3957 // CHECK3-NEXT: [[DOTCAPTURE_EXPR_2:%.*]] = alloca i32, align 4 3958 // CHECK3-NEXT: [[I:%.*]] = alloca i32, align 4 3959 // CHECK3-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 3960 // CHECK3-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4 3961 // CHECK3-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 3962 // CHECK3-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 3963 // CHECK3-NEXT: [[I4:%.*]] = alloca i32, align 4 3964 // CHECK3-NEXT: [[REF_TMP:%.*]] = alloca [[CLASS_ANON_4:%.*]], align 4 3965 // CHECK3-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 4 3966 // CHECK3-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 4 3967 // CHECK3-NEXT: store i32 [[DOTPREVIOUS_LB_]], i32* [[DOTPREVIOUS_LB__ADDR]], align 4 3968 // CHECK3-NEXT: store i32 [[DOTPREVIOUS_UB_]], i32* [[DOTPREVIOUS_UB__ADDR]], align 4 3969 // CHECK3-NEXT: store i32* [[N]], i32** [[N_ADDR]], align 4 3970 // CHECK3-NEXT: store double** [[A]], double*** [[A_ADDR]], align 4 3971 // CHECK3-NEXT: store double** [[B]], double*** [[B_ADDR]], align 4 3972 // CHECK3-NEXT: store double** [[C]], double*** [[C_ADDR]], align 4 3973 // CHECK3-NEXT: store i32 [[DOTCAPTURE_EXPR_]], i32* [[DOTCAPTURE_EXPR__ADDR]], align 4 3974 // CHECK3-NEXT: [[TMP0:%.*]] = load i32*, i32** [[N_ADDR]], align 4 3975 // CHECK3-NEXT: [[TMP1:%.*]] = load double**, double*** [[A_ADDR]], align 4 3976 // CHECK3-NEXT: [[TMP2:%.*]] = load double**, double*** [[B_ADDR]], align 4 3977 // CHECK3-NEXT: [[TMP3:%.*]] = load double**, double*** [[C_ADDR]], align 4 3978 // CHECK3-NEXT: [[TMP4:%.*]] = load i32, i32* [[TMP0]], align 4 3979 // CHECK3-NEXT: store i32 [[TMP4]], i32* [[DOTCAPTURE_EXPR_1]], align 4 3980 // CHECK3-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 3981 // CHECK3-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP5]], 0 3982 // CHECK3-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 3983 // CHECK3-NEXT: [[SUB3:%.*]] = sub nsw i32 [[DIV]], 1 3984 // CHECK3-NEXT: store i32 [[SUB3]], i32* [[DOTCAPTURE_EXPR_2]], align 4 3985 // CHECK3-NEXT: store i32 0, i32* [[I]], align 4 3986 // CHECK3-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 3987 // CHECK3-NEXT: [[CMP:%.*]] = icmp slt i32 0, [[TMP6]] 3988 // CHECK3-NEXT: br i1 [[CMP]], label [[OMP_PRECOND_THEN:%.*]], label [[OMP_PRECOND_END:%.*]] 3989 // CHECK3: omp.precond.then: 3990 // CHECK3-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4 3991 // CHECK3-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_2]], align 4 3992 // CHECK3-NEXT: store i32 [[TMP7]], i32* [[DOTOMP_UB]], align 4 3993 // CHECK3-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTPREVIOUS_LB__ADDR]], align 4 3994 // CHECK3-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTPREVIOUS_UB__ADDR]], align 4 3995 // CHECK3-NEXT: store i32 [[TMP8]], i32* [[DOTOMP_LB]], align 4 3996 // CHECK3-NEXT: store i32 [[TMP9]], i32* [[DOTOMP_UB]], align 4 3997 // CHECK3-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4 3998 // CHECK3-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4 3999 // CHECK3-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR__ADDR]], align 4 4000 // CHECK3-NEXT: [[TMP11:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 4 4001 // CHECK3-NEXT: [[TMP12:%.*]] = load i32, i32* [[TMP11]], align 4 4002 // CHECK3-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB2]], i32 [[TMP12]], i32 33, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 [[TMP10]]) 4003 // CHECK3-NEXT: br label [[OMP_DISPATCH_COND:%.*]] 4004 // CHECK3: omp.dispatch.cond: 4005 // CHECK3-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 4006 // CHECK3-NEXT: [[TMP14:%.*]] = load i32, i32* [[DOTPREVIOUS_UB__ADDR]], align 4 4007 // CHECK3-NEXT: [[CMP5:%.*]] = icmp sgt i32 [[TMP13]], [[TMP14]] 4008 // CHECK3-NEXT: br i1 [[CMP5]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] 4009 // CHECK3: cond.true: 4010 // CHECK3-NEXT: [[TMP15:%.*]] = load i32, i32* [[DOTPREVIOUS_UB__ADDR]], align 4 4011 // CHECK3-NEXT: br label [[COND_END:%.*]] 4012 // CHECK3: cond.false: 4013 // CHECK3-NEXT: [[TMP16:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 4014 // CHECK3-NEXT: br label [[COND_END]] 4015 // CHECK3: cond.end: 4016 // CHECK3-NEXT: [[COND:%.*]] = phi i32 [ [[TMP15]], [[COND_TRUE]] ], [ [[TMP16]], [[COND_FALSE]] ] 4017 // CHECK3-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4 4018 // CHECK3-NEXT: [[TMP17:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4 4019 // CHECK3-NEXT: store i32 [[TMP17]], i32* [[DOTOMP_IV]], align 4 4020 // CHECK3-NEXT: [[TMP18:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 4021 // CHECK3-NEXT: [[TMP19:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 4022 // CHECK3-NEXT: [[CMP6:%.*]] = icmp sle i32 [[TMP18]], [[TMP19]] 4023 // CHECK3-NEXT: br i1 [[CMP6]], label [[OMP_DISPATCH_BODY:%.*]], label [[OMP_DISPATCH_END:%.*]] 4024 // CHECK3: omp.dispatch.body: 4025 // CHECK3-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] 4026 // CHECK3: omp.inner.for.cond: 4027 // CHECK3-NEXT: [[TMP20:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !41 4028 // CHECK3-NEXT: [[TMP21:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4, !llvm.access.group !41 4029 // CHECK3-NEXT: [[CMP7:%.*]] = icmp sle i32 [[TMP20]], [[TMP21]] 4030 // CHECK3-NEXT: br i1 [[CMP7]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] 4031 // CHECK3: omp.inner.for.body: 4032 // CHECK3-NEXT: [[TMP22:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !41 4033 // CHECK3-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP22]], 1 4034 // CHECK3-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] 4035 // CHECK3-NEXT: store i32 [[ADD]], i32* [[I4]], align 4, !llvm.access.group !41 4036 // CHECK3-NEXT: [[TMP23:%.*]] = load double*, double** [[TMP2]], align 4, !llvm.access.group !41 4037 // CHECK3-NEXT: [[TMP24:%.*]] = load i32, i32* [[I4]], align 4, !llvm.access.group !41 4038 // CHECK3-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds double, double* [[TMP23]], i32 [[TMP24]] 4039 // CHECK3-NEXT: [[TMP25:%.*]] = load double, double* [[ARRAYIDX]], align 4, !llvm.access.group !41 4040 // CHECK3-NEXT: [[TMP26:%.*]] = load double*, double** [[TMP3]], align 4, !llvm.access.group !41 4041 // CHECK3-NEXT: [[TMP27:%.*]] = load i32, i32* [[I4]], align 4, !llvm.access.group !41 4042 // CHECK3-NEXT: [[ARRAYIDX8:%.*]] = getelementptr inbounds double, double* [[TMP26]], i32 [[TMP27]] 4043 // CHECK3-NEXT: [[TMP28:%.*]] = load double, double* [[ARRAYIDX8]], align 4, !llvm.access.group !41 4044 // CHECK3-NEXT: [[ADD9:%.*]] = fadd double [[TMP25]], [[TMP28]] 4045 // CHECK3-NEXT: [[TMP29:%.*]] = load double*, double** [[TMP1]], align 4, !llvm.access.group !41 4046 // CHECK3-NEXT: [[TMP30:%.*]] = load i32, i32* [[I4]], align 4, !llvm.access.group !41 4047 // CHECK3-NEXT: [[ARRAYIDX10:%.*]] = getelementptr inbounds double, double* [[TMP29]], i32 [[TMP30]] 4048 // CHECK3-NEXT: store double [[ADD9]], double* [[ARRAYIDX10]], align 4, !llvm.access.group !41 4049 // CHECK3-NEXT: [[TMP31:%.*]] = getelementptr inbounds [[CLASS_ANON_4]], %class.anon.4* [[REF_TMP]], i32 0, i32 0 4050 // CHECK3-NEXT: store double** [[TMP1]], double*** [[TMP31]], align 4, !llvm.access.group !41 4051 // CHECK3-NEXT: [[TMP32:%.*]] = getelementptr inbounds [[CLASS_ANON_4]], %class.anon.4* [[REF_TMP]], i32 0, i32 1 4052 // CHECK3-NEXT: store i32* [[I4]], i32** [[TMP32]], align 4, !llvm.access.group !41 4053 // CHECK3-NEXT: [[TMP33:%.*]] = getelementptr inbounds [[CLASS_ANON_4]], %class.anon.4* [[REF_TMP]], i32 0, i32 2 4054 // CHECK3-NEXT: store double** [[TMP2]], double*** [[TMP33]], align 4, !llvm.access.group !41 4055 // CHECK3-NEXT: [[TMP34:%.*]] = getelementptr inbounds [[CLASS_ANON_4]], %class.anon.4* [[REF_TMP]], i32 0, i32 3 4056 // CHECK3-NEXT: store double** [[TMP3]], double*** [[TMP34]], align 4, !llvm.access.group !41 4057 // CHECK3-NEXT: call void @"_ZZZ4mainENK3$_0clEvENKUlvE3_clEv"(%class.anon.4* noundef nonnull align 4 dereferenceable(16) [[REF_TMP]]), !llvm.access.group !41 4058 // CHECK3-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] 4059 // CHECK3: omp.body.continue: 4060 // CHECK3-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] 4061 // CHECK3: omp.inner.for.inc: 4062 // CHECK3-NEXT: [[TMP35:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !41 4063 // CHECK3-NEXT: [[ADD11:%.*]] = add nsw i32 [[TMP35]], 1 4064 // CHECK3-NEXT: store i32 [[ADD11]], i32* [[DOTOMP_IV]], align 4, !llvm.access.group !41 4065 // CHECK3-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP42:![0-9]+]] 4066 // CHECK3: omp.inner.for.end: 4067 // CHECK3-NEXT: br label [[OMP_DISPATCH_INC:%.*]] 4068 // CHECK3: omp.dispatch.inc: 4069 // CHECK3-NEXT: [[TMP36:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4 4070 // CHECK3-NEXT: [[TMP37:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4 4071 // CHECK3-NEXT: [[ADD12:%.*]] = add nsw i32 [[TMP36]], [[TMP37]] 4072 // CHECK3-NEXT: store i32 [[ADD12]], i32* [[DOTOMP_LB]], align 4 4073 // CHECK3-NEXT: [[TMP38:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 4074 // CHECK3-NEXT: [[TMP39:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4 4075 // CHECK3-NEXT: [[ADD13:%.*]] = add nsw i32 [[TMP38]], [[TMP39]] 4076 // CHECK3-NEXT: store i32 [[ADD13]], i32* [[DOTOMP_UB]], align 4 4077 // CHECK3-NEXT: br label [[OMP_DISPATCH_COND]] 4078 // CHECK3: omp.dispatch.end: 4079 // CHECK3-NEXT: [[TMP40:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 4 4080 // CHECK3-NEXT: [[TMP41:%.*]] = load i32, i32* [[TMP40]], align 4 4081 // CHECK3-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP41]]) 4082 // CHECK3-NEXT: [[TMP42:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4 4083 // CHECK3-NEXT: [[TMP43:%.*]] = icmp ne i32 [[TMP42]], 0 4084 // CHECK3-NEXT: br i1 [[TMP43]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]] 4085 // CHECK3: .omp.final.then: 4086 // CHECK3-NEXT: [[TMP44:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 4087 // CHECK3-NEXT: [[SUB14:%.*]] = sub nsw i32 [[TMP44]], 0 4088 // CHECK3-NEXT: [[DIV15:%.*]] = sdiv i32 [[SUB14]], 1 4089 // CHECK3-NEXT: [[MUL16:%.*]] = mul nsw i32 [[DIV15]], 1 4090 // CHECK3-NEXT: [[ADD17:%.*]] = add nsw i32 0, [[MUL16]] 4091 // CHECK3-NEXT: store i32 [[ADD17]], i32* [[I4]], align 4 4092 // CHECK3-NEXT: br label [[DOTOMP_FINAL_DONE]] 4093 // CHECK3: .omp.final.done: 4094 // CHECK3-NEXT: br label [[OMP_PRECOND_END]] 4095 // CHECK3: omp.precond.end: 4096 // CHECK3-NEXT: ret void 4097 // 4098 // 4099 // CHECK3-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l299 4100 // CHECK3-SAME: (i32 noundef [[N:%.*]], double* noundef [[A:%.*]], double* noundef [[B:%.*]], double* noundef [[C:%.*]]) #[[ATTR2]] { 4101 // CHECK3-NEXT: entry: 4102 // CHECK3-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4 4103 // CHECK3-NEXT: [[A_ADDR:%.*]] = alloca double*, align 4 4104 // CHECK3-NEXT: [[B_ADDR:%.*]] = alloca double*, align 4 4105 // CHECK3-NEXT: [[C_ADDR:%.*]] = alloca double*, align 4 4106 // CHECK3-NEXT: store i32 [[N]], i32* [[N_ADDR]], align 4 4107 // CHECK3-NEXT: store double* [[A]], double** [[A_ADDR]], align 4 4108 // CHECK3-NEXT: store double* [[B]], double** [[B_ADDR]], align 4 4109 // CHECK3-NEXT: store double* [[C]], double** [[C_ADDR]], align 4 4110 // CHECK3-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 4, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32*, double**, double**, double**)* @.omp_outlined..18 to void (i32*, i32*, ...)*), i32* [[N_ADDR]], double** [[A_ADDR]], double** [[B_ADDR]], double** [[C_ADDR]]) 4111 // CHECK3-NEXT: ret void 4112 // 4113 // 4114 // CHECK3-LABEL: define {{[^@]+}}@.omp_outlined..18 4115 // CHECK3-SAME: (i32* noalias noundef [[DOTGLOBAL_TID_:%.*]], i32* noalias noundef [[DOTBOUND_TID_:%.*]], i32* noundef nonnull align 4 dereferenceable(4) [[N:%.*]], double** noundef nonnull align 4 dereferenceable(4) [[A:%.*]], double** noundef nonnull align 4 dereferenceable(4) [[B:%.*]], double** noundef nonnull align 4 dereferenceable(4) [[C:%.*]]) #[[ATTR2]] { 4116 // CHECK3-NEXT: entry: 4117 // CHECK3-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4 4118 // CHECK3-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4 4119 // CHECK3-NEXT: [[N_ADDR:%.*]] = alloca i32*, align 4 4120 // CHECK3-NEXT: [[A_ADDR:%.*]] = alloca double**, align 4 4121 // CHECK3-NEXT: [[B_ADDR:%.*]] = alloca double**, align 4 4122 // CHECK3-NEXT: [[C_ADDR:%.*]] = alloca double**, align 4 4123 // CHECK3-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 4124 // CHECK3-NEXT: [[TMP:%.*]] = alloca i32, align 4 4125 // CHECK3-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 4126 // CHECK3-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 4127 // CHECK3-NEXT: [[I:%.*]] = alloca i32, align 4 4128 // CHECK3-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4 4129 // CHECK3-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4 4130 // CHECK3-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 4131 // CHECK3-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 4132 // CHECK3-NEXT: [[I3:%.*]] = alloca i32, align 4 4133 // CHECK3-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 4 4134 // CHECK3-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 4 4135 // CHECK3-NEXT: store i32* [[N]], i32** [[N_ADDR]], align 4 4136 // CHECK3-NEXT: store double** [[A]], double*** [[A_ADDR]], align 4 4137 // CHECK3-NEXT: store double** [[B]], double*** [[B_ADDR]], align 4 4138 // CHECK3-NEXT: store double** [[C]], double*** [[C_ADDR]], align 4 4139 // CHECK3-NEXT: [[TMP0:%.*]] = load i32*, i32** [[N_ADDR]], align 4 4140 // CHECK3-NEXT: [[TMP1:%.*]] = load double**, double*** [[A_ADDR]], align 4 4141 // CHECK3-NEXT: [[TMP2:%.*]] = load double**, double*** [[B_ADDR]], align 4 4142 // CHECK3-NEXT: [[TMP3:%.*]] = load double**, double*** [[C_ADDR]], align 4 4143 // CHECK3-NEXT: [[TMP4:%.*]] = load i32, i32* [[TMP0]], align 4 4144 // CHECK3-NEXT: store i32 [[TMP4]], i32* [[DOTCAPTURE_EXPR_]], align 4 4145 // CHECK3-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 4146 // CHECK3-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP5]], 0 4147 // CHECK3-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 4148 // CHECK3-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1 4149 // CHECK3-NEXT: store i32 [[SUB2]], i32* [[DOTCAPTURE_EXPR_1]], align 4 4150 // CHECK3-NEXT: store i32 0, i32* [[I]], align 4 4151 // CHECK3-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 4152 // CHECK3-NEXT: [[CMP:%.*]] = icmp slt i32 0, [[TMP6]] 4153 // CHECK3-NEXT: br i1 [[CMP]], label [[OMP_PRECOND_THEN:%.*]], label [[OMP_PRECOND_END:%.*]] 4154 // CHECK3: omp.precond.then: 4155 // CHECK3-NEXT: store i32 0, i32* [[DOTOMP_COMB_LB]], align 4 4156 // CHECK3-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 4157 // CHECK3-NEXT: store i32 [[TMP7]], i32* [[DOTOMP_COMB_UB]], align 4 4158 // CHECK3-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4 4159 // CHECK3-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4 4160 // CHECK3-NEXT: [[TMP8:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 4 4161 // CHECK3-NEXT: [[TMP9:%.*]] = load i32, i32* [[TMP8]], align 4 4162 // CHECK3-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1]], i32 [[TMP9]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_COMB_LB]], i32* [[DOTOMP_COMB_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1) 4163 // CHECK3-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 4164 // CHECK3-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 4165 // CHECK3-NEXT: [[CMP4:%.*]] = icmp sgt i32 [[TMP10]], [[TMP11]] 4166 // CHECK3-NEXT: br i1 [[CMP4]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] 4167 // CHECK3: cond.true: 4168 // CHECK3-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 4169 // CHECK3-NEXT: br label [[COND_END:%.*]] 4170 // CHECK3: cond.false: 4171 // CHECK3-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 4172 // CHECK3-NEXT: br label [[COND_END]] 4173 // CHECK3: cond.end: 4174 // CHECK3-NEXT: [[COND:%.*]] = phi i32 [ [[TMP12]], [[COND_TRUE]] ], [ [[TMP13]], [[COND_FALSE]] ] 4175 // CHECK3-NEXT: store i32 [[COND]], i32* [[DOTOMP_COMB_UB]], align 4 4176 // CHECK3-NEXT: [[TMP14:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4 4177 // CHECK3-NEXT: store i32 [[TMP14]], i32* [[DOTOMP_IV]], align 4 4178 // CHECK3-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] 4179 // CHECK3: omp.inner.for.cond: 4180 // CHECK3-NEXT: [[TMP15:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !44 4181 // CHECK3-NEXT: [[TMP16:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4, !llvm.access.group !44 4182 // CHECK3-NEXT: [[CMP5:%.*]] = icmp sle i32 [[TMP15]], [[TMP16]] 4183 // CHECK3-NEXT: br i1 [[CMP5]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] 4184 // CHECK3: omp.inner.for.body: 4185 // CHECK3-NEXT: [[TMP17:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4, !llvm.access.group !44 4186 // CHECK3-NEXT: [[TMP18:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4, !llvm.access.group !44 4187 // CHECK3-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 6, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32, i32, i32*, double**, double**, double**)* @.omp_outlined..19 to void (i32*, i32*, ...)*), i32 [[TMP17]], i32 [[TMP18]], i32* [[TMP0]], double** [[TMP1]], double** [[TMP2]], double** [[TMP3]]), !llvm.access.group !44 4188 // CHECK3-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] 4189 // CHECK3: omp.inner.for.inc: 4190 // CHECK3-NEXT: [[TMP19:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !44 4191 // CHECK3-NEXT: [[TMP20:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4, !llvm.access.group !44 4192 // CHECK3-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP19]], [[TMP20]] 4193 // CHECK3-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4, !llvm.access.group !44 4194 // CHECK3-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP45:![0-9]+]] 4195 // CHECK3: omp.inner.for.end: 4196 // CHECK3-NEXT: br label [[OMP_LOOP_EXIT:%.*]] 4197 // CHECK3: omp.loop.exit: 4198 // CHECK3-NEXT: [[TMP21:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 4 4199 // CHECK3-NEXT: [[TMP22:%.*]] = load i32, i32* [[TMP21]], align 4 4200 // CHECK3-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP22]]) 4201 // CHECK3-NEXT: [[TMP23:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4 4202 // CHECK3-NEXT: [[TMP24:%.*]] = icmp ne i32 [[TMP23]], 0 4203 // CHECK3-NEXT: br i1 [[TMP24]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]] 4204 // CHECK3: .omp.final.then: 4205 // CHECK3-NEXT: [[TMP25:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 4206 // CHECK3-NEXT: [[SUB6:%.*]] = sub nsw i32 [[TMP25]], 0 4207 // CHECK3-NEXT: [[DIV7:%.*]] = sdiv i32 [[SUB6]], 1 4208 // CHECK3-NEXT: [[MUL:%.*]] = mul nsw i32 [[DIV7]], 1 4209 // CHECK3-NEXT: [[ADD8:%.*]] = add nsw i32 0, [[MUL]] 4210 // CHECK3-NEXT: store i32 [[ADD8]], i32* [[I3]], align 4 4211 // CHECK3-NEXT: br label [[DOTOMP_FINAL_DONE]] 4212 // CHECK3: .omp.final.done: 4213 // CHECK3-NEXT: br label [[OMP_PRECOND_END]] 4214 // CHECK3: omp.precond.end: 4215 // CHECK3-NEXT: ret void 4216 // 4217 // 4218 // CHECK3-LABEL: define {{[^@]+}}@.omp_outlined..19 4219 // CHECK3-SAME: (i32* noalias noundef [[DOTGLOBAL_TID_:%.*]], i32* noalias noundef [[DOTBOUND_TID_:%.*]], i32 noundef [[DOTPREVIOUS_LB_:%.*]], i32 noundef [[DOTPREVIOUS_UB_:%.*]], i32* noundef nonnull align 4 dereferenceable(4) [[N:%.*]], double** noundef nonnull align 4 dereferenceable(4) [[A:%.*]], double** noundef nonnull align 4 dereferenceable(4) [[B:%.*]], double** noundef nonnull align 4 dereferenceable(4) [[C:%.*]]) #[[ATTR2]] { 4220 // CHECK3-NEXT: entry: 4221 // CHECK3-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4 4222 // CHECK3-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4 4223 // CHECK3-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i32, align 4 4224 // CHECK3-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i32, align 4 4225 // CHECK3-NEXT: [[N_ADDR:%.*]] = alloca i32*, align 4 4226 // CHECK3-NEXT: [[A_ADDR:%.*]] = alloca double**, align 4 4227 // CHECK3-NEXT: [[B_ADDR:%.*]] = alloca double**, align 4 4228 // CHECK3-NEXT: [[C_ADDR:%.*]] = alloca double**, align 4 4229 // CHECK3-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 4230 // CHECK3-NEXT: [[TMP:%.*]] = alloca i32, align 4 4231 // CHECK3-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 4232 // CHECK3-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 4233 // CHECK3-NEXT: [[I:%.*]] = alloca i32, align 4 4234 // CHECK3-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 4235 // CHECK3-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4 4236 // CHECK3-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 4237 // CHECK3-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 4238 // CHECK3-NEXT: [[I3:%.*]] = alloca i32, align 4 4239 // CHECK3-NEXT: [[REF_TMP:%.*]] = alloca [[CLASS_ANON_5:%.*]], align 4 4240 // CHECK3-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 4 4241 // CHECK3-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 4 4242 // CHECK3-NEXT: store i32 [[DOTPREVIOUS_LB_]], i32* [[DOTPREVIOUS_LB__ADDR]], align 4 4243 // CHECK3-NEXT: store i32 [[DOTPREVIOUS_UB_]], i32* [[DOTPREVIOUS_UB__ADDR]], align 4 4244 // CHECK3-NEXT: store i32* [[N]], i32** [[N_ADDR]], align 4 4245 // CHECK3-NEXT: store double** [[A]], double*** [[A_ADDR]], align 4 4246 // CHECK3-NEXT: store double** [[B]], double*** [[B_ADDR]], align 4 4247 // CHECK3-NEXT: store double** [[C]], double*** [[C_ADDR]], align 4 4248 // CHECK3-NEXT: [[TMP0:%.*]] = load i32*, i32** [[N_ADDR]], align 4 4249 // CHECK3-NEXT: [[TMP1:%.*]] = load double**, double*** [[A_ADDR]], align 4 4250 // CHECK3-NEXT: [[TMP2:%.*]] = load double**, double*** [[B_ADDR]], align 4 4251 // CHECK3-NEXT: [[TMP3:%.*]] = load double**, double*** [[C_ADDR]], align 4 4252 // CHECK3-NEXT: [[TMP4:%.*]] = load i32, i32* [[TMP0]], align 4 4253 // CHECK3-NEXT: store i32 [[TMP4]], i32* [[DOTCAPTURE_EXPR_]], align 4 4254 // CHECK3-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 4255 // CHECK3-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP5]], 0 4256 // CHECK3-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 4257 // CHECK3-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1 4258 // CHECK3-NEXT: store i32 [[SUB2]], i32* [[DOTCAPTURE_EXPR_1]], align 4 4259 // CHECK3-NEXT: store i32 0, i32* [[I]], align 4 4260 // CHECK3-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 4261 // CHECK3-NEXT: [[CMP:%.*]] = icmp slt i32 0, [[TMP6]] 4262 // CHECK3-NEXT: br i1 [[CMP]], label [[OMP_PRECOND_THEN:%.*]], label [[OMP_PRECOND_END:%.*]] 4263 // CHECK3: omp.precond.then: 4264 // CHECK3-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4 4265 // CHECK3-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 4266 // CHECK3-NEXT: store i32 [[TMP7]], i32* [[DOTOMP_UB]], align 4 4267 // CHECK3-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTPREVIOUS_LB__ADDR]], align 4 4268 // CHECK3-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTPREVIOUS_UB__ADDR]], align 4 4269 // CHECK3-NEXT: store i32 [[TMP8]], i32* [[DOTOMP_LB]], align 4 4270 // CHECK3-NEXT: store i32 [[TMP9]], i32* [[DOTOMP_UB]], align 4 4271 // CHECK3-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4 4272 // CHECK3-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4 4273 // CHECK3-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4 4274 // CHECK3-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 4275 // CHECK3-NEXT: [[TMP12:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 4 4276 // CHECK3-NEXT: [[TMP13:%.*]] = load i32, i32* [[TMP12]], align 4 4277 // CHECK3-NEXT: call void @__kmpc_dispatch_init_4(%struct.ident_t* @[[GLOB3]], i32 [[TMP13]], i32 35, i32 [[TMP10]], i32 [[TMP11]], i32 1, i32 1) 4278 // CHECK3-NEXT: br label [[OMP_DISPATCH_COND:%.*]] 4279 // CHECK3: omp.dispatch.cond: 4280 // CHECK3-NEXT: [[TMP14:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 4 4281 // CHECK3-NEXT: [[TMP15:%.*]] = load i32, i32* [[TMP14]], align 4 4282 // CHECK3-NEXT: [[TMP16:%.*]] = call i32 @__kmpc_dispatch_next_4(%struct.ident_t* @[[GLOB3]], i32 [[TMP15]], i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]]) 4283 // CHECK3-NEXT: [[TOBOOL:%.*]] = icmp ne i32 [[TMP16]], 0 4284 // CHECK3-NEXT: br i1 [[TOBOOL]], label [[OMP_DISPATCH_BODY:%.*]], label [[OMP_DISPATCH_END:%.*]] 4285 // CHECK3: omp.dispatch.body: 4286 // CHECK3-NEXT: [[TMP17:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4 4287 // CHECK3-NEXT: store i32 [[TMP17]], i32* [[DOTOMP_IV]], align 4 4288 // CHECK3-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] 4289 // CHECK3: omp.inner.for.cond: 4290 // CHECK3-NEXT: [[TMP18:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !47 4291 // CHECK3-NEXT: [[TMP19:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4, !llvm.access.group !47 4292 // CHECK3-NEXT: [[CMP4:%.*]] = icmp sle i32 [[TMP18]], [[TMP19]] 4293 // CHECK3-NEXT: br i1 [[CMP4]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] 4294 // CHECK3: omp.inner.for.body: 4295 // CHECK3-NEXT: [[TMP20:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !47 4296 // CHECK3-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP20]], 1 4297 // CHECK3-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] 4298 // CHECK3-NEXT: store i32 [[ADD]], i32* [[I3]], align 4, !llvm.access.group !47 4299 // CHECK3-NEXT: [[TMP21:%.*]] = load double*, double** [[TMP2]], align 4, !llvm.access.group !47 4300 // CHECK3-NEXT: [[TMP22:%.*]] = load i32, i32* [[I3]], align 4, !llvm.access.group !47 4301 // CHECK3-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds double, double* [[TMP21]], i32 [[TMP22]] 4302 // CHECK3-NEXT: [[TMP23:%.*]] = load double, double* [[ARRAYIDX]], align 4, !llvm.access.group !47 4303 // CHECK3-NEXT: [[TMP24:%.*]] = load double*, double** [[TMP3]], align 4, !llvm.access.group !47 4304 // CHECK3-NEXT: [[TMP25:%.*]] = load i32, i32* [[I3]], align 4, !llvm.access.group !47 4305 // CHECK3-NEXT: [[ARRAYIDX5:%.*]] = getelementptr inbounds double, double* [[TMP24]], i32 [[TMP25]] 4306 // CHECK3-NEXT: [[TMP26:%.*]] = load double, double* [[ARRAYIDX5]], align 4, !llvm.access.group !47 4307 // CHECK3-NEXT: [[ADD6:%.*]] = fadd double [[TMP23]], [[TMP26]] 4308 // CHECK3-NEXT: [[TMP27:%.*]] = load double*, double** [[TMP1]], align 4, !llvm.access.group !47 4309 // CHECK3-NEXT: [[TMP28:%.*]] = load i32, i32* [[I3]], align 4, !llvm.access.group !47 4310 // CHECK3-NEXT: [[ARRAYIDX7:%.*]] = getelementptr inbounds double, double* [[TMP27]], i32 [[TMP28]] 4311 // CHECK3-NEXT: store double [[ADD6]], double* [[ARRAYIDX7]], align 4, !llvm.access.group !47 4312 // CHECK3-NEXT: [[TMP29:%.*]] = getelementptr inbounds [[CLASS_ANON_5]], %class.anon.5* [[REF_TMP]], i32 0, i32 0 4313 // CHECK3-NEXT: store double** [[TMP1]], double*** [[TMP29]], align 4, !llvm.access.group !47 4314 // CHECK3-NEXT: [[TMP30:%.*]] = getelementptr inbounds [[CLASS_ANON_5]], %class.anon.5* [[REF_TMP]], i32 0, i32 1 4315 // CHECK3-NEXT: store i32* [[I3]], i32** [[TMP30]], align 4, !llvm.access.group !47 4316 // CHECK3-NEXT: [[TMP31:%.*]] = getelementptr inbounds [[CLASS_ANON_5]], %class.anon.5* [[REF_TMP]], i32 0, i32 2 4317 // CHECK3-NEXT: store double** [[TMP2]], double*** [[TMP31]], align 4, !llvm.access.group !47 4318 // CHECK3-NEXT: [[TMP32:%.*]] = getelementptr inbounds [[CLASS_ANON_5]], %class.anon.5* [[REF_TMP]], i32 0, i32 3 4319 // CHECK3-NEXT: store double** [[TMP3]], double*** [[TMP32]], align 4, !llvm.access.group !47 4320 // CHECK3-NEXT: call void @"_ZZZ4mainENK3$_0clEvENKUlvE4_clEv"(%class.anon.5* noundef nonnull align 4 dereferenceable(16) [[REF_TMP]]), !llvm.access.group !47 4321 // CHECK3-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] 4322 // CHECK3: omp.body.continue: 4323 // CHECK3-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] 4324 // CHECK3: omp.inner.for.inc: 4325 // CHECK3-NEXT: [[TMP33:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !47 4326 // CHECK3-NEXT: [[ADD8:%.*]] = add nsw i32 [[TMP33]], 1 4327 // CHECK3-NEXT: store i32 [[ADD8]], i32* [[DOTOMP_IV]], align 4, !llvm.access.group !47 4328 // CHECK3-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP48:![0-9]+]] 4329 // CHECK3: omp.inner.for.end: 4330 // CHECK3-NEXT: br label [[OMP_DISPATCH_INC:%.*]] 4331 // CHECK3: omp.dispatch.inc: 4332 // CHECK3-NEXT: br label [[OMP_DISPATCH_COND]] 4333 // CHECK3: omp.dispatch.end: 4334 // CHECK3-NEXT: [[TMP34:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4 4335 // CHECK3-NEXT: [[TMP35:%.*]] = icmp ne i32 [[TMP34]], 0 4336 // CHECK3-NEXT: br i1 [[TMP35]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]] 4337 // CHECK3: .omp.final.then: 4338 // CHECK3-NEXT: [[TMP36:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 4339 // CHECK3-NEXT: [[SUB9:%.*]] = sub nsw i32 [[TMP36]], 0 4340 // CHECK3-NEXT: [[DIV10:%.*]] = sdiv i32 [[SUB9]], 1 4341 // CHECK3-NEXT: [[MUL11:%.*]] = mul nsw i32 [[DIV10]], 1 4342 // CHECK3-NEXT: [[ADD12:%.*]] = add nsw i32 0, [[MUL11]] 4343 // CHECK3-NEXT: store i32 [[ADD12]], i32* [[I3]], align 4 4344 // CHECK3-NEXT: br label [[DOTOMP_FINAL_DONE]] 4345 // CHECK3: .omp.final.done: 4346 // CHECK3-NEXT: br label [[OMP_PRECOND_END]] 4347 // CHECK3: omp.precond.end: 4348 // CHECK3-NEXT: ret void 4349 // 4350 // 4351 // CHECK3-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l328 4352 // CHECK3-SAME: (i32 noundef [[CH:%.*]], i32 noundef [[N:%.*]], double* noundef [[A:%.*]], double* noundef [[B:%.*]], double* noundef [[C:%.*]]) #[[ATTR2]] { 4353 // CHECK3-NEXT: entry: 4354 // CHECK3-NEXT: [[CH_ADDR:%.*]] = alloca i32, align 4 4355 // CHECK3-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4 4356 // CHECK3-NEXT: [[A_ADDR:%.*]] = alloca double*, align 4 4357 // CHECK3-NEXT: [[B_ADDR:%.*]] = alloca double*, align 4 4358 // CHECK3-NEXT: [[C_ADDR:%.*]] = alloca double*, align 4 4359 // CHECK3-NEXT: store i32 [[CH]], i32* [[CH_ADDR]], align 4 4360 // CHECK3-NEXT: store i32 [[N]], i32* [[N_ADDR]], align 4 4361 // CHECK3-NEXT: store double* [[A]], double** [[A_ADDR]], align 4 4362 // CHECK3-NEXT: store double* [[B]], double** [[B_ADDR]], align 4 4363 // CHECK3-NEXT: store double* [[C]], double** [[C_ADDR]], align 4 4364 // CHECK3-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 5, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32*, i32*, double**, double**, double**)* @.omp_outlined..22 to void (i32*, i32*, ...)*), i32* [[CH_ADDR]], i32* [[N_ADDR]], double** [[A_ADDR]], double** [[B_ADDR]], double** [[C_ADDR]]) 4365 // CHECK3-NEXT: ret void 4366 // 4367 // 4368 // CHECK3-LABEL: define {{[^@]+}}@.omp_outlined..22 4369 // CHECK3-SAME: (i32* noalias noundef [[DOTGLOBAL_TID_:%.*]], i32* noalias noundef [[DOTBOUND_TID_:%.*]], i32* noundef nonnull align 4 dereferenceable(4) [[CH:%.*]], i32* noundef nonnull align 4 dereferenceable(4) [[N:%.*]], double** noundef nonnull align 4 dereferenceable(4) [[A:%.*]], double** noundef nonnull align 4 dereferenceable(4) [[B:%.*]], double** noundef nonnull align 4 dereferenceable(4) [[C:%.*]]) #[[ATTR2]] { 4370 // CHECK3-NEXT: entry: 4371 // CHECK3-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4 4372 // CHECK3-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4 4373 // CHECK3-NEXT: [[CH_ADDR:%.*]] = alloca i32*, align 4 4374 // CHECK3-NEXT: [[N_ADDR:%.*]] = alloca i32*, align 4 4375 // CHECK3-NEXT: [[A_ADDR:%.*]] = alloca double**, align 4 4376 // CHECK3-NEXT: [[B_ADDR:%.*]] = alloca double**, align 4 4377 // CHECK3-NEXT: [[C_ADDR:%.*]] = alloca double**, align 4 4378 // CHECK3-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 4379 // CHECK3-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 4380 // CHECK3-NEXT: [[TMP:%.*]] = alloca i32, align 4 4381 // CHECK3-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 4382 // CHECK3-NEXT: [[DOTCAPTURE_EXPR_2:%.*]] = alloca i32, align 4 4383 // CHECK3-NEXT: [[I:%.*]] = alloca i32, align 4 4384 // CHECK3-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4 4385 // CHECK3-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4 4386 // CHECK3-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 4387 // CHECK3-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 4388 // CHECK3-NEXT: [[I4:%.*]] = alloca i32, align 4 4389 // CHECK3-NEXT: [[DOTCAPTURE_EXPR__CASTED:%.*]] = alloca i32, align 4 4390 // CHECK3-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 4 4391 // CHECK3-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 4 4392 // CHECK3-NEXT: store i32* [[CH]], i32** [[CH_ADDR]], align 4 4393 // CHECK3-NEXT: store i32* [[N]], i32** [[N_ADDR]], align 4 4394 // CHECK3-NEXT: store double** [[A]], double*** [[A_ADDR]], align 4 4395 // CHECK3-NEXT: store double** [[B]], double*** [[B_ADDR]], align 4 4396 // CHECK3-NEXT: store double** [[C]], double*** [[C_ADDR]], align 4 4397 // CHECK3-NEXT: [[TMP0:%.*]] = load i32*, i32** [[CH_ADDR]], align 4 4398 // CHECK3-NEXT: [[TMP1:%.*]] = load i32*, i32** [[N_ADDR]], align 4 4399 // CHECK3-NEXT: [[TMP2:%.*]] = load double**, double*** [[A_ADDR]], align 4 4400 // CHECK3-NEXT: [[TMP3:%.*]] = load double**, double*** [[B_ADDR]], align 4 4401 // CHECK3-NEXT: [[TMP4:%.*]] = load double**, double*** [[C_ADDR]], align 4 4402 // CHECK3-NEXT: [[TMP5:%.*]] = load i32, i32* [[TMP0]], align 4 4403 // CHECK3-NEXT: store i32 [[TMP5]], i32* [[DOTCAPTURE_EXPR_]], align 4 4404 // CHECK3-NEXT: [[TMP6:%.*]] = load i32, i32* [[TMP1]], align 4 4405 // CHECK3-NEXT: store i32 [[TMP6]], i32* [[DOTCAPTURE_EXPR_1]], align 4 4406 // CHECK3-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 4407 // CHECK3-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP7]], 0 4408 // CHECK3-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 4409 // CHECK3-NEXT: [[SUB3:%.*]] = sub nsw i32 [[DIV]], 1 4410 // CHECK3-NEXT: store i32 [[SUB3]], i32* [[DOTCAPTURE_EXPR_2]], align 4 4411 // CHECK3-NEXT: store i32 0, i32* [[I]], align 4 4412 // CHECK3-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 4413 // CHECK3-NEXT: [[CMP:%.*]] = icmp slt i32 0, [[TMP8]] 4414 // CHECK3-NEXT: br i1 [[CMP]], label [[OMP_PRECOND_THEN:%.*]], label [[OMP_PRECOND_END:%.*]] 4415 // CHECK3: omp.precond.then: 4416 // CHECK3-NEXT: store i32 0, i32* [[DOTOMP_COMB_LB]], align 4 4417 // CHECK3-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_2]], align 4 4418 // CHECK3-NEXT: store i32 [[TMP9]], i32* [[DOTOMP_COMB_UB]], align 4 4419 // CHECK3-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4 4420 // CHECK3-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4 4421 // CHECK3-NEXT: [[TMP10:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 4 4422 // CHECK3-NEXT: [[TMP11:%.*]] = load i32, i32* [[TMP10]], align 4 4423 // CHECK3-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1]], i32 [[TMP11]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_COMB_LB]], i32* [[DOTOMP_COMB_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1) 4424 // CHECK3-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 4425 // CHECK3-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_2]], align 4 4426 // CHECK3-NEXT: [[CMP5:%.*]] = icmp sgt i32 [[TMP12]], [[TMP13]] 4427 // CHECK3-NEXT: br i1 [[CMP5]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] 4428 // CHECK3: cond.true: 4429 // CHECK3-NEXT: [[TMP14:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_2]], align 4 4430 // CHECK3-NEXT: br label [[COND_END:%.*]] 4431 // CHECK3: cond.false: 4432 // CHECK3-NEXT: [[TMP15:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 4433 // CHECK3-NEXT: br label [[COND_END]] 4434 // CHECK3: cond.end: 4435 // CHECK3-NEXT: [[COND:%.*]] = phi i32 [ [[TMP14]], [[COND_TRUE]] ], [ [[TMP15]], [[COND_FALSE]] ] 4436 // CHECK3-NEXT: store i32 [[COND]], i32* [[DOTOMP_COMB_UB]], align 4 4437 // CHECK3-NEXT: [[TMP16:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4 4438 // CHECK3-NEXT: store i32 [[TMP16]], i32* [[DOTOMP_IV]], align 4 4439 // CHECK3-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] 4440 // CHECK3: omp.inner.for.cond: 4441 // CHECK3-NEXT: [[TMP17:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !50 4442 // CHECK3-NEXT: [[TMP18:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4, !llvm.access.group !50 4443 // CHECK3-NEXT: [[CMP6:%.*]] = icmp sle i32 [[TMP17]], [[TMP18]] 4444 // CHECK3-NEXT: br i1 [[CMP6]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] 4445 // CHECK3: omp.inner.for.body: 4446 // CHECK3-NEXT: [[TMP19:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4, !llvm.access.group !50 4447 // CHECK3-NEXT: [[TMP20:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4, !llvm.access.group !50 4448 // CHECK3-NEXT: [[TMP21:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4, !llvm.access.group !50 4449 // CHECK3-NEXT: store i32 [[TMP21]], i32* [[DOTCAPTURE_EXPR__CASTED]], align 4, !llvm.access.group !50 4450 // CHECK3-NEXT: [[TMP22:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR__CASTED]], align 4, !llvm.access.group !50 4451 // CHECK3-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 7, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32, i32, i32*, double**, double**, double**, i32)* @.omp_outlined..23 to void (i32*, i32*, ...)*), i32 [[TMP19]], i32 [[TMP20]], i32* [[TMP1]], double** [[TMP2]], double** [[TMP3]], double** [[TMP4]], i32 [[TMP22]]), !llvm.access.group !50 4452 // CHECK3-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] 4453 // CHECK3: omp.inner.for.inc: 4454 // CHECK3-NEXT: [[TMP23:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !50 4455 // CHECK3-NEXT: [[TMP24:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4, !llvm.access.group !50 4456 // CHECK3-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP23]], [[TMP24]] 4457 // CHECK3-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4, !llvm.access.group !50 4458 // CHECK3-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP51:![0-9]+]] 4459 // CHECK3: omp.inner.for.end: 4460 // CHECK3-NEXT: br label [[OMP_LOOP_EXIT:%.*]] 4461 // CHECK3: omp.loop.exit: 4462 // CHECK3-NEXT: [[TMP25:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 4 4463 // CHECK3-NEXT: [[TMP26:%.*]] = load i32, i32* [[TMP25]], align 4 4464 // CHECK3-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP26]]) 4465 // CHECK3-NEXT: [[TMP27:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4 4466 // CHECK3-NEXT: [[TMP28:%.*]] = icmp ne i32 [[TMP27]], 0 4467 // CHECK3-NEXT: br i1 [[TMP28]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]] 4468 // CHECK3: .omp.final.then: 4469 // CHECK3-NEXT: [[TMP29:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 4470 // CHECK3-NEXT: [[SUB7:%.*]] = sub nsw i32 [[TMP29]], 0 4471 // CHECK3-NEXT: [[DIV8:%.*]] = sdiv i32 [[SUB7]], 1 4472 // CHECK3-NEXT: [[MUL:%.*]] = mul nsw i32 [[DIV8]], 1 4473 // CHECK3-NEXT: [[ADD9:%.*]] = add nsw i32 0, [[MUL]] 4474 // CHECK3-NEXT: store i32 [[ADD9]], i32* [[I4]], align 4 4475 // CHECK3-NEXT: br label [[DOTOMP_FINAL_DONE]] 4476 // CHECK3: .omp.final.done: 4477 // CHECK3-NEXT: br label [[OMP_PRECOND_END]] 4478 // CHECK3: omp.precond.end: 4479 // CHECK3-NEXT: ret void 4480 // 4481 // 4482 // CHECK3-LABEL: define {{[^@]+}}@.omp_outlined..23 4483 // CHECK3-SAME: (i32* noalias noundef [[DOTGLOBAL_TID_:%.*]], i32* noalias noundef [[DOTBOUND_TID_:%.*]], i32 noundef [[DOTPREVIOUS_LB_:%.*]], i32 noundef [[DOTPREVIOUS_UB_:%.*]], i32* noundef nonnull align 4 dereferenceable(4) [[N:%.*]], double** noundef nonnull align 4 dereferenceable(4) [[A:%.*]], double** noundef nonnull align 4 dereferenceable(4) [[B:%.*]], double** noundef nonnull align 4 dereferenceable(4) [[C:%.*]], i32 noundef [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR2]] { 4484 // CHECK3-NEXT: entry: 4485 // CHECK3-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4 4486 // CHECK3-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4 4487 // CHECK3-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i32, align 4 4488 // CHECK3-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i32, align 4 4489 // CHECK3-NEXT: [[N_ADDR:%.*]] = alloca i32*, align 4 4490 // CHECK3-NEXT: [[A_ADDR:%.*]] = alloca double**, align 4 4491 // CHECK3-NEXT: [[B_ADDR:%.*]] = alloca double**, align 4 4492 // CHECK3-NEXT: [[C_ADDR:%.*]] = alloca double**, align 4 4493 // CHECK3-NEXT: [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i32, align 4 4494 // CHECK3-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 4495 // CHECK3-NEXT: [[TMP:%.*]] = alloca i32, align 4 4496 // CHECK3-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 4497 // CHECK3-NEXT: [[DOTCAPTURE_EXPR_2:%.*]] = alloca i32, align 4 4498 // CHECK3-NEXT: [[I:%.*]] = alloca i32, align 4 4499 // CHECK3-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 4500 // CHECK3-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4 4501 // CHECK3-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 4502 // CHECK3-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 4503 // CHECK3-NEXT: [[I4:%.*]] = alloca i32, align 4 4504 // CHECK3-NEXT: [[REF_TMP:%.*]] = alloca [[CLASS_ANON_6:%.*]], align 4 4505 // CHECK3-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 4 4506 // CHECK3-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 4 4507 // CHECK3-NEXT: store i32 [[DOTPREVIOUS_LB_]], i32* [[DOTPREVIOUS_LB__ADDR]], align 4 4508 // CHECK3-NEXT: store i32 [[DOTPREVIOUS_UB_]], i32* [[DOTPREVIOUS_UB__ADDR]], align 4 4509 // CHECK3-NEXT: store i32* [[N]], i32** [[N_ADDR]], align 4 4510 // CHECK3-NEXT: store double** [[A]], double*** [[A_ADDR]], align 4 4511 // CHECK3-NEXT: store double** [[B]], double*** [[B_ADDR]], align 4 4512 // CHECK3-NEXT: store double** [[C]], double*** [[C_ADDR]], align 4 4513 // CHECK3-NEXT: store i32 [[DOTCAPTURE_EXPR_]], i32* [[DOTCAPTURE_EXPR__ADDR]], align 4 4514 // CHECK3-NEXT: [[TMP0:%.*]] = load i32*, i32** [[N_ADDR]], align 4 4515 // CHECK3-NEXT: [[TMP1:%.*]] = load double**, double*** [[A_ADDR]], align 4 4516 // CHECK3-NEXT: [[TMP2:%.*]] = load double**, double*** [[B_ADDR]], align 4 4517 // CHECK3-NEXT: [[TMP3:%.*]] = load double**, double*** [[C_ADDR]], align 4 4518 // CHECK3-NEXT: [[TMP4:%.*]] = load i32, i32* [[TMP0]], align 4 4519 // CHECK3-NEXT: store i32 [[TMP4]], i32* [[DOTCAPTURE_EXPR_1]], align 4 4520 // CHECK3-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 4521 // CHECK3-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP5]], 0 4522 // CHECK3-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 4523 // CHECK3-NEXT: [[SUB3:%.*]] = sub nsw i32 [[DIV]], 1 4524 // CHECK3-NEXT: store i32 [[SUB3]], i32* [[DOTCAPTURE_EXPR_2]], align 4 4525 // CHECK3-NEXT: store i32 0, i32* [[I]], align 4 4526 // CHECK3-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 4527 // CHECK3-NEXT: [[CMP:%.*]] = icmp slt i32 0, [[TMP6]] 4528 // CHECK3-NEXT: br i1 [[CMP]], label [[OMP_PRECOND_THEN:%.*]], label [[OMP_PRECOND_END:%.*]] 4529 // CHECK3: omp.precond.then: 4530 // CHECK3-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4 4531 // CHECK3-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_2]], align 4 4532 // CHECK3-NEXT: store i32 [[TMP7]], i32* [[DOTOMP_UB]], align 4 4533 // CHECK3-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTPREVIOUS_LB__ADDR]], align 4 4534 // CHECK3-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTPREVIOUS_UB__ADDR]], align 4 4535 // CHECK3-NEXT: store i32 [[TMP8]], i32* [[DOTOMP_LB]], align 4 4536 // CHECK3-NEXT: store i32 [[TMP9]], i32* [[DOTOMP_UB]], align 4 4537 // CHECK3-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4 4538 // CHECK3-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4 4539 // CHECK3-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR__ADDR]], align 4 4540 // CHECK3-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4 4541 // CHECK3-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 4542 // CHECK3-NEXT: [[TMP13:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 4 4543 // CHECK3-NEXT: [[TMP14:%.*]] = load i32, i32* [[TMP13]], align 4 4544 // CHECK3-NEXT: call void @__kmpc_dispatch_init_4(%struct.ident_t* @[[GLOB3]], i32 [[TMP14]], i32 35, i32 [[TMP11]], i32 [[TMP12]], i32 1, i32 [[TMP10]]) 4545 // CHECK3-NEXT: br label [[OMP_DISPATCH_COND:%.*]] 4546 // CHECK3: omp.dispatch.cond: 4547 // CHECK3-NEXT: [[TMP15:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 4 4548 // CHECK3-NEXT: [[TMP16:%.*]] = load i32, i32* [[TMP15]], align 4 4549 // CHECK3-NEXT: [[TMP17:%.*]] = call i32 @__kmpc_dispatch_next_4(%struct.ident_t* @[[GLOB3]], i32 [[TMP16]], i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]]) 4550 // CHECK3-NEXT: [[TOBOOL:%.*]] = icmp ne i32 [[TMP17]], 0 4551 // CHECK3-NEXT: br i1 [[TOBOOL]], label [[OMP_DISPATCH_BODY:%.*]], label [[OMP_DISPATCH_END:%.*]] 4552 // CHECK3: omp.dispatch.body: 4553 // CHECK3-NEXT: [[TMP18:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4 4554 // CHECK3-NEXT: store i32 [[TMP18]], i32* [[DOTOMP_IV]], align 4 4555 // CHECK3-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] 4556 // CHECK3: omp.inner.for.cond: 4557 // CHECK3-NEXT: [[TMP19:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !53 4558 // CHECK3-NEXT: [[TMP20:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4, !llvm.access.group !53 4559 // CHECK3-NEXT: [[CMP5:%.*]] = icmp sle i32 [[TMP19]], [[TMP20]] 4560 // CHECK3-NEXT: br i1 [[CMP5]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] 4561 // CHECK3: omp.inner.for.body: 4562 // CHECK3-NEXT: [[TMP21:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !53 4563 // CHECK3-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP21]], 1 4564 // CHECK3-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] 4565 // CHECK3-NEXT: store i32 [[ADD]], i32* [[I4]], align 4, !llvm.access.group !53 4566 // CHECK3-NEXT: [[TMP22:%.*]] = load double*, double** [[TMP2]], align 4, !llvm.access.group !53 4567 // CHECK3-NEXT: [[TMP23:%.*]] = load i32, i32* [[I4]], align 4, !llvm.access.group !53 4568 // CHECK3-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds double, double* [[TMP22]], i32 [[TMP23]] 4569 // CHECK3-NEXT: [[TMP24:%.*]] = load double, double* [[ARRAYIDX]], align 4, !llvm.access.group !53 4570 // CHECK3-NEXT: [[TMP25:%.*]] = load double*, double** [[TMP3]], align 4, !llvm.access.group !53 4571 // CHECK3-NEXT: [[TMP26:%.*]] = load i32, i32* [[I4]], align 4, !llvm.access.group !53 4572 // CHECK3-NEXT: [[ARRAYIDX6:%.*]] = getelementptr inbounds double, double* [[TMP25]], i32 [[TMP26]] 4573 // CHECK3-NEXT: [[TMP27:%.*]] = load double, double* [[ARRAYIDX6]], align 4, !llvm.access.group !53 4574 // CHECK3-NEXT: [[ADD7:%.*]] = fadd double [[TMP24]], [[TMP27]] 4575 // CHECK3-NEXT: [[TMP28:%.*]] = load double*, double** [[TMP1]], align 4, !llvm.access.group !53 4576 // CHECK3-NEXT: [[TMP29:%.*]] = load i32, i32* [[I4]], align 4, !llvm.access.group !53 4577 // CHECK3-NEXT: [[ARRAYIDX8:%.*]] = getelementptr inbounds double, double* [[TMP28]], i32 [[TMP29]] 4578 // CHECK3-NEXT: store double [[ADD7]], double* [[ARRAYIDX8]], align 4, !llvm.access.group !53 4579 // CHECK3-NEXT: [[TMP30:%.*]] = getelementptr inbounds [[CLASS_ANON_6]], %class.anon.6* [[REF_TMP]], i32 0, i32 0 4580 // CHECK3-NEXT: store double** [[TMP1]], double*** [[TMP30]], align 4, !llvm.access.group !53 4581 // CHECK3-NEXT: [[TMP31:%.*]] = getelementptr inbounds [[CLASS_ANON_6]], %class.anon.6* [[REF_TMP]], i32 0, i32 1 4582 // CHECK3-NEXT: store i32* [[I4]], i32** [[TMP31]], align 4, !llvm.access.group !53 4583 // CHECK3-NEXT: [[TMP32:%.*]] = getelementptr inbounds [[CLASS_ANON_6]], %class.anon.6* [[REF_TMP]], i32 0, i32 2 4584 // CHECK3-NEXT: store double** [[TMP2]], double*** [[TMP32]], align 4, !llvm.access.group !53 4585 // CHECK3-NEXT: [[TMP33:%.*]] = getelementptr inbounds [[CLASS_ANON_6]], %class.anon.6* [[REF_TMP]], i32 0, i32 3 4586 // CHECK3-NEXT: store double** [[TMP3]], double*** [[TMP33]], align 4, !llvm.access.group !53 4587 // CHECK3-NEXT: call void @"_ZZZ4mainENK3$_0clEvENKUlvE5_clEv"(%class.anon.6* noundef nonnull align 4 dereferenceable(16) [[REF_TMP]]), !llvm.access.group !53 4588 // CHECK3-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] 4589 // CHECK3: omp.body.continue: 4590 // CHECK3-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] 4591 // CHECK3: omp.inner.for.inc: 4592 // CHECK3-NEXT: [[TMP34:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !53 4593 // CHECK3-NEXT: [[ADD9:%.*]] = add nsw i32 [[TMP34]], 1 4594 // CHECK3-NEXT: store i32 [[ADD9]], i32* [[DOTOMP_IV]], align 4, !llvm.access.group !53 4595 // CHECK3-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP54:![0-9]+]] 4596 // CHECK3: omp.inner.for.end: 4597 // CHECK3-NEXT: br label [[OMP_DISPATCH_INC:%.*]] 4598 // CHECK3: omp.dispatch.inc: 4599 // CHECK3-NEXT: br label [[OMP_DISPATCH_COND]] 4600 // CHECK3: omp.dispatch.end: 4601 // CHECK3-NEXT: [[TMP35:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4 4602 // CHECK3-NEXT: [[TMP36:%.*]] = icmp ne i32 [[TMP35]], 0 4603 // CHECK3-NEXT: br i1 [[TMP36]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]] 4604 // CHECK3: .omp.final.then: 4605 // CHECK3-NEXT: [[TMP37:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 4606 // CHECK3-NEXT: [[SUB10:%.*]] = sub nsw i32 [[TMP37]], 0 4607 // CHECK3-NEXT: [[DIV11:%.*]] = sdiv i32 [[SUB10]], 1 4608 // CHECK3-NEXT: [[MUL12:%.*]] = mul nsw i32 [[DIV11]], 1 4609 // CHECK3-NEXT: [[ADD13:%.*]] = add nsw i32 0, [[MUL12]] 4610 // CHECK3-NEXT: store i32 [[ADD13]], i32* [[I4]], align 4 4611 // CHECK3-NEXT: br label [[DOTOMP_FINAL_DONE]] 4612 // CHECK3: .omp.final.done: 4613 // CHECK3-NEXT: br label [[OMP_PRECOND_END]] 4614 // CHECK3: omp.precond.end: 4615 // CHECK3-NEXT: ret void 4616 // 4617 // 4618 // CHECK3-LABEL: define {{[^@]+}}@.omp_offloading.requires_reg 4619 // CHECK3-SAME: () #[[ATTR4:[0-9]+]] { 4620 // CHECK3-NEXT: entry: 4621 // CHECK3-NEXT: call void @__tgt_register_requires(i64 1) 4622 // CHECK3-NEXT: ret void 4623 // 4624 // 4625 // CHECK5-LABEL: define {{[^@]+}}@main 4626 // CHECK5-SAME: () #[[ATTR0:[0-9]+]] { 4627 // CHECK5-NEXT: entry: 4628 // CHECK5-NEXT: [[RETVAL:%.*]] = alloca i32, align 4 4629 // CHECK5-NEXT: [[A:%.*]] = alloca double*, align 8 4630 // CHECK5-NEXT: [[B:%.*]] = alloca double*, align 8 4631 // CHECK5-NEXT: [[C:%.*]] = alloca double*, align 8 4632 // CHECK5-NEXT: [[N:%.*]] = alloca i32, align 4 4633 // CHECK5-NEXT: [[CH:%.*]] = alloca i32, align 4 4634 // CHECK5-NEXT: [[REF_TMP:%.*]] = alloca [[CLASS_ANON:%.*]], align 8 4635 // CHECK5-NEXT: store i32 0, i32* [[RETVAL]], align 4 4636 // CHECK5-NEXT: store i32 10000, i32* [[N]], align 4 4637 // CHECK5-NEXT: store i32 100, i32* [[CH]], align 4 4638 // CHECK5-NEXT: [[TMP0:%.*]] = getelementptr inbounds [[CLASS_ANON]], %class.anon* [[REF_TMP]], i32 0, i32 0 4639 // CHECK5-NEXT: store i32* [[N]], i32** [[TMP0]], align 8 4640 // CHECK5-NEXT: [[TMP1:%.*]] = getelementptr inbounds [[CLASS_ANON]], %class.anon* [[REF_TMP]], i32 0, i32 1 4641 // CHECK5-NEXT: store double** [[A]], double*** [[TMP1]], align 8 4642 // CHECK5-NEXT: [[TMP2:%.*]] = getelementptr inbounds [[CLASS_ANON]], %class.anon* [[REF_TMP]], i32 0, i32 2 4643 // CHECK5-NEXT: store double** [[B]], double*** [[TMP2]], align 8 4644 // CHECK5-NEXT: [[TMP3:%.*]] = getelementptr inbounds [[CLASS_ANON]], %class.anon* [[REF_TMP]], i32 0, i32 3 4645 // CHECK5-NEXT: store double** [[C]], double*** [[TMP3]], align 8 4646 // CHECK5-NEXT: [[TMP4:%.*]] = getelementptr inbounds [[CLASS_ANON]], %class.anon* [[REF_TMP]], i32 0, i32 4 4647 // CHECK5-NEXT: store i32* [[CH]], i32** [[TMP4]], align 8 4648 // CHECK5-NEXT: call void @"_ZZ4mainENK3$_0clEv"(%class.anon* noundef nonnull align 8 dereferenceable(40) [[REF_TMP]]) 4649 // CHECK5-NEXT: ret i32 0 4650 // 4651 // 4652 // CHECK7-LABEL: define {{[^@]+}}@main 4653 // CHECK7-SAME: () #[[ATTR0:[0-9]+]] { 4654 // CHECK7-NEXT: entry: 4655 // CHECK7-NEXT: [[RETVAL:%.*]] = alloca i32, align 4 4656 // CHECK7-NEXT: [[A:%.*]] = alloca double*, align 4 4657 // CHECK7-NEXT: [[B:%.*]] = alloca double*, align 4 4658 // CHECK7-NEXT: [[C:%.*]] = alloca double*, align 4 4659 // CHECK7-NEXT: [[N:%.*]] = alloca i32, align 4 4660 // CHECK7-NEXT: [[CH:%.*]] = alloca i32, align 4 4661 // CHECK7-NEXT: [[REF_TMP:%.*]] = alloca [[CLASS_ANON:%.*]], align 4 4662 // CHECK7-NEXT: store i32 0, i32* [[RETVAL]], align 4 4663 // CHECK7-NEXT: store i32 10000, i32* [[N]], align 4 4664 // CHECK7-NEXT: store i32 100, i32* [[CH]], align 4 4665 // CHECK7-NEXT: [[TMP0:%.*]] = getelementptr inbounds [[CLASS_ANON]], %class.anon* [[REF_TMP]], i32 0, i32 0 4666 // CHECK7-NEXT: store i32* [[N]], i32** [[TMP0]], align 4 4667 // CHECK7-NEXT: [[TMP1:%.*]] = getelementptr inbounds [[CLASS_ANON]], %class.anon* [[REF_TMP]], i32 0, i32 1 4668 // CHECK7-NEXT: store double** [[A]], double*** [[TMP1]], align 4 4669 // CHECK7-NEXT: [[TMP2:%.*]] = getelementptr inbounds [[CLASS_ANON]], %class.anon* [[REF_TMP]], i32 0, i32 2 4670 // CHECK7-NEXT: store double** [[B]], double*** [[TMP2]], align 4 4671 // CHECK7-NEXT: [[TMP3:%.*]] = getelementptr inbounds [[CLASS_ANON]], %class.anon* [[REF_TMP]], i32 0, i32 3 4672 // CHECK7-NEXT: store double** [[C]], double*** [[TMP3]], align 4 4673 // CHECK7-NEXT: [[TMP4:%.*]] = getelementptr inbounds [[CLASS_ANON]], %class.anon* [[REF_TMP]], i32 0, i32 4 4674 // CHECK7-NEXT: store i32* [[CH]], i32** [[TMP4]], align 4 4675 // CHECK7-NEXT: call void @"_ZZ4mainENK3$_0clEv"(%class.anon* noundef nonnull align 4 dereferenceable(20) [[REF_TMP]]) 4676 // CHECK7-NEXT: ret i32 0 4677 // 4678 // 4679 // CHECK9-LABEL: define {{[^@]+}}@main 4680 // CHECK9-SAME: () #[[ATTR0:[0-9]+]] { 4681 // CHECK9-NEXT: entry: 4682 // CHECK9-NEXT: [[RETVAL:%.*]] = alloca i32, align 4 4683 // CHECK9-NEXT: [[A:%.*]] = alloca double*, align 8 4684 // CHECK9-NEXT: [[B:%.*]] = alloca double*, align 8 4685 // CHECK9-NEXT: [[C:%.*]] = alloca double*, align 8 4686 // CHECK9-NEXT: [[N:%.*]] = alloca i32, align 4 4687 // CHECK9-NEXT: [[CH:%.*]] = alloca i32, align 4 4688 // CHECK9-NEXT: [[N_CASTED:%.*]] = alloca i64, align 8 4689 // CHECK9-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [4 x i8*], align 8 4690 // CHECK9-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [4 x i8*], align 8 4691 // CHECK9-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [4 x i8*], align 8 4692 // CHECK9-NEXT: [[TMP:%.*]] = alloca i32, align 4 4693 // CHECK9-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 4694 // CHECK9-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 4695 // CHECK9-NEXT: [[N_CASTED3:%.*]] = alloca i64, align 8 4696 // CHECK9-NEXT: [[DOTOFFLOAD_BASEPTRS5:%.*]] = alloca [4 x i8*], align 8 4697 // CHECK9-NEXT: [[DOTOFFLOAD_PTRS6:%.*]] = alloca [4 x i8*], align 8 4698 // CHECK9-NEXT: [[DOTOFFLOAD_MAPPERS7:%.*]] = alloca [4 x i8*], align 8 4699 // CHECK9-NEXT: [[_TMP8:%.*]] = alloca i32, align 4 4700 // CHECK9-NEXT: [[DOTCAPTURE_EXPR_9:%.*]] = alloca i32, align 4 4701 // CHECK9-NEXT: [[DOTCAPTURE_EXPR_10:%.*]] = alloca i32, align 4 4702 // CHECK9-NEXT: [[CH_CASTED:%.*]] = alloca i64, align 8 4703 // CHECK9-NEXT: [[N_CASTED19:%.*]] = alloca i64, align 8 4704 // CHECK9-NEXT: [[DOTOFFLOAD_BASEPTRS21:%.*]] = alloca [5 x i8*], align 8 4705 // CHECK9-NEXT: [[DOTOFFLOAD_PTRS22:%.*]] = alloca [5 x i8*], align 8 4706 // CHECK9-NEXT: [[DOTOFFLOAD_MAPPERS23:%.*]] = alloca [5 x i8*], align 8 4707 // CHECK9-NEXT: [[_TMP24:%.*]] = alloca i32, align 4 4708 // CHECK9-NEXT: [[DOTCAPTURE_EXPR_25:%.*]] = alloca i32, align 4 4709 // CHECK9-NEXT: [[DOTCAPTURE_EXPR_26:%.*]] = alloca i32, align 4 4710 // CHECK9-NEXT: [[N_CASTED34:%.*]] = alloca i64, align 8 4711 // CHECK9-NEXT: [[DOTOFFLOAD_BASEPTRS36:%.*]] = alloca [4 x i8*], align 8 4712 // CHECK9-NEXT: [[DOTOFFLOAD_PTRS37:%.*]] = alloca [4 x i8*], align 8 4713 // CHECK9-NEXT: [[DOTOFFLOAD_MAPPERS38:%.*]] = alloca [4 x i8*], align 8 4714 // CHECK9-NEXT: [[_TMP39:%.*]] = alloca i32, align 4 4715 // CHECK9-NEXT: [[DOTCAPTURE_EXPR_40:%.*]] = alloca i32, align 4 4716 // CHECK9-NEXT: [[DOTCAPTURE_EXPR_41:%.*]] = alloca i32, align 4 4717 // CHECK9-NEXT: [[CH_CASTED49:%.*]] = alloca i64, align 8 4718 // CHECK9-NEXT: [[N_CASTED51:%.*]] = alloca i64, align 8 4719 // CHECK9-NEXT: [[DOTOFFLOAD_BASEPTRS53:%.*]] = alloca [5 x i8*], align 8 4720 // CHECK9-NEXT: [[DOTOFFLOAD_PTRS54:%.*]] = alloca [5 x i8*], align 8 4721 // CHECK9-NEXT: [[DOTOFFLOAD_MAPPERS55:%.*]] = alloca [5 x i8*], align 8 4722 // CHECK9-NEXT: [[_TMP56:%.*]] = alloca i32, align 4 4723 // CHECK9-NEXT: [[DOTCAPTURE_EXPR_57:%.*]] = alloca i32, align 4 4724 // CHECK9-NEXT: [[DOTCAPTURE_EXPR_58:%.*]] = alloca i32, align 4 4725 // CHECK9-NEXT: [[N_CASTED66:%.*]] = alloca i64, align 8 4726 // CHECK9-NEXT: [[DOTOFFLOAD_BASEPTRS68:%.*]] = alloca [4 x i8*], align 8 4727 // CHECK9-NEXT: [[DOTOFFLOAD_PTRS69:%.*]] = alloca [4 x i8*], align 8 4728 // CHECK9-NEXT: [[DOTOFFLOAD_MAPPERS70:%.*]] = alloca [4 x i8*], align 8 4729 // CHECK9-NEXT: [[_TMP71:%.*]] = alloca i32, align 4 4730 // CHECK9-NEXT: [[DOTCAPTURE_EXPR_72:%.*]] = alloca i32, align 4 4731 // CHECK9-NEXT: [[DOTCAPTURE_EXPR_73:%.*]] = alloca i32, align 4 4732 // CHECK9-NEXT: [[CH_CASTED81:%.*]] = alloca i64, align 8 4733 // CHECK9-NEXT: [[N_CASTED83:%.*]] = alloca i64, align 8 4734 // CHECK9-NEXT: [[DOTOFFLOAD_BASEPTRS85:%.*]] = alloca [5 x i8*], align 8 4735 // CHECK9-NEXT: [[DOTOFFLOAD_PTRS86:%.*]] = alloca [5 x i8*], align 8 4736 // CHECK9-NEXT: [[DOTOFFLOAD_MAPPERS87:%.*]] = alloca [5 x i8*], align 8 4737 // CHECK9-NEXT: [[_TMP88:%.*]] = alloca i32, align 4 4738 // CHECK9-NEXT: [[DOTCAPTURE_EXPR_89:%.*]] = alloca i32, align 4 4739 // CHECK9-NEXT: [[DOTCAPTURE_EXPR_90:%.*]] = alloca i32, align 4 4740 // CHECK9-NEXT: store i32 0, i32* [[RETVAL]], align 4 4741 // CHECK9-NEXT: store i32 10000, i32* [[N]], align 4 4742 // CHECK9-NEXT: store i32 100, i32* [[CH]], align 4 4743 // CHECK9-NEXT: [[TMP0:%.*]] = load i32, i32* [[N]], align 4 4744 // CHECK9-NEXT: [[CONV:%.*]] = bitcast i64* [[N_CASTED]] to i32* 4745 // CHECK9-NEXT: store i32 [[TMP0]], i32* [[CONV]], align 4 4746 // CHECK9-NEXT: [[TMP1:%.*]] = load i64, i64* [[N_CASTED]], align 8 4747 // CHECK9-NEXT: [[TMP2:%.*]] = load double*, double** [[A]], align 8 4748 // CHECK9-NEXT: [[TMP3:%.*]] = load double*, double** [[B]], align 8 4749 // CHECK9-NEXT: [[TMP4:%.*]] = load double*, double** [[C]], align 8 4750 // CHECK9-NEXT: [[TMP5:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 4751 // CHECK9-NEXT: [[TMP6:%.*]] = bitcast i8** [[TMP5]] to i64* 4752 // CHECK9-NEXT: store i64 [[TMP1]], i64* [[TMP6]], align 8 4753 // CHECK9-NEXT: [[TMP7:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 4754 // CHECK9-NEXT: [[TMP8:%.*]] = bitcast i8** [[TMP7]] to i64* 4755 // CHECK9-NEXT: store i64 [[TMP1]], i64* [[TMP8]], align 8 4756 // CHECK9-NEXT: [[TMP9:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0 4757 // CHECK9-NEXT: store i8* null, i8** [[TMP9]], align 8 4758 // CHECK9-NEXT: [[TMP10:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1 4759 // CHECK9-NEXT: [[TMP11:%.*]] = bitcast i8** [[TMP10]] to double** 4760 // CHECK9-NEXT: store double* [[TMP2]], double** [[TMP11]], align 8 4761 // CHECK9-NEXT: [[TMP12:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1 4762 // CHECK9-NEXT: [[TMP13:%.*]] = bitcast i8** [[TMP12]] to double** 4763 // CHECK9-NEXT: store double* [[TMP2]], double** [[TMP13]], align 8 4764 // CHECK9-NEXT: [[TMP14:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 1 4765 // CHECK9-NEXT: store i8* null, i8** [[TMP14]], align 8 4766 // CHECK9-NEXT: [[TMP15:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2 4767 // CHECK9-NEXT: [[TMP16:%.*]] = bitcast i8** [[TMP15]] to double** 4768 // CHECK9-NEXT: store double* [[TMP3]], double** [[TMP16]], align 8 4769 // CHECK9-NEXT: [[TMP17:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2 4770 // CHECK9-NEXT: [[TMP18:%.*]] = bitcast i8** [[TMP17]] to double** 4771 // CHECK9-NEXT: store double* [[TMP3]], double** [[TMP18]], align 8 4772 // CHECK9-NEXT: [[TMP19:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2 4773 // CHECK9-NEXT: store i8* null, i8** [[TMP19]], align 8 4774 // CHECK9-NEXT: [[TMP20:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 3 4775 // CHECK9-NEXT: [[TMP21:%.*]] = bitcast i8** [[TMP20]] to double** 4776 // CHECK9-NEXT: store double* [[TMP4]], double** [[TMP21]], align 8 4777 // CHECK9-NEXT: [[TMP22:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 3 4778 // CHECK9-NEXT: [[TMP23:%.*]] = bitcast i8** [[TMP22]] to double** 4779 // CHECK9-NEXT: store double* [[TMP4]], double** [[TMP23]], align 8 4780 // CHECK9-NEXT: [[TMP24:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 3 4781 // CHECK9-NEXT: store i8* null, i8** [[TMP24]], align 8 4782 // CHECK9-NEXT: [[TMP25:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 4783 // CHECK9-NEXT: [[TMP26:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 4784 // CHECK9-NEXT: [[TMP27:%.*]] = load i32, i32* [[N]], align 4 4785 // CHECK9-NEXT: store i32 [[TMP27]], i32* [[DOTCAPTURE_EXPR_]], align 4 4786 // CHECK9-NEXT: [[TMP28:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 4787 // CHECK9-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP28]], 0 4788 // CHECK9-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 4789 // CHECK9-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1 4790 // CHECK9-NEXT: store i32 [[SUB2]], i32* [[DOTCAPTURE_EXPR_1]], align 4 4791 // CHECK9-NEXT: [[TMP29:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 4792 // CHECK9-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP29]], 1 4793 // CHECK9-NEXT: [[TMP30:%.*]] = zext i32 [[ADD]] to i64 4794 // CHECK9-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8 4795 // CHECK9-NEXT: [[TMP31:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS]], i32 0, i32 0 4796 // CHECK9-NEXT: store i32 1, i32* [[TMP31]], align 4 4797 // CHECK9-NEXT: [[TMP32:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS]], i32 0, i32 1 4798 // CHECK9-NEXT: store i32 4, i32* [[TMP32]], align 4 4799 // CHECK9-NEXT: [[TMP33:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS]], i32 0, i32 2 4800 // CHECK9-NEXT: store i8** [[TMP25]], i8*** [[TMP33]], align 8 4801 // CHECK9-NEXT: [[TMP34:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS]], i32 0, i32 3 4802 // CHECK9-NEXT: store i8** [[TMP26]], i8*** [[TMP34]], align 8 4803 // CHECK9-NEXT: [[TMP35:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS]], i32 0, i32 4 4804 // CHECK9-NEXT: store i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_sizes, i32 0, i32 0), i64** [[TMP35]], align 8 4805 // CHECK9-NEXT: [[TMP36:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS]], i32 0, i32 5 4806 // CHECK9-NEXT: store i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_maptypes, i32 0, i32 0), i64** [[TMP36]], align 8 4807 // CHECK9-NEXT: [[TMP37:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS]], i32 0, i32 6 4808 // CHECK9-NEXT: store i8** null, i8*** [[TMP37]], align 8 4809 // CHECK9-NEXT: [[TMP38:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS]], i32 0, i32 7 4810 // CHECK9-NEXT: store i8** null, i8*** [[TMP38]], align 8 4811 // CHECK9-NEXT: [[TMP39:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS]], i32 0, i32 8 4812 // CHECK9-NEXT: store i64 [[TMP30]], i64* [[TMP39]], align 8 4813 // CHECK9-NEXT: [[TMP40:%.*]] = call i32 @__tgt_target_kernel(%struct.ident_t* @[[GLOB3:[0-9]+]], i64 -1, i32 0, i32 1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l368.region_id, %struct.__tgt_kernel_arguments* [[KERNEL_ARGS]]) 4814 // CHECK9-NEXT: [[TMP41:%.*]] = icmp ne i32 [[TMP40]], 0 4815 // CHECK9-NEXT: br i1 [[TMP41]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] 4816 // CHECK9: omp_offload.failed: 4817 // CHECK9-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l368(i64 [[TMP1]], double* [[TMP2]], double* [[TMP3]], double* [[TMP4]]) #[[ATTR2:[0-9]+]] 4818 // CHECK9-NEXT: br label [[OMP_OFFLOAD_CONT]] 4819 // CHECK9: omp_offload.cont: 4820 // CHECK9-NEXT: [[TMP42:%.*]] = load i32, i32* [[N]], align 4 4821 // CHECK9-NEXT: [[CONV4:%.*]] = bitcast i64* [[N_CASTED3]] to i32* 4822 // CHECK9-NEXT: store i32 [[TMP42]], i32* [[CONV4]], align 4 4823 // CHECK9-NEXT: [[TMP43:%.*]] = load i64, i64* [[N_CASTED3]], align 8 4824 // CHECK9-NEXT: [[TMP44:%.*]] = load double*, double** [[A]], align 8 4825 // CHECK9-NEXT: [[TMP45:%.*]] = load double*, double** [[B]], align 8 4826 // CHECK9-NEXT: [[TMP46:%.*]] = load double*, double** [[C]], align 8 4827 // CHECK9-NEXT: [[TMP47:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS5]], i32 0, i32 0 4828 // CHECK9-NEXT: [[TMP48:%.*]] = bitcast i8** [[TMP47]] to i64* 4829 // CHECK9-NEXT: store i64 [[TMP43]], i64* [[TMP48]], align 8 4830 // CHECK9-NEXT: [[TMP49:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS6]], i32 0, i32 0 4831 // CHECK9-NEXT: [[TMP50:%.*]] = bitcast i8** [[TMP49]] to i64* 4832 // CHECK9-NEXT: store i64 [[TMP43]], i64* [[TMP50]], align 8 4833 // CHECK9-NEXT: [[TMP51:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS7]], i64 0, i64 0 4834 // CHECK9-NEXT: store i8* null, i8** [[TMP51]], align 8 4835 // CHECK9-NEXT: [[TMP52:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS5]], i32 0, i32 1 4836 // CHECK9-NEXT: [[TMP53:%.*]] = bitcast i8** [[TMP52]] to double** 4837 // CHECK9-NEXT: store double* [[TMP44]], double** [[TMP53]], align 8 4838 // CHECK9-NEXT: [[TMP54:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS6]], i32 0, i32 1 4839 // CHECK9-NEXT: [[TMP55:%.*]] = bitcast i8** [[TMP54]] to double** 4840 // CHECK9-NEXT: store double* [[TMP44]], double** [[TMP55]], align 8 4841 // CHECK9-NEXT: [[TMP56:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS7]], i64 0, i64 1 4842 // CHECK9-NEXT: store i8* null, i8** [[TMP56]], align 8 4843 // CHECK9-NEXT: [[TMP57:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS5]], i32 0, i32 2 4844 // CHECK9-NEXT: [[TMP58:%.*]] = bitcast i8** [[TMP57]] to double** 4845 // CHECK9-NEXT: store double* [[TMP45]], double** [[TMP58]], align 8 4846 // CHECK9-NEXT: [[TMP59:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS6]], i32 0, i32 2 4847 // CHECK9-NEXT: [[TMP60:%.*]] = bitcast i8** [[TMP59]] to double** 4848 // CHECK9-NEXT: store double* [[TMP45]], double** [[TMP60]], align 8 4849 // CHECK9-NEXT: [[TMP61:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS7]], i64 0, i64 2 4850 // CHECK9-NEXT: store i8* null, i8** [[TMP61]], align 8 4851 // CHECK9-NEXT: [[TMP62:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS5]], i32 0, i32 3 4852 // CHECK9-NEXT: [[TMP63:%.*]] = bitcast i8** [[TMP62]] to double** 4853 // CHECK9-NEXT: store double* [[TMP46]], double** [[TMP63]], align 8 4854 // CHECK9-NEXT: [[TMP64:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS6]], i32 0, i32 3 4855 // CHECK9-NEXT: [[TMP65:%.*]] = bitcast i8** [[TMP64]] to double** 4856 // CHECK9-NEXT: store double* [[TMP46]], double** [[TMP65]], align 8 4857 // CHECK9-NEXT: [[TMP66:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS7]], i64 0, i64 3 4858 // CHECK9-NEXT: store i8* null, i8** [[TMP66]], align 8 4859 // CHECK9-NEXT: [[TMP67:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS5]], i32 0, i32 0 4860 // CHECK9-NEXT: [[TMP68:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS6]], i32 0, i32 0 4861 // CHECK9-NEXT: [[TMP69:%.*]] = load i32, i32* [[N]], align 4 4862 // CHECK9-NEXT: store i32 [[TMP69]], i32* [[DOTCAPTURE_EXPR_9]], align 4 4863 // CHECK9-NEXT: [[TMP70:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_9]], align 4 4864 // CHECK9-NEXT: [[SUB11:%.*]] = sub nsw i32 [[TMP70]], 0 4865 // CHECK9-NEXT: [[DIV12:%.*]] = sdiv i32 [[SUB11]], 1 4866 // CHECK9-NEXT: [[SUB13:%.*]] = sub nsw i32 [[DIV12]], 1 4867 // CHECK9-NEXT: store i32 [[SUB13]], i32* [[DOTCAPTURE_EXPR_10]], align 4 4868 // CHECK9-NEXT: [[TMP71:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_10]], align 4 4869 // CHECK9-NEXT: [[ADD14:%.*]] = add nsw i32 [[TMP71]], 1 4870 // CHECK9-NEXT: [[TMP72:%.*]] = zext i32 [[ADD14]] to i64 4871 // CHECK9-NEXT: [[KERNEL_ARGS15:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS]], align 8 4872 // CHECK9-NEXT: [[TMP73:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS15]], i32 0, i32 0 4873 // CHECK9-NEXT: store i32 1, i32* [[TMP73]], align 4 4874 // CHECK9-NEXT: [[TMP74:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS15]], i32 0, i32 1 4875 // CHECK9-NEXT: store i32 4, i32* [[TMP74]], align 4 4876 // CHECK9-NEXT: [[TMP75:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS15]], i32 0, i32 2 4877 // CHECK9-NEXT: store i8** [[TMP67]], i8*** [[TMP75]], align 8 4878 // CHECK9-NEXT: [[TMP76:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS15]], i32 0, i32 3 4879 // CHECK9-NEXT: store i8** [[TMP68]], i8*** [[TMP76]], align 8 4880 // CHECK9-NEXT: [[TMP77:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS15]], i32 0, i32 4 4881 // CHECK9-NEXT: store i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_sizes.4, i32 0, i32 0), i64** [[TMP77]], align 8 4882 // CHECK9-NEXT: [[TMP78:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS15]], i32 0, i32 5 4883 // CHECK9-NEXT: store i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_maptypes.5, i32 0, i32 0), i64** [[TMP78]], align 8 4884 // CHECK9-NEXT: [[TMP79:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS15]], i32 0, i32 6 4885 // CHECK9-NEXT: store i8** null, i8*** [[TMP79]], align 8 4886 // CHECK9-NEXT: [[TMP80:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS15]], i32 0, i32 7 4887 // CHECK9-NEXT: store i8** null, i8*** [[TMP80]], align 8 4888 // CHECK9-NEXT: [[TMP81:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS15]], i32 0, i32 8 4889 // CHECK9-NEXT: store i64 [[TMP72]], i64* [[TMP81]], align 8 4890 // CHECK9-NEXT: [[TMP82:%.*]] = call i32 @__tgt_target_kernel(%struct.ident_t* @[[GLOB3]], i64 -1, i32 0, i32 1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l407.region_id, %struct.__tgt_kernel_arguments* [[KERNEL_ARGS15]]) 4891 // CHECK9-NEXT: [[TMP83:%.*]] = icmp ne i32 [[TMP82]], 0 4892 // CHECK9-NEXT: br i1 [[TMP83]], label [[OMP_OFFLOAD_FAILED16:%.*]], label [[OMP_OFFLOAD_CONT17:%.*]] 4893 // CHECK9: omp_offload.failed16: 4894 // CHECK9-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l407(i64 [[TMP43]], double* [[TMP44]], double* [[TMP45]], double* [[TMP46]]) #[[ATTR2]] 4895 // CHECK9-NEXT: br label [[OMP_OFFLOAD_CONT17]] 4896 // CHECK9: omp_offload.cont17: 4897 // CHECK9-NEXT: [[TMP84:%.*]] = load i32, i32* [[CH]], align 4 4898 // CHECK9-NEXT: [[CONV18:%.*]] = bitcast i64* [[CH_CASTED]] to i32* 4899 // CHECK9-NEXT: store i32 [[TMP84]], i32* [[CONV18]], align 4 4900 // CHECK9-NEXT: [[TMP85:%.*]] = load i64, i64* [[CH_CASTED]], align 8 4901 // CHECK9-NEXT: [[TMP86:%.*]] = load i32, i32* [[N]], align 4 4902 // CHECK9-NEXT: [[CONV20:%.*]] = bitcast i64* [[N_CASTED19]] to i32* 4903 // CHECK9-NEXT: store i32 [[TMP86]], i32* [[CONV20]], align 4 4904 // CHECK9-NEXT: [[TMP87:%.*]] = load i64, i64* [[N_CASTED19]], align 8 4905 // CHECK9-NEXT: [[TMP88:%.*]] = load double*, double** [[A]], align 8 4906 // CHECK9-NEXT: [[TMP89:%.*]] = load double*, double** [[B]], align 8 4907 // CHECK9-NEXT: [[TMP90:%.*]] = load double*, double** [[C]], align 8 4908 // CHECK9-NEXT: [[TMP91:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS21]], i32 0, i32 0 4909 // CHECK9-NEXT: [[TMP92:%.*]] = bitcast i8** [[TMP91]] to i64* 4910 // CHECK9-NEXT: store i64 [[TMP85]], i64* [[TMP92]], align 8 4911 // CHECK9-NEXT: [[TMP93:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS22]], i32 0, i32 0 4912 // CHECK9-NEXT: [[TMP94:%.*]] = bitcast i8** [[TMP93]] to i64* 4913 // CHECK9-NEXT: store i64 [[TMP85]], i64* [[TMP94]], align 8 4914 // CHECK9-NEXT: [[TMP95:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS23]], i64 0, i64 0 4915 // CHECK9-NEXT: store i8* null, i8** [[TMP95]], align 8 4916 // CHECK9-NEXT: [[TMP96:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS21]], i32 0, i32 1 4917 // CHECK9-NEXT: [[TMP97:%.*]] = bitcast i8** [[TMP96]] to i64* 4918 // CHECK9-NEXT: store i64 [[TMP87]], i64* [[TMP97]], align 8 4919 // CHECK9-NEXT: [[TMP98:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS22]], i32 0, i32 1 4920 // CHECK9-NEXT: [[TMP99:%.*]] = bitcast i8** [[TMP98]] to i64* 4921 // CHECK9-NEXT: store i64 [[TMP87]], i64* [[TMP99]], align 8 4922 // CHECK9-NEXT: [[TMP100:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS23]], i64 0, i64 1 4923 // CHECK9-NEXT: store i8* null, i8** [[TMP100]], align 8 4924 // CHECK9-NEXT: [[TMP101:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS21]], i32 0, i32 2 4925 // CHECK9-NEXT: [[TMP102:%.*]] = bitcast i8** [[TMP101]] to double** 4926 // CHECK9-NEXT: store double* [[TMP88]], double** [[TMP102]], align 8 4927 // CHECK9-NEXT: [[TMP103:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS22]], i32 0, i32 2 4928 // CHECK9-NEXT: [[TMP104:%.*]] = bitcast i8** [[TMP103]] to double** 4929 // CHECK9-NEXT: store double* [[TMP88]], double** [[TMP104]], align 8 4930 // CHECK9-NEXT: [[TMP105:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS23]], i64 0, i64 2 4931 // CHECK9-NEXT: store i8* null, i8** [[TMP105]], align 8 4932 // CHECK9-NEXT: [[TMP106:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS21]], i32 0, i32 3 4933 // CHECK9-NEXT: [[TMP107:%.*]] = bitcast i8** [[TMP106]] to double** 4934 // CHECK9-NEXT: store double* [[TMP89]], double** [[TMP107]], align 8 4935 // CHECK9-NEXT: [[TMP108:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS22]], i32 0, i32 3 4936 // CHECK9-NEXT: [[TMP109:%.*]] = bitcast i8** [[TMP108]] to double** 4937 // CHECK9-NEXT: store double* [[TMP89]], double** [[TMP109]], align 8 4938 // CHECK9-NEXT: [[TMP110:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS23]], i64 0, i64 3 4939 // CHECK9-NEXT: store i8* null, i8** [[TMP110]], align 8 4940 // CHECK9-NEXT: [[TMP111:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS21]], i32 0, i32 4 4941 // CHECK9-NEXT: [[TMP112:%.*]] = bitcast i8** [[TMP111]] to double** 4942 // CHECK9-NEXT: store double* [[TMP90]], double** [[TMP112]], align 8 4943 // CHECK9-NEXT: [[TMP113:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS22]], i32 0, i32 4 4944 // CHECK9-NEXT: [[TMP114:%.*]] = bitcast i8** [[TMP113]] to double** 4945 // CHECK9-NEXT: store double* [[TMP90]], double** [[TMP114]], align 8 4946 // CHECK9-NEXT: [[TMP115:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS23]], i64 0, i64 4 4947 // CHECK9-NEXT: store i8* null, i8** [[TMP115]], align 8 4948 // CHECK9-NEXT: [[TMP116:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS21]], i32 0, i32 0 4949 // CHECK9-NEXT: [[TMP117:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS22]], i32 0, i32 0 4950 // CHECK9-NEXT: [[TMP118:%.*]] = load i32, i32* [[N]], align 4 4951 // CHECK9-NEXT: store i32 [[TMP118]], i32* [[DOTCAPTURE_EXPR_25]], align 4 4952 // CHECK9-NEXT: [[TMP119:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_25]], align 4 4953 // CHECK9-NEXT: [[SUB27:%.*]] = sub nsw i32 [[TMP119]], 0 4954 // CHECK9-NEXT: [[DIV28:%.*]] = sdiv i32 [[SUB27]], 1 4955 // CHECK9-NEXT: [[SUB29:%.*]] = sub nsw i32 [[DIV28]], 1 4956 // CHECK9-NEXT: store i32 [[SUB29]], i32* [[DOTCAPTURE_EXPR_26]], align 4 4957 // CHECK9-NEXT: [[TMP120:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_26]], align 4 4958 // CHECK9-NEXT: [[ADD30:%.*]] = add nsw i32 [[TMP120]], 1 4959 // CHECK9-NEXT: [[TMP121:%.*]] = zext i32 [[ADD30]] to i64 4960 // CHECK9-NEXT: [[KERNEL_ARGS31:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS]], align 8 4961 // CHECK9-NEXT: [[TMP122:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS31]], i32 0, i32 0 4962 // CHECK9-NEXT: store i32 1, i32* [[TMP122]], align 4 4963 // CHECK9-NEXT: [[TMP123:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS31]], i32 0, i32 1 4964 // CHECK9-NEXT: store i32 5, i32* [[TMP123]], align 4 4965 // CHECK9-NEXT: [[TMP124:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS31]], i32 0, i32 2 4966 // CHECK9-NEXT: store i8** [[TMP116]], i8*** [[TMP124]], align 8 4967 // CHECK9-NEXT: [[TMP125:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS31]], i32 0, i32 3 4968 // CHECK9-NEXT: store i8** [[TMP117]], i8*** [[TMP125]], align 8 4969 // CHECK9-NEXT: [[TMP126:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS31]], i32 0, i32 4 4970 // CHECK9-NEXT: store i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_sizes.8, i32 0, i32 0), i64** [[TMP126]], align 8 4971 // CHECK9-NEXT: [[TMP127:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS31]], i32 0, i32 5 4972 // CHECK9-NEXT: store i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_maptypes.9, i32 0, i32 0), i64** [[TMP127]], align 8 4973 // CHECK9-NEXT: [[TMP128:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS31]], i32 0, i32 6 4974 // CHECK9-NEXT: store i8** null, i8*** [[TMP128]], align 8 4975 // CHECK9-NEXT: [[TMP129:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS31]], i32 0, i32 7 4976 // CHECK9-NEXT: store i8** null, i8*** [[TMP129]], align 8 4977 // CHECK9-NEXT: [[TMP130:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS31]], i32 0, i32 8 4978 // CHECK9-NEXT: store i64 [[TMP121]], i64* [[TMP130]], align 8 4979 // CHECK9-NEXT: [[TMP131:%.*]] = call i32 @__tgt_target_kernel(%struct.ident_t* @[[GLOB3]], i64 -1, i32 0, i32 1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l446.region_id, %struct.__tgt_kernel_arguments* [[KERNEL_ARGS31]]) 4980 // CHECK9-NEXT: [[TMP132:%.*]] = icmp ne i32 [[TMP131]], 0 4981 // CHECK9-NEXT: br i1 [[TMP132]], label [[OMP_OFFLOAD_FAILED32:%.*]], label [[OMP_OFFLOAD_CONT33:%.*]] 4982 // CHECK9: omp_offload.failed32: 4983 // CHECK9-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l446(i64 [[TMP85]], i64 [[TMP87]], double* [[TMP88]], double* [[TMP89]], double* [[TMP90]]) #[[ATTR2]] 4984 // CHECK9-NEXT: br label [[OMP_OFFLOAD_CONT33]] 4985 // CHECK9: omp_offload.cont33: 4986 // CHECK9-NEXT: [[TMP133:%.*]] = load i32, i32* [[N]], align 4 4987 // CHECK9-NEXT: [[CONV35:%.*]] = bitcast i64* [[N_CASTED34]] to i32* 4988 // CHECK9-NEXT: store i32 [[TMP133]], i32* [[CONV35]], align 4 4989 // CHECK9-NEXT: [[TMP134:%.*]] = load i64, i64* [[N_CASTED34]], align 8 4990 // CHECK9-NEXT: [[TMP135:%.*]] = load double*, double** [[A]], align 8 4991 // CHECK9-NEXT: [[TMP136:%.*]] = load double*, double** [[B]], align 8 4992 // CHECK9-NEXT: [[TMP137:%.*]] = load double*, double** [[C]], align 8 4993 // CHECK9-NEXT: [[TMP138:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS36]], i32 0, i32 0 4994 // CHECK9-NEXT: [[TMP139:%.*]] = bitcast i8** [[TMP138]] to i64* 4995 // CHECK9-NEXT: store i64 [[TMP134]], i64* [[TMP139]], align 8 4996 // CHECK9-NEXT: [[TMP140:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS37]], i32 0, i32 0 4997 // CHECK9-NEXT: [[TMP141:%.*]] = bitcast i8** [[TMP140]] to i64* 4998 // CHECK9-NEXT: store i64 [[TMP134]], i64* [[TMP141]], align 8 4999 // CHECK9-NEXT: [[TMP142:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS38]], i64 0, i64 0 5000 // CHECK9-NEXT: store i8* null, i8** [[TMP142]], align 8 5001 // CHECK9-NEXT: [[TMP143:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS36]], i32 0, i32 1 5002 // CHECK9-NEXT: [[TMP144:%.*]] = bitcast i8** [[TMP143]] to double** 5003 // CHECK9-NEXT: store double* [[TMP135]], double** [[TMP144]], align 8 5004 // CHECK9-NEXT: [[TMP145:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS37]], i32 0, i32 1 5005 // CHECK9-NEXT: [[TMP146:%.*]] = bitcast i8** [[TMP145]] to double** 5006 // CHECK9-NEXT: store double* [[TMP135]], double** [[TMP146]], align 8 5007 // CHECK9-NEXT: [[TMP147:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS38]], i64 0, i64 1 5008 // CHECK9-NEXT: store i8* null, i8** [[TMP147]], align 8 5009 // CHECK9-NEXT: [[TMP148:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS36]], i32 0, i32 2 5010 // CHECK9-NEXT: [[TMP149:%.*]] = bitcast i8** [[TMP148]] to double** 5011 // CHECK9-NEXT: store double* [[TMP136]], double** [[TMP149]], align 8 5012 // CHECK9-NEXT: [[TMP150:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS37]], i32 0, i32 2 5013 // CHECK9-NEXT: [[TMP151:%.*]] = bitcast i8** [[TMP150]] to double** 5014 // CHECK9-NEXT: store double* [[TMP136]], double** [[TMP151]], align 8 5015 // CHECK9-NEXT: [[TMP152:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS38]], i64 0, i64 2 5016 // CHECK9-NEXT: store i8* null, i8** [[TMP152]], align 8 5017 // CHECK9-NEXT: [[TMP153:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS36]], i32 0, i32 3 5018 // CHECK9-NEXT: [[TMP154:%.*]] = bitcast i8** [[TMP153]] to double** 5019 // CHECK9-NEXT: store double* [[TMP137]], double** [[TMP154]], align 8 5020 // CHECK9-NEXT: [[TMP155:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS37]], i32 0, i32 3 5021 // CHECK9-NEXT: [[TMP156:%.*]] = bitcast i8** [[TMP155]] to double** 5022 // CHECK9-NEXT: store double* [[TMP137]], double** [[TMP156]], align 8 5023 // CHECK9-NEXT: [[TMP157:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS38]], i64 0, i64 3 5024 // CHECK9-NEXT: store i8* null, i8** [[TMP157]], align 8 5025 // CHECK9-NEXT: [[TMP158:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS36]], i32 0, i32 0 5026 // CHECK9-NEXT: [[TMP159:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS37]], i32 0, i32 0 5027 // CHECK9-NEXT: [[TMP160:%.*]] = load i32, i32* [[N]], align 4 5028 // CHECK9-NEXT: store i32 [[TMP160]], i32* [[DOTCAPTURE_EXPR_40]], align 4 5029 // CHECK9-NEXT: [[TMP161:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_40]], align 4 5030 // CHECK9-NEXT: [[SUB42:%.*]] = sub nsw i32 [[TMP161]], 0 5031 // CHECK9-NEXT: [[DIV43:%.*]] = sdiv i32 [[SUB42]], 1 5032 // CHECK9-NEXT: [[SUB44:%.*]] = sub nsw i32 [[DIV43]], 1 5033 // CHECK9-NEXT: store i32 [[SUB44]], i32* [[DOTCAPTURE_EXPR_41]], align 4 5034 // CHECK9-NEXT: [[TMP162:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_41]], align 4 5035 // CHECK9-NEXT: [[ADD45:%.*]] = add nsw i32 [[TMP162]], 1 5036 // CHECK9-NEXT: [[TMP163:%.*]] = zext i32 [[ADD45]] to i64 5037 // CHECK9-NEXT: [[KERNEL_ARGS46:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS]], align 8 5038 // CHECK9-NEXT: [[TMP164:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS46]], i32 0, i32 0 5039 // CHECK9-NEXT: store i32 1, i32* [[TMP164]], align 4 5040 // CHECK9-NEXT: [[TMP165:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS46]], i32 0, i32 1 5041 // CHECK9-NEXT: store i32 4, i32* [[TMP165]], align 4 5042 // CHECK9-NEXT: [[TMP166:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS46]], i32 0, i32 2 5043 // CHECK9-NEXT: store i8** [[TMP158]], i8*** [[TMP166]], align 8 5044 // CHECK9-NEXT: [[TMP167:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS46]], i32 0, i32 3 5045 // CHECK9-NEXT: store i8** [[TMP159]], i8*** [[TMP167]], align 8 5046 // CHECK9-NEXT: [[TMP168:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS46]], i32 0, i32 4 5047 // CHECK9-NEXT: store i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_sizes.12, i32 0, i32 0), i64** [[TMP168]], align 8 5048 // CHECK9-NEXT: [[TMP169:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS46]], i32 0, i32 5 5049 // CHECK9-NEXT: store i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_maptypes.13, i32 0, i32 0), i64** [[TMP169]], align 8 5050 // CHECK9-NEXT: [[TMP170:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS46]], i32 0, i32 6 5051 // CHECK9-NEXT: store i8** null, i8*** [[TMP170]], align 8 5052 // CHECK9-NEXT: [[TMP171:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS46]], i32 0, i32 7 5053 // CHECK9-NEXT: store i8** null, i8*** [[TMP171]], align 8 5054 // CHECK9-NEXT: [[TMP172:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS46]], i32 0, i32 8 5055 // CHECK9-NEXT: store i64 [[TMP163]], i64* [[TMP172]], align 8 5056 // CHECK9-NEXT: [[TMP173:%.*]] = call i32 @__tgt_target_kernel(%struct.ident_t* @[[GLOB3]], i64 -1, i32 0, i32 1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l477.region_id, %struct.__tgt_kernel_arguments* [[KERNEL_ARGS46]]) 5057 // CHECK9-NEXT: [[TMP174:%.*]] = icmp ne i32 [[TMP173]], 0 5058 // CHECK9-NEXT: br i1 [[TMP174]], label [[OMP_OFFLOAD_FAILED47:%.*]], label [[OMP_OFFLOAD_CONT48:%.*]] 5059 // CHECK9: omp_offload.failed47: 5060 // CHECK9-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l477(i64 [[TMP134]], double* [[TMP135]], double* [[TMP136]], double* [[TMP137]]) #[[ATTR2]] 5061 // CHECK9-NEXT: br label [[OMP_OFFLOAD_CONT48]] 5062 // CHECK9: omp_offload.cont48: 5063 // CHECK9-NEXT: [[TMP175:%.*]] = load i32, i32* [[CH]], align 4 5064 // CHECK9-NEXT: [[CONV50:%.*]] = bitcast i64* [[CH_CASTED49]] to i32* 5065 // CHECK9-NEXT: store i32 [[TMP175]], i32* [[CONV50]], align 4 5066 // CHECK9-NEXT: [[TMP176:%.*]] = load i64, i64* [[CH_CASTED49]], align 8 5067 // CHECK9-NEXT: [[TMP177:%.*]] = load i32, i32* [[N]], align 4 5068 // CHECK9-NEXT: [[CONV52:%.*]] = bitcast i64* [[N_CASTED51]] to i32* 5069 // CHECK9-NEXT: store i32 [[TMP177]], i32* [[CONV52]], align 4 5070 // CHECK9-NEXT: [[TMP178:%.*]] = load i64, i64* [[N_CASTED51]], align 8 5071 // CHECK9-NEXT: [[TMP179:%.*]] = load double*, double** [[A]], align 8 5072 // CHECK9-NEXT: [[TMP180:%.*]] = load double*, double** [[B]], align 8 5073 // CHECK9-NEXT: [[TMP181:%.*]] = load double*, double** [[C]], align 8 5074 // CHECK9-NEXT: [[TMP182:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS53]], i32 0, i32 0 5075 // CHECK9-NEXT: [[TMP183:%.*]] = bitcast i8** [[TMP182]] to i64* 5076 // CHECK9-NEXT: store i64 [[TMP176]], i64* [[TMP183]], align 8 5077 // CHECK9-NEXT: [[TMP184:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS54]], i32 0, i32 0 5078 // CHECK9-NEXT: [[TMP185:%.*]] = bitcast i8** [[TMP184]] to i64* 5079 // CHECK9-NEXT: store i64 [[TMP176]], i64* [[TMP185]], align 8 5080 // CHECK9-NEXT: [[TMP186:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS55]], i64 0, i64 0 5081 // CHECK9-NEXT: store i8* null, i8** [[TMP186]], align 8 5082 // CHECK9-NEXT: [[TMP187:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS53]], i32 0, i32 1 5083 // CHECK9-NEXT: [[TMP188:%.*]] = bitcast i8** [[TMP187]] to i64* 5084 // CHECK9-NEXT: store i64 [[TMP178]], i64* [[TMP188]], align 8 5085 // CHECK9-NEXT: [[TMP189:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS54]], i32 0, i32 1 5086 // CHECK9-NEXT: [[TMP190:%.*]] = bitcast i8** [[TMP189]] to i64* 5087 // CHECK9-NEXT: store i64 [[TMP178]], i64* [[TMP190]], align 8 5088 // CHECK9-NEXT: [[TMP191:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS55]], i64 0, i64 1 5089 // CHECK9-NEXT: store i8* null, i8** [[TMP191]], align 8 5090 // CHECK9-NEXT: [[TMP192:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS53]], i32 0, i32 2 5091 // CHECK9-NEXT: [[TMP193:%.*]] = bitcast i8** [[TMP192]] to double** 5092 // CHECK9-NEXT: store double* [[TMP179]], double** [[TMP193]], align 8 5093 // CHECK9-NEXT: [[TMP194:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS54]], i32 0, i32 2 5094 // CHECK9-NEXT: [[TMP195:%.*]] = bitcast i8** [[TMP194]] to double** 5095 // CHECK9-NEXT: store double* [[TMP179]], double** [[TMP195]], align 8 5096 // CHECK9-NEXT: [[TMP196:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS55]], i64 0, i64 2 5097 // CHECK9-NEXT: store i8* null, i8** [[TMP196]], align 8 5098 // CHECK9-NEXT: [[TMP197:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS53]], i32 0, i32 3 5099 // CHECK9-NEXT: [[TMP198:%.*]] = bitcast i8** [[TMP197]] to double** 5100 // CHECK9-NEXT: store double* [[TMP180]], double** [[TMP198]], align 8 5101 // CHECK9-NEXT: [[TMP199:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS54]], i32 0, i32 3 5102 // CHECK9-NEXT: [[TMP200:%.*]] = bitcast i8** [[TMP199]] to double** 5103 // CHECK9-NEXT: store double* [[TMP180]], double** [[TMP200]], align 8 5104 // CHECK9-NEXT: [[TMP201:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS55]], i64 0, i64 3 5105 // CHECK9-NEXT: store i8* null, i8** [[TMP201]], align 8 5106 // CHECK9-NEXT: [[TMP202:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS53]], i32 0, i32 4 5107 // CHECK9-NEXT: [[TMP203:%.*]] = bitcast i8** [[TMP202]] to double** 5108 // CHECK9-NEXT: store double* [[TMP181]], double** [[TMP203]], align 8 5109 // CHECK9-NEXT: [[TMP204:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS54]], i32 0, i32 4 5110 // CHECK9-NEXT: [[TMP205:%.*]] = bitcast i8** [[TMP204]] to double** 5111 // CHECK9-NEXT: store double* [[TMP181]], double** [[TMP205]], align 8 5112 // CHECK9-NEXT: [[TMP206:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS55]], i64 0, i64 4 5113 // CHECK9-NEXT: store i8* null, i8** [[TMP206]], align 8 5114 // CHECK9-NEXT: [[TMP207:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS53]], i32 0, i32 0 5115 // CHECK9-NEXT: [[TMP208:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS54]], i32 0, i32 0 5116 // CHECK9-NEXT: [[TMP209:%.*]] = load i32, i32* [[N]], align 4 5117 // CHECK9-NEXT: store i32 [[TMP209]], i32* [[DOTCAPTURE_EXPR_57]], align 4 5118 // CHECK9-NEXT: [[TMP210:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_57]], align 4 5119 // CHECK9-NEXT: [[SUB59:%.*]] = sub nsw i32 [[TMP210]], 0 5120 // CHECK9-NEXT: [[DIV60:%.*]] = sdiv i32 [[SUB59]], 1 5121 // CHECK9-NEXT: [[SUB61:%.*]] = sub nsw i32 [[DIV60]], 1 5122 // CHECK9-NEXT: store i32 [[SUB61]], i32* [[DOTCAPTURE_EXPR_58]], align 4 5123 // CHECK9-NEXT: [[TMP211:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_58]], align 4 5124 // CHECK9-NEXT: [[ADD62:%.*]] = add nsw i32 [[TMP211]], 1 5125 // CHECK9-NEXT: [[TMP212:%.*]] = zext i32 [[ADD62]] to i64 5126 // CHECK9-NEXT: [[KERNEL_ARGS63:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS]], align 8 5127 // CHECK9-NEXT: [[TMP213:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS63]], i32 0, i32 0 5128 // CHECK9-NEXT: store i32 1, i32* [[TMP213]], align 4 5129 // CHECK9-NEXT: [[TMP214:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS63]], i32 0, i32 1 5130 // CHECK9-NEXT: store i32 5, i32* [[TMP214]], align 4 5131 // CHECK9-NEXT: [[TMP215:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS63]], i32 0, i32 2 5132 // CHECK9-NEXT: store i8** [[TMP207]], i8*** [[TMP215]], align 8 5133 // CHECK9-NEXT: [[TMP216:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS63]], i32 0, i32 3 5134 // CHECK9-NEXT: store i8** [[TMP208]], i8*** [[TMP216]], align 8 5135 // CHECK9-NEXT: [[TMP217:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS63]], i32 0, i32 4 5136 // CHECK9-NEXT: store i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_sizes.16, i32 0, i32 0), i64** [[TMP217]], align 8 5137 // CHECK9-NEXT: [[TMP218:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS63]], i32 0, i32 5 5138 // CHECK9-NEXT: store i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_maptypes.17, i32 0, i32 0), i64** [[TMP218]], align 8 5139 // CHECK9-NEXT: [[TMP219:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS63]], i32 0, i32 6 5140 // CHECK9-NEXT: store i8** null, i8*** [[TMP219]], align 8 5141 // CHECK9-NEXT: [[TMP220:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS63]], i32 0, i32 7 5142 // CHECK9-NEXT: store i8** null, i8*** [[TMP220]], align 8 5143 // CHECK9-NEXT: [[TMP221:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS63]], i32 0, i32 8 5144 // CHECK9-NEXT: store i64 [[TMP212]], i64* [[TMP221]], align 8 5145 // CHECK9-NEXT: [[TMP222:%.*]] = call i32 @__tgt_target_kernel(%struct.ident_t* @[[GLOB3]], i64 -1, i32 0, i32 1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l505.region_id, %struct.__tgt_kernel_arguments* [[KERNEL_ARGS63]]) 5146 // CHECK9-NEXT: [[TMP223:%.*]] = icmp ne i32 [[TMP222]], 0 5147 // CHECK9-NEXT: br i1 [[TMP223]], label [[OMP_OFFLOAD_FAILED64:%.*]], label [[OMP_OFFLOAD_CONT65:%.*]] 5148 // CHECK9: omp_offload.failed64: 5149 // CHECK9-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l505(i64 [[TMP176]], i64 [[TMP178]], double* [[TMP179]], double* [[TMP180]], double* [[TMP181]]) #[[ATTR2]] 5150 // CHECK9-NEXT: br label [[OMP_OFFLOAD_CONT65]] 5151 // CHECK9: omp_offload.cont65: 5152 // CHECK9-NEXT: [[TMP224:%.*]] = load i32, i32* [[N]], align 4 5153 // CHECK9-NEXT: [[CONV67:%.*]] = bitcast i64* [[N_CASTED66]] to i32* 5154 // CHECK9-NEXT: store i32 [[TMP224]], i32* [[CONV67]], align 4 5155 // CHECK9-NEXT: [[TMP225:%.*]] = load i64, i64* [[N_CASTED66]], align 8 5156 // CHECK9-NEXT: [[TMP226:%.*]] = load double*, double** [[A]], align 8 5157 // CHECK9-NEXT: [[TMP227:%.*]] = load double*, double** [[B]], align 8 5158 // CHECK9-NEXT: [[TMP228:%.*]] = load double*, double** [[C]], align 8 5159 // CHECK9-NEXT: [[TMP229:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS68]], i32 0, i32 0 5160 // CHECK9-NEXT: [[TMP230:%.*]] = bitcast i8** [[TMP229]] to i64* 5161 // CHECK9-NEXT: store i64 [[TMP225]], i64* [[TMP230]], align 8 5162 // CHECK9-NEXT: [[TMP231:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS69]], i32 0, i32 0 5163 // CHECK9-NEXT: [[TMP232:%.*]] = bitcast i8** [[TMP231]] to i64* 5164 // CHECK9-NEXT: store i64 [[TMP225]], i64* [[TMP232]], align 8 5165 // CHECK9-NEXT: [[TMP233:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS70]], i64 0, i64 0 5166 // CHECK9-NEXT: store i8* null, i8** [[TMP233]], align 8 5167 // CHECK9-NEXT: [[TMP234:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS68]], i32 0, i32 1 5168 // CHECK9-NEXT: [[TMP235:%.*]] = bitcast i8** [[TMP234]] to double** 5169 // CHECK9-NEXT: store double* [[TMP226]], double** [[TMP235]], align 8 5170 // CHECK9-NEXT: [[TMP236:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS69]], i32 0, i32 1 5171 // CHECK9-NEXT: [[TMP237:%.*]] = bitcast i8** [[TMP236]] to double** 5172 // CHECK9-NEXT: store double* [[TMP226]], double** [[TMP237]], align 8 5173 // CHECK9-NEXT: [[TMP238:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS70]], i64 0, i64 1 5174 // CHECK9-NEXT: store i8* null, i8** [[TMP238]], align 8 5175 // CHECK9-NEXT: [[TMP239:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS68]], i32 0, i32 2 5176 // CHECK9-NEXT: [[TMP240:%.*]] = bitcast i8** [[TMP239]] to double** 5177 // CHECK9-NEXT: store double* [[TMP227]], double** [[TMP240]], align 8 5178 // CHECK9-NEXT: [[TMP241:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS69]], i32 0, i32 2 5179 // CHECK9-NEXT: [[TMP242:%.*]] = bitcast i8** [[TMP241]] to double** 5180 // CHECK9-NEXT: store double* [[TMP227]], double** [[TMP242]], align 8 5181 // CHECK9-NEXT: [[TMP243:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS70]], i64 0, i64 2 5182 // CHECK9-NEXT: store i8* null, i8** [[TMP243]], align 8 5183 // CHECK9-NEXT: [[TMP244:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS68]], i32 0, i32 3 5184 // CHECK9-NEXT: [[TMP245:%.*]] = bitcast i8** [[TMP244]] to double** 5185 // CHECK9-NEXT: store double* [[TMP228]], double** [[TMP245]], align 8 5186 // CHECK9-NEXT: [[TMP246:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS69]], i32 0, i32 3 5187 // CHECK9-NEXT: [[TMP247:%.*]] = bitcast i8** [[TMP246]] to double** 5188 // CHECK9-NEXT: store double* [[TMP228]], double** [[TMP247]], align 8 5189 // CHECK9-NEXT: [[TMP248:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS70]], i64 0, i64 3 5190 // CHECK9-NEXT: store i8* null, i8** [[TMP248]], align 8 5191 // CHECK9-NEXT: [[TMP249:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS68]], i32 0, i32 0 5192 // CHECK9-NEXT: [[TMP250:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS69]], i32 0, i32 0 5193 // CHECK9-NEXT: [[TMP251:%.*]] = load i32, i32* [[N]], align 4 5194 // CHECK9-NEXT: store i32 [[TMP251]], i32* [[DOTCAPTURE_EXPR_72]], align 4 5195 // CHECK9-NEXT: [[TMP252:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_72]], align 4 5196 // CHECK9-NEXT: [[SUB74:%.*]] = sub nsw i32 [[TMP252]], 0 5197 // CHECK9-NEXT: [[DIV75:%.*]] = sdiv i32 [[SUB74]], 1 5198 // CHECK9-NEXT: [[SUB76:%.*]] = sub nsw i32 [[DIV75]], 1 5199 // CHECK9-NEXT: store i32 [[SUB76]], i32* [[DOTCAPTURE_EXPR_73]], align 4 5200 // CHECK9-NEXT: [[TMP253:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_73]], align 4 5201 // CHECK9-NEXT: [[ADD77:%.*]] = add nsw i32 [[TMP253]], 1 5202 // CHECK9-NEXT: [[TMP254:%.*]] = zext i32 [[ADD77]] to i64 5203 // CHECK9-NEXT: [[KERNEL_ARGS78:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS]], align 8 5204 // CHECK9-NEXT: [[TMP255:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS78]], i32 0, i32 0 5205 // CHECK9-NEXT: store i32 1, i32* [[TMP255]], align 4 5206 // CHECK9-NEXT: [[TMP256:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS78]], i32 0, i32 1 5207 // CHECK9-NEXT: store i32 4, i32* [[TMP256]], align 4 5208 // CHECK9-NEXT: [[TMP257:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS78]], i32 0, i32 2 5209 // CHECK9-NEXT: store i8** [[TMP249]], i8*** [[TMP257]], align 8 5210 // CHECK9-NEXT: [[TMP258:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS78]], i32 0, i32 3 5211 // CHECK9-NEXT: store i8** [[TMP250]], i8*** [[TMP258]], align 8 5212 // CHECK9-NEXT: [[TMP259:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS78]], i32 0, i32 4 5213 // CHECK9-NEXT: store i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_sizes.20, i32 0, i32 0), i64** [[TMP259]], align 8 5214 // CHECK9-NEXT: [[TMP260:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS78]], i32 0, i32 5 5215 // CHECK9-NEXT: store i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_maptypes.21, i32 0, i32 0), i64** [[TMP260]], align 8 5216 // CHECK9-NEXT: [[TMP261:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS78]], i32 0, i32 6 5217 // CHECK9-NEXT: store i8** null, i8*** [[TMP261]], align 8 5218 // CHECK9-NEXT: [[TMP262:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS78]], i32 0, i32 7 5219 // CHECK9-NEXT: store i8** null, i8*** [[TMP262]], align 8 5220 // CHECK9-NEXT: [[TMP263:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS78]], i32 0, i32 8 5221 // CHECK9-NEXT: store i64 [[TMP254]], i64* [[TMP263]], align 8 5222 // CHECK9-NEXT: [[TMP264:%.*]] = call i32 @__tgt_target_kernel(%struct.ident_t* @[[GLOB3]], i64 -1, i32 0, i32 1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l535.region_id, %struct.__tgt_kernel_arguments* [[KERNEL_ARGS78]]) 5223 // CHECK9-NEXT: [[TMP265:%.*]] = icmp ne i32 [[TMP264]], 0 5224 // CHECK9-NEXT: br i1 [[TMP265]], label [[OMP_OFFLOAD_FAILED79:%.*]], label [[OMP_OFFLOAD_CONT80:%.*]] 5225 // CHECK9: omp_offload.failed79: 5226 // CHECK9-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l535(i64 [[TMP225]], double* [[TMP226]], double* [[TMP227]], double* [[TMP228]]) #[[ATTR2]] 5227 // CHECK9-NEXT: br label [[OMP_OFFLOAD_CONT80]] 5228 // CHECK9: omp_offload.cont80: 5229 // CHECK9-NEXT: [[TMP266:%.*]] = load i32, i32* [[CH]], align 4 5230 // CHECK9-NEXT: [[CONV82:%.*]] = bitcast i64* [[CH_CASTED81]] to i32* 5231 // CHECK9-NEXT: store i32 [[TMP266]], i32* [[CONV82]], align 4 5232 // CHECK9-NEXT: [[TMP267:%.*]] = load i64, i64* [[CH_CASTED81]], align 8 5233 // CHECK9-NEXT: [[TMP268:%.*]] = load i32, i32* [[N]], align 4 5234 // CHECK9-NEXT: [[CONV84:%.*]] = bitcast i64* [[N_CASTED83]] to i32* 5235 // CHECK9-NEXT: store i32 [[TMP268]], i32* [[CONV84]], align 4 5236 // CHECK9-NEXT: [[TMP269:%.*]] = load i64, i64* [[N_CASTED83]], align 8 5237 // CHECK9-NEXT: [[TMP270:%.*]] = load double*, double** [[A]], align 8 5238 // CHECK9-NEXT: [[TMP271:%.*]] = load double*, double** [[B]], align 8 5239 // CHECK9-NEXT: [[TMP272:%.*]] = load double*, double** [[C]], align 8 5240 // CHECK9-NEXT: [[TMP273:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS85]], i32 0, i32 0 5241 // CHECK9-NEXT: [[TMP274:%.*]] = bitcast i8** [[TMP273]] to i64* 5242 // CHECK9-NEXT: store i64 [[TMP267]], i64* [[TMP274]], align 8 5243 // CHECK9-NEXT: [[TMP275:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS86]], i32 0, i32 0 5244 // CHECK9-NEXT: [[TMP276:%.*]] = bitcast i8** [[TMP275]] to i64* 5245 // CHECK9-NEXT: store i64 [[TMP267]], i64* [[TMP276]], align 8 5246 // CHECK9-NEXT: [[TMP277:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS87]], i64 0, i64 0 5247 // CHECK9-NEXT: store i8* null, i8** [[TMP277]], align 8 5248 // CHECK9-NEXT: [[TMP278:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS85]], i32 0, i32 1 5249 // CHECK9-NEXT: [[TMP279:%.*]] = bitcast i8** [[TMP278]] to i64* 5250 // CHECK9-NEXT: store i64 [[TMP269]], i64* [[TMP279]], align 8 5251 // CHECK9-NEXT: [[TMP280:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS86]], i32 0, i32 1 5252 // CHECK9-NEXT: [[TMP281:%.*]] = bitcast i8** [[TMP280]] to i64* 5253 // CHECK9-NEXT: store i64 [[TMP269]], i64* [[TMP281]], align 8 5254 // CHECK9-NEXT: [[TMP282:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS87]], i64 0, i64 1 5255 // CHECK9-NEXT: store i8* null, i8** [[TMP282]], align 8 5256 // CHECK9-NEXT: [[TMP283:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS85]], i32 0, i32 2 5257 // CHECK9-NEXT: [[TMP284:%.*]] = bitcast i8** [[TMP283]] to double** 5258 // CHECK9-NEXT: store double* [[TMP270]], double** [[TMP284]], align 8 5259 // CHECK9-NEXT: [[TMP285:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS86]], i32 0, i32 2 5260 // CHECK9-NEXT: [[TMP286:%.*]] = bitcast i8** [[TMP285]] to double** 5261 // CHECK9-NEXT: store double* [[TMP270]], double** [[TMP286]], align 8 5262 // CHECK9-NEXT: [[TMP287:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS87]], i64 0, i64 2 5263 // CHECK9-NEXT: store i8* null, i8** [[TMP287]], align 8 5264 // CHECK9-NEXT: [[TMP288:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS85]], i32 0, i32 3 5265 // CHECK9-NEXT: [[TMP289:%.*]] = bitcast i8** [[TMP288]] to double** 5266 // CHECK9-NEXT: store double* [[TMP271]], double** [[TMP289]], align 8 5267 // CHECK9-NEXT: [[TMP290:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS86]], i32 0, i32 3 5268 // CHECK9-NEXT: [[TMP291:%.*]] = bitcast i8** [[TMP290]] to double** 5269 // CHECK9-NEXT: store double* [[TMP271]], double** [[TMP291]], align 8 5270 // CHECK9-NEXT: [[TMP292:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS87]], i64 0, i64 3 5271 // CHECK9-NEXT: store i8* null, i8** [[TMP292]], align 8 5272 // CHECK9-NEXT: [[TMP293:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS85]], i32 0, i32 4 5273 // CHECK9-NEXT: [[TMP294:%.*]] = bitcast i8** [[TMP293]] to double** 5274 // CHECK9-NEXT: store double* [[TMP272]], double** [[TMP294]], align 8 5275 // CHECK9-NEXT: [[TMP295:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS86]], i32 0, i32 4 5276 // CHECK9-NEXT: [[TMP296:%.*]] = bitcast i8** [[TMP295]] to double** 5277 // CHECK9-NEXT: store double* [[TMP272]], double** [[TMP296]], align 8 5278 // CHECK9-NEXT: [[TMP297:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS87]], i64 0, i64 4 5279 // CHECK9-NEXT: store i8* null, i8** [[TMP297]], align 8 5280 // CHECK9-NEXT: [[TMP298:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS85]], i32 0, i32 0 5281 // CHECK9-NEXT: [[TMP299:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS86]], i32 0, i32 0 5282 // CHECK9-NEXT: [[TMP300:%.*]] = load i32, i32* [[N]], align 4 5283 // CHECK9-NEXT: store i32 [[TMP300]], i32* [[DOTCAPTURE_EXPR_89]], align 4 5284 // CHECK9-NEXT: [[TMP301:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_89]], align 4 5285 // CHECK9-NEXT: [[SUB91:%.*]] = sub nsw i32 [[TMP301]], 0 5286 // CHECK9-NEXT: [[DIV92:%.*]] = sdiv i32 [[SUB91]], 1 5287 // CHECK9-NEXT: [[SUB93:%.*]] = sub nsw i32 [[DIV92]], 1 5288 // CHECK9-NEXT: store i32 [[SUB93]], i32* [[DOTCAPTURE_EXPR_90]], align 4 5289 // CHECK9-NEXT: [[TMP302:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_90]], align 4 5290 // CHECK9-NEXT: [[ADD94:%.*]] = add nsw i32 [[TMP302]], 1 5291 // CHECK9-NEXT: [[TMP303:%.*]] = zext i32 [[ADD94]] to i64 5292 // CHECK9-NEXT: [[KERNEL_ARGS95:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS]], align 8 5293 // CHECK9-NEXT: [[TMP304:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS95]], i32 0, i32 0 5294 // CHECK9-NEXT: store i32 1, i32* [[TMP304]], align 4 5295 // CHECK9-NEXT: [[TMP305:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS95]], i32 0, i32 1 5296 // CHECK9-NEXT: store i32 5, i32* [[TMP305]], align 4 5297 // CHECK9-NEXT: [[TMP306:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS95]], i32 0, i32 2 5298 // CHECK9-NEXT: store i8** [[TMP298]], i8*** [[TMP306]], align 8 5299 // CHECK9-NEXT: [[TMP307:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS95]], i32 0, i32 3 5300 // CHECK9-NEXT: store i8** [[TMP299]], i8*** [[TMP307]], align 8 5301 // CHECK9-NEXT: [[TMP308:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS95]], i32 0, i32 4 5302 // CHECK9-NEXT: store i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_sizes.24, i32 0, i32 0), i64** [[TMP308]], align 8 5303 // CHECK9-NEXT: [[TMP309:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS95]], i32 0, i32 5 5304 // CHECK9-NEXT: store i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_maptypes.25, i32 0, i32 0), i64** [[TMP309]], align 8 5305 // CHECK9-NEXT: [[TMP310:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS95]], i32 0, i32 6 5306 // CHECK9-NEXT: store i8** null, i8*** [[TMP310]], align 8 5307 // CHECK9-NEXT: [[TMP311:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS95]], i32 0, i32 7 5308 // CHECK9-NEXT: store i8** null, i8*** [[TMP311]], align 8 5309 // CHECK9-NEXT: [[TMP312:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS95]], i32 0, i32 8 5310 // CHECK9-NEXT: store i64 [[TMP303]], i64* [[TMP312]], align 8 5311 // CHECK9-NEXT: [[TMP313:%.*]] = call i32 @__tgt_target_kernel(%struct.ident_t* @[[GLOB3]], i64 -1, i32 0, i32 1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l561.region_id, %struct.__tgt_kernel_arguments* [[KERNEL_ARGS95]]) 5312 // CHECK9-NEXT: [[TMP314:%.*]] = icmp ne i32 [[TMP313]], 0 5313 // CHECK9-NEXT: br i1 [[TMP314]], label [[OMP_OFFLOAD_FAILED96:%.*]], label [[OMP_OFFLOAD_CONT97:%.*]] 5314 // CHECK9: omp_offload.failed96: 5315 // CHECK9-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l561(i64 [[TMP267]], i64 [[TMP269]], double* [[TMP270]], double* [[TMP271]], double* [[TMP272]]) #[[ATTR2]] 5316 // CHECK9-NEXT: br label [[OMP_OFFLOAD_CONT97]] 5317 // CHECK9: omp_offload.cont97: 5318 // CHECK9-NEXT: [[CALL:%.*]] = call noundef signext i32 @_Z5tmainIiET_v() 5319 // CHECK9-NEXT: ret i32 [[CALL]] 5320 // 5321 // 5322 // CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l368 5323 // CHECK9-SAME: (i64 noundef [[N:%.*]], double* noundef [[A:%.*]], double* noundef [[B:%.*]], double* noundef [[C:%.*]]) #[[ATTR1:[0-9]+]] { 5324 // CHECK9-NEXT: entry: 5325 // CHECK9-NEXT: [[N_ADDR:%.*]] = alloca i64, align 8 5326 // CHECK9-NEXT: [[A_ADDR:%.*]] = alloca double*, align 8 5327 // CHECK9-NEXT: [[B_ADDR:%.*]] = alloca double*, align 8 5328 // CHECK9-NEXT: [[C_ADDR:%.*]] = alloca double*, align 8 5329 // CHECK9-NEXT: store i64 [[N]], i64* [[N_ADDR]], align 8 5330 // CHECK9-NEXT: store double* [[A]], double** [[A_ADDR]], align 8 5331 // CHECK9-NEXT: store double* [[B]], double** [[B_ADDR]], align 8 5332 // CHECK9-NEXT: store double* [[C]], double** [[C_ADDR]], align 8 5333 // CHECK9-NEXT: [[CONV:%.*]] = bitcast i64* [[N_ADDR]] to i32* 5334 // CHECK9-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 4, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32*, double**, double**, double**)* @.omp_outlined. to void (i32*, i32*, ...)*), i32* [[CONV]], double** [[A_ADDR]], double** [[B_ADDR]], double** [[C_ADDR]]) 5335 // CHECK9-NEXT: ret void 5336 // 5337 // 5338 // CHECK9-LABEL: define {{[^@]+}}@.omp_outlined. 5339 // CHECK9-SAME: (i32* noalias noundef [[DOTGLOBAL_TID_:%.*]], i32* noalias noundef [[DOTBOUND_TID_:%.*]], i32* noundef nonnull align 4 dereferenceable(4) [[N:%.*]], double** noundef nonnull align 8 dereferenceable(8) [[A:%.*]], double** noundef nonnull align 8 dereferenceable(8) [[B:%.*]], double** noundef nonnull align 8 dereferenceable(8) [[C:%.*]]) #[[ATTR1]] { 5340 // CHECK9-NEXT: entry: 5341 // CHECK9-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 5342 // CHECK9-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 5343 // CHECK9-NEXT: [[N_ADDR:%.*]] = alloca i32*, align 8 5344 // CHECK9-NEXT: [[A_ADDR:%.*]] = alloca double**, align 8 5345 // CHECK9-NEXT: [[B_ADDR:%.*]] = alloca double**, align 8 5346 // CHECK9-NEXT: [[C_ADDR:%.*]] = alloca double**, align 8 5347 // CHECK9-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 5348 // CHECK9-NEXT: [[TMP:%.*]] = alloca i32, align 4 5349 // CHECK9-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 5350 // CHECK9-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 5351 // CHECK9-NEXT: [[I:%.*]] = alloca i32, align 4 5352 // CHECK9-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4 5353 // CHECK9-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4 5354 // CHECK9-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 5355 // CHECK9-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 5356 // CHECK9-NEXT: [[I3:%.*]] = alloca i32, align 4 5357 // CHECK9-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8 5358 // CHECK9-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8 5359 // CHECK9-NEXT: store i32* [[N]], i32** [[N_ADDR]], align 8 5360 // CHECK9-NEXT: store double** [[A]], double*** [[A_ADDR]], align 8 5361 // CHECK9-NEXT: store double** [[B]], double*** [[B_ADDR]], align 8 5362 // CHECK9-NEXT: store double** [[C]], double*** [[C_ADDR]], align 8 5363 // CHECK9-NEXT: [[TMP0:%.*]] = load i32*, i32** [[N_ADDR]], align 8 5364 // CHECK9-NEXT: [[TMP1:%.*]] = load double**, double*** [[A_ADDR]], align 8 5365 // CHECK9-NEXT: [[TMP2:%.*]] = load double**, double*** [[B_ADDR]], align 8 5366 // CHECK9-NEXT: [[TMP3:%.*]] = load double**, double*** [[C_ADDR]], align 8 5367 // CHECK9-NEXT: [[TMP4:%.*]] = load i32, i32* [[TMP0]], align 4 5368 // CHECK9-NEXT: store i32 [[TMP4]], i32* [[DOTCAPTURE_EXPR_]], align 4 5369 // CHECK9-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 5370 // CHECK9-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP5]], 0 5371 // CHECK9-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 5372 // CHECK9-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1 5373 // CHECK9-NEXT: store i32 [[SUB2]], i32* [[DOTCAPTURE_EXPR_1]], align 4 5374 // CHECK9-NEXT: store i32 0, i32* [[I]], align 4 5375 // CHECK9-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 5376 // CHECK9-NEXT: [[CMP:%.*]] = icmp slt i32 0, [[TMP6]] 5377 // CHECK9-NEXT: br i1 [[CMP]], label [[OMP_PRECOND_THEN:%.*]], label [[OMP_PRECOND_END:%.*]] 5378 // CHECK9: omp.precond.then: 5379 // CHECK9-NEXT: store i32 0, i32* [[DOTOMP_COMB_LB]], align 4 5380 // CHECK9-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 5381 // CHECK9-NEXT: store i32 [[TMP7]], i32* [[DOTOMP_COMB_UB]], align 4 5382 // CHECK9-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4 5383 // CHECK9-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4 5384 // CHECK9-NEXT: [[TMP8:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 5385 // CHECK9-NEXT: [[TMP9:%.*]] = load i32, i32* [[TMP8]], align 4 5386 // CHECK9-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1:[0-9]+]], i32 [[TMP9]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_COMB_LB]], i32* [[DOTOMP_COMB_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1) 5387 // CHECK9-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 5388 // CHECK9-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 5389 // CHECK9-NEXT: [[CMP4:%.*]] = icmp sgt i32 [[TMP10]], [[TMP11]] 5390 // CHECK9-NEXT: br i1 [[CMP4]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] 5391 // CHECK9: cond.true: 5392 // CHECK9-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 5393 // CHECK9-NEXT: br label [[COND_END:%.*]] 5394 // CHECK9: cond.false: 5395 // CHECK9-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 5396 // CHECK9-NEXT: br label [[COND_END]] 5397 // CHECK9: cond.end: 5398 // CHECK9-NEXT: [[COND:%.*]] = phi i32 [ [[TMP12]], [[COND_TRUE]] ], [ [[TMP13]], [[COND_FALSE]] ] 5399 // CHECK9-NEXT: store i32 [[COND]], i32* [[DOTOMP_COMB_UB]], align 4 5400 // CHECK9-NEXT: [[TMP14:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4 5401 // CHECK9-NEXT: store i32 [[TMP14]], i32* [[DOTOMP_IV]], align 4 5402 // CHECK9-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] 5403 // CHECK9: omp.inner.for.cond: 5404 // CHECK9-NEXT: [[TMP15:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !17 5405 // CHECK9-NEXT: [[TMP16:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4, !llvm.access.group !17 5406 // CHECK9-NEXT: [[CMP5:%.*]] = icmp sle i32 [[TMP15]], [[TMP16]] 5407 // CHECK9-NEXT: br i1 [[CMP5]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] 5408 // CHECK9: omp.inner.for.body: 5409 // CHECK9-NEXT: [[TMP17:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4, !llvm.access.group !17 5410 // CHECK9-NEXT: [[TMP18:%.*]] = zext i32 [[TMP17]] to i64 5411 // CHECK9-NEXT: [[TMP19:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4, !llvm.access.group !17 5412 // CHECK9-NEXT: [[TMP20:%.*]] = zext i32 [[TMP19]] to i64 5413 // CHECK9-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 6, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64, i32*, double**, double**, double**)* @.omp_outlined..1 to void (i32*, i32*, ...)*), i64 [[TMP18]], i64 [[TMP20]], i32* [[TMP0]], double** [[TMP1]], double** [[TMP2]], double** [[TMP3]]), !llvm.access.group !17 5414 // CHECK9-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] 5415 // CHECK9: omp.inner.for.inc: 5416 // CHECK9-NEXT: [[TMP21:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !17 5417 // CHECK9-NEXT: [[TMP22:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4, !llvm.access.group !17 5418 // CHECK9-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP21]], [[TMP22]] 5419 // CHECK9-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4, !llvm.access.group !17 5420 // CHECK9-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP18:![0-9]+]] 5421 // CHECK9: omp.inner.for.end: 5422 // CHECK9-NEXT: br label [[OMP_LOOP_EXIT:%.*]] 5423 // CHECK9: omp.loop.exit: 5424 // CHECK9-NEXT: [[TMP23:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 5425 // CHECK9-NEXT: [[TMP24:%.*]] = load i32, i32* [[TMP23]], align 4 5426 // CHECK9-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP24]]) 5427 // CHECK9-NEXT: [[TMP25:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4 5428 // CHECK9-NEXT: [[TMP26:%.*]] = icmp ne i32 [[TMP25]], 0 5429 // CHECK9-NEXT: br i1 [[TMP26]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]] 5430 // CHECK9: .omp.final.then: 5431 // CHECK9-NEXT: [[TMP27:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 5432 // CHECK9-NEXT: [[SUB6:%.*]] = sub nsw i32 [[TMP27]], 0 5433 // CHECK9-NEXT: [[DIV7:%.*]] = sdiv i32 [[SUB6]], 1 5434 // CHECK9-NEXT: [[MUL:%.*]] = mul nsw i32 [[DIV7]], 1 5435 // CHECK9-NEXT: [[ADD8:%.*]] = add nsw i32 0, [[MUL]] 5436 // CHECK9-NEXT: store i32 [[ADD8]], i32* [[I3]], align 4 5437 // CHECK9-NEXT: br label [[DOTOMP_FINAL_DONE]] 5438 // CHECK9: .omp.final.done: 5439 // CHECK9-NEXT: br label [[OMP_PRECOND_END]] 5440 // CHECK9: omp.precond.end: 5441 // CHECK9-NEXT: ret void 5442 // 5443 // 5444 // CHECK9-LABEL: define {{[^@]+}}@.omp_outlined..1 5445 // CHECK9-SAME: (i32* noalias noundef [[DOTGLOBAL_TID_:%.*]], i32* noalias noundef [[DOTBOUND_TID_:%.*]], i64 noundef [[DOTPREVIOUS_LB_:%.*]], i64 noundef [[DOTPREVIOUS_UB_:%.*]], i32* noundef nonnull align 4 dereferenceable(4) [[N:%.*]], double** noundef nonnull align 8 dereferenceable(8) [[A:%.*]], double** noundef nonnull align 8 dereferenceable(8) [[B:%.*]], double** noundef nonnull align 8 dereferenceable(8) [[C:%.*]]) #[[ATTR1]] { 5446 // CHECK9-NEXT: entry: 5447 // CHECK9-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 5448 // CHECK9-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 5449 // CHECK9-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8 5450 // CHECK9-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8 5451 // CHECK9-NEXT: [[N_ADDR:%.*]] = alloca i32*, align 8 5452 // CHECK9-NEXT: [[A_ADDR:%.*]] = alloca double**, align 8 5453 // CHECK9-NEXT: [[B_ADDR:%.*]] = alloca double**, align 8 5454 // CHECK9-NEXT: [[C_ADDR:%.*]] = alloca double**, align 8 5455 // CHECK9-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 5456 // CHECK9-NEXT: [[TMP:%.*]] = alloca i32, align 4 5457 // CHECK9-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 5458 // CHECK9-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 5459 // CHECK9-NEXT: [[I:%.*]] = alloca i32, align 4 5460 // CHECK9-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 5461 // CHECK9-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4 5462 // CHECK9-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 5463 // CHECK9-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 5464 // CHECK9-NEXT: [[I4:%.*]] = alloca i32, align 4 5465 // CHECK9-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8 5466 // CHECK9-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8 5467 // CHECK9-NEXT: store i64 [[DOTPREVIOUS_LB_]], i64* [[DOTPREVIOUS_LB__ADDR]], align 8 5468 // CHECK9-NEXT: store i64 [[DOTPREVIOUS_UB_]], i64* [[DOTPREVIOUS_UB__ADDR]], align 8 5469 // CHECK9-NEXT: store i32* [[N]], i32** [[N_ADDR]], align 8 5470 // CHECK9-NEXT: store double** [[A]], double*** [[A_ADDR]], align 8 5471 // CHECK9-NEXT: store double** [[B]], double*** [[B_ADDR]], align 8 5472 // CHECK9-NEXT: store double** [[C]], double*** [[C_ADDR]], align 8 5473 // CHECK9-NEXT: [[TMP0:%.*]] = load i32*, i32** [[N_ADDR]], align 8 5474 // CHECK9-NEXT: [[TMP1:%.*]] = load double**, double*** [[A_ADDR]], align 8 5475 // CHECK9-NEXT: [[TMP2:%.*]] = load double**, double*** [[B_ADDR]], align 8 5476 // CHECK9-NEXT: [[TMP3:%.*]] = load double**, double*** [[C_ADDR]], align 8 5477 // CHECK9-NEXT: [[TMP4:%.*]] = load i32, i32* [[TMP0]], align 4 5478 // CHECK9-NEXT: store i32 [[TMP4]], i32* [[DOTCAPTURE_EXPR_]], align 4 5479 // CHECK9-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 5480 // CHECK9-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP5]], 0 5481 // CHECK9-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 5482 // CHECK9-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1 5483 // CHECK9-NEXT: store i32 [[SUB2]], i32* [[DOTCAPTURE_EXPR_1]], align 4 5484 // CHECK9-NEXT: store i32 0, i32* [[I]], align 4 5485 // CHECK9-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 5486 // CHECK9-NEXT: [[CMP:%.*]] = icmp slt i32 0, [[TMP6]] 5487 // CHECK9-NEXT: br i1 [[CMP]], label [[OMP_PRECOND_THEN:%.*]], label [[OMP_PRECOND_END:%.*]] 5488 // CHECK9: omp.precond.then: 5489 // CHECK9-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4 5490 // CHECK9-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 5491 // CHECK9-NEXT: store i32 [[TMP7]], i32* [[DOTOMP_UB]], align 4 5492 // CHECK9-NEXT: [[TMP8:%.*]] = load i64, i64* [[DOTPREVIOUS_LB__ADDR]], align 8 5493 // CHECK9-NEXT: [[CONV:%.*]] = trunc i64 [[TMP8]] to i32 5494 // CHECK9-NEXT: [[TMP9:%.*]] = load i64, i64* [[DOTPREVIOUS_UB__ADDR]], align 8 5495 // CHECK9-NEXT: [[CONV3:%.*]] = trunc i64 [[TMP9]] to i32 5496 // CHECK9-NEXT: store i32 [[CONV]], i32* [[DOTOMP_LB]], align 4 5497 // CHECK9-NEXT: store i32 [[CONV3]], i32* [[DOTOMP_UB]], align 4 5498 // CHECK9-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4 5499 // CHECK9-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4 5500 // CHECK9-NEXT: [[TMP10:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 5501 // CHECK9-NEXT: [[TMP11:%.*]] = load i32, i32* [[TMP10]], align 4 5502 // CHECK9-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB2:[0-9]+]], i32 [[TMP11]], i32 34, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1) 5503 // CHECK9-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 5504 // CHECK9-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 5505 // CHECK9-NEXT: [[CMP5:%.*]] = icmp sgt i32 [[TMP12]], [[TMP13]] 5506 // CHECK9-NEXT: br i1 [[CMP5]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] 5507 // CHECK9: cond.true: 5508 // CHECK9-NEXT: [[TMP14:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 5509 // CHECK9-NEXT: br label [[COND_END:%.*]] 5510 // CHECK9: cond.false: 5511 // CHECK9-NEXT: [[TMP15:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 5512 // CHECK9-NEXT: br label [[COND_END]] 5513 // CHECK9: cond.end: 5514 // CHECK9-NEXT: [[COND:%.*]] = phi i32 [ [[TMP14]], [[COND_TRUE]] ], [ [[TMP15]], [[COND_FALSE]] ] 5515 // CHECK9-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4 5516 // CHECK9-NEXT: [[TMP16:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4 5517 // CHECK9-NEXT: store i32 [[TMP16]], i32* [[DOTOMP_IV]], align 4 5518 // CHECK9-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] 5519 // CHECK9: omp.inner.for.cond: 5520 // CHECK9-NEXT: [[TMP17:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !21 5521 // CHECK9-NEXT: [[TMP18:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4, !llvm.access.group !21 5522 // CHECK9-NEXT: [[CMP6:%.*]] = icmp sle i32 [[TMP17]], [[TMP18]] 5523 // CHECK9-NEXT: br i1 [[CMP6]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] 5524 // CHECK9: omp.inner.for.body: 5525 // CHECK9-NEXT: [[TMP19:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !21 5526 // CHECK9-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP19]], 1 5527 // CHECK9-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] 5528 // CHECK9-NEXT: store i32 [[ADD]], i32* [[I4]], align 4, !llvm.access.group !21 5529 // CHECK9-NEXT: [[TMP20:%.*]] = load double*, double** [[TMP2]], align 8, !llvm.access.group !21 5530 // CHECK9-NEXT: [[TMP21:%.*]] = load i32, i32* [[I4]], align 4, !llvm.access.group !21 5531 // CHECK9-NEXT: [[IDXPROM:%.*]] = sext i32 [[TMP21]] to i64 5532 // CHECK9-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds double, double* [[TMP20]], i64 [[IDXPROM]] 5533 // CHECK9-NEXT: [[TMP22:%.*]] = load double, double* [[ARRAYIDX]], align 8, !llvm.access.group !21 5534 // CHECK9-NEXT: [[TMP23:%.*]] = load double*, double** [[TMP3]], align 8, !llvm.access.group !21 5535 // CHECK9-NEXT: [[TMP24:%.*]] = load i32, i32* [[I4]], align 4, !llvm.access.group !21 5536 // CHECK9-NEXT: [[IDXPROM7:%.*]] = sext i32 [[TMP24]] to i64 5537 // CHECK9-NEXT: [[ARRAYIDX8:%.*]] = getelementptr inbounds double, double* [[TMP23]], i64 [[IDXPROM7]] 5538 // CHECK9-NEXT: [[TMP25:%.*]] = load double, double* [[ARRAYIDX8]], align 8, !llvm.access.group !21 5539 // CHECK9-NEXT: [[ADD9:%.*]] = fadd double [[TMP22]], [[TMP25]] 5540 // CHECK9-NEXT: [[TMP26:%.*]] = load double*, double** [[TMP1]], align 8, !llvm.access.group !21 5541 // CHECK9-NEXT: [[TMP27:%.*]] = load i32, i32* [[I4]], align 4, !llvm.access.group !21 5542 // CHECK9-NEXT: [[IDXPROM10:%.*]] = sext i32 [[TMP27]] to i64 5543 // CHECK9-NEXT: [[ARRAYIDX11:%.*]] = getelementptr inbounds double, double* [[TMP26]], i64 [[IDXPROM10]] 5544 // CHECK9-NEXT: store double [[ADD9]], double* [[ARRAYIDX11]], align 8, !llvm.access.group !21 5545 // CHECK9-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] 5546 // CHECK9: omp.body.continue: 5547 // CHECK9-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] 5548 // CHECK9: omp.inner.for.inc: 5549 // CHECK9-NEXT: [[TMP28:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !21 5550 // CHECK9-NEXT: [[ADD12:%.*]] = add nsw i32 [[TMP28]], 1 5551 // CHECK9-NEXT: store i32 [[ADD12]], i32* [[DOTOMP_IV]], align 4, !llvm.access.group !21 5552 // CHECK9-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP22:![0-9]+]] 5553 // CHECK9: omp.inner.for.end: 5554 // CHECK9-NEXT: br label [[OMP_LOOP_EXIT:%.*]] 5555 // CHECK9: omp.loop.exit: 5556 // CHECK9-NEXT: [[TMP29:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 5557 // CHECK9-NEXT: [[TMP30:%.*]] = load i32, i32* [[TMP29]], align 4 5558 // CHECK9-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP30]]) 5559 // CHECK9-NEXT: [[TMP31:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4 5560 // CHECK9-NEXT: [[TMP32:%.*]] = icmp ne i32 [[TMP31]], 0 5561 // CHECK9-NEXT: br i1 [[TMP32]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]] 5562 // CHECK9: .omp.final.then: 5563 // CHECK9-NEXT: [[TMP33:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 5564 // CHECK9-NEXT: [[SUB13:%.*]] = sub nsw i32 [[TMP33]], 0 5565 // CHECK9-NEXT: [[DIV14:%.*]] = sdiv i32 [[SUB13]], 1 5566 // CHECK9-NEXT: [[MUL15:%.*]] = mul nsw i32 [[DIV14]], 1 5567 // CHECK9-NEXT: [[ADD16:%.*]] = add nsw i32 0, [[MUL15]] 5568 // CHECK9-NEXT: store i32 [[ADD16]], i32* [[I4]], align 4 5569 // CHECK9-NEXT: br label [[DOTOMP_FINAL_DONE]] 5570 // CHECK9: .omp.final.done: 5571 // CHECK9-NEXT: br label [[OMP_PRECOND_END]] 5572 // CHECK9: omp.precond.end: 5573 // CHECK9-NEXT: ret void 5574 // 5575 // 5576 // CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l407 5577 // CHECK9-SAME: (i64 noundef [[N:%.*]], double* noundef [[A:%.*]], double* noundef [[B:%.*]], double* noundef [[C:%.*]]) #[[ATTR1]] { 5578 // CHECK9-NEXT: entry: 5579 // CHECK9-NEXT: [[N_ADDR:%.*]] = alloca i64, align 8 5580 // CHECK9-NEXT: [[A_ADDR:%.*]] = alloca double*, align 8 5581 // CHECK9-NEXT: [[B_ADDR:%.*]] = alloca double*, align 8 5582 // CHECK9-NEXT: [[C_ADDR:%.*]] = alloca double*, align 8 5583 // CHECK9-NEXT: store i64 [[N]], i64* [[N_ADDR]], align 8 5584 // CHECK9-NEXT: store double* [[A]], double** [[A_ADDR]], align 8 5585 // CHECK9-NEXT: store double* [[B]], double** [[B_ADDR]], align 8 5586 // CHECK9-NEXT: store double* [[C]], double** [[C_ADDR]], align 8 5587 // CHECK9-NEXT: [[CONV:%.*]] = bitcast i64* [[N_ADDR]] to i32* 5588 // CHECK9-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 4, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32*, double**, double**, double**)* @.omp_outlined..2 to void (i32*, i32*, ...)*), i32* [[CONV]], double** [[A_ADDR]], double** [[B_ADDR]], double** [[C_ADDR]]) 5589 // CHECK9-NEXT: ret void 5590 // 5591 // 5592 // CHECK9-LABEL: define {{[^@]+}}@.omp_outlined..2 5593 // CHECK9-SAME: (i32* noalias noundef [[DOTGLOBAL_TID_:%.*]], i32* noalias noundef [[DOTBOUND_TID_:%.*]], i32* noundef nonnull align 4 dereferenceable(4) [[N:%.*]], double** noundef nonnull align 8 dereferenceable(8) [[A:%.*]], double** noundef nonnull align 8 dereferenceable(8) [[B:%.*]], double** noundef nonnull align 8 dereferenceable(8) [[C:%.*]]) #[[ATTR1]] { 5594 // CHECK9-NEXT: entry: 5595 // CHECK9-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 5596 // CHECK9-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 5597 // CHECK9-NEXT: [[N_ADDR:%.*]] = alloca i32*, align 8 5598 // CHECK9-NEXT: [[A_ADDR:%.*]] = alloca double**, align 8 5599 // CHECK9-NEXT: [[B_ADDR:%.*]] = alloca double**, align 8 5600 // CHECK9-NEXT: [[C_ADDR:%.*]] = alloca double**, align 8 5601 // CHECK9-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 5602 // CHECK9-NEXT: [[TMP:%.*]] = alloca i32, align 4 5603 // CHECK9-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 5604 // CHECK9-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 5605 // CHECK9-NEXT: [[I:%.*]] = alloca i32, align 4 5606 // CHECK9-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4 5607 // CHECK9-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4 5608 // CHECK9-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 5609 // CHECK9-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 5610 // CHECK9-NEXT: [[I3:%.*]] = alloca i32, align 4 5611 // CHECK9-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8 5612 // CHECK9-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8 5613 // CHECK9-NEXT: store i32* [[N]], i32** [[N_ADDR]], align 8 5614 // CHECK9-NEXT: store double** [[A]], double*** [[A_ADDR]], align 8 5615 // CHECK9-NEXT: store double** [[B]], double*** [[B_ADDR]], align 8 5616 // CHECK9-NEXT: store double** [[C]], double*** [[C_ADDR]], align 8 5617 // CHECK9-NEXT: [[TMP0:%.*]] = load i32*, i32** [[N_ADDR]], align 8 5618 // CHECK9-NEXT: [[TMP1:%.*]] = load double**, double*** [[A_ADDR]], align 8 5619 // CHECK9-NEXT: [[TMP2:%.*]] = load double**, double*** [[B_ADDR]], align 8 5620 // CHECK9-NEXT: [[TMP3:%.*]] = load double**, double*** [[C_ADDR]], align 8 5621 // CHECK9-NEXT: [[TMP4:%.*]] = load i32, i32* [[TMP0]], align 4 5622 // CHECK9-NEXT: store i32 [[TMP4]], i32* [[DOTCAPTURE_EXPR_]], align 4 5623 // CHECK9-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 5624 // CHECK9-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP5]], 0 5625 // CHECK9-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 5626 // CHECK9-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1 5627 // CHECK9-NEXT: store i32 [[SUB2]], i32* [[DOTCAPTURE_EXPR_1]], align 4 5628 // CHECK9-NEXT: store i32 0, i32* [[I]], align 4 5629 // CHECK9-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 5630 // CHECK9-NEXT: [[CMP:%.*]] = icmp slt i32 0, [[TMP6]] 5631 // CHECK9-NEXT: br i1 [[CMP]], label [[OMP_PRECOND_THEN:%.*]], label [[OMP_PRECOND_END:%.*]] 5632 // CHECK9: omp.precond.then: 5633 // CHECK9-NEXT: store i32 0, i32* [[DOTOMP_COMB_LB]], align 4 5634 // CHECK9-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 5635 // CHECK9-NEXT: store i32 [[TMP7]], i32* [[DOTOMP_COMB_UB]], align 4 5636 // CHECK9-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4 5637 // CHECK9-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4 5638 // CHECK9-NEXT: [[TMP8:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 5639 // CHECK9-NEXT: [[TMP9:%.*]] = load i32, i32* [[TMP8]], align 4 5640 // CHECK9-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1]], i32 [[TMP9]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_COMB_LB]], i32* [[DOTOMP_COMB_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1) 5641 // CHECK9-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 5642 // CHECK9-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 5643 // CHECK9-NEXT: [[CMP4:%.*]] = icmp sgt i32 [[TMP10]], [[TMP11]] 5644 // CHECK9-NEXT: br i1 [[CMP4]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] 5645 // CHECK9: cond.true: 5646 // CHECK9-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 5647 // CHECK9-NEXT: br label [[COND_END:%.*]] 5648 // CHECK9: cond.false: 5649 // CHECK9-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 5650 // CHECK9-NEXT: br label [[COND_END]] 5651 // CHECK9: cond.end: 5652 // CHECK9-NEXT: [[COND:%.*]] = phi i32 [ [[TMP12]], [[COND_TRUE]] ], [ [[TMP13]], [[COND_FALSE]] ] 5653 // CHECK9-NEXT: store i32 [[COND]], i32* [[DOTOMP_COMB_UB]], align 4 5654 // CHECK9-NEXT: [[TMP14:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4 5655 // CHECK9-NEXT: store i32 [[TMP14]], i32* [[DOTOMP_IV]], align 4 5656 // CHECK9-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] 5657 // CHECK9: omp.inner.for.cond: 5658 // CHECK9-NEXT: [[TMP15:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !26 5659 // CHECK9-NEXT: [[TMP16:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4, !llvm.access.group !26 5660 // CHECK9-NEXT: [[CMP5:%.*]] = icmp sle i32 [[TMP15]], [[TMP16]] 5661 // CHECK9-NEXT: br i1 [[CMP5]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] 5662 // CHECK9: omp.inner.for.body: 5663 // CHECK9-NEXT: [[TMP17:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4, !llvm.access.group !26 5664 // CHECK9-NEXT: [[TMP18:%.*]] = zext i32 [[TMP17]] to i64 5665 // CHECK9-NEXT: [[TMP19:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4, !llvm.access.group !26 5666 // CHECK9-NEXT: [[TMP20:%.*]] = zext i32 [[TMP19]] to i64 5667 // CHECK9-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 6, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64, i32*, double**, double**, double**)* @.omp_outlined..3 to void (i32*, i32*, ...)*), i64 [[TMP18]], i64 [[TMP20]], i32* [[TMP0]], double** [[TMP1]], double** [[TMP2]], double** [[TMP3]]), !llvm.access.group !26 5668 // CHECK9-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] 5669 // CHECK9: omp.inner.for.inc: 5670 // CHECK9-NEXT: [[TMP21:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !26 5671 // CHECK9-NEXT: [[TMP22:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4, !llvm.access.group !26 5672 // CHECK9-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP21]], [[TMP22]] 5673 // CHECK9-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4, !llvm.access.group !26 5674 // CHECK9-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP27:![0-9]+]] 5675 // CHECK9: omp.inner.for.end: 5676 // CHECK9-NEXT: br label [[OMP_LOOP_EXIT:%.*]] 5677 // CHECK9: omp.loop.exit: 5678 // CHECK9-NEXT: [[TMP23:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 5679 // CHECK9-NEXT: [[TMP24:%.*]] = load i32, i32* [[TMP23]], align 4 5680 // CHECK9-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP24]]) 5681 // CHECK9-NEXT: [[TMP25:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4 5682 // CHECK9-NEXT: [[TMP26:%.*]] = icmp ne i32 [[TMP25]], 0 5683 // CHECK9-NEXT: br i1 [[TMP26]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]] 5684 // CHECK9: .omp.final.then: 5685 // CHECK9-NEXT: [[TMP27:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 5686 // CHECK9-NEXT: [[SUB6:%.*]] = sub nsw i32 [[TMP27]], 0 5687 // CHECK9-NEXT: [[DIV7:%.*]] = sdiv i32 [[SUB6]], 1 5688 // CHECK9-NEXT: [[MUL:%.*]] = mul nsw i32 [[DIV7]], 1 5689 // CHECK9-NEXT: [[ADD8:%.*]] = add nsw i32 0, [[MUL]] 5690 // CHECK9-NEXT: store i32 [[ADD8]], i32* [[I3]], align 4 5691 // CHECK9-NEXT: br label [[DOTOMP_FINAL_DONE]] 5692 // CHECK9: .omp.final.done: 5693 // CHECK9-NEXT: br label [[OMP_PRECOND_END]] 5694 // CHECK9: omp.precond.end: 5695 // CHECK9-NEXT: ret void 5696 // 5697 // 5698 // CHECK9-LABEL: define {{[^@]+}}@.omp_outlined..3 5699 // CHECK9-SAME: (i32* noalias noundef [[DOTGLOBAL_TID_:%.*]], i32* noalias noundef [[DOTBOUND_TID_:%.*]], i64 noundef [[DOTPREVIOUS_LB_:%.*]], i64 noundef [[DOTPREVIOUS_UB_:%.*]], i32* noundef nonnull align 4 dereferenceable(4) [[N:%.*]], double** noundef nonnull align 8 dereferenceable(8) [[A:%.*]], double** noundef nonnull align 8 dereferenceable(8) [[B:%.*]], double** noundef nonnull align 8 dereferenceable(8) [[C:%.*]]) #[[ATTR1]] { 5700 // CHECK9-NEXT: entry: 5701 // CHECK9-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 5702 // CHECK9-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 5703 // CHECK9-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8 5704 // CHECK9-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8 5705 // CHECK9-NEXT: [[N_ADDR:%.*]] = alloca i32*, align 8 5706 // CHECK9-NEXT: [[A_ADDR:%.*]] = alloca double**, align 8 5707 // CHECK9-NEXT: [[B_ADDR:%.*]] = alloca double**, align 8 5708 // CHECK9-NEXT: [[C_ADDR:%.*]] = alloca double**, align 8 5709 // CHECK9-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 5710 // CHECK9-NEXT: [[TMP:%.*]] = alloca i32, align 4 5711 // CHECK9-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 5712 // CHECK9-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 5713 // CHECK9-NEXT: [[I:%.*]] = alloca i32, align 4 5714 // CHECK9-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 5715 // CHECK9-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4 5716 // CHECK9-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 5717 // CHECK9-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 5718 // CHECK9-NEXT: [[I4:%.*]] = alloca i32, align 4 5719 // CHECK9-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8 5720 // CHECK9-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8 5721 // CHECK9-NEXT: store i64 [[DOTPREVIOUS_LB_]], i64* [[DOTPREVIOUS_LB__ADDR]], align 8 5722 // CHECK9-NEXT: store i64 [[DOTPREVIOUS_UB_]], i64* [[DOTPREVIOUS_UB__ADDR]], align 8 5723 // CHECK9-NEXT: store i32* [[N]], i32** [[N_ADDR]], align 8 5724 // CHECK9-NEXT: store double** [[A]], double*** [[A_ADDR]], align 8 5725 // CHECK9-NEXT: store double** [[B]], double*** [[B_ADDR]], align 8 5726 // CHECK9-NEXT: store double** [[C]], double*** [[C_ADDR]], align 8 5727 // CHECK9-NEXT: [[TMP0:%.*]] = load i32*, i32** [[N_ADDR]], align 8 5728 // CHECK9-NEXT: [[TMP1:%.*]] = load double**, double*** [[A_ADDR]], align 8 5729 // CHECK9-NEXT: [[TMP2:%.*]] = load double**, double*** [[B_ADDR]], align 8 5730 // CHECK9-NEXT: [[TMP3:%.*]] = load double**, double*** [[C_ADDR]], align 8 5731 // CHECK9-NEXT: [[TMP4:%.*]] = load i32, i32* [[TMP0]], align 4 5732 // CHECK9-NEXT: store i32 [[TMP4]], i32* [[DOTCAPTURE_EXPR_]], align 4 5733 // CHECK9-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 5734 // CHECK9-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP5]], 0 5735 // CHECK9-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 5736 // CHECK9-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1 5737 // CHECK9-NEXT: store i32 [[SUB2]], i32* [[DOTCAPTURE_EXPR_1]], align 4 5738 // CHECK9-NEXT: store i32 0, i32* [[I]], align 4 5739 // CHECK9-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 5740 // CHECK9-NEXT: [[CMP:%.*]] = icmp slt i32 0, [[TMP6]] 5741 // CHECK9-NEXT: br i1 [[CMP]], label [[OMP_PRECOND_THEN:%.*]], label [[OMP_PRECOND_END:%.*]] 5742 // CHECK9: omp.precond.then: 5743 // CHECK9-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4 5744 // CHECK9-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 5745 // CHECK9-NEXT: store i32 [[TMP7]], i32* [[DOTOMP_UB]], align 4 5746 // CHECK9-NEXT: [[TMP8:%.*]] = load i64, i64* [[DOTPREVIOUS_LB__ADDR]], align 8 5747 // CHECK9-NEXT: [[CONV:%.*]] = trunc i64 [[TMP8]] to i32 5748 // CHECK9-NEXT: [[TMP9:%.*]] = load i64, i64* [[DOTPREVIOUS_UB__ADDR]], align 8 5749 // CHECK9-NEXT: [[CONV3:%.*]] = trunc i64 [[TMP9]] to i32 5750 // CHECK9-NEXT: store i32 [[CONV]], i32* [[DOTOMP_LB]], align 4 5751 // CHECK9-NEXT: store i32 [[CONV3]], i32* [[DOTOMP_UB]], align 4 5752 // CHECK9-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4 5753 // CHECK9-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4 5754 // CHECK9-NEXT: [[TMP10:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 5755 // CHECK9-NEXT: [[TMP11:%.*]] = load i32, i32* [[TMP10]], align 4 5756 // CHECK9-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB2]], i32 [[TMP11]], i32 34, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1) 5757 // CHECK9-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 5758 // CHECK9-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 5759 // CHECK9-NEXT: [[CMP5:%.*]] = icmp sgt i32 [[TMP12]], [[TMP13]] 5760 // CHECK9-NEXT: br i1 [[CMP5]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] 5761 // CHECK9: cond.true: 5762 // CHECK9-NEXT: [[TMP14:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 5763 // CHECK9-NEXT: br label [[COND_END:%.*]] 5764 // CHECK9: cond.false: 5765 // CHECK9-NEXT: [[TMP15:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 5766 // CHECK9-NEXT: br label [[COND_END]] 5767 // CHECK9: cond.end: 5768 // CHECK9-NEXT: [[COND:%.*]] = phi i32 [ [[TMP14]], [[COND_TRUE]] ], [ [[TMP15]], [[COND_FALSE]] ] 5769 // CHECK9-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4 5770 // CHECK9-NEXT: [[TMP16:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4 5771 // CHECK9-NEXT: store i32 [[TMP16]], i32* [[DOTOMP_IV]], align 4 5772 // CHECK9-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] 5773 // CHECK9: omp.inner.for.cond: 5774 // CHECK9-NEXT: [[TMP17:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !29 5775 // CHECK9-NEXT: [[TMP18:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4, !llvm.access.group !29 5776 // CHECK9-NEXT: [[CMP6:%.*]] = icmp sle i32 [[TMP17]], [[TMP18]] 5777 // CHECK9-NEXT: br i1 [[CMP6]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] 5778 // CHECK9: omp.inner.for.body: 5779 // CHECK9-NEXT: [[TMP19:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !29 5780 // CHECK9-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP19]], 1 5781 // CHECK9-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] 5782 // CHECK9-NEXT: store i32 [[ADD]], i32* [[I4]], align 4, !llvm.access.group !29 5783 // CHECK9-NEXT: [[TMP20:%.*]] = load double*, double** [[TMP2]], align 8, !llvm.access.group !29 5784 // CHECK9-NEXT: [[TMP21:%.*]] = load i32, i32* [[I4]], align 4, !llvm.access.group !29 5785 // CHECK9-NEXT: [[IDXPROM:%.*]] = sext i32 [[TMP21]] to i64 5786 // CHECK9-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds double, double* [[TMP20]], i64 [[IDXPROM]] 5787 // CHECK9-NEXT: [[TMP22:%.*]] = load double, double* [[ARRAYIDX]], align 8, !llvm.access.group !29 5788 // CHECK9-NEXT: [[TMP23:%.*]] = load double*, double** [[TMP3]], align 8, !llvm.access.group !29 5789 // CHECK9-NEXT: [[TMP24:%.*]] = load i32, i32* [[I4]], align 4, !llvm.access.group !29 5790 // CHECK9-NEXT: [[IDXPROM7:%.*]] = sext i32 [[TMP24]] to i64 5791 // CHECK9-NEXT: [[ARRAYIDX8:%.*]] = getelementptr inbounds double, double* [[TMP23]], i64 [[IDXPROM7]] 5792 // CHECK9-NEXT: [[TMP25:%.*]] = load double, double* [[ARRAYIDX8]], align 8, !llvm.access.group !29 5793 // CHECK9-NEXT: [[ADD9:%.*]] = fadd double [[TMP22]], [[TMP25]] 5794 // CHECK9-NEXT: [[TMP26:%.*]] = load double*, double** [[TMP1]], align 8, !llvm.access.group !29 5795 // CHECK9-NEXT: [[TMP27:%.*]] = load i32, i32* [[I4]], align 4, !llvm.access.group !29 5796 // CHECK9-NEXT: [[IDXPROM10:%.*]] = sext i32 [[TMP27]] to i64 5797 // CHECK9-NEXT: [[ARRAYIDX11:%.*]] = getelementptr inbounds double, double* [[TMP26]], i64 [[IDXPROM10]] 5798 // CHECK9-NEXT: store double [[ADD9]], double* [[ARRAYIDX11]], align 8, !llvm.access.group !29 5799 // CHECK9-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] 5800 // CHECK9: omp.body.continue: 5801 // CHECK9-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] 5802 // CHECK9: omp.inner.for.inc: 5803 // CHECK9-NEXT: [[TMP28:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !29 5804 // CHECK9-NEXT: [[ADD12:%.*]] = add nsw i32 [[TMP28]], 1 5805 // CHECK9-NEXT: store i32 [[ADD12]], i32* [[DOTOMP_IV]], align 4, !llvm.access.group !29 5806 // CHECK9-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP30:![0-9]+]] 5807 // CHECK9: omp.inner.for.end: 5808 // CHECK9-NEXT: br label [[OMP_LOOP_EXIT:%.*]] 5809 // CHECK9: omp.loop.exit: 5810 // CHECK9-NEXT: [[TMP29:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 5811 // CHECK9-NEXT: [[TMP30:%.*]] = load i32, i32* [[TMP29]], align 4 5812 // CHECK9-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP30]]) 5813 // CHECK9-NEXT: [[TMP31:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4 5814 // CHECK9-NEXT: [[TMP32:%.*]] = icmp ne i32 [[TMP31]], 0 5815 // CHECK9-NEXT: br i1 [[TMP32]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]] 5816 // CHECK9: .omp.final.then: 5817 // CHECK9-NEXT: [[TMP33:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 5818 // CHECK9-NEXT: [[SUB13:%.*]] = sub nsw i32 [[TMP33]], 0 5819 // CHECK9-NEXT: [[DIV14:%.*]] = sdiv i32 [[SUB13]], 1 5820 // CHECK9-NEXT: [[MUL15:%.*]] = mul nsw i32 [[DIV14]], 1 5821 // CHECK9-NEXT: [[ADD16:%.*]] = add nsw i32 0, [[MUL15]] 5822 // CHECK9-NEXT: store i32 [[ADD16]], i32* [[I4]], align 4 5823 // CHECK9-NEXT: br label [[DOTOMP_FINAL_DONE]] 5824 // CHECK9: .omp.final.done: 5825 // CHECK9-NEXT: br label [[OMP_PRECOND_END]] 5826 // CHECK9: omp.precond.end: 5827 // CHECK9-NEXT: ret void 5828 // 5829 // 5830 // CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l446 5831 // CHECK9-SAME: (i64 noundef [[CH:%.*]], i64 noundef [[N:%.*]], double* noundef [[A:%.*]], double* noundef [[B:%.*]], double* noundef [[C:%.*]]) #[[ATTR1]] { 5832 // CHECK9-NEXT: entry: 5833 // CHECK9-NEXT: [[CH_ADDR:%.*]] = alloca i64, align 8 5834 // CHECK9-NEXT: [[N_ADDR:%.*]] = alloca i64, align 8 5835 // CHECK9-NEXT: [[A_ADDR:%.*]] = alloca double*, align 8 5836 // CHECK9-NEXT: [[B_ADDR:%.*]] = alloca double*, align 8 5837 // CHECK9-NEXT: [[C_ADDR:%.*]] = alloca double*, align 8 5838 // CHECK9-NEXT: store i64 [[CH]], i64* [[CH_ADDR]], align 8 5839 // CHECK9-NEXT: store i64 [[N]], i64* [[N_ADDR]], align 8 5840 // CHECK9-NEXT: store double* [[A]], double** [[A_ADDR]], align 8 5841 // CHECK9-NEXT: store double* [[B]], double** [[B_ADDR]], align 8 5842 // CHECK9-NEXT: store double* [[C]], double** [[C_ADDR]], align 8 5843 // CHECK9-NEXT: [[CONV:%.*]] = bitcast i64* [[CH_ADDR]] to i32* 5844 // CHECK9-NEXT: [[CONV1:%.*]] = bitcast i64* [[N_ADDR]] to i32* 5845 // CHECK9-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 5, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32*, i32*, double**, double**, double**)* @.omp_outlined..6 to void (i32*, i32*, ...)*), i32* [[CONV]], i32* [[CONV1]], double** [[A_ADDR]], double** [[B_ADDR]], double** [[C_ADDR]]) 5846 // CHECK9-NEXT: ret void 5847 // 5848 // 5849 // CHECK9-LABEL: define {{[^@]+}}@.omp_outlined..6 5850 // CHECK9-SAME: (i32* noalias noundef [[DOTGLOBAL_TID_:%.*]], i32* noalias noundef [[DOTBOUND_TID_:%.*]], i32* noundef nonnull align 4 dereferenceable(4) [[CH:%.*]], i32* noundef nonnull align 4 dereferenceable(4) [[N:%.*]], double** noundef nonnull align 8 dereferenceable(8) [[A:%.*]], double** noundef nonnull align 8 dereferenceable(8) [[B:%.*]], double** noundef nonnull align 8 dereferenceable(8) [[C:%.*]]) #[[ATTR1]] { 5851 // CHECK9-NEXT: entry: 5852 // CHECK9-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 5853 // CHECK9-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 5854 // CHECK9-NEXT: [[CH_ADDR:%.*]] = alloca i32*, align 8 5855 // CHECK9-NEXT: [[N_ADDR:%.*]] = alloca i32*, align 8 5856 // CHECK9-NEXT: [[A_ADDR:%.*]] = alloca double**, align 8 5857 // CHECK9-NEXT: [[B_ADDR:%.*]] = alloca double**, align 8 5858 // CHECK9-NEXT: [[C_ADDR:%.*]] = alloca double**, align 8 5859 // CHECK9-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 5860 // CHECK9-NEXT: [[TMP:%.*]] = alloca i32, align 4 5861 // CHECK9-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 5862 // CHECK9-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 5863 // CHECK9-NEXT: [[I:%.*]] = alloca i32, align 4 5864 // CHECK9-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4 5865 // CHECK9-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4 5866 // CHECK9-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 5867 // CHECK9-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 5868 // CHECK9-NEXT: [[I3:%.*]] = alloca i32, align 4 5869 // CHECK9-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8 5870 // CHECK9-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8 5871 // CHECK9-NEXT: store i32* [[CH]], i32** [[CH_ADDR]], align 8 5872 // CHECK9-NEXT: store i32* [[N]], i32** [[N_ADDR]], align 8 5873 // CHECK9-NEXT: store double** [[A]], double*** [[A_ADDR]], align 8 5874 // CHECK9-NEXT: store double** [[B]], double*** [[B_ADDR]], align 8 5875 // CHECK9-NEXT: store double** [[C]], double*** [[C_ADDR]], align 8 5876 // CHECK9-NEXT: [[TMP0:%.*]] = load i32*, i32** [[CH_ADDR]], align 8 5877 // CHECK9-NEXT: [[TMP1:%.*]] = load i32*, i32** [[N_ADDR]], align 8 5878 // CHECK9-NEXT: [[TMP2:%.*]] = load double**, double*** [[A_ADDR]], align 8 5879 // CHECK9-NEXT: [[TMP3:%.*]] = load double**, double*** [[B_ADDR]], align 8 5880 // CHECK9-NEXT: [[TMP4:%.*]] = load double**, double*** [[C_ADDR]], align 8 5881 // CHECK9-NEXT: [[TMP5:%.*]] = load i32, i32* [[TMP1]], align 4 5882 // CHECK9-NEXT: store i32 [[TMP5]], i32* [[DOTCAPTURE_EXPR_]], align 4 5883 // CHECK9-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 5884 // CHECK9-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP6]], 0 5885 // CHECK9-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 5886 // CHECK9-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1 5887 // CHECK9-NEXT: store i32 [[SUB2]], i32* [[DOTCAPTURE_EXPR_1]], align 4 5888 // CHECK9-NEXT: store i32 0, i32* [[I]], align 4 5889 // CHECK9-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 5890 // CHECK9-NEXT: [[CMP:%.*]] = icmp slt i32 0, [[TMP7]] 5891 // CHECK9-NEXT: br i1 [[CMP]], label [[OMP_PRECOND_THEN:%.*]], label [[OMP_PRECOND_END:%.*]] 5892 // CHECK9: omp.precond.then: 5893 // CHECK9-NEXT: store i32 0, i32* [[DOTOMP_COMB_LB]], align 4 5894 // CHECK9-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 5895 // CHECK9-NEXT: store i32 [[TMP8]], i32* [[DOTOMP_COMB_UB]], align 4 5896 // CHECK9-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4 5897 // CHECK9-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4 5898 // CHECK9-NEXT: [[TMP9:%.*]] = load i32, i32* [[TMP0]], align 4 5899 // CHECK9-NEXT: [[TMP10:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 5900 // CHECK9-NEXT: [[TMP11:%.*]] = load i32, i32* [[TMP10]], align 4 5901 // CHECK9-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1]], i32 [[TMP11]], i32 91, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_COMB_LB]], i32* [[DOTOMP_COMB_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 [[TMP9]]) 5902 // CHECK9-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 5903 // CHECK9-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 5904 // CHECK9-NEXT: [[CMP4:%.*]] = icmp sgt i32 [[TMP12]], [[TMP13]] 5905 // CHECK9-NEXT: br i1 [[CMP4]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] 5906 // CHECK9: cond.true: 5907 // CHECK9-NEXT: [[TMP14:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 5908 // CHECK9-NEXT: br label [[COND_END:%.*]] 5909 // CHECK9: cond.false: 5910 // CHECK9-NEXT: [[TMP15:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 5911 // CHECK9-NEXT: br label [[COND_END]] 5912 // CHECK9: cond.end: 5913 // CHECK9-NEXT: [[COND:%.*]] = phi i32 [ [[TMP14]], [[COND_TRUE]] ], [ [[TMP15]], [[COND_FALSE]] ] 5914 // CHECK9-NEXT: store i32 [[COND]], i32* [[DOTOMP_COMB_UB]], align 4 5915 // CHECK9-NEXT: [[TMP16:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4 5916 // CHECK9-NEXT: store i32 [[TMP16]], i32* [[DOTOMP_IV]], align 4 5917 // CHECK9-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] 5918 // CHECK9: omp.inner.for.cond: 5919 // CHECK9-NEXT: [[TMP17:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !32 5920 // CHECK9-NEXT: [[TMP18:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4, !llvm.access.group !32 5921 // CHECK9-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP18]], 1 5922 // CHECK9-NEXT: [[CMP5:%.*]] = icmp slt i32 [[TMP17]], [[ADD]] 5923 // CHECK9-NEXT: br i1 [[CMP5]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] 5924 // CHECK9: omp.inner.for.body: 5925 // CHECK9-NEXT: [[TMP19:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4, !llvm.access.group !32 5926 // CHECK9-NEXT: [[TMP20:%.*]] = zext i32 [[TMP19]] to i64 5927 // CHECK9-NEXT: [[TMP21:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4, !llvm.access.group !32 5928 // CHECK9-NEXT: [[TMP22:%.*]] = zext i32 [[TMP21]] to i64 5929 // CHECK9-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 6, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64, i32*, double**, double**, double**)* @.omp_outlined..7 to void (i32*, i32*, ...)*), i64 [[TMP20]], i64 [[TMP22]], i32* [[TMP1]], double** [[TMP2]], double** [[TMP3]], double** [[TMP4]]), !llvm.access.group !32 5930 // CHECK9-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] 5931 // CHECK9: omp.inner.for.inc: 5932 // CHECK9-NEXT: [[TMP23:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !32 5933 // CHECK9-NEXT: [[TMP24:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4, !llvm.access.group !32 5934 // CHECK9-NEXT: [[ADD6:%.*]] = add nsw i32 [[TMP23]], [[TMP24]] 5935 // CHECK9-NEXT: store i32 [[ADD6]], i32* [[DOTOMP_IV]], align 4, !llvm.access.group !32 5936 // CHECK9-NEXT: [[TMP25:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4, !llvm.access.group !32 5937 // CHECK9-NEXT: [[TMP26:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4, !llvm.access.group !32 5938 // CHECK9-NEXT: [[ADD7:%.*]] = add nsw i32 [[TMP25]], [[TMP26]] 5939 // CHECK9-NEXT: store i32 [[ADD7]], i32* [[DOTOMP_COMB_LB]], align 4, !llvm.access.group !32 5940 // CHECK9-NEXT: [[TMP27:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4, !llvm.access.group !32 5941 // CHECK9-NEXT: [[TMP28:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4, !llvm.access.group !32 5942 // CHECK9-NEXT: [[ADD8:%.*]] = add nsw i32 [[TMP27]], [[TMP28]] 5943 // CHECK9-NEXT: store i32 [[ADD8]], i32* [[DOTOMP_COMB_UB]], align 4, !llvm.access.group !32 5944 // CHECK9-NEXT: [[TMP29:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4, !llvm.access.group !32 5945 // CHECK9-NEXT: [[TMP30:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4, !llvm.access.group !32 5946 // CHECK9-NEXT: [[CMP9:%.*]] = icmp sgt i32 [[TMP29]], [[TMP30]] 5947 // CHECK9-NEXT: br i1 [[CMP9]], label [[COND_TRUE10:%.*]], label [[COND_FALSE11:%.*]] 5948 // CHECK9: cond.true10: 5949 // CHECK9-NEXT: [[TMP31:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4, !llvm.access.group !32 5950 // CHECK9-NEXT: br label [[COND_END12:%.*]] 5951 // CHECK9: cond.false11: 5952 // CHECK9-NEXT: [[TMP32:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4, !llvm.access.group !32 5953 // CHECK9-NEXT: br label [[COND_END12]] 5954 // CHECK9: cond.end12: 5955 // CHECK9-NEXT: [[COND13:%.*]] = phi i32 [ [[TMP31]], [[COND_TRUE10]] ], [ [[TMP32]], [[COND_FALSE11]] ] 5956 // CHECK9-NEXT: store i32 [[COND13]], i32* [[DOTOMP_COMB_UB]], align 4, !llvm.access.group !32 5957 // CHECK9-NEXT: [[TMP33:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4, !llvm.access.group !32 5958 // CHECK9-NEXT: store i32 [[TMP33]], i32* [[DOTOMP_IV]], align 4, !llvm.access.group !32 5959 // CHECK9-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP33:![0-9]+]] 5960 // CHECK9: omp.inner.for.end: 5961 // CHECK9-NEXT: br label [[OMP_LOOP_EXIT:%.*]] 5962 // CHECK9: omp.loop.exit: 5963 // CHECK9-NEXT: [[TMP34:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 5964 // CHECK9-NEXT: [[TMP35:%.*]] = load i32, i32* [[TMP34]], align 4 5965 // CHECK9-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP35]]) 5966 // CHECK9-NEXT: [[TMP36:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4 5967 // CHECK9-NEXT: [[TMP37:%.*]] = icmp ne i32 [[TMP36]], 0 5968 // CHECK9-NEXT: br i1 [[TMP37]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]] 5969 // CHECK9: .omp.final.then: 5970 // CHECK9-NEXT: [[TMP38:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 5971 // CHECK9-NEXT: [[SUB14:%.*]] = sub nsw i32 [[TMP38]], 0 5972 // CHECK9-NEXT: [[DIV15:%.*]] = sdiv i32 [[SUB14]], 1 5973 // CHECK9-NEXT: [[MUL:%.*]] = mul nsw i32 [[DIV15]], 1 5974 // CHECK9-NEXT: [[ADD16:%.*]] = add nsw i32 0, [[MUL]] 5975 // CHECK9-NEXT: store i32 [[ADD16]], i32* [[I3]], align 4 5976 // CHECK9-NEXT: br label [[DOTOMP_FINAL_DONE]] 5977 // CHECK9: .omp.final.done: 5978 // CHECK9-NEXT: br label [[OMP_PRECOND_END]] 5979 // CHECK9: omp.precond.end: 5980 // CHECK9-NEXT: ret void 5981 // 5982 // 5983 // CHECK9-LABEL: define {{[^@]+}}@.omp_outlined..7 5984 // CHECK9-SAME: (i32* noalias noundef [[DOTGLOBAL_TID_:%.*]], i32* noalias noundef [[DOTBOUND_TID_:%.*]], i64 noundef [[DOTPREVIOUS_LB_:%.*]], i64 noundef [[DOTPREVIOUS_UB_:%.*]], i32* noundef nonnull align 4 dereferenceable(4) [[N:%.*]], double** noundef nonnull align 8 dereferenceable(8) [[A:%.*]], double** noundef nonnull align 8 dereferenceable(8) [[B:%.*]], double** noundef nonnull align 8 dereferenceable(8) [[C:%.*]]) #[[ATTR1]] { 5985 // CHECK9-NEXT: entry: 5986 // CHECK9-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 5987 // CHECK9-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 5988 // CHECK9-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8 5989 // CHECK9-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8 5990 // CHECK9-NEXT: [[N_ADDR:%.*]] = alloca i32*, align 8 5991 // CHECK9-NEXT: [[A_ADDR:%.*]] = alloca double**, align 8 5992 // CHECK9-NEXT: [[B_ADDR:%.*]] = alloca double**, align 8 5993 // CHECK9-NEXT: [[C_ADDR:%.*]] = alloca double**, align 8 5994 // CHECK9-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 5995 // CHECK9-NEXT: [[TMP:%.*]] = alloca i32, align 4 5996 // CHECK9-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 5997 // CHECK9-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 5998 // CHECK9-NEXT: [[I:%.*]] = alloca i32, align 4 5999 // CHECK9-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 6000 // CHECK9-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4 6001 // CHECK9-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 6002 // CHECK9-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 6003 // CHECK9-NEXT: [[I4:%.*]] = alloca i32, align 4 6004 // CHECK9-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8 6005 // CHECK9-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8 6006 // CHECK9-NEXT: store i64 [[DOTPREVIOUS_LB_]], i64* [[DOTPREVIOUS_LB__ADDR]], align 8 6007 // CHECK9-NEXT: store i64 [[DOTPREVIOUS_UB_]], i64* [[DOTPREVIOUS_UB__ADDR]], align 8 6008 // CHECK9-NEXT: store i32* [[N]], i32** [[N_ADDR]], align 8 6009 // CHECK9-NEXT: store double** [[A]], double*** [[A_ADDR]], align 8 6010 // CHECK9-NEXT: store double** [[B]], double*** [[B_ADDR]], align 8 6011 // CHECK9-NEXT: store double** [[C]], double*** [[C_ADDR]], align 8 6012 // CHECK9-NEXT: [[TMP0:%.*]] = load i32*, i32** [[N_ADDR]], align 8 6013 // CHECK9-NEXT: [[TMP1:%.*]] = load double**, double*** [[A_ADDR]], align 8 6014 // CHECK9-NEXT: [[TMP2:%.*]] = load double**, double*** [[B_ADDR]], align 8 6015 // CHECK9-NEXT: [[TMP3:%.*]] = load double**, double*** [[C_ADDR]], align 8 6016 // CHECK9-NEXT: [[TMP4:%.*]] = load i32, i32* [[TMP0]], align 4 6017 // CHECK9-NEXT: store i32 [[TMP4]], i32* [[DOTCAPTURE_EXPR_]], align 4 6018 // CHECK9-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 6019 // CHECK9-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP5]], 0 6020 // CHECK9-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 6021 // CHECK9-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1 6022 // CHECK9-NEXT: store i32 [[SUB2]], i32* [[DOTCAPTURE_EXPR_1]], align 4 6023 // CHECK9-NEXT: store i32 0, i32* [[I]], align 4 6024 // CHECK9-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 6025 // CHECK9-NEXT: [[CMP:%.*]] = icmp slt i32 0, [[TMP6]] 6026 // CHECK9-NEXT: br i1 [[CMP]], label [[OMP_PRECOND_THEN:%.*]], label [[OMP_PRECOND_END:%.*]] 6027 // CHECK9: omp.precond.then: 6028 // CHECK9-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4 6029 // CHECK9-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 6030 // CHECK9-NEXT: store i32 [[TMP7]], i32* [[DOTOMP_UB]], align 4 6031 // CHECK9-NEXT: [[TMP8:%.*]] = load i64, i64* [[DOTPREVIOUS_LB__ADDR]], align 8 6032 // CHECK9-NEXT: [[CONV:%.*]] = trunc i64 [[TMP8]] to i32 6033 // CHECK9-NEXT: [[TMP9:%.*]] = load i64, i64* [[DOTPREVIOUS_UB__ADDR]], align 8 6034 // CHECK9-NEXT: [[CONV3:%.*]] = trunc i64 [[TMP9]] to i32 6035 // CHECK9-NEXT: store i32 [[CONV]], i32* [[DOTOMP_LB]], align 4 6036 // CHECK9-NEXT: store i32 [[CONV3]], i32* [[DOTOMP_UB]], align 4 6037 // CHECK9-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4 6038 // CHECK9-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4 6039 // CHECK9-NEXT: [[TMP10:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 6040 // CHECK9-NEXT: [[TMP11:%.*]] = load i32, i32* [[TMP10]], align 4 6041 // CHECK9-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB2]], i32 [[TMP11]], i32 34, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1) 6042 // CHECK9-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 6043 // CHECK9-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 6044 // CHECK9-NEXT: [[CMP5:%.*]] = icmp sgt i32 [[TMP12]], [[TMP13]] 6045 // CHECK9-NEXT: br i1 [[CMP5]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] 6046 // CHECK9: cond.true: 6047 // CHECK9-NEXT: [[TMP14:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 6048 // CHECK9-NEXT: br label [[COND_END:%.*]] 6049 // CHECK9: cond.false: 6050 // CHECK9-NEXT: [[TMP15:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 6051 // CHECK9-NEXT: br label [[COND_END]] 6052 // CHECK9: cond.end: 6053 // CHECK9-NEXT: [[COND:%.*]] = phi i32 [ [[TMP14]], [[COND_TRUE]] ], [ [[TMP15]], [[COND_FALSE]] ] 6054 // CHECK9-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4 6055 // CHECK9-NEXT: [[TMP16:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4 6056 // CHECK9-NEXT: store i32 [[TMP16]], i32* [[DOTOMP_IV]], align 4 6057 // CHECK9-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] 6058 // CHECK9: omp.inner.for.cond: 6059 // CHECK9-NEXT: [[TMP17:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !35 6060 // CHECK9-NEXT: [[TMP18:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4, !llvm.access.group !35 6061 // CHECK9-NEXT: [[CMP6:%.*]] = icmp sle i32 [[TMP17]], [[TMP18]] 6062 // CHECK9-NEXT: br i1 [[CMP6]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] 6063 // CHECK9: omp.inner.for.body: 6064 // CHECK9-NEXT: [[TMP19:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !35 6065 // CHECK9-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP19]], 1 6066 // CHECK9-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] 6067 // CHECK9-NEXT: store i32 [[ADD]], i32* [[I4]], align 4, !llvm.access.group !35 6068 // CHECK9-NEXT: [[TMP20:%.*]] = load double*, double** [[TMP2]], align 8, !llvm.access.group !35 6069 // CHECK9-NEXT: [[TMP21:%.*]] = load i32, i32* [[I4]], align 4, !llvm.access.group !35 6070 // CHECK9-NEXT: [[IDXPROM:%.*]] = sext i32 [[TMP21]] to i64 6071 // CHECK9-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds double, double* [[TMP20]], i64 [[IDXPROM]] 6072 // CHECK9-NEXT: [[TMP22:%.*]] = load double, double* [[ARRAYIDX]], align 8, !llvm.access.group !35 6073 // CHECK9-NEXT: [[TMP23:%.*]] = load double*, double** [[TMP3]], align 8, !llvm.access.group !35 6074 // CHECK9-NEXT: [[TMP24:%.*]] = load i32, i32* [[I4]], align 4, !llvm.access.group !35 6075 // CHECK9-NEXT: [[IDXPROM7:%.*]] = sext i32 [[TMP24]] to i64 6076 // CHECK9-NEXT: [[ARRAYIDX8:%.*]] = getelementptr inbounds double, double* [[TMP23]], i64 [[IDXPROM7]] 6077 // CHECK9-NEXT: [[TMP25:%.*]] = load double, double* [[ARRAYIDX8]], align 8, !llvm.access.group !35 6078 // CHECK9-NEXT: [[ADD9:%.*]] = fadd double [[TMP22]], [[TMP25]] 6079 // CHECK9-NEXT: [[TMP26:%.*]] = load double*, double** [[TMP1]], align 8, !llvm.access.group !35 6080 // CHECK9-NEXT: [[TMP27:%.*]] = load i32, i32* [[I4]], align 4, !llvm.access.group !35 6081 // CHECK9-NEXT: [[IDXPROM10:%.*]] = sext i32 [[TMP27]] to i64 6082 // CHECK9-NEXT: [[ARRAYIDX11:%.*]] = getelementptr inbounds double, double* [[TMP26]], i64 [[IDXPROM10]] 6083 // CHECK9-NEXT: store double [[ADD9]], double* [[ARRAYIDX11]], align 8, !llvm.access.group !35 6084 // CHECK9-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] 6085 // CHECK9: omp.body.continue: 6086 // CHECK9-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] 6087 // CHECK9: omp.inner.for.inc: 6088 // CHECK9-NEXT: [[TMP28:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !35 6089 // CHECK9-NEXT: [[ADD12:%.*]] = add nsw i32 [[TMP28]], 1 6090 // CHECK9-NEXT: store i32 [[ADD12]], i32* [[DOTOMP_IV]], align 4, !llvm.access.group !35 6091 // CHECK9-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP36:![0-9]+]] 6092 // CHECK9: omp.inner.for.end: 6093 // CHECK9-NEXT: br label [[OMP_LOOP_EXIT:%.*]] 6094 // CHECK9: omp.loop.exit: 6095 // CHECK9-NEXT: [[TMP29:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 6096 // CHECK9-NEXT: [[TMP30:%.*]] = load i32, i32* [[TMP29]], align 4 6097 // CHECK9-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP30]]) 6098 // CHECK9-NEXT: [[TMP31:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4 6099 // CHECK9-NEXT: [[TMP32:%.*]] = icmp ne i32 [[TMP31]], 0 6100 // CHECK9-NEXT: br i1 [[TMP32]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]] 6101 // CHECK9: .omp.final.then: 6102 // CHECK9-NEXT: [[TMP33:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 6103 // CHECK9-NEXT: [[SUB13:%.*]] = sub nsw i32 [[TMP33]], 0 6104 // CHECK9-NEXT: [[DIV14:%.*]] = sdiv i32 [[SUB13]], 1 6105 // CHECK9-NEXT: [[MUL15:%.*]] = mul nsw i32 [[DIV14]], 1 6106 // CHECK9-NEXT: [[ADD16:%.*]] = add nsw i32 0, [[MUL15]] 6107 // CHECK9-NEXT: store i32 [[ADD16]], i32* [[I4]], align 4 6108 // CHECK9-NEXT: br label [[DOTOMP_FINAL_DONE]] 6109 // CHECK9: .omp.final.done: 6110 // CHECK9-NEXT: br label [[OMP_PRECOND_END]] 6111 // CHECK9: omp.precond.end: 6112 // CHECK9-NEXT: ret void 6113 // 6114 // 6115 // CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l477 6116 // CHECK9-SAME: (i64 noundef [[N:%.*]], double* noundef [[A:%.*]], double* noundef [[B:%.*]], double* noundef [[C:%.*]]) #[[ATTR1]] { 6117 // CHECK9-NEXT: entry: 6118 // CHECK9-NEXT: [[N_ADDR:%.*]] = alloca i64, align 8 6119 // CHECK9-NEXT: [[A_ADDR:%.*]] = alloca double*, align 8 6120 // CHECK9-NEXT: [[B_ADDR:%.*]] = alloca double*, align 8 6121 // CHECK9-NEXT: [[C_ADDR:%.*]] = alloca double*, align 8 6122 // CHECK9-NEXT: store i64 [[N]], i64* [[N_ADDR]], align 8 6123 // CHECK9-NEXT: store double* [[A]], double** [[A_ADDR]], align 8 6124 // CHECK9-NEXT: store double* [[B]], double** [[B_ADDR]], align 8 6125 // CHECK9-NEXT: store double* [[C]], double** [[C_ADDR]], align 8 6126 // CHECK9-NEXT: [[CONV:%.*]] = bitcast i64* [[N_ADDR]] to i32* 6127 // CHECK9-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 4, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32*, double**, double**, double**)* @.omp_outlined..10 to void (i32*, i32*, ...)*), i32* [[CONV]], double** [[A_ADDR]], double** [[B_ADDR]], double** [[C_ADDR]]) 6128 // CHECK9-NEXT: ret void 6129 // 6130 // 6131 // CHECK9-LABEL: define {{[^@]+}}@.omp_outlined..10 6132 // CHECK9-SAME: (i32* noalias noundef [[DOTGLOBAL_TID_:%.*]], i32* noalias noundef [[DOTBOUND_TID_:%.*]], i32* noundef nonnull align 4 dereferenceable(4) [[N:%.*]], double** noundef nonnull align 8 dereferenceable(8) [[A:%.*]], double** noundef nonnull align 8 dereferenceable(8) [[B:%.*]], double** noundef nonnull align 8 dereferenceable(8) [[C:%.*]]) #[[ATTR1]] { 6133 // CHECK9-NEXT: entry: 6134 // CHECK9-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 6135 // CHECK9-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 6136 // CHECK9-NEXT: [[N_ADDR:%.*]] = alloca i32*, align 8 6137 // CHECK9-NEXT: [[A_ADDR:%.*]] = alloca double**, align 8 6138 // CHECK9-NEXT: [[B_ADDR:%.*]] = alloca double**, align 8 6139 // CHECK9-NEXT: [[C_ADDR:%.*]] = alloca double**, align 8 6140 // CHECK9-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 6141 // CHECK9-NEXT: [[TMP:%.*]] = alloca i32, align 4 6142 // CHECK9-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 6143 // CHECK9-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 6144 // CHECK9-NEXT: [[I:%.*]] = alloca i32, align 4 6145 // CHECK9-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4 6146 // CHECK9-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4 6147 // CHECK9-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 6148 // CHECK9-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 6149 // CHECK9-NEXT: [[I3:%.*]] = alloca i32, align 4 6150 // CHECK9-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8 6151 // CHECK9-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8 6152 // CHECK9-NEXT: store i32* [[N]], i32** [[N_ADDR]], align 8 6153 // CHECK9-NEXT: store double** [[A]], double*** [[A_ADDR]], align 8 6154 // CHECK9-NEXT: store double** [[B]], double*** [[B_ADDR]], align 8 6155 // CHECK9-NEXT: store double** [[C]], double*** [[C_ADDR]], align 8 6156 // CHECK9-NEXT: [[TMP0:%.*]] = load i32*, i32** [[N_ADDR]], align 8 6157 // CHECK9-NEXT: [[TMP1:%.*]] = load double**, double*** [[A_ADDR]], align 8 6158 // CHECK9-NEXT: [[TMP2:%.*]] = load double**, double*** [[B_ADDR]], align 8 6159 // CHECK9-NEXT: [[TMP3:%.*]] = load double**, double*** [[C_ADDR]], align 8 6160 // CHECK9-NEXT: [[TMP4:%.*]] = load i32, i32* [[TMP0]], align 4 6161 // CHECK9-NEXT: store i32 [[TMP4]], i32* [[DOTCAPTURE_EXPR_]], align 4 6162 // CHECK9-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 6163 // CHECK9-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP5]], 0 6164 // CHECK9-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 6165 // CHECK9-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1 6166 // CHECK9-NEXT: store i32 [[SUB2]], i32* [[DOTCAPTURE_EXPR_1]], align 4 6167 // CHECK9-NEXT: store i32 0, i32* [[I]], align 4 6168 // CHECK9-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 6169 // CHECK9-NEXT: [[CMP:%.*]] = icmp slt i32 0, [[TMP6]] 6170 // CHECK9-NEXT: br i1 [[CMP]], label [[OMP_PRECOND_THEN:%.*]], label [[OMP_PRECOND_END:%.*]] 6171 // CHECK9: omp.precond.then: 6172 // CHECK9-NEXT: store i32 0, i32* [[DOTOMP_COMB_LB]], align 4 6173 // CHECK9-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 6174 // CHECK9-NEXT: store i32 [[TMP7]], i32* [[DOTOMP_COMB_UB]], align 4 6175 // CHECK9-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4 6176 // CHECK9-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4 6177 // CHECK9-NEXT: [[TMP8:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 6178 // CHECK9-NEXT: [[TMP9:%.*]] = load i32, i32* [[TMP8]], align 4 6179 // CHECK9-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1]], i32 [[TMP9]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_COMB_LB]], i32* [[DOTOMP_COMB_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1) 6180 // CHECK9-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 6181 // CHECK9-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 6182 // CHECK9-NEXT: [[CMP4:%.*]] = icmp sgt i32 [[TMP10]], [[TMP11]] 6183 // CHECK9-NEXT: br i1 [[CMP4]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] 6184 // CHECK9: cond.true: 6185 // CHECK9-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 6186 // CHECK9-NEXT: br label [[COND_END:%.*]] 6187 // CHECK9: cond.false: 6188 // CHECK9-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 6189 // CHECK9-NEXT: br label [[COND_END]] 6190 // CHECK9: cond.end: 6191 // CHECK9-NEXT: [[COND:%.*]] = phi i32 [ [[TMP12]], [[COND_TRUE]] ], [ [[TMP13]], [[COND_FALSE]] ] 6192 // CHECK9-NEXT: store i32 [[COND]], i32* [[DOTOMP_COMB_UB]], align 4 6193 // CHECK9-NEXT: [[TMP14:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4 6194 // CHECK9-NEXT: store i32 [[TMP14]], i32* [[DOTOMP_IV]], align 4 6195 // CHECK9-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] 6196 // CHECK9: omp.inner.for.cond: 6197 // CHECK9-NEXT: [[TMP15:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !38 6198 // CHECK9-NEXT: [[TMP16:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4, !llvm.access.group !38 6199 // CHECK9-NEXT: [[CMP5:%.*]] = icmp sle i32 [[TMP15]], [[TMP16]] 6200 // CHECK9-NEXT: br i1 [[CMP5]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] 6201 // CHECK9: omp.inner.for.body: 6202 // CHECK9-NEXT: [[TMP17:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4, !llvm.access.group !38 6203 // CHECK9-NEXT: [[TMP18:%.*]] = zext i32 [[TMP17]] to i64 6204 // CHECK9-NEXT: [[TMP19:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4, !llvm.access.group !38 6205 // CHECK9-NEXT: [[TMP20:%.*]] = zext i32 [[TMP19]] to i64 6206 // CHECK9-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 6, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64, i32*, double**, double**, double**)* @.omp_outlined..11 to void (i32*, i32*, ...)*), i64 [[TMP18]], i64 [[TMP20]], i32* [[TMP0]], double** [[TMP1]], double** [[TMP2]], double** [[TMP3]]), !llvm.access.group !38 6207 // CHECK9-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] 6208 // CHECK9: omp.inner.for.inc: 6209 // CHECK9-NEXT: [[TMP21:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !38 6210 // CHECK9-NEXT: [[TMP22:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4, !llvm.access.group !38 6211 // CHECK9-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP21]], [[TMP22]] 6212 // CHECK9-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4, !llvm.access.group !38 6213 // CHECK9-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP39:![0-9]+]] 6214 // CHECK9: omp.inner.for.end: 6215 // CHECK9-NEXT: br label [[OMP_LOOP_EXIT:%.*]] 6216 // CHECK9: omp.loop.exit: 6217 // CHECK9-NEXT: [[TMP23:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 6218 // CHECK9-NEXT: [[TMP24:%.*]] = load i32, i32* [[TMP23]], align 4 6219 // CHECK9-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP24]]) 6220 // CHECK9-NEXT: [[TMP25:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4 6221 // CHECK9-NEXT: [[TMP26:%.*]] = icmp ne i32 [[TMP25]], 0 6222 // CHECK9-NEXT: br i1 [[TMP26]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]] 6223 // CHECK9: .omp.final.then: 6224 // CHECK9-NEXT: [[TMP27:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 6225 // CHECK9-NEXT: [[SUB6:%.*]] = sub nsw i32 [[TMP27]], 0 6226 // CHECK9-NEXT: [[DIV7:%.*]] = sdiv i32 [[SUB6]], 1 6227 // CHECK9-NEXT: [[MUL:%.*]] = mul nsw i32 [[DIV7]], 1 6228 // CHECK9-NEXT: [[ADD8:%.*]] = add nsw i32 0, [[MUL]] 6229 // CHECK9-NEXT: store i32 [[ADD8]], i32* [[I3]], align 4 6230 // CHECK9-NEXT: br label [[DOTOMP_FINAL_DONE]] 6231 // CHECK9: .omp.final.done: 6232 // CHECK9-NEXT: br label [[OMP_PRECOND_END]] 6233 // CHECK9: omp.precond.end: 6234 // CHECK9-NEXT: ret void 6235 // 6236 // 6237 // CHECK9-LABEL: define {{[^@]+}}@.omp_outlined..11 6238 // CHECK9-SAME: (i32* noalias noundef [[DOTGLOBAL_TID_:%.*]], i32* noalias noundef [[DOTBOUND_TID_:%.*]], i64 noundef [[DOTPREVIOUS_LB_:%.*]], i64 noundef [[DOTPREVIOUS_UB_:%.*]], i32* noundef nonnull align 4 dereferenceable(4) [[N:%.*]], double** noundef nonnull align 8 dereferenceable(8) [[A:%.*]], double** noundef nonnull align 8 dereferenceable(8) [[B:%.*]], double** noundef nonnull align 8 dereferenceable(8) [[C:%.*]]) #[[ATTR1]] { 6239 // CHECK9-NEXT: entry: 6240 // CHECK9-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 6241 // CHECK9-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 6242 // CHECK9-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8 6243 // CHECK9-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8 6244 // CHECK9-NEXT: [[N_ADDR:%.*]] = alloca i32*, align 8 6245 // CHECK9-NEXT: [[A_ADDR:%.*]] = alloca double**, align 8 6246 // CHECK9-NEXT: [[B_ADDR:%.*]] = alloca double**, align 8 6247 // CHECK9-NEXT: [[C_ADDR:%.*]] = alloca double**, align 8 6248 // CHECK9-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 6249 // CHECK9-NEXT: [[TMP:%.*]] = alloca i32, align 4 6250 // CHECK9-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 6251 // CHECK9-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 6252 // CHECK9-NEXT: [[I:%.*]] = alloca i32, align 4 6253 // CHECK9-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 6254 // CHECK9-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4 6255 // CHECK9-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 6256 // CHECK9-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 6257 // CHECK9-NEXT: [[I4:%.*]] = alloca i32, align 4 6258 // CHECK9-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8 6259 // CHECK9-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8 6260 // CHECK9-NEXT: store i64 [[DOTPREVIOUS_LB_]], i64* [[DOTPREVIOUS_LB__ADDR]], align 8 6261 // CHECK9-NEXT: store i64 [[DOTPREVIOUS_UB_]], i64* [[DOTPREVIOUS_UB__ADDR]], align 8 6262 // CHECK9-NEXT: store i32* [[N]], i32** [[N_ADDR]], align 8 6263 // CHECK9-NEXT: store double** [[A]], double*** [[A_ADDR]], align 8 6264 // CHECK9-NEXT: store double** [[B]], double*** [[B_ADDR]], align 8 6265 // CHECK9-NEXT: store double** [[C]], double*** [[C_ADDR]], align 8 6266 // CHECK9-NEXT: [[TMP0:%.*]] = load i32*, i32** [[N_ADDR]], align 8 6267 // CHECK9-NEXT: [[TMP1:%.*]] = load double**, double*** [[A_ADDR]], align 8 6268 // CHECK9-NEXT: [[TMP2:%.*]] = load double**, double*** [[B_ADDR]], align 8 6269 // CHECK9-NEXT: [[TMP3:%.*]] = load double**, double*** [[C_ADDR]], align 8 6270 // CHECK9-NEXT: [[TMP4:%.*]] = load i32, i32* [[TMP0]], align 4 6271 // CHECK9-NEXT: store i32 [[TMP4]], i32* [[DOTCAPTURE_EXPR_]], align 4 6272 // CHECK9-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 6273 // CHECK9-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP5]], 0 6274 // CHECK9-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 6275 // CHECK9-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1 6276 // CHECK9-NEXT: store i32 [[SUB2]], i32* [[DOTCAPTURE_EXPR_1]], align 4 6277 // CHECK9-NEXT: store i32 0, i32* [[I]], align 4 6278 // CHECK9-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 6279 // CHECK9-NEXT: [[CMP:%.*]] = icmp slt i32 0, [[TMP6]] 6280 // CHECK9-NEXT: br i1 [[CMP]], label [[OMP_PRECOND_THEN:%.*]], label [[OMP_PRECOND_END:%.*]] 6281 // CHECK9: omp.precond.then: 6282 // CHECK9-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4 6283 // CHECK9-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 6284 // CHECK9-NEXT: store i32 [[TMP7]], i32* [[DOTOMP_UB]], align 4 6285 // CHECK9-NEXT: [[TMP8:%.*]] = load i64, i64* [[DOTPREVIOUS_LB__ADDR]], align 8 6286 // CHECK9-NEXT: [[CONV:%.*]] = trunc i64 [[TMP8]] to i32 6287 // CHECK9-NEXT: [[TMP9:%.*]] = load i64, i64* [[DOTPREVIOUS_UB__ADDR]], align 8 6288 // CHECK9-NEXT: [[CONV3:%.*]] = trunc i64 [[TMP9]] to i32 6289 // CHECK9-NEXT: store i32 [[CONV]], i32* [[DOTOMP_LB]], align 4 6290 // CHECK9-NEXT: store i32 [[CONV3]], i32* [[DOTOMP_UB]], align 4 6291 // CHECK9-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4 6292 // CHECK9-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4 6293 // CHECK9-NEXT: [[TMP10:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 6294 // CHECK9-NEXT: [[TMP11:%.*]] = load i32, i32* [[TMP10]], align 4 6295 // CHECK9-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB2]], i32 [[TMP11]], i32 34, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1) 6296 // CHECK9-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 6297 // CHECK9-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 6298 // CHECK9-NEXT: [[CMP5:%.*]] = icmp sgt i32 [[TMP12]], [[TMP13]] 6299 // CHECK9-NEXT: br i1 [[CMP5]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] 6300 // CHECK9: cond.true: 6301 // CHECK9-NEXT: [[TMP14:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 6302 // CHECK9-NEXT: br label [[COND_END:%.*]] 6303 // CHECK9: cond.false: 6304 // CHECK9-NEXT: [[TMP15:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 6305 // CHECK9-NEXT: br label [[COND_END]] 6306 // CHECK9: cond.end: 6307 // CHECK9-NEXT: [[COND:%.*]] = phi i32 [ [[TMP14]], [[COND_TRUE]] ], [ [[TMP15]], [[COND_FALSE]] ] 6308 // CHECK9-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4 6309 // CHECK9-NEXT: [[TMP16:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4 6310 // CHECK9-NEXT: store i32 [[TMP16]], i32* [[DOTOMP_IV]], align 4 6311 // CHECK9-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] 6312 // CHECK9: omp.inner.for.cond: 6313 // CHECK9-NEXT: [[TMP17:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !41 6314 // CHECK9-NEXT: [[TMP18:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4, !llvm.access.group !41 6315 // CHECK9-NEXT: [[CMP6:%.*]] = icmp sle i32 [[TMP17]], [[TMP18]] 6316 // CHECK9-NEXT: br i1 [[CMP6]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] 6317 // CHECK9: omp.inner.for.body: 6318 // CHECK9-NEXT: [[TMP19:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !41 6319 // CHECK9-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP19]], 1 6320 // CHECK9-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] 6321 // CHECK9-NEXT: store i32 [[ADD]], i32* [[I4]], align 4, !llvm.access.group !41 6322 // CHECK9-NEXT: [[TMP20:%.*]] = load double*, double** [[TMP2]], align 8, !llvm.access.group !41 6323 // CHECK9-NEXT: [[TMP21:%.*]] = load i32, i32* [[I4]], align 4, !llvm.access.group !41 6324 // CHECK9-NEXT: [[IDXPROM:%.*]] = sext i32 [[TMP21]] to i64 6325 // CHECK9-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds double, double* [[TMP20]], i64 [[IDXPROM]] 6326 // CHECK9-NEXT: [[TMP22:%.*]] = load double, double* [[ARRAYIDX]], align 8, !llvm.access.group !41 6327 // CHECK9-NEXT: [[TMP23:%.*]] = load double*, double** [[TMP3]], align 8, !llvm.access.group !41 6328 // CHECK9-NEXT: [[TMP24:%.*]] = load i32, i32* [[I4]], align 4, !llvm.access.group !41 6329 // CHECK9-NEXT: [[IDXPROM7:%.*]] = sext i32 [[TMP24]] to i64 6330 // CHECK9-NEXT: [[ARRAYIDX8:%.*]] = getelementptr inbounds double, double* [[TMP23]], i64 [[IDXPROM7]] 6331 // CHECK9-NEXT: [[TMP25:%.*]] = load double, double* [[ARRAYIDX8]], align 8, !llvm.access.group !41 6332 // CHECK9-NEXT: [[ADD9:%.*]] = fadd double [[TMP22]], [[TMP25]] 6333 // CHECK9-NEXT: [[TMP26:%.*]] = load double*, double** [[TMP1]], align 8, !llvm.access.group !41 6334 // CHECK9-NEXT: [[TMP27:%.*]] = load i32, i32* [[I4]], align 4, !llvm.access.group !41 6335 // CHECK9-NEXT: [[IDXPROM10:%.*]] = sext i32 [[TMP27]] to i64 6336 // CHECK9-NEXT: [[ARRAYIDX11:%.*]] = getelementptr inbounds double, double* [[TMP26]], i64 [[IDXPROM10]] 6337 // CHECK9-NEXT: store double [[ADD9]], double* [[ARRAYIDX11]], align 8, !llvm.access.group !41 6338 // CHECK9-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] 6339 // CHECK9: omp.body.continue: 6340 // CHECK9-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] 6341 // CHECK9: omp.inner.for.inc: 6342 // CHECK9-NEXT: [[TMP28:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !41 6343 // CHECK9-NEXT: [[ADD12:%.*]] = add nsw i32 [[TMP28]], 1 6344 // CHECK9-NEXT: store i32 [[ADD12]], i32* [[DOTOMP_IV]], align 4, !llvm.access.group !41 6345 // CHECK9-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP42:![0-9]+]] 6346 // CHECK9: omp.inner.for.end: 6347 // CHECK9-NEXT: br label [[OMP_LOOP_EXIT:%.*]] 6348 // CHECK9: omp.loop.exit: 6349 // CHECK9-NEXT: [[TMP29:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 6350 // CHECK9-NEXT: [[TMP30:%.*]] = load i32, i32* [[TMP29]], align 4 6351 // CHECK9-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP30]]) 6352 // CHECK9-NEXT: [[TMP31:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4 6353 // CHECK9-NEXT: [[TMP32:%.*]] = icmp ne i32 [[TMP31]], 0 6354 // CHECK9-NEXT: br i1 [[TMP32]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]] 6355 // CHECK9: .omp.final.then: 6356 // CHECK9-NEXT: [[TMP33:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 6357 // CHECK9-NEXT: [[SUB13:%.*]] = sub nsw i32 [[TMP33]], 0 6358 // CHECK9-NEXT: [[DIV14:%.*]] = sdiv i32 [[SUB13]], 1 6359 // CHECK9-NEXT: [[MUL15:%.*]] = mul nsw i32 [[DIV14]], 1 6360 // CHECK9-NEXT: [[ADD16:%.*]] = add nsw i32 0, [[MUL15]] 6361 // CHECK9-NEXT: store i32 [[ADD16]], i32* [[I4]], align 4 6362 // CHECK9-NEXT: br label [[DOTOMP_FINAL_DONE]] 6363 // CHECK9: .omp.final.done: 6364 // CHECK9-NEXT: br label [[OMP_PRECOND_END]] 6365 // CHECK9: omp.precond.end: 6366 // CHECK9-NEXT: ret void 6367 // 6368 // 6369 // CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l505 6370 // CHECK9-SAME: (i64 noundef [[CH:%.*]], i64 noundef [[N:%.*]], double* noundef [[A:%.*]], double* noundef [[B:%.*]], double* noundef [[C:%.*]]) #[[ATTR1]] { 6371 // CHECK9-NEXT: entry: 6372 // CHECK9-NEXT: [[CH_ADDR:%.*]] = alloca i64, align 8 6373 // CHECK9-NEXT: [[N_ADDR:%.*]] = alloca i64, align 8 6374 // CHECK9-NEXT: [[A_ADDR:%.*]] = alloca double*, align 8 6375 // CHECK9-NEXT: [[B_ADDR:%.*]] = alloca double*, align 8 6376 // CHECK9-NEXT: [[C_ADDR:%.*]] = alloca double*, align 8 6377 // CHECK9-NEXT: store i64 [[CH]], i64* [[CH_ADDR]], align 8 6378 // CHECK9-NEXT: store i64 [[N]], i64* [[N_ADDR]], align 8 6379 // CHECK9-NEXT: store double* [[A]], double** [[A_ADDR]], align 8 6380 // CHECK9-NEXT: store double* [[B]], double** [[B_ADDR]], align 8 6381 // CHECK9-NEXT: store double* [[C]], double** [[C_ADDR]], align 8 6382 // CHECK9-NEXT: [[CONV:%.*]] = bitcast i64* [[CH_ADDR]] to i32* 6383 // CHECK9-NEXT: [[CONV1:%.*]] = bitcast i64* [[N_ADDR]] to i32* 6384 // CHECK9-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 5, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32*, i32*, double**, double**, double**)* @.omp_outlined..14 to void (i32*, i32*, ...)*), i32* [[CONV]], i32* [[CONV1]], double** [[A_ADDR]], double** [[B_ADDR]], double** [[C_ADDR]]) 6385 // CHECK9-NEXT: ret void 6386 // 6387 // 6388 // CHECK9-LABEL: define {{[^@]+}}@.omp_outlined..14 6389 // CHECK9-SAME: (i32* noalias noundef [[DOTGLOBAL_TID_:%.*]], i32* noalias noundef [[DOTBOUND_TID_:%.*]], i32* noundef nonnull align 4 dereferenceable(4) [[CH:%.*]], i32* noundef nonnull align 4 dereferenceable(4) [[N:%.*]], double** noundef nonnull align 8 dereferenceable(8) [[A:%.*]], double** noundef nonnull align 8 dereferenceable(8) [[B:%.*]], double** noundef nonnull align 8 dereferenceable(8) [[C:%.*]]) #[[ATTR1]] { 6390 // CHECK9-NEXT: entry: 6391 // CHECK9-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 6392 // CHECK9-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 6393 // CHECK9-NEXT: [[CH_ADDR:%.*]] = alloca i32*, align 8 6394 // CHECK9-NEXT: [[N_ADDR:%.*]] = alloca i32*, align 8 6395 // CHECK9-NEXT: [[A_ADDR:%.*]] = alloca double**, align 8 6396 // CHECK9-NEXT: [[B_ADDR:%.*]] = alloca double**, align 8 6397 // CHECK9-NEXT: [[C_ADDR:%.*]] = alloca double**, align 8 6398 // CHECK9-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 6399 // CHECK9-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 6400 // CHECK9-NEXT: [[TMP:%.*]] = alloca i32, align 4 6401 // CHECK9-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 6402 // CHECK9-NEXT: [[DOTCAPTURE_EXPR_2:%.*]] = alloca i32, align 4 6403 // CHECK9-NEXT: [[I:%.*]] = alloca i32, align 4 6404 // CHECK9-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4 6405 // CHECK9-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4 6406 // CHECK9-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 6407 // CHECK9-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 6408 // CHECK9-NEXT: [[I4:%.*]] = alloca i32, align 4 6409 // CHECK9-NEXT: [[DOTCAPTURE_EXPR__CASTED:%.*]] = alloca i64, align 8 6410 // CHECK9-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8 6411 // CHECK9-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8 6412 // CHECK9-NEXT: store i32* [[CH]], i32** [[CH_ADDR]], align 8 6413 // CHECK9-NEXT: store i32* [[N]], i32** [[N_ADDR]], align 8 6414 // CHECK9-NEXT: store double** [[A]], double*** [[A_ADDR]], align 8 6415 // CHECK9-NEXT: store double** [[B]], double*** [[B_ADDR]], align 8 6416 // CHECK9-NEXT: store double** [[C]], double*** [[C_ADDR]], align 8 6417 // CHECK9-NEXT: [[TMP0:%.*]] = load i32*, i32** [[CH_ADDR]], align 8 6418 // CHECK9-NEXT: [[TMP1:%.*]] = load i32*, i32** [[N_ADDR]], align 8 6419 // CHECK9-NEXT: [[TMP2:%.*]] = load double**, double*** [[A_ADDR]], align 8 6420 // CHECK9-NEXT: [[TMP3:%.*]] = load double**, double*** [[B_ADDR]], align 8 6421 // CHECK9-NEXT: [[TMP4:%.*]] = load double**, double*** [[C_ADDR]], align 8 6422 // CHECK9-NEXT: [[TMP5:%.*]] = load i32, i32* [[TMP0]], align 4 6423 // CHECK9-NEXT: store i32 [[TMP5]], i32* [[DOTCAPTURE_EXPR_]], align 4 6424 // CHECK9-NEXT: [[TMP6:%.*]] = load i32, i32* [[TMP1]], align 4 6425 // CHECK9-NEXT: store i32 [[TMP6]], i32* [[DOTCAPTURE_EXPR_1]], align 4 6426 // CHECK9-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 6427 // CHECK9-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP7]], 0 6428 // CHECK9-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 6429 // CHECK9-NEXT: [[SUB3:%.*]] = sub nsw i32 [[DIV]], 1 6430 // CHECK9-NEXT: store i32 [[SUB3]], i32* [[DOTCAPTURE_EXPR_2]], align 4 6431 // CHECK9-NEXT: store i32 0, i32* [[I]], align 4 6432 // CHECK9-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 6433 // CHECK9-NEXT: [[CMP:%.*]] = icmp slt i32 0, [[TMP8]] 6434 // CHECK9-NEXT: br i1 [[CMP]], label [[OMP_PRECOND_THEN:%.*]], label [[OMP_PRECOND_END:%.*]] 6435 // CHECK9: omp.precond.then: 6436 // CHECK9-NEXT: store i32 0, i32* [[DOTOMP_COMB_LB]], align 4 6437 // CHECK9-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_2]], align 4 6438 // CHECK9-NEXT: store i32 [[TMP9]], i32* [[DOTOMP_COMB_UB]], align 4 6439 // CHECK9-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4 6440 // CHECK9-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4 6441 // CHECK9-NEXT: [[TMP10:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 6442 // CHECK9-NEXT: [[TMP11:%.*]] = load i32, i32* [[TMP10]], align 4 6443 // CHECK9-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1]], i32 [[TMP11]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_COMB_LB]], i32* [[DOTOMP_COMB_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1) 6444 // CHECK9-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 6445 // CHECK9-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_2]], align 4 6446 // CHECK9-NEXT: [[CMP5:%.*]] = icmp sgt i32 [[TMP12]], [[TMP13]] 6447 // CHECK9-NEXT: br i1 [[CMP5]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] 6448 // CHECK9: cond.true: 6449 // CHECK9-NEXT: [[TMP14:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_2]], align 4 6450 // CHECK9-NEXT: br label [[COND_END:%.*]] 6451 // CHECK9: cond.false: 6452 // CHECK9-NEXT: [[TMP15:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 6453 // CHECK9-NEXT: br label [[COND_END]] 6454 // CHECK9: cond.end: 6455 // CHECK9-NEXT: [[COND:%.*]] = phi i32 [ [[TMP14]], [[COND_TRUE]] ], [ [[TMP15]], [[COND_FALSE]] ] 6456 // CHECK9-NEXT: store i32 [[COND]], i32* [[DOTOMP_COMB_UB]], align 4 6457 // CHECK9-NEXT: [[TMP16:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4 6458 // CHECK9-NEXT: store i32 [[TMP16]], i32* [[DOTOMP_IV]], align 4 6459 // CHECK9-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] 6460 // CHECK9: omp.inner.for.cond: 6461 // CHECK9-NEXT: [[TMP17:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !44 6462 // CHECK9-NEXT: [[TMP18:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4, !llvm.access.group !44 6463 // CHECK9-NEXT: [[CMP6:%.*]] = icmp sle i32 [[TMP17]], [[TMP18]] 6464 // CHECK9-NEXT: br i1 [[CMP6]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] 6465 // CHECK9: omp.inner.for.body: 6466 // CHECK9-NEXT: [[TMP19:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4, !llvm.access.group !44 6467 // CHECK9-NEXT: [[TMP20:%.*]] = zext i32 [[TMP19]] to i64 6468 // CHECK9-NEXT: [[TMP21:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4, !llvm.access.group !44 6469 // CHECK9-NEXT: [[TMP22:%.*]] = zext i32 [[TMP21]] to i64 6470 // CHECK9-NEXT: [[TMP23:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4, !llvm.access.group !44 6471 // CHECK9-NEXT: [[CONV:%.*]] = bitcast i64* [[DOTCAPTURE_EXPR__CASTED]] to i32* 6472 // CHECK9-NEXT: store i32 [[TMP23]], i32* [[CONV]], align 4, !llvm.access.group !44 6473 // CHECK9-NEXT: [[TMP24:%.*]] = load i64, i64* [[DOTCAPTURE_EXPR__CASTED]], align 8, !llvm.access.group !44 6474 // CHECK9-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 7, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64, i32*, double**, double**, double**, i64)* @.omp_outlined..15 to void (i32*, i32*, ...)*), i64 [[TMP20]], i64 [[TMP22]], i32* [[TMP1]], double** [[TMP2]], double** [[TMP3]], double** [[TMP4]], i64 [[TMP24]]), !llvm.access.group !44 6475 // CHECK9-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] 6476 // CHECK9: omp.inner.for.inc: 6477 // CHECK9-NEXT: [[TMP25:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !44 6478 // CHECK9-NEXT: [[TMP26:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4, !llvm.access.group !44 6479 // CHECK9-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP25]], [[TMP26]] 6480 // CHECK9-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4, !llvm.access.group !44 6481 // CHECK9-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP45:![0-9]+]] 6482 // CHECK9: omp.inner.for.end: 6483 // CHECK9-NEXT: br label [[OMP_LOOP_EXIT:%.*]] 6484 // CHECK9: omp.loop.exit: 6485 // CHECK9-NEXT: [[TMP27:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 6486 // CHECK9-NEXT: [[TMP28:%.*]] = load i32, i32* [[TMP27]], align 4 6487 // CHECK9-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP28]]) 6488 // CHECK9-NEXT: [[TMP29:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4 6489 // CHECK9-NEXT: [[TMP30:%.*]] = icmp ne i32 [[TMP29]], 0 6490 // CHECK9-NEXT: br i1 [[TMP30]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]] 6491 // CHECK9: .omp.final.then: 6492 // CHECK9-NEXT: [[TMP31:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 6493 // CHECK9-NEXT: [[SUB7:%.*]] = sub nsw i32 [[TMP31]], 0 6494 // CHECK9-NEXT: [[DIV8:%.*]] = sdiv i32 [[SUB7]], 1 6495 // CHECK9-NEXT: [[MUL:%.*]] = mul nsw i32 [[DIV8]], 1 6496 // CHECK9-NEXT: [[ADD9:%.*]] = add nsw i32 0, [[MUL]] 6497 // CHECK9-NEXT: store i32 [[ADD9]], i32* [[I4]], align 4 6498 // CHECK9-NEXT: br label [[DOTOMP_FINAL_DONE]] 6499 // CHECK9: .omp.final.done: 6500 // CHECK9-NEXT: br label [[OMP_PRECOND_END]] 6501 // CHECK9: omp.precond.end: 6502 // CHECK9-NEXT: ret void 6503 // 6504 // 6505 // CHECK9-LABEL: define {{[^@]+}}@.omp_outlined..15 6506 // CHECK9-SAME: (i32* noalias noundef [[DOTGLOBAL_TID_:%.*]], i32* noalias noundef [[DOTBOUND_TID_:%.*]], i64 noundef [[DOTPREVIOUS_LB_:%.*]], i64 noundef [[DOTPREVIOUS_UB_:%.*]], i32* noundef nonnull align 4 dereferenceable(4) [[N:%.*]], double** noundef nonnull align 8 dereferenceable(8) [[A:%.*]], double** noundef nonnull align 8 dereferenceable(8) [[B:%.*]], double** noundef nonnull align 8 dereferenceable(8) [[C:%.*]], i64 noundef [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR1]] { 6507 // CHECK9-NEXT: entry: 6508 // CHECK9-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 6509 // CHECK9-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 6510 // CHECK9-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8 6511 // CHECK9-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8 6512 // CHECK9-NEXT: [[N_ADDR:%.*]] = alloca i32*, align 8 6513 // CHECK9-NEXT: [[A_ADDR:%.*]] = alloca double**, align 8 6514 // CHECK9-NEXT: [[B_ADDR:%.*]] = alloca double**, align 8 6515 // CHECK9-NEXT: [[C_ADDR:%.*]] = alloca double**, align 8 6516 // CHECK9-NEXT: [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i64, align 8 6517 // CHECK9-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 6518 // CHECK9-NEXT: [[TMP:%.*]] = alloca i32, align 4 6519 // CHECK9-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 6520 // CHECK9-NEXT: [[DOTCAPTURE_EXPR_2:%.*]] = alloca i32, align 4 6521 // CHECK9-NEXT: [[I:%.*]] = alloca i32, align 4 6522 // CHECK9-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 6523 // CHECK9-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4 6524 // CHECK9-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 6525 // CHECK9-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 6526 // CHECK9-NEXT: [[I6:%.*]] = alloca i32, align 4 6527 // CHECK9-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8 6528 // CHECK9-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8 6529 // CHECK9-NEXT: store i64 [[DOTPREVIOUS_LB_]], i64* [[DOTPREVIOUS_LB__ADDR]], align 8 6530 // CHECK9-NEXT: store i64 [[DOTPREVIOUS_UB_]], i64* [[DOTPREVIOUS_UB__ADDR]], align 8 6531 // CHECK9-NEXT: store i32* [[N]], i32** [[N_ADDR]], align 8 6532 // CHECK9-NEXT: store double** [[A]], double*** [[A_ADDR]], align 8 6533 // CHECK9-NEXT: store double** [[B]], double*** [[B_ADDR]], align 8 6534 // CHECK9-NEXT: store double** [[C]], double*** [[C_ADDR]], align 8 6535 // CHECK9-NEXT: store i64 [[DOTCAPTURE_EXPR_]], i64* [[DOTCAPTURE_EXPR__ADDR]], align 8 6536 // CHECK9-NEXT: [[TMP0:%.*]] = load i32*, i32** [[N_ADDR]], align 8 6537 // CHECK9-NEXT: [[TMP1:%.*]] = load double**, double*** [[A_ADDR]], align 8 6538 // CHECK9-NEXT: [[TMP2:%.*]] = load double**, double*** [[B_ADDR]], align 8 6539 // CHECK9-NEXT: [[TMP3:%.*]] = load double**, double*** [[C_ADDR]], align 8 6540 // CHECK9-NEXT: [[CONV:%.*]] = bitcast i64* [[DOTCAPTURE_EXPR__ADDR]] to i32* 6541 // CHECK9-NEXT: [[TMP4:%.*]] = load i32, i32* [[TMP0]], align 4 6542 // CHECK9-NEXT: store i32 [[TMP4]], i32* [[DOTCAPTURE_EXPR_1]], align 4 6543 // CHECK9-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 6544 // CHECK9-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP5]], 0 6545 // CHECK9-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 6546 // CHECK9-NEXT: [[SUB3:%.*]] = sub nsw i32 [[DIV]], 1 6547 // CHECK9-NEXT: store i32 [[SUB3]], i32* [[DOTCAPTURE_EXPR_2]], align 4 6548 // CHECK9-NEXT: store i32 0, i32* [[I]], align 4 6549 // CHECK9-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 6550 // CHECK9-NEXT: [[CMP:%.*]] = icmp slt i32 0, [[TMP6]] 6551 // CHECK9-NEXT: br i1 [[CMP]], label [[OMP_PRECOND_THEN:%.*]], label [[OMP_PRECOND_END:%.*]] 6552 // CHECK9: omp.precond.then: 6553 // CHECK9-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4 6554 // CHECK9-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_2]], align 4 6555 // CHECK9-NEXT: store i32 [[TMP7]], i32* [[DOTOMP_UB]], align 4 6556 // CHECK9-NEXT: [[TMP8:%.*]] = load i64, i64* [[DOTPREVIOUS_LB__ADDR]], align 8 6557 // CHECK9-NEXT: [[CONV4:%.*]] = trunc i64 [[TMP8]] to i32 6558 // CHECK9-NEXT: [[TMP9:%.*]] = load i64, i64* [[DOTPREVIOUS_UB__ADDR]], align 8 6559 // CHECK9-NEXT: [[CONV5:%.*]] = trunc i64 [[TMP9]] to i32 6560 // CHECK9-NEXT: store i32 [[CONV4]], i32* [[DOTOMP_LB]], align 4 6561 // CHECK9-NEXT: store i32 [[CONV5]], i32* [[DOTOMP_UB]], align 4 6562 // CHECK9-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4 6563 // CHECK9-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4 6564 // CHECK9-NEXT: [[TMP10:%.*]] = load i32, i32* [[CONV]], align 4 6565 // CHECK9-NEXT: [[TMP11:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 6566 // CHECK9-NEXT: [[TMP12:%.*]] = load i32, i32* [[TMP11]], align 4 6567 // CHECK9-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB2]], i32 [[TMP12]], i32 33, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 [[TMP10]]) 6568 // CHECK9-NEXT: br label [[OMP_DISPATCH_COND:%.*]] 6569 // CHECK9: omp.dispatch.cond: 6570 // CHECK9-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 6571 // CHECK9-NEXT: [[TMP14:%.*]] = load i64, i64* [[DOTPREVIOUS_UB__ADDR]], align 8 6572 // CHECK9-NEXT: [[CONV7:%.*]] = trunc i64 [[TMP14]] to i32 6573 // CHECK9-NEXT: [[CMP8:%.*]] = icmp sgt i32 [[TMP13]], [[CONV7]] 6574 // CHECK9-NEXT: br i1 [[CMP8]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] 6575 // CHECK9: cond.true: 6576 // CHECK9-NEXT: [[TMP15:%.*]] = load i64, i64* [[DOTPREVIOUS_UB__ADDR]], align 8 6577 // CHECK9-NEXT: [[CONV9:%.*]] = trunc i64 [[TMP15]] to i32 6578 // CHECK9-NEXT: br label [[COND_END:%.*]] 6579 // CHECK9: cond.false: 6580 // CHECK9-NEXT: [[TMP16:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 6581 // CHECK9-NEXT: br label [[COND_END]] 6582 // CHECK9: cond.end: 6583 // CHECK9-NEXT: [[COND:%.*]] = phi i32 [ [[CONV9]], [[COND_TRUE]] ], [ [[TMP16]], [[COND_FALSE]] ] 6584 // CHECK9-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4 6585 // CHECK9-NEXT: [[TMP17:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4 6586 // CHECK9-NEXT: store i32 [[TMP17]], i32* [[DOTOMP_IV]], align 4 6587 // CHECK9-NEXT: [[TMP18:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 6588 // CHECK9-NEXT: [[TMP19:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 6589 // CHECK9-NEXT: [[CMP10:%.*]] = icmp sle i32 [[TMP18]], [[TMP19]] 6590 // CHECK9-NEXT: br i1 [[CMP10]], label [[OMP_DISPATCH_BODY:%.*]], label [[OMP_DISPATCH_END:%.*]] 6591 // CHECK9: omp.dispatch.body: 6592 // CHECK9-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] 6593 // CHECK9: omp.inner.for.cond: 6594 // CHECK9-NEXT: [[TMP20:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !47 6595 // CHECK9-NEXT: [[TMP21:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4, !llvm.access.group !47 6596 // CHECK9-NEXT: [[CMP11:%.*]] = icmp sle i32 [[TMP20]], [[TMP21]] 6597 // CHECK9-NEXT: br i1 [[CMP11]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] 6598 // CHECK9: omp.inner.for.body: 6599 // CHECK9-NEXT: [[TMP22:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !47 6600 // CHECK9-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP22]], 1 6601 // CHECK9-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] 6602 // CHECK9-NEXT: store i32 [[ADD]], i32* [[I6]], align 4, !llvm.access.group !47 6603 // CHECK9-NEXT: [[TMP23:%.*]] = load double*, double** [[TMP2]], align 8, !llvm.access.group !47 6604 // CHECK9-NEXT: [[TMP24:%.*]] = load i32, i32* [[I6]], align 4, !llvm.access.group !47 6605 // CHECK9-NEXT: [[IDXPROM:%.*]] = sext i32 [[TMP24]] to i64 6606 // CHECK9-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds double, double* [[TMP23]], i64 [[IDXPROM]] 6607 // CHECK9-NEXT: [[TMP25:%.*]] = load double, double* [[ARRAYIDX]], align 8, !llvm.access.group !47 6608 // CHECK9-NEXT: [[TMP26:%.*]] = load double*, double** [[TMP3]], align 8, !llvm.access.group !47 6609 // CHECK9-NEXT: [[TMP27:%.*]] = load i32, i32* [[I6]], align 4, !llvm.access.group !47 6610 // CHECK9-NEXT: [[IDXPROM12:%.*]] = sext i32 [[TMP27]] to i64 6611 // CHECK9-NEXT: [[ARRAYIDX13:%.*]] = getelementptr inbounds double, double* [[TMP26]], i64 [[IDXPROM12]] 6612 // CHECK9-NEXT: [[TMP28:%.*]] = load double, double* [[ARRAYIDX13]], align 8, !llvm.access.group !47 6613 // CHECK9-NEXT: [[ADD14:%.*]] = fadd double [[TMP25]], [[TMP28]] 6614 // CHECK9-NEXT: [[TMP29:%.*]] = load double*, double** [[TMP1]], align 8, !llvm.access.group !47 6615 // CHECK9-NEXT: [[TMP30:%.*]] = load i32, i32* [[I6]], align 4, !llvm.access.group !47 6616 // CHECK9-NEXT: [[IDXPROM15:%.*]] = sext i32 [[TMP30]] to i64 6617 // CHECK9-NEXT: [[ARRAYIDX16:%.*]] = getelementptr inbounds double, double* [[TMP29]], i64 [[IDXPROM15]] 6618 // CHECK9-NEXT: store double [[ADD14]], double* [[ARRAYIDX16]], align 8, !llvm.access.group !47 6619 // CHECK9-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] 6620 // CHECK9: omp.body.continue: 6621 // CHECK9-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] 6622 // CHECK9: omp.inner.for.inc: 6623 // CHECK9-NEXT: [[TMP31:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !47 6624 // CHECK9-NEXT: [[ADD17:%.*]] = add nsw i32 [[TMP31]], 1 6625 // CHECK9-NEXT: store i32 [[ADD17]], i32* [[DOTOMP_IV]], align 4, !llvm.access.group !47 6626 // CHECK9-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP48:![0-9]+]] 6627 // CHECK9: omp.inner.for.end: 6628 // CHECK9-NEXT: br label [[OMP_DISPATCH_INC:%.*]] 6629 // CHECK9: omp.dispatch.inc: 6630 // CHECK9-NEXT: [[TMP32:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4 6631 // CHECK9-NEXT: [[TMP33:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4 6632 // CHECK9-NEXT: [[ADD18:%.*]] = add nsw i32 [[TMP32]], [[TMP33]] 6633 // CHECK9-NEXT: store i32 [[ADD18]], i32* [[DOTOMP_LB]], align 4 6634 // CHECK9-NEXT: [[TMP34:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 6635 // CHECK9-NEXT: [[TMP35:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4 6636 // CHECK9-NEXT: [[ADD19:%.*]] = add nsw i32 [[TMP34]], [[TMP35]] 6637 // CHECK9-NEXT: store i32 [[ADD19]], i32* [[DOTOMP_UB]], align 4 6638 // CHECK9-NEXT: br label [[OMP_DISPATCH_COND]] 6639 // CHECK9: omp.dispatch.end: 6640 // CHECK9-NEXT: [[TMP36:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 6641 // CHECK9-NEXT: [[TMP37:%.*]] = load i32, i32* [[TMP36]], align 4 6642 // CHECK9-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP37]]) 6643 // CHECK9-NEXT: [[TMP38:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4 6644 // CHECK9-NEXT: [[TMP39:%.*]] = icmp ne i32 [[TMP38]], 0 6645 // CHECK9-NEXT: br i1 [[TMP39]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]] 6646 // CHECK9: .omp.final.then: 6647 // CHECK9-NEXT: [[TMP40:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 6648 // CHECK9-NEXT: [[SUB20:%.*]] = sub nsw i32 [[TMP40]], 0 6649 // CHECK9-NEXT: [[DIV21:%.*]] = sdiv i32 [[SUB20]], 1 6650 // CHECK9-NEXT: [[MUL22:%.*]] = mul nsw i32 [[DIV21]], 1 6651 // CHECK9-NEXT: [[ADD23:%.*]] = add nsw i32 0, [[MUL22]] 6652 // CHECK9-NEXT: store i32 [[ADD23]], i32* [[I6]], align 4 6653 // CHECK9-NEXT: br label [[DOTOMP_FINAL_DONE]] 6654 // CHECK9: .omp.final.done: 6655 // CHECK9-NEXT: br label [[OMP_PRECOND_END]] 6656 // CHECK9: omp.precond.end: 6657 // CHECK9-NEXT: ret void 6658 // 6659 // 6660 // CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l535 6661 // CHECK9-SAME: (i64 noundef [[N:%.*]], double* noundef [[A:%.*]], double* noundef [[B:%.*]], double* noundef [[C:%.*]]) #[[ATTR1]] { 6662 // CHECK9-NEXT: entry: 6663 // CHECK9-NEXT: [[N_ADDR:%.*]] = alloca i64, align 8 6664 // CHECK9-NEXT: [[A_ADDR:%.*]] = alloca double*, align 8 6665 // CHECK9-NEXT: [[B_ADDR:%.*]] = alloca double*, align 8 6666 // CHECK9-NEXT: [[C_ADDR:%.*]] = alloca double*, align 8 6667 // CHECK9-NEXT: store i64 [[N]], i64* [[N_ADDR]], align 8 6668 // CHECK9-NEXT: store double* [[A]], double** [[A_ADDR]], align 8 6669 // CHECK9-NEXT: store double* [[B]], double** [[B_ADDR]], align 8 6670 // CHECK9-NEXT: store double* [[C]], double** [[C_ADDR]], align 8 6671 // CHECK9-NEXT: [[CONV:%.*]] = bitcast i64* [[N_ADDR]] to i32* 6672 // CHECK9-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 4, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32*, double**, double**, double**)* @.omp_outlined..18 to void (i32*, i32*, ...)*), i32* [[CONV]], double** [[A_ADDR]], double** [[B_ADDR]], double** [[C_ADDR]]) 6673 // CHECK9-NEXT: ret void 6674 // 6675 // 6676 // CHECK9-LABEL: define {{[^@]+}}@.omp_outlined..18 6677 // CHECK9-SAME: (i32* noalias noundef [[DOTGLOBAL_TID_:%.*]], i32* noalias noundef [[DOTBOUND_TID_:%.*]], i32* noundef nonnull align 4 dereferenceable(4) [[N:%.*]], double** noundef nonnull align 8 dereferenceable(8) [[A:%.*]], double** noundef nonnull align 8 dereferenceable(8) [[B:%.*]], double** noundef nonnull align 8 dereferenceable(8) [[C:%.*]]) #[[ATTR1]] { 6678 // CHECK9-NEXT: entry: 6679 // CHECK9-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 6680 // CHECK9-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 6681 // CHECK9-NEXT: [[N_ADDR:%.*]] = alloca i32*, align 8 6682 // CHECK9-NEXT: [[A_ADDR:%.*]] = alloca double**, align 8 6683 // CHECK9-NEXT: [[B_ADDR:%.*]] = alloca double**, align 8 6684 // CHECK9-NEXT: [[C_ADDR:%.*]] = alloca double**, align 8 6685 // CHECK9-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 6686 // CHECK9-NEXT: [[TMP:%.*]] = alloca i32, align 4 6687 // CHECK9-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 6688 // CHECK9-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 6689 // CHECK9-NEXT: [[I:%.*]] = alloca i32, align 4 6690 // CHECK9-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4 6691 // CHECK9-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4 6692 // CHECK9-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 6693 // CHECK9-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 6694 // CHECK9-NEXT: [[I3:%.*]] = alloca i32, align 4 6695 // CHECK9-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8 6696 // CHECK9-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8 6697 // CHECK9-NEXT: store i32* [[N]], i32** [[N_ADDR]], align 8 6698 // CHECK9-NEXT: store double** [[A]], double*** [[A_ADDR]], align 8 6699 // CHECK9-NEXT: store double** [[B]], double*** [[B_ADDR]], align 8 6700 // CHECK9-NEXT: store double** [[C]], double*** [[C_ADDR]], align 8 6701 // CHECK9-NEXT: [[TMP0:%.*]] = load i32*, i32** [[N_ADDR]], align 8 6702 // CHECK9-NEXT: [[TMP1:%.*]] = load double**, double*** [[A_ADDR]], align 8 6703 // CHECK9-NEXT: [[TMP2:%.*]] = load double**, double*** [[B_ADDR]], align 8 6704 // CHECK9-NEXT: [[TMP3:%.*]] = load double**, double*** [[C_ADDR]], align 8 6705 // CHECK9-NEXT: [[TMP4:%.*]] = load i32, i32* [[TMP0]], align 4 6706 // CHECK9-NEXT: store i32 [[TMP4]], i32* [[DOTCAPTURE_EXPR_]], align 4 6707 // CHECK9-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 6708 // CHECK9-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP5]], 0 6709 // CHECK9-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 6710 // CHECK9-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1 6711 // CHECK9-NEXT: store i32 [[SUB2]], i32* [[DOTCAPTURE_EXPR_1]], align 4 6712 // CHECK9-NEXT: store i32 0, i32* [[I]], align 4 6713 // CHECK9-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 6714 // CHECK9-NEXT: [[CMP:%.*]] = icmp slt i32 0, [[TMP6]] 6715 // CHECK9-NEXT: br i1 [[CMP]], label [[OMP_PRECOND_THEN:%.*]], label [[OMP_PRECOND_END:%.*]] 6716 // CHECK9: omp.precond.then: 6717 // CHECK9-NEXT: store i32 0, i32* [[DOTOMP_COMB_LB]], align 4 6718 // CHECK9-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 6719 // CHECK9-NEXT: store i32 [[TMP7]], i32* [[DOTOMP_COMB_UB]], align 4 6720 // CHECK9-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4 6721 // CHECK9-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4 6722 // CHECK9-NEXT: [[TMP8:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 6723 // CHECK9-NEXT: [[TMP9:%.*]] = load i32, i32* [[TMP8]], align 4 6724 // CHECK9-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1]], i32 [[TMP9]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_COMB_LB]], i32* [[DOTOMP_COMB_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1) 6725 // CHECK9-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 6726 // CHECK9-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 6727 // CHECK9-NEXT: [[CMP4:%.*]] = icmp sgt i32 [[TMP10]], [[TMP11]] 6728 // CHECK9-NEXT: br i1 [[CMP4]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] 6729 // CHECK9: cond.true: 6730 // CHECK9-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 6731 // CHECK9-NEXT: br label [[COND_END:%.*]] 6732 // CHECK9: cond.false: 6733 // CHECK9-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 6734 // CHECK9-NEXT: br label [[COND_END]] 6735 // CHECK9: cond.end: 6736 // CHECK9-NEXT: [[COND:%.*]] = phi i32 [ [[TMP12]], [[COND_TRUE]] ], [ [[TMP13]], [[COND_FALSE]] ] 6737 // CHECK9-NEXT: store i32 [[COND]], i32* [[DOTOMP_COMB_UB]], align 4 6738 // CHECK9-NEXT: [[TMP14:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4 6739 // CHECK9-NEXT: store i32 [[TMP14]], i32* [[DOTOMP_IV]], align 4 6740 // CHECK9-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] 6741 // CHECK9: omp.inner.for.cond: 6742 // CHECK9-NEXT: [[TMP15:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !50 6743 // CHECK9-NEXT: [[TMP16:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4, !llvm.access.group !50 6744 // CHECK9-NEXT: [[CMP5:%.*]] = icmp sle i32 [[TMP15]], [[TMP16]] 6745 // CHECK9-NEXT: br i1 [[CMP5]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] 6746 // CHECK9: omp.inner.for.body: 6747 // CHECK9-NEXT: [[TMP17:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4, !llvm.access.group !50 6748 // CHECK9-NEXT: [[TMP18:%.*]] = zext i32 [[TMP17]] to i64 6749 // CHECK9-NEXT: [[TMP19:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4, !llvm.access.group !50 6750 // CHECK9-NEXT: [[TMP20:%.*]] = zext i32 [[TMP19]] to i64 6751 // CHECK9-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 6, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64, i32*, double**, double**, double**)* @.omp_outlined..19 to void (i32*, i32*, ...)*), i64 [[TMP18]], i64 [[TMP20]], i32* [[TMP0]], double** [[TMP1]], double** [[TMP2]], double** [[TMP3]]), !llvm.access.group !50 6752 // CHECK9-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] 6753 // CHECK9: omp.inner.for.inc: 6754 // CHECK9-NEXT: [[TMP21:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !50 6755 // CHECK9-NEXT: [[TMP22:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4, !llvm.access.group !50 6756 // CHECK9-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP21]], [[TMP22]] 6757 // CHECK9-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4, !llvm.access.group !50 6758 // CHECK9-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP51:![0-9]+]] 6759 // CHECK9: omp.inner.for.end: 6760 // CHECK9-NEXT: br label [[OMP_LOOP_EXIT:%.*]] 6761 // CHECK9: omp.loop.exit: 6762 // CHECK9-NEXT: [[TMP23:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 6763 // CHECK9-NEXT: [[TMP24:%.*]] = load i32, i32* [[TMP23]], align 4 6764 // CHECK9-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP24]]) 6765 // CHECK9-NEXT: [[TMP25:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4 6766 // CHECK9-NEXT: [[TMP26:%.*]] = icmp ne i32 [[TMP25]], 0 6767 // CHECK9-NEXT: br i1 [[TMP26]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]] 6768 // CHECK9: .omp.final.then: 6769 // CHECK9-NEXT: [[TMP27:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 6770 // CHECK9-NEXT: [[SUB6:%.*]] = sub nsw i32 [[TMP27]], 0 6771 // CHECK9-NEXT: [[DIV7:%.*]] = sdiv i32 [[SUB6]], 1 6772 // CHECK9-NEXT: [[MUL:%.*]] = mul nsw i32 [[DIV7]], 1 6773 // CHECK9-NEXT: [[ADD8:%.*]] = add nsw i32 0, [[MUL]] 6774 // CHECK9-NEXT: store i32 [[ADD8]], i32* [[I3]], align 4 6775 // CHECK9-NEXT: br label [[DOTOMP_FINAL_DONE]] 6776 // CHECK9: .omp.final.done: 6777 // CHECK9-NEXT: br label [[OMP_PRECOND_END]] 6778 // CHECK9: omp.precond.end: 6779 // CHECK9-NEXT: ret void 6780 // 6781 // 6782 // CHECK9-LABEL: define {{[^@]+}}@.omp_outlined..19 6783 // CHECK9-SAME: (i32* noalias noundef [[DOTGLOBAL_TID_:%.*]], i32* noalias noundef [[DOTBOUND_TID_:%.*]], i64 noundef [[DOTPREVIOUS_LB_:%.*]], i64 noundef [[DOTPREVIOUS_UB_:%.*]], i32* noundef nonnull align 4 dereferenceable(4) [[N:%.*]], double** noundef nonnull align 8 dereferenceable(8) [[A:%.*]], double** noundef nonnull align 8 dereferenceable(8) [[B:%.*]], double** noundef nonnull align 8 dereferenceable(8) [[C:%.*]]) #[[ATTR1]] { 6784 // CHECK9-NEXT: entry: 6785 // CHECK9-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 6786 // CHECK9-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 6787 // CHECK9-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8 6788 // CHECK9-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8 6789 // CHECK9-NEXT: [[N_ADDR:%.*]] = alloca i32*, align 8 6790 // CHECK9-NEXT: [[A_ADDR:%.*]] = alloca double**, align 8 6791 // CHECK9-NEXT: [[B_ADDR:%.*]] = alloca double**, align 8 6792 // CHECK9-NEXT: [[C_ADDR:%.*]] = alloca double**, align 8 6793 // CHECK9-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 6794 // CHECK9-NEXT: [[TMP:%.*]] = alloca i32, align 4 6795 // CHECK9-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 6796 // CHECK9-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 6797 // CHECK9-NEXT: [[I:%.*]] = alloca i32, align 4 6798 // CHECK9-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 6799 // CHECK9-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4 6800 // CHECK9-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 6801 // CHECK9-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 6802 // CHECK9-NEXT: [[I4:%.*]] = alloca i32, align 4 6803 // CHECK9-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8 6804 // CHECK9-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8 6805 // CHECK9-NEXT: store i64 [[DOTPREVIOUS_LB_]], i64* [[DOTPREVIOUS_LB__ADDR]], align 8 6806 // CHECK9-NEXT: store i64 [[DOTPREVIOUS_UB_]], i64* [[DOTPREVIOUS_UB__ADDR]], align 8 6807 // CHECK9-NEXT: store i32* [[N]], i32** [[N_ADDR]], align 8 6808 // CHECK9-NEXT: store double** [[A]], double*** [[A_ADDR]], align 8 6809 // CHECK9-NEXT: store double** [[B]], double*** [[B_ADDR]], align 8 6810 // CHECK9-NEXT: store double** [[C]], double*** [[C_ADDR]], align 8 6811 // CHECK9-NEXT: [[TMP0:%.*]] = load i32*, i32** [[N_ADDR]], align 8 6812 // CHECK9-NEXT: [[TMP1:%.*]] = load double**, double*** [[A_ADDR]], align 8 6813 // CHECK9-NEXT: [[TMP2:%.*]] = load double**, double*** [[B_ADDR]], align 8 6814 // CHECK9-NEXT: [[TMP3:%.*]] = load double**, double*** [[C_ADDR]], align 8 6815 // CHECK9-NEXT: [[TMP4:%.*]] = load i32, i32* [[TMP0]], align 4 6816 // CHECK9-NEXT: store i32 [[TMP4]], i32* [[DOTCAPTURE_EXPR_]], align 4 6817 // CHECK9-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 6818 // CHECK9-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP5]], 0 6819 // CHECK9-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 6820 // CHECK9-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1 6821 // CHECK9-NEXT: store i32 [[SUB2]], i32* [[DOTCAPTURE_EXPR_1]], align 4 6822 // CHECK9-NEXT: store i32 0, i32* [[I]], align 4 6823 // CHECK9-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 6824 // CHECK9-NEXT: [[CMP:%.*]] = icmp slt i32 0, [[TMP6]] 6825 // CHECK9-NEXT: br i1 [[CMP]], label [[OMP_PRECOND_THEN:%.*]], label [[OMP_PRECOND_END:%.*]] 6826 // CHECK9: omp.precond.then: 6827 // CHECK9-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4 6828 // CHECK9-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 6829 // CHECK9-NEXT: store i32 [[TMP7]], i32* [[DOTOMP_UB]], align 4 6830 // CHECK9-NEXT: [[TMP8:%.*]] = load i64, i64* [[DOTPREVIOUS_LB__ADDR]], align 8 6831 // CHECK9-NEXT: [[CONV:%.*]] = trunc i64 [[TMP8]] to i32 6832 // CHECK9-NEXT: [[TMP9:%.*]] = load i64, i64* [[DOTPREVIOUS_UB__ADDR]], align 8 6833 // CHECK9-NEXT: [[CONV3:%.*]] = trunc i64 [[TMP9]] to i32 6834 // CHECK9-NEXT: store i32 [[CONV]], i32* [[DOTOMP_LB]], align 4 6835 // CHECK9-NEXT: store i32 [[CONV3]], i32* [[DOTOMP_UB]], align 4 6836 // CHECK9-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4 6837 // CHECK9-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4 6838 // CHECK9-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4 6839 // CHECK9-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 6840 // CHECK9-NEXT: [[TMP12:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 6841 // CHECK9-NEXT: [[TMP13:%.*]] = load i32, i32* [[TMP12]], align 4 6842 // CHECK9-NEXT: call void @__kmpc_dispatch_init_4(%struct.ident_t* @[[GLOB3]], i32 [[TMP13]], i32 35, i32 [[TMP10]], i32 [[TMP11]], i32 1, i32 1) 6843 // CHECK9-NEXT: br label [[OMP_DISPATCH_COND:%.*]] 6844 // CHECK9: omp.dispatch.cond: 6845 // CHECK9-NEXT: [[TMP14:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 6846 // CHECK9-NEXT: [[TMP15:%.*]] = load i32, i32* [[TMP14]], align 4 6847 // CHECK9-NEXT: [[TMP16:%.*]] = call i32 @__kmpc_dispatch_next_4(%struct.ident_t* @[[GLOB3]], i32 [[TMP15]], i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]]) 6848 // CHECK9-NEXT: [[TOBOOL:%.*]] = icmp ne i32 [[TMP16]], 0 6849 // CHECK9-NEXT: br i1 [[TOBOOL]], label [[OMP_DISPATCH_BODY:%.*]], label [[OMP_DISPATCH_END:%.*]] 6850 // CHECK9: omp.dispatch.body: 6851 // CHECK9-NEXT: [[TMP17:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4 6852 // CHECK9-NEXT: store i32 [[TMP17]], i32* [[DOTOMP_IV]], align 4 6853 // CHECK9-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] 6854 // CHECK9: omp.inner.for.cond: 6855 // CHECK9-NEXT: [[TMP18:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !53 6856 // CHECK9-NEXT: [[TMP19:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4, !llvm.access.group !53 6857 // CHECK9-NEXT: [[CMP5:%.*]] = icmp sle i32 [[TMP18]], [[TMP19]] 6858 // CHECK9-NEXT: br i1 [[CMP5]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] 6859 // CHECK9: omp.inner.for.body: 6860 // CHECK9-NEXT: [[TMP20:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !53 6861 // CHECK9-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP20]], 1 6862 // CHECK9-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] 6863 // CHECK9-NEXT: store i32 [[ADD]], i32* [[I4]], align 4, !llvm.access.group !53 6864 // CHECK9-NEXT: [[TMP21:%.*]] = load double*, double** [[TMP2]], align 8, !llvm.access.group !53 6865 // CHECK9-NEXT: [[TMP22:%.*]] = load i32, i32* [[I4]], align 4, !llvm.access.group !53 6866 // CHECK9-NEXT: [[IDXPROM:%.*]] = sext i32 [[TMP22]] to i64 6867 // CHECK9-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds double, double* [[TMP21]], i64 [[IDXPROM]] 6868 // CHECK9-NEXT: [[TMP23:%.*]] = load double, double* [[ARRAYIDX]], align 8, !llvm.access.group !53 6869 // CHECK9-NEXT: [[TMP24:%.*]] = load double*, double** [[TMP3]], align 8, !llvm.access.group !53 6870 // CHECK9-NEXT: [[TMP25:%.*]] = load i32, i32* [[I4]], align 4, !llvm.access.group !53 6871 // CHECK9-NEXT: [[IDXPROM6:%.*]] = sext i32 [[TMP25]] to i64 6872 // CHECK9-NEXT: [[ARRAYIDX7:%.*]] = getelementptr inbounds double, double* [[TMP24]], i64 [[IDXPROM6]] 6873 // CHECK9-NEXT: [[TMP26:%.*]] = load double, double* [[ARRAYIDX7]], align 8, !llvm.access.group !53 6874 // CHECK9-NEXT: [[ADD8:%.*]] = fadd double [[TMP23]], [[TMP26]] 6875 // CHECK9-NEXT: [[TMP27:%.*]] = load double*, double** [[TMP1]], align 8, !llvm.access.group !53 6876 // CHECK9-NEXT: [[TMP28:%.*]] = load i32, i32* [[I4]], align 4, !llvm.access.group !53 6877 // CHECK9-NEXT: [[IDXPROM9:%.*]] = sext i32 [[TMP28]] to i64 6878 // CHECK9-NEXT: [[ARRAYIDX10:%.*]] = getelementptr inbounds double, double* [[TMP27]], i64 [[IDXPROM9]] 6879 // CHECK9-NEXT: store double [[ADD8]], double* [[ARRAYIDX10]], align 8, !llvm.access.group !53 6880 // CHECK9-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] 6881 // CHECK9: omp.body.continue: 6882 // CHECK9-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] 6883 // CHECK9: omp.inner.for.inc: 6884 // CHECK9-NEXT: [[TMP29:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !53 6885 // CHECK9-NEXT: [[ADD11:%.*]] = add nsw i32 [[TMP29]], 1 6886 // CHECK9-NEXT: store i32 [[ADD11]], i32* [[DOTOMP_IV]], align 4, !llvm.access.group !53 6887 // CHECK9-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP54:![0-9]+]] 6888 // CHECK9: omp.inner.for.end: 6889 // CHECK9-NEXT: br label [[OMP_DISPATCH_INC:%.*]] 6890 // CHECK9: omp.dispatch.inc: 6891 // CHECK9-NEXT: br label [[OMP_DISPATCH_COND]] 6892 // CHECK9: omp.dispatch.end: 6893 // CHECK9-NEXT: [[TMP30:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4 6894 // CHECK9-NEXT: [[TMP31:%.*]] = icmp ne i32 [[TMP30]], 0 6895 // CHECK9-NEXT: br i1 [[TMP31]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]] 6896 // CHECK9: .omp.final.then: 6897 // CHECK9-NEXT: [[TMP32:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 6898 // CHECK9-NEXT: [[SUB12:%.*]] = sub nsw i32 [[TMP32]], 0 6899 // CHECK9-NEXT: [[DIV13:%.*]] = sdiv i32 [[SUB12]], 1 6900 // CHECK9-NEXT: [[MUL14:%.*]] = mul nsw i32 [[DIV13]], 1 6901 // CHECK9-NEXT: [[ADD15:%.*]] = add nsw i32 0, [[MUL14]] 6902 // CHECK9-NEXT: store i32 [[ADD15]], i32* [[I4]], align 4 6903 // CHECK9-NEXT: br label [[DOTOMP_FINAL_DONE]] 6904 // CHECK9: .omp.final.done: 6905 // CHECK9-NEXT: br label [[OMP_PRECOND_END]] 6906 // CHECK9: omp.precond.end: 6907 // CHECK9-NEXT: ret void 6908 // 6909 // 6910 // CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l561 6911 // CHECK9-SAME: (i64 noundef [[CH:%.*]], i64 noundef [[N:%.*]], double* noundef [[A:%.*]], double* noundef [[B:%.*]], double* noundef [[C:%.*]]) #[[ATTR1]] { 6912 // CHECK9-NEXT: entry: 6913 // CHECK9-NEXT: [[CH_ADDR:%.*]] = alloca i64, align 8 6914 // CHECK9-NEXT: [[N_ADDR:%.*]] = alloca i64, align 8 6915 // CHECK9-NEXT: [[A_ADDR:%.*]] = alloca double*, align 8 6916 // CHECK9-NEXT: [[B_ADDR:%.*]] = alloca double*, align 8 6917 // CHECK9-NEXT: [[C_ADDR:%.*]] = alloca double*, align 8 6918 // CHECK9-NEXT: store i64 [[CH]], i64* [[CH_ADDR]], align 8 6919 // CHECK9-NEXT: store i64 [[N]], i64* [[N_ADDR]], align 8 6920 // CHECK9-NEXT: store double* [[A]], double** [[A_ADDR]], align 8 6921 // CHECK9-NEXT: store double* [[B]], double** [[B_ADDR]], align 8 6922 // CHECK9-NEXT: store double* [[C]], double** [[C_ADDR]], align 8 6923 // CHECK9-NEXT: [[CONV:%.*]] = bitcast i64* [[CH_ADDR]] to i32* 6924 // CHECK9-NEXT: [[CONV1:%.*]] = bitcast i64* [[N_ADDR]] to i32* 6925 // CHECK9-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 5, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32*, i32*, double**, double**, double**)* @.omp_outlined..22 to void (i32*, i32*, ...)*), i32* [[CONV]], i32* [[CONV1]], double** [[A_ADDR]], double** [[B_ADDR]], double** [[C_ADDR]]) 6926 // CHECK9-NEXT: ret void 6927 // 6928 // 6929 // CHECK9-LABEL: define {{[^@]+}}@.omp_outlined..22 6930 // CHECK9-SAME: (i32* noalias noundef [[DOTGLOBAL_TID_:%.*]], i32* noalias noundef [[DOTBOUND_TID_:%.*]], i32* noundef nonnull align 4 dereferenceable(4) [[CH:%.*]], i32* noundef nonnull align 4 dereferenceable(4) [[N:%.*]], double** noundef nonnull align 8 dereferenceable(8) [[A:%.*]], double** noundef nonnull align 8 dereferenceable(8) [[B:%.*]], double** noundef nonnull align 8 dereferenceable(8) [[C:%.*]]) #[[ATTR1]] { 6931 // CHECK9-NEXT: entry: 6932 // CHECK9-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 6933 // CHECK9-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 6934 // CHECK9-NEXT: [[CH_ADDR:%.*]] = alloca i32*, align 8 6935 // CHECK9-NEXT: [[N_ADDR:%.*]] = alloca i32*, align 8 6936 // CHECK9-NEXT: [[A_ADDR:%.*]] = alloca double**, align 8 6937 // CHECK9-NEXT: [[B_ADDR:%.*]] = alloca double**, align 8 6938 // CHECK9-NEXT: [[C_ADDR:%.*]] = alloca double**, align 8 6939 // CHECK9-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 6940 // CHECK9-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 6941 // CHECK9-NEXT: [[TMP:%.*]] = alloca i32, align 4 6942 // CHECK9-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 6943 // CHECK9-NEXT: [[DOTCAPTURE_EXPR_2:%.*]] = alloca i32, align 4 6944 // CHECK9-NEXT: [[I:%.*]] = alloca i32, align 4 6945 // CHECK9-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4 6946 // CHECK9-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4 6947 // CHECK9-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 6948 // CHECK9-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 6949 // CHECK9-NEXT: [[I4:%.*]] = alloca i32, align 4 6950 // CHECK9-NEXT: [[DOTCAPTURE_EXPR__CASTED:%.*]] = alloca i64, align 8 6951 // CHECK9-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8 6952 // CHECK9-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8 6953 // CHECK9-NEXT: store i32* [[CH]], i32** [[CH_ADDR]], align 8 6954 // CHECK9-NEXT: store i32* [[N]], i32** [[N_ADDR]], align 8 6955 // CHECK9-NEXT: store double** [[A]], double*** [[A_ADDR]], align 8 6956 // CHECK9-NEXT: store double** [[B]], double*** [[B_ADDR]], align 8 6957 // CHECK9-NEXT: store double** [[C]], double*** [[C_ADDR]], align 8 6958 // CHECK9-NEXT: [[TMP0:%.*]] = load i32*, i32** [[CH_ADDR]], align 8 6959 // CHECK9-NEXT: [[TMP1:%.*]] = load i32*, i32** [[N_ADDR]], align 8 6960 // CHECK9-NEXT: [[TMP2:%.*]] = load double**, double*** [[A_ADDR]], align 8 6961 // CHECK9-NEXT: [[TMP3:%.*]] = load double**, double*** [[B_ADDR]], align 8 6962 // CHECK9-NEXT: [[TMP4:%.*]] = load double**, double*** [[C_ADDR]], align 8 6963 // CHECK9-NEXT: [[TMP5:%.*]] = load i32, i32* [[TMP0]], align 4 6964 // CHECK9-NEXT: store i32 [[TMP5]], i32* [[DOTCAPTURE_EXPR_]], align 4 6965 // CHECK9-NEXT: [[TMP6:%.*]] = load i32, i32* [[TMP1]], align 4 6966 // CHECK9-NEXT: store i32 [[TMP6]], i32* [[DOTCAPTURE_EXPR_1]], align 4 6967 // CHECK9-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 6968 // CHECK9-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP7]], 0 6969 // CHECK9-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 6970 // CHECK9-NEXT: [[SUB3:%.*]] = sub nsw i32 [[DIV]], 1 6971 // CHECK9-NEXT: store i32 [[SUB3]], i32* [[DOTCAPTURE_EXPR_2]], align 4 6972 // CHECK9-NEXT: store i32 0, i32* [[I]], align 4 6973 // CHECK9-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 6974 // CHECK9-NEXT: [[CMP:%.*]] = icmp slt i32 0, [[TMP8]] 6975 // CHECK9-NEXT: br i1 [[CMP]], label [[OMP_PRECOND_THEN:%.*]], label [[OMP_PRECOND_END:%.*]] 6976 // CHECK9: omp.precond.then: 6977 // CHECK9-NEXT: store i32 0, i32* [[DOTOMP_COMB_LB]], align 4 6978 // CHECK9-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_2]], align 4 6979 // CHECK9-NEXT: store i32 [[TMP9]], i32* [[DOTOMP_COMB_UB]], align 4 6980 // CHECK9-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4 6981 // CHECK9-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4 6982 // CHECK9-NEXT: [[TMP10:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 6983 // CHECK9-NEXT: [[TMP11:%.*]] = load i32, i32* [[TMP10]], align 4 6984 // CHECK9-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1]], i32 [[TMP11]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_COMB_LB]], i32* [[DOTOMP_COMB_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1) 6985 // CHECK9-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 6986 // CHECK9-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_2]], align 4 6987 // CHECK9-NEXT: [[CMP5:%.*]] = icmp sgt i32 [[TMP12]], [[TMP13]] 6988 // CHECK9-NEXT: br i1 [[CMP5]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] 6989 // CHECK9: cond.true: 6990 // CHECK9-NEXT: [[TMP14:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_2]], align 4 6991 // CHECK9-NEXT: br label [[COND_END:%.*]] 6992 // CHECK9: cond.false: 6993 // CHECK9-NEXT: [[TMP15:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 6994 // CHECK9-NEXT: br label [[COND_END]] 6995 // CHECK9: cond.end: 6996 // CHECK9-NEXT: [[COND:%.*]] = phi i32 [ [[TMP14]], [[COND_TRUE]] ], [ [[TMP15]], [[COND_FALSE]] ] 6997 // CHECK9-NEXT: store i32 [[COND]], i32* [[DOTOMP_COMB_UB]], align 4 6998 // CHECK9-NEXT: [[TMP16:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4 6999 // CHECK9-NEXT: store i32 [[TMP16]], i32* [[DOTOMP_IV]], align 4 7000 // CHECK9-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] 7001 // CHECK9: omp.inner.for.cond: 7002 // CHECK9-NEXT: [[TMP17:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !56 7003 // CHECK9-NEXT: [[TMP18:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4, !llvm.access.group !56 7004 // CHECK9-NEXT: [[CMP6:%.*]] = icmp sle i32 [[TMP17]], [[TMP18]] 7005 // CHECK9-NEXT: br i1 [[CMP6]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] 7006 // CHECK9: omp.inner.for.body: 7007 // CHECK9-NEXT: [[TMP19:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4, !llvm.access.group !56 7008 // CHECK9-NEXT: [[TMP20:%.*]] = zext i32 [[TMP19]] to i64 7009 // CHECK9-NEXT: [[TMP21:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4, !llvm.access.group !56 7010 // CHECK9-NEXT: [[TMP22:%.*]] = zext i32 [[TMP21]] to i64 7011 // CHECK9-NEXT: [[TMP23:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4, !llvm.access.group !56 7012 // CHECK9-NEXT: [[CONV:%.*]] = bitcast i64* [[DOTCAPTURE_EXPR__CASTED]] to i32* 7013 // CHECK9-NEXT: store i32 [[TMP23]], i32* [[CONV]], align 4, !llvm.access.group !56 7014 // CHECK9-NEXT: [[TMP24:%.*]] = load i64, i64* [[DOTCAPTURE_EXPR__CASTED]], align 8, !llvm.access.group !56 7015 // CHECK9-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 7, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64, i32*, double**, double**, double**, i64)* @.omp_outlined..23 to void (i32*, i32*, ...)*), i64 [[TMP20]], i64 [[TMP22]], i32* [[TMP1]], double** [[TMP2]], double** [[TMP3]], double** [[TMP4]], i64 [[TMP24]]), !llvm.access.group !56 7016 // CHECK9-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] 7017 // CHECK9: omp.inner.for.inc: 7018 // CHECK9-NEXT: [[TMP25:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !56 7019 // CHECK9-NEXT: [[TMP26:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4, !llvm.access.group !56 7020 // CHECK9-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP25]], [[TMP26]] 7021 // CHECK9-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4, !llvm.access.group !56 7022 // CHECK9-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP57:![0-9]+]] 7023 // CHECK9: omp.inner.for.end: 7024 // CHECK9-NEXT: br label [[OMP_LOOP_EXIT:%.*]] 7025 // CHECK9: omp.loop.exit: 7026 // CHECK9-NEXT: [[TMP27:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 7027 // CHECK9-NEXT: [[TMP28:%.*]] = load i32, i32* [[TMP27]], align 4 7028 // CHECK9-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP28]]) 7029 // CHECK9-NEXT: [[TMP29:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4 7030 // CHECK9-NEXT: [[TMP30:%.*]] = icmp ne i32 [[TMP29]], 0 7031 // CHECK9-NEXT: br i1 [[TMP30]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]] 7032 // CHECK9: .omp.final.then: 7033 // CHECK9-NEXT: [[TMP31:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 7034 // CHECK9-NEXT: [[SUB7:%.*]] = sub nsw i32 [[TMP31]], 0 7035 // CHECK9-NEXT: [[DIV8:%.*]] = sdiv i32 [[SUB7]], 1 7036 // CHECK9-NEXT: [[MUL:%.*]] = mul nsw i32 [[DIV8]], 1 7037 // CHECK9-NEXT: [[ADD9:%.*]] = add nsw i32 0, [[MUL]] 7038 // CHECK9-NEXT: store i32 [[ADD9]], i32* [[I4]], align 4 7039 // CHECK9-NEXT: br label [[DOTOMP_FINAL_DONE]] 7040 // CHECK9: .omp.final.done: 7041 // CHECK9-NEXT: br label [[OMP_PRECOND_END]] 7042 // CHECK9: omp.precond.end: 7043 // CHECK9-NEXT: ret void 7044 // 7045 // 7046 // CHECK9-LABEL: define {{[^@]+}}@.omp_outlined..23 7047 // CHECK9-SAME: (i32* noalias noundef [[DOTGLOBAL_TID_:%.*]], i32* noalias noundef [[DOTBOUND_TID_:%.*]], i64 noundef [[DOTPREVIOUS_LB_:%.*]], i64 noundef [[DOTPREVIOUS_UB_:%.*]], i32* noundef nonnull align 4 dereferenceable(4) [[N:%.*]], double** noundef nonnull align 8 dereferenceable(8) [[A:%.*]], double** noundef nonnull align 8 dereferenceable(8) [[B:%.*]], double** noundef nonnull align 8 dereferenceable(8) [[C:%.*]], i64 noundef [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR1]] { 7048 // CHECK9-NEXT: entry: 7049 // CHECK9-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 7050 // CHECK9-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 7051 // CHECK9-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8 7052 // CHECK9-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8 7053 // CHECK9-NEXT: [[N_ADDR:%.*]] = alloca i32*, align 8 7054 // CHECK9-NEXT: [[A_ADDR:%.*]] = alloca double**, align 8 7055 // CHECK9-NEXT: [[B_ADDR:%.*]] = alloca double**, align 8 7056 // CHECK9-NEXT: [[C_ADDR:%.*]] = alloca double**, align 8 7057 // CHECK9-NEXT: [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i64, align 8 7058 // CHECK9-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 7059 // CHECK9-NEXT: [[TMP:%.*]] = alloca i32, align 4 7060 // CHECK9-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 7061 // CHECK9-NEXT: [[DOTCAPTURE_EXPR_2:%.*]] = alloca i32, align 4 7062 // CHECK9-NEXT: [[I:%.*]] = alloca i32, align 4 7063 // CHECK9-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 7064 // CHECK9-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4 7065 // CHECK9-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 7066 // CHECK9-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 7067 // CHECK9-NEXT: [[I6:%.*]] = alloca i32, align 4 7068 // CHECK9-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8 7069 // CHECK9-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8 7070 // CHECK9-NEXT: store i64 [[DOTPREVIOUS_LB_]], i64* [[DOTPREVIOUS_LB__ADDR]], align 8 7071 // CHECK9-NEXT: store i64 [[DOTPREVIOUS_UB_]], i64* [[DOTPREVIOUS_UB__ADDR]], align 8 7072 // CHECK9-NEXT: store i32* [[N]], i32** [[N_ADDR]], align 8 7073 // CHECK9-NEXT: store double** [[A]], double*** [[A_ADDR]], align 8 7074 // CHECK9-NEXT: store double** [[B]], double*** [[B_ADDR]], align 8 7075 // CHECK9-NEXT: store double** [[C]], double*** [[C_ADDR]], align 8 7076 // CHECK9-NEXT: store i64 [[DOTCAPTURE_EXPR_]], i64* [[DOTCAPTURE_EXPR__ADDR]], align 8 7077 // CHECK9-NEXT: [[TMP0:%.*]] = load i32*, i32** [[N_ADDR]], align 8 7078 // CHECK9-NEXT: [[TMP1:%.*]] = load double**, double*** [[A_ADDR]], align 8 7079 // CHECK9-NEXT: [[TMP2:%.*]] = load double**, double*** [[B_ADDR]], align 8 7080 // CHECK9-NEXT: [[TMP3:%.*]] = load double**, double*** [[C_ADDR]], align 8 7081 // CHECK9-NEXT: [[CONV:%.*]] = bitcast i64* [[DOTCAPTURE_EXPR__ADDR]] to i32* 7082 // CHECK9-NEXT: [[TMP4:%.*]] = load i32, i32* [[TMP0]], align 4 7083 // CHECK9-NEXT: store i32 [[TMP4]], i32* [[DOTCAPTURE_EXPR_1]], align 4 7084 // CHECK9-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 7085 // CHECK9-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP5]], 0 7086 // CHECK9-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 7087 // CHECK9-NEXT: [[SUB3:%.*]] = sub nsw i32 [[DIV]], 1 7088 // CHECK9-NEXT: store i32 [[SUB3]], i32* [[DOTCAPTURE_EXPR_2]], align 4 7089 // CHECK9-NEXT: store i32 0, i32* [[I]], align 4 7090 // CHECK9-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 7091 // CHECK9-NEXT: [[CMP:%.*]] = icmp slt i32 0, [[TMP6]] 7092 // CHECK9-NEXT: br i1 [[CMP]], label [[OMP_PRECOND_THEN:%.*]], label [[OMP_PRECOND_END:%.*]] 7093 // CHECK9: omp.precond.then: 7094 // CHECK9-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4 7095 // CHECK9-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_2]], align 4 7096 // CHECK9-NEXT: store i32 [[TMP7]], i32* [[DOTOMP_UB]], align 4 7097 // CHECK9-NEXT: [[TMP8:%.*]] = load i64, i64* [[DOTPREVIOUS_LB__ADDR]], align 8 7098 // CHECK9-NEXT: [[CONV4:%.*]] = trunc i64 [[TMP8]] to i32 7099 // CHECK9-NEXT: [[TMP9:%.*]] = load i64, i64* [[DOTPREVIOUS_UB__ADDR]], align 8 7100 // CHECK9-NEXT: [[CONV5:%.*]] = trunc i64 [[TMP9]] to i32 7101 // CHECK9-NEXT: store i32 [[CONV4]], i32* [[DOTOMP_LB]], align 4 7102 // CHECK9-NEXT: store i32 [[CONV5]], i32* [[DOTOMP_UB]], align 4 7103 // CHECK9-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4 7104 // CHECK9-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4 7105 // CHECK9-NEXT: [[TMP10:%.*]] = load i32, i32* [[CONV]], align 4 7106 // CHECK9-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4 7107 // CHECK9-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 7108 // CHECK9-NEXT: [[TMP13:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 7109 // CHECK9-NEXT: [[TMP14:%.*]] = load i32, i32* [[TMP13]], align 4 7110 // CHECK9-NEXT: call void @__kmpc_dispatch_init_4(%struct.ident_t* @[[GLOB3]], i32 [[TMP14]], i32 35, i32 [[TMP11]], i32 [[TMP12]], i32 1, i32 [[TMP10]]) 7111 // CHECK9-NEXT: br label [[OMP_DISPATCH_COND:%.*]] 7112 // CHECK9: omp.dispatch.cond: 7113 // CHECK9-NEXT: [[TMP15:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 7114 // CHECK9-NEXT: [[TMP16:%.*]] = load i32, i32* [[TMP15]], align 4 7115 // CHECK9-NEXT: [[TMP17:%.*]] = call i32 @__kmpc_dispatch_next_4(%struct.ident_t* @[[GLOB3]], i32 [[TMP16]], i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]]) 7116 // CHECK9-NEXT: [[TOBOOL:%.*]] = icmp ne i32 [[TMP17]], 0 7117 // CHECK9-NEXT: br i1 [[TOBOOL]], label [[OMP_DISPATCH_BODY:%.*]], label [[OMP_DISPATCH_END:%.*]] 7118 // CHECK9: omp.dispatch.body: 7119 // CHECK9-NEXT: [[TMP18:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4 7120 // CHECK9-NEXT: store i32 [[TMP18]], i32* [[DOTOMP_IV]], align 4 7121 // CHECK9-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] 7122 // CHECK9: omp.inner.for.cond: 7123 // CHECK9-NEXT: [[TMP19:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !59 7124 // CHECK9-NEXT: [[TMP20:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4, !llvm.access.group !59 7125 // CHECK9-NEXT: [[CMP7:%.*]] = icmp sle i32 [[TMP19]], [[TMP20]] 7126 // CHECK9-NEXT: br i1 [[CMP7]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] 7127 // CHECK9: omp.inner.for.body: 7128 // CHECK9-NEXT: [[TMP21:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !59 7129 // CHECK9-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP21]], 1 7130 // CHECK9-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] 7131 // CHECK9-NEXT: store i32 [[ADD]], i32* [[I6]], align 4, !llvm.access.group !59 7132 // CHECK9-NEXT: [[TMP22:%.*]] = load double*, double** [[TMP2]], align 8, !llvm.access.group !59 7133 // CHECK9-NEXT: [[TMP23:%.*]] = load i32, i32* [[I6]], align 4, !llvm.access.group !59 7134 // CHECK9-NEXT: [[IDXPROM:%.*]] = sext i32 [[TMP23]] to i64 7135 // CHECK9-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds double, double* [[TMP22]], i64 [[IDXPROM]] 7136 // CHECK9-NEXT: [[TMP24:%.*]] = load double, double* [[ARRAYIDX]], align 8, !llvm.access.group !59 7137 // CHECK9-NEXT: [[TMP25:%.*]] = load double*, double** [[TMP3]], align 8, !llvm.access.group !59 7138 // CHECK9-NEXT: [[TMP26:%.*]] = load i32, i32* [[I6]], align 4, !llvm.access.group !59 7139 // CHECK9-NEXT: [[IDXPROM8:%.*]] = sext i32 [[TMP26]] to i64 7140 // CHECK9-NEXT: [[ARRAYIDX9:%.*]] = getelementptr inbounds double, double* [[TMP25]], i64 [[IDXPROM8]] 7141 // CHECK9-NEXT: [[TMP27:%.*]] = load double, double* [[ARRAYIDX9]], align 8, !llvm.access.group !59 7142 // CHECK9-NEXT: [[ADD10:%.*]] = fadd double [[TMP24]], [[TMP27]] 7143 // CHECK9-NEXT: [[TMP28:%.*]] = load double*, double** [[TMP1]], align 8, !llvm.access.group !59 7144 // CHECK9-NEXT: [[TMP29:%.*]] = load i32, i32* [[I6]], align 4, !llvm.access.group !59 7145 // CHECK9-NEXT: [[IDXPROM11:%.*]] = sext i32 [[TMP29]] to i64 7146 // CHECK9-NEXT: [[ARRAYIDX12:%.*]] = getelementptr inbounds double, double* [[TMP28]], i64 [[IDXPROM11]] 7147 // CHECK9-NEXT: store double [[ADD10]], double* [[ARRAYIDX12]], align 8, !llvm.access.group !59 7148 // CHECK9-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] 7149 // CHECK9: omp.body.continue: 7150 // CHECK9-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] 7151 // CHECK9: omp.inner.for.inc: 7152 // CHECK9-NEXT: [[TMP30:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !59 7153 // CHECK9-NEXT: [[ADD13:%.*]] = add nsw i32 [[TMP30]], 1 7154 // CHECK9-NEXT: store i32 [[ADD13]], i32* [[DOTOMP_IV]], align 4, !llvm.access.group !59 7155 // CHECK9-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP60:![0-9]+]] 7156 // CHECK9: omp.inner.for.end: 7157 // CHECK9-NEXT: br label [[OMP_DISPATCH_INC:%.*]] 7158 // CHECK9: omp.dispatch.inc: 7159 // CHECK9-NEXT: br label [[OMP_DISPATCH_COND]] 7160 // CHECK9: omp.dispatch.end: 7161 // CHECK9-NEXT: [[TMP31:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4 7162 // CHECK9-NEXT: [[TMP32:%.*]] = icmp ne i32 [[TMP31]], 0 7163 // CHECK9-NEXT: br i1 [[TMP32]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]] 7164 // CHECK9: .omp.final.then: 7165 // CHECK9-NEXT: [[TMP33:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 7166 // CHECK9-NEXT: [[SUB14:%.*]] = sub nsw i32 [[TMP33]], 0 7167 // CHECK9-NEXT: [[DIV15:%.*]] = sdiv i32 [[SUB14]], 1 7168 // CHECK9-NEXT: [[MUL16:%.*]] = mul nsw i32 [[DIV15]], 1 7169 // CHECK9-NEXT: [[ADD17:%.*]] = add nsw i32 0, [[MUL16]] 7170 // CHECK9-NEXT: store i32 [[ADD17]], i32* [[I6]], align 4 7171 // CHECK9-NEXT: br label [[DOTOMP_FINAL_DONE]] 7172 // CHECK9: .omp.final.done: 7173 // CHECK9-NEXT: br label [[OMP_PRECOND_END]] 7174 // CHECK9: omp.precond.end: 7175 // CHECK9-NEXT: ret void 7176 // 7177 // 7178 // CHECK9-LABEL: define {{[^@]+}}@_Z5tmainIiET_v 7179 // CHECK9-SAME: () #[[ATTR3:[0-9]+]] comdat { 7180 // CHECK9-NEXT: entry: 7181 // CHECK9-NEXT: [[A:%.*]] = alloca i32*, align 8 7182 // CHECK9-NEXT: [[B:%.*]] = alloca i32*, align 8 7183 // CHECK9-NEXT: [[C:%.*]] = alloca i32*, align 8 7184 // CHECK9-NEXT: [[N:%.*]] = alloca i32, align 4 7185 // CHECK9-NEXT: [[CH:%.*]] = alloca i32, align 4 7186 // CHECK9-NEXT: [[N_CASTED:%.*]] = alloca i64, align 8 7187 // CHECK9-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [4 x i8*], align 8 7188 // CHECK9-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [4 x i8*], align 8 7189 // CHECK9-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [4 x i8*], align 8 7190 // CHECK9-NEXT: [[TMP:%.*]] = alloca i32, align 4 7191 // CHECK9-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 7192 // CHECK9-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 7193 // CHECK9-NEXT: [[N_CASTED3:%.*]] = alloca i64, align 8 7194 // CHECK9-NEXT: [[DOTOFFLOAD_BASEPTRS5:%.*]] = alloca [4 x i8*], align 8 7195 // CHECK9-NEXT: [[DOTOFFLOAD_PTRS6:%.*]] = alloca [4 x i8*], align 8 7196 // CHECK9-NEXT: [[DOTOFFLOAD_MAPPERS7:%.*]] = alloca [4 x i8*], align 8 7197 // CHECK9-NEXT: [[_TMP8:%.*]] = alloca i32, align 4 7198 // CHECK9-NEXT: [[DOTCAPTURE_EXPR_9:%.*]] = alloca i32, align 4 7199 // CHECK9-NEXT: [[DOTCAPTURE_EXPR_10:%.*]] = alloca i32, align 4 7200 // CHECK9-NEXT: [[CH_CASTED:%.*]] = alloca i64, align 8 7201 // CHECK9-NEXT: [[N_CASTED19:%.*]] = alloca i64, align 8 7202 // CHECK9-NEXT: [[DOTOFFLOAD_BASEPTRS21:%.*]] = alloca [5 x i8*], align 8 7203 // CHECK9-NEXT: [[DOTOFFLOAD_PTRS22:%.*]] = alloca [5 x i8*], align 8 7204 // CHECK9-NEXT: [[DOTOFFLOAD_MAPPERS23:%.*]] = alloca [5 x i8*], align 8 7205 // CHECK9-NEXT: [[_TMP24:%.*]] = alloca i32, align 4 7206 // CHECK9-NEXT: [[DOTCAPTURE_EXPR_25:%.*]] = alloca i32, align 4 7207 // CHECK9-NEXT: [[DOTCAPTURE_EXPR_26:%.*]] = alloca i32, align 4 7208 // CHECK9-NEXT: [[N_CASTED34:%.*]] = alloca i64, align 8 7209 // CHECK9-NEXT: [[DOTOFFLOAD_BASEPTRS36:%.*]] = alloca [4 x i8*], align 8 7210 // CHECK9-NEXT: [[DOTOFFLOAD_PTRS37:%.*]] = alloca [4 x i8*], align 8 7211 // CHECK9-NEXT: [[DOTOFFLOAD_MAPPERS38:%.*]] = alloca [4 x i8*], align 8 7212 // CHECK9-NEXT: [[_TMP39:%.*]] = alloca i32, align 4 7213 // CHECK9-NEXT: [[DOTCAPTURE_EXPR_40:%.*]] = alloca i32, align 4 7214 // CHECK9-NEXT: [[DOTCAPTURE_EXPR_41:%.*]] = alloca i32, align 4 7215 // CHECK9-NEXT: [[CH_CASTED49:%.*]] = alloca i64, align 8 7216 // CHECK9-NEXT: [[N_CASTED51:%.*]] = alloca i64, align 8 7217 // CHECK9-NEXT: [[DOTOFFLOAD_BASEPTRS53:%.*]] = alloca [5 x i8*], align 8 7218 // CHECK9-NEXT: [[DOTOFFLOAD_PTRS54:%.*]] = alloca [5 x i8*], align 8 7219 // CHECK9-NEXT: [[DOTOFFLOAD_MAPPERS55:%.*]] = alloca [5 x i8*], align 8 7220 // CHECK9-NEXT: [[_TMP56:%.*]] = alloca i32, align 4 7221 // CHECK9-NEXT: [[DOTCAPTURE_EXPR_57:%.*]] = alloca i32, align 4 7222 // CHECK9-NEXT: [[DOTCAPTURE_EXPR_58:%.*]] = alloca i32, align 4 7223 // CHECK9-NEXT: [[N_CASTED66:%.*]] = alloca i64, align 8 7224 // CHECK9-NEXT: [[DOTOFFLOAD_BASEPTRS68:%.*]] = alloca [4 x i8*], align 8 7225 // CHECK9-NEXT: [[DOTOFFLOAD_PTRS69:%.*]] = alloca [4 x i8*], align 8 7226 // CHECK9-NEXT: [[DOTOFFLOAD_MAPPERS70:%.*]] = alloca [4 x i8*], align 8 7227 // CHECK9-NEXT: [[_TMP71:%.*]] = alloca i32, align 4 7228 // CHECK9-NEXT: [[DOTCAPTURE_EXPR_72:%.*]] = alloca i32, align 4 7229 // CHECK9-NEXT: [[DOTCAPTURE_EXPR_73:%.*]] = alloca i32, align 4 7230 // CHECK9-NEXT: [[CH_CASTED81:%.*]] = alloca i64, align 8 7231 // CHECK9-NEXT: [[N_CASTED83:%.*]] = alloca i64, align 8 7232 // CHECK9-NEXT: [[DOTOFFLOAD_BASEPTRS85:%.*]] = alloca [5 x i8*], align 8 7233 // CHECK9-NEXT: [[DOTOFFLOAD_PTRS86:%.*]] = alloca [5 x i8*], align 8 7234 // CHECK9-NEXT: [[DOTOFFLOAD_MAPPERS87:%.*]] = alloca [5 x i8*], align 8 7235 // CHECK9-NEXT: [[_TMP88:%.*]] = alloca i32, align 4 7236 // CHECK9-NEXT: [[DOTCAPTURE_EXPR_89:%.*]] = alloca i32, align 4 7237 // CHECK9-NEXT: [[DOTCAPTURE_EXPR_90:%.*]] = alloca i32, align 4 7238 // CHECK9-NEXT: store i32 10000, i32* [[N]], align 4 7239 // CHECK9-NEXT: store i32 100, i32* [[CH]], align 4 7240 // CHECK9-NEXT: [[TMP0:%.*]] = load i32, i32* [[N]], align 4 7241 // CHECK9-NEXT: [[CONV:%.*]] = bitcast i64* [[N_CASTED]] to i32* 7242 // CHECK9-NEXT: store i32 [[TMP0]], i32* [[CONV]], align 4 7243 // CHECK9-NEXT: [[TMP1:%.*]] = load i64, i64* [[N_CASTED]], align 8 7244 // CHECK9-NEXT: [[TMP2:%.*]] = load i32*, i32** [[A]], align 8 7245 // CHECK9-NEXT: [[TMP3:%.*]] = load i32*, i32** [[B]], align 8 7246 // CHECK9-NEXT: [[TMP4:%.*]] = load i32*, i32** [[C]], align 8 7247 // CHECK9-NEXT: [[TMP5:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 7248 // CHECK9-NEXT: [[TMP6:%.*]] = bitcast i8** [[TMP5]] to i64* 7249 // CHECK9-NEXT: store i64 [[TMP1]], i64* [[TMP6]], align 8 7250 // CHECK9-NEXT: [[TMP7:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 7251 // CHECK9-NEXT: [[TMP8:%.*]] = bitcast i8** [[TMP7]] to i64* 7252 // CHECK9-NEXT: store i64 [[TMP1]], i64* [[TMP8]], align 8 7253 // CHECK9-NEXT: [[TMP9:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0 7254 // CHECK9-NEXT: store i8* null, i8** [[TMP9]], align 8 7255 // CHECK9-NEXT: [[TMP10:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1 7256 // CHECK9-NEXT: [[TMP11:%.*]] = bitcast i8** [[TMP10]] to i32** 7257 // CHECK9-NEXT: store i32* [[TMP2]], i32** [[TMP11]], align 8 7258 // CHECK9-NEXT: [[TMP12:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1 7259 // CHECK9-NEXT: [[TMP13:%.*]] = bitcast i8** [[TMP12]] to i32** 7260 // CHECK9-NEXT: store i32* [[TMP2]], i32** [[TMP13]], align 8 7261 // CHECK9-NEXT: [[TMP14:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 1 7262 // CHECK9-NEXT: store i8* null, i8** [[TMP14]], align 8 7263 // CHECK9-NEXT: [[TMP15:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2 7264 // CHECK9-NEXT: [[TMP16:%.*]] = bitcast i8** [[TMP15]] to i32** 7265 // CHECK9-NEXT: store i32* [[TMP3]], i32** [[TMP16]], align 8 7266 // CHECK9-NEXT: [[TMP17:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2 7267 // CHECK9-NEXT: [[TMP18:%.*]] = bitcast i8** [[TMP17]] to i32** 7268 // CHECK9-NEXT: store i32* [[TMP3]], i32** [[TMP18]], align 8 7269 // CHECK9-NEXT: [[TMP19:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2 7270 // CHECK9-NEXT: store i8* null, i8** [[TMP19]], align 8 7271 // CHECK9-NEXT: [[TMP20:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 3 7272 // CHECK9-NEXT: [[TMP21:%.*]] = bitcast i8** [[TMP20]] to i32** 7273 // CHECK9-NEXT: store i32* [[TMP4]], i32** [[TMP21]], align 8 7274 // CHECK9-NEXT: [[TMP22:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 3 7275 // CHECK9-NEXT: [[TMP23:%.*]] = bitcast i8** [[TMP22]] to i32** 7276 // CHECK9-NEXT: store i32* [[TMP4]], i32** [[TMP23]], align 8 7277 // CHECK9-NEXT: [[TMP24:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 3 7278 // CHECK9-NEXT: store i8* null, i8** [[TMP24]], align 8 7279 // CHECK9-NEXT: [[TMP25:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 7280 // CHECK9-NEXT: [[TMP26:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 7281 // CHECK9-NEXT: [[TMP27:%.*]] = load i32, i32* [[N]], align 4 7282 // CHECK9-NEXT: store i32 [[TMP27]], i32* [[DOTCAPTURE_EXPR_]], align 4 7283 // CHECK9-NEXT: [[TMP28:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 7284 // CHECK9-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP28]], 0 7285 // CHECK9-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 7286 // CHECK9-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1 7287 // CHECK9-NEXT: store i32 [[SUB2]], i32* [[DOTCAPTURE_EXPR_1]], align 4 7288 // CHECK9-NEXT: [[TMP29:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 7289 // CHECK9-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP29]], 1 7290 // CHECK9-NEXT: [[TMP30:%.*]] = zext i32 [[ADD]] to i64 7291 // CHECK9-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8 7292 // CHECK9-NEXT: [[TMP31:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS]], i32 0, i32 0 7293 // CHECK9-NEXT: store i32 1, i32* [[TMP31]], align 4 7294 // CHECK9-NEXT: [[TMP32:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS]], i32 0, i32 1 7295 // CHECK9-NEXT: store i32 4, i32* [[TMP32]], align 4 7296 // CHECK9-NEXT: [[TMP33:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS]], i32 0, i32 2 7297 // CHECK9-NEXT: store i8** [[TMP25]], i8*** [[TMP33]], align 8 7298 // CHECK9-NEXT: [[TMP34:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS]], i32 0, i32 3 7299 // CHECK9-NEXT: store i8** [[TMP26]], i8*** [[TMP34]], align 8 7300 // CHECK9-NEXT: [[TMP35:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS]], i32 0, i32 4 7301 // CHECK9-NEXT: store i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_sizes.28, i32 0, i32 0), i64** [[TMP35]], align 8 7302 // CHECK9-NEXT: [[TMP36:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS]], i32 0, i32 5 7303 // CHECK9-NEXT: store i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_maptypes.29, i32 0, i32 0), i64** [[TMP36]], align 8 7304 // CHECK9-NEXT: [[TMP37:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS]], i32 0, i32 6 7305 // CHECK9-NEXT: store i8** null, i8*** [[TMP37]], align 8 7306 // CHECK9-NEXT: [[TMP38:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS]], i32 0, i32 7 7307 // CHECK9-NEXT: store i8** null, i8*** [[TMP38]], align 8 7308 // CHECK9-NEXT: [[TMP39:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS]], i32 0, i32 8 7309 // CHECK9-NEXT: store i64 [[TMP30]], i64* [[TMP39]], align 8 7310 // CHECK9-NEXT: [[TMP40:%.*]] = call i32 @__tgt_target_kernel(%struct.ident_t* @[[GLOB3]], i64 -1, i32 0, i32 1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiET_v_l42.region_id, %struct.__tgt_kernel_arguments* [[KERNEL_ARGS]]) 7311 // CHECK9-NEXT: [[TMP41:%.*]] = icmp ne i32 [[TMP40]], 0 7312 // CHECK9-NEXT: br i1 [[TMP41]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] 7313 // CHECK9: omp_offload.failed: 7314 // CHECK9-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiET_v_l42(i64 [[TMP1]], i32* [[TMP2]], i32* [[TMP3]], i32* [[TMP4]]) #[[ATTR2]] 7315 // CHECK9-NEXT: br label [[OMP_OFFLOAD_CONT]] 7316 // CHECK9: omp_offload.cont: 7317 // CHECK9-NEXT: [[TMP42:%.*]] = load i32, i32* [[N]], align 4 7318 // CHECK9-NEXT: [[CONV4:%.*]] = bitcast i64* [[N_CASTED3]] to i32* 7319 // CHECK9-NEXT: store i32 [[TMP42]], i32* [[CONV4]], align 4 7320 // CHECK9-NEXT: [[TMP43:%.*]] = load i64, i64* [[N_CASTED3]], align 8 7321 // CHECK9-NEXT: [[TMP44:%.*]] = load i32*, i32** [[A]], align 8 7322 // CHECK9-NEXT: [[TMP45:%.*]] = load i32*, i32** [[B]], align 8 7323 // CHECK9-NEXT: [[TMP46:%.*]] = load i32*, i32** [[C]], align 8 7324 // CHECK9-NEXT: [[TMP47:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS5]], i32 0, i32 0 7325 // CHECK9-NEXT: [[TMP48:%.*]] = bitcast i8** [[TMP47]] to i64* 7326 // CHECK9-NEXT: store i64 [[TMP43]], i64* [[TMP48]], align 8 7327 // CHECK9-NEXT: [[TMP49:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS6]], i32 0, i32 0 7328 // CHECK9-NEXT: [[TMP50:%.*]] = bitcast i8** [[TMP49]] to i64* 7329 // CHECK9-NEXT: store i64 [[TMP43]], i64* [[TMP50]], align 8 7330 // CHECK9-NEXT: [[TMP51:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS7]], i64 0, i64 0 7331 // CHECK9-NEXT: store i8* null, i8** [[TMP51]], align 8 7332 // CHECK9-NEXT: [[TMP52:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS5]], i32 0, i32 1 7333 // CHECK9-NEXT: [[TMP53:%.*]] = bitcast i8** [[TMP52]] to i32** 7334 // CHECK9-NEXT: store i32* [[TMP44]], i32** [[TMP53]], align 8 7335 // CHECK9-NEXT: [[TMP54:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS6]], i32 0, i32 1 7336 // CHECK9-NEXT: [[TMP55:%.*]] = bitcast i8** [[TMP54]] to i32** 7337 // CHECK9-NEXT: store i32* [[TMP44]], i32** [[TMP55]], align 8 7338 // CHECK9-NEXT: [[TMP56:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS7]], i64 0, i64 1 7339 // CHECK9-NEXT: store i8* null, i8** [[TMP56]], align 8 7340 // CHECK9-NEXT: [[TMP57:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS5]], i32 0, i32 2 7341 // CHECK9-NEXT: [[TMP58:%.*]] = bitcast i8** [[TMP57]] to i32** 7342 // CHECK9-NEXT: store i32* [[TMP45]], i32** [[TMP58]], align 8 7343 // CHECK9-NEXT: [[TMP59:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS6]], i32 0, i32 2 7344 // CHECK9-NEXT: [[TMP60:%.*]] = bitcast i8** [[TMP59]] to i32** 7345 // CHECK9-NEXT: store i32* [[TMP45]], i32** [[TMP60]], align 8 7346 // CHECK9-NEXT: [[TMP61:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS7]], i64 0, i64 2 7347 // CHECK9-NEXT: store i8* null, i8** [[TMP61]], align 8 7348 // CHECK9-NEXT: [[TMP62:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS5]], i32 0, i32 3 7349 // CHECK9-NEXT: [[TMP63:%.*]] = bitcast i8** [[TMP62]] to i32** 7350 // CHECK9-NEXT: store i32* [[TMP46]], i32** [[TMP63]], align 8 7351 // CHECK9-NEXT: [[TMP64:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS6]], i32 0, i32 3 7352 // CHECK9-NEXT: [[TMP65:%.*]] = bitcast i8** [[TMP64]] to i32** 7353 // CHECK9-NEXT: store i32* [[TMP46]], i32** [[TMP65]], align 8 7354 // CHECK9-NEXT: [[TMP66:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS7]], i64 0, i64 3 7355 // CHECK9-NEXT: store i8* null, i8** [[TMP66]], align 8 7356 // CHECK9-NEXT: [[TMP67:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS5]], i32 0, i32 0 7357 // CHECK9-NEXT: [[TMP68:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS6]], i32 0, i32 0 7358 // CHECK9-NEXT: [[TMP69:%.*]] = load i32, i32* [[N]], align 4 7359 // CHECK9-NEXT: store i32 [[TMP69]], i32* [[DOTCAPTURE_EXPR_9]], align 4 7360 // CHECK9-NEXT: [[TMP70:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_9]], align 4 7361 // CHECK9-NEXT: [[SUB11:%.*]] = sub nsw i32 [[TMP70]], 0 7362 // CHECK9-NEXT: [[DIV12:%.*]] = sdiv i32 [[SUB11]], 1 7363 // CHECK9-NEXT: [[SUB13:%.*]] = sub nsw i32 [[DIV12]], 1 7364 // CHECK9-NEXT: store i32 [[SUB13]], i32* [[DOTCAPTURE_EXPR_10]], align 4 7365 // CHECK9-NEXT: [[TMP71:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_10]], align 4 7366 // CHECK9-NEXT: [[ADD14:%.*]] = add nsw i32 [[TMP71]], 1 7367 // CHECK9-NEXT: [[TMP72:%.*]] = zext i32 [[ADD14]] to i64 7368 // CHECK9-NEXT: [[KERNEL_ARGS15:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS]], align 8 7369 // CHECK9-NEXT: [[TMP73:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS15]], i32 0, i32 0 7370 // CHECK9-NEXT: store i32 1, i32* [[TMP73]], align 4 7371 // CHECK9-NEXT: [[TMP74:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS15]], i32 0, i32 1 7372 // CHECK9-NEXT: store i32 4, i32* [[TMP74]], align 4 7373 // CHECK9-NEXT: [[TMP75:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS15]], i32 0, i32 2 7374 // CHECK9-NEXT: store i8** [[TMP67]], i8*** [[TMP75]], align 8 7375 // CHECK9-NEXT: [[TMP76:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS15]], i32 0, i32 3 7376 // CHECK9-NEXT: store i8** [[TMP68]], i8*** [[TMP76]], align 8 7377 // CHECK9-NEXT: [[TMP77:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS15]], i32 0, i32 4 7378 // CHECK9-NEXT: store i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_sizes.32, i32 0, i32 0), i64** [[TMP77]], align 8 7379 // CHECK9-NEXT: [[TMP78:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS15]], i32 0, i32 5 7380 // CHECK9-NEXT: store i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_maptypes.33, i32 0, i32 0), i64** [[TMP78]], align 8 7381 // CHECK9-NEXT: [[TMP79:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS15]], i32 0, i32 6 7382 // CHECK9-NEXT: store i8** null, i8*** [[TMP79]], align 8 7383 // CHECK9-NEXT: [[TMP80:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS15]], i32 0, i32 7 7384 // CHECK9-NEXT: store i8** null, i8*** [[TMP80]], align 8 7385 // CHECK9-NEXT: [[TMP81:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS15]], i32 0, i32 8 7386 // CHECK9-NEXT: store i64 [[TMP72]], i64* [[TMP81]], align 8 7387 // CHECK9-NEXT: [[TMP82:%.*]] = call i32 @__tgt_target_kernel(%struct.ident_t* @[[GLOB3]], i64 -1, i32 0, i32 1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiET_v_l50.region_id, %struct.__tgt_kernel_arguments* [[KERNEL_ARGS15]]) 7388 // CHECK9-NEXT: [[TMP83:%.*]] = icmp ne i32 [[TMP82]], 0 7389 // CHECK9-NEXT: br i1 [[TMP83]], label [[OMP_OFFLOAD_FAILED16:%.*]], label [[OMP_OFFLOAD_CONT17:%.*]] 7390 // CHECK9: omp_offload.failed16: 7391 // CHECK9-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiET_v_l50(i64 [[TMP43]], i32* [[TMP44]], i32* [[TMP45]], i32* [[TMP46]]) #[[ATTR2]] 7392 // CHECK9-NEXT: br label [[OMP_OFFLOAD_CONT17]] 7393 // CHECK9: omp_offload.cont17: 7394 // CHECK9-NEXT: [[TMP84:%.*]] = load i32, i32* [[CH]], align 4 7395 // CHECK9-NEXT: [[CONV18:%.*]] = bitcast i64* [[CH_CASTED]] to i32* 7396 // CHECK9-NEXT: store i32 [[TMP84]], i32* [[CONV18]], align 4 7397 // CHECK9-NEXT: [[TMP85:%.*]] = load i64, i64* [[CH_CASTED]], align 8 7398 // CHECK9-NEXT: [[TMP86:%.*]] = load i32, i32* [[N]], align 4 7399 // CHECK9-NEXT: [[CONV20:%.*]] = bitcast i64* [[N_CASTED19]] to i32* 7400 // CHECK9-NEXT: store i32 [[TMP86]], i32* [[CONV20]], align 4 7401 // CHECK9-NEXT: [[TMP87:%.*]] = load i64, i64* [[N_CASTED19]], align 8 7402 // CHECK9-NEXT: [[TMP88:%.*]] = load i32*, i32** [[A]], align 8 7403 // CHECK9-NEXT: [[TMP89:%.*]] = load i32*, i32** [[B]], align 8 7404 // CHECK9-NEXT: [[TMP90:%.*]] = load i32*, i32** [[C]], align 8 7405 // CHECK9-NEXT: [[TMP91:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS21]], i32 0, i32 0 7406 // CHECK9-NEXT: [[TMP92:%.*]] = bitcast i8** [[TMP91]] to i64* 7407 // CHECK9-NEXT: store i64 [[TMP85]], i64* [[TMP92]], align 8 7408 // CHECK9-NEXT: [[TMP93:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS22]], i32 0, i32 0 7409 // CHECK9-NEXT: [[TMP94:%.*]] = bitcast i8** [[TMP93]] to i64* 7410 // CHECK9-NEXT: store i64 [[TMP85]], i64* [[TMP94]], align 8 7411 // CHECK9-NEXT: [[TMP95:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS23]], i64 0, i64 0 7412 // CHECK9-NEXT: store i8* null, i8** [[TMP95]], align 8 7413 // CHECK9-NEXT: [[TMP96:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS21]], i32 0, i32 1 7414 // CHECK9-NEXT: [[TMP97:%.*]] = bitcast i8** [[TMP96]] to i64* 7415 // CHECK9-NEXT: store i64 [[TMP87]], i64* [[TMP97]], align 8 7416 // CHECK9-NEXT: [[TMP98:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS22]], i32 0, i32 1 7417 // CHECK9-NEXT: [[TMP99:%.*]] = bitcast i8** [[TMP98]] to i64* 7418 // CHECK9-NEXT: store i64 [[TMP87]], i64* [[TMP99]], align 8 7419 // CHECK9-NEXT: [[TMP100:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS23]], i64 0, i64 1 7420 // CHECK9-NEXT: store i8* null, i8** [[TMP100]], align 8 7421 // CHECK9-NEXT: [[TMP101:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS21]], i32 0, i32 2 7422 // CHECK9-NEXT: [[TMP102:%.*]] = bitcast i8** [[TMP101]] to i32** 7423 // CHECK9-NEXT: store i32* [[TMP88]], i32** [[TMP102]], align 8 7424 // CHECK9-NEXT: [[TMP103:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS22]], i32 0, i32 2 7425 // CHECK9-NEXT: [[TMP104:%.*]] = bitcast i8** [[TMP103]] to i32** 7426 // CHECK9-NEXT: store i32* [[TMP88]], i32** [[TMP104]], align 8 7427 // CHECK9-NEXT: [[TMP105:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS23]], i64 0, i64 2 7428 // CHECK9-NEXT: store i8* null, i8** [[TMP105]], align 8 7429 // CHECK9-NEXT: [[TMP106:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS21]], i32 0, i32 3 7430 // CHECK9-NEXT: [[TMP107:%.*]] = bitcast i8** [[TMP106]] to i32** 7431 // CHECK9-NEXT: store i32* [[TMP89]], i32** [[TMP107]], align 8 7432 // CHECK9-NEXT: [[TMP108:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS22]], i32 0, i32 3 7433 // CHECK9-NEXT: [[TMP109:%.*]] = bitcast i8** [[TMP108]] to i32** 7434 // CHECK9-NEXT: store i32* [[TMP89]], i32** [[TMP109]], align 8 7435 // CHECK9-NEXT: [[TMP110:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS23]], i64 0, i64 3 7436 // CHECK9-NEXT: store i8* null, i8** [[TMP110]], align 8 7437 // CHECK9-NEXT: [[TMP111:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS21]], i32 0, i32 4 7438 // CHECK9-NEXT: [[TMP112:%.*]] = bitcast i8** [[TMP111]] to i32** 7439 // CHECK9-NEXT: store i32* [[TMP90]], i32** [[TMP112]], align 8 7440 // CHECK9-NEXT: [[TMP113:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS22]], i32 0, i32 4 7441 // CHECK9-NEXT: [[TMP114:%.*]] = bitcast i8** [[TMP113]] to i32** 7442 // CHECK9-NEXT: store i32* [[TMP90]], i32** [[TMP114]], align 8 7443 // CHECK9-NEXT: [[TMP115:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS23]], i64 0, i64 4 7444 // CHECK9-NEXT: store i8* null, i8** [[TMP115]], align 8 7445 // CHECK9-NEXT: [[TMP116:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS21]], i32 0, i32 0 7446 // CHECK9-NEXT: [[TMP117:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS22]], i32 0, i32 0 7447 // CHECK9-NEXT: [[TMP118:%.*]] = load i32, i32* [[N]], align 4 7448 // CHECK9-NEXT: store i32 [[TMP118]], i32* [[DOTCAPTURE_EXPR_25]], align 4 7449 // CHECK9-NEXT: [[TMP119:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_25]], align 4 7450 // CHECK9-NEXT: [[SUB27:%.*]] = sub nsw i32 [[TMP119]], 0 7451 // CHECK9-NEXT: [[DIV28:%.*]] = sdiv i32 [[SUB27]], 1 7452 // CHECK9-NEXT: [[SUB29:%.*]] = sub nsw i32 [[DIV28]], 1 7453 // CHECK9-NEXT: store i32 [[SUB29]], i32* [[DOTCAPTURE_EXPR_26]], align 4 7454 // CHECK9-NEXT: [[TMP120:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_26]], align 4 7455 // CHECK9-NEXT: [[ADD30:%.*]] = add nsw i32 [[TMP120]], 1 7456 // CHECK9-NEXT: [[TMP121:%.*]] = zext i32 [[ADD30]] to i64 7457 // CHECK9-NEXT: [[KERNEL_ARGS31:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS]], align 8 7458 // CHECK9-NEXT: [[TMP122:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS31]], i32 0, i32 0 7459 // CHECK9-NEXT: store i32 1, i32* [[TMP122]], align 4 7460 // CHECK9-NEXT: [[TMP123:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS31]], i32 0, i32 1 7461 // CHECK9-NEXT: store i32 5, i32* [[TMP123]], align 4 7462 // CHECK9-NEXT: [[TMP124:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS31]], i32 0, i32 2 7463 // CHECK9-NEXT: store i8** [[TMP116]], i8*** [[TMP124]], align 8 7464 // CHECK9-NEXT: [[TMP125:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS31]], i32 0, i32 3 7465 // CHECK9-NEXT: store i8** [[TMP117]], i8*** [[TMP125]], align 8 7466 // CHECK9-NEXT: [[TMP126:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS31]], i32 0, i32 4 7467 // CHECK9-NEXT: store i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_sizes.36, i32 0, i32 0), i64** [[TMP126]], align 8 7468 // CHECK9-NEXT: [[TMP127:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS31]], i32 0, i32 5 7469 // CHECK9-NEXT: store i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_maptypes.37, i32 0, i32 0), i64** [[TMP127]], align 8 7470 // CHECK9-NEXT: [[TMP128:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS31]], i32 0, i32 6 7471 // CHECK9-NEXT: store i8** null, i8*** [[TMP128]], align 8 7472 // CHECK9-NEXT: [[TMP129:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS31]], i32 0, i32 7 7473 // CHECK9-NEXT: store i8** null, i8*** [[TMP129]], align 8 7474 // CHECK9-NEXT: [[TMP130:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS31]], i32 0, i32 8 7475 // CHECK9-NEXT: store i64 [[TMP121]], i64* [[TMP130]], align 8 7476 // CHECK9-NEXT: [[TMP131:%.*]] = call i32 @__tgt_target_kernel(%struct.ident_t* @[[GLOB3]], i64 -1, i32 0, i32 1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiET_v_l58.region_id, %struct.__tgt_kernel_arguments* [[KERNEL_ARGS31]]) 7477 // CHECK9-NEXT: [[TMP132:%.*]] = icmp ne i32 [[TMP131]], 0 7478 // CHECK9-NEXT: br i1 [[TMP132]], label [[OMP_OFFLOAD_FAILED32:%.*]], label [[OMP_OFFLOAD_CONT33:%.*]] 7479 // CHECK9: omp_offload.failed32: 7480 // CHECK9-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiET_v_l58(i64 [[TMP85]], i64 [[TMP87]], i32* [[TMP88]], i32* [[TMP89]], i32* [[TMP90]]) #[[ATTR2]] 7481 // CHECK9-NEXT: br label [[OMP_OFFLOAD_CONT33]] 7482 // CHECK9: omp_offload.cont33: 7483 // CHECK9-NEXT: [[TMP133:%.*]] = load i32, i32* [[N]], align 4 7484 // CHECK9-NEXT: [[CONV35:%.*]] = bitcast i64* [[N_CASTED34]] to i32* 7485 // CHECK9-NEXT: store i32 [[TMP133]], i32* [[CONV35]], align 4 7486 // CHECK9-NEXT: [[TMP134:%.*]] = load i64, i64* [[N_CASTED34]], align 8 7487 // CHECK9-NEXT: [[TMP135:%.*]] = load i32*, i32** [[A]], align 8 7488 // CHECK9-NEXT: [[TMP136:%.*]] = load i32*, i32** [[B]], align 8 7489 // CHECK9-NEXT: [[TMP137:%.*]] = load i32*, i32** [[C]], align 8 7490 // CHECK9-NEXT: [[TMP138:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS36]], i32 0, i32 0 7491 // CHECK9-NEXT: [[TMP139:%.*]] = bitcast i8** [[TMP138]] to i64* 7492 // CHECK9-NEXT: store i64 [[TMP134]], i64* [[TMP139]], align 8 7493 // CHECK9-NEXT: [[TMP140:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS37]], i32 0, i32 0 7494 // CHECK9-NEXT: [[TMP141:%.*]] = bitcast i8** [[TMP140]] to i64* 7495 // CHECK9-NEXT: store i64 [[TMP134]], i64* [[TMP141]], align 8 7496 // CHECK9-NEXT: [[TMP142:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS38]], i64 0, i64 0 7497 // CHECK9-NEXT: store i8* null, i8** [[TMP142]], align 8 7498 // CHECK9-NEXT: [[TMP143:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS36]], i32 0, i32 1 7499 // CHECK9-NEXT: [[TMP144:%.*]] = bitcast i8** [[TMP143]] to i32** 7500 // CHECK9-NEXT: store i32* [[TMP135]], i32** [[TMP144]], align 8 7501 // CHECK9-NEXT: [[TMP145:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS37]], i32 0, i32 1 7502 // CHECK9-NEXT: [[TMP146:%.*]] = bitcast i8** [[TMP145]] to i32** 7503 // CHECK9-NEXT: store i32* [[TMP135]], i32** [[TMP146]], align 8 7504 // CHECK9-NEXT: [[TMP147:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS38]], i64 0, i64 1 7505 // CHECK9-NEXT: store i8* null, i8** [[TMP147]], align 8 7506 // CHECK9-NEXT: [[TMP148:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS36]], i32 0, i32 2 7507 // CHECK9-NEXT: [[TMP149:%.*]] = bitcast i8** [[TMP148]] to i32** 7508 // CHECK9-NEXT: store i32* [[TMP136]], i32** [[TMP149]], align 8 7509 // CHECK9-NEXT: [[TMP150:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS37]], i32 0, i32 2 7510 // CHECK9-NEXT: [[TMP151:%.*]] = bitcast i8** [[TMP150]] to i32** 7511 // CHECK9-NEXT: store i32* [[TMP136]], i32** [[TMP151]], align 8 7512 // CHECK9-NEXT: [[TMP152:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS38]], i64 0, i64 2 7513 // CHECK9-NEXT: store i8* null, i8** [[TMP152]], align 8 7514 // CHECK9-NEXT: [[TMP153:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS36]], i32 0, i32 3 7515 // CHECK9-NEXT: [[TMP154:%.*]] = bitcast i8** [[TMP153]] to i32** 7516 // CHECK9-NEXT: store i32* [[TMP137]], i32** [[TMP154]], align 8 7517 // CHECK9-NEXT: [[TMP155:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS37]], i32 0, i32 3 7518 // CHECK9-NEXT: [[TMP156:%.*]] = bitcast i8** [[TMP155]] to i32** 7519 // CHECK9-NEXT: store i32* [[TMP137]], i32** [[TMP156]], align 8 7520 // CHECK9-NEXT: [[TMP157:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS38]], i64 0, i64 3 7521 // CHECK9-NEXT: store i8* null, i8** [[TMP157]], align 8 7522 // CHECK9-NEXT: [[TMP158:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS36]], i32 0, i32 0 7523 // CHECK9-NEXT: [[TMP159:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS37]], i32 0, i32 0 7524 // CHECK9-NEXT: [[TMP160:%.*]] = load i32, i32* [[N]], align 4 7525 // CHECK9-NEXT: store i32 [[TMP160]], i32* [[DOTCAPTURE_EXPR_40]], align 4 7526 // CHECK9-NEXT: [[TMP161:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_40]], align 4 7527 // CHECK9-NEXT: [[SUB42:%.*]] = sub nsw i32 [[TMP161]], 0 7528 // CHECK9-NEXT: [[DIV43:%.*]] = sdiv i32 [[SUB42]], 1 7529 // CHECK9-NEXT: [[SUB44:%.*]] = sub nsw i32 [[DIV43]], 1 7530 // CHECK9-NEXT: store i32 [[SUB44]], i32* [[DOTCAPTURE_EXPR_41]], align 4 7531 // CHECK9-NEXT: [[TMP162:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_41]], align 4 7532 // CHECK9-NEXT: [[ADD45:%.*]] = add nsw i32 [[TMP162]], 1 7533 // CHECK9-NEXT: [[TMP163:%.*]] = zext i32 [[ADD45]] to i64 7534 // CHECK9-NEXT: [[KERNEL_ARGS46:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS]], align 8 7535 // CHECK9-NEXT: [[TMP164:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS46]], i32 0, i32 0 7536 // CHECK9-NEXT: store i32 1, i32* [[TMP164]], align 4 7537 // CHECK9-NEXT: [[TMP165:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS46]], i32 0, i32 1 7538 // CHECK9-NEXT: store i32 4, i32* [[TMP165]], align 4 7539 // CHECK9-NEXT: [[TMP166:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS46]], i32 0, i32 2 7540 // CHECK9-NEXT: store i8** [[TMP158]], i8*** [[TMP166]], align 8 7541 // CHECK9-NEXT: [[TMP167:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS46]], i32 0, i32 3 7542 // CHECK9-NEXT: store i8** [[TMP159]], i8*** [[TMP167]], align 8 7543 // CHECK9-NEXT: [[TMP168:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS46]], i32 0, i32 4 7544 // CHECK9-NEXT: store i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_sizes.40, i32 0, i32 0), i64** [[TMP168]], align 8 7545 // CHECK9-NEXT: [[TMP169:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS46]], i32 0, i32 5 7546 // CHECK9-NEXT: store i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_maptypes.41, i32 0, i32 0), i64** [[TMP169]], align 8 7547 // CHECK9-NEXT: [[TMP170:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS46]], i32 0, i32 6 7548 // CHECK9-NEXT: store i8** null, i8*** [[TMP170]], align 8 7549 // CHECK9-NEXT: [[TMP171:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS46]], i32 0, i32 7 7550 // CHECK9-NEXT: store i8** null, i8*** [[TMP171]], align 8 7551 // CHECK9-NEXT: [[TMP172:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS46]], i32 0, i32 8 7552 // CHECK9-NEXT: store i64 [[TMP163]], i64* [[TMP172]], align 8 7553 // CHECK9-NEXT: [[TMP173:%.*]] = call i32 @__tgt_target_kernel(%struct.ident_t* @[[GLOB3]], i64 -1, i32 0, i32 1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiET_v_l66.region_id, %struct.__tgt_kernel_arguments* [[KERNEL_ARGS46]]) 7554 // CHECK9-NEXT: [[TMP174:%.*]] = icmp ne i32 [[TMP173]], 0 7555 // CHECK9-NEXT: br i1 [[TMP174]], label [[OMP_OFFLOAD_FAILED47:%.*]], label [[OMP_OFFLOAD_CONT48:%.*]] 7556 // CHECK9: omp_offload.failed47: 7557 // CHECK9-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiET_v_l66(i64 [[TMP134]], i32* [[TMP135]], i32* [[TMP136]], i32* [[TMP137]]) #[[ATTR2]] 7558 // CHECK9-NEXT: br label [[OMP_OFFLOAD_CONT48]] 7559 // CHECK9: omp_offload.cont48: 7560 // CHECK9-NEXT: [[TMP175:%.*]] = load i32, i32* [[CH]], align 4 7561 // CHECK9-NEXT: [[CONV50:%.*]] = bitcast i64* [[CH_CASTED49]] to i32* 7562 // CHECK9-NEXT: store i32 [[TMP175]], i32* [[CONV50]], align 4 7563 // CHECK9-NEXT: [[TMP176:%.*]] = load i64, i64* [[CH_CASTED49]], align 8 7564 // CHECK9-NEXT: [[TMP177:%.*]] = load i32, i32* [[N]], align 4 7565 // CHECK9-NEXT: [[CONV52:%.*]] = bitcast i64* [[N_CASTED51]] to i32* 7566 // CHECK9-NEXT: store i32 [[TMP177]], i32* [[CONV52]], align 4 7567 // CHECK9-NEXT: [[TMP178:%.*]] = load i64, i64* [[N_CASTED51]], align 8 7568 // CHECK9-NEXT: [[TMP179:%.*]] = load i32*, i32** [[A]], align 8 7569 // CHECK9-NEXT: [[TMP180:%.*]] = load i32*, i32** [[B]], align 8 7570 // CHECK9-NEXT: [[TMP181:%.*]] = load i32*, i32** [[C]], align 8 7571 // CHECK9-NEXT: [[TMP182:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS53]], i32 0, i32 0 7572 // CHECK9-NEXT: [[TMP183:%.*]] = bitcast i8** [[TMP182]] to i64* 7573 // CHECK9-NEXT: store i64 [[TMP176]], i64* [[TMP183]], align 8 7574 // CHECK9-NEXT: [[TMP184:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS54]], i32 0, i32 0 7575 // CHECK9-NEXT: [[TMP185:%.*]] = bitcast i8** [[TMP184]] to i64* 7576 // CHECK9-NEXT: store i64 [[TMP176]], i64* [[TMP185]], align 8 7577 // CHECK9-NEXT: [[TMP186:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS55]], i64 0, i64 0 7578 // CHECK9-NEXT: store i8* null, i8** [[TMP186]], align 8 7579 // CHECK9-NEXT: [[TMP187:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS53]], i32 0, i32 1 7580 // CHECK9-NEXT: [[TMP188:%.*]] = bitcast i8** [[TMP187]] to i64* 7581 // CHECK9-NEXT: store i64 [[TMP178]], i64* [[TMP188]], align 8 7582 // CHECK9-NEXT: [[TMP189:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS54]], i32 0, i32 1 7583 // CHECK9-NEXT: [[TMP190:%.*]] = bitcast i8** [[TMP189]] to i64* 7584 // CHECK9-NEXT: store i64 [[TMP178]], i64* [[TMP190]], align 8 7585 // CHECK9-NEXT: [[TMP191:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS55]], i64 0, i64 1 7586 // CHECK9-NEXT: store i8* null, i8** [[TMP191]], align 8 7587 // CHECK9-NEXT: [[TMP192:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS53]], i32 0, i32 2 7588 // CHECK9-NEXT: [[TMP193:%.*]] = bitcast i8** [[TMP192]] to i32** 7589 // CHECK9-NEXT: store i32* [[TMP179]], i32** [[TMP193]], align 8 7590 // CHECK9-NEXT: [[TMP194:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS54]], i32 0, i32 2 7591 // CHECK9-NEXT: [[TMP195:%.*]] = bitcast i8** [[TMP194]] to i32** 7592 // CHECK9-NEXT: store i32* [[TMP179]], i32** [[TMP195]], align 8 7593 // CHECK9-NEXT: [[TMP196:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS55]], i64 0, i64 2 7594 // CHECK9-NEXT: store i8* null, i8** [[TMP196]], align 8 7595 // CHECK9-NEXT: [[TMP197:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS53]], i32 0, i32 3 7596 // CHECK9-NEXT: [[TMP198:%.*]] = bitcast i8** [[TMP197]] to i32** 7597 // CHECK9-NEXT: store i32* [[TMP180]], i32** [[TMP198]], align 8 7598 // CHECK9-NEXT: [[TMP199:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS54]], i32 0, i32 3 7599 // CHECK9-NEXT: [[TMP200:%.*]] = bitcast i8** [[TMP199]] to i32** 7600 // CHECK9-NEXT: store i32* [[TMP180]], i32** [[TMP200]], align 8 7601 // CHECK9-NEXT: [[TMP201:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS55]], i64 0, i64 3 7602 // CHECK9-NEXT: store i8* null, i8** [[TMP201]], align 8 7603 // CHECK9-NEXT: [[TMP202:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS53]], i32 0, i32 4 7604 // CHECK9-NEXT: [[TMP203:%.*]] = bitcast i8** [[TMP202]] to i32** 7605 // CHECK9-NEXT: store i32* [[TMP181]], i32** [[TMP203]], align 8 7606 // CHECK9-NEXT: [[TMP204:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS54]], i32 0, i32 4 7607 // CHECK9-NEXT: [[TMP205:%.*]] = bitcast i8** [[TMP204]] to i32** 7608 // CHECK9-NEXT: store i32* [[TMP181]], i32** [[TMP205]], align 8 7609 // CHECK9-NEXT: [[TMP206:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS55]], i64 0, i64 4 7610 // CHECK9-NEXT: store i8* null, i8** [[TMP206]], align 8 7611 // CHECK9-NEXT: [[TMP207:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS53]], i32 0, i32 0 7612 // CHECK9-NEXT: [[TMP208:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS54]], i32 0, i32 0 7613 // CHECK9-NEXT: [[TMP209:%.*]] = load i32, i32* [[N]], align 4 7614 // CHECK9-NEXT: store i32 [[TMP209]], i32* [[DOTCAPTURE_EXPR_57]], align 4 7615 // CHECK9-NEXT: [[TMP210:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_57]], align 4 7616 // CHECK9-NEXT: [[SUB59:%.*]] = sub nsw i32 [[TMP210]], 0 7617 // CHECK9-NEXT: [[DIV60:%.*]] = sdiv i32 [[SUB59]], 1 7618 // CHECK9-NEXT: [[SUB61:%.*]] = sub nsw i32 [[DIV60]], 1 7619 // CHECK9-NEXT: store i32 [[SUB61]], i32* [[DOTCAPTURE_EXPR_58]], align 4 7620 // CHECK9-NEXT: [[TMP211:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_58]], align 4 7621 // CHECK9-NEXT: [[ADD62:%.*]] = add nsw i32 [[TMP211]], 1 7622 // CHECK9-NEXT: [[TMP212:%.*]] = zext i32 [[ADD62]] to i64 7623 // CHECK9-NEXT: [[KERNEL_ARGS63:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS]], align 8 7624 // CHECK9-NEXT: [[TMP213:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS63]], i32 0, i32 0 7625 // CHECK9-NEXT: store i32 1, i32* [[TMP213]], align 4 7626 // CHECK9-NEXT: [[TMP214:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS63]], i32 0, i32 1 7627 // CHECK9-NEXT: store i32 5, i32* [[TMP214]], align 4 7628 // CHECK9-NEXT: [[TMP215:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS63]], i32 0, i32 2 7629 // CHECK9-NEXT: store i8** [[TMP207]], i8*** [[TMP215]], align 8 7630 // CHECK9-NEXT: [[TMP216:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS63]], i32 0, i32 3 7631 // CHECK9-NEXT: store i8** [[TMP208]], i8*** [[TMP216]], align 8 7632 // CHECK9-NEXT: [[TMP217:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS63]], i32 0, i32 4 7633 // CHECK9-NEXT: store i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_sizes.44, i32 0, i32 0), i64** [[TMP217]], align 8 7634 // CHECK9-NEXT: [[TMP218:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS63]], i32 0, i32 5 7635 // CHECK9-NEXT: store i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_maptypes.45, i32 0, i32 0), i64** [[TMP218]], align 8 7636 // CHECK9-NEXT: [[TMP219:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS63]], i32 0, i32 6 7637 // CHECK9-NEXT: store i8** null, i8*** [[TMP219]], align 8 7638 // CHECK9-NEXT: [[TMP220:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS63]], i32 0, i32 7 7639 // CHECK9-NEXT: store i8** null, i8*** [[TMP220]], align 8 7640 // CHECK9-NEXT: [[TMP221:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS63]], i32 0, i32 8 7641 // CHECK9-NEXT: store i64 [[TMP212]], i64* [[TMP221]], align 8 7642 // CHECK9-NEXT: [[TMP222:%.*]] = call i32 @__tgt_target_kernel(%struct.ident_t* @[[GLOB3]], i64 -1, i32 0, i32 1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiET_v_l74.region_id, %struct.__tgt_kernel_arguments* [[KERNEL_ARGS63]]) 7643 // CHECK9-NEXT: [[TMP223:%.*]] = icmp ne i32 [[TMP222]], 0 7644 // CHECK9-NEXT: br i1 [[TMP223]], label [[OMP_OFFLOAD_FAILED64:%.*]], label [[OMP_OFFLOAD_CONT65:%.*]] 7645 // CHECK9: omp_offload.failed64: 7646 // CHECK9-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiET_v_l74(i64 [[TMP176]], i64 [[TMP178]], i32* [[TMP179]], i32* [[TMP180]], i32* [[TMP181]]) #[[ATTR2]] 7647 // CHECK9-NEXT: br label [[OMP_OFFLOAD_CONT65]] 7648 // CHECK9: omp_offload.cont65: 7649 // CHECK9-NEXT: [[TMP224:%.*]] = load i32, i32* [[N]], align 4 7650 // CHECK9-NEXT: [[CONV67:%.*]] = bitcast i64* [[N_CASTED66]] to i32* 7651 // CHECK9-NEXT: store i32 [[TMP224]], i32* [[CONV67]], align 4 7652 // CHECK9-NEXT: [[TMP225:%.*]] = load i64, i64* [[N_CASTED66]], align 8 7653 // CHECK9-NEXT: [[TMP226:%.*]] = load i32*, i32** [[A]], align 8 7654 // CHECK9-NEXT: [[TMP227:%.*]] = load i32*, i32** [[B]], align 8 7655 // CHECK9-NEXT: [[TMP228:%.*]] = load i32*, i32** [[C]], align 8 7656 // CHECK9-NEXT: [[TMP229:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS68]], i32 0, i32 0 7657 // CHECK9-NEXT: [[TMP230:%.*]] = bitcast i8** [[TMP229]] to i64* 7658 // CHECK9-NEXT: store i64 [[TMP225]], i64* [[TMP230]], align 8 7659 // CHECK9-NEXT: [[TMP231:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS69]], i32 0, i32 0 7660 // CHECK9-NEXT: [[TMP232:%.*]] = bitcast i8** [[TMP231]] to i64* 7661 // CHECK9-NEXT: store i64 [[TMP225]], i64* [[TMP232]], align 8 7662 // CHECK9-NEXT: [[TMP233:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS70]], i64 0, i64 0 7663 // CHECK9-NEXT: store i8* null, i8** [[TMP233]], align 8 7664 // CHECK9-NEXT: [[TMP234:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS68]], i32 0, i32 1 7665 // CHECK9-NEXT: [[TMP235:%.*]] = bitcast i8** [[TMP234]] to i32** 7666 // CHECK9-NEXT: store i32* [[TMP226]], i32** [[TMP235]], align 8 7667 // CHECK9-NEXT: [[TMP236:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS69]], i32 0, i32 1 7668 // CHECK9-NEXT: [[TMP237:%.*]] = bitcast i8** [[TMP236]] to i32** 7669 // CHECK9-NEXT: store i32* [[TMP226]], i32** [[TMP237]], align 8 7670 // CHECK9-NEXT: [[TMP238:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS70]], i64 0, i64 1 7671 // CHECK9-NEXT: store i8* null, i8** [[TMP238]], align 8 7672 // CHECK9-NEXT: [[TMP239:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS68]], i32 0, i32 2 7673 // CHECK9-NEXT: [[TMP240:%.*]] = bitcast i8** [[TMP239]] to i32** 7674 // CHECK9-NEXT: store i32* [[TMP227]], i32** [[TMP240]], align 8 7675 // CHECK9-NEXT: [[TMP241:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS69]], i32 0, i32 2 7676 // CHECK9-NEXT: [[TMP242:%.*]] = bitcast i8** [[TMP241]] to i32** 7677 // CHECK9-NEXT: store i32* [[TMP227]], i32** [[TMP242]], align 8 7678 // CHECK9-NEXT: [[TMP243:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS70]], i64 0, i64 2 7679 // CHECK9-NEXT: store i8* null, i8** [[TMP243]], align 8 7680 // CHECK9-NEXT: [[TMP244:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS68]], i32 0, i32 3 7681 // CHECK9-NEXT: [[TMP245:%.*]] = bitcast i8** [[TMP244]] to i32** 7682 // CHECK9-NEXT: store i32* [[TMP228]], i32** [[TMP245]], align 8 7683 // CHECK9-NEXT: [[TMP246:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS69]], i32 0, i32 3 7684 // CHECK9-NEXT: [[TMP247:%.*]] = bitcast i8** [[TMP246]] to i32** 7685 // CHECK9-NEXT: store i32* [[TMP228]], i32** [[TMP247]], align 8 7686 // CHECK9-NEXT: [[TMP248:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS70]], i64 0, i64 3 7687 // CHECK9-NEXT: store i8* null, i8** [[TMP248]], align 8 7688 // CHECK9-NEXT: [[TMP249:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS68]], i32 0, i32 0 7689 // CHECK9-NEXT: [[TMP250:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS69]], i32 0, i32 0 7690 // CHECK9-NEXT: [[TMP251:%.*]] = load i32, i32* [[N]], align 4 7691 // CHECK9-NEXT: store i32 [[TMP251]], i32* [[DOTCAPTURE_EXPR_72]], align 4 7692 // CHECK9-NEXT: [[TMP252:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_72]], align 4 7693 // CHECK9-NEXT: [[SUB74:%.*]] = sub nsw i32 [[TMP252]], 0 7694 // CHECK9-NEXT: [[DIV75:%.*]] = sdiv i32 [[SUB74]], 1 7695 // CHECK9-NEXT: [[SUB76:%.*]] = sub nsw i32 [[DIV75]], 1 7696 // CHECK9-NEXT: store i32 [[SUB76]], i32* [[DOTCAPTURE_EXPR_73]], align 4 7697 // CHECK9-NEXT: [[TMP253:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_73]], align 4 7698 // CHECK9-NEXT: [[ADD77:%.*]] = add nsw i32 [[TMP253]], 1 7699 // CHECK9-NEXT: [[TMP254:%.*]] = zext i32 [[ADD77]] to i64 7700 // CHECK9-NEXT: [[KERNEL_ARGS78:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS]], align 8 7701 // CHECK9-NEXT: [[TMP255:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS78]], i32 0, i32 0 7702 // CHECK9-NEXT: store i32 1, i32* [[TMP255]], align 4 7703 // CHECK9-NEXT: [[TMP256:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS78]], i32 0, i32 1 7704 // CHECK9-NEXT: store i32 4, i32* [[TMP256]], align 4 7705 // CHECK9-NEXT: [[TMP257:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS78]], i32 0, i32 2 7706 // CHECK9-NEXT: store i8** [[TMP249]], i8*** [[TMP257]], align 8 7707 // CHECK9-NEXT: [[TMP258:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS78]], i32 0, i32 3 7708 // CHECK9-NEXT: store i8** [[TMP250]], i8*** [[TMP258]], align 8 7709 // CHECK9-NEXT: [[TMP259:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS78]], i32 0, i32 4 7710 // CHECK9-NEXT: store i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_sizes.48, i32 0, i32 0), i64** [[TMP259]], align 8 7711 // CHECK9-NEXT: [[TMP260:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS78]], i32 0, i32 5 7712 // CHECK9-NEXT: store i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_maptypes.49, i32 0, i32 0), i64** [[TMP260]], align 8 7713 // CHECK9-NEXT: [[TMP261:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS78]], i32 0, i32 6 7714 // CHECK9-NEXT: store i8** null, i8*** [[TMP261]], align 8 7715 // CHECK9-NEXT: [[TMP262:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS78]], i32 0, i32 7 7716 // CHECK9-NEXT: store i8** null, i8*** [[TMP262]], align 8 7717 // CHECK9-NEXT: [[TMP263:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS78]], i32 0, i32 8 7718 // CHECK9-NEXT: store i64 [[TMP254]], i64* [[TMP263]], align 8 7719 // CHECK9-NEXT: [[TMP264:%.*]] = call i32 @__tgt_target_kernel(%struct.ident_t* @[[GLOB3]], i64 -1, i32 0, i32 1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiET_v_l82.region_id, %struct.__tgt_kernel_arguments* [[KERNEL_ARGS78]]) 7720 // CHECK9-NEXT: [[TMP265:%.*]] = icmp ne i32 [[TMP264]], 0 7721 // CHECK9-NEXT: br i1 [[TMP265]], label [[OMP_OFFLOAD_FAILED79:%.*]], label [[OMP_OFFLOAD_CONT80:%.*]] 7722 // CHECK9: omp_offload.failed79: 7723 // CHECK9-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiET_v_l82(i64 [[TMP225]], i32* [[TMP226]], i32* [[TMP227]], i32* [[TMP228]]) #[[ATTR2]] 7724 // CHECK9-NEXT: br label [[OMP_OFFLOAD_CONT80]] 7725 // CHECK9: omp_offload.cont80: 7726 // CHECK9-NEXT: [[TMP266:%.*]] = load i32, i32* [[CH]], align 4 7727 // CHECK9-NEXT: [[CONV82:%.*]] = bitcast i64* [[CH_CASTED81]] to i32* 7728 // CHECK9-NEXT: store i32 [[TMP266]], i32* [[CONV82]], align 4 7729 // CHECK9-NEXT: [[TMP267:%.*]] = load i64, i64* [[CH_CASTED81]], align 8 7730 // CHECK9-NEXT: [[TMP268:%.*]] = load i32, i32* [[N]], align 4 7731 // CHECK9-NEXT: [[CONV84:%.*]] = bitcast i64* [[N_CASTED83]] to i32* 7732 // CHECK9-NEXT: store i32 [[TMP268]], i32* [[CONV84]], align 4 7733 // CHECK9-NEXT: [[TMP269:%.*]] = load i64, i64* [[N_CASTED83]], align 8 7734 // CHECK9-NEXT: [[TMP270:%.*]] = load i32*, i32** [[A]], align 8 7735 // CHECK9-NEXT: [[TMP271:%.*]] = load i32*, i32** [[B]], align 8 7736 // CHECK9-NEXT: [[TMP272:%.*]] = load i32*, i32** [[C]], align 8 7737 // CHECK9-NEXT: [[TMP273:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS85]], i32 0, i32 0 7738 // CHECK9-NEXT: [[TMP274:%.*]] = bitcast i8** [[TMP273]] to i64* 7739 // CHECK9-NEXT: store i64 [[TMP267]], i64* [[TMP274]], align 8 7740 // CHECK9-NEXT: [[TMP275:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS86]], i32 0, i32 0 7741 // CHECK9-NEXT: [[TMP276:%.*]] = bitcast i8** [[TMP275]] to i64* 7742 // CHECK9-NEXT: store i64 [[TMP267]], i64* [[TMP276]], align 8 7743 // CHECK9-NEXT: [[TMP277:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS87]], i64 0, i64 0 7744 // CHECK9-NEXT: store i8* null, i8** [[TMP277]], align 8 7745 // CHECK9-NEXT: [[TMP278:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS85]], i32 0, i32 1 7746 // CHECK9-NEXT: [[TMP279:%.*]] = bitcast i8** [[TMP278]] to i64* 7747 // CHECK9-NEXT: store i64 [[TMP269]], i64* [[TMP279]], align 8 7748 // CHECK9-NEXT: [[TMP280:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS86]], i32 0, i32 1 7749 // CHECK9-NEXT: [[TMP281:%.*]] = bitcast i8** [[TMP280]] to i64* 7750 // CHECK9-NEXT: store i64 [[TMP269]], i64* [[TMP281]], align 8 7751 // CHECK9-NEXT: [[TMP282:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS87]], i64 0, i64 1 7752 // CHECK9-NEXT: store i8* null, i8** [[TMP282]], align 8 7753 // CHECK9-NEXT: [[TMP283:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS85]], i32 0, i32 2 7754 // CHECK9-NEXT: [[TMP284:%.*]] = bitcast i8** [[TMP283]] to i32** 7755 // CHECK9-NEXT: store i32* [[TMP270]], i32** [[TMP284]], align 8 7756 // CHECK9-NEXT: [[TMP285:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS86]], i32 0, i32 2 7757 // CHECK9-NEXT: [[TMP286:%.*]] = bitcast i8** [[TMP285]] to i32** 7758 // CHECK9-NEXT: store i32* [[TMP270]], i32** [[TMP286]], align 8 7759 // CHECK9-NEXT: [[TMP287:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS87]], i64 0, i64 2 7760 // CHECK9-NEXT: store i8* null, i8** [[TMP287]], align 8 7761 // CHECK9-NEXT: [[TMP288:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS85]], i32 0, i32 3 7762 // CHECK9-NEXT: [[TMP289:%.*]] = bitcast i8** [[TMP288]] to i32** 7763 // CHECK9-NEXT: store i32* [[TMP271]], i32** [[TMP289]], align 8 7764 // CHECK9-NEXT: [[TMP290:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS86]], i32 0, i32 3 7765 // CHECK9-NEXT: [[TMP291:%.*]] = bitcast i8** [[TMP290]] to i32** 7766 // CHECK9-NEXT: store i32* [[TMP271]], i32** [[TMP291]], align 8 7767 // CHECK9-NEXT: [[TMP292:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS87]], i64 0, i64 3 7768 // CHECK9-NEXT: store i8* null, i8** [[TMP292]], align 8 7769 // CHECK9-NEXT: [[TMP293:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS85]], i32 0, i32 4 7770 // CHECK9-NEXT: [[TMP294:%.*]] = bitcast i8** [[TMP293]] to i32** 7771 // CHECK9-NEXT: store i32* [[TMP272]], i32** [[TMP294]], align 8 7772 // CHECK9-NEXT: [[TMP295:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS86]], i32 0, i32 4 7773 // CHECK9-NEXT: [[TMP296:%.*]] = bitcast i8** [[TMP295]] to i32** 7774 // CHECK9-NEXT: store i32* [[TMP272]], i32** [[TMP296]], align 8 7775 // CHECK9-NEXT: [[TMP297:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS87]], i64 0, i64 4 7776 // CHECK9-NEXT: store i8* null, i8** [[TMP297]], align 8 7777 // CHECK9-NEXT: [[TMP298:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS85]], i32 0, i32 0 7778 // CHECK9-NEXT: [[TMP299:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS86]], i32 0, i32 0 7779 // CHECK9-NEXT: [[TMP300:%.*]] = load i32, i32* [[N]], align 4 7780 // CHECK9-NEXT: store i32 [[TMP300]], i32* [[DOTCAPTURE_EXPR_89]], align 4 7781 // CHECK9-NEXT: [[TMP301:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_89]], align 4 7782 // CHECK9-NEXT: [[SUB91:%.*]] = sub nsw i32 [[TMP301]], 0 7783 // CHECK9-NEXT: [[DIV92:%.*]] = sdiv i32 [[SUB91]], 1 7784 // CHECK9-NEXT: [[SUB93:%.*]] = sub nsw i32 [[DIV92]], 1 7785 // CHECK9-NEXT: store i32 [[SUB93]], i32* [[DOTCAPTURE_EXPR_90]], align 4 7786 // CHECK9-NEXT: [[TMP302:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_90]], align 4 7787 // CHECK9-NEXT: [[ADD94:%.*]] = add nsw i32 [[TMP302]], 1 7788 // CHECK9-NEXT: [[TMP303:%.*]] = zext i32 [[ADD94]] to i64 7789 // CHECK9-NEXT: [[KERNEL_ARGS95:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS]], align 8 7790 // CHECK9-NEXT: [[TMP304:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS95]], i32 0, i32 0 7791 // CHECK9-NEXT: store i32 1, i32* [[TMP304]], align 4 7792 // CHECK9-NEXT: [[TMP305:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS95]], i32 0, i32 1 7793 // CHECK9-NEXT: store i32 5, i32* [[TMP305]], align 4 7794 // CHECK9-NEXT: [[TMP306:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS95]], i32 0, i32 2 7795 // CHECK9-NEXT: store i8** [[TMP298]], i8*** [[TMP306]], align 8 7796 // CHECK9-NEXT: [[TMP307:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS95]], i32 0, i32 3 7797 // CHECK9-NEXT: store i8** [[TMP299]], i8*** [[TMP307]], align 8 7798 // CHECK9-NEXT: [[TMP308:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS95]], i32 0, i32 4 7799 // CHECK9-NEXT: store i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_sizes.52, i32 0, i32 0), i64** [[TMP308]], align 8 7800 // CHECK9-NEXT: [[TMP309:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS95]], i32 0, i32 5 7801 // CHECK9-NEXT: store i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_maptypes.53, i32 0, i32 0), i64** [[TMP309]], align 8 7802 // CHECK9-NEXT: [[TMP310:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS95]], i32 0, i32 6 7803 // CHECK9-NEXT: store i8** null, i8*** [[TMP310]], align 8 7804 // CHECK9-NEXT: [[TMP311:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS95]], i32 0, i32 7 7805 // CHECK9-NEXT: store i8** null, i8*** [[TMP311]], align 8 7806 // CHECK9-NEXT: [[TMP312:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS95]], i32 0, i32 8 7807 // CHECK9-NEXT: store i64 [[TMP303]], i64* [[TMP312]], align 8 7808 // CHECK9-NEXT: [[TMP313:%.*]] = call i32 @__tgt_target_kernel(%struct.ident_t* @[[GLOB3]], i64 -1, i32 0, i32 1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiET_v_l90.region_id, %struct.__tgt_kernel_arguments* [[KERNEL_ARGS95]]) 7809 // CHECK9-NEXT: [[TMP314:%.*]] = icmp ne i32 [[TMP313]], 0 7810 // CHECK9-NEXT: br i1 [[TMP314]], label [[OMP_OFFLOAD_FAILED96:%.*]], label [[OMP_OFFLOAD_CONT97:%.*]] 7811 // CHECK9: omp_offload.failed96: 7812 // CHECK9-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiET_v_l90(i64 [[TMP267]], i64 [[TMP269]], i32* [[TMP270]], i32* [[TMP271]], i32* [[TMP272]]) #[[ATTR2]] 7813 // CHECK9-NEXT: br label [[OMP_OFFLOAD_CONT97]] 7814 // CHECK9: omp_offload.cont97: 7815 // CHECK9-NEXT: ret i32 0 7816 // 7817 // 7818 // CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiET_v_l42 7819 // CHECK9-SAME: (i64 noundef [[N:%.*]], i32* noundef [[A:%.*]], i32* noundef [[B:%.*]], i32* noundef [[C:%.*]]) #[[ATTR1]] { 7820 // CHECK9-NEXT: entry: 7821 // CHECK9-NEXT: [[N_ADDR:%.*]] = alloca i64, align 8 7822 // CHECK9-NEXT: [[A_ADDR:%.*]] = alloca i32*, align 8 7823 // CHECK9-NEXT: [[B_ADDR:%.*]] = alloca i32*, align 8 7824 // CHECK9-NEXT: [[C_ADDR:%.*]] = alloca i32*, align 8 7825 // CHECK9-NEXT: store i64 [[N]], i64* [[N_ADDR]], align 8 7826 // CHECK9-NEXT: store i32* [[A]], i32** [[A_ADDR]], align 8 7827 // CHECK9-NEXT: store i32* [[B]], i32** [[B_ADDR]], align 8 7828 // CHECK9-NEXT: store i32* [[C]], i32** [[C_ADDR]], align 8 7829 // CHECK9-NEXT: [[CONV:%.*]] = bitcast i64* [[N_ADDR]] to i32* 7830 // CHECK9-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 4, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32*, i32**, i32**, i32**)* @.omp_outlined..26 to void (i32*, i32*, ...)*), i32* [[CONV]], i32** [[A_ADDR]], i32** [[B_ADDR]], i32** [[C_ADDR]]) 7831 // CHECK9-NEXT: ret void 7832 // 7833 // 7834 // CHECK9-LABEL: define {{[^@]+}}@.omp_outlined..26 7835 // CHECK9-SAME: (i32* noalias noundef [[DOTGLOBAL_TID_:%.*]], i32* noalias noundef [[DOTBOUND_TID_:%.*]], i32* noundef nonnull align 4 dereferenceable(4) [[N:%.*]], i32** noundef nonnull align 8 dereferenceable(8) [[A:%.*]], i32** noundef nonnull align 8 dereferenceable(8) [[B:%.*]], i32** noundef nonnull align 8 dereferenceable(8) [[C:%.*]]) #[[ATTR1]] { 7836 // CHECK9-NEXT: entry: 7837 // CHECK9-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 7838 // CHECK9-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 7839 // CHECK9-NEXT: [[N_ADDR:%.*]] = alloca i32*, align 8 7840 // CHECK9-NEXT: [[A_ADDR:%.*]] = alloca i32**, align 8 7841 // CHECK9-NEXT: [[B_ADDR:%.*]] = alloca i32**, align 8 7842 // CHECK9-NEXT: [[C_ADDR:%.*]] = alloca i32**, align 8 7843 // CHECK9-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 7844 // CHECK9-NEXT: [[TMP:%.*]] = alloca i32, align 4 7845 // CHECK9-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 7846 // CHECK9-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 7847 // CHECK9-NEXT: [[I:%.*]] = alloca i32, align 4 7848 // CHECK9-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4 7849 // CHECK9-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4 7850 // CHECK9-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 7851 // CHECK9-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 7852 // CHECK9-NEXT: [[I3:%.*]] = alloca i32, align 4 7853 // CHECK9-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8 7854 // CHECK9-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8 7855 // CHECK9-NEXT: store i32* [[N]], i32** [[N_ADDR]], align 8 7856 // CHECK9-NEXT: store i32** [[A]], i32*** [[A_ADDR]], align 8 7857 // CHECK9-NEXT: store i32** [[B]], i32*** [[B_ADDR]], align 8 7858 // CHECK9-NEXT: store i32** [[C]], i32*** [[C_ADDR]], align 8 7859 // CHECK9-NEXT: [[TMP0:%.*]] = load i32*, i32** [[N_ADDR]], align 8 7860 // CHECK9-NEXT: [[TMP1:%.*]] = load i32**, i32*** [[A_ADDR]], align 8 7861 // CHECK9-NEXT: [[TMP2:%.*]] = load i32**, i32*** [[B_ADDR]], align 8 7862 // CHECK9-NEXT: [[TMP3:%.*]] = load i32**, i32*** [[C_ADDR]], align 8 7863 // CHECK9-NEXT: [[TMP4:%.*]] = load i32, i32* [[TMP0]], align 4 7864 // CHECK9-NEXT: store i32 [[TMP4]], i32* [[DOTCAPTURE_EXPR_]], align 4 7865 // CHECK9-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 7866 // CHECK9-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP5]], 0 7867 // CHECK9-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 7868 // CHECK9-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1 7869 // CHECK9-NEXT: store i32 [[SUB2]], i32* [[DOTCAPTURE_EXPR_1]], align 4 7870 // CHECK9-NEXT: store i32 0, i32* [[I]], align 4 7871 // CHECK9-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 7872 // CHECK9-NEXT: [[CMP:%.*]] = icmp slt i32 0, [[TMP6]] 7873 // CHECK9-NEXT: br i1 [[CMP]], label [[OMP_PRECOND_THEN:%.*]], label [[OMP_PRECOND_END:%.*]] 7874 // CHECK9: omp.precond.then: 7875 // CHECK9-NEXT: store i32 0, i32* [[DOTOMP_COMB_LB]], align 4 7876 // CHECK9-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 7877 // CHECK9-NEXT: store i32 [[TMP7]], i32* [[DOTOMP_COMB_UB]], align 4 7878 // CHECK9-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4 7879 // CHECK9-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4 7880 // CHECK9-NEXT: [[TMP8:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 7881 // CHECK9-NEXT: [[TMP9:%.*]] = load i32, i32* [[TMP8]], align 4 7882 // CHECK9-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1]], i32 [[TMP9]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_COMB_LB]], i32* [[DOTOMP_COMB_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1) 7883 // CHECK9-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 7884 // CHECK9-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 7885 // CHECK9-NEXT: [[CMP4:%.*]] = icmp sgt i32 [[TMP10]], [[TMP11]] 7886 // CHECK9-NEXT: br i1 [[CMP4]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] 7887 // CHECK9: cond.true: 7888 // CHECK9-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 7889 // CHECK9-NEXT: br label [[COND_END:%.*]] 7890 // CHECK9: cond.false: 7891 // CHECK9-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 7892 // CHECK9-NEXT: br label [[COND_END]] 7893 // CHECK9: cond.end: 7894 // CHECK9-NEXT: [[COND:%.*]] = phi i32 [ [[TMP12]], [[COND_TRUE]] ], [ [[TMP13]], [[COND_FALSE]] ] 7895 // CHECK9-NEXT: store i32 [[COND]], i32* [[DOTOMP_COMB_UB]], align 4 7896 // CHECK9-NEXT: [[TMP14:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4 7897 // CHECK9-NEXT: store i32 [[TMP14]], i32* [[DOTOMP_IV]], align 4 7898 // CHECK9-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] 7899 // CHECK9: omp.inner.for.cond: 7900 // CHECK9-NEXT: [[TMP15:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !62 7901 // CHECK9-NEXT: [[TMP16:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4, !llvm.access.group !62 7902 // CHECK9-NEXT: [[CMP5:%.*]] = icmp sle i32 [[TMP15]], [[TMP16]] 7903 // CHECK9-NEXT: br i1 [[CMP5]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] 7904 // CHECK9: omp.inner.for.body: 7905 // CHECK9-NEXT: [[TMP17:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4, !llvm.access.group !62 7906 // CHECK9-NEXT: [[TMP18:%.*]] = zext i32 [[TMP17]] to i64 7907 // CHECK9-NEXT: [[TMP19:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4, !llvm.access.group !62 7908 // CHECK9-NEXT: [[TMP20:%.*]] = zext i32 [[TMP19]] to i64 7909 // CHECK9-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 6, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64, i32*, i32**, i32**, i32**)* @.omp_outlined..27 to void (i32*, i32*, ...)*), i64 [[TMP18]], i64 [[TMP20]], i32* [[TMP0]], i32** [[TMP1]], i32** [[TMP2]], i32** [[TMP3]]), !llvm.access.group !62 7910 // CHECK9-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] 7911 // CHECK9: omp.inner.for.inc: 7912 // CHECK9-NEXT: [[TMP21:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !62 7913 // CHECK9-NEXT: [[TMP22:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4, !llvm.access.group !62 7914 // CHECK9-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP21]], [[TMP22]] 7915 // CHECK9-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4, !llvm.access.group !62 7916 // CHECK9-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP63:![0-9]+]] 7917 // CHECK9: omp.inner.for.end: 7918 // CHECK9-NEXT: br label [[OMP_LOOP_EXIT:%.*]] 7919 // CHECK9: omp.loop.exit: 7920 // CHECK9-NEXT: [[TMP23:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 7921 // CHECK9-NEXT: [[TMP24:%.*]] = load i32, i32* [[TMP23]], align 4 7922 // CHECK9-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP24]]) 7923 // CHECK9-NEXT: [[TMP25:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4 7924 // CHECK9-NEXT: [[TMP26:%.*]] = icmp ne i32 [[TMP25]], 0 7925 // CHECK9-NEXT: br i1 [[TMP26]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]] 7926 // CHECK9: .omp.final.then: 7927 // CHECK9-NEXT: [[TMP27:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 7928 // CHECK9-NEXT: [[SUB6:%.*]] = sub nsw i32 [[TMP27]], 0 7929 // CHECK9-NEXT: [[DIV7:%.*]] = sdiv i32 [[SUB6]], 1 7930 // CHECK9-NEXT: [[MUL:%.*]] = mul nsw i32 [[DIV7]], 1 7931 // CHECK9-NEXT: [[ADD8:%.*]] = add nsw i32 0, [[MUL]] 7932 // CHECK9-NEXT: store i32 [[ADD8]], i32* [[I3]], align 4 7933 // CHECK9-NEXT: br label [[DOTOMP_FINAL_DONE]] 7934 // CHECK9: .omp.final.done: 7935 // CHECK9-NEXT: br label [[OMP_PRECOND_END]] 7936 // CHECK9: omp.precond.end: 7937 // CHECK9-NEXT: ret void 7938 // 7939 // 7940 // CHECK9-LABEL: define {{[^@]+}}@.omp_outlined..27 7941 // CHECK9-SAME: (i32* noalias noundef [[DOTGLOBAL_TID_:%.*]], i32* noalias noundef [[DOTBOUND_TID_:%.*]], i64 noundef [[DOTPREVIOUS_LB_:%.*]], i64 noundef [[DOTPREVIOUS_UB_:%.*]], i32* noundef nonnull align 4 dereferenceable(4) [[N:%.*]], i32** noundef nonnull align 8 dereferenceable(8) [[A:%.*]], i32** noundef nonnull align 8 dereferenceable(8) [[B:%.*]], i32** noundef nonnull align 8 dereferenceable(8) [[C:%.*]]) #[[ATTR1]] { 7942 // CHECK9-NEXT: entry: 7943 // CHECK9-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 7944 // CHECK9-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 7945 // CHECK9-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8 7946 // CHECK9-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8 7947 // CHECK9-NEXT: [[N_ADDR:%.*]] = alloca i32*, align 8 7948 // CHECK9-NEXT: [[A_ADDR:%.*]] = alloca i32**, align 8 7949 // CHECK9-NEXT: [[B_ADDR:%.*]] = alloca i32**, align 8 7950 // CHECK9-NEXT: [[C_ADDR:%.*]] = alloca i32**, align 8 7951 // CHECK9-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 7952 // CHECK9-NEXT: [[TMP:%.*]] = alloca i32, align 4 7953 // CHECK9-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 7954 // CHECK9-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 7955 // CHECK9-NEXT: [[I:%.*]] = alloca i32, align 4 7956 // CHECK9-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 7957 // CHECK9-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4 7958 // CHECK9-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 7959 // CHECK9-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 7960 // CHECK9-NEXT: [[I4:%.*]] = alloca i32, align 4 7961 // CHECK9-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8 7962 // CHECK9-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8 7963 // CHECK9-NEXT: store i64 [[DOTPREVIOUS_LB_]], i64* [[DOTPREVIOUS_LB__ADDR]], align 8 7964 // CHECK9-NEXT: store i64 [[DOTPREVIOUS_UB_]], i64* [[DOTPREVIOUS_UB__ADDR]], align 8 7965 // CHECK9-NEXT: store i32* [[N]], i32** [[N_ADDR]], align 8 7966 // CHECK9-NEXT: store i32** [[A]], i32*** [[A_ADDR]], align 8 7967 // CHECK9-NEXT: store i32** [[B]], i32*** [[B_ADDR]], align 8 7968 // CHECK9-NEXT: store i32** [[C]], i32*** [[C_ADDR]], align 8 7969 // CHECK9-NEXT: [[TMP0:%.*]] = load i32*, i32** [[N_ADDR]], align 8 7970 // CHECK9-NEXT: [[TMP1:%.*]] = load i32**, i32*** [[A_ADDR]], align 8 7971 // CHECK9-NEXT: [[TMP2:%.*]] = load i32**, i32*** [[B_ADDR]], align 8 7972 // CHECK9-NEXT: [[TMP3:%.*]] = load i32**, i32*** [[C_ADDR]], align 8 7973 // CHECK9-NEXT: [[TMP4:%.*]] = load i32, i32* [[TMP0]], align 4 7974 // CHECK9-NEXT: store i32 [[TMP4]], i32* [[DOTCAPTURE_EXPR_]], align 4 7975 // CHECK9-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 7976 // CHECK9-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP5]], 0 7977 // CHECK9-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 7978 // CHECK9-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1 7979 // CHECK9-NEXT: store i32 [[SUB2]], i32* [[DOTCAPTURE_EXPR_1]], align 4 7980 // CHECK9-NEXT: store i32 0, i32* [[I]], align 4 7981 // CHECK9-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 7982 // CHECK9-NEXT: [[CMP:%.*]] = icmp slt i32 0, [[TMP6]] 7983 // CHECK9-NEXT: br i1 [[CMP]], label [[OMP_PRECOND_THEN:%.*]], label [[OMP_PRECOND_END:%.*]] 7984 // CHECK9: omp.precond.then: 7985 // CHECK9-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4 7986 // CHECK9-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 7987 // CHECK9-NEXT: store i32 [[TMP7]], i32* [[DOTOMP_UB]], align 4 7988 // CHECK9-NEXT: [[TMP8:%.*]] = load i64, i64* [[DOTPREVIOUS_LB__ADDR]], align 8 7989 // CHECK9-NEXT: [[CONV:%.*]] = trunc i64 [[TMP8]] to i32 7990 // CHECK9-NEXT: [[TMP9:%.*]] = load i64, i64* [[DOTPREVIOUS_UB__ADDR]], align 8 7991 // CHECK9-NEXT: [[CONV3:%.*]] = trunc i64 [[TMP9]] to i32 7992 // CHECK9-NEXT: store i32 [[CONV]], i32* [[DOTOMP_LB]], align 4 7993 // CHECK9-NEXT: store i32 [[CONV3]], i32* [[DOTOMP_UB]], align 4 7994 // CHECK9-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4 7995 // CHECK9-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4 7996 // CHECK9-NEXT: [[TMP10:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 7997 // CHECK9-NEXT: [[TMP11:%.*]] = load i32, i32* [[TMP10]], align 4 7998 // CHECK9-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB2]], i32 [[TMP11]], i32 34, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1) 7999 // CHECK9-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 8000 // CHECK9-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 8001 // CHECK9-NEXT: [[CMP5:%.*]] = icmp sgt i32 [[TMP12]], [[TMP13]] 8002 // CHECK9-NEXT: br i1 [[CMP5]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] 8003 // CHECK9: cond.true: 8004 // CHECK9-NEXT: [[TMP14:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 8005 // CHECK9-NEXT: br label [[COND_END:%.*]] 8006 // CHECK9: cond.false: 8007 // CHECK9-NEXT: [[TMP15:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 8008 // CHECK9-NEXT: br label [[COND_END]] 8009 // CHECK9: cond.end: 8010 // CHECK9-NEXT: [[COND:%.*]] = phi i32 [ [[TMP14]], [[COND_TRUE]] ], [ [[TMP15]], [[COND_FALSE]] ] 8011 // CHECK9-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4 8012 // CHECK9-NEXT: [[TMP16:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4 8013 // CHECK9-NEXT: store i32 [[TMP16]], i32* [[DOTOMP_IV]], align 4 8014 // CHECK9-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] 8015 // CHECK9: omp.inner.for.cond: 8016 // CHECK9-NEXT: [[TMP17:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !65 8017 // CHECK9-NEXT: [[TMP18:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4, !llvm.access.group !65 8018 // CHECK9-NEXT: [[CMP6:%.*]] = icmp sle i32 [[TMP17]], [[TMP18]] 8019 // CHECK9-NEXT: br i1 [[CMP6]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] 8020 // CHECK9: omp.inner.for.body: 8021 // CHECK9-NEXT: [[TMP19:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !65 8022 // CHECK9-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP19]], 1 8023 // CHECK9-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] 8024 // CHECK9-NEXT: store i32 [[ADD]], i32* [[I4]], align 4, !llvm.access.group !65 8025 // CHECK9-NEXT: [[TMP20:%.*]] = load i32*, i32** [[TMP2]], align 8, !llvm.access.group !65 8026 // CHECK9-NEXT: [[TMP21:%.*]] = load i32, i32* [[I4]], align 4, !llvm.access.group !65 8027 // CHECK9-NEXT: [[IDXPROM:%.*]] = sext i32 [[TMP21]] to i64 8028 // CHECK9-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i32, i32* [[TMP20]], i64 [[IDXPROM]] 8029 // CHECK9-NEXT: [[TMP22:%.*]] = load i32, i32* [[ARRAYIDX]], align 4, !llvm.access.group !65 8030 // CHECK9-NEXT: [[TMP23:%.*]] = load i32*, i32** [[TMP3]], align 8, !llvm.access.group !65 8031 // CHECK9-NEXT: [[TMP24:%.*]] = load i32, i32* [[I4]], align 4, !llvm.access.group !65 8032 // CHECK9-NEXT: [[IDXPROM7:%.*]] = sext i32 [[TMP24]] to i64 8033 // CHECK9-NEXT: [[ARRAYIDX8:%.*]] = getelementptr inbounds i32, i32* [[TMP23]], i64 [[IDXPROM7]] 8034 // CHECK9-NEXT: [[TMP25:%.*]] = load i32, i32* [[ARRAYIDX8]], align 4, !llvm.access.group !65 8035 // CHECK9-NEXT: [[ADD9:%.*]] = add nsw i32 [[TMP22]], [[TMP25]] 8036 // CHECK9-NEXT: [[TMP26:%.*]] = load i32*, i32** [[TMP1]], align 8, !llvm.access.group !65 8037 // CHECK9-NEXT: [[TMP27:%.*]] = load i32, i32* [[I4]], align 4, !llvm.access.group !65 8038 // CHECK9-NEXT: [[IDXPROM10:%.*]] = sext i32 [[TMP27]] to i64 8039 // CHECK9-NEXT: [[ARRAYIDX11:%.*]] = getelementptr inbounds i32, i32* [[TMP26]], i64 [[IDXPROM10]] 8040 // CHECK9-NEXT: store i32 [[ADD9]], i32* [[ARRAYIDX11]], align 4, !llvm.access.group !65 8041 // CHECK9-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] 8042 // CHECK9: omp.body.continue: 8043 // CHECK9-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] 8044 // CHECK9: omp.inner.for.inc: 8045 // CHECK9-NEXT: [[TMP28:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !65 8046 // CHECK9-NEXT: [[ADD12:%.*]] = add nsw i32 [[TMP28]], 1 8047 // CHECK9-NEXT: store i32 [[ADD12]], i32* [[DOTOMP_IV]], align 4, !llvm.access.group !65 8048 // CHECK9-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP66:![0-9]+]] 8049 // CHECK9: omp.inner.for.end: 8050 // CHECK9-NEXT: br label [[OMP_LOOP_EXIT:%.*]] 8051 // CHECK9: omp.loop.exit: 8052 // CHECK9-NEXT: [[TMP29:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 8053 // CHECK9-NEXT: [[TMP30:%.*]] = load i32, i32* [[TMP29]], align 4 8054 // CHECK9-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP30]]) 8055 // CHECK9-NEXT: [[TMP31:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4 8056 // CHECK9-NEXT: [[TMP32:%.*]] = icmp ne i32 [[TMP31]], 0 8057 // CHECK9-NEXT: br i1 [[TMP32]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]] 8058 // CHECK9: .omp.final.then: 8059 // CHECK9-NEXT: [[TMP33:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 8060 // CHECK9-NEXT: [[SUB13:%.*]] = sub nsw i32 [[TMP33]], 0 8061 // CHECK9-NEXT: [[DIV14:%.*]] = sdiv i32 [[SUB13]], 1 8062 // CHECK9-NEXT: [[MUL15:%.*]] = mul nsw i32 [[DIV14]], 1 8063 // CHECK9-NEXT: [[ADD16:%.*]] = add nsw i32 0, [[MUL15]] 8064 // CHECK9-NEXT: store i32 [[ADD16]], i32* [[I4]], align 4 8065 // CHECK9-NEXT: br label [[DOTOMP_FINAL_DONE]] 8066 // CHECK9: .omp.final.done: 8067 // CHECK9-NEXT: br label [[OMP_PRECOND_END]] 8068 // CHECK9: omp.precond.end: 8069 // CHECK9-NEXT: ret void 8070 // 8071 // 8072 // CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiET_v_l50 8073 // CHECK9-SAME: (i64 noundef [[N:%.*]], i32* noundef [[A:%.*]], i32* noundef [[B:%.*]], i32* noundef [[C:%.*]]) #[[ATTR1]] { 8074 // CHECK9-NEXT: entry: 8075 // CHECK9-NEXT: [[N_ADDR:%.*]] = alloca i64, align 8 8076 // CHECK9-NEXT: [[A_ADDR:%.*]] = alloca i32*, align 8 8077 // CHECK9-NEXT: [[B_ADDR:%.*]] = alloca i32*, align 8 8078 // CHECK9-NEXT: [[C_ADDR:%.*]] = alloca i32*, align 8 8079 // CHECK9-NEXT: store i64 [[N]], i64* [[N_ADDR]], align 8 8080 // CHECK9-NEXT: store i32* [[A]], i32** [[A_ADDR]], align 8 8081 // CHECK9-NEXT: store i32* [[B]], i32** [[B_ADDR]], align 8 8082 // CHECK9-NEXT: store i32* [[C]], i32** [[C_ADDR]], align 8 8083 // CHECK9-NEXT: [[CONV:%.*]] = bitcast i64* [[N_ADDR]] to i32* 8084 // CHECK9-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 4, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32*, i32**, i32**, i32**)* @.omp_outlined..30 to void (i32*, i32*, ...)*), i32* [[CONV]], i32** [[A_ADDR]], i32** [[B_ADDR]], i32** [[C_ADDR]]) 8085 // CHECK9-NEXT: ret void 8086 // 8087 // 8088 // CHECK9-LABEL: define {{[^@]+}}@.omp_outlined..30 8089 // CHECK9-SAME: (i32* noalias noundef [[DOTGLOBAL_TID_:%.*]], i32* noalias noundef [[DOTBOUND_TID_:%.*]], i32* noundef nonnull align 4 dereferenceable(4) [[N:%.*]], i32** noundef nonnull align 8 dereferenceable(8) [[A:%.*]], i32** noundef nonnull align 8 dereferenceable(8) [[B:%.*]], i32** noundef nonnull align 8 dereferenceable(8) [[C:%.*]]) #[[ATTR1]] { 8090 // CHECK9-NEXT: entry: 8091 // CHECK9-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 8092 // CHECK9-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 8093 // CHECK9-NEXT: [[N_ADDR:%.*]] = alloca i32*, align 8 8094 // CHECK9-NEXT: [[A_ADDR:%.*]] = alloca i32**, align 8 8095 // CHECK9-NEXT: [[B_ADDR:%.*]] = alloca i32**, align 8 8096 // CHECK9-NEXT: [[C_ADDR:%.*]] = alloca i32**, align 8 8097 // CHECK9-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 8098 // CHECK9-NEXT: [[TMP:%.*]] = alloca i32, align 4 8099 // CHECK9-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 8100 // CHECK9-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 8101 // CHECK9-NEXT: [[I:%.*]] = alloca i32, align 4 8102 // CHECK9-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4 8103 // CHECK9-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4 8104 // CHECK9-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 8105 // CHECK9-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 8106 // CHECK9-NEXT: [[I3:%.*]] = alloca i32, align 4 8107 // CHECK9-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8 8108 // CHECK9-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8 8109 // CHECK9-NEXT: store i32* [[N]], i32** [[N_ADDR]], align 8 8110 // CHECK9-NEXT: store i32** [[A]], i32*** [[A_ADDR]], align 8 8111 // CHECK9-NEXT: store i32** [[B]], i32*** [[B_ADDR]], align 8 8112 // CHECK9-NEXT: store i32** [[C]], i32*** [[C_ADDR]], align 8 8113 // CHECK9-NEXT: [[TMP0:%.*]] = load i32*, i32** [[N_ADDR]], align 8 8114 // CHECK9-NEXT: [[TMP1:%.*]] = load i32**, i32*** [[A_ADDR]], align 8 8115 // CHECK9-NEXT: [[TMP2:%.*]] = load i32**, i32*** [[B_ADDR]], align 8 8116 // CHECK9-NEXT: [[TMP3:%.*]] = load i32**, i32*** [[C_ADDR]], align 8 8117 // CHECK9-NEXT: [[TMP4:%.*]] = load i32, i32* [[TMP0]], align 4 8118 // CHECK9-NEXT: store i32 [[TMP4]], i32* [[DOTCAPTURE_EXPR_]], align 4 8119 // CHECK9-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 8120 // CHECK9-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP5]], 0 8121 // CHECK9-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 8122 // CHECK9-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1 8123 // CHECK9-NEXT: store i32 [[SUB2]], i32* [[DOTCAPTURE_EXPR_1]], align 4 8124 // CHECK9-NEXT: store i32 0, i32* [[I]], align 4 8125 // CHECK9-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 8126 // CHECK9-NEXT: [[CMP:%.*]] = icmp slt i32 0, [[TMP6]] 8127 // CHECK9-NEXT: br i1 [[CMP]], label [[OMP_PRECOND_THEN:%.*]], label [[OMP_PRECOND_END:%.*]] 8128 // CHECK9: omp.precond.then: 8129 // CHECK9-NEXT: store i32 0, i32* [[DOTOMP_COMB_LB]], align 4 8130 // CHECK9-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 8131 // CHECK9-NEXT: store i32 [[TMP7]], i32* [[DOTOMP_COMB_UB]], align 4 8132 // CHECK9-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4 8133 // CHECK9-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4 8134 // CHECK9-NEXT: [[TMP8:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 8135 // CHECK9-NEXT: [[TMP9:%.*]] = load i32, i32* [[TMP8]], align 4 8136 // CHECK9-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1]], i32 [[TMP9]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_COMB_LB]], i32* [[DOTOMP_COMB_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1) 8137 // CHECK9-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 8138 // CHECK9-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 8139 // CHECK9-NEXT: [[CMP4:%.*]] = icmp sgt i32 [[TMP10]], [[TMP11]] 8140 // CHECK9-NEXT: br i1 [[CMP4]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] 8141 // CHECK9: cond.true: 8142 // CHECK9-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 8143 // CHECK9-NEXT: br label [[COND_END:%.*]] 8144 // CHECK9: cond.false: 8145 // CHECK9-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 8146 // CHECK9-NEXT: br label [[COND_END]] 8147 // CHECK9: cond.end: 8148 // CHECK9-NEXT: [[COND:%.*]] = phi i32 [ [[TMP12]], [[COND_TRUE]] ], [ [[TMP13]], [[COND_FALSE]] ] 8149 // CHECK9-NEXT: store i32 [[COND]], i32* [[DOTOMP_COMB_UB]], align 4 8150 // CHECK9-NEXT: [[TMP14:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4 8151 // CHECK9-NEXT: store i32 [[TMP14]], i32* [[DOTOMP_IV]], align 4 8152 // CHECK9-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] 8153 // CHECK9: omp.inner.for.cond: 8154 // CHECK9-NEXT: [[TMP15:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !68 8155 // CHECK9-NEXT: [[TMP16:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4, !llvm.access.group !68 8156 // CHECK9-NEXT: [[CMP5:%.*]] = icmp sle i32 [[TMP15]], [[TMP16]] 8157 // CHECK9-NEXT: br i1 [[CMP5]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] 8158 // CHECK9: omp.inner.for.body: 8159 // CHECK9-NEXT: [[TMP17:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4, !llvm.access.group !68 8160 // CHECK9-NEXT: [[TMP18:%.*]] = zext i32 [[TMP17]] to i64 8161 // CHECK9-NEXT: [[TMP19:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4, !llvm.access.group !68 8162 // CHECK9-NEXT: [[TMP20:%.*]] = zext i32 [[TMP19]] to i64 8163 // CHECK9-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 6, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64, i32*, i32**, i32**, i32**)* @.omp_outlined..31 to void (i32*, i32*, ...)*), i64 [[TMP18]], i64 [[TMP20]], i32* [[TMP0]], i32** [[TMP1]], i32** [[TMP2]], i32** [[TMP3]]), !llvm.access.group !68 8164 // CHECK9-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] 8165 // CHECK9: omp.inner.for.inc: 8166 // CHECK9-NEXT: [[TMP21:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !68 8167 // CHECK9-NEXT: [[TMP22:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4, !llvm.access.group !68 8168 // CHECK9-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP21]], [[TMP22]] 8169 // CHECK9-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4, !llvm.access.group !68 8170 // CHECK9-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP69:![0-9]+]] 8171 // CHECK9: omp.inner.for.end: 8172 // CHECK9-NEXT: br label [[OMP_LOOP_EXIT:%.*]] 8173 // CHECK9: omp.loop.exit: 8174 // CHECK9-NEXT: [[TMP23:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 8175 // CHECK9-NEXT: [[TMP24:%.*]] = load i32, i32* [[TMP23]], align 4 8176 // CHECK9-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP24]]) 8177 // CHECK9-NEXT: [[TMP25:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4 8178 // CHECK9-NEXT: [[TMP26:%.*]] = icmp ne i32 [[TMP25]], 0 8179 // CHECK9-NEXT: br i1 [[TMP26]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]] 8180 // CHECK9: .omp.final.then: 8181 // CHECK9-NEXT: [[TMP27:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 8182 // CHECK9-NEXT: [[SUB6:%.*]] = sub nsw i32 [[TMP27]], 0 8183 // CHECK9-NEXT: [[DIV7:%.*]] = sdiv i32 [[SUB6]], 1 8184 // CHECK9-NEXT: [[MUL:%.*]] = mul nsw i32 [[DIV7]], 1 8185 // CHECK9-NEXT: [[ADD8:%.*]] = add nsw i32 0, [[MUL]] 8186 // CHECK9-NEXT: store i32 [[ADD8]], i32* [[I3]], align 4 8187 // CHECK9-NEXT: br label [[DOTOMP_FINAL_DONE]] 8188 // CHECK9: .omp.final.done: 8189 // CHECK9-NEXT: br label [[OMP_PRECOND_END]] 8190 // CHECK9: omp.precond.end: 8191 // CHECK9-NEXT: ret void 8192 // 8193 // 8194 // CHECK9-LABEL: define {{[^@]+}}@.omp_outlined..31 8195 // CHECK9-SAME: (i32* noalias noundef [[DOTGLOBAL_TID_:%.*]], i32* noalias noundef [[DOTBOUND_TID_:%.*]], i64 noundef [[DOTPREVIOUS_LB_:%.*]], i64 noundef [[DOTPREVIOUS_UB_:%.*]], i32* noundef nonnull align 4 dereferenceable(4) [[N:%.*]], i32** noundef nonnull align 8 dereferenceable(8) [[A:%.*]], i32** noundef nonnull align 8 dereferenceable(8) [[B:%.*]], i32** noundef nonnull align 8 dereferenceable(8) [[C:%.*]]) #[[ATTR1]] { 8196 // CHECK9-NEXT: entry: 8197 // CHECK9-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 8198 // CHECK9-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 8199 // CHECK9-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8 8200 // CHECK9-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8 8201 // CHECK9-NEXT: [[N_ADDR:%.*]] = alloca i32*, align 8 8202 // CHECK9-NEXT: [[A_ADDR:%.*]] = alloca i32**, align 8 8203 // CHECK9-NEXT: [[B_ADDR:%.*]] = alloca i32**, align 8 8204 // CHECK9-NEXT: [[C_ADDR:%.*]] = alloca i32**, align 8 8205 // CHECK9-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 8206 // CHECK9-NEXT: [[TMP:%.*]] = alloca i32, align 4 8207 // CHECK9-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 8208 // CHECK9-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 8209 // CHECK9-NEXT: [[I:%.*]] = alloca i32, align 4 8210 // CHECK9-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 8211 // CHECK9-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4 8212 // CHECK9-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 8213 // CHECK9-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 8214 // CHECK9-NEXT: [[I4:%.*]] = alloca i32, align 4 8215 // CHECK9-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8 8216 // CHECK9-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8 8217 // CHECK9-NEXT: store i64 [[DOTPREVIOUS_LB_]], i64* [[DOTPREVIOUS_LB__ADDR]], align 8 8218 // CHECK9-NEXT: store i64 [[DOTPREVIOUS_UB_]], i64* [[DOTPREVIOUS_UB__ADDR]], align 8 8219 // CHECK9-NEXT: store i32* [[N]], i32** [[N_ADDR]], align 8 8220 // CHECK9-NEXT: store i32** [[A]], i32*** [[A_ADDR]], align 8 8221 // CHECK9-NEXT: store i32** [[B]], i32*** [[B_ADDR]], align 8 8222 // CHECK9-NEXT: store i32** [[C]], i32*** [[C_ADDR]], align 8 8223 // CHECK9-NEXT: [[TMP0:%.*]] = load i32*, i32** [[N_ADDR]], align 8 8224 // CHECK9-NEXT: [[TMP1:%.*]] = load i32**, i32*** [[A_ADDR]], align 8 8225 // CHECK9-NEXT: [[TMP2:%.*]] = load i32**, i32*** [[B_ADDR]], align 8 8226 // CHECK9-NEXT: [[TMP3:%.*]] = load i32**, i32*** [[C_ADDR]], align 8 8227 // CHECK9-NEXT: [[TMP4:%.*]] = load i32, i32* [[TMP0]], align 4 8228 // CHECK9-NEXT: store i32 [[TMP4]], i32* [[DOTCAPTURE_EXPR_]], align 4 8229 // CHECK9-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 8230 // CHECK9-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP5]], 0 8231 // CHECK9-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 8232 // CHECK9-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1 8233 // CHECK9-NEXT: store i32 [[SUB2]], i32* [[DOTCAPTURE_EXPR_1]], align 4 8234 // CHECK9-NEXT: store i32 0, i32* [[I]], align 4 8235 // CHECK9-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 8236 // CHECK9-NEXT: [[CMP:%.*]] = icmp slt i32 0, [[TMP6]] 8237 // CHECK9-NEXT: br i1 [[CMP]], label [[OMP_PRECOND_THEN:%.*]], label [[OMP_PRECOND_END:%.*]] 8238 // CHECK9: omp.precond.then: 8239 // CHECK9-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4 8240 // CHECK9-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 8241 // CHECK9-NEXT: store i32 [[TMP7]], i32* [[DOTOMP_UB]], align 4 8242 // CHECK9-NEXT: [[TMP8:%.*]] = load i64, i64* [[DOTPREVIOUS_LB__ADDR]], align 8 8243 // CHECK9-NEXT: [[CONV:%.*]] = trunc i64 [[TMP8]] to i32 8244 // CHECK9-NEXT: [[TMP9:%.*]] = load i64, i64* [[DOTPREVIOUS_UB__ADDR]], align 8 8245 // CHECK9-NEXT: [[CONV3:%.*]] = trunc i64 [[TMP9]] to i32 8246 // CHECK9-NEXT: store i32 [[CONV]], i32* [[DOTOMP_LB]], align 4 8247 // CHECK9-NEXT: store i32 [[CONV3]], i32* [[DOTOMP_UB]], align 4 8248 // CHECK9-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4 8249 // CHECK9-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4 8250 // CHECK9-NEXT: [[TMP10:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 8251 // CHECK9-NEXT: [[TMP11:%.*]] = load i32, i32* [[TMP10]], align 4 8252 // CHECK9-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB2]], i32 [[TMP11]], i32 34, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1) 8253 // CHECK9-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 8254 // CHECK9-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 8255 // CHECK9-NEXT: [[CMP5:%.*]] = icmp sgt i32 [[TMP12]], [[TMP13]] 8256 // CHECK9-NEXT: br i1 [[CMP5]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] 8257 // CHECK9: cond.true: 8258 // CHECK9-NEXT: [[TMP14:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 8259 // CHECK9-NEXT: br label [[COND_END:%.*]] 8260 // CHECK9: cond.false: 8261 // CHECK9-NEXT: [[TMP15:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 8262 // CHECK9-NEXT: br label [[COND_END]] 8263 // CHECK9: cond.end: 8264 // CHECK9-NEXT: [[COND:%.*]] = phi i32 [ [[TMP14]], [[COND_TRUE]] ], [ [[TMP15]], [[COND_FALSE]] ] 8265 // CHECK9-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4 8266 // CHECK9-NEXT: [[TMP16:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4 8267 // CHECK9-NEXT: store i32 [[TMP16]], i32* [[DOTOMP_IV]], align 4 8268 // CHECK9-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] 8269 // CHECK9: omp.inner.for.cond: 8270 // CHECK9-NEXT: [[TMP17:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !71 8271 // CHECK9-NEXT: [[TMP18:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4, !llvm.access.group !71 8272 // CHECK9-NEXT: [[CMP6:%.*]] = icmp sle i32 [[TMP17]], [[TMP18]] 8273 // CHECK9-NEXT: br i1 [[CMP6]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] 8274 // CHECK9: omp.inner.for.body: 8275 // CHECK9-NEXT: [[TMP19:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !71 8276 // CHECK9-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP19]], 1 8277 // CHECK9-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] 8278 // CHECK9-NEXT: store i32 [[ADD]], i32* [[I4]], align 4, !llvm.access.group !71 8279 // CHECK9-NEXT: [[TMP20:%.*]] = load i32*, i32** [[TMP2]], align 8, !llvm.access.group !71 8280 // CHECK9-NEXT: [[TMP21:%.*]] = load i32, i32* [[I4]], align 4, !llvm.access.group !71 8281 // CHECK9-NEXT: [[IDXPROM:%.*]] = sext i32 [[TMP21]] to i64 8282 // CHECK9-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i32, i32* [[TMP20]], i64 [[IDXPROM]] 8283 // CHECK9-NEXT: [[TMP22:%.*]] = load i32, i32* [[ARRAYIDX]], align 4, !llvm.access.group !71 8284 // CHECK9-NEXT: [[TMP23:%.*]] = load i32*, i32** [[TMP3]], align 8, !llvm.access.group !71 8285 // CHECK9-NEXT: [[TMP24:%.*]] = load i32, i32* [[I4]], align 4, !llvm.access.group !71 8286 // CHECK9-NEXT: [[IDXPROM7:%.*]] = sext i32 [[TMP24]] to i64 8287 // CHECK9-NEXT: [[ARRAYIDX8:%.*]] = getelementptr inbounds i32, i32* [[TMP23]], i64 [[IDXPROM7]] 8288 // CHECK9-NEXT: [[TMP25:%.*]] = load i32, i32* [[ARRAYIDX8]], align 4, !llvm.access.group !71 8289 // CHECK9-NEXT: [[ADD9:%.*]] = add nsw i32 [[TMP22]], [[TMP25]] 8290 // CHECK9-NEXT: [[TMP26:%.*]] = load i32*, i32** [[TMP1]], align 8, !llvm.access.group !71 8291 // CHECK9-NEXT: [[TMP27:%.*]] = load i32, i32* [[I4]], align 4, !llvm.access.group !71 8292 // CHECK9-NEXT: [[IDXPROM10:%.*]] = sext i32 [[TMP27]] to i64 8293 // CHECK9-NEXT: [[ARRAYIDX11:%.*]] = getelementptr inbounds i32, i32* [[TMP26]], i64 [[IDXPROM10]] 8294 // CHECK9-NEXT: store i32 [[ADD9]], i32* [[ARRAYIDX11]], align 4, !llvm.access.group !71 8295 // CHECK9-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] 8296 // CHECK9: omp.body.continue: 8297 // CHECK9-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] 8298 // CHECK9: omp.inner.for.inc: 8299 // CHECK9-NEXT: [[TMP28:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !71 8300 // CHECK9-NEXT: [[ADD12:%.*]] = add nsw i32 [[TMP28]], 1 8301 // CHECK9-NEXT: store i32 [[ADD12]], i32* [[DOTOMP_IV]], align 4, !llvm.access.group !71 8302 // CHECK9-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP72:![0-9]+]] 8303 // CHECK9: omp.inner.for.end: 8304 // CHECK9-NEXT: br label [[OMP_LOOP_EXIT:%.*]] 8305 // CHECK9: omp.loop.exit: 8306 // CHECK9-NEXT: [[TMP29:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 8307 // CHECK9-NEXT: [[TMP30:%.*]] = load i32, i32* [[TMP29]], align 4 8308 // CHECK9-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP30]]) 8309 // CHECK9-NEXT: [[TMP31:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4 8310 // CHECK9-NEXT: [[TMP32:%.*]] = icmp ne i32 [[TMP31]], 0 8311 // CHECK9-NEXT: br i1 [[TMP32]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]] 8312 // CHECK9: .omp.final.then: 8313 // CHECK9-NEXT: [[TMP33:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 8314 // CHECK9-NEXT: [[SUB13:%.*]] = sub nsw i32 [[TMP33]], 0 8315 // CHECK9-NEXT: [[DIV14:%.*]] = sdiv i32 [[SUB13]], 1 8316 // CHECK9-NEXT: [[MUL15:%.*]] = mul nsw i32 [[DIV14]], 1 8317 // CHECK9-NEXT: [[ADD16:%.*]] = add nsw i32 0, [[MUL15]] 8318 // CHECK9-NEXT: store i32 [[ADD16]], i32* [[I4]], align 4 8319 // CHECK9-NEXT: br label [[DOTOMP_FINAL_DONE]] 8320 // CHECK9: .omp.final.done: 8321 // CHECK9-NEXT: br label [[OMP_PRECOND_END]] 8322 // CHECK9: omp.precond.end: 8323 // CHECK9-NEXT: ret void 8324 // 8325 // 8326 // CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiET_v_l58 8327 // CHECK9-SAME: (i64 noundef [[CH:%.*]], i64 noundef [[N:%.*]], i32* noundef [[A:%.*]], i32* noundef [[B:%.*]], i32* noundef [[C:%.*]]) #[[ATTR1]] { 8328 // CHECK9-NEXT: entry: 8329 // CHECK9-NEXT: [[CH_ADDR:%.*]] = alloca i64, align 8 8330 // CHECK9-NEXT: [[N_ADDR:%.*]] = alloca i64, align 8 8331 // CHECK9-NEXT: [[A_ADDR:%.*]] = alloca i32*, align 8 8332 // CHECK9-NEXT: [[B_ADDR:%.*]] = alloca i32*, align 8 8333 // CHECK9-NEXT: [[C_ADDR:%.*]] = alloca i32*, align 8 8334 // CHECK9-NEXT: store i64 [[CH]], i64* [[CH_ADDR]], align 8 8335 // CHECK9-NEXT: store i64 [[N]], i64* [[N_ADDR]], align 8 8336 // CHECK9-NEXT: store i32* [[A]], i32** [[A_ADDR]], align 8 8337 // CHECK9-NEXT: store i32* [[B]], i32** [[B_ADDR]], align 8 8338 // CHECK9-NEXT: store i32* [[C]], i32** [[C_ADDR]], align 8 8339 // CHECK9-NEXT: [[CONV:%.*]] = bitcast i64* [[CH_ADDR]] to i32* 8340 // CHECK9-NEXT: [[CONV1:%.*]] = bitcast i64* [[N_ADDR]] to i32* 8341 // CHECK9-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 5, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32*, i32*, i32**, i32**, i32**)* @.omp_outlined..34 to void (i32*, i32*, ...)*), i32* [[CONV]], i32* [[CONV1]], i32** [[A_ADDR]], i32** [[B_ADDR]], i32** [[C_ADDR]]) 8342 // CHECK9-NEXT: ret void 8343 // 8344 // 8345 // CHECK9-LABEL: define {{[^@]+}}@.omp_outlined..34 8346 // CHECK9-SAME: (i32* noalias noundef [[DOTGLOBAL_TID_:%.*]], i32* noalias noundef [[DOTBOUND_TID_:%.*]], i32* noundef nonnull align 4 dereferenceable(4) [[CH:%.*]], i32* noundef nonnull align 4 dereferenceable(4) [[N:%.*]], i32** noundef nonnull align 8 dereferenceable(8) [[A:%.*]], i32** noundef nonnull align 8 dereferenceable(8) [[B:%.*]], i32** noundef nonnull align 8 dereferenceable(8) [[C:%.*]]) #[[ATTR1]] { 8347 // CHECK9-NEXT: entry: 8348 // CHECK9-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 8349 // CHECK9-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 8350 // CHECK9-NEXT: [[CH_ADDR:%.*]] = alloca i32*, align 8 8351 // CHECK9-NEXT: [[N_ADDR:%.*]] = alloca i32*, align 8 8352 // CHECK9-NEXT: [[A_ADDR:%.*]] = alloca i32**, align 8 8353 // CHECK9-NEXT: [[B_ADDR:%.*]] = alloca i32**, align 8 8354 // CHECK9-NEXT: [[C_ADDR:%.*]] = alloca i32**, align 8 8355 // CHECK9-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 8356 // CHECK9-NEXT: [[TMP:%.*]] = alloca i32, align 4 8357 // CHECK9-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 8358 // CHECK9-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 8359 // CHECK9-NEXT: [[I:%.*]] = alloca i32, align 4 8360 // CHECK9-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4 8361 // CHECK9-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4 8362 // CHECK9-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 8363 // CHECK9-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 8364 // CHECK9-NEXT: [[I3:%.*]] = alloca i32, align 4 8365 // CHECK9-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8 8366 // CHECK9-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8 8367 // CHECK9-NEXT: store i32* [[CH]], i32** [[CH_ADDR]], align 8 8368 // CHECK9-NEXT: store i32* [[N]], i32** [[N_ADDR]], align 8 8369 // CHECK9-NEXT: store i32** [[A]], i32*** [[A_ADDR]], align 8 8370 // CHECK9-NEXT: store i32** [[B]], i32*** [[B_ADDR]], align 8 8371 // CHECK9-NEXT: store i32** [[C]], i32*** [[C_ADDR]], align 8 8372 // CHECK9-NEXT: [[TMP0:%.*]] = load i32*, i32** [[CH_ADDR]], align 8 8373 // CHECK9-NEXT: [[TMP1:%.*]] = load i32*, i32** [[N_ADDR]], align 8 8374 // CHECK9-NEXT: [[TMP2:%.*]] = load i32**, i32*** [[A_ADDR]], align 8 8375 // CHECK9-NEXT: [[TMP3:%.*]] = load i32**, i32*** [[B_ADDR]], align 8 8376 // CHECK9-NEXT: [[TMP4:%.*]] = load i32**, i32*** [[C_ADDR]], align 8 8377 // CHECK9-NEXT: [[TMP5:%.*]] = load i32, i32* [[TMP1]], align 4 8378 // CHECK9-NEXT: store i32 [[TMP5]], i32* [[DOTCAPTURE_EXPR_]], align 4 8379 // CHECK9-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 8380 // CHECK9-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP6]], 0 8381 // CHECK9-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 8382 // CHECK9-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1 8383 // CHECK9-NEXT: store i32 [[SUB2]], i32* [[DOTCAPTURE_EXPR_1]], align 4 8384 // CHECK9-NEXT: store i32 0, i32* [[I]], align 4 8385 // CHECK9-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 8386 // CHECK9-NEXT: [[CMP:%.*]] = icmp slt i32 0, [[TMP7]] 8387 // CHECK9-NEXT: br i1 [[CMP]], label [[OMP_PRECOND_THEN:%.*]], label [[OMP_PRECOND_END:%.*]] 8388 // CHECK9: omp.precond.then: 8389 // CHECK9-NEXT: store i32 0, i32* [[DOTOMP_COMB_LB]], align 4 8390 // CHECK9-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 8391 // CHECK9-NEXT: store i32 [[TMP8]], i32* [[DOTOMP_COMB_UB]], align 4 8392 // CHECK9-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4 8393 // CHECK9-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4 8394 // CHECK9-NEXT: [[TMP9:%.*]] = load i32, i32* [[TMP0]], align 4 8395 // CHECK9-NEXT: [[TMP10:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 8396 // CHECK9-NEXT: [[TMP11:%.*]] = load i32, i32* [[TMP10]], align 4 8397 // CHECK9-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1]], i32 [[TMP11]], i32 91, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_COMB_LB]], i32* [[DOTOMP_COMB_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 [[TMP9]]) 8398 // CHECK9-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 8399 // CHECK9-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 8400 // CHECK9-NEXT: [[CMP4:%.*]] = icmp sgt i32 [[TMP12]], [[TMP13]] 8401 // CHECK9-NEXT: br i1 [[CMP4]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] 8402 // CHECK9: cond.true: 8403 // CHECK9-NEXT: [[TMP14:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 8404 // CHECK9-NEXT: br label [[COND_END:%.*]] 8405 // CHECK9: cond.false: 8406 // CHECK9-NEXT: [[TMP15:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 8407 // CHECK9-NEXT: br label [[COND_END]] 8408 // CHECK9: cond.end: 8409 // CHECK9-NEXT: [[COND:%.*]] = phi i32 [ [[TMP14]], [[COND_TRUE]] ], [ [[TMP15]], [[COND_FALSE]] ] 8410 // CHECK9-NEXT: store i32 [[COND]], i32* [[DOTOMP_COMB_UB]], align 4 8411 // CHECK9-NEXT: [[TMP16:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4 8412 // CHECK9-NEXT: store i32 [[TMP16]], i32* [[DOTOMP_IV]], align 4 8413 // CHECK9-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] 8414 // CHECK9: omp.inner.for.cond: 8415 // CHECK9-NEXT: [[TMP17:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !74 8416 // CHECK9-NEXT: [[TMP18:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4, !llvm.access.group !74 8417 // CHECK9-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP18]], 1 8418 // CHECK9-NEXT: [[CMP5:%.*]] = icmp slt i32 [[TMP17]], [[ADD]] 8419 // CHECK9-NEXT: br i1 [[CMP5]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] 8420 // CHECK9: omp.inner.for.body: 8421 // CHECK9-NEXT: [[TMP19:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4, !llvm.access.group !74 8422 // CHECK9-NEXT: [[TMP20:%.*]] = zext i32 [[TMP19]] to i64 8423 // CHECK9-NEXT: [[TMP21:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4, !llvm.access.group !74 8424 // CHECK9-NEXT: [[TMP22:%.*]] = zext i32 [[TMP21]] to i64 8425 // CHECK9-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 6, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64, i32*, i32**, i32**, i32**)* @.omp_outlined..35 to void (i32*, i32*, ...)*), i64 [[TMP20]], i64 [[TMP22]], i32* [[TMP1]], i32** [[TMP2]], i32** [[TMP3]], i32** [[TMP4]]), !llvm.access.group !74 8426 // CHECK9-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] 8427 // CHECK9: omp.inner.for.inc: 8428 // CHECK9-NEXT: [[TMP23:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !74 8429 // CHECK9-NEXT: [[TMP24:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4, !llvm.access.group !74 8430 // CHECK9-NEXT: [[ADD6:%.*]] = add nsw i32 [[TMP23]], [[TMP24]] 8431 // CHECK9-NEXT: store i32 [[ADD6]], i32* [[DOTOMP_IV]], align 4, !llvm.access.group !74 8432 // CHECK9-NEXT: [[TMP25:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4, !llvm.access.group !74 8433 // CHECK9-NEXT: [[TMP26:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4, !llvm.access.group !74 8434 // CHECK9-NEXT: [[ADD7:%.*]] = add nsw i32 [[TMP25]], [[TMP26]] 8435 // CHECK9-NEXT: store i32 [[ADD7]], i32* [[DOTOMP_COMB_LB]], align 4, !llvm.access.group !74 8436 // CHECK9-NEXT: [[TMP27:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4, !llvm.access.group !74 8437 // CHECK9-NEXT: [[TMP28:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4, !llvm.access.group !74 8438 // CHECK9-NEXT: [[ADD8:%.*]] = add nsw i32 [[TMP27]], [[TMP28]] 8439 // CHECK9-NEXT: store i32 [[ADD8]], i32* [[DOTOMP_COMB_UB]], align 4, !llvm.access.group !74 8440 // CHECK9-NEXT: [[TMP29:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4, !llvm.access.group !74 8441 // CHECK9-NEXT: [[TMP30:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4, !llvm.access.group !74 8442 // CHECK9-NEXT: [[CMP9:%.*]] = icmp sgt i32 [[TMP29]], [[TMP30]] 8443 // CHECK9-NEXT: br i1 [[CMP9]], label [[COND_TRUE10:%.*]], label [[COND_FALSE11:%.*]] 8444 // CHECK9: cond.true10: 8445 // CHECK9-NEXT: [[TMP31:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4, !llvm.access.group !74 8446 // CHECK9-NEXT: br label [[COND_END12:%.*]] 8447 // CHECK9: cond.false11: 8448 // CHECK9-NEXT: [[TMP32:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4, !llvm.access.group !74 8449 // CHECK9-NEXT: br label [[COND_END12]] 8450 // CHECK9: cond.end12: 8451 // CHECK9-NEXT: [[COND13:%.*]] = phi i32 [ [[TMP31]], [[COND_TRUE10]] ], [ [[TMP32]], [[COND_FALSE11]] ] 8452 // CHECK9-NEXT: store i32 [[COND13]], i32* [[DOTOMP_COMB_UB]], align 4, !llvm.access.group !74 8453 // CHECK9-NEXT: [[TMP33:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4, !llvm.access.group !74 8454 // CHECK9-NEXT: store i32 [[TMP33]], i32* [[DOTOMP_IV]], align 4, !llvm.access.group !74 8455 // CHECK9-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP75:![0-9]+]] 8456 // CHECK9: omp.inner.for.end: 8457 // CHECK9-NEXT: br label [[OMP_LOOP_EXIT:%.*]] 8458 // CHECK9: omp.loop.exit: 8459 // CHECK9-NEXT: [[TMP34:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 8460 // CHECK9-NEXT: [[TMP35:%.*]] = load i32, i32* [[TMP34]], align 4 8461 // CHECK9-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP35]]) 8462 // CHECK9-NEXT: [[TMP36:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4 8463 // CHECK9-NEXT: [[TMP37:%.*]] = icmp ne i32 [[TMP36]], 0 8464 // CHECK9-NEXT: br i1 [[TMP37]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]] 8465 // CHECK9: .omp.final.then: 8466 // CHECK9-NEXT: [[TMP38:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 8467 // CHECK9-NEXT: [[SUB14:%.*]] = sub nsw i32 [[TMP38]], 0 8468 // CHECK9-NEXT: [[DIV15:%.*]] = sdiv i32 [[SUB14]], 1 8469 // CHECK9-NEXT: [[MUL:%.*]] = mul nsw i32 [[DIV15]], 1 8470 // CHECK9-NEXT: [[ADD16:%.*]] = add nsw i32 0, [[MUL]] 8471 // CHECK9-NEXT: store i32 [[ADD16]], i32* [[I3]], align 4 8472 // CHECK9-NEXT: br label [[DOTOMP_FINAL_DONE]] 8473 // CHECK9: .omp.final.done: 8474 // CHECK9-NEXT: br label [[OMP_PRECOND_END]] 8475 // CHECK9: omp.precond.end: 8476 // CHECK9-NEXT: ret void 8477 // 8478 // 8479 // CHECK9-LABEL: define {{[^@]+}}@.omp_outlined..35 8480 // CHECK9-SAME: (i32* noalias noundef [[DOTGLOBAL_TID_:%.*]], i32* noalias noundef [[DOTBOUND_TID_:%.*]], i64 noundef [[DOTPREVIOUS_LB_:%.*]], i64 noundef [[DOTPREVIOUS_UB_:%.*]], i32* noundef nonnull align 4 dereferenceable(4) [[N:%.*]], i32** noundef nonnull align 8 dereferenceable(8) [[A:%.*]], i32** noundef nonnull align 8 dereferenceable(8) [[B:%.*]], i32** noundef nonnull align 8 dereferenceable(8) [[C:%.*]]) #[[ATTR1]] { 8481 // CHECK9-NEXT: entry: 8482 // CHECK9-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 8483 // CHECK9-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 8484 // CHECK9-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8 8485 // CHECK9-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8 8486 // CHECK9-NEXT: [[N_ADDR:%.*]] = alloca i32*, align 8 8487 // CHECK9-NEXT: [[A_ADDR:%.*]] = alloca i32**, align 8 8488 // CHECK9-NEXT: [[B_ADDR:%.*]] = alloca i32**, align 8 8489 // CHECK9-NEXT: [[C_ADDR:%.*]] = alloca i32**, align 8 8490 // CHECK9-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 8491 // CHECK9-NEXT: [[TMP:%.*]] = alloca i32, align 4 8492 // CHECK9-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 8493 // CHECK9-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 8494 // CHECK9-NEXT: [[I:%.*]] = alloca i32, align 4 8495 // CHECK9-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 8496 // CHECK9-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4 8497 // CHECK9-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 8498 // CHECK9-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 8499 // CHECK9-NEXT: [[I4:%.*]] = alloca i32, align 4 8500 // CHECK9-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8 8501 // CHECK9-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8 8502 // CHECK9-NEXT: store i64 [[DOTPREVIOUS_LB_]], i64* [[DOTPREVIOUS_LB__ADDR]], align 8 8503 // CHECK9-NEXT: store i64 [[DOTPREVIOUS_UB_]], i64* [[DOTPREVIOUS_UB__ADDR]], align 8 8504 // CHECK9-NEXT: store i32* [[N]], i32** [[N_ADDR]], align 8 8505 // CHECK9-NEXT: store i32** [[A]], i32*** [[A_ADDR]], align 8 8506 // CHECK9-NEXT: store i32** [[B]], i32*** [[B_ADDR]], align 8 8507 // CHECK9-NEXT: store i32** [[C]], i32*** [[C_ADDR]], align 8 8508 // CHECK9-NEXT: [[TMP0:%.*]] = load i32*, i32** [[N_ADDR]], align 8 8509 // CHECK9-NEXT: [[TMP1:%.*]] = load i32**, i32*** [[A_ADDR]], align 8 8510 // CHECK9-NEXT: [[TMP2:%.*]] = load i32**, i32*** [[B_ADDR]], align 8 8511 // CHECK9-NEXT: [[TMP3:%.*]] = load i32**, i32*** [[C_ADDR]], align 8 8512 // CHECK9-NEXT: [[TMP4:%.*]] = load i32, i32* [[TMP0]], align 4 8513 // CHECK9-NEXT: store i32 [[TMP4]], i32* [[DOTCAPTURE_EXPR_]], align 4 8514 // CHECK9-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 8515 // CHECK9-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP5]], 0 8516 // CHECK9-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 8517 // CHECK9-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1 8518 // CHECK9-NEXT: store i32 [[SUB2]], i32* [[DOTCAPTURE_EXPR_1]], align 4 8519 // CHECK9-NEXT: store i32 0, i32* [[I]], align 4 8520 // CHECK9-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 8521 // CHECK9-NEXT: [[CMP:%.*]] = icmp slt i32 0, [[TMP6]] 8522 // CHECK9-NEXT: br i1 [[CMP]], label [[OMP_PRECOND_THEN:%.*]], label [[OMP_PRECOND_END:%.*]] 8523 // CHECK9: omp.precond.then: 8524 // CHECK9-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4 8525 // CHECK9-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 8526 // CHECK9-NEXT: store i32 [[TMP7]], i32* [[DOTOMP_UB]], align 4 8527 // CHECK9-NEXT: [[TMP8:%.*]] = load i64, i64* [[DOTPREVIOUS_LB__ADDR]], align 8 8528 // CHECK9-NEXT: [[CONV:%.*]] = trunc i64 [[TMP8]] to i32 8529 // CHECK9-NEXT: [[TMP9:%.*]] = load i64, i64* [[DOTPREVIOUS_UB__ADDR]], align 8 8530 // CHECK9-NEXT: [[CONV3:%.*]] = trunc i64 [[TMP9]] to i32 8531 // CHECK9-NEXT: store i32 [[CONV]], i32* [[DOTOMP_LB]], align 4 8532 // CHECK9-NEXT: store i32 [[CONV3]], i32* [[DOTOMP_UB]], align 4 8533 // CHECK9-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4 8534 // CHECK9-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4 8535 // CHECK9-NEXT: [[TMP10:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 8536 // CHECK9-NEXT: [[TMP11:%.*]] = load i32, i32* [[TMP10]], align 4 8537 // CHECK9-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB2]], i32 [[TMP11]], i32 34, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1) 8538 // CHECK9-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 8539 // CHECK9-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 8540 // CHECK9-NEXT: [[CMP5:%.*]] = icmp sgt i32 [[TMP12]], [[TMP13]] 8541 // CHECK9-NEXT: br i1 [[CMP5]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] 8542 // CHECK9: cond.true: 8543 // CHECK9-NEXT: [[TMP14:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 8544 // CHECK9-NEXT: br label [[COND_END:%.*]] 8545 // CHECK9: cond.false: 8546 // CHECK9-NEXT: [[TMP15:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 8547 // CHECK9-NEXT: br label [[COND_END]] 8548 // CHECK9: cond.end: 8549 // CHECK9-NEXT: [[COND:%.*]] = phi i32 [ [[TMP14]], [[COND_TRUE]] ], [ [[TMP15]], [[COND_FALSE]] ] 8550 // CHECK9-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4 8551 // CHECK9-NEXT: [[TMP16:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4 8552 // CHECK9-NEXT: store i32 [[TMP16]], i32* [[DOTOMP_IV]], align 4 8553 // CHECK9-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] 8554 // CHECK9: omp.inner.for.cond: 8555 // CHECK9-NEXT: [[TMP17:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !77 8556 // CHECK9-NEXT: [[TMP18:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4, !llvm.access.group !77 8557 // CHECK9-NEXT: [[CMP6:%.*]] = icmp sle i32 [[TMP17]], [[TMP18]] 8558 // CHECK9-NEXT: br i1 [[CMP6]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] 8559 // CHECK9: omp.inner.for.body: 8560 // CHECK9-NEXT: [[TMP19:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !77 8561 // CHECK9-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP19]], 1 8562 // CHECK9-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] 8563 // CHECK9-NEXT: store i32 [[ADD]], i32* [[I4]], align 4, !llvm.access.group !77 8564 // CHECK9-NEXT: [[TMP20:%.*]] = load i32*, i32** [[TMP2]], align 8, !llvm.access.group !77 8565 // CHECK9-NEXT: [[TMP21:%.*]] = load i32, i32* [[I4]], align 4, !llvm.access.group !77 8566 // CHECK9-NEXT: [[IDXPROM:%.*]] = sext i32 [[TMP21]] to i64 8567 // CHECK9-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i32, i32* [[TMP20]], i64 [[IDXPROM]] 8568 // CHECK9-NEXT: [[TMP22:%.*]] = load i32, i32* [[ARRAYIDX]], align 4, !llvm.access.group !77 8569 // CHECK9-NEXT: [[TMP23:%.*]] = load i32*, i32** [[TMP3]], align 8, !llvm.access.group !77 8570 // CHECK9-NEXT: [[TMP24:%.*]] = load i32, i32* [[I4]], align 4, !llvm.access.group !77 8571 // CHECK9-NEXT: [[IDXPROM7:%.*]] = sext i32 [[TMP24]] to i64 8572 // CHECK9-NEXT: [[ARRAYIDX8:%.*]] = getelementptr inbounds i32, i32* [[TMP23]], i64 [[IDXPROM7]] 8573 // CHECK9-NEXT: [[TMP25:%.*]] = load i32, i32* [[ARRAYIDX8]], align 4, !llvm.access.group !77 8574 // CHECK9-NEXT: [[ADD9:%.*]] = add nsw i32 [[TMP22]], [[TMP25]] 8575 // CHECK9-NEXT: [[TMP26:%.*]] = load i32*, i32** [[TMP1]], align 8, !llvm.access.group !77 8576 // CHECK9-NEXT: [[TMP27:%.*]] = load i32, i32* [[I4]], align 4, !llvm.access.group !77 8577 // CHECK9-NEXT: [[IDXPROM10:%.*]] = sext i32 [[TMP27]] to i64 8578 // CHECK9-NEXT: [[ARRAYIDX11:%.*]] = getelementptr inbounds i32, i32* [[TMP26]], i64 [[IDXPROM10]] 8579 // CHECK9-NEXT: store i32 [[ADD9]], i32* [[ARRAYIDX11]], align 4, !llvm.access.group !77 8580 // CHECK9-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] 8581 // CHECK9: omp.body.continue: 8582 // CHECK9-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] 8583 // CHECK9: omp.inner.for.inc: 8584 // CHECK9-NEXT: [[TMP28:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !77 8585 // CHECK9-NEXT: [[ADD12:%.*]] = add nsw i32 [[TMP28]], 1 8586 // CHECK9-NEXT: store i32 [[ADD12]], i32* [[DOTOMP_IV]], align 4, !llvm.access.group !77 8587 // CHECK9-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP78:![0-9]+]] 8588 // CHECK9: omp.inner.for.end: 8589 // CHECK9-NEXT: br label [[OMP_LOOP_EXIT:%.*]] 8590 // CHECK9: omp.loop.exit: 8591 // CHECK9-NEXT: [[TMP29:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 8592 // CHECK9-NEXT: [[TMP30:%.*]] = load i32, i32* [[TMP29]], align 4 8593 // CHECK9-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP30]]) 8594 // CHECK9-NEXT: [[TMP31:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4 8595 // CHECK9-NEXT: [[TMP32:%.*]] = icmp ne i32 [[TMP31]], 0 8596 // CHECK9-NEXT: br i1 [[TMP32]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]] 8597 // CHECK9: .omp.final.then: 8598 // CHECK9-NEXT: [[TMP33:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 8599 // CHECK9-NEXT: [[SUB13:%.*]] = sub nsw i32 [[TMP33]], 0 8600 // CHECK9-NEXT: [[DIV14:%.*]] = sdiv i32 [[SUB13]], 1 8601 // CHECK9-NEXT: [[MUL15:%.*]] = mul nsw i32 [[DIV14]], 1 8602 // CHECK9-NEXT: [[ADD16:%.*]] = add nsw i32 0, [[MUL15]] 8603 // CHECK9-NEXT: store i32 [[ADD16]], i32* [[I4]], align 4 8604 // CHECK9-NEXT: br label [[DOTOMP_FINAL_DONE]] 8605 // CHECK9: .omp.final.done: 8606 // CHECK9-NEXT: br label [[OMP_PRECOND_END]] 8607 // CHECK9: omp.precond.end: 8608 // CHECK9-NEXT: ret void 8609 // 8610 // 8611 // CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiET_v_l66 8612 // CHECK9-SAME: (i64 noundef [[N:%.*]], i32* noundef [[A:%.*]], i32* noundef [[B:%.*]], i32* noundef [[C:%.*]]) #[[ATTR1]] { 8613 // CHECK9-NEXT: entry: 8614 // CHECK9-NEXT: [[N_ADDR:%.*]] = alloca i64, align 8 8615 // CHECK9-NEXT: [[A_ADDR:%.*]] = alloca i32*, align 8 8616 // CHECK9-NEXT: [[B_ADDR:%.*]] = alloca i32*, align 8 8617 // CHECK9-NEXT: [[C_ADDR:%.*]] = alloca i32*, align 8 8618 // CHECK9-NEXT: store i64 [[N]], i64* [[N_ADDR]], align 8 8619 // CHECK9-NEXT: store i32* [[A]], i32** [[A_ADDR]], align 8 8620 // CHECK9-NEXT: store i32* [[B]], i32** [[B_ADDR]], align 8 8621 // CHECK9-NEXT: store i32* [[C]], i32** [[C_ADDR]], align 8 8622 // CHECK9-NEXT: [[CONV:%.*]] = bitcast i64* [[N_ADDR]] to i32* 8623 // CHECK9-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 4, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32*, i32**, i32**, i32**)* @.omp_outlined..38 to void (i32*, i32*, ...)*), i32* [[CONV]], i32** [[A_ADDR]], i32** [[B_ADDR]], i32** [[C_ADDR]]) 8624 // CHECK9-NEXT: ret void 8625 // 8626 // 8627 // CHECK9-LABEL: define {{[^@]+}}@.omp_outlined..38 8628 // CHECK9-SAME: (i32* noalias noundef [[DOTGLOBAL_TID_:%.*]], i32* noalias noundef [[DOTBOUND_TID_:%.*]], i32* noundef nonnull align 4 dereferenceable(4) [[N:%.*]], i32** noundef nonnull align 8 dereferenceable(8) [[A:%.*]], i32** noundef nonnull align 8 dereferenceable(8) [[B:%.*]], i32** noundef nonnull align 8 dereferenceable(8) [[C:%.*]]) #[[ATTR1]] { 8629 // CHECK9-NEXT: entry: 8630 // CHECK9-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 8631 // CHECK9-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 8632 // CHECK9-NEXT: [[N_ADDR:%.*]] = alloca i32*, align 8 8633 // CHECK9-NEXT: [[A_ADDR:%.*]] = alloca i32**, align 8 8634 // CHECK9-NEXT: [[B_ADDR:%.*]] = alloca i32**, align 8 8635 // CHECK9-NEXT: [[C_ADDR:%.*]] = alloca i32**, align 8 8636 // CHECK9-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 8637 // CHECK9-NEXT: [[TMP:%.*]] = alloca i32, align 4 8638 // CHECK9-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 8639 // CHECK9-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 8640 // CHECK9-NEXT: [[I:%.*]] = alloca i32, align 4 8641 // CHECK9-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4 8642 // CHECK9-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4 8643 // CHECK9-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 8644 // CHECK9-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 8645 // CHECK9-NEXT: [[I3:%.*]] = alloca i32, align 4 8646 // CHECK9-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8 8647 // CHECK9-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8 8648 // CHECK9-NEXT: store i32* [[N]], i32** [[N_ADDR]], align 8 8649 // CHECK9-NEXT: store i32** [[A]], i32*** [[A_ADDR]], align 8 8650 // CHECK9-NEXT: store i32** [[B]], i32*** [[B_ADDR]], align 8 8651 // CHECK9-NEXT: store i32** [[C]], i32*** [[C_ADDR]], align 8 8652 // CHECK9-NEXT: [[TMP0:%.*]] = load i32*, i32** [[N_ADDR]], align 8 8653 // CHECK9-NEXT: [[TMP1:%.*]] = load i32**, i32*** [[A_ADDR]], align 8 8654 // CHECK9-NEXT: [[TMP2:%.*]] = load i32**, i32*** [[B_ADDR]], align 8 8655 // CHECK9-NEXT: [[TMP3:%.*]] = load i32**, i32*** [[C_ADDR]], align 8 8656 // CHECK9-NEXT: [[TMP4:%.*]] = load i32, i32* [[TMP0]], align 4 8657 // CHECK9-NEXT: store i32 [[TMP4]], i32* [[DOTCAPTURE_EXPR_]], align 4 8658 // CHECK9-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 8659 // CHECK9-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP5]], 0 8660 // CHECK9-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 8661 // CHECK9-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1 8662 // CHECK9-NEXT: store i32 [[SUB2]], i32* [[DOTCAPTURE_EXPR_1]], align 4 8663 // CHECK9-NEXT: store i32 0, i32* [[I]], align 4 8664 // CHECK9-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 8665 // CHECK9-NEXT: [[CMP:%.*]] = icmp slt i32 0, [[TMP6]] 8666 // CHECK9-NEXT: br i1 [[CMP]], label [[OMP_PRECOND_THEN:%.*]], label [[OMP_PRECOND_END:%.*]] 8667 // CHECK9: omp.precond.then: 8668 // CHECK9-NEXT: store i32 0, i32* [[DOTOMP_COMB_LB]], align 4 8669 // CHECK9-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 8670 // CHECK9-NEXT: store i32 [[TMP7]], i32* [[DOTOMP_COMB_UB]], align 4 8671 // CHECK9-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4 8672 // CHECK9-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4 8673 // CHECK9-NEXT: [[TMP8:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 8674 // CHECK9-NEXT: [[TMP9:%.*]] = load i32, i32* [[TMP8]], align 4 8675 // CHECK9-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1]], i32 [[TMP9]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_COMB_LB]], i32* [[DOTOMP_COMB_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1) 8676 // CHECK9-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 8677 // CHECK9-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 8678 // CHECK9-NEXT: [[CMP4:%.*]] = icmp sgt i32 [[TMP10]], [[TMP11]] 8679 // CHECK9-NEXT: br i1 [[CMP4]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] 8680 // CHECK9: cond.true: 8681 // CHECK9-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 8682 // CHECK9-NEXT: br label [[COND_END:%.*]] 8683 // CHECK9: cond.false: 8684 // CHECK9-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 8685 // CHECK9-NEXT: br label [[COND_END]] 8686 // CHECK9: cond.end: 8687 // CHECK9-NEXT: [[COND:%.*]] = phi i32 [ [[TMP12]], [[COND_TRUE]] ], [ [[TMP13]], [[COND_FALSE]] ] 8688 // CHECK9-NEXT: store i32 [[COND]], i32* [[DOTOMP_COMB_UB]], align 4 8689 // CHECK9-NEXT: [[TMP14:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4 8690 // CHECK9-NEXT: store i32 [[TMP14]], i32* [[DOTOMP_IV]], align 4 8691 // CHECK9-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] 8692 // CHECK9: omp.inner.for.cond: 8693 // CHECK9-NEXT: [[TMP15:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !80 8694 // CHECK9-NEXT: [[TMP16:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4, !llvm.access.group !80 8695 // CHECK9-NEXT: [[CMP5:%.*]] = icmp sle i32 [[TMP15]], [[TMP16]] 8696 // CHECK9-NEXT: br i1 [[CMP5]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] 8697 // CHECK9: omp.inner.for.body: 8698 // CHECK9-NEXT: [[TMP17:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4, !llvm.access.group !80 8699 // CHECK9-NEXT: [[TMP18:%.*]] = zext i32 [[TMP17]] to i64 8700 // CHECK9-NEXT: [[TMP19:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4, !llvm.access.group !80 8701 // CHECK9-NEXT: [[TMP20:%.*]] = zext i32 [[TMP19]] to i64 8702 // CHECK9-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 6, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64, i32*, i32**, i32**, i32**)* @.omp_outlined..39 to void (i32*, i32*, ...)*), i64 [[TMP18]], i64 [[TMP20]], i32* [[TMP0]], i32** [[TMP1]], i32** [[TMP2]], i32** [[TMP3]]), !llvm.access.group !80 8703 // CHECK9-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] 8704 // CHECK9: omp.inner.for.inc: 8705 // CHECK9-NEXT: [[TMP21:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !80 8706 // CHECK9-NEXT: [[TMP22:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4, !llvm.access.group !80 8707 // CHECK9-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP21]], [[TMP22]] 8708 // CHECK9-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4, !llvm.access.group !80 8709 // CHECK9-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP81:![0-9]+]] 8710 // CHECK9: omp.inner.for.end: 8711 // CHECK9-NEXT: br label [[OMP_LOOP_EXIT:%.*]] 8712 // CHECK9: omp.loop.exit: 8713 // CHECK9-NEXT: [[TMP23:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 8714 // CHECK9-NEXT: [[TMP24:%.*]] = load i32, i32* [[TMP23]], align 4 8715 // CHECK9-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP24]]) 8716 // CHECK9-NEXT: [[TMP25:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4 8717 // CHECK9-NEXT: [[TMP26:%.*]] = icmp ne i32 [[TMP25]], 0 8718 // CHECK9-NEXT: br i1 [[TMP26]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]] 8719 // CHECK9: .omp.final.then: 8720 // CHECK9-NEXT: [[TMP27:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 8721 // CHECK9-NEXT: [[SUB6:%.*]] = sub nsw i32 [[TMP27]], 0 8722 // CHECK9-NEXT: [[DIV7:%.*]] = sdiv i32 [[SUB6]], 1 8723 // CHECK9-NEXT: [[MUL:%.*]] = mul nsw i32 [[DIV7]], 1 8724 // CHECK9-NEXT: [[ADD8:%.*]] = add nsw i32 0, [[MUL]] 8725 // CHECK9-NEXT: store i32 [[ADD8]], i32* [[I3]], align 4 8726 // CHECK9-NEXT: br label [[DOTOMP_FINAL_DONE]] 8727 // CHECK9: .omp.final.done: 8728 // CHECK9-NEXT: br label [[OMP_PRECOND_END]] 8729 // CHECK9: omp.precond.end: 8730 // CHECK9-NEXT: ret void 8731 // 8732 // 8733 // CHECK9-LABEL: define {{[^@]+}}@.omp_outlined..39 8734 // CHECK9-SAME: (i32* noalias noundef [[DOTGLOBAL_TID_:%.*]], i32* noalias noundef [[DOTBOUND_TID_:%.*]], i64 noundef [[DOTPREVIOUS_LB_:%.*]], i64 noundef [[DOTPREVIOUS_UB_:%.*]], i32* noundef nonnull align 4 dereferenceable(4) [[N:%.*]], i32** noundef nonnull align 8 dereferenceable(8) [[A:%.*]], i32** noundef nonnull align 8 dereferenceable(8) [[B:%.*]], i32** noundef nonnull align 8 dereferenceable(8) [[C:%.*]]) #[[ATTR1]] { 8735 // CHECK9-NEXT: entry: 8736 // CHECK9-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 8737 // CHECK9-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 8738 // CHECK9-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8 8739 // CHECK9-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8 8740 // CHECK9-NEXT: [[N_ADDR:%.*]] = alloca i32*, align 8 8741 // CHECK9-NEXT: [[A_ADDR:%.*]] = alloca i32**, align 8 8742 // CHECK9-NEXT: [[B_ADDR:%.*]] = alloca i32**, align 8 8743 // CHECK9-NEXT: [[C_ADDR:%.*]] = alloca i32**, align 8 8744 // CHECK9-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 8745 // CHECK9-NEXT: [[TMP:%.*]] = alloca i32, align 4 8746 // CHECK9-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 8747 // CHECK9-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 8748 // CHECK9-NEXT: [[I:%.*]] = alloca i32, align 4 8749 // CHECK9-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 8750 // CHECK9-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4 8751 // CHECK9-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 8752 // CHECK9-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 8753 // CHECK9-NEXT: [[I4:%.*]] = alloca i32, align 4 8754 // CHECK9-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8 8755 // CHECK9-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8 8756 // CHECK9-NEXT: store i64 [[DOTPREVIOUS_LB_]], i64* [[DOTPREVIOUS_LB__ADDR]], align 8 8757 // CHECK9-NEXT: store i64 [[DOTPREVIOUS_UB_]], i64* [[DOTPREVIOUS_UB__ADDR]], align 8 8758 // CHECK9-NEXT: store i32* [[N]], i32** [[N_ADDR]], align 8 8759 // CHECK9-NEXT: store i32** [[A]], i32*** [[A_ADDR]], align 8 8760 // CHECK9-NEXT: store i32** [[B]], i32*** [[B_ADDR]], align 8 8761 // CHECK9-NEXT: store i32** [[C]], i32*** [[C_ADDR]], align 8 8762 // CHECK9-NEXT: [[TMP0:%.*]] = load i32*, i32** [[N_ADDR]], align 8 8763 // CHECK9-NEXT: [[TMP1:%.*]] = load i32**, i32*** [[A_ADDR]], align 8 8764 // CHECK9-NEXT: [[TMP2:%.*]] = load i32**, i32*** [[B_ADDR]], align 8 8765 // CHECK9-NEXT: [[TMP3:%.*]] = load i32**, i32*** [[C_ADDR]], align 8 8766 // CHECK9-NEXT: [[TMP4:%.*]] = load i32, i32* [[TMP0]], align 4 8767 // CHECK9-NEXT: store i32 [[TMP4]], i32* [[DOTCAPTURE_EXPR_]], align 4 8768 // CHECK9-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 8769 // CHECK9-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP5]], 0 8770 // CHECK9-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 8771 // CHECK9-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1 8772 // CHECK9-NEXT: store i32 [[SUB2]], i32* [[DOTCAPTURE_EXPR_1]], align 4 8773 // CHECK9-NEXT: store i32 0, i32* [[I]], align 4 8774 // CHECK9-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 8775 // CHECK9-NEXT: [[CMP:%.*]] = icmp slt i32 0, [[TMP6]] 8776 // CHECK9-NEXT: br i1 [[CMP]], label [[OMP_PRECOND_THEN:%.*]], label [[OMP_PRECOND_END:%.*]] 8777 // CHECK9: omp.precond.then: 8778 // CHECK9-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4 8779 // CHECK9-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 8780 // CHECK9-NEXT: store i32 [[TMP7]], i32* [[DOTOMP_UB]], align 4 8781 // CHECK9-NEXT: [[TMP8:%.*]] = load i64, i64* [[DOTPREVIOUS_LB__ADDR]], align 8 8782 // CHECK9-NEXT: [[CONV:%.*]] = trunc i64 [[TMP8]] to i32 8783 // CHECK9-NEXT: [[TMP9:%.*]] = load i64, i64* [[DOTPREVIOUS_UB__ADDR]], align 8 8784 // CHECK9-NEXT: [[CONV3:%.*]] = trunc i64 [[TMP9]] to i32 8785 // CHECK9-NEXT: store i32 [[CONV]], i32* [[DOTOMP_LB]], align 4 8786 // CHECK9-NEXT: store i32 [[CONV3]], i32* [[DOTOMP_UB]], align 4 8787 // CHECK9-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4 8788 // CHECK9-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4 8789 // CHECK9-NEXT: [[TMP10:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 8790 // CHECK9-NEXT: [[TMP11:%.*]] = load i32, i32* [[TMP10]], align 4 8791 // CHECK9-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB2]], i32 [[TMP11]], i32 34, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1) 8792 // CHECK9-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 8793 // CHECK9-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 8794 // CHECK9-NEXT: [[CMP5:%.*]] = icmp sgt i32 [[TMP12]], [[TMP13]] 8795 // CHECK9-NEXT: br i1 [[CMP5]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] 8796 // CHECK9: cond.true: 8797 // CHECK9-NEXT: [[TMP14:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 8798 // CHECK9-NEXT: br label [[COND_END:%.*]] 8799 // CHECK9: cond.false: 8800 // CHECK9-NEXT: [[TMP15:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 8801 // CHECK9-NEXT: br label [[COND_END]] 8802 // CHECK9: cond.end: 8803 // CHECK9-NEXT: [[COND:%.*]] = phi i32 [ [[TMP14]], [[COND_TRUE]] ], [ [[TMP15]], [[COND_FALSE]] ] 8804 // CHECK9-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4 8805 // CHECK9-NEXT: [[TMP16:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4 8806 // CHECK9-NEXT: store i32 [[TMP16]], i32* [[DOTOMP_IV]], align 4 8807 // CHECK9-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] 8808 // CHECK9: omp.inner.for.cond: 8809 // CHECK9-NEXT: [[TMP17:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !83 8810 // CHECK9-NEXT: [[TMP18:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4, !llvm.access.group !83 8811 // CHECK9-NEXT: [[CMP6:%.*]] = icmp sle i32 [[TMP17]], [[TMP18]] 8812 // CHECK9-NEXT: br i1 [[CMP6]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] 8813 // CHECK9: omp.inner.for.body: 8814 // CHECK9-NEXT: [[TMP19:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !83 8815 // CHECK9-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP19]], 1 8816 // CHECK9-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] 8817 // CHECK9-NEXT: store i32 [[ADD]], i32* [[I4]], align 4, !llvm.access.group !83 8818 // CHECK9-NEXT: [[TMP20:%.*]] = load i32*, i32** [[TMP2]], align 8, !llvm.access.group !83 8819 // CHECK9-NEXT: [[TMP21:%.*]] = load i32, i32* [[I4]], align 4, !llvm.access.group !83 8820 // CHECK9-NEXT: [[IDXPROM:%.*]] = sext i32 [[TMP21]] to i64 8821 // CHECK9-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i32, i32* [[TMP20]], i64 [[IDXPROM]] 8822 // CHECK9-NEXT: [[TMP22:%.*]] = load i32, i32* [[ARRAYIDX]], align 4, !llvm.access.group !83 8823 // CHECK9-NEXT: [[TMP23:%.*]] = load i32*, i32** [[TMP3]], align 8, !llvm.access.group !83 8824 // CHECK9-NEXT: [[TMP24:%.*]] = load i32, i32* [[I4]], align 4, !llvm.access.group !83 8825 // CHECK9-NEXT: [[IDXPROM7:%.*]] = sext i32 [[TMP24]] to i64 8826 // CHECK9-NEXT: [[ARRAYIDX8:%.*]] = getelementptr inbounds i32, i32* [[TMP23]], i64 [[IDXPROM7]] 8827 // CHECK9-NEXT: [[TMP25:%.*]] = load i32, i32* [[ARRAYIDX8]], align 4, !llvm.access.group !83 8828 // CHECK9-NEXT: [[ADD9:%.*]] = add nsw i32 [[TMP22]], [[TMP25]] 8829 // CHECK9-NEXT: [[TMP26:%.*]] = load i32*, i32** [[TMP1]], align 8, !llvm.access.group !83 8830 // CHECK9-NEXT: [[TMP27:%.*]] = load i32, i32* [[I4]], align 4, !llvm.access.group !83 8831 // CHECK9-NEXT: [[IDXPROM10:%.*]] = sext i32 [[TMP27]] to i64 8832 // CHECK9-NEXT: [[ARRAYIDX11:%.*]] = getelementptr inbounds i32, i32* [[TMP26]], i64 [[IDXPROM10]] 8833 // CHECK9-NEXT: store i32 [[ADD9]], i32* [[ARRAYIDX11]], align 4, !llvm.access.group !83 8834 // CHECK9-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] 8835 // CHECK9: omp.body.continue: 8836 // CHECK9-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] 8837 // CHECK9: omp.inner.for.inc: 8838 // CHECK9-NEXT: [[TMP28:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !83 8839 // CHECK9-NEXT: [[ADD12:%.*]] = add nsw i32 [[TMP28]], 1 8840 // CHECK9-NEXT: store i32 [[ADD12]], i32* [[DOTOMP_IV]], align 4, !llvm.access.group !83 8841 // CHECK9-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP84:![0-9]+]] 8842 // CHECK9: omp.inner.for.end: 8843 // CHECK9-NEXT: br label [[OMP_LOOP_EXIT:%.*]] 8844 // CHECK9: omp.loop.exit: 8845 // CHECK9-NEXT: [[TMP29:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 8846 // CHECK9-NEXT: [[TMP30:%.*]] = load i32, i32* [[TMP29]], align 4 8847 // CHECK9-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP30]]) 8848 // CHECK9-NEXT: [[TMP31:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4 8849 // CHECK9-NEXT: [[TMP32:%.*]] = icmp ne i32 [[TMP31]], 0 8850 // CHECK9-NEXT: br i1 [[TMP32]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]] 8851 // CHECK9: .omp.final.then: 8852 // CHECK9-NEXT: [[TMP33:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 8853 // CHECK9-NEXT: [[SUB13:%.*]] = sub nsw i32 [[TMP33]], 0 8854 // CHECK9-NEXT: [[DIV14:%.*]] = sdiv i32 [[SUB13]], 1 8855 // CHECK9-NEXT: [[MUL15:%.*]] = mul nsw i32 [[DIV14]], 1 8856 // CHECK9-NEXT: [[ADD16:%.*]] = add nsw i32 0, [[MUL15]] 8857 // CHECK9-NEXT: store i32 [[ADD16]], i32* [[I4]], align 4 8858 // CHECK9-NEXT: br label [[DOTOMP_FINAL_DONE]] 8859 // CHECK9: .omp.final.done: 8860 // CHECK9-NEXT: br label [[OMP_PRECOND_END]] 8861 // CHECK9: omp.precond.end: 8862 // CHECK9-NEXT: ret void 8863 // 8864 // 8865 // CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiET_v_l74 8866 // CHECK9-SAME: (i64 noundef [[CH:%.*]], i64 noundef [[N:%.*]], i32* noundef [[A:%.*]], i32* noundef [[B:%.*]], i32* noundef [[C:%.*]]) #[[ATTR1]] { 8867 // CHECK9-NEXT: entry: 8868 // CHECK9-NEXT: [[CH_ADDR:%.*]] = alloca i64, align 8 8869 // CHECK9-NEXT: [[N_ADDR:%.*]] = alloca i64, align 8 8870 // CHECK9-NEXT: [[A_ADDR:%.*]] = alloca i32*, align 8 8871 // CHECK9-NEXT: [[B_ADDR:%.*]] = alloca i32*, align 8 8872 // CHECK9-NEXT: [[C_ADDR:%.*]] = alloca i32*, align 8 8873 // CHECK9-NEXT: store i64 [[CH]], i64* [[CH_ADDR]], align 8 8874 // CHECK9-NEXT: store i64 [[N]], i64* [[N_ADDR]], align 8 8875 // CHECK9-NEXT: store i32* [[A]], i32** [[A_ADDR]], align 8 8876 // CHECK9-NEXT: store i32* [[B]], i32** [[B_ADDR]], align 8 8877 // CHECK9-NEXT: store i32* [[C]], i32** [[C_ADDR]], align 8 8878 // CHECK9-NEXT: [[CONV:%.*]] = bitcast i64* [[CH_ADDR]] to i32* 8879 // CHECK9-NEXT: [[CONV1:%.*]] = bitcast i64* [[N_ADDR]] to i32* 8880 // CHECK9-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 5, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32*, i32*, i32**, i32**, i32**)* @.omp_outlined..42 to void (i32*, i32*, ...)*), i32* [[CONV]], i32* [[CONV1]], i32** [[A_ADDR]], i32** [[B_ADDR]], i32** [[C_ADDR]]) 8881 // CHECK9-NEXT: ret void 8882 // 8883 // 8884 // CHECK9-LABEL: define {{[^@]+}}@.omp_outlined..42 8885 // CHECK9-SAME: (i32* noalias noundef [[DOTGLOBAL_TID_:%.*]], i32* noalias noundef [[DOTBOUND_TID_:%.*]], i32* noundef nonnull align 4 dereferenceable(4) [[CH:%.*]], i32* noundef nonnull align 4 dereferenceable(4) [[N:%.*]], i32** noundef nonnull align 8 dereferenceable(8) [[A:%.*]], i32** noundef nonnull align 8 dereferenceable(8) [[B:%.*]], i32** noundef nonnull align 8 dereferenceable(8) [[C:%.*]]) #[[ATTR1]] { 8886 // CHECK9-NEXT: entry: 8887 // CHECK9-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 8888 // CHECK9-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 8889 // CHECK9-NEXT: [[CH_ADDR:%.*]] = alloca i32*, align 8 8890 // CHECK9-NEXT: [[N_ADDR:%.*]] = alloca i32*, align 8 8891 // CHECK9-NEXT: [[A_ADDR:%.*]] = alloca i32**, align 8 8892 // CHECK9-NEXT: [[B_ADDR:%.*]] = alloca i32**, align 8 8893 // CHECK9-NEXT: [[C_ADDR:%.*]] = alloca i32**, align 8 8894 // CHECK9-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 8895 // CHECK9-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 8896 // CHECK9-NEXT: [[TMP:%.*]] = alloca i32, align 4 8897 // CHECK9-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 8898 // CHECK9-NEXT: [[DOTCAPTURE_EXPR_2:%.*]] = alloca i32, align 4 8899 // CHECK9-NEXT: [[I:%.*]] = alloca i32, align 4 8900 // CHECK9-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4 8901 // CHECK9-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4 8902 // CHECK9-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 8903 // CHECK9-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 8904 // CHECK9-NEXT: [[I4:%.*]] = alloca i32, align 4 8905 // CHECK9-NEXT: [[DOTCAPTURE_EXPR__CASTED:%.*]] = alloca i64, align 8 8906 // CHECK9-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8 8907 // CHECK9-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8 8908 // CHECK9-NEXT: store i32* [[CH]], i32** [[CH_ADDR]], align 8 8909 // CHECK9-NEXT: store i32* [[N]], i32** [[N_ADDR]], align 8 8910 // CHECK9-NEXT: store i32** [[A]], i32*** [[A_ADDR]], align 8 8911 // CHECK9-NEXT: store i32** [[B]], i32*** [[B_ADDR]], align 8 8912 // CHECK9-NEXT: store i32** [[C]], i32*** [[C_ADDR]], align 8 8913 // CHECK9-NEXT: [[TMP0:%.*]] = load i32*, i32** [[CH_ADDR]], align 8 8914 // CHECK9-NEXT: [[TMP1:%.*]] = load i32*, i32** [[N_ADDR]], align 8 8915 // CHECK9-NEXT: [[TMP2:%.*]] = load i32**, i32*** [[A_ADDR]], align 8 8916 // CHECK9-NEXT: [[TMP3:%.*]] = load i32**, i32*** [[B_ADDR]], align 8 8917 // CHECK9-NEXT: [[TMP4:%.*]] = load i32**, i32*** [[C_ADDR]], align 8 8918 // CHECK9-NEXT: [[TMP5:%.*]] = load i32, i32* [[TMP0]], align 4 8919 // CHECK9-NEXT: store i32 [[TMP5]], i32* [[DOTCAPTURE_EXPR_]], align 4 8920 // CHECK9-NEXT: [[TMP6:%.*]] = load i32, i32* [[TMP1]], align 4 8921 // CHECK9-NEXT: store i32 [[TMP6]], i32* [[DOTCAPTURE_EXPR_1]], align 4 8922 // CHECK9-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 8923 // CHECK9-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP7]], 0 8924 // CHECK9-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 8925 // CHECK9-NEXT: [[SUB3:%.*]] = sub nsw i32 [[DIV]], 1 8926 // CHECK9-NEXT: store i32 [[SUB3]], i32* [[DOTCAPTURE_EXPR_2]], align 4 8927 // CHECK9-NEXT: store i32 0, i32* [[I]], align 4 8928 // CHECK9-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 8929 // CHECK9-NEXT: [[CMP:%.*]] = icmp slt i32 0, [[TMP8]] 8930 // CHECK9-NEXT: br i1 [[CMP]], label [[OMP_PRECOND_THEN:%.*]], label [[OMP_PRECOND_END:%.*]] 8931 // CHECK9: omp.precond.then: 8932 // CHECK9-NEXT: store i32 0, i32* [[DOTOMP_COMB_LB]], align 4 8933 // CHECK9-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_2]], align 4 8934 // CHECK9-NEXT: store i32 [[TMP9]], i32* [[DOTOMP_COMB_UB]], align 4 8935 // CHECK9-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4 8936 // CHECK9-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4 8937 // CHECK9-NEXT: [[TMP10:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 8938 // CHECK9-NEXT: [[TMP11:%.*]] = load i32, i32* [[TMP10]], align 4 8939 // CHECK9-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1]], i32 [[TMP11]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_COMB_LB]], i32* [[DOTOMP_COMB_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1) 8940 // CHECK9-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 8941 // CHECK9-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_2]], align 4 8942 // CHECK9-NEXT: [[CMP5:%.*]] = icmp sgt i32 [[TMP12]], [[TMP13]] 8943 // CHECK9-NEXT: br i1 [[CMP5]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] 8944 // CHECK9: cond.true: 8945 // CHECK9-NEXT: [[TMP14:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_2]], align 4 8946 // CHECK9-NEXT: br label [[COND_END:%.*]] 8947 // CHECK9: cond.false: 8948 // CHECK9-NEXT: [[TMP15:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 8949 // CHECK9-NEXT: br label [[COND_END]] 8950 // CHECK9: cond.end: 8951 // CHECK9-NEXT: [[COND:%.*]] = phi i32 [ [[TMP14]], [[COND_TRUE]] ], [ [[TMP15]], [[COND_FALSE]] ] 8952 // CHECK9-NEXT: store i32 [[COND]], i32* [[DOTOMP_COMB_UB]], align 4 8953 // CHECK9-NEXT: [[TMP16:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4 8954 // CHECK9-NEXT: store i32 [[TMP16]], i32* [[DOTOMP_IV]], align 4 8955 // CHECK9-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] 8956 // CHECK9: omp.inner.for.cond: 8957 // CHECK9-NEXT: [[TMP17:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !86 8958 // CHECK9-NEXT: [[TMP18:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4, !llvm.access.group !86 8959 // CHECK9-NEXT: [[CMP6:%.*]] = icmp sle i32 [[TMP17]], [[TMP18]] 8960 // CHECK9-NEXT: br i1 [[CMP6]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] 8961 // CHECK9: omp.inner.for.body: 8962 // CHECK9-NEXT: [[TMP19:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4, !llvm.access.group !86 8963 // CHECK9-NEXT: [[TMP20:%.*]] = zext i32 [[TMP19]] to i64 8964 // CHECK9-NEXT: [[TMP21:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4, !llvm.access.group !86 8965 // CHECK9-NEXT: [[TMP22:%.*]] = zext i32 [[TMP21]] to i64 8966 // CHECK9-NEXT: [[TMP23:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4, !llvm.access.group !86 8967 // CHECK9-NEXT: [[CONV:%.*]] = bitcast i64* [[DOTCAPTURE_EXPR__CASTED]] to i32* 8968 // CHECK9-NEXT: store i32 [[TMP23]], i32* [[CONV]], align 4, !llvm.access.group !86 8969 // CHECK9-NEXT: [[TMP24:%.*]] = load i64, i64* [[DOTCAPTURE_EXPR__CASTED]], align 8, !llvm.access.group !86 8970 // CHECK9-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 7, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64, i32*, i32**, i32**, i32**, i64)* @.omp_outlined..43 to void (i32*, i32*, ...)*), i64 [[TMP20]], i64 [[TMP22]], i32* [[TMP1]], i32** [[TMP2]], i32** [[TMP3]], i32** [[TMP4]], i64 [[TMP24]]), !llvm.access.group !86 8971 // CHECK9-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] 8972 // CHECK9: omp.inner.for.inc: 8973 // CHECK9-NEXT: [[TMP25:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !86 8974 // CHECK9-NEXT: [[TMP26:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4, !llvm.access.group !86 8975 // CHECK9-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP25]], [[TMP26]] 8976 // CHECK9-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4, !llvm.access.group !86 8977 // CHECK9-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP87:![0-9]+]] 8978 // CHECK9: omp.inner.for.end: 8979 // CHECK9-NEXT: br label [[OMP_LOOP_EXIT:%.*]] 8980 // CHECK9: omp.loop.exit: 8981 // CHECK9-NEXT: [[TMP27:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 8982 // CHECK9-NEXT: [[TMP28:%.*]] = load i32, i32* [[TMP27]], align 4 8983 // CHECK9-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP28]]) 8984 // CHECK9-NEXT: [[TMP29:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4 8985 // CHECK9-NEXT: [[TMP30:%.*]] = icmp ne i32 [[TMP29]], 0 8986 // CHECK9-NEXT: br i1 [[TMP30]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]] 8987 // CHECK9: .omp.final.then: 8988 // CHECK9-NEXT: [[TMP31:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 8989 // CHECK9-NEXT: [[SUB7:%.*]] = sub nsw i32 [[TMP31]], 0 8990 // CHECK9-NEXT: [[DIV8:%.*]] = sdiv i32 [[SUB7]], 1 8991 // CHECK9-NEXT: [[MUL:%.*]] = mul nsw i32 [[DIV8]], 1 8992 // CHECK9-NEXT: [[ADD9:%.*]] = add nsw i32 0, [[MUL]] 8993 // CHECK9-NEXT: store i32 [[ADD9]], i32* [[I4]], align 4 8994 // CHECK9-NEXT: br label [[DOTOMP_FINAL_DONE]] 8995 // CHECK9: .omp.final.done: 8996 // CHECK9-NEXT: br label [[OMP_PRECOND_END]] 8997 // CHECK9: omp.precond.end: 8998 // CHECK9-NEXT: ret void 8999 // 9000 // 9001 // CHECK9-LABEL: define {{[^@]+}}@.omp_outlined..43 9002 // CHECK9-SAME: (i32* noalias noundef [[DOTGLOBAL_TID_:%.*]], i32* noalias noundef [[DOTBOUND_TID_:%.*]], i64 noundef [[DOTPREVIOUS_LB_:%.*]], i64 noundef [[DOTPREVIOUS_UB_:%.*]], i32* noundef nonnull align 4 dereferenceable(4) [[N:%.*]], i32** noundef nonnull align 8 dereferenceable(8) [[A:%.*]], i32** noundef nonnull align 8 dereferenceable(8) [[B:%.*]], i32** noundef nonnull align 8 dereferenceable(8) [[C:%.*]], i64 noundef [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR1]] { 9003 // CHECK9-NEXT: entry: 9004 // CHECK9-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 9005 // CHECK9-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 9006 // CHECK9-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8 9007 // CHECK9-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8 9008 // CHECK9-NEXT: [[N_ADDR:%.*]] = alloca i32*, align 8 9009 // CHECK9-NEXT: [[A_ADDR:%.*]] = alloca i32**, align 8 9010 // CHECK9-NEXT: [[B_ADDR:%.*]] = alloca i32**, align 8 9011 // CHECK9-NEXT: [[C_ADDR:%.*]] = alloca i32**, align 8 9012 // CHECK9-NEXT: [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i64, align 8 9013 // CHECK9-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 9014 // CHECK9-NEXT: [[TMP:%.*]] = alloca i32, align 4 9015 // CHECK9-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 9016 // CHECK9-NEXT: [[DOTCAPTURE_EXPR_2:%.*]] = alloca i32, align 4 9017 // CHECK9-NEXT: [[I:%.*]] = alloca i32, align 4 9018 // CHECK9-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 9019 // CHECK9-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4 9020 // CHECK9-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 9021 // CHECK9-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 9022 // CHECK9-NEXT: [[I6:%.*]] = alloca i32, align 4 9023 // CHECK9-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8 9024 // CHECK9-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8 9025 // CHECK9-NEXT: store i64 [[DOTPREVIOUS_LB_]], i64* [[DOTPREVIOUS_LB__ADDR]], align 8 9026 // CHECK9-NEXT: store i64 [[DOTPREVIOUS_UB_]], i64* [[DOTPREVIOUS_UB__ADDR]], align 8 9027 // CHECK9-NEXT: store i32* [[N]], i32** [[N_ADDR]], align 8 9028 // CHECK9-NEXT: store i32** [[A]], i32*** [[A_ADDR]], align 8 9029 // CHECK9-NEXT: store i32** [[B]], i32*** [[B_ADDR]], align 8 9030 // CHECK9-NEXT: store i32** [[C]], i32*** [[C_ADDR]], align 8 9031 // CHECK9-NEXT: store i64 [[DOTCAPTURE_EXPR_]], i64* [[DOTCAPTURE_EXPR__ADDR]], align 8 9032 // CHECK9-NEXT: [[TMP0:%.*]] = load i32*, i32** [[N_ADDR]], align 8 9033 // CHECK9-NEXT: [[TMP1:%.*]] = load i32**, i32*** [[A_ADDR]], align 8 9034 // CHECK9-NEXT: [[TMP2:%.*]] = load i32**, i32*** [[B_ADDR]], align 8 9035 // CHECK9-NEXT: [[TMP3:%.*]] = load i32**, i32*** [[C_ADDR]], align 8 9036 // CHECK9-NEXT: [[CONV:%.*]] = bitcast i64* [[DOTCAPTURE_EXPR__ADDR]] to i32* 9037 // CHECK9-NEXT: [[TMP4:%.*]] = load i32, i32* [[TMP0]], align 4 9038 // CHECK9-NEXT: store i32 [[TMP4]], i32* [[DOTCAPTURE_EXPR_1]], align 4 9039 // CHECK9-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 9040 // CHECK9-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP5]], 0 9041 // CHECK9-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 9042 // CHECK9-NEXT: [[SUB3:%.*]] = sub nsw i32 [[DIV]], 1 9043 // CHECK9-NEXT: store i32 [[SUB3]], i32* [[DOTCAPTURE_EXPR_2]], align 4 9044 // CHECK9-NEXT: store i32 0, i32* [[I]], align 4 9045 // CHECK9-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 9046 // CHECK9-NEXT: [[CMP:%.*]] = icmp slt i32 0, [[TMP6]] 9047 // CHECK9-NEXT: br i1 [[CMP]], label [[OMP_PRECOND_THEN:%.*]], label [[OMP_PRECOND_END:%.*]] 9048 // CHECK9: omp.precond.then: 9049 // CHECK9-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4 9050 // CHECK9-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_2]], align 4 9051 // CHECK9-NEXT: store i32 [[TMP7]], i32* [[DOTOMP_UB]], align 4 9052 // CHECK9-NEXT: [[TMP8:%.*]] = load i64, i64* [[DOTPREVIOUS_LB__ADDR]], align 8 9053 // CHECK9-NEXT: [[CONV4:%.*]] = trunc i64 [[TMP8]] to i32 9054 // CHECK9-NEXT: [[TMP9:%.*]] = load i64, i64* [[DOTPREVIOUS_UB__ADDR]], align 8 9055 // CHECK9-NEXT: [[CONV5:%.*]] = trunc i64 [[TMP9]] to i32 9056 // CHECK9-NEXT: store i32 [[CONV4]], i32* [[DOTOMP_LB]], align 4 9057 // CHECK9-NEXT: store i32 [[CONV5]], i32* [[DOTOMP_UB]], align 4 9058 // CHECK9-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4 9059 // CHECK9-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4 9060 // CHECK9-NEXT: [[TMP10:%.*]] = load i32, i32* [[CONV]], align 4 9061 // CHECK9-NEXT: [[TMP11:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 9062 // CHECK9-NEXT: [[TMP12:%.*]] = load i32, i32* [[TMP11]], align 4 9063 // CHECK9-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB2]], i32 [[TMP12]], i32 33, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 [[TMP10]]) 9064 // CHECK9-NEXT: br label [[OMP_DISPATCH_COND:%.*]] 9065 // CHECK9: omp.dispatch.cond: 9066 // CHECK9-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 9067 // CHECK9-NEXT: [[TMP14:%.*]] = load i64, i64* [[DOTPREVIOUS_UB__ADDR]], align 8 9068 // CHECK9-NEXT: [[CONV7:%.*]] = trunc i64 [[TMP14]] to i32 9069 // CHECK9-NEXT: [[CMP8:%.*]] = icmp sgt i32 [[TMP13]], [[CONV7]] 9070 // CHECK9-NEXT: br i1 [[CMP8]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] 9071 // CHECK9: cond.true: 9072 // CHECK9-NEXT: [[TMP15:%.*]] = load i64, i64* [[DOTPREVIOUS_UB__ADDR]], align 8 9073 // CHECK9-NEXT: [[CONV9:%.*]] = trunc i64 [[TMP15]] to i32 9074 // CHECK9-NEXT: br label [[COND_END:%.*]] 9075 // CHECK9: cond.false: 9076 // CHECK9-NEXT: [[TMP16:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 9077 // CHECK9-NEXT: br label [[COND_END]] 9078 // CHECK9: cond.end: 9079 // CHECK9-NEXT: [[COND:%.*]] = phi i32 [ [[CONV9]], [[COND_TRUE]] ], [ [[TMP16]], [[COND_FALSE]] ] 9080 // CHECK9-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4 9081 // CHECK9-NEXT: [[TMP17:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4 9082 // CHECK9-NEXT: store i32 [[TMP17]], i32* [[DOTOMP_IV]], align 4 9083 // CHECK9-NEXT: [[TMP18:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 9084 // CHECK9-NEXT: [[TMP19:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 9085 // CHECK9-NEXT: [[CMP10:%.*]] = icmp sle i32 [[TMP18]], [[TMP19]] 9086 // CHECK9-NEXT: br i1 [[CMP10]], label [[OMP_DISPATCH_BODY:%.*]], label [[OMP_DISPATCH_END:%.*]] 9087 // CHECK9: omp.dispatch.body: 9088 // CHECK9-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] 9089 // CHECK9: omp.inner.for.cond: 9090 // CHECK9-NEXT: [[TMP20:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !89 9091 // CHECK9-NEXT: [[TMP21:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4, !llvm.access.group !89 9092 // CHECK9-NEXT: [[CMP11:%.*]] = icmp sle i32 [[TMP20]], [[TMP21]] 9093 // CHECK9-NEXT: br i1 [[CMP11]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] 9094 // CHECK9: omp.inner.for.body: 9095 // CHECK9-NEXT: [[TMP22:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !89 9096 // CHECK9-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP22]], 1 9097 // CHECK9-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] 9098 // CHECK9-NEXT: store i32 [[ADD]], i32* [[I6]], align 4, !llvm.access.group !89 9099 // CHECK9-NEXT: [[TMP23:%.*]] = load i32*, i32** [[TMP2]], align 8, !llvm.access.group !89 9100 // CHECK9-NEXT: [[TMP24:%.*]] = load i32, i32* [[I6]], align 4, !llvm.access.group !89 9101 // CHECK9-NEXT: [[IDXPROM:%.*]] = sext i32 [[TMP24]] to i64 9102 // CHECK9-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i32, i32* [[TMP23]], i64 [[IDXPROM]] 9103 // CHECK9-NEXT: [[TMP25:%.*]] = load i32, i32* [[ARRAYIDX]], align 4, !llvm.access.group !89 9104 // CHECK9-NEXT: [[TMP26:%.*]] = load i32*, i32** [[TMP3]], align 8, !llvm.access.group !89 9105 // CHECK9-NEXT: [[TMP27:%.*]] = load i32, i32* [[I6]], align 4, !llvm.access.group !89 9106 // CHECK9-NEXT: [[IDXPROM12:%.*]] = sext i32 [[TMP27]] to i64 9107 // CHECK9-NEXT: [[ARRAYIDX13:%.*]] = getelementptr inbounds i32, i32* [[TMP26]], i64 [[IDXPROM12]] 9108 // CHECK9-NEXT: [[TMP28:%.*]] = load i32, i32* [[ARRAYIDX13]], align 4, !llvm.access.group !89 9109 // CHECK9-NEXT: [[ADD14:%.*]] = add nsw i32 [[TMP25]], [[TMP28]] 9110 // CHECK9-NEXT: [[TMP29:%.*]] = load i32*, i32** [[TMP1]], align 8, !llvm.access.group !89 9111 // CHECK9-NEXT: [[TMP30:%.*]] = load i32, i32* [[I6]], align 4, !llvm.access.group !89 9112 // CHECK9-NEXT: [[IDXPROM15:%.*]] = sext i32 [[TMP30]] to i64 9113 // CHECK9-NEXT: [[ARRAYIDX16:%.*]] = getelementptr inbounds i32, i32* [[TMP29]], i64 [[IDXPROM15]] 9114 // CHECK9-NEXT: store i32 [[ADD14]], i32* [[ARRAYIDX16]], align 4, !llvm.access.group !89 9115 // CHECK9-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] 9116 // CHECK9: omp.body.continue: 9117 // CHECK9-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] 9118 // CHECK9: omp.inner.for.inc: 9119 // CHECK9-NEXT: [[TMP31:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !89 9120 // CHECK9-NEXT: [[ADD17:%.*]] = add nsw i32 [[TMP31]], 1 9121 // CHECK9-NEXT: store i32 [[ADD17]], i32* [[DOTOMP_IV]], align 4, !llvm.access.group !89 9122 // CHECK9-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP90:![0-9]+]] 9123 // CHECK9: omp.inner.for.end: 9124 // CHECK9-NEXT: br label [[OMP_DISPATCH_INC:%.*]] 9125 // CHECK9: omp.dispatch.inc: 9126 // CHECK9-NEXT: [[TMP32:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4 9127 // CHECK9-NEXT: [[TMP33:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4 9128 // CHECK9-NEXT: [[ADD18:%.*]] = add nsw i32 [[TMP32]], [[TMP33]] 9129 // CHECK9-NEXT: store i32 [[ADD18]], i32* [[DOTOMP_LB]], align 4 9130 // CHECK9-NEXT: [[TMP34:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 9131 // CHECK9-NEXT: [[TMP35:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4 9132 // CHECK9-NEXT: [[ADD19:%.*]] = add nsw i32 [[TMP34]], [[TMP35]] 9133 // CHECK9-NEXT: store i32 [[ADD19]], i32* [[DOTOMP_UB]], align 4 9134 // CHECK9-NEXT: br label [[OMP_DISPATCH_COND]] 9135 // CHECK9: omp.dispatch.end: 9136 // CHECK9-NEXT: [[TMP36:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 9137 // CHECK9-NEXT: [[TMP37:%.*]] = load i32, i32* [[TMP36]], align 4 9138 // CHECK9-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP37]]) 9139 // CHECK9-NEXT: [[TMP38:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4 9140 // CHECK9-NEXT: [[TMP39:%.*]] = icmp ne i32 [[TMP38]], 0 9141 // CHECK9-NEXT: br i1 [[TMP39]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]] 9142 // CHECK9: .omp.final.then: 9143 // CHECK9-NEXT: [[TMP40:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 9144 // CHECK9-NEXT: [[SUB20:%.*]] = sub nsw i32 [[TMP40]], 0 9145 // CHECK9-NEXT: [[DIV21:%.*]] = sdiv i32 [[SUB20]], 1 9146 // CHECK9-NEXT: [[MUL22:%.*]] = mul nsw i32 [[DIV21]], 1 9147 // CHECK9-NEXT: [[ADD23:%.*]] = add nsw i32 0, [[MUL22]] 9148 // CHECK9-NEXT: store i32 [[ADD23]], i32* [[I6]], align 4 9149 // CHECK9-NEXT: br label [[DOTOMP_FINAL_DONE]] 9150 // CHECK9: .omp.final.done: 9151 // CHECK9-NEXT: br label [[OMP_PRECOND_END]] 9152 // CHECK9: omp.precond.end: 9153 // CHECK9-NEXT: ret void 9154 // 9155 // 9156 // CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiET_v_l82 9157 // CHECK9-SAME: (i64 noundef [[N:%.*]], i32* noundef [[A:%.*]], i32* noundef [[B:%.*]], i32* noundef [[C:%.*]]) #[[ATTR1]] { 9158 // CHECK9-NEXT: entry: 9159 // CHECK9-NEXT: [[N_ADDR:%.*]] = alloca i64, align 8 9160 // CHECK9-NEXT: [[A_ADDR:%.*]] = alloca i32*, align 8 9161 // CHECK9-NEXT: [[B_ADDR:%.*]] = alloca i32*, align 8 9162 // CHECK9-NEXT: [[C_ADDR:%.*]] = alloca i32*, align 8 9163 // CHECK9-NEXT: store i64 [[N]], i64* [[N_ADDR]], align 8 9164 // CHECK9-NEXT: store i32* [[A]], i32** [[A_ADDR]], align 8 9165 // CHECK9-NEXT: store i32* [[B]], i32** [[B_ADDR]], align 8 9166 // CHECK9-NEXT: store i32* [[C]], i32** [[C_ADDR]], align 8 9167 // CHECK9-NEXT: [[CONV:%.*]] = bitcast i64* [[N_ADDR]] to i32* 9168 // CHECK9-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 4, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32*, i32**, i32**, i32**)* @.omp_outlined..46 to void (i32*, i32*, ...)*), i32* [[CONV]], i32** [[A_ADDR]], i32** [[B_ADDR]], i32** [[C_ADDR]]) 9169 // CHECK9-NEXT: ret void 9170 // 9171 // 9172 // CHECK9-LABEL: define {{[^@]+}}@.omp_outlined..46 9173 // CHECK9-SAME: (i32* noalias noundef [[DOTGLOBAL_TID_:%.*]], i32* noalias noundef [[DOTBOUND_TID_:%.*]], i32* noundef nonnull align 4 dereferenceable(4) [[N:%.*]], i32** noundef nonnull align 8 dereferenceable(8) [[A:%.*]], i32** noundef nonnull align 8 dereferenceable(8) [[B:%.*]], i32** noundef nonnull align 8 dereferenceable(8) [[C:%.*]]) #[[ATTR1]] { 9174 // CHECK9-NEXT: entry: 9175 // CHECK9-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 9176 // CHECK9-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 9177 // CHECK9-NEXT: [[N_ADDR:%.*]] = alloca i32*, align 8 9178 // CHECK9-NEXT: [[A_ADDR:%.*]] = alloca i32**, align 8 9179 // CHECK9-NEXT: [[B_ADDR:%.*]] = alloca i32**, align 8 9180 // CHECK9-NEXT: [[C_ADDR:%.*]] = alloca i32**, align 8 9181 // CHECK9-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 9182 // CHECK9-NEXT: [[TMP:%.*]] = alloca i32, align 4 9183 // CHECK9-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 9184 // CHECK9-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 9185 // CHECK9-NEXT: [[I:%.*]] = alloca i32, align 4 9186 // CHECK9-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4 9187 // CHECK9-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4 9188 // CHECK9-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 9189 // CHECK9-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 9190 // CHECK9-NEXT: [[I3:%.*]] = alloca i32, align 4 9191 // CHECK9-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8 9192 // CHECK9-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8 9193 // CHECK9-NEXT: store i32* [[N]], i32** [[N_ADDR]], align 8 9194 // CHECK9-NEXT: store i32** [[A]], i32*** [[A_ADDR]], align 8 9195 // CHECK9-NEXT: store i32** [[B]], i32*** [[B_ADDR]], align 8 9196 // CHECK9-NEXT: store i32** [[C]], i32*** [[C_ADDR]], align 8 9197 // CHECK9-NEXT: [[TMP0:%.*]] = load i32*, i32** [[N_ADDR]], align 8 9198 // CHECK9-NEXT: [[TMP1:%.*]] = load i32**, i32*** [[A_ADDR]], align 8 9199 // CHECK9-NEXT: [[TMP2:%.*]] = load i32**, i32*** [[B_ADDR]], align 8 9200 // CHECK9-NEXT: [[TMP3:%.*]] = load i32**, i32*** [[C_ADDR]], align 8 9201 // CHECK9-NEXT: [[TMP4:%.*]] = load i32, i32* [[TMP0]], align 4 9202 // CHECK9-NEXT: store i32 [[TMP4]], i32* [[DOTCAPTURE_EXPR_]], align 4 9203 // CHECK9-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 9204 // CHECK9-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP5]], 0 9205 // CHECK9-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 9206 // CHECK9-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1 9207 // CHECK9-NEXT: store i32 [[SUB2]], i32* [[DOTCAPTURE_EXPR_1]], align 4 9208 // CHECK9-NEXT: store i32 0, i32* [[I]], align 4 9209 // CHECK9-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 9210 // CHECK9-NEXT: [[CMP:%.*]] = icmp slt i32 0, [[TMP6]] 9211 // CHECK9-NEXT: br i1 [[CMP]], label [[OMP_PRECOND_THEN:%.*]], label [[OMP_PRECOND_END:%.*]] 9212 // CHECK9: omp.precond.then: 9213 // CHECK9-NEXT: store i32 0, i32* [[DOTOMP_COMB_LB]], align 4 9214 // CHECK9-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 9215 // CHECK9-NEXT: store i32 [[TMP7]], i32* [[DOTOMP_COMB_UB]], align 4 9216 // CHECK9-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4 9217 // CHECK9-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4 9218 // CHECK9-NEXT: [[TMP8:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 9219 // CHECK9-NEXT: [[TMP9:%.*]] = load i32, i32* [[TMP8]], align 4 9220 // CHECK9-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1]], i32 [[TMP9]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_COMB_LB]], i32* [[DOTOMP_COMB_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1) 9221 // CHECK9-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 9222 // CHECK9-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 9223 // CHECK9-NEXT: [[CMP4:%.*]] = icmp sgt i32 [[TMP10]], [[TMP11]] 9224 // CHECK9-NEXT: br i1 [[CMP4]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] 9225 // CHECK9: cond.true: 9226 // CHECK9-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 9227 // CHECK9-NEXT: br label [[COND_END:%.*]] 9228 // CHECK9: cond.false: 9229 // CHECK9-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 9230 // CHECK9-NEXT: br label [[COND_END]] 9231 // CHECK9: cond.end: 9232 // CHECK9-NEXT: [[COND:%.*]] = phi i32 [ [[TMP12]], [[COND_TRUE]] ], [ [[TMP13]], [[COND_FALSE]] ] 9233 // CHECK9-NEXT: store i32 [[COND]], i32* [[DOTOMP_COMB_UB]], align 4 9234 // CHECK9-NEXT: [[TMP14:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4 9235 // CHECK9-NEXT: store i32 [[TMP14]], i32* [[DOTOMP_IV]], align 4 9236 // CHECK9-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] 9237 // CHECK9: omp.inner.for.cond: 9238 // CHECK9-NEXT: [[TMP15:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !92 9239 // CHECK9-NEXT: [[TMP16:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4, !llvm.access.group !92 9240 // CHECK9-NEXT: [[CMP5:%.*]] = icmp sle i32 [[TMP15]], [[TMP16]] 9241 // CHECK9-NEXT: br i1 [[CMP5]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] 9242 // CHECK9: omp.inner.for.body: 9243 // CHECK9-NEXT: [[TMP17:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4, !llvm.access.group !92 9244 // CHECK9-NEXT: [[TMP18:%.*]] = zext i32 [[TMP17]] to i64 9245 // CHECK9-NEXT: [[TMP19:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4, !llvm.access.group !92 9246 // CHECK9-NEXT: [[TMP20:%.*]] = zext i32 [[TMP19]] to i64 9247 // CHECK9-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 6, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64, i32*, i32**, i32**, i32**)* @.omp_outlined..47 to void (i32*, i32*, ...)*), i64 [[TMP18]], i64 [[TMP20]], i32* [[TMP0]], i32** [[TMP1]], i32** [[TMP2]], i32** [[TMP3]]), !llvm.access.group !92 9248 // CHECK9-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] 9249 // CHECK9: omp.inner.for.inc: 9250 // CHECK9-NEXT: [[TMP21:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !92 9251 // CHECK9-NEXT: [[TMP22:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4, !llvm.access.group !92 9252 // CHECK9-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP21]], [[TMP22]] 9253 // CHECK9-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4, !llvm.access.group !92 9254 // CHECK9-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP93:![0-9]+]] 9255 // CHECK9: omp.inner.for.end: 9256 // CHECK9-NEXT: br label [[OMP_LOOP_EXIT:%.*]] 9257 // CHECK9: omp.loop.exit: 9258 // CHECK9-NEXT: [[TMP23:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 9259 // CHECK9-NEXT: [[TMP24:%.*]] = load i32, i32* [[TMP23]], align 4 9260 // CHECK9-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP24]]) 9261 // CHECK9-NEXT: [[TMP25:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4 9262 // CHECK9-NEXT: [[TMP26:%.*]] = icmp ne i32 [[TMP25]], 0 9263 // CHECK9-NEXT: br i1 [[TMP26]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]] 9264 // CHECK9: .omp.final.then: 9265 // CHECK9-NEXT: [[TMP27:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 9266 // CHECK9-NEXT: [[SUB6:%.*]] = sub nsw i32 [[TMP27]], 0 9267 // CHECK9-NEXT: [[DIV7:%.*]] = sdiv i32 [[SUB6]], 1 9268 // CHECK9-NEXT: [[MUL:%.*]] = mul nsw i32 [[DIV7]], 1 9269 // CHECK9-NEXT: [[ADD8:%.*]] = add nsw i32 0, [[MUL]] 9270 // CHECK9-NEXT: store i32 [[ADD8]], i32* [[I3]], align 4 9271 // CHECK9-NEXT: br label [[DOTOMP_FINAL_DONE]] 9272 // CHECK9: .omp.final.done: 9273 // CHECK9-NEXT: br label [[OMP_PRECOND_END]] 9274 // CHECK9: omp.precond.end: 9275 // CHECK9-NEXT: ret void 9276 // 9277 // 9278 // CHECK9-LABEL: define {{[^@]+}}@.omp_outlined..47 9279 // CHECK9-SAME: (i32* noalias noundef [[DOTGLOBAL_TID_:%.*]], i32* noalias noundef [[DOTBOUND_TID_:%.*]], i64 noundef [[DOTPREVIOUS_LB_:%.*]], i64 noundef [[DOTPREVIOUS_UB_:%.*]], i32* noundef nonnull align 4 dereferenceable(4) [[N:%.*]], i32** noundef nonnull align 8 dereferenceable(8) [[A:%.*]], i32** noundef nonnull align 8 dereferenceable(8) [[B:%.*]], i32** noundef nonnull align 8 dereferenceable(8) [[C:%.*]]) #[[ATTR1]] { 9280 // CHECK9-NEXT: entry: 9281 // CHECK9-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 9282 // CHECK9-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 9283 // CHECK9-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8 9284 // CHECK9-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8 9285 // CHECK9-NEXT: [[N_ADDR:%.*]] = alloca i32*, align 8 9286 // CHECK9-NEXT: [[A_ADDR:%.*]] = alloca i32**, align 8 9287 // CHECK9-NEXT: [[B_ADDR:%.*]] = alloca i32**, align 8 9288 // CHECK9-NEXT: [[C_ADDR:%.*]] = alloca i32**, align 8 9289 // CHECK9-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 9290 // CHECK9-NEXT: [[TMP:%.*]] = alloca i32, align 4 9291 // CHECK9-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 9292 // CHECK9-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 9293 // CHECK9-NEXT: [[I:%.*]] = alloca i32, align 4 9294 // CHECK9-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 9295 // CHECK9-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4 9296 // CHECK9-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 9297 // CHECK9-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 9298 // CHECK9-NEXT: [[I4:%.*]] = alloca i32, align 4 9299 // CHECK9-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8 9300 // CHECK9-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8 9301 // CHECK9-NEXT: store i64 [[DOTPREVIOUS_LB_]], i64* [[DOTPREVIOUS_LB__ADDR]], align 8 9302 // CHECK9-NEXT: store i64 [[DOTPREVIOUS_UB_]], i64* [[DOTPREVIOUS_UB__ADDR]], align 8 9303 // CHECK9-NEXT: store i32* [[N]], i32** [[N_ADDR]], align 8 9304 // CHECK9-NEXT: store i32** [[A]], i32*** [[A_ADDR]], align 8 9305 // CHECK9-NEXT: store i32** [[B]], i32*** [[B_ADDR]], align 8 9306 // CHECK9-NEXT: store i32** [[C]], i32*** [[C_ADDR]], align 8 9307 // CHECK9-NEXT: [[TMP0:%.*]] = load i32*, i32** [[N_ADDR]], align 8 9308 // CHECK9-NEXT: [[TMP1:%.*]] = load i32**, i32*** [[A_ADDR]], align 8 9309 // CHECK9-NEXT: [[TMP2:%.*]] = load i32**, i32*** [[B_ADDR]], align 8 9310 // CHECK9-NEXT: [[TMP3:%.*]] = load i32**, i32*** [[C_ADDR]], align 8 9311 // CHECK9-NEXT: [[TMP4:%.*]] = load i32, i32* [[TMP0]], align 4 9312 // CHECK9-NEXT: store i32 [[TMP4]], i32* [[DOTCAPTURE_EXPR_]], align 4 9313 // CHECK9-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 9314 // CHECK9-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP5]], 0 9315 // CHECK9-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 9316 // CHECK9-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1 9317 // CHECK9-NEXT: store i32 [[SUB2]], i32* [[DOTCAPTURE_EXPR_1]], align 4 9318 // CHECK9-NEXT: store i32 0, i32* [[I]], align 4 9319 // CHECK9-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 9320 // CHECK9-NEXT: [[CMP:%.*]] = icmp slt i32 0, [[TMP6]] 9321 // CHECK9-NEXT: br i1 [[CMP]], label [[OMP_PRECOND_THEN:%.*]], label [[OMP_PRECOND_END:%.*]] 9322 // CHECK9: omp.precond.then: 9323 // CHECK9-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4 9324 // CHECK9-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 9325 // CHECK9-NEXT: store i32 [[TMP7]], i32* [[DOTOMP_UB]], align 4 9326 // CHECK9-NEXT: [[TMP8:%.*]] = load i64, i64* [[DOTPREVIOUS_LB__ADDR]], align 8 9327 // CHECK9-NEXT: [[CONV:%.*]] = trunc i64 [[TMP8]] to i32 9328 // CHECK9-NEXT: [[TMP9:%.*]] = load i64, i64* [[DOTPREVIOUS_UB__ADDR]], align 8 9329 // CHECK9-NEXT: [[CONV3:%.*]] = trunc i64 [[TMP9]] to i32 9330 // CHECK9-NEXT: store i32 [[CONV]], i32* [[DOTOMP_LB]], align 4 9331 // CHECK9-NEXT: store i32 [[CONV3]], i32* [[DOTOMP_UB]], align 4 9332 // CHECK9-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4 9333 // CHECK9-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4 9334 // CHECK9-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4 9335 // CHECK9-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 9336 // CHECK9-NEXT: [[TMP12:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 9337 // CHECK9-NEXT: [[TMP13:%.*]] = load i32, i32* [[TMP12]], align 4 9338 // CHECK9-NEXT: call void @__kmpc_dispatch_init_4(%struct.ident_t* @[[GLOB3]], i32 [[TMP13]], i32 35, i32 [[TMP10]], i32 [[TMP11]], i32 1, i32 1) 9339 // CHECK9-NEXT: br label [[OMP_DISPATCH_COND:%.*]] 9340 // CHECK9: omp.dispatch.cond: 9341 // CHECK9-NEXT: [[TMP14:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 9342 // CHECK9-NEXT: [[TMP15:%.*]] = load i32, i32* [[TMP14]], align 4 9343 // CHECK9-NEXT: [[TMP16:%.*]] = call i32 @__kmpc_dispatch_next_4(%struct.ident_t* @[[GLOB3]], i32 [[TMP15]], i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]]) 9344 // CHECK9-NEXT: [[TOBOOL:%.*]] = icmp ne i32 [[TMP16]], 0 9345 // CHECK9-NEXT: br i1 [[TOBOOL]], label [[OMP_DISPATCH_BODY:%.*]], label [[OMP_DISPATCH_END:%.*]] 9346 // CHECK9: omp.dispatch.body: 9347 // CHECK9-NEXT: [[TMP17:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4 9348 // CHECK9-NEXT: store i32 [[TMP17]], i32* [[DOTOMP_IV]], align 4 9349 // CHECK9-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] 9350 // CHECK9: omp.inner.for.cond: 9351 // CHECK9-NEXT: [[TMP18:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !95 9352 // CHECK9-NEXT: [[TMP19:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4, !llvm.access.group !95 9353 // CHECK9-NEXT: [[CMP5:%.*]] = icmp sle i32 [[TMP18]], [[TMP19]] 9354 // CHECK9-NEXT: br i1 [[CMP5]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] 9355 // CHECK9: omp.inner.for.body: 9356 // CHECK9-NEXT: [[TMP20:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !95 9357 // CHECK9-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP20]], 1 9358 // CHECK9-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] 9359 // CHECK9-NEXT: store i32 [[ADD]], i32* [[I4]], align 4, !llvm.access.group !95 9360 // CHECK9-NEXT: [[TMP21:%.*]] = load i32*, i32** [[TMP2]], align 8, !llvm.access.group !95 9361 // CHECK9-NEXT: [[TMP22:%.*]] = load i32, i32* [[I4]], align 4, !llvm.access.group !95 9362 // CHECK9-NEXT: [[IDXPROM:%.*]] = sext i32 [[TMP22]] to i64 9363 // CHECK9-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i32, i32* [[TMP21]], i64 [[IDXPROM]] 9364 // CHECK9-NEXT: [[TMP23:%.*]] = load i32, i32* [[ARRAYIDX]], align 4, !llvm.access.group !95 9365 // CHECK9-NEXT: [[TMP24:%.*]] = load i32*, i32** [[TMP3]], align 8, !llvm.access.group !95 9366 // CHECK9-NEXT: [[TMP25:%.*]] = load i32, i32* [[I4]], align 4, !llvm.access.group !95 9367 // CHECK9-NEXT: [[IDXPROM6:%.*]] = sext i32 [[TMP25]] to i64 9368 // CHECK9-NEXT: [[ARRAYIDX7:%.*]] = getelementptr inbounds i32, i32* [[TMP24]], i64 [[IDXPROM6]] 9369 // CHECK9-NEXT: [[TMP26:%.*]] = load i32, i32* [[ARRAYIDX7]], align 4, !llvm.access.group !95 9370 // CHECK9-NEXT: [[ADD8:%.*]] = add nsw i32 [[TMP23]], [[TMP26]] 9371 // CHECK9-NEXT: [[TMP27:%.*]] = load i32*, i32** [[TMP1]], align 8, !llvm.access.group !95 9372 // CHECK9-NEXT: [[TMP28:%.*]] = load i32, i32* [[I4]], align 4, !llvm.access.group !95 9373 // CHECK9-NEXT: [[IDXPROM9:%.*]] = sext i32 [[TMP28]] to i64 9374 // CHECK9-NEXT: [[ARRAYIDX10:%.*]] = getelementptr inbounds i32, i32* [[TMP27]], i64 [[IDXPROM9]] 9375 // CHECK9-NEXT: store i32 [[ADD8]], i32* [[ARRAYIDX10]], align 4, !llvm.access.group !95 9376 // CHECK9-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] 9377 // CHECK9: omp.body.continue: 9378 // CHECK9-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] 9379 // CHECK9: omp.inner.for.inc: 9380 // CHECK9-NEXT: [[TMP29:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !95 9381 // CHECK9-NEXT: [[ADD11:%.*]] = add nsw i32 [[TMP29]], 1 9382 // CHECK9-NEXT: store i32 [[ADD11]], i32* [[DOTOMP_IV]], align 4, !llvm.access.group !95 9383 // CHECK9-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP96:![0-9]+]] 9384 // CHECK9: omp.inner.for.end: 9385 // CHECK9-NEXT: br label [[OMP_DISPATCH_INC:%.*]] 9386 // CHECK9: omp.dispatch.inc: 9387 // CHECK9-NEXT: br label [[OMP_DISPATCH_COND]] 9388 // CHECK9: omp.dispatch.end: 9389 // CHECK9-NEXT: [[TMP30:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4 9390 // CHECK9-NEXT: [[TMP31:%.*]] = icmp ne i32 [[TMP30]], 0 9391 // CHECK9-NEXT: br i1 [[TMP31]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]] 9392 // CHECK9: .omp.final.then: 9393 // CHECK9-NEXT: [[TMP32:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 9394 // CHECK9-NEXT: [[SUB12:%.*]] = sub nsw i32 [[TMP32]], 0 9395 // CHECK9-NEXT: [[DIV13:%.*]] = sdiv i32 [[SUB12]], 1 9396 // CHECK9-NEXT: [[MUL14:%.*]] = mul nsw i32 [[DIV13]], 1 9397 // CHECK9-NEXT: [[ADD15:%.*]] = add nsw i32 0, [[MUL14]] 9398 // CHECK9-NEXT: store i32 [[ADD15]], i32* [[I4]], align 4 9399 // CHECK9-NEXT: br label [[DOTOMP_FINAL_DONE]] 9400 // CHECK9: .omp.final.done: 9401 // CHECK9-NEXT: br label [[OMP_PRECOND_END]] 9402 // CHECK9: omp.precond.end: 9403 // CHECK9-NEXT: ret void 9404 // 9405 // 9406 // CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiET_v_l90 9407 // CHECK9-SAME: (i64 noundef [[CH:%.*]], i64 noundef [[N:%.*]], i32* noundef [[A:%.*]], i32* noundef [[B:%.*]], i32* noundef [[C:%.*]]) #[[ATTR1]] { 9408 // CHECK9-NEXT: entry: 9409 // CHECK9-NEXT: [[CH_ADDR:%.*]] = alloca i64, align 8 9410 // CHECK9-NEXT: [[N_ADDR:%.*]] = alloca i64, align 8 9411 // CHECK9-NEXT: [[A_ADDR:%.*]] = alloca i32*, align 8 9412 // CHECK9-NEXT: [[B_ADDR:%.*]] = alloca i32*, align 8 9413 // CHECK9-NEXT: [[C_ADDR:%.*]] = alloca i32*, align 8 9414 // CHECK9-NEXT: store i64 [[CH]], i64* [[CH_ADDR]], align 8 9415 // CHECK9-NEXT: store i64 [[N]], i64* [[N_ADDR]], align 8 9416 // CHECK9-NEXT: store i32* [[A]], i32** [[A_ADDR]], align 8 9417 // CHECK9-NEXT: store i32* [[B]], i32** [[B_ADDR]], align 8 9418 // CHECK9-NEXT: store i32* [[C]], i32** [[C_ADDR]], align 8 9419 // CHECK9-NEXT: [[CONV:%.*]] = bitcast i64* [[CH_ADDR]] to i32* 9420 // CHECK9-NEXT: [[CONV1:%.*]] = bitcast i64* [[N_ADDR]] to i32* 9421 // CHECK9-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 5, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32*, i32*, i32**, i32**, i32**)* @.omp_outlined..50 to void (i32*, i32*, ...)*), i32* [[CONV]], i32* [[CONV1]], i32** [[A_ADDR]], i32** [[B_ADDR]], i32** [[C_ADDR]]) 9422 // CHECK9-NEXT: ret void 9423 // 9424 // 9425 // CHECK9-LABEL: define {{[^@]+}}@.omp_outlined..50 9426 // CHECK9-SAME: (i32* noalias noundef [[DOTGLOBAL_TID_:%.*]], i32* noalias noundef [[DOTBOUND_TID_:%.*]], i32* noundef nonnull align 4 dereferenceable(4) [[CH:%.*]], i32* noundef nonnull align 4 dereferenceable(4) [[N:%.*]], i32** noundef nonnull align 8 dereferenceable(8) [[A:%.*]], i32** noundef nonnull align 8 dereferenceable(8) [[B:%.*]], i32** noundef nonnull align 8 dereferenceable(8) [[C:%.*]]) #[[ATTR1]] { 9427 // CHECK9-NEXT: entry: 9428 // CHECK9-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 9429 // CHECK9-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 9430 // CHECK9-NEXT: [[CH_ADDR:%.*]] = alloca i32*, align 8 9431 // CHECK9-NEXT: [[N_ADDR:%.*]] = alloca i32*, align 8 9432 // CHECK9-NEXT: [[A_ADDR:%.*]] = alloca i32**, align 8 9433 // CHECK9-NEXT: [[B_ADDR:%.*]] = alloca i32**, align 8 9434 // CHECK9-NEXT: [[C_ADDR:%.*]] = alloca i32**, align 8 9435 // CHECK9-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 9436 // CHECK9-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 9437 // CHECK9-NEXT: [[TMP:%.*]] = alloca i32, align 4 9438 // CHECK9-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 9439 // CHECK9-NEXT: [[DOTCAPTURE_EXPR_2:%.*]] = alloca i32, align 4 9440 // CHECK9-NEXT: [[I:%.*]] = alloca i32, align 4 9441 // CHECK9-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4 9442 // CHECK9-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4 9443 // CHECK9-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 9444 // CHECK9-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 9445 // CHECK9-NEXT: [[I4:%.*]] = alloca i32, align 4 9446 // CHECK9-NEXT: [[DOTCAPTURE_EXPR__CASTED:%.*]] = alloca i64, align 8 9447 // CHECK9-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8 9448 // CHECK9-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8 9449 // CHECK9-NEXT: store i32* [[CH]], i32** [[CH_ADDR]], align 8 9450 // CHECK9-NEXT: store i32* [[N]], i32** [[N_ADDR]], align 8 9451 // CHECK9-NEXT: store i32** [[A]], i32*** [[A_ADDR]], align 8 9452 // CHECK9-NEXT: store i32** [[B]], i32*** [[B_ADDR]], align 8 9453 // CHECK9-NEXT: store i32** [[C]], i32*** [[C_ADDR]], align 8 9454 // CHECK9-NEXT: [[TMP0:%.*]] = load i32*, i32** [[CH_ADDR]], align 8 9455 // CHECK9-NEXT: [[TMP1:%.*]] = load i32*, i32** [[N_ADDR]], align 8 9456 // CHECK9-NEXT: [[TMP2:%.*]] = load i32**, i32*** [[A_ADDR]], align 8 9457 // CHECK9-NEXT: [[TMP3:%.*]] = load i32**, i32*** [[B_ADDR]], align 8 9458 // CHECK9-NEXT: [[TMP4:%.*]] = load i32**, i32*** [[C_ADDR]], align 8 9459 // CHECK9-NEXT: [[TMP5:%.*]] = load i32, i32* [[TMP0]], align 4 9460 // CHECK9-NEXT: store i32 [[TMP5]], i32* [[DOTCAPTURE_EXPR_]], align 4 9461 // CHECK9-NEXT: [[TMP6:%.*]] = load i32, i32* [[TMP1]], align 4 9462 // CHECK9-NEXT: store i32 [[TMP6]], i32* [[DOTCAPTURE_EXPR_1]], align 4 9463 // CHECK9-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 9464 // CHECK9-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP7]], 0 9465 // CHECK9-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 9466 // CHECK9-NEXT: [[SUB3:%.*]] = sub nsw i32 [[DIV]], 1 9467 // CHECK9-NEXT: store i32 [[SUB3]], i32* [[DOTCAPTURE_EXPR_2]], align 4 9468 // CHECK9-NEXT: store i32 0, i32* [[I]], align 4 9469 // CHECK9-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 9470 // CHECK9-NEXT: [[CMP:%.*]] = icmp slt i32 0, [[TMP8]] 9471 // CHECK9-NEXT: br i1 [[CMP]], label [[OMP_PRECOND_THEN:%.*]], label [[OMP_PRECOND_END:%.*]] 9472 // CHECK9: omp.precond.then: 9473 // CHECK9-NEXT: store i32 0, i32* [[DOTOMP_COMB_LB]], align 4 9474 // CHECK9-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_2]], align 4 9475 // CHECK9-NEXT: store i32 [[TMP9]], i32* [[DOTOMP_COMB_UB]], align 4 9476 // CHECK9-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4 9477 // CHECK9-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4 9478 // CHECK9-NEXT: [[TMP10:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 9479 // CHECK9-NEXT: [[TMP11:%.*]] = load i32, i32* [[TMP10]], align 4 9480 // CHECK9-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1]], i32 [[TMP11]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_COMB_LB]], i32* [[DOTOMP_COMB_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1) 9481 // CHECK9-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 9482 // CHECK9-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_2]], align 4 9483 // CHECK9-NEXT: [[CMP5:%.*]] = icmp sgt i32 [[TMP12]], [[TMP13]] 9484 // CHECK9-NEXT: br i1 [[CMP5]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] 9485 // CHECK9: cond.true: 9486 // CHECK9-NEXT: [[TMP14:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_2]], align 4 9487 // CHECK9-NEXT: br label [[COND_END:%.*]] 9488 // CHECK9: cond.false: 9489 // CHECK9-NEXT: [[TMP15:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 9490 // CHECK9-NEXT: br label [[COND_END]] 9491 // CHECK9: cond.end: 9492 // CHECK9-NEXT: [[COND:%.*]] = phi i32 [ [[TMP14]], [[COND_TRUE]] ], [ [[TMP15]], [[COND_FALSE]] ] 9493 // CHECK9-NEXT: store i32 [[COND]], i32* [[DOTOMP_COMB_UB]], align 4 9494 // CHECK9-NEXT: [[TMP16:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4 9495 // CHECK9-NEXT: store i32 [[TMP16]], i32* [[DOTOMP_IV]], align 4 9496 // CHECK9-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] 9497 // CHECK9: omp.inner.for.cond: 9498 // CHECK9-NEXT: [[TMP17:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !98 9499 // CHECK9-NEXT: [[TMP18:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4, !llvm.access.group !98 9500 // CHECK9-NEXT: [[CMP6:%.*]] = icmp sle i32 [[TMP17]], [[TMP18]] 9501 // CHECK9-NEXT: br i1 [[CMP6]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] 9502 // CHECK9: omp.inner.for.body: 9503 // CHECK9-NEXT: [[TMP19:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4, !llvm.access.group !98 9504 // CHECK9-NEXT: [[TMP20:%.*]] = zext i32 [[TMP19]] to i64 9505 // CHECK9-NEXT: [[TMP21:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4, !llvm.access.group !98 9506 // CHECK9-NEXT: [[TMP22:%.*]] = zext i32 [[TMP21]] to i64 9507 // CHECK9-NEXT: [[TMP23:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4, !llvm.access.group !98 9508 // CHECK9-NEXT: [[CONV:%.*]] = bitcast i64* [[DOTCAPTURE_EXPR__CASTED]] to i32* 9509 // CHECK9-NEXT: store i32 [[TMP23]], i32* [[CONV]], align 4, !llvm.access.group !98 9510 // CHECK9-NEXT: [[TMP24:%.*]] = load i64, i64* [[DOTCAPTURE_EXPR__CASTED]], align 8, !llvm.access.group !98 9511 // CHECK9-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 7, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64, i32*, i32**, i32**, i32**, i64)* @.omp_outlined..51 to void (i32*, i32*, ...)*), i64 [[TMP20]], i64 [[TMP22]], i32* [[TMP1]], i32** [[TMP2]], i32** [[TMP3]], i32** [[TMP4]], i64 [[TMP24]]), !llvm.access.group !98 9512 // CHECK9-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] 9513 // CHECK9: omp.inner.for.inc: 9514 // CHECK9-NEXT: [[TMP25:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !98 9515 // CHECK9-NEXT: [[TMP26:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4, !llvm.access.group !98 9516 // CHECK9-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP25]], [[TMP26]] 9517 // CHECK9-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4, !llvm.access.group !98 9518 // CHECK9-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP99:![0-9]+]] 9519 // CHECK9: omp.inner.for.end: 9520 // CHECK9-NEXT: br label [[OMP_LOOP_EXIT:%.*]] 9521 // CHECK9: omp.loop.exit: 9522 // CHECK9-NEXT: [[TMP27:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 9523 // CHECK9-NEXT: [[TMP28:%.*]] = load i32, i32* [[TMP27]], align 4 9524 // CHECK9-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP28]]) 9525 // CHECK9-NEXT: [[TMP29:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4 9526 // CHECK9-NEXT: [[TMP30:%.*]] = icmp ne i32 [[TMP29]], 0 9527 // CHECK9-NEXT: br i1 [[TMP30]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]] 9528 // CHECK9: .omp.final.then: 9529 // CHECK9-NEXT: [[TMP31:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 9530 // CHECK9-NEXT: [[SUB7:%.*]] = sub nsw i32 [[TMP31]], 0 9531 // CHECK9-NEXT: [[DIV8:%.*]] = sdiv i32 [[SUB7]], 1 9532 // CHECK9-NEXT: [[MUL:%.*]] = mul nsw i32 [[DIV8]], 1 9533 // CHECK9-NEXT: [[ADD9:%.*]] = add nsw i32 0, [[MUL]] 9534 // CHECK9-NEXT: store i32 [[ADD9]], i32* [[I4]], align 4 9535 // CHECK9-NEXT: br label [[DOTOMP_FINAL_DONE]] 9536 // CHECK9: .omp.final.done: 9537 // CHECK9-NEXT: br label [[OMP_PRECOND_END]] 9538 // CHECK9: omp.precond.end: 9539 // CHECK9-NEXT: ret void 9540 // 9541 // 9542 // CHECK9-LABEL: define {{[^@]+}}@.omp_outlined..51 9543 // CHECK9-SAME: (i32* noalias noundef [[DOTGLOBAL_TID_:%.*]], i32* noalias noundef [[DOTBOUND_TID_:%.*]], i64 noundef [[DOTPREVIOUS_LB_:%.*]], i64 noundef [[DOTPREVIOUS_UB_:%.*]], i32* noundef nonnull align 4 dereferenceable(4) [[N:%.*]], i32** noundef nonnull align 8 dereferenceable(8) [[A:%.*]], i32** noundef nonnull align 8 dereferenceable(8) [[B:%.*]], i32** noundef nonnull align 8 dereferenceable(8) [[C:%.*]], i64 noundef [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR1]] { 9544 // CHECK9-NEXT: entry: 9545 // CHECK9-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 9546 // CHECK9-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 9547 // CHECK9-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8 9548 // CHECK9-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8 9549 // CHECK9-NEXT: [[N_ADDR:%.*]] = alloca i32*, align 8 9550 // CHECK9-NEXT: [[A_ADDR:%.*]] = alloca i32**, align 8 9551 // CHECK9-NEXT: [[B_ADDR:%.*]] = alloca i32**, align 8 9552 // CHECK9-NEXT: [[C_ADDR:%.*]] = alloca i32**, align 8 9553 // CHECK9-NEXT: [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i64, align 8 9554 // CHECK9-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 9555 // CHECK9-NEXT: [[TMP:%.*]] = alloca i32, align 4 9556 // CHECK9-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 9557 // CHECK9-NEXT: [[DOTCAPTURE_EXPR_2:%.*]] = alloca i32, align 4 9558 // CHECK9-NEXT: [[I:%.*]] = alloca i32, align 4 9559 // CHECK9-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 9560 // CHECK9-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4 9561 // CHECK9-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 9562 // CHECK9-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 9563 // CHECK9-NEXT: [[I6:%.*]] = alloca i32, align 4 9564 // CHECK9-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8 9565 // CHECK9-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8 9566 // CHECK9-NEXT: store i64 [[DOTPREVIOUS_LB_]], i64* [[DOTPREVIOUS_LB__ADDR]], align 8 9567 // CHECK9-NEXT: store i64 [[DOTPREVIOUS_UB_]], i64* [[DOTPREVIOUS_UB__ADDR]], align 8 9568 // CHECK9-NEXT: store i32* [[N]], i32** [[N_ADDR]], align 8 9569 // CHECK9-NEXT: store i32** [[A]], i32*** [[A_ADDR]], align 8 9570 // CHECK9-NEXT: store i32** [[B]], i32*** [[B_ADDR]], align 8 9571 // CHECK9-NEXT: store i32** [[C]], i32*** [[C_ADDR]], align 8 9572 // CHECK9-NEXT: store i64 [[DOTCAPTURE_EXPR_]], i64* [[DOTCAPTURE_EXPR__ADDR]], align 8 9573 // CHECK9-NEXT: [[TMP0:%.*]] = load i32*, i32** [[N_ADDR]], align 8 9574 // CHECK9-NEXT: [[TMP1:%.*]] = load i32**, i32*** [[A_ADDR]], align 8 9575 // CHECK9-NEXT: [[TMP2:%.*]] = load i32**, i32*** [[B_ADDR]], align 8 9576 // CHECK9-NEXT: [[TMP3:%.*]] = load i32**, i32*** [[C_ADDR]], align 8 9577 // CHECK9-NEXT: [[CONV:%.*]] = bitcast i64* [[DOTCAPTURE_EXPR__ADDR]] to i32* 9578 // CHECK9-NEXT: [[TMP4:%.*]] = load i32, i32* [[TMP0]], align 4 9579 // CHECK9-NEXT: store i32 [[TMP4]], i32* [[DOTCAPTURE_EXPR_1]], align 4 9580 // CHECK9-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 9581 // CHECK9-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP5]], 0 9582 // CHECK9-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 9583 // CHECK9-NEXT: [[SUB3:%.*]] = sub nsw i32 [[DIV]], 1 9584 // CHECK9-NEXT: store i32 [[SUB3]], i32* [[DOTCAPTURE_EXPR_2]], align 4 9585 // CHECK9-NEXT: store i32 0, i32* [[I]], align 4 9586 // CHECK9-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 9587 // CHECK9-NEXT: [[CMP:%.*]] = icmp slt i32 0, [[TMP6]] 9588 // CHECK9-NEXT: br i1 [[CMP]], label [[OMP_PRECOND_THEN:%.*]], label [[OMP_PRECOND_END:%.*]] 9589 // CHECK9: omp.precond.then: 9590 // CHECK9-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4 9591 // CHECK9-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_2]], align 4 9592 // CHECK9-NEXT: store i32 [[TMP7]], i32* [[DOTOMP_UB]], align 4 9593 // CHECK9-NEXT: [[TMP8:%.*]] = load i64, i64* [[DOTPREVIOUS_LB__ADDR]], align 8 9594 // CHECK9-NEXT: [[CONV4:%.*]] = trunc i64 [[TMP8]] to i32 9595 // CHECK9-NEXT: [[TMP9:%.*]] = load i64, i64* [[DOTPREVIOUS_UB__ADDR]], align 8 9596 // CHECK9-NEXT: [[CONV5:%.*]] = trunc i64 [[TMP9]] to i32 9597 // CHECK9-NEXT: store i32 [[CONV4]], i32* [[DOTOMP_LB]], align 4 9598 // CHECK9-NEXT: store i32 [[CONV5]], i32* [[DOTOMP_UB]], align 4 9599 // CHECK9-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4 9600 // CHECK9-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4 9601 // CHECK9-NEXT: [[TMP10:%.*]] = load i32, i32* [[CONV]], align 4 9602 // CHECK9-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4 9603 // CHECK9-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 9604 // CHECK9-NEXT: [[TMP13:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 9605 // CHECK9-NEXT: [[TMP14:%.*]] = load i32, i32* [[TMP13]], align 4 9606 // CHECK9-NEXT: call void @__kmpc_dispatch_init_4(%struct.ident_t* @[[GLOB3]], i32 [[TMP14]], i32 35, i32 [[TMP11]], i32 [[TMP12]], i32 1, i32 [[TMP10]]) 9607 // CHECK9-NEXT: br label [[OMP_DISPATCH_COND:%.*]] 9608 // CHECK9: omp.dispatch.cond: 9609 // CHECK9-NEXT: [[TMP15:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 9610 // CHECK9-NEXT: [[TMP16:%.*]] = load i32, i32* [[TMP15]], align 4 9611 // CHECK9-NEXT: [[TMP17:%.*]] = call i32 @__kmpc_dispatch_next_4(%struct.ident_t* @[[GLOB3]], i32 [[TMP16]], i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]]) 9612 // CHECK9-NEXT: [[TOBOOL:%.*]] = icmp ne i32 [[TMP17]], 0 9613 // CHECK9-NEXT: br i1 [[TOBOOL]], label [[OMP_DISPATCH_BODY:%.*]], label [[OMP_DISPATCH_END:%.*]] 9614 // CHECK9: omp.dispatch.body: 9615 // CHECK9-NEXT: [[TMP18:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4 9616 // CHECK9-NEXT: store i32 [[TMP18]], i32* [[DOTOMP_IV]], align 4 9617 // CHECK9-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] 9618 // CHECK9: omp.inner.for.cond: 9619 // CHECK9-NEXT: [[TMP19:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !101 9620 // CHECK9-NEXT: [[TMP20:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4, !llvm.access.group !101 9621 // CHECK9-NEXT: [[CMP7:%.*]] = icmp sle i32 [[TMP19]], [[TMP20]] 9622 // CHECK9-NEXT: br i1 [[CMP7]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] 9623 // CHECK9: omp.inner.for.body: 9624 // CHECK9-NEXT: [[TMP21:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !101 9625 // CHECK9-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP21]], 1 9626 // CHECK9-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] 9627 // CHECK9-NEXT: store i32 [[ADD]], i32* [[I6]], align 4, !llvm.access.group !101 9628 // CHECK9-NEXT: [[TMP22:%.*]] = load i32*, i32** [[TMP2]], align 8, !llvm.access.group !101 9629 // CHECK9-NEXT: [[TMP23:%.*]] = load i32, i32* [[I6]], align 4, !llvm.access.group !101 9630 // CHECK9-NEXT: [[IDXPROM:%.*]] = sext i32 [[TMP23]] to i64 9631 // CHECK9-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i32, i32* [[TMP22]], i64 [[IDXPROM]] 9632 // CHECK9-NEXT: [[TMP24:%.*]] = load i32, i32* [[ARRAYIDX]], align 4, !llvm.access.group !101 9633 // CHECK9-NEXT: [[TMP25:%.*]] = load i32*, i32** [[TMP3]], align 8, !llvm.access.group !101 9634 // CHECK9-NEXT: [[TMP26:%.*]] = load i32, i32* [[I6]], align 4, !llvm.access.group !101 9635 // CHECK9-NEXT: [[IDXPROM8:%.*]] = sext i32 [[TMP26]] to i64 9636 // CHECK9-NEXT: [[ARRAYIDX9:%.*]] = getelementptr inbounds i32, i32* [[TMP25]], i64 [[IDXPROM8]] 9637 // CHECK9-NEXT: [[TMP27:%.*]] = load i32, i32* [[ARRAYIDX9]], align 4, !llvm.access.group !101 9638 // CHECK9-NEXT: [[ADD10:%.*]] = add nsw i32 [[TMP24]], [[TMP27]] 9639 // CHECK9-NEXT: [[TMP28:%.*]] = load i32*, i32** [[TMP1]], align 8, !llvm.access.group !101 9640 // CHECK9-NEXT: [[TMP29:%.*]] = load i32, i32* [[I6]], align 4, !llvm.access.group !101 9641 // CHECK9-NEXT: [[IDXPROM11:%.*]] = sext i32 [[TMP29]] to i64 9642 // CHECK9-NEXT: [[ARRAYIDX12:%.*]] = getelementptr inbounds i32, i32* [[TMP28]], i64 [[IDXPROM11]] 9643 // CHECK9-NEXT: store i32 [[ADD10]], i32* [[ARRAYIDX12]], align 4, !llvm.access.group !101 9644 // CHECK9-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] 9645 // CHECK9: omp.body.continue: 9646 // CHECK9-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] 9647 // CHECK9: omp.inner.for.inc: 9648 // CHECK9-NEXT: [[TMP30:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !101 9649 // CHECK9-NEXT: [[ADD13:%.*]] = add nsw i32 [[TMP30]], 1 9650 // CHECK9-NEXT: store i32 [[ADD13]], i32* [[DOTOMP_IV]], align 4, !llvm.access.group !101 9651 // CHECK9-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP102:![0-9]+]] 9652 // CHECK9: omp.inner.for.end: 9653 // CHECK9-NEXT: br label [[OMP_DISPATCH_INC:%.*]] 9654 // CHECK9: omp.dispatch.inc: 9655 // CHECK9-NEXT: br label [[OMP_DISPATCH_COND]] 9656 // CHECK9: omp.dispatch.end: 9657 // CHECK9-NEXT: [[TMP31:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4 9658 // CHECK9-NEXT: [[TMP32:%.*]] = icmp ne i32 [[TMP31]], 0 9659 // CHECK9-NEXT: br i1 [[TMP32]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]] 9660 // CHECK9: .omp.final.then: 9661 // CHECK9-NEXT: [[TMP33:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 9662 // CHECK9-NEXT: [[SUB14:%.*]] = sub nsw i32 [[TMP33]], 0 9663 // CHECK9-NEXT: [[DIV15:%.*]] = sdiv i32 [[SUB14]], 1 9664 // CHECK9-NEXT: [[MUL16:%.*]] = mul nsw i32 [[DIV15]], 1 9665 // CHECK9-NEXT: [[ADD17:%.*]] = add nsw i32 0, [[MUL16]] 9666 // CHECK9-NEXT: store i32 [[ADD17]], i32* [[I6]], align 4 9667 // CHECK9-NEXT: br label [[DOTOMP_FINAL_DONE]] 9668 // CHECK9: .omp.final.done: 9669 // CHECK9-NEXT: br label [[OMP_PRECOND_END]] 9670 // CHECK9: omp.precond.end: 9671 // CHECK9-NEXT: ret void 9672 // 9673 // 9674 // CHECK9-LABEL: define {{[^@]+}}@.omp_offloading.requires_reg 9675 // CHECK9-SAME: () #[[ATTR4:[0-9]+]] { 9676 // CHECK9-NEXT: entry: 9677 // CHECK9-NEXT: call void @__tgt_register_requires(i64 1) 9678 // CHECK9-NEXT: ret void 9679 // 9680 // 9681 // CHECK11-LABEL: define {{[^@]+}}@main 9682 // CHECK11-SAME: () #[[ATTR0:[0-9]+]] { 9683 // CHECK11-NEXT: entry: 9684 // CHECK11-NEXT: [[RETVAL:%.*]] = alloca i32, align 4 9685 // CHECK11-NEXT: [[A:%.*]] = alloca double*, align 4 9686 // CHECK11-NEXT: [[B:%.*]] = alloca double*, align 4 9687 // CHECK11-NEXT: [[C:%.*]] = alloca double*, align 4 9688 // CHECK11-NEXT: [[N:%.*]] = alloca i32, align 4 9689 // CHECK11-NEXT: [[CH:%.*]] = alloca i32, align 4 9690 // CHECK11-NEXT: [[N_CASTED:%.*]] = alloca i32, align 4 9691 // CHECK11-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [4 x i8*], align 4 9692 // CHECK11-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [4 x i8*], align 4 9693 // CHECK11-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [4 x i8*], align 4 9694 // CHECK11-NEXT: [[TMP:%.*]] = alloca i32, align 4 9695 // CHECK11-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 9696 // CHECK11-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 9697 // CHECK11-NEXT: [[N_CASTED3:%.*]] = alloca i32, align 4 9698 // CHECK11-NEXT: [[DOTOFFLOAD_BASEPTRS4:%.*]] = alloca [4 x i8*], align 4 9699 // CHECK11-NEXT: [[DOTOFFLOAD_PTRS5:%.*]] = alloca [4 x i8*], align 4 9700 // CHECK11-NEXT: [[DOTOFFLOAD_MAPPERS6:%.*]] = alloca [4 x i8*], align 4 9701 // CHECK11-NEXT: [[_TMP7:%.*]] = alloca i32, align 4 9702 // CHECK11-NEXT: [[DOTCAPTURE_EXPR_8:%.*]] = alloca i32, align 4 9703 // CHECK11-NEXT: [[DOTCAPTURE_EXPR_9:%.*]] = alloca i32, align 4 9704 // CHECK11-NEXT: [[CH_CASTED:%.*]] = alloca i32, align 4 9705 // CHECK11-NEXT: [[N_CASTED17:%.*]] = alloca i32, align 4 9706 // CHECK11-NEXT: [[DOTOFFLOAD_BASEPTRS18:%.*]] = alloca [5 x i8*], align 4 9707 // CHECK11-NEXT: [[DOTOFFLOAD_PTRS19:%.*]] = alloca [5 x i8*], align 4 9708 // CHECK11-NEXT: [[DOTOFFLOAD_MAPPERS20:%.*]] = alloca [5 x i8*], align 4 9709 // CHECK11-NEXT: [[_TMP21:%.*]] = alloca i32, align 4 9710 // CHECK11-NEXT: [[DOTCAPTURE_EXPR_22:%.*]] = alloca i32, align 4 9711 // CHECK11-NEXT: [[DOTCAPTURE_EXPR_23:%.*]] = alloca i32, align 4 9712 // CHECK11-NEXT: [[N_CASTED31:%.*]] = alloca i32, align 4 9713 // CHECK11-NEXT: [[DOTOFFLOAD_BASEPTRS32:%.*]] = alloca [4 x i8*], align 4 9714 // CHECK11-NEXT: [[DOTOFFLOAD_PTRS33:%.*]] = alloca [4 x i8*], align 4 9715 // CHECK11-NEXT: [[DOTOFFLOAD_MAPPERS34:%.*]] = alloca [4 x i8*], align 4 9716 // CHECK11-NEXT: [[_TMP35:%.*]] = alloca i32, align 4 9717 // CHECK11-NEXT: [[DOTCAPTURE_EXPR_36:%.*]] = alloca i32, align 4 9718 // CHECK11-NEXT: [[DOTCAPTURE_EXPR_37:%.*]] = alloca i32, align 4 9719 // CHECK11-NEXT: [[CH_CASTED45:%.*]] = alloca i32, align 4 9720 // CHECK11-NEXT: [[N_CASTED46:%.*]] = alloca i32, align 4 9721 // CHECK11-NEXT: [[DOTOFFLOAD_BASEPTRS47:%.*]] = alloca [5 x i8*], align 4 9722 // CHECK11-NEXT: [[DOTOFFLOAD_PTRS48:%.*]] = alloca [5 x i8*], align 4 9723 // CHECK11-NEXT: [[DOTOFFLOAD_MAPPERS49:%.*]] = alloca [5 x i8*], align 4 9724 // CHECK11-NEXT: [[_TMP50:%.*]] = alloca i32, align 4 9725 // CHECK11-NEXT: [[DOTCAPTURE_EXPR_51:%.*]] = alloca i32, align 4 9726 // CHECK11-NEXT: [[DOTCAPTURE_EXPR_52:%.*]] = alloca i32, align 4 9727 // CHECK11-NEXT: [[N_CASTED60:%.*]] = alloca i32, align 4 9728 // CHECK11-NEXT: [[DOTOFFLOAD_BASEPTRS61:%.*]] = alloca [4 x i8*], align 4 9729 // CHECK11-NEXT: [[DOTOFFLOAD_PTRS62:%.*]] = alloca [4 x i8*], align 4 9730 // CHECK11-NEXT: [[DOTOFFLOAD_MAPPERS63:%.*]] = alloca [4 x i8*], align 4 9731 // CHECK11-NEXT: [[_TMP64:%.*]] = alloca i32, align 4 9732 // CHECK11-NEXT: [[DOTCAPTURE_EXPR_65:%.*]] = alloca i32, align 4 9733 // CHECK11-NEXT: [[DOTCAPTURE_EXPR_66:%.*]] = alloca i32, align 4 9734 // CHECK11-NEXT: [[CH_CASTED74:%.*]] = alloca i32, align 4 9735 // CHECK11-NEXT: [[N_CASTED75:%.*]] = alloca i32, align 4 9736 // CHECK11-NEXT: [[DOTOFFLOAD_BASEPTRS76:%.*]] = alloca [5 x i8*], align 4 9737 // CHECK11-NEXT: [[DOTOFFLOAD_PTRS77:%.*]] = alloca [5 x i8*], align 4 9738 // CHECK11-NEXT: [[DOTOFFLOAD_MAPPERS78:%.*]] = alloca [5 x i8*], align 4 9739 // CHECK11-NEXT: [[_TMP79:%.*]] = alloca i32, align 4 9740 // CHECK11-NEXT: [[DOTCAPTURE_EXPR_80:%.*]] = alloca i32, align 4 9741 // CHECK11-NEXT: [[DOTCAPTURE_EXPR_81:%.*]] = alloca i32, align 4 9742 // CHECK11-NEXT: store i32 0, i32* [[RETVAL]], align 4 9743 // CHECK11-NEXT: store i32 10000, i32* [[N]], align 4 9744 // CHECK11-NEXT: store i32 100, i32* [[CH]], align 4 9745 // CHECK11-NEXT: [[TMP0:%.*]] = load i32, i32* [[N]], align 4 9746 // CHECK11-NEXT: store i32 [[TMP0]], i32* [[N_CASTED]], align 4 9747 // CHECK11-NEXT: [[TMP1:%.*]] = load i32, i32* [[N_CASTED]], align 4 9748 // CHECK11-NEXT: [[TMP2:%.*]] = load double*, double** [[A]], align 4 9749 // CHECK11-NEXT: [[TMP3:%.*]] = load double*, double** [[B]], align 4 9750 // CHECK11-NEXT: [[TMP4:%.*]] = load double*, double** [[C]], align 4 9751 // CHECK11-NEXT: [[TMP5:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 9752 // CHECK11-NEXT: [[TMP6:%.*]] = bitcast i8** [[TMP5]] to i32* 9753 // CHECK11-NEXT: store i32 [[TMP1]], i32* [[TMP6]], align 4 9754 // CHECK11-NEXT: [[TMP7:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 9755 // CHECK11-NEXT: [[TMP8:%.*]] = bitcast i8** [[TMP7]] to i32* 9756 // CHECK11-NEXT: store i32 [[TMP1]], i32* [[TMP8]], align 4 9757 // CHECK11-NEXT: [[TMP9:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0 9758 // CHECK11-NEXT: store i8* null, i8** [[TMP9]], align 4 9759 // CHECK11-NEXT: [[TMP10:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1 9760 // CHECK11-NEXT: [[TMP11:%.*]] = bitcast i8** [[TMP10]] to double** 9761 // CHECK11-NEXT: store double* [[TMP2]], double** [[TMP11]], align 4 9762 // CHECK11-NEXT: [[TMP12:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1 9763 // CHECK11-NEXT: [[TMP13:%.*]] = bitcast i8** [[TMP12]] to double** 9764 // CHECK11-NEXT: store double* [[TMP2]], double** [[TMP13]], align 4 9765 // CHECK11-NEXT: [[TMP14:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 1 9766 // CHECK11-NEXT: store i8* null, i8** [[TMP14]], align 4 9767 // CHECK11-NEXT: [[TMP15:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2 9768 // CHECK11-NEXT: [[TMP16:%.*]] = bitcast i8** [[TMP15]] to double** 9769 // CHECK11-NEXT: store double* [[TMP3]], double** [[TMP16]], align 4 9770 // CHECK11-NEXT: [[TMP17:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2 9771 // CHECK11-NEXT: [[TMP18:%.*]] = bitcast i8** [[TMP17]] to double** 9772 // CHECK11-NEXT: store double* [[TMP3]], double** [[TMP18]], align 4 9773 // CHECK11-NEXT: [[TMP19:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2 9774 // CHECK11-NEXT: store i8* null, i8** [[TMP19]], align 4 9775 // CHECK11-NEXT: [[TMP20:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 3 9776 // CHECK11-NEXT: [[TMP21:%.*]] = bitcast i8** [[TMP20]] to double** 9777 // CHECK11-NEXT: store double* [[TMP4]], double** [[TMP21]], align 4 9778 // CHECK11-NEXT: [[TMP22:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 3 9779 // CHECK11-NEXT: [[TMP23:%.*]] = bitcast i8** [[TMP22]] to double** 9780 // CHECK11-NEXT: store double* [[TMP4]], double** [[TMP23]], align 4 9781 // CHECK11-NEXT: [[TMP24:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 3 9782 // CHECK11-NEXT: store i8* null, i8** [[TMP24]], align 4 9783 // CHECK11-NEXT: [[TMP25:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 9784 // CHECK11-NEXT: [[TMP26:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 9785 // CHECK11-NEXT: [[TMP27:%.*]] = load i32, i32* [[N]], align 4 9786 // CHECK11-NEXT: store i32 [[TMP27]], i32* [[DOTCAPTURE_EXPR_]], align 4 9787 // CHECK11-NEXT: [[TMP28:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 9788 // CHECK11-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP28]], 0 9789 // CHECK11-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 9790 // CHECK11-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1 9791 // CHECK11-NEXT: store i32 [[SUB2]], i32* [[DOTCAPTURE_EXPR_1]], align 4 9792 // CHECK11-NEXT: [[TMP29:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 9793 // CHECK11-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP29]], 1 9794 // CHECK11-NEXT: [[TMP30:%.*]] = zext i32 [[ADD]] to i64 9795 // CHECK11-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8 9796 // CHECK11-NEXT: [[TMP31:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS]], i32 0, i32 0 9797 // CHECK11-NEXT: store i32 1, i32* [[TMP31]], align 4 9798 // CHECK11-NEXT: [[TMP32:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS]], i32 0, i32 1 9799 // CHECK11-NEXT: store i32 4, i32* [[TMP32]], align 4 9800 // CHECK11-NEXT: [[TMP33:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS]], i32 0, i32 2 9801 // CHECK11-NEXT: store i8** [[TMP25]], i8*** [[TMP33]], align 4 9802 // CHECK11-NEXT: [[TMP34:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS]], i32 0, i32 3 9803 // CHECK11-NEXT: store i8** [[TMP26]], i8*** [[TMP34]], align 4 9804 // CHECK11-NEXT: [[TMP35:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS]], i32 0, i32 4 9805 // CHECK11-NEXT: store i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_sizes, i32 0, i32 0), i64** [[TMP35]], align 4 9806 // CHECK11-NEXT: [[TMP36:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS]], i32 0, i32 5 9807 // CHECK11-NEXT: store i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_maptypes, i32 0, i32 0), i64** [[TMP36]], align 4 9808 // CHECK11-NEXT: [[TMP37:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS]], i32 0, i32 6 9809 // CHECK11-NEXT: store i8** null, i8*** [[TMP37]], align 4 9810 // CHECK11-NEXT: [[TMP38:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS]], i32 0, i32 7 9811 // CHECK11-NEXT: store i8** null, i8*** [[TMP38]], align 4 9812 // CHECK11-NEXT: [[TMP39:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS]], i32 0, i32 8 9813 // CHECK11-NEXT: store i64 [[TMP30]], i64* [[TMP39]], align 8 9814 // CHECK11-NEXT: [[TMP40:%.*]] = call i32 @__tgt_target_kernel(%struct.ident_t* @[[GLOB3:[0-9]+]], i64 -1, i32 0, i32 1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l368.region_id, %struct.__tgt_kernel_arguments* [[KERNEL_ARGS]]) 9815 // CHECK11-NEXT: [[TMP41:%.*]] = icmp ne i32 [[TMP40]], 0 9816 // CHECK11-NEXT: br i1 [[TMP41]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] 9817 // CHECK11: omp_offload.failed: 9818 // CHECK11-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l368(i32 [[TMP1]], double* [[TMP2]], double* [[TMP3]], double* [[TMP4]]) #[[ATTR2:[0-9]+]] 9819 // CHECK11-NEXT: br label [[OMP_OFFLOAD_CONT]] 9820 // CHECK11: omp_offload.cont: 9821 // CHECK11-NEXT: [[TMP42:%.*]] = load i32, i32* [[N]], align 4 9822 // CHECK11-NEXT: store i32 [[TMP42]], i32* [[N_CASTED3]], align 4 9823 // CHECK11-NEXT: [[TMP43:%.*]] = load i32, i32* [[N_CASTED3]], align 4 9824 // CHECK11-NEXT: [[TMP44:%.*]] = load double*, double** [[A]], align 4 9825 // CHECK11-NEXT: [[TMP45:%.*]] = load double*, double** [[B]], align 4 9826 // CHECK11-NEXT: [[TMP46:%.*]] = load double*, double** [[C]], align 4 9827 // CHECK11-NEXT: [[TMP47:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS4]], i32 0, i32 0 9828 // CHECK11-NEXT: [[TMP48:%.*]] = bitcast i8** [[TMP47]] to i32* 9829 // CHECK11-NEXT: store i32 [[TMP43]], i32* [[TMP48]], align 4 9830 // CHECK11-NEXT: [[TMP49:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS5]], i32 0, i32 0 9831 // CHECK11-NEXT: [[TMP50:%.*]] = bitcast i8** [[TMP49]] to i32* 9832 // CHECK11-NEXT: store i32 [[TMP43]], i32* [[TMP50]], align 4 9833 // CHECK11-NEXT: [[TMP51:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS6]], i32 0, i32 0 9834 // CHECK11-NEXT: store i8* null, i8** [[TMP51]], align 4 9835 // CHECK11-NEXT: [[TMP52:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS4]], i32 0, i32 1 9836 // CHECK11-NEXT: [[TMP53:%.*]] = bitcast i8** [[TMP52]] to double** 9837 // CHECK11-NEXT: store double* [[TMP44]], double** [[TMP53]], align 4 9838 // CHECK11-NEXT: [[TMP54:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS5]], i32 0, i32 1 9839 // CHECK11-NEXT: [[TMP55:%.*]] = bitcast i8** [[TMP54]] to double** 9840 // CHECK11-NEXT: store double* [[TMP44]], double** [[TMP55]], align 4 9841 // CHECK11-NEXT: [[TMP56:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS6]], i32 0, i32 1 9842 // CHECK11-NEXT: store i8* null, i8** [[TMP56]], align 4 9843 // CHECK11-NEXT: [[TMP57:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS4]], i32 0, i32 2 9844 // CHECK11-NEXT: [[TMP58:%.*]] = bitcast i8** [[TMP57]] to double** 9845 // CHECK11-NEXT: store double* [[TMP45]], double** [[TMP58]], align 4 9846 // CHECK11-NEXT: [[TMP59:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS5]], i32 0, i32 2 9847 // CHECK11-NEXT: [[TMP60:%.*]] = bitcast i8** [[TMP59]] to double** 9848 // CHECK11-NEXT: store double* [[TMP45]], double** [[TMP60]], align 4 9849 // CHECK11-NEXT: [[TMP61:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS6]], i32 0, i32 2 9850 // CHECK11-NEXT: store i8* null, i8** [[TMP61]], align 4 9851 // CHECK11-NEXT: [[TMP62:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS4]], i32 0, i32 3 9852 // CHECK11-NEXT: [[TMP63:%.*]] = bitcast i8** [[TMP62]] to double** 9853 // CHECK11-NEXT: store double* [[TMP46]], double** [[TMP63]], align 4 9854 // CHECK11-NEXT: [[TMP64:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS5]], i32 0, i32 3 9855 // CHECK11-NEXT: [[TMP65:%.*]] = bitcast i8** [[TMP64]] to double** 9856 // CHECK11-NEXT: store double* [[TMP46]], double** [[TMP65]], align 4 9857 // CHECK11-NEXT: [[TMP66:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS6]], i32 0, i32 3 9858 // CHECK11-NEXT: store i8* null, i8** [[TMP66]], align 4 9859 // CHECK11-NEXT: [[TMP67:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS4]], i32 0, i32 0 9860 // CHECK11-NEXT: [[TMP68:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS5]], i32 0, i32 0 9861 // CHECK11-NEXT: [[TMP69:%.*]] = load i32, i32* [[N]], align 4 9862 // CHECK11-NEXT: store i32 [[TMP69]], i32* [[DOTCAPTURE_EXPR_8]], align 4 9863 // CHECK11-NEXT: [[TMP70:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_8]], align 4 9864 // CHECK11-NEXT: [[SUB10:%.*]] = sub nsw i32 [[TMP70]], 0 9865 // CHECK11-NEXT: [[DIV11:%.*]] = sdiv i32 [[SUB10]], 1 9866 // CHECK11-NEXT: [[SUB12:%.*]] = sub nsw i32 [[DIV11]], 1 9867 // CHECK11-NEXT: store i32 [[SUB12]], i32* [[DOTCAPTURE_EXPR_9]], align 4 9868 // CHECK11-NEXT: [[TMP71:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_9]], align 4 9869 // CHECK11-NEXT: [[ADD13:%.*]] = add nsw i32 [[TMP71]], 1 9870 // CHECK11-NEXT: [[TMP72:%.*]] = zext i32 [[ADD13]] to i64 9871 // CHECK11-NEXT: [[KERNEL_ARGS14:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS]], align 8 9872 // CHECK11-NEXT: [[TMP73:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS14]], i32 0, i32 0 9873 // CHECK11-NEXT: store i32 1, i32* [[TMP73]], align 4 9874 // CHECK11-NEXT: [[TMP74:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS14]], i32 0, i32 1 9875 // CHECK11-NEXT: store i32 4, i32* [[TMP74]], align 4 9876 // CHECK11-NEXT: [[TMP75:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS14]], i32 0, i32 2 9877 // CHECK11-NEXT: store i8** [[TMP67]], i8*** [[TMP75]], align 4 9878 // CHECK11-NEXT: [[TMP76:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS14]], i32 0, i32 3 9879 // CHECK11-NEXT: store i8** [[TMP68]], i8*** [[TMP76]], align 4 9880 // CHECK11-NEXT: [[TMP77:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS14]], i32 0, i32 4 9881 // CHECK11-NEXT: store i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_sizes.4, i32 0, i32 0), i64** [[TMP77]], align 4 9882 // CHECK11-NEXT: [[TMP78:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS14]], i32 0, i32 5 9883 // CHECK11-NEXT: store i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_maptypes.5, i32 0, i32 0), i64** [[TMP78]], align 4 9884 // CHECK11-NEXT: [[TMP79:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS14]], i32 0, i32 6 9885 // CHECK11-NEXT: store i8** null, i8*** [[TMP79]], align 4 9886 // CHECK11-NEXT: [[TMP80:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS14]], i32 0, i32 7 9887 // CHECK11-NEXT: store i8** null, i8*** [[TMP80]], align 4 9888 // CHECK11-NEXT: [[TMP81:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS14]], i32 0, i32 8 9889 // CHECK11-NEXT: store i64 [[TMP72]], i64* [[TMP81]], align 8 9890 // CHECK11-NEXT: [[TMP82:%.*]] = call i32 @__tgt_target_kernel(%struct.ident_t* @[[GLOB3]], i64 -1, i32 0, i32 1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l407.region_id, %struct.__tgt_kernel_arguments* [[KERNEL_ARGS14]]) 9891 // CHECK11-NEXT: [[TMP83:%.*]] = icmp ne i32 [[TMP82]], 0 9892 // CHECK11-NEXT: br i1 [[TMP83]], label [[OMP_OFFLOAD_FAILED15:%.*]], label [[OMP_OFFLOAD_CONT16:%.*]] 9893 // CHECK11: omp_offload.failed15: 9894 // CHECK11-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l407(i32 [[TMP43]], double* [[TMP44]], double* [[TMP45]], double* [[TMP46]]) #[[ATTR2]] 9895 // CHECK11-NEXT: br label [[OMP_OFFLOAD_CONT16]] 9896 // CHECK11: omp_offload.cont16: 9897 // CHECK11-NEXT: [[TMP84:%.*]] = load i32, i32* [[CH]], align 4 9898 // CHECK11-NEXT: store i32 [[TMP84]], i32* [[CH_CASTED]], align 4 9899 // CHECK11-NEXT: [[TMP85:%.*]] = load i32, i32* [[CH_CASTED]], align 4 9900 // CHECK11-NEXT: [[TMP86:%.*]] = load i32, i32* [[N]], align 4 9901 // CHECK11-NEXT: store i32 [[TMP86]], i32* [[N_CASTED17]], align 4 9902 // CHECK11-NEXT: [[TMP87:%.*]] = load i32, i32* [[N_CASTED17]], align 4 9903 // CHECK11-NEXT: [[TMP88:%.*]] = load double*, double** [[A]], align 4 9904 // CHECK11-NEXT: [[TMP89:%.*]] = load double*, double** [[B]], align 4 9905 // CHECK11-NEXT: [[TMP90:%.*]] = load double*, double** [[C]], align 4 9906 // CHECK11-NEXT: [[TMP91:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS18]], i32 0, i32 0 9907 // CHECK11-NEXT: [[TMP92:%.*]] = bitcast i8** [[TMP91]] to i32* 9908 // CHECK11-NEXT: store i32 [[TMP85]], i32* [[TMP92]], align 4 9909 // CHECK11-NEXT: [[TMP93:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS19]], i32 0, i32 0 9910 // CHECK11-NEXT: [[TMP94:%.*]] = bitcast i8** [[TMP93]] to i32* 9911 // CHECK11-NEXT: store i32 [[TMP85]], i32* [[TMP94]], align 4 9912 // CHECK11-NEXT: [[TMP95:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS20]], i32 0, i32 0 9913 // CHECK11-NEXT: store i8* null, i8** [[TMP95]], align 4 9914 // CHECK11-NEXT: [[TMP96:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS18]], i32 0, i32 1 9915 // CHECK11-NEXT: [[TMP97:%.*]] = bitcast i8** [[TMP96]] to i32* 9916 // CHECK11-NEXT: store i32 [[TMP87]], i32* [[TMP97]], align 4 9917 // CHECK11-NEXT: [[TMP98:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS19]], i32 0, i32 1 9918 // CHECK11-NEXT: [[TMP99:%.*]] = bitcast i8** [[TMP98]] to i32* 9919 // CHECK11-NEXT: store i32 [[TMP87]], i32* [[TMP99]], align 4 9920 // CHECK11-NEXT: [[TMP100:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS20]], i32 0, i32 1 9921 // CHECK11-NEXT: store i8* null, i8** [[TMP100]], align 4 9922 // CHECK11-NEXT: [[TMP101:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS18]], i32 0, i32 2 9923 // CHECK11-NEXT: [[TMP102:%.*]] = bitcast i8** [[TMP101]] to double** 9924 // CHECK11-NEXT: store double* [[TMP88]], double** [[TMP102]], align 4 9925 // CHECK11-NEXT: [[TMP103:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS19]], i32 0, i32 2 9926 // CHECK11-NEXT: [[TMP104:%.*]] = bitcast i8** [[TMP103]] to double** 9927 // CHECK11-NEXT: store double* [[TMP88]], double** [[TMP104]], align 4 9928 // CHECK11-NEXT: [[TMP105:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS20]], i32 0, i32 2 9929 // CHECK11-NEXT: store i8* null, i8** [[TMP105]], align 4 9930 // CHECK11-NEXT: [[TMP106:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS18]], i32 0, i32 3 9931 // CHECK11-NEXT: [[TMP107:%.*]] = bitcast i8** [[TMP106]] to double** 9932 // CHECK11-NEXT: store double* [[TMP89]], double** [[TMP107]], align 4 9933 // CHECK11-NEXT: [[TMP108:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS19]], i32 0, i32 3 9934 // CHECK11-NEXT: [[TMP109:%.*]] = bitcast i8** [[TMP108]] to double** 9935 // CHECK11-NEXT: store double* [[TMP89]], double** [[TMP109]], align 4 9936 // CHECK11-NEXT: [[TMP110:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS20]], i32 0, i32 3 9937 // CHECK11-NEXT: store i8* null, i8** [[TMP110]], align 4 9938 // CHECK11-NEXT: [[TMP111:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS18]], i32 0, i32 4 9939 // CHECK11-NEXT: [[TMP112:%.*]] = bitcast i8** [[TMP111]] to double** 9940 // CHECK11-NEXT: store double* [[TMP90]], double** [[TMP112]], align 4 9941 // CHECK11-NEXT: [[TMP113:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS19]], i32 0, i32 4 9942 // CHECK11-NEXT: [[TMP114:%.*]] = bitcast i8** [[TMP113]] to double** 9943 // CHECK11-NEXT: store double* [[TMP90]], double** [[TMP114]], align 4 9944 // CHECK11-NEXT: [[TMP115:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS20]], i32 0, i32 4 9945 // CHECK11-NEXT: store i8* null, i8** [[TMP115]], align 4 9946 // CHECK11-NEXT: [[TMP116:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS18]], i32 0, i32 0 9947 // CHECK11-NEXT: [[TMP117:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS19]], i32 0, i32 0 9948 // CHECK11-NEXT: [[TMP118:%.*]] = load i32, i32* [[N]], align 4 9949 // CHECK11-NEXT: store i32 [[TMP118]], i32* [[DOTCAPTURE_EXPR_22]], align 4 9950 // CHECK11-NEXT: [[TMP119:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_22]], align 4 9951 // CHECK11-NEXT: [[SUB24:%.*]] = sub nsw i32 [[TMP119]], 0 9952 // CHECK11-NEXT: [[DIV25:%.*]] = sdiv i32 [[SUB24]], 1 9953 // CHECK11-NEXT: [[SUB26:%.*]] = sub nsw i32 [[DIV25]], 1 9954 // CHECK11-NEXT: store i32 [[SUB26]], i32* [[DOTCAPTURE_EXPR_23]], align 4 9955 // CHECK11-NEXT: [[TMP120:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_23]], align 4 9956 // CHECK11-NEXT: [[ADD27:%.*]] = add nsw i32 [[TMP120]], 1 9957 // CHECK11-NEXT: [[TMP121:%.*]] = zext i32 [[ADD27]] to i64 9958 // CHECK11-NEXT: [[KERNEL_ARGS28:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS]], align 8 9959 // CHECK11-NEXT: [[TMP122:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS28]], i32 0, i32 0 9960 // CHECK11-NEXT: store i32 1, i32* [[TMP122]], align 4 9961 // CHECK11-NEXT: [[TMP123:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS28]], i32 0, i32 1 9962 // CHECK11-NEXT: store i32 5, i32* [[TMP123]], align 4 9963 // CHECK11-NEXT: [[TMP124:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS28]], i32 0, i32 2 9964 // CHECK11-NEXT: store i8** [[TMP116]], i8*** [[TMP124]], align 4 9965 // CHECK11-NEXT: [[TMP125:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS28]], i32 0, i32 3 9966 // CHECK11-NEXT: store i8** [[TMP117]], i8*** [[TMP125]], align 4 9967 // CHECK11-NEXT: [[TMP126:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS28]], i32 0, i32 4 9968 // CHECK11-NEXT: store i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_sizes.8, i32 0, i32 0), i64** [[TMP126]], align 4 9969 // CHECK11-NEXT: [[TMP127:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS28]], i32 0, i32 5 9970 // CHECK11-NEXT: store i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_maptypes.9, i32 0, i32 0), i64** [[TMP127]], align 4 9971 // CHECK11-NEXT: [[TMP128:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS28]], i32 0, i32 6 9972 // CHECK11-NEXT: store i8** null, i8*** [[TMP128]], align 4 9973 // CHECK11-NEXT: [[TMP129:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS28]], i32 0, i32 7 9974 // CHECK11-NEXT: store i8** null, i8*** [[TMP129]], align 4 9975 // CHECK11-NEXT: [[TMP130:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS28]], i32 0, i32 8 9976 // CHECK11-NEXT: store i64 [[TMP121]], i64* [[TMP130]], align 8 9977 // CHECK11-NEXT: [[TMP131:%.*]] = call i32 @__tgt_target_kernel(%struct.ident_t* @[[GLOB3]], i64 -1, i32 0, i32 1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l446.region_id, %struct.__tgt_kernel_arguments* [[KERNEL_ARGS28]]) 9978 // CHECK11-NEXT: [[TMP132:%.*]] = icmp ne i32 [[TMP131]], 0 9979 // CHECK11-NEXT: br i1 [[TMP132]], label [[OMP_OFFLOAD_FAILED29:%.*]], label [[OMP_OFFLOAD_CONT30:%.*]] 9980 // CHECK11: omp_offload.failed29: 9981 // CHECK11-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l446(i32 [[TMP85]], i32 [[TMP87]], double* [[TMP88]], double* [[TMP89]], double* [[TMP90]]) #[[ATTR2]] 9982 // CHECK11-NEXT: br label [[OMP_OFFLOAD_CONT30]] 9983 // CHECK11: omp_offload.cont30: 9984 // CHECK11-NEXT: [[TMP133:%.*]] = load i32, i32* [[N]], align 4 9985 // CHECK11-NEXT: store i32 [[TMP133]], i32* [[N_CASTED31]], align 4 9986 // CHECK11-NEXT: [[TMP134:%.*]] = load i32, i32* [[N_CASTED31]], align 4 9987 // CHECK11-NEXT: [[TMP135:%.*]] = load double*, double** [[A]], align 4 9988 // CHECK11-NEXT: [[TMP136:%.*]] = load double*, double** [[B]], align 4 9989 // CHECK11-NEXT: [[TMP137:%.*]] = load double*, double** [[C]], align 4 9990 // CHECK11-NEXT: [[TMP138:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS32]], i32 0, i32 0 9991 // CHECK11-NEXT: [[TMP139:%.*]] = bitcast i8** [[TMP138]] to i32* 9992 // CHECK11-NEXT: store i32 [[TMP134]], i32* [[TMP139]], align 4 9993 // CHECK11-NEXT: [[TMP140:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS33]], i32 0, i32 0 9994 // CHECK11-NEXT: [[TMP141:%.*]] = bitcast i8** [[TMP140]] to i32* 9995 // CHECK11-NEXT: store i32 [[TMP134]], i32* [[TMP141]], align 4 9996 // CHECK11-NEXT: [[TMP142:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS34]], i32 0, i32 0 9997 // CHECK11-NEXT: store i8* null, i8** [[TMP142]], align 4 9998 // CHECK11-NEXT: [[TMP143:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS32]], i32 0, i32 1 9999 // CHECK11-NEXT: [[TMP144:%.*]] = bitcast i8** [[TMP143]] to double** 10000 // CHECK11-NEXT: store double* [[TMP135]], double** [[TMP144]], align 4 10001 // CHECK11-NEXT: [[TMP145:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS33]], i32 0, i32 1 10002 // CHECK11-NEXT: [[TMP146:%.*]] = bitcast i8** [[TMP145]] to double** 10003 // CHECK11-NEXT: store double* [[TMP135]], double** [[TMP146]], align 4 10004 // CHECK11-NEXT: [[TMP147:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS34]], i32 0, i32 1 10005 // CHECK11-NEXT: store i8* null, i8** [[TMP147]], align 4 10006 // CHECK11-NEXT: [[TMP148:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS32]], i32 0, i32 2 10007 // CHECK11-NEXT: [[TMP149:%.*]] = bitcast i8** [[TMP148]] to double** 10008 // CHECK11-NEXT: store double* [[TMP136]], double** [[TMP149]], align 4 10009 // CHECK11-NEXT: [[TMP150:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS33]], i32 0, i32 2 10010 // CHECK11-NEXT: [[TMP151:%.*]] = bitcast i8** [[TMP150]] to double** 10011 // CHECK11-NEXT: store double* [[TMP136]], double** [[TMP151]], align 4 10012 // CHECK11-NEXT: [[TMP152:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS34]], i32 0, i32 2 10013 // CHECK11-NEXT: store i8* null, i8** [[TMP152]], align 4 10014 // CHECK11-NEXT: [[TMP153:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS32]], i32 0, i32 3 10015 // CHECK11-NEXT: [[TMP154:%.*]] = bitcast i8** [[TMP153]] to double** 10016 // CHECK11-NEXT: store double* [[TMP137]], double** [[TMP154]], align 4 10017 // CHECK11-NEXT: [[TMP155:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS33]], i32 0, i32 3 10018 // CHECK11-NEXT: [[TMP156:%.*]] = bitcast i8** [[TMP155]] to double** 10019 // CHECK11-NEXT: store double* [[TMP137]], double** [[TMP156]], align 4 10020 // CHECK11-NEXT: [[TMP157:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS34]], i32 0, i32 3 10021 // CHECK11-NEXT: store i8* null, i8** [[TMP157]], align 4 10022 // CHECK11-NEXT: [[TMP158:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS32]], i32 0, i32 0 10023 // CHECK11-NEXT: [[TMP159:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS33]], i32 0, i32 0 10024 // CHECK11-NEXT: [[TMP160:%.*]] = load i32, i32* [[N]], align 4 10025 // CHECK11-NEXT: store i32 [[TMP160]], i32* [[DOTCAPTURE_EXPR_36]], align 4 10026 // CHECK11-NEXT: [[TMP161:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_36]], align 4 10027 // CHECK11-NEXT: [[SUB38:%.*]] = sub nsw i32 [[TMP161]], 0 10028 // CHECK11-NEXT: [[DIV39:%.*]] = sdiv i32 [[SUB38]], 1 10029 // CHECK11-NEXT: [[SUB40:%.*]] = sub nsw i32 [[DIV39]], 1 10030 // CHECK11-NEXT: store i32 [[SUB40]], i32* [[DOTCAPTURE_EXPR_37]], align 4 10031 // CHECK11-NEXT: [[TMP162:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_37]], align 4 10032 // CHECK11-NEXT: [[ADD41:%.*]] = add nsw i32 [[TMP162]], 1 10033 // CHECK11-NEXT: [[TMP163:%.*]] = zext i32 [[ADD41]] to i64 10034 // CHECK11-NEXT: [[KERNEL_ARGS42:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS]], align 8 10035 // CHECK11-NEXT: [[TMP164:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS42]], i32 0, i32 0 10036 // CHECK11-NEXT: store i32 1, i32* [[TMP164]], align 4 10037 // CHECK11-NEXT: [[TMP165:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS42]], i32 0, i32 1 10038 // CHECK11-NEXT: store i32 4, i32* [[TMP165]], align 4 10039 // CHECK11-NEXT: [[TMP166:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS42]], i32 0, i32 2 10040 // CHECK11-NEXT: store i8** [[TMP158]], i8*** [[TMP166]], align 4 10041 // CHECK11-NEXT: [[TMP167:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS42]], i32 0, i32 3 10042 // CHECK11-NEXT: store i8** [[TMP159]], i8*** [[TMP167]], align 4 10043 // CHECK11-NEXT: [[TMP168:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS42]], i32 0, i32 4 10044 // CHECK11-NEXT: store i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_sizes.12, i32 0, i32 0), i64** [[TMP168]], align 4 10045 // CHECK11-NEXT: [[TMP169:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS42]], i32 0, i32 5 10046 // CHECK11-NEXT: store i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_maptypes.13, i32 0, i32 0), i64** [[TMP169]], align 4 10047 // CHECK11-NEXT: [[TMP170:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS42]], i32 0, i32 6 10048 // CHECK11-NEXT: store i8** null, i8*** [[TMP170]], align 4 10049 // CHECK11-NEXT: [[TMP171:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS42]], i32 0, i32 7 10050 // CHECK11-NEXT: store i8** null, i8*** [[TMP171]], align 4 10051 // CHECK11-NEXT: [[TMP172:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS42]], i32 0, i32 8 10052 // CHECK11-NEXT: store i64 [[TMP163]], i64* [[TMP172]], align 8 10053 // CHECK11-NEXT: [[TMP173:%.*]] = call i32 @__tgt_target_kernel(%struct.ident_t* @[[GLOB3]], i64 -1, i32 0, i32 1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l477.region_id, %struct.__tgt_kernel_arguments* [[KERNEL_ARGS42]]) 10054 // CHECK11-NEXT: [[TMP174:%.*]] = icmp ne i32 [[TMP173]], 0 10055 // CHECK11-NEXT: br i1 [[TMP174]], label [[OMP_OFFLOAD_FAILED43:%.*]], label [[OMP_OFFLOAD_CONT44:%.*]] 10056 // CHECK11: omp_offload.failed43: 10057 // CHECK11-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l477(i32 [[TMP134]], double* [[TMP135]], double* [[TMP136]], double* [[TMP137]]) #[[ATTR2]] 10058 // CHECK11-NEXT: br label [[OMP_OFFLOAD_CONT44]] 10059 // CHECK11: omp_offload.cont44: 10060 // CHECK11-NEXT: [[TMP175:%.*]] = load i32, i32* [[CH]], align 4 10061 // CHECK11-NEXT: store i32 [[TMP175]], i32* [[CH_CASTED45]], align 4 10062 // CHECK11-NEXT: [[TMP176:%.*]] = load i32, i32* [[CH_CASTED45]], align 4 10063 // CHECK11-NEXT: [[TMP177:%.*]] = load i32, i32* [[N]], align 4 10064 // CHECK11-NEXT: store i32 [[TMP177]], i32* [[N_CASTED46]], align 4 10065 // CHECK11-NEXT: [[TMP178:%.*]] = load i32, i32* [[N_CASTED46]], align 4 10066 // CHECK11-NEXT: [[TMP179:%.*]] = load double*, double** [[A]], align 4 10067 // CHECK11-NEXT: [[TMP180:%.*]] = load double*, double** [[B]], align 4 10068 // CHECK11-NEXT: [[TMP181:%.*]] = load double*, double** [[C]], align 4 10069 // CHECK11-NEXT: [[TMP182:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS47]], i32 0, i32 0 10070 // CHECK11-NEXT: [[TMP183:%.*]] = bitcast i8** [[TMP182]] to i32* 10071 // CHECK11-NEXT: store i32 [[TMP176]], i32* [[TMP183]], align 4 10072 // CHECK11-NEXT: [[TMP184:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS48]], i32 0, i32 0 10073 // CHECK11-NEXT: [[TMP185:%.*]] = bitcast i8** [[TMP184]] to i32* 10074 // CHECK11-NEXT: store i32 [[TMP176]], i32* [[TMP185]], align 4 10075 // CHECK11-NEXT: [[TMP186:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS49]], i32 0, i32 0 10076 // CHECK11-NEXT: store i8* null, i8** [[TMP186]], align 4 10077 // CHECK11-NEXT: [[TMP187:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS47]], i32 0, i32 1 10078 // CHECK11-NEXT: [[TMP188:%.*]] = bitcast i8** [[TMP187]] to i32* 10079 // CHECK11-NEXT: store i32 [[TMP178]], i32* [[TMP188]], align 4 10080 // CHECK11-NEXT: [[TMP189:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS48]], i32 0, i32 1 10081 // CHECK11-NEXT: [[TMP190:%.*]] = bitcast i8** [[TMP189]] to i32* 10082 // CHECK11-NEXT: store i32 [[TMP178]], i32* [[TMP190]], align 4 10083 // CHECK11-NEXT: [[TMP191:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS49]], i32 0, i32 1 10084 // CHECK11-NEXT: store i8* null, i8** [[TMP191]], align 4 10085 // CHECK11-NEXT: [[TMP192:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS47]], i32 0, i32 2 10086 // CHECK11-NEXT: [[TMP193:%.*]] = bitcast i8** [[TMP192]] to double** 10087 // CHECK11-NEXT: store double* [[TMP179]], double** [[TMP193]], align 4 10088 // CHECK11-NEXT: [[TMP194:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS48]], i32 0, i32 2 10089 // CHECK11-NEXT: [[TMP195:%.*]] = bitcast i8** [[TMP194]] to double** 10090 // CHECK11-NEXT: store double* [[TMP179]], double** [[TMP195]], align 4 10091 // CHECK11-NEXT: [[TMP196:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS49]], i32 0, i32 2 10092 // CHECK11-NEXT: store i8* null, i8** [[TMP196]], align 4 10093 // CHECK11-NEXT: [[TMP197:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS47]], i32 0, i32 3 10094 // CHECK11-NEXT: [[TMP198:%.*]] = bitcast i8** [[TMP197]] to double** 10095 // CHECK11-NEXT: store double* [[TMP180]], double** [[TMP198]], align 4 10096 // CHECK11-NEXT: [[TMP199:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS48]], i32 0, i32 3 10097 // CHECK11-NEXT: [[TMP200:%.*]] = bitcast i8** [[TMP199]] to double** 10098 // CHECK11-NEXT: store double* [[TMP180]], double** [[TMP200]], align 4 10099 // CHECK11-NEXT: [[TMP201:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS49]], i32 0, i32 3 10100 // CHECK11-NEXT: store i8* null, i8** [[TMP201]], align 4 10101 // CHECK11-NEXT: [[TMP202:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS47]], i32 0, i32 4 10102 // CHECK11-NEXT: [[TMP203:%.*]] = bitcast i8** [[TMP202]] to double** 10103 // CHECK11-NEXT: store double* [[TMP181]], double** [[TMP203]], align 4 10104 // CHECK11-NEXT: [[TMP204:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS48]], i32 0, i32 4 10105 // CHECK11-NEXT: [[TMP205:%.*]] = bitcast i8** [[TMP204]] to double** 10106 // CHECK11-NEXT: store double* [[TMP181]], double** [[TMP205]], align 4 10107 // CHECK11-NEXT: [[TMP206:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS49]], i32 0, i32 4 10108 // CHECK11-NEXT: store i8* null, i8** [[TMP206]], align 4 10109 // CHECK11-NEXT: [[TMP207:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS47]], i32 0, i32 0 10110 // CHECK11-NEXT: [[TMP208:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS48]], i32 0, i32 0 10111 // CHECK11-NEXT: [[TMP209:%.*]] = load i32, i32* [[N]], align 4 10112 // CHECK11-NEXT: store i32 [[TMP209]], i32* [[DOTCAPTURE_EXPR_51]], align 4 10113 // CHECK11-NEXT: [[TMP210:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_51]], align 4 10114 // CHECK11-NEXT: [[SUB53:%.*]] = sub nsw i32 [[TMP210]], 0 10115 // CHECK11-NEXT: [[DIV54:%.*]] = sdiv i32 [[SUB53]], 1 10116 // CHECK11-NEXT: [[SUB55:%.*]] = sub nsw i32 [[DIV54]], 1 10117 // CHECK11-NEXT: store i32 [[SUB55]], i32* [[DOTCAPTURE_EXPR_52]], align 4 10118 // CHECK11-NEXT: [[TMP211:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_52]], align 4 10119 // CHECK11-NEXT: [[ADD56:%.*]] = add nsw i32 [[TMP211]], 1 10120 // CHECK11-NEXT: [[TMP212:%.*]] = zext i32 [[ADD56]] to i64 10121 // CHECK11-NEXT: [[KERNEL_ARGS57:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS]], align 8 10122 // CHECK11-NEXT: [[TMP213:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS57]], i32 0, i32 0 10123 // CHECK11-NEXT: store i32 1, i32* [[TMP213]], align 4 10124 // CHECK11-NEXT: [[TMP214:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS57]], i32 0, i32 1 10125 // CHECK11-NEXT: store i32 5, i32* [[TMP214]], align 4 10126 // CHECK11-NEXT: [[TMP215:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS57]], i32 0, i32 2 10127 // CHECK11-NEXT: store i8** [[TMP207]], i8*** [[TMP215]], align 4 10128 // CHECK11-NEXT: [[TMP216:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS57]], i32 0, i32 3 10129 // CHECK11-NEXT: store i8** [[TMP208]], i8*** [[TMP216]], align 4 10130 // CHECK11-NEXT: [[TMP217:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS57]], i32 0, i32 4 10131 // CHECK11-NEXT: store i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_sizes.16, i32 0, i32 0), i64** [[TMP217]], align 4 10132 // CHECK11-NEXT: [[TMP218:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS57]], i32 0, i32 5 10133 // CHECK11-NEXT: store i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_maptypes.17, i32 0, i32 0), i64** [[TMP218]], align 4 10134 // CHECK11-NEXT: [[TMP219:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS57]], i32 0, i32 6 10135 // CHECK11-NEXT: store i8** null, i8*** [[TMP219]], align 4 10136 // CHECK11-NEXT: [[TMP220:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS57]], i32 0, i32 7 10137 // CHECK11-NEXT: store i8** null, i8*** [[TMP220]], align 4 10138 // CHECK11-NEXT: [[TMP221:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS57]], i32 0, i32 8 10139 // CHECK11-NEXT: store i64 [[TMP212]], i64* [[TMP221]], align 8 10140 // CHECK11-NEXT: [[TMP222:%.*]] = call i32 @__tgt_target_kernel(%struct.ident_t* @[[GLOB3]], i64 -1, i32 0, i32 1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l505.region_id, %struct.__tgt_kernel_arguments* [[KERNEL_ARGS57]]) 10141 // CHECK11-NEXT: [[TMP223:%.*]] = icmp ne i32 [[TMP222]], 0 10142 // CHECK11-NEXT: br i1 [[TMP223]], label [[OMP_OFFLOAD_FAILED58:%.*]], label [[OMP_OFFLOAD_CONT59:%.*]] 10143 // CHECK11: omp_offload.failed58: 10144 // CHECK11-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l505(i32 [[TMP176]], i32 [[TMP178]], double* [[TMP179]], double* [[TMP180]], double* [[TMP181]]) #[[ATTR2]] 10145 // CHECK11-NEXT: br label [[OMP_OFFLOAD_CONT59]] 10146 // CHECK11: omp_offload.cont59: 10147 // CHECK11-NEXT: [[TMP224:%.*]] = load i32, i32* [[N]], align 4 10148 // CHECK11-NEXT: store i32 [[TMP224]], i32* [[N_CASTED60]], align 4 10149 // CHECK11-NEXT: [[TMP225:%.*]] = load i32, i32* [[N_CASTED60]], align 4 10150 // CHECK11-NEXT: [[TMP226:%.*]] = load double*, double** [[A]], align 4 10151 // CHECK11-NEXT: [[TMP227:%.*]] = load double*, double** [[B]], align 4 10152 // CHECK11-NEXT: [[TMP228:%.*]] = load double*, double** [[C]], align 4 10153 // CHECK11-NEXT: [[TMP229:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS61]], i32 0, i32 0 10154 // CHECK11-NEXT: [[TMP230:%.*]] = bitcast i8** [[TMP229]] to i32* 10155 // CHECK11-NEXT: store i32 [[TMP225]], i32* [[TMP230]], align 4 10156 // CHECK11-NEXT: [[TMP231:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS62]], i32 0, i32 0 10157 // CHECK11-NEXT: [[TMP232:%.*]] = bitcast i8** [[TMP231]] to i32* 10158 // CHECK11-NEXT: store i32 [[TMP225]], i32* [[TMP232]], align 4 10159 // CHECK11-NEXT: [[TMP233:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS63]], i32 0, i32 0 10160 // CHECK11-NEXT: store i8* null, i8** [[TMP233]], align 4 10161 // CHECK11-NEXT: [[TMP234:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS61]], i32 0, i32 1 10162 // CHECK11-NEXT: [[TMP235:%.*]] = bitcast i8** [[TMP234]] to double** 10163 // CHECK11-NEXT: store double* [[TMP226]], double** [[TMP235]], align 4 10164 // CHECK11-NEXT: [[TMP236:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS62]], i32 0, i32 1 10165 // CHECK11-NEXT: [[TMP237:%.*]] = bitcast i8** [[TMP236]] to double** 10166 // CHECK11-NEXT: store double* [[TMP226]], double** [[TMP237]], align 4 10167 // CHECK11-NEXT: [[TMP238:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS63]], i32 0, i32 1 10168 // CHECK11-NEXT: store i8* null, i8** [[TMP238]], align 4 10169 // CHECK11-NEXT: [[TMP239:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS61]], i32 0, i32 2 10170 // CHECK11-NEXT: [[TMP240:%.*]] = bitcast i8** [[TMP239]] to double** 10171 // CHECK11-NEXT: store double* [[TMP227]], double** [[TMP240]], align 4 10172 // CHECK11-NEXT: [[TMP241:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS62]], i32 0, i32 2 10173 // CHECK11-NEXT: [[TMP242:%.*]] = bitcast i8** [[TMP241]] to double** 10174 // CHECK11-NEXT: store double* [[TMP227]], double** [[TMP242]], align 4 10175 // CHECK11-NEXT: [[TMP243:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS63]], i32 0, i32 2 10176 // CHECK11-NEXT: store i8* null, i8** [[TMP243]], align 4 10177 // CHECK11-NEXT: [[TMP244:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS61]], i32 0, i32 3 10178 // CHECK11-NEXT: [[TMP245:%.*]] = bitcast i8** [[TMP244]] to double** 10179 // CHECK11-NEXT: store double* [[TMP228]], double** [[TMP245]], align 4 10180 // CHECK11-NEXT: [[TMP246:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS62]], i32 0, i32 3 10181 // CHECK11-NEXT: [[TMP247:%.*]] = bitcast i8** [[TMP246]] to double** 10182 // CHECK11-NEXT: store double* [[TMP228]], double** [[TMP247]], align 4 10183 // CHECK11-NEXT: [[TMP248:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS63]], i32 0, i32 3 10184 // CHECK11-NEXT: store i8* null, i8** [[TMP248]], align 4 10185 // CHECK11-NEXT: [[TMP249:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS61]], i32 0, i32 0 10186 // CHECK11-NEXT: [[TMP250:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS62]], i32 0, i32 0 10187 // CHECK11-NEXT: [[TMP251:%.*]] = load i32, i32* [[N]], align 4 10188 // CHECK11-NEXT: store i32 [[TMP251]], i32* [[DOTCAPTURE_EXPR_65]], align 4 10189 // CHECK11-NEXT: [[TMP252:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_65]], align 4 10190 // CHECK11-NEXT: [[SUB67:%.*]] = sub nsw i32 [[TMP252]], 0 10191 // CHECK11-NEXT: [[DIV68:%.*]] = sdiv i32 [[SUB67]], 1 10192 // CHECK11-NEXT: [[SUB69:%.*]] = sub nsw i32 [[DIV68]], 1 10193 // CHECK11-NEXT: store i32 [[SUB69]], i32* [[DOTCAPTURE_EXPR_66]], align 4 10194 // CHECK11-NEXT: [[TMP253:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_66]], align 4 10195 // CHECK11-NEXT: [[ADD70:%.*]] = add nsw i32 [[TMP253]], 1 10196 // CHECK11-NEXT: [[TMP254:%.*]] = zext i32 [[ADD70]] to i64 10197 // CHECK11-NEXT: [[KERNEL_ARGS71:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS]], align 8 10198 // CHECK11-NEXT: [[TMP255:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS71]], i32 0, i32 0 10199 // CHECK11-NEXT: store i32 1, i32* [[TMP255]], align 4 10200 // CHECK11-NEXT: [[TMP256:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS71]], i32 0, i32 1 10201 // CHECK11-NEXT: store i32 4, i32* [[TMP256]], align 4 10202 // CHECK11-NEXT: [[TMP257:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS71]], i32 0, i32 2 10203 // CHECK11-NEXT: store i8** [[TMP249]], i8*** [[TMP257]], align 4 10204 // CHECK11-NEXT: [[TMP258:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS71]], i32 0, i32 3 10205 // CHECK11-NEXT: store i8** [[TMP250]], i8*** [[TMP258]], align 4 10206 // CHECK11-NEXT: [[TMP259:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS71]], i32 0, i32 4 10207 // CHECK11-NEXT: store i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_sizes.20, i32 0, i32 0), i64** [[TMP259]], align 4 10208 // CHECK11-NEXT: [[TMP260:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS71]], i32 0, i32 5 10209 // CHECK11-NEXT: store i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_maptypes.21, i32 0, i32 0), i64** [[TMP260]], align 4 10210 // CHECK11-NEXT: [[TMP261:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS71]], i32 0, i32 6 10211 // CHECK11-NEXT: store i8** null, i8*** [[TMP261]], align 4 10212 // CHECK11-NEXT: [[TMP262:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS71]], i32 0, i32 7 10213 // CHECK11-NEXT: store i8** null, i8*** [[TMP262]], align 4 10214 // CHECK11-NEXT: [[TMP263:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS71]], i32 0, i32 8 10215 // CHECK11-NEXT: store i64 [[TMP254]], i64* [[TMP263]], align 8 10216 // CHECK11-NEXT: [[TMP264:%.*]] = call i32 @__tgt_target_kernel(%struct.ident_t* @[[GLOB3]], i64 -1, i32 0, i32 1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l535.region_id, %struct.__tgt_kernel_arguments* [[KERNEL_ARGS71]]) 10217 // CHECK11-NEXT: [[TMP265:%.*]] = icmp ne i32 [[TMP264]], 0 10218 // CHECK11-NEXT: br i1 [[TMP265]], label [[OMP_OFFLOAD_FAILED72:%.*]], label [[OMP_OFFLOAD_CONT73:%.*]] 10219 // CHECK11: omp_offload.failed72: 10220 // CHECK11-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l535(i32 [[TMP225]], double* [[TMP226]], double* [[TMP227]], double* [[TMP228]]) #[[ATTR2]] 10221 // CHECK11-NEXT: br label [[OMP_OFFLOAD_CONT73]] 10222 // CHECK11: omp_offload.cont73: 10223 // CHECK11-NEXT: [[TMP266:%.*]] = load i32, i32* [[CH]], align 4 10224 // CHECK11-NEXT: store i32 [[TMP266]], i32* [[CH_CASTED74]], align 4 10225 // CHECK11-NEXT: [[TMP267:%.*]] = load i32, i32* [[CH_CASTED74]], align 4 10226 // CHECK11-NEXT: [[TMP268:%.*]] = load i32, i32* [[N]], align 4 10227 // CHECK11-NEXT: store i32 [[TMP268]], i32* [[N_CASTED75]], align 4 10228 // CHECK11-NEXT: [[TMP269:%.*]] = load i32, i32* [[N_CASTED75]], align 4 10229 // CHECK11-NEXT: [[TMP270:%.*]] = load double*, double** [[A]], align 4 10230 // CHECK11-NEXT: [[TMP271:%.*]] = load double*, double** [[B]], align 4 10231 // CHECK11-NEXT: [[TMP272:%.*]] = load double*, double** [[C]], align 4 10232 // CHECK11-NEXT: [[TMP273:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS76]], i32 0, i32 0 10233 // CHECK11-NEXT: [[TMP274:%.*]] = bitcast i8** [[TMP273]] to i32* 10234 // CHECK11-NEXT: store i32 [[TMP267]], i32* [[TMP274]], align 4 10235 // CHECK11-NEXT: [[TMP275:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS77]], i32 0, i32 0 10236 // CHECK11-NEXT: [[TMP276:%.*]] = bitcast i8** [[TMP275]] to i32* 10237 // CHECK11-NEXT: store i32 [[TMP267]], i32* [[TMP276]], align 4 10238 // CHECK11-NEXT: [[TMP277:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS78]], i32 0, i32 0 10239 // CHECK11-NEXT: store i8* null, i8** [[TMP277]], align 4 10240 // CHECK11-NEXT: [[TMP278:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS76]], i32 0, i32 1 10241 // CHECK11-NEXT: [[TMP279:%.*]] = bitcast i8** [[TMP278]] to i32* 10242 // CHECK11-NEXT: store i32 [[TMP269]], i32* [[TMP279]], align 4 10243 // CHECK11-NEXT: [[TMP280:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS77]], i32 0, i32 1 10244 // CHECK11-NEXT: [[TMP281:%.*]] = bitcast i8** [[TMP280]] to i32* 10245 // CHECK11-NEXT: store i32 [[TMP269]], i32* [[TMP281]], align 4 10246 // CHECK11-NEXT: [[TMP282:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS78]], i32 0, i32 1 10247 // CHECK11-NEXT: store i8* null, i8** [[TMP282]], align 4 10248 // CHECK11-NEXT: [[TMP283:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS76]], i32 0, i32 2 10249 // CHECK11-NEXT: [[TMP284:%.*]] = bitcast i8** [[TMP283]] to double** 10250 // CHECK11-NEXT: store double* [[TMP270]], double** [[TMP284]], align 4 10251 // CHECK11-NEXT: [[TMP285:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS77]], i32 0, i32 2 10252 // CHECK11-NEXT: [[TMP286:%.*]] = bitcast i8** [[TMP285]] to double** 10253 // CHECK11-NEXT: store double* [[TMP270]], double** [[TMP286]], align 4 10254 // CHECK11-NEXT: [[TMP287:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS78]], i32 0, i32 2 10255 // CHECK11-NEXT: store i8* null, i8** [[TMP287]], align 4 10256 // CHECK11-NEXT: [[TMP288:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS76]], i32 0, i32 3 10257 // CHECK11-NEXT: [[TMP289:%.*]] = bitcast i8** [[TMP288]] to double** 10258 // CHECK11-NEXT: store double* [[TMP271]], double** [[TMP289]], align 4 10259 // CHECK11-NEXT: [[TMP290:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS77]], i32 0, i32 3 10260 // CHECK11-NEXT: [[TMP291:%.*]] = bitcast i8** [[TMP290]] to double** 10261 // CHECK11-NEXT: store double* [[TMP271]], double** [[TMP291]], align 4 10262 // CHECK11-NEXT: [[TMP292:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS78]], i32 0, i32 3 10263 // CHECK11-NEXT: store i8* null, i8** [[TMP292]], align 4 10264 // CHECK11-NEXT: [[TMP293:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS76]], i32 0, i32 4 10265 // CHECK11-NEXT: [[TMP294:%.*]] = bitcast i8** [[TMP293]] to double** 10266 // CHECK11-NEXT: store double* [[TMP272]], double** [[TMP294]], align 4 10267 // CHECK11-NEXT: [[TMP295:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS77]], i32 0, i32 4 10268 // CHECK11-NEXT: [[TMP296:%.*]] = bitcast i8** [[TMP295]] to double** 10269 // CHECK11-NEXT: store double* [[TMP272]], double** [[TMP296]], align 4 10270 // CHECK11-NEXT: [[TMP297:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS78]], i32 0, i32 4 10271 // CHECK11-NEXT: store i8* null, i8** [[TMP297]], align 4 10272 // CHECK11-NEXT: [[TMP298:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS76]], i32 0, i32 0 10273 // CHECK11-NEXT: [[TMP299:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS77]], i32 0, i32 0 10274 // CHECK11-NEXT: [[TMP300:%.*]] = load i32, i32* [[N]], align 4 10275 // CHECK11-NEXT: store i32 [[TMP300]], i32* [[DOTCAPTURE_EXPR_80]], align 4 10276 // CHECK11-NEXT: [[TMP301:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_80]], align 4 10277 // CHECK11-NEXT: [[SUB82:%.*]] = sub nsw i32 [[TMP301]], 0 10278 // CHECK11-NEXT: [[DIV83:%.*]] = sdiv i32 [[SUB82]], 1 10279 // CHECK11-NEXT: [[SUB84:%.*]] = sub nsw i32 [[DIV83]], 1 10280 // CHECK11-NEXT: store i32 [[SUB84]], i32* [[DOTCAPTURE_EXPR_81]], align 4 10281 // CHECK11-NEXT: [[TMP302:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_81]], align 4 10282 // CHECK11-NEXT: [[ADD85:%.*]] = add nsw i32 [[TMP302]], 1 10283 // CHECK11-NEXT: [[TMP303:%.*]] = zext i32 [[ADD85]] to i64 10284 // CHECK11-NEXT: [[KERNEL_ARGS86:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS]], align 8 10285 // CHECK11-NEXT: [[TMP304:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS86]], i32 0, i32 0 10286 // CHECK11-NEXT: store i32 1, i32* [[TMP304]], align 4 10287 // CHECK11-NEXT: [[TMP305:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS86]], i32 0, i32 1 10288 // CHECK11-NEXT: store i32 5, i32* [[TMP305]], align 4 10289 // CHECK11-NEXT: [[TMP306:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS86]], i32 0, i32 2 10290 // CHECK11-NEXT: store i8** [[TMP298]], i8*** [[TMP306]], align 4 10291 // CHECK11-NEXT: [[TMP307:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS86]], i32 0, i32 3 10292 // CHECK11-NEXT: store i8** [[TMP299]], i8*** [[TMP307]], align 4 10293 // CHECK11-NEXT: [[TMP308:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS86]], i32 0, i32 4 10294 // CHECK11-NEXT: store i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_sizes.24, i32 0, i32 0), i64** [[TMP308]], align 4 10295 // CHECK11-NEXT: [[TMP309:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS86]], i32 0, i32 5 10296 // CHECK11-NEXT: store i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_maptypes.25, i32 0, i32 0), i64** [[TMP309]], align 4 10297 // CHECK11-NEXT: [[TMP310:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS86]], i32 0, i32 6 10298 // CHECK11-NEXT: store i8** null, i8*** [[TMP310]], align 4 10299 // CHECK11-NEXT: [[TMP311:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS86]], i32 0, i32 7 10300 // CHECK11-NEXT: store i8** null, i8*** [[TMP311]], align 4 10301 // CHECK11-NEXT: [[TMP312:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS86]], i32 0, i32 8 10302 // CHECK11-NEXT: store i64 [[TMP303]], i64* [[TMP312]], align 8 10303 // CHECK11-NEXT: [[TMP313:%.*]] = call i32 @__tgt_target_kernel(%struct.ident_t* @[[GLOB3]], i64 -1, i32 0, i32 1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l561.region_id, %struct.__tgt_kernel_arguments* [[KERNEL_ARGS86]]) 10304 // CHECK11-NEXT: [[TMP314:%.*]] = icmp ne i32 [[TMP313]], 0 10305 // CHECK11-NEXT: br i1 [[TMP314]], label [[OMP_OFFLOAD_FAILED87:%.*]], label [[OMP_OFFLOAD_CONT88:%.*]] 10306 // CHECK11: omp_offload.failed87: 10307 // CHECK11-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l561(i32 [[TMP267]], i32 [[TMP269]], double* [[TMP270]], double* [[TMP271]], double* [[TMP272]]) #[[ATTR2]] 10308 // CHECK11-NEXT: br label [[OMP_OFFLOAD_CONT88]] 10309 // CHECK11: omp_offload.cont88: 10310 // CHECK11-NEXT: [[CALL:%.*]] = call noundef i32 @_Z5tmainIiET_v() 10311 // CHECK11-NEXT: ret i32 [[CALL]] 10312 // 10313 // 10314 // CHECK11-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l368 10315 // CHECK11-SAME: (i32 noundef [[N:%.*]], double* noundef [[A:%.*]], double* noundef [[B:%.*]], double* noundef [[C:%.*]]) #[[ATTR1:[0-9]+]] { 10316 // CHECK11-NEXT: entry: 10317 // CHECK11-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4 10318 // CHECK11-NEXT: [[A_ADDR:%.*]] = alloca double*, align 4 10319 // CHECK11-NEXT: [[B_ADDR:%.*]] = alloca double*, align 4 10320 // CHECK11-NEXT: [[C_ADDR:%.*]] = alloca double*, align 4 10321 // CHECK11-NEXT: store i32 [[N]], i32* [[N_ADDR]], align 4 10322 // CHECK11-NEXT: store double* [[A]], double** [[A_ADDR]], align 4 10323 // CHECK11-NEXT: store double* [[B]], double** [[B_ADDR]], align 4 10324 // CHECK11-NEXT: store double* [[C]], double** [[C_ADDR]], align 4 10325 // CHECK11-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 4, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32*, double**, double**, double**)* @.omp_outlined. to void (i32*, i32*, ...)*), i32* [[N_ADDR]], double** [[A_ADDR]], double** [[B_ADDR]], double** [[C_ADDR]]) 10326 // CHECK11-NEXT: ret void 10327 // 10328 // 10329 // CHECK11-LABEL: define {{[^@]+}}@.omp_outlined. 10330 // CHECK11-SAME: (i32* noalias noundef [[DOTGLOBAL_TID_:%.*]], i32* noalias noundef [[DOTBOUND_TID_:%.*]], i32* noundef nonnull align 4 dereferenceable(4) [[N:%.*]], double** noundef nonnull align 4 dereferenceable(4) [[A:%.*]], double** noundef nonnull align 4 dereferenceable(4) [[B:%.*]], double** noundef nonnull align 4 dereferenceable(4) [[C:%.*]]) #[[ATTR1]] { 10331 // CHECK11-NEXT: entry: 10332 // CHECK11-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4 10333 // CHECK11-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4 10334 // CHECK11-NEXT: [[N_ADDR:%.*]] = alloca i32*, align 4 10335 // CHECK11-NEXT: [[A_ADDR:%.*]] = alloca double**, align 4 10336 // CHECK11-NEXT: [[B_ADDR:%.*]] = alloca double**, align 4 10337 // CHECK11-NEXT: [[C_ADDR:%.*]] = alloca double**, align 4 10338 // CHECK11-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 10339 // CHECK11-NEXT: [[TMP:%.*]] = alloca i32, align 4 10340 // CHECK11-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 10341 // CHECK11-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 10342 // CHECK11-NEXT: [[I:%.*]] = alloca i32, align 4 10343 // CHECK11-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4 10344 // CHECK11-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4 10345 // CHECK11-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 10346 // CHECK11-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 10347 // CHECK11-NEXT: [[I3:%.*]] = alloca i32, align 4 10348 // CHECK11-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 4 10349 // CHECK11-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 4 10350 // CHECK11-NEXT: store i32* [[N]], i32** [[N_ADDR]], align 4 10351 // CHECK11-NEXT: store double** [[A]], double*** [[A_ADDR]], align 4 10352 // CHECK11-NEXT: store double** [[B]], double*** [[B_ADDR]], align 4 10353 // CHECK11-NEXT: store double** [[C]], double*** [[C_ADDR]], align 4 10354 // CHECK11-NEXT: [[TMP0:%.*]] = load i32*, i32** [[N_ADDR]], align 4 10355 // CHECK11-NEXT: [[TMP1:%.*]] = load double**, double*** [[A_ADDR]], align 4 10356 // CHECK11-NEXT: [[TMP2:%.*]] = load double**, double*** [[B_ADDR]], align 4 10357 // CHECK11-NEXT: [[TMP3:%.*]] = load double**, double*** [[C_ADDR]], align 4 10358 // CHECK11-NEXT: [[TMP4:%.*]] = load i32, i32* [[TMP0]], align 4 10359 // CHECK11-NEXT: store i32 [[TMP4]], i32* [[DOTCAPTURE_EXPR_]], align 4 10360 // CHECK11-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 10361 // CHECK11-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP5]], 0 10362 // CHECK11-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 10363 // CHECK11-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1 10364 // CHECK11-NEXT: store i32 [[SUB2]], i32* [[DOTCAPTURE_EXPR_1]], align 4 10365 // CHECK11-NEXT: store i32 0, i32* [[I]], align 4 10366 // CHECK11-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 10367 // CHECK11-NEXT: [[CMP:%.*]] = icmp slt i32 0, [[TMP6]] 10368 // CHECK11-NEXT: br i1 [[CMP]], label [[OMP_PRECOND_THEN:%.*]], label [[OMP_PRECOND_END:%.*]] 10369 // CHECK11: omp.precond.then: 10370 // CHECK11-NEXT: store i32 0, i32* [[DOTOMP_COMB_LB]], align 4 10371 // CHECK11-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 10372 // CHECK11-NEXT: store i32 [[TMP7]], i32* [[DOTOMP_COMB_UB]], align 4 10373 // CHECK11-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4 10374 // CHECK11-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4 10375 // CHECK11-NEXT: [[TMP8:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 4 10376 // CHECK11-NEXT: [[TMP9:%.*]] = load i32, i32* [[TMP8]], align 4 10377 // CHECK11-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1:[0-9]+]], i32 [[TMP9]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_COMB_LB]], i32* [[DOTOMP_COMB_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1) 10378 // CHECK11-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 10379 // CHECK11-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 10380 // CHECK11-NEXT: [[CMP4:%.*]] = icmp sgt i32 [[TMP10]], [[TMP11]] 10381 // CHECK11-NEXT: br i1 [[CMP4]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] 10382 // CHECK11: cond.true: 10383 // CHECK11-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 10384 // CHECK11-NEXT: br label [[COND_END:%.*]] 10385 // CHECK11: cond.false: 10386 // CHECK11-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 10387 // CHECK11-NEXT: br label [[COND_END]] 10388 // CHECK11: cond.end: 10389 // CHECK11-NEXT: [[COND:%.*]] = phi i32 [ [[TMP12]], [[COND_TRUE]] ], [ [[TMP13]], [[COND_FALSE]] ] 10390 // CHECK11-NEXT: store i32 [[COND]], i32* [[DOTOMP_COMB_UB]], align 4 10391 // CHECK11-NEXT: [[TMP14:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4 10392 // CHECK11-NEXT: store i32 [[TMP14]], i32* [[DOTOMP_IV]], align 4 10393 // CHECK11-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] 10394 // CHECK11: omp.inner.for.cond: 10395 // CHECK11-NEXT: [[TMP15:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !18 10396 // CHECK11-NEXT: [[TMP16:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4, !llvm.access.group !18 10397 // CHECK11-NEXT: [[CMP5:%.*]] = icmp sle i32 [[TMP15]], [[TMP16]] 10398 // CHECK11-NEXT: br i1 [[CMP5]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] 10399 // CHECK11: omp.inner.for.body: 10400 // CHECK11-NEXT: [[TMP17:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4, !llvm.access.group !18 10401 // CHECK11-NEXT: [[TMP18:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4, !llvm.access.group !18 10402 // CHECK11-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 6, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32, i32, i32*, double**, double**, double**)* @.omp_outlined..1 to void (i32*, i32*, ...)*), i32 [[TMP17]], i32 [[TMP18]], i32* [[TMP0]], double** [[TMP1]], double** [[TMP2]], double** [[TMP3]]), !llvm.access.group !18 10403 // CHECK11-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] 10404 // CHECK11: omp.inner.for.inc: 10405 // CHECK11-NEXT: [[TMP19:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !18 10406 // CHECK11-NEXT: [[TMP20:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4, !llvm.access.group !18 10407 // CHECK11-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP19]], [[TMP20]] 10408 // CHECK11-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4, !llvm.access.group !18 10409 // CHECK11-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP19:![0-9]+]] 10410 // CHECK11: omp.inner.for.end: 10411 // CHECK11-NEXT: br label [[OMP_LOOP_EXIT:%.*]] 10412 // CHECK11: omp.loop.exit: 10413 // CHECK11-NEXT: [[TMP21:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 4 10414 // CHECK11-NEXT: [[TMP22:%.*]] = load i32, i32* [[TMP21]], align 4 10415 // CHECK11-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP22]]) 10416 // CHECK11-NEXT: [[TMP23:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4 10417 // CHECK11-NEXT: [[TMP24:%.*]] = icmp ne i32 [[TMP23]], 0 10418 // CHECK11-NEXT: br i1 [[TMP24]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]] 10419 // CHECK11: .omp.final.then: 10420 // CHECK11-NEXT: [[TMP25:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 10421 // CHECK11-NEXT: [[SUB6:%.*]] = sub nsw i32 [[TMP25]], 0 10422 // CHECK11-NEXT: [[DIV7:%.*]] = sdiv i32 [[SUB6]], 1 10423 // CHECK11-NEXT: [[MUL:%.*]] = mul nsw i32 [[DIV7]], 1 10424 // CHECK11-NEXT: [[ADD8:%.*]] = add nsw i32 0, [[MUL]] 10425 // CHECK11-NEXT: store i32 [[ADD8]], i32* [[I3]], align 4 10426 // CHECK11-NEXT: br label [[DOTOMP_FINAL_DONE]] 10427 // CHECK11: .omp.final.done: 10428 // CHECK11-NEXT: br label [[OMP_PRECOND_END]] 10429 // CHECK11: omp.precond.end: 10430 // CHECK11-NEXT: ret void 10431 // 10432 // 10433 // CHECK11-LABEL: define {{[^@]+}}@.omp_outlined..1 10434 // CHECK11-SAME: (i32* noalias noundef [[DOTGLOBAL_TID_:%.*]], i32* noalias noundef [[DOTBOUND_TID_:%.*]], i32 noundef [[DOTPREVIOUS_LB_:%.*]], i32 noundef [[DOTPREVIOUS_UB_:%.*]], i32* noundef nonnull align 4 dereferenceable(4) [[N:%.*]], double** noundef nonnull align 4 dereferenceable(4) [[A:%.*]], double** noundef nonnull align 4 dereferenceable(4) [[B:%.*]], double** noundef nonnull align 4 dereferenceable(4) [[C:%.*]]) #[[ATTR1]] { 10435 // CHECK11-NEXT: entry: 10436 // CHECK11-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4 10437 // CHECK11-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4 10438 // CHECK11-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i32, align 4 10439 // CHECK11-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i32, align 4 10440 // CHECK11-NEXT: [[N_ADDR:%.*]] = alloca i32*, align 4 10441 // CHECK11-NEXT: [[A_ADDR:%.*]] = alloca double**, align 4 10442 // CHECK11-NEXT: [[B_ADDR:%.*]] = alloca double**, align 4 10443 // CHECK11-NEXT: [[C_ADDR:%.*]] = alloca double**, align 4 10444 // CHECK11-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 10445 // CHECK11-NEXT: [[TMP:%.*]] = alloca i32, align 4 10446 // CHECK11-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 10447 // CHECK11-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 10448 // CHECK11-NEXT: [[I:%.*]] = alloca i32, align 4 10449 // CHECK11-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 10450 // CHECK11-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4 10451 // CHECK11-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 10452 // CHECK11-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 10453 // CHECK11-NEXT: [[I3:%.*]] = alloca i32, align 4 10454 // CHECK11-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 4 10455 // CHECK11-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 4 10456 // CHECK11-NEXT: store i32 [[DOTPREVIOUS_LB_]], i32* [[DOTPREVIOUS_LB__ADDR]], align 4 10457 // CHECK11-NEXT: store i32 [[DOTPREVIOUS_UB_]], i32* [[DOTPREVIOUS_UB__ADDR]], align 4 10458 // CHECK11-NEXT: store i32* [[N]], i32** [[N_ADDR]], align 4 10459 // CHECK11-NEXT: store double** [[A]], double*** [[A_ADDR]], align 4 10460 // CHECK11-NEXT: store double** [[B]], double*** [[B_ADDR]], align 4 10461 // CHECK11-NEXT: store double** [[C]], double*** [[C_ADDR]], align 4 10462 // CHECK11-NEXT: [[TMP0:%.*]] = load i32*, i32** [[N_ADDR]], align 4 10463 // CHECK11-NEXT: [[TMP1:%.*]] = load double**, double*** [[A_ADDR]], align 4 10464 // CHECK11-NEXT: [[TMP2:%.*]] = load double**, double*** [[B_ADDR]], align 4 10465 // CHECK11-NEXT: [[TMP3:%.*]] = load double**, double*** [[C_ADDR]], align 4 10466 // CHECK11-NEXT: [[TMP4:%.*]] = load i32, i32* [[TMP0]], align 4 10467 // CHECK11-NEXT: store i32 [[TMP4]], i32* [[DOTCAPTURE_EXPR_]], align 4 10468 // CHECK11-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 10469 // CHECK11-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP5]], 0 10470 // CHECK11-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 10471 // CHECK11-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1 10472 // CHECK11-NEXT: store i32 [[SUB2]], i32* [[DOTCAPTURE_EXPR_1]], align 4 10473 // CHECK11-NEXT: store i32 0, i32* [[I]], align 4 10474 // CHECK11-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 10475 // CHECK11-NEXT: [[CMP:%.*]] = icmp slt i32 0, [[TMP6]] 10476 // CHECK11-NEXT: br i1 [[CMP]], label [[OMP_PRECOND_THEN:%.*]], label [[OMP_PRECOND_END:%.*]] 10477 // CHECK11: omp.precond.then: 10478 // CHECK11-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4 10479 // CHECK11-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 10480 // CHECK11-NEXT: store i32 [[TMP7]], i32* [[DOTOMP_UB]], align 4 10481 // CHECK11-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTPREVIOUS_LB__ADDR]], align 4 10482 // CHECK11-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTPREVIOUS_UB__ADDR]], align 4 10483 // CHECK11-NEXT: store i32 [[TMP8]], i32* [[DOTOMP_LB]], align 4 10484 // CHECK11-NEXT: store i32 [[TMP9]], i32* [[DOTOMP_UB]], align 4 10485 // CHECK11-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4 10486 // CHECK11-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4 10487 // CHECK11-NEXT: [[TMP10:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 4 10488 // CHECK11-NEXT: [[TMP11:%.*]] = load i32, i32* [[TMP10]], align 4 10489 // CHECK11-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB2:[0-9]+]], i32 [[TMP11]], i32 34, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1) 10490 // CHECK11-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 10491 // CHECK11-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 10492 // CHECK11-NEXT: [[CMP4:%.*]] = icmp sgt i32 [[TMP12]], [[TMP13]] 10493 // CHECK11-NEXT: br i1 [[CMP4]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] 10494 // CHECK11: cond.true: 10495 // CHECK11-NEXT: [[TMP14:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 10496 // CHECK11-NEXT: br label [[COND_END:%.*]] 10497 // CHECK11: cond.false: 10498 // CHECK11-NEXT: [[TMP15:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 10499 // CHECK11-NEXT: br label [[COND_END]] 10500 // CHECK11: cond.end: 10501 // CHECK11-NEXT: [[COND:%.*]] = phi i32 [ [[TMP14]], [[COND_TRUE]] ], [ [[TMP15]], [[COND_FALSE]] ] 10502 // CHECK11-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4 10503 // CHECK11-NEXT: [[TMP16:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4 10504 // CHECK11-NEXT: store i32 [[TMP16]], i32* [[DOTOMP_IV]], align 4 10505 // CHECK11-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] 10506 // CHECK11: omp.inner.for.cond: 10507 // CHECK11-NEXT: [[TMP17:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !22 10508 // CHECK11-NEXT: [[TMP18:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4, !llvm.access.group !22 10509 // CHECK11-NEXT: [[CMP5:%.*]] = icmp sle i32 [[TMP17]], [[TMP18]] 10510 // CHECK11-NEXT: br i1 [[CMP5]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] 10511 // CHECK11: omp.inner.for.body: 10512 // CHECK11-NEXT: [[TMP19:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !22 10513 // CHECK11-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP19]], 1 10514 // CHECK11-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] 10515 // CHECK11-NEXT: store i32 [[ADD]], i32* [[I3]], align 4, !llvm.access.group !22 10516 // CHECK11-NEXT: [[TMP20:%.*]] = load double*, double** [[TMP2]], align 4, !llvm.access.group !22 10517 // CHECK11-NEXT: [[TMP21:%.*]] = load i32, i32* [[I3]], align 4, !llvm.access.group !22 10518 // CHECK11-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds double, double* [[TMP20]], i32 [[TMP21]] 10519 // CHECK11-NEXT: [[TMP22:%.*]] = load double, double* [[ARRAYIDX]], align 4, !llvm.access.group !22 10520 // CHECK11-NEXT: [[TMP23:%.*]] = load double*, double** [[TMP3]], align 4, !llvm.access.group !22 10521 // CHECK11-NEXT: [[TMP24:%.*]] = load i32, i32* [[I3]], align 4, !llvm.access.group !22 10522 // CHECK11-NEXT: [[ARRAYIDX6:%.*]] = getelementptr inbounds double, double* [[TMP23]], i32 [[TMP24]] 10523 // CHECK11-NEXT: [[TMP25:%.*]] = load double, double* [[ARRAYIDX6]], align 4, !llvm.access.group !22 10524 // CHECK11-NEXT: [[ADD7:%.*]] = fadd double [[TMP22]], [[TMP25]] 10525 // CHECK11-NEXT: [[TMP26:%.*]] = load double*, double** [[TMP1]], align 4, !llvm.access.group !22 10526 // CHECK11-NEXT: [[TMP27:%.*]] = load i32, i32* [[I3]], align 4, !llvm.access.group !22 10527 // CHECK11-NEXT: [[ARRAYIDX8:%.*]] = getelementptr inbounds double, double* [[TMP26]], i32 [[TMP27]] 10528 // CHECK11-NEXT: store double [[ADD7]], double* [[ARRAYIDX8]], align 4, !llvm.access.group !22 10529 // CHECK11-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] 10530 // CHECK11: omp.body.continue: 10531 // CHECK11-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] 10532 // CHECK11: omp.inner.for.inc: 10533 // CHECK11-NEXT: [[TMP28:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !22 10534 // CHECK11-NEXT: [[ADD9:%.*]] = add nsw i32 [[TMP28]], 1 10535 // CHECK11-NEXT: store i32 [[ADD9]], i32* [[DOTOMP_IV]], align 4, !llvm.access.group !22 10536 // CHECK11-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP23:![0-9]+]] 10537 // CHECK11: omp.inner.for.end: 10538 // CHECK11-NEXT: br label [[OMP_LOOP_EXIT:%.*]] 10539 // CHECK11: omp.loop.exit: 10540 // CHECK11-NEXT: [[TMP29:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 4 10541 // CHECK11-NEXT: [[TMP30:%.*]] = load i32, i32* [[TMP29]], align 4 10542 // CHECK11-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP30]]) 10543 // CHECK11-NEXT: [[TMP31:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4 10544 // CHECK11-NEXT: [[TMP32:%.*]] = icmp ne i32 [[TMP31]], 0 10545 // CHECK11-NEXT: br i1 [[TMP32]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]] 10546 // CHECK11: .omp.final.then: 10547 // CHECK11-NEXT: [[TMP33:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 10548 // CHECK11-NEXT: [[SUB10:%.*]] = sub nsw i32 [[TMP33]], 0 10549 // CHECK11-NEXT: [[DIV11:%.*]] = sdiv i32 [[SUB10]], 1 10550 // CHECK11-NEXT: [[MUL12:%.*]] = mul nsw i32 [[DIV11]], 1 10551 // CHECK11-NEXT: [[ADD13:%.*]] = add nsw i32 0, [[MUL12]] 10552 // CHECK11-NEXT: store i32 [[ADD13]], i32* [[I3]], align 4 10553 // CHECK11-NEXT: br label [[DOTOMP_FINAL_DONE]] 10554 // CHECK11: .omp.final.done: 10555 // CHECK11-NEXT: br label [[OMP_PRECOND_END]] 10556 // CHECK11: omp.precond.end: 10557 // CHECK11-NEXT: ret void 10558 // 10559 // 10560 // CHECK11-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l407 10561 // CHECK11-SAME: (i32 noundef [[N:%.*]], double* noundef [[A:%.*]], double* noundef [[B:%.*]], double* noundef [[C:%.*]]) #[[ATTR1]] { 10562 // CHECK11-NEXT: entry: 10563 // CHECK11-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4 10564 // CHECK11-NEXT: [[A_ADDR:%.*]] = alloca double*, align 4 10565 // CHECK11-NEXT: [[B_ADDR:%.*]] = alloca double*, align 4 10566 // CHECK11-NEXT: [[C_ADDR:%.*]] = alloca double*, align 4 10567 // CHECK11-NEXT: store i32 [[N]], i32* [[N_ADDR]], align 4 10568 // CHECK11-NEXT: store double* [[A]], double** [[A_ADDR]], align 4 10569 // CHECK11-NEXT: store double* [[B]], double** [[B_ADDR]], align 4 10570 // CHECK11-NEXT: store double* [[C]], double** [[C_ADDR]], align 4 10571 // CHECK11-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 4, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32*, double**, double**, double**)* @.omp_outlined..2 to void (i32*, i32*, ...)*), i32* [[N_ADDR]], double** [[A_ADDR]], double** [[B_ADDR]], double** [[C_ADDR]]) 10572 // CHECK11-NEXT: ret void 10573 // 10574 // 10575 // CHECK11-LABEL: define {{[^@]+}}@.omp_outlined..2 10576 // CHECK11-SAME: (i32* noalias noundef [[DOTGLOBAL_TID_:%.*]], i32* noalias noundef [[DOTBOUND_TID_:%.*]], i32* noundef nonnull align 4 dereferenceable(4) [[N:%.*]], double** noundef nonnull align 4 dereferenceable(4) [[A:%.*]], double** noundef nonnull align 4 dereferenceable(4) [[B:%.*]], double** noundef nonnull align 4 dereferenceable(4) [[C:%.*]]) #[[ATTR1]] { 10577 // CHECK11-NEXT: entry: 10578 // CHECK11-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4 10579 // CHECK11-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4 10580 // CHECK11-NEXT: [[N_ADDR:%.*]] = alloca i32*, align 4 10581 // CHECK11-NEXT: [[A_ADDR:%.*]] = alloca double**, align 4 10582 // CHECK11-NEXT: [[B_ADDR:%.*]] = alloca double**, align 4 10583 // CHECK11-NEXT: [[C_ADDR:%.*]] = alloca double**, align 4 10584 // CHECK11-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 10585 // CHECK11-NEXT: [[TMP:%.*]] = alloca i32, align 4 10586 // CHECK11-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 10587 // CHECK11-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 10588 // CHECK11-NEXT: [[I:%.*]] = alloca i32, align 4 10589 // CHECK11-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4 10590 // CHECK11-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4 10591 // CHECK11-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 10592 // CHECK11-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 10593 // CHECK11-NEXT: [[I3:%.*]] = alloca i32, align 4 10594 // CHECK11-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 4 10595 // CHECK11-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 4 10596 // CHECK11-NEXT: store i32* [[N]], i32** [[N_ADDR]], align 4 10597 // CHECK11-NEXT: store double** [[A]], double*** [[A_ADDR]], align 4 10598 // CHECK11-NEXT: store double** [[B]], double*** [[B_ADDR]], align 4 10599 // CHECK11-NEXT: store double** [[C]], double*** [[C_ADDR]], align 4 10600 // CHECK11-NEXT: [[TMP0:%.*]] = load i32*, i32** [[N_ADDR]], align 4 10601 // CHECK11-NEXT: [[TMP1:%.*]] = load double**, double*** [[A_ADDR]], align 4 10602 // CHECK11-NEXT: [[TMP2:%.*]] = load double**, double*** [[B_ADDR]], align 4 10603 // CHECK11-NEXT: [[TMP3:%.*]] = load double**, double*** [[C_ADDR]], align 4 10604 // CHECK11-NEXT: [[TMP4:%.*]] = load i32, i32* [[TMP0]], align 4 10605 // CHECK11-NEXT: store i32 [[TMP4]], i32* [[DOTCAPTURE_EXPR_]], align 4 10606 // CHECK11-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 10607 // CHECK11-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP5]], 0 10608 // CHECK11-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 10609 // CHECK11-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1 10610 // CHECK11-NEXT: store i32 [[SUB2]], i32* [[DOTCAPTURE_EXPR_1]], align 4 10611 // CHECK11-NEXT: store i32 0, i32* [[I]], align 4 10612 // CHECK11-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 10613 // CHECK11-NEXT: [[CMP:%.*]] = icmp slt i32 0, [[TMP6]] 10614 // CHECK11-NEXT: br i1 [[CMP]], label [[OMP_PRECOND_THEN:%.*]], label [[OMP_PRECOND_END:%.*]] 10615 // CHECK11: omp.precond.then: 10616 // CHECK11-NEXT: store i32 0, i32* [[DOTOMP_COMB_LB]], align 4 10617 // CHECK11-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 10618 // CHECK11-NEXT: store i32 [[TMP7]], i32* [[DOTOMP_COMB_UB]], align 4 10619 // CHECK11-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4 10620 // CHECK11-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4 10621 // CHECK11-NEXT: [[TMP8:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 4 10622 // CHECK11-NEXT: [[TMP9:%.*]] = load i32, i32* [[TMP8]], align 4 10623 // CHECK11-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1]], i32 [[TMP9]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_COMB_LB]], i32* [[DOTOMP_COMB_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1) 10624 // CHECK11-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 10625 // CHECK11-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 10626 // CHECK11-NEXT: [[CMP4:%.*]] = icmp sgt i32 [[TMP10]], [[TMP11]] 10627 // CHECK11-NEXT: br i1 [[CMP4]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] 10628 // CHECK11: cond.true: 10629 // CHECK11-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 10630 // CHECK11-NEXT: br label [[COND_END:%.*]] 10631 // CHECK11: cond.false: 10632 // CHECK11-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 10633 // CHECK11-NEXT: br label [[COND_END]] 10634 // CHECK11: cond.end: 10635 // CHECK11-NEXT: [[COND:%.*]] = phi i32 [ [[TMP12]], [[COND_TRUE]] ], [ [[TMP13]], [[COND_FALSE]] ] 10636 // CHECK11-NEXT: store i32 [[COND]], i32* [[DOTOMP_COMB_UB]], align 4 10637 // CHECK11-NEXT: [[TMP14:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4 10638 // CHECK11-NEXT: store i32 [[TMP14]], i32* [[DOTOMP_IV]], align 4 10639 // CHECK11-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] 10640 // CHECK11: omp.inner.for.cond: 10641 // CHECK11-NEXT: [[TMP15:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !27 10642 // CHECK11-NEXT: [[TMP16:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4, !llvm.access.group !27 10643 // CHECK11-NEXT: [[CMP5:%.*]] = icmp sle i32 [[TMP15]], [[TMP16]] 10644 // CHECK11-NEXT: br i1 [[CMP5]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] 10645 // CHECK11: omp.inner.for.body: 10646 // CHECK11-NEXT: [[TMP17:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4, !llvm.access.group !27 10647 // CHECK11-NEXT: [[TMP18:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4, !llvm.access.group !27 10648 // CHECK11-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 6, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32, i32, i32*, double**, double**, double**)* @.omp_outlined..3 to void (i32*, i32*, ...)*), i32 [[TMP17]], i32 [[TMP18]], i32* [[TMP0]], double** [[TMP1]], double** [[TMP2]], double** [[TMP3]]), !llvm.access.group !27 10649 // CHECK11-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] 10650 // CHECK11: omp.inner.for.inc: 10651 // CHECK11-NEXT: [[TMP19:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !27 10652 // CHECK11-NEXT: [[TMP20:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4, !llvm.access.group !27 10653 // CHECK11-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP19]], [[TMP20]] 10654 // CHECK11-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4, !llvm.access.group !27 10655 // CHECK11-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP28:![0-9]+]] 10656 // CHECK11: omp.inner.for.end: 10657 // CHECK11-NEXT: br label [[OMP_LOOP_EXIT:%.*]] 10658 // CHECK11: omp.loop.exit: 10659 // CHECK11-NEXT: [[TMP21:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 4 10660 // CHECK11-NEXT: [[TMP22:%.*]] = load i32, i32* [[TMP21]], align 4 10661 // CHECK11-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP22]]) 10662 // CHECK11-NEXT: [[TMP23:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4 10663 // CHECK11-NEXT: [[TMP24:%.*]] = icmp ne i32 [[TMP23]], 0 10664 // CHECK11-NEXT: br i1 [[TMP24]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]] 10665 // CHECK11: .omp.final.then: 10666 // CHECK11-NEXT: [[TMP25:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 10667 // CHECK11-NEXT: [[SUB6:%.*]] = sub nsw i32 [[TMP25]], 0 10668 // CHECK11-NEXT: [[DIV7:%.*]] = sdiv i32 [[SUB6]], 1 10669 // CHECK11-NEXT: [[MUL:%.*]] = mul nsw i32 [[DIV7]], 1 10670 // CHECK11-NEXT: [[ADD8:%.*]] = add nsw i32 0, [[MUL]] 10671 // CHECK11-NEXT: store i32 [[ADD8]], i32* [[I3]], align 4 10672 // CHECK11-NEXT: br label [[DOTOMP_FINAL_DONE]] 10673 // CHECK11: .omp.final.done: 10674 // CHECK11-NEXT: br label [[OMP_PRECOND_END]] 10675 // CHECK11: omp.precond.end: 10676 // CHECK11-NEXT: ret void 10677 // 10678 // 10679 // CHECK11-LABEL: define {{[^@]+}}@.omp_outlined..3 10680 // CHECK11-SAME: (i32* noalias noundef [[DOTGLOBAL_TID_:%.*]], i32* noalias noundef [[DOTBOUND_TID_:%.*]], i32 noundef [[DOTPREVIOUS_LB_:%.*]], i32 noundef [[DOTPREVIOUS_UB_:%.*]], i32* noundef nonnull align 4 dereferenceable(4) [[N:%.*]], double** noundef nonnull align 4 dereferenceable(4) [[A:%.*]], double** noundef nonnull align 4 dereferenceable(4) [[B:%.*]], double** noundef nonnull align 4 dereferenceable(4) [[C:%.*]]) #[[ATTR1]] { 10681 // CHECK11-NEXT: entry: 10682 // CHECK11-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4 10683 // CHECK11-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4 10684 // CHECK11-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i32, align 4 10685 // CHECK11-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i32, align 4 10686 // CHECK11-NEXT: [[N_ADDR:%.*]] = alloca i32*, align 4 10687 // CHECK11-NEXT: [[A_ADDR:%.*]] = alloca double**, align 4 10688 // CHECK11-NEXT: [[B_ADDR:%.*]] = alloca double**, align 4 10689 // CHECK11-NEXT: [[C_ADDR:%.*]] = alloca double**, align 4 10690 // CHECK11-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 10691 // CHECK11-NEXT: [[TMP:%.*]] = alloca i32, align 4 10692 // CHECK11-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 10693 // CHECK11-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 10694 // CHECK11-NEXT: [[I:%.*]] = alloca i32, align 4 10695 // CHECK11-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 10696 // CHECK11-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4 10697 // CHECK11-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 10698 // CHECK11-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 10699 // CHECK11-NEXT: [[I3:%.*]] = alloca i32, align 4 10700 // CHECK11-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 4 10701 // CHECK11-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 4 10702 // CHECK11-NEXT: store i32 [[DOTPREVIOUS_LB_]], i32* [[DOTPREVIOUS_LB__ADDR]], align 4 10703 // CHECK11-NEXT: store i32 [[DOTPREVIOUS_UB_]], i32* [[DOTPREVIOUS_UB__ADDR]], align 4 10704 // CHECK11-NEXT: store i32* [[N]], i32** [[N_ADDR]], align 4 10705 // CHECK11-NEXT: store double** [[A]], double*** [[A_ADDR]], align 4 10706 // CHECK11-NEXT: store double** [[B]], double*** [[B_ADDR]], align 4 10707 // CHECK11-NEXT: store double** [[C]], double*** [[C_ADDR]], align 4 10708 // CHECK11-NEXT: [[TMP0:%.*]] = load i32*, i32** [[N_ADDR]], align 4 10709 // CHECK11-NEXT: [[TMP1:%.*]] = load double**, double*** [[A_ADDR]], align 4 10710 // CHECK11-NEXT: [[TMP2:%.*]] = load double**, double*** [[B_ADDR]], align 4 10711 // CHECK11-NEXT: [[TMP3:%.*]] = load double**, double*** [[C_ADDR]], align 4 10712 // CHECK11-NEXT: [[TMP4:%.*]] = load i32, i32* [[TMP0]], align 4 10713 // CHECK11-NEXT: store i32 [[TMP4]], i32* [[DOTCAPTURE_EXPR_]], align 4 10714 // CHECK11-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 10715 // CHECK11-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP5]], 0 10716 // CHECK11-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 10717 // CHECK11-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1 10718 // CHECK11-NEXT: store i32 [[SUB2]], i32* [[DOTCAPTURE_EXPR_1]], align 4 10719 // CHECK11-NEXT: store i32 0, i32* [[I]], align 4 10720 // CHECK11-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 10721 // CHECK11-NEXT: [[CMP:%.*]] = icmp slt i32 0, [[TMP6]] 10722 // CHECK11-NEXT: br i1 [[CMP]], label [[OMP_PRECOND_THEN:%.*]], label [[OMP_PRECOND_END:%.*]] 10723 // CHECK11: omp.precond.then: 10724 // CHECK11-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4 10725 // CHECK11-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 10726 // CHECK11-NEXT: store i32 [[TMP7]], i32* [[DOTOMP_UB]], align 4 10727 // CHECK11-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTPREVIOUS_LB__ADDR]], align 4 10728 // CHECK11-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTPREVIOUS_UB__ADDR]], align 4 10729 // CHECK11-NEXT: store i32 [[TMP8]], i32* [[DOTOMP_LB]], align 4 10730 // CHECK11-NEXT: store i32 [[TMP9]], i32* [[DOTOMP_UB]], align 4 10731 // CHECK11-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4 10732 // CHECK11-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4 10733 // CHECK11-NEXT: [[TMP10:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 4 10734 // CHECK11-NEXT: [[TMP11:%.*]] = load i32, i32* [[TMP10]], align 4 10735 // CHECK11-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB2]], i32 [[TMP11]], i32 34, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1) 10736 // CHECK11-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 10737 // CHECK11-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 10738 // CHECK11-NEXT: [[CMP4:%.*]] = icmp sgt i32 [[TMP12]], [[TMP13]] 10739 // CHECK11-NEXT: br i1 [[CMP4]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] 10740 // CHECK11: cond.true: 10741 // CHECK11-NEXT: [[TMP14:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 10742 // CHECK11-NEXT: br label [[COND_END:%.*]] 10743 // CHECK11: cond.false: 10744 // CHECK11-NEXT: [[TMP15:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 10745 // CHECK11-NEXT: br label [[COND_END]] 10746 // CHECK11: cond.end: 10747 // CHECK11-NEXT: [[COND:%.*]] = phi i32 [ [[TMP14]], [[COND_TRUE]] ], [ [[TMP15]], [[COND_FALSE]] ] 10748 // CHECK11-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4 10749 // CHECK11-NEXT: [[TMP16:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4 10750 // CHECK11-NEXT: store i32 [[TMP16]], i32* [[DOTOMP_IV]], align 4 10751 // CHECK11-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] 10752 // CHECK11: omp.inner.for.cond: 10753 // CHECK11-NEXT: [[TMP17:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !30 10754 // CHECK11-NEXT: [[TMP18:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4, !llvm.access.group !30 10755 // CHECK11-NEXT: [[CMP5:%.*]] = icmp sle i32 [[TMP17]], [[TMP18]] 10756 // CHECK11-NEXT: br i1 [[CMP5]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] 10757 // CHECK11: omp.inner.for.body: 10758 // CHECK11-NEXT: [[TMP19:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !30 10759 // CHECK11-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP19]], 1 10760 // CHECK11-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] 10761 // CHECK11-NEXT: store i32 [[ADD]], i32* [[I3]], align 4, !llvm.access.group !30 10762 // CHECK11-NEXT: [[TMP20:%.*]] = load double*, double** [[TMP2]], align 4, !llvm.access.group !30 10763 // CHECK11-NEXT: [[TMP21:%.*]] = load i32, i32* [[I3]], align 4, !llvm.access.group !30 10764 // CHECK11-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds double, double* [[TMP20]], i32 [[TMP21]] 10765 // CHECK11-NEXT: [[TMP22:%.*]] = load double, double* [[ARRAYIDX]], align 4, !llvm.access.group !30 10766 // CHECK11-NEXT: [[TMP23:%.*]] = load double*, double** [[TMP3]], align 4, !llvm.access.group !30 10767 // CHECK11-NEXT: [[TMP24:%.*]] = load i32, i32* [[I3]], align 4, !llvm.access.group !30 10768 // CHECK11-NEXT: [[ARRAYIDX6:%.*]] = getelementptr inbounds double, double* [[TMP23]], i32 [[TMP24]] 10769 // CHECK11-NEXT: [[TMP25:%.*]] = load double, double* [[ARRAYIDX6]], align 4, !llvm.access.group !30 10770 // CHECK11-NEXT: [[ADD7:%.*]] = fadd double [[TMP22]], [[TMP25]] 10771 // CHECK11-NEXT: [[TMP26:%.*]] = load double*, double** [[TMP1]], align 4, !llvm.access.group !30 10772 // CHECK11-NEXT: [[TMP27:%.*]] = load i32, i32* [[I3]], align 4, !llvm.access.group !30 10773 // CHECK11-NEXT: [[ARRAYIDX8:%.*]] = getelementptr inbounds double, double* [[TMP26]], i32 [[TMP27]] 10774 // CHECK11-NEXT: store double [[ADD7]], double* [[ARRAYIDX8]], align 4, !llvm.access.group !30 10775 // CHECK11-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] 10776 // CHECK11: omp.body.continue: 10777 // CHECK11-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] 10778 // CHECK11: omp.inner.for.inc: 10779 // CHECK11-NEXT: [[TMP28:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !30 10780 // CHECK11-NEXT: [[ADD9:%.*]] = add nsw i32 [[TMP28]], 1 10781 // CHECK11-NEXT: store i32 [[ADD9]], i32* [[DOTOMP_IV]], align 4, !llvm.access.group !30 10782 // CHECK11-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP31:![0-9]+]] 10783 // CHECK11: omp.inner.for.end: 10784 // CHECK11-NEXT: br label [[OMP_LOOP_EXIT:%.*]] 10785 // CHECK11: omp.loop.exit: 10786 // CHECK11-NEXT: [[TMP29:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 4 10787 // CHECK11-NEXT: [[TMP30:%.*]] = load i32, i32* [[TMP29]], align 4 10788 // CHECK11-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP30]]) 10789 // CHECK11-NEXT: [[TMP31:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4 10790 // CHECK11-NEXT: [[TMP32:%.*]] = icmp ne i32 [[TMP31]], 0 10791 // CHECK11-NEXT: br i1 [[TMP32]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]] 10792 // CHECK11: .omp.final.then: 10793 // CHECK11-NEXT: [[TMP33:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 10794 // CHECK11-NEXT: [[SUB10:%.*]] = sub nsw i32 [[TMP33]], 0 10795 // CHECK11-NEXT: [[DIV11:%.*]] = sdiv i32 [[SUB10]], 1 10796 // CHECK11-NEXT: [[MUL12:%.*]] = mul nsw i32 [[DIV11]], 1 10797 // CHECK11-NEXT: [[ADD13:%.*]] = add nsw i32 0, [[MUL12]] 10798 // CHECK11-NEXT: store i32 [[ADD13]], i32* [[I3]], align 4 10799 // CHECK11-NEXT: br label [[DOTOMP_FINAL_DONE]] 10800 // CHECK11: .omp.final.done: 10801 // CHECK11-NEXT: br label [[OMP_PRECOND_END]] 10802 // CHECK11: omp.precond.end: 10803 // CHECK11-NEXT: ret void 10804 // 10805 // 10806 // CHECK11-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l446 10807 // CHECK11-SAME: (i32 noundef [[CH:%.*]], i32 noundef [[N:%.*]], double* noundef [[A:%.*]], double* noundef [[B:%.*]], double* noundef [[C:%.*]]) #[[ATTR1]] { 10808 // CHECK11-NEXT: entry: 10809 // CHECK11-NEXT: [[CH_ADDR:%.*]] = alloca i32, align 4 10810 // CHECK11-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4 10811 // CHECK11-NEXT: [[A_ADDR:%.*]] = alloca double*, align 4 10812 // CHECK11-NEXT: [[B_ADDR:%.*]] = alloca double*, align 4 10813 // CHECK11-NEXT: [[C_ADDR:%.*]] = alloca double*, align 4 10814 // CHECK11-NEXT: store i32 [[CH]], i32* [[CH_ADDR]], align 4 10815 // CHECK11-NEXT: store i32 [[N]], i32* [[N_ADDR]], align 4 10816 // CHECK11-NEXT: store double* [[A]], double** [[A_ADDR]], align 4 10817 // CHECK11-NEXT: store double* [[B]], double** [[B_ADDR]], align 4 10818 // CHECK11-NEXT: store double* [[C]], double** [[C_ADDR]], align 4 10819 // CHECK11-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 5, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32*, i32*, double**, double**, double**)* @.omp_outlined..6 to void (i32*, i32*, ...)*), i32* [[CH_ADDR]], i32* [[N_ADDR]], double** [[A_ADDR]], double** [[B_ADDR]], double** [[C_ADDR]]) 10820 // CHECK11-NEXT: ret void 10821 // 10822 // 10823 // CHECK11-LABEL: define {{[^@]+}}@.omp_outlined..6 10824 // CHECK11-SAME: (i32* noalias noundef [[DOTGLOBAL_TID_:%.*]], i32* noalias noundef [[DOTBOUND_TID_:%.*]], i32* noundef nonnull align 4 dereferenceable(4) [[CH:%.*]], i32* noundef nonnull align 4 dereferenceable(4) [[N:%.*]], double** noundef nonnull align 4 dereferenceable(4) [[A:%.*]], double** noundef nonnull align 4 dereferenceable(4) [[B:%.*]], double** noundef nonnull align 4 dereferenceable(4) [[C:%.*]]) #[[ATTR1]] { 10825 // CHECK11-NEXT: entry: 10826 // CHECK11-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4 10827 // CHECK11-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4 10828 // CHECK11-NEXT: [[CH_ADDR:%.*]] = alloca i32*, align 4 10829 // CHECK11-NEXT: [[N_ADDR:%.*]] = alloca i32*, align 4 10830 // CHECK11-NEXT: [[A_ADDR:%.*]] = alloca double**, align 4 10831 // CHECK11-NEXT: [[B_ADDR:%.*]] = alloca double**, align 4 10832 // CHECK11-NEXT: [[C_ADDR:%.*]] = alloca double**, align 4 10833 // CHECK11-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 10834 // CHECK11-NEXT: [[TMP:%.*]] = alloca i32, align 4 10835 // CHECK11-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 10836 // CHECK11-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 10837 // CHECK11-NEXT: [[I:%.*]] = alloca i32, align 4 10838 // CHECK11-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4 10839 // CHECK11-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4 10840 // CHECK11-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 10841 // CHECK11-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 10842 // CHECK11-NEXT: [[I3:%.*]] = alloca i32, align 4 10843 // CHECK11-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 4 10844 // CHECK11-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 4 10845 // CHECK11-NEXT: store i32* [[CH]], i32** [[CH_ADDR]], align 4 10846 // CHECK11-NEXT: store i32* [[N]], i32** [[N_ADDR]], align 4 10847 // CHECK11-NEXT: store double** [[A]], double*** [[A_ADDR]], align 4 10848 // CHECK11-NEXT: store double** [[B]], double*** [[B_ADDR]], align 4 10849 // CHECK11-NEXT: store double** [[C]], double*** [[C_ADDR]], align 4 10850 // CHECK11-NEXT: [[TMP0:%.*]] = load i32*, i32** [[CH_ADDR]], align 4 10851 // CHECK11-NEXT: [[TMP1:%.*]] = load i32*, i32** [[N_ADDR]], align 4 10852 // CHECK11-NEXT: [[TMP2:%.*]] = load double**, double*** [[A_ADDR]], align 4 10853 // CHECK11-NEXT: [[TMP3:%.*]] = load double**, double*** [[B_ADDR]], align 4 10854 // CHECK11-NEXT: [[TMP4:%.*]] = load double**, double*** [[C_ADDR]], align 4 10855 // CHECK11-NEXT: [[TMP5:%.*]] = load i32, i32* [[TMP1]], align 4 10856 // CHECK11-NEXT: store i32 [[TMP5]], i32* [[DOTCAPTURE_EXPR_]], align 4 10857 // CHECK11-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 10858 // CHECK11-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP6]], 0 10859 // CHECK11-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 10860 // CHECK11-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1 10861 // CHECK11-NEXT: store i32 [[SUB2]], i32* [[DOTCAPTURE_EXPR_1]], align 4 10862 // CHECK11-NEXT: store i32 0, i32* [[I]], align 4 10863 // CHECK11-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 10864 // CHECK11-NEXT: [[CMP:%.*]] = icmp slt i32 0, [[TMP7]] 10865 // CHECK11-NEXT: br i1 [[CMP]], label [[OMP_PRECOND_THEN:%.*]], label [[OMP_PRECOND_END:%.*]] 10866 // CHECK11: omp.precond.then: 10867 // CHECK11-NEXT: store i32 0, i32* [[DOTOMP_COMB_LB]], align 4 10868 // CHECK11-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 10869 // CHECK11-NEXT: store i32 [[TMP8]], i32* [[DOTOMP_COMB_UB]], align 4 10870 // CHECK11-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4 10871 // CHECK11-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4 10872 // CHECK11-NEXT: [[TMP9:%.*]] = load i32, i32* [[TMP0]], align 4 10873 // CHECK11-NEXT: [[TMP10:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 4 10874 // CHECK11-NEXT: [[TMP11:%.*]] = load i32, i32* [[TMP10]], align 4 10875 // CHECK11-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1]], i32 [[TMP11]], i32 91, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_COMB_LB]], i32* [[DOTOMP_COMB_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 [[TMP9]]) 10876 // CHECK11-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 10877 // CHECK11-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 10878 // CHECK11-NEXT: [[CMP4:%.*]] = icmp sgt i32 [[TMP12]], [[TMP13]] 10879 // CHECK11-NEXT: br i1 [[CMP4]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] 10880 // CHECK11: cond.true: 10881 // CHECK11-NEXT: [[TMP14:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 10882 // CHECK11-NEXT: br label [[COND_END:%.*]] 10883 // CHECK11: cond.false: 10884 // CHECK11-NEXT: [[TMP15:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 10885 // CHECK11-NEXT: br label [[COND_END]] 10886 // CHECK11: cond.end: 10887 // CHECK11-NEXT: [[COND:%.*]] = phi i32 [ [[TMP14]], [[COND_TRUE]] ], [ [[TMP15]], [[COND_FALSE]] ] 10888 // CHECK11-NEXT: store i32 [[COND]], i32* [[DOTOMP_COMB_UB]], align 4 10889 // CHECK11-NEXT: [[TMP16:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4 10890 // CHECK11-NEXT: store i32 [[TMP16]], i32* [[DOTOMP_IV]], align 4 10891 // CHECK11-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] 10892 // CHECK11: omp.inner.for.cond: 10893 // CHECK11-NEXT: [[TMP17:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !33 10894 // CHECK11-NEXT: [[TMP18:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4, !llvm.access.group !33 10895 // CHECK11-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP18]], 1 10896 // CHECK11-NEXT: [[CMP5:%.*]] = icmp slt i32 [[TMP17]], [[ADD]] 10897 // CHECK11-NEXT: br i1 [[CMP5]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] 10898 // CHECK11: omp.inner.for.body: 10899 // CHECK11-NEXT: [[TMP19:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4, !llvm.access.group !33 10900 // CHECK11-NEXT: [[TMP20:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4, !llvm.access.group !33 10901 // CHECK11-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 6, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32, i32, i32*, double**, double**, double**)* @.omp_outlined..7 to void (i32*, i32*, ...)*), i32 [[TMP19]], i32 [[TMP20]], i32* [[TMP1]], double** [[TMP2]], double** [[TMP3]], double** [[TMP4]]), !llvm.access.group !33 10902 // CHECK11-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] 10903 // CHECK11: omp.inner.for.inc: 10904 // CHECK11-NEXT: [[TMP21:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !33 10905 // CHECK11-NEXT: [[TMP22:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4, !llvm.access.group !33 10906 // CHECK11-NEXT: [[ADD6:%.*]] = add nsw i32 [[TMP21]], [[TMP22]] 10907 // CHECK11-NEXT: store i32 [[ADD6]], i32* [[DOTOMP_IV]], align 4, !llvm.access.group !33 10908 // CHECK11-NEXT: [[TMP23:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4, !llvm.access.group !33 10909 // CHECK11-NEXT: [[TMP24:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4, !llvm.access.group !33 10910 // CHECK11-NEXT: [[ADD7:%.*]] = add nsw i32 [[TMP23]], [[TMP24]] 10911 // CHECK11-NEXT: store i32 [[ADD7]], i32* [[DOTOMP_COMB_LB]], align 4, !llvm.access.group !33 10912 // CHECK11-NEXT: [[TMP25:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4, !llvm.access.group !33 10913 // CHECK11-NEXT: [[TMP26:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4, !llvm.access.group !33 10914 // CHECK11-NEXT: [[ADD8:%.*]] = add nsw i32 [[TMP25]], [[TMP26]] 10915 // CHECK11-NEXT: store i32 [[ADD8]], i32* [[DOTOMP_COMB_UB]], align 4, !llvm.access.group !33 10916 // CHECK11-NEXT: [[TMP27:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4, !llvm.access.group !33 10917 // CHECK11-NEXT: [[TMP28:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4, !llvm.access.group !33 10918 // CHECK11-NEXT: [[CMP9:%.*]] = icmp sgt i32 [[TMP27]], [[TMP28]] 10919 // CHECK11-NEXT: br i1 [[CMP9]], label [[COND_TRUE10:%.*]], label [[COND_FALSE11:%.*]] 10920 // CHECK11: cond.true10: 10921 // CHECK11-NEXT: [[TMP29:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4, !llvm.access.group !33 10922 // CHECK11-NEXT: br label [[COND_END12:%.*]] 10923 // CHECK11: cond.false11: 10924 // CHECK11-NEXT: [[TMP30:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4, !llvm.access.group !33 10925 // CHECK11-NEXT: br label [[COND_END12]] 10926 // CHECK11: cond.end12: 10927 // CHECK11-NEXT: [[COND13:%.*]] = phi i32 [ [[TMP29]], [[COND_TRUE10]] ], [ [[TMP30]], [[COND_FALSE11]] ] 10928 // CHECK11-NEXT: store i32 [[COND13]], i32* [[DOTOMP_COMB_UB]], align 4, !llvm.access.group !33 10929 // CHECK11-NEXT: [[TMP31:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4, !llvm.access.group !33 10930 // CHECK11-NEXT: store i32 [[TMP31]], i32* [[DOTOMP_IV]], align 4, !llvm.access.group !33 10931 // CHECK11-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP34:![0-9]+]] 10932 // CHECK11: omp.inner.for.end: 10933 // CHECK11-NEXT: br label [[OMP_LOOP_EXIT:%.*]] 10934 // CHECK11: omp.loop.exit: 10935 // CHECK11-NEXT: [[TMP32:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 4 10936 // CHECK11-NEXT: [[TMP33:%.*]] = load i32, i32* [[TMP32]], align 4 10937 // CHECK11-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP33]]) 10938 // CHECK11-NEXT: [[TMP34:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4 10939 // CHECK11-NEXT: [[TMP35:%.*]] = icmp ne i32 [[TMP34]], 0 10940 // CHECK11-NEXT: br i1 [[TMP35]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]] 10941 // CHECK11: .omp.final.then: 10942 // CHECK11-NEXT: [[TMP36:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 10943 // CHECK11-NEXT: [[SUB14:%.*]] = sub nsw i32 [[TMP36]], 0 10944 // CHECK11-NEXT: [[DIV15:%.*]] = sdiv i32 [[SUB14]], 1 10945 // CHECK11-NEXT: [[MUL:%.*]] = mul nsw i32 [[DIV15]], 1 10946 // CHECK11-NEXT: [[ADD16:%.*]] = add nsw i32 0, [[MUL]] 10947 // CHECK11-NEXT: store i32 [[ADD16]], i32* [[I3]], align 4 10948 // CHECK11-NEXT: br label [[DOTOMP_FINAL_DONE]] 10949 // CHECK11: .omp.final.done: 10950 // CHECK11-NEXT: br label [[OMP_PRECOND_END]] 10951 // CHECK11: omp.precond.end: 10952 // CHECK11-NEXT: ret void 10953 // 10954 // 10955 // CHECK11-LABEL: define {{[^@]+}}@.omp_outlined..7 10956 // CHECK11-SAME: (i32* noalias noundef [[DOTGLOBAL_TID_:%.*]], i32* noalias noundef [[DOTBOUND_TID_:%.*]], i32 noundef [[DOTPREVIOUS_LB_:%.*]], i32 noundef [[DOTPREVIOUS_UB_:%.*]], i32* noundef nonnull align 4 dereferenceable(4) [[N:%.*]], double** noundef nonnull align 4 dereferenceable(4) [[A:%.*]], double** noundef nonnull align 4 dereferenceable(4) [[B:%.*]], double** noundef nonnull align 4 dereferenceable(4) [[C:%.*]]) #[[ATTR1]] { 10957 // CHECK11-NEXT: entry: 10958 // CHECK11-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4 10959 // CHECK11-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4 10960 // CHECK11-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i32, align 4 10961 // CHECK11-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i32, align 4 10962 // CHECK11-NEXT: [[N_ADDR:%.*]] = alloca i32*, align 4 10963 // CHECK11-NEXT: [[A_ADDR:%.*]] = alloca double**, align 4 10964 // CHECK11-NEXT: [[B_ADDR:%.*]] = alloca double**, align 4 10965 // CHECK11-NEXT: [[C_ADDR:%.*]] = alloca double**, align 4 10966 // CHECK11-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 10967 // CHECK11-NEXT: [[TMP:%.*]] = alloca i32, align 4 10968 // CHECK11-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 10969 // CHECK11-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 10970 // CHECK11-NEXT: [[I:%.*]] = alloca i32, align 4 10971 // CHECK11-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 10972 // CHECK11-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4 10973 // CHECK11-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 10974 // CHECK11-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 10975 // CHECK11-NEXT: [[I3:%.*]] = alloca i32, align 4 10976 // CHECK11-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 4 10977 // CHECK11-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 4 10978 // CHECK11-NEXT: store i32 [[DOTPREVIOUS_LB_]], i32* [[DOTPREVIOUS_LB__ADDR]], align 4 10979 // CHECK11-NEXT: store i32 [[DOTPREVIOUS_UB_]], i32* [[DOTPREVIOUS_UB__ADDR]], align 4 10980 // CHECK11-NEXT: store i32* [[N]], i32** [[N_ADDR]], align 4 10981 // CHECK11-NEXT: store double** [[A]], double*** [[A_ADDR]], align 4 10982 // CHECK11-NEXT: store double** [[B]], double*** [[B_ADDR]], align 4 10983 // CHECK11-NEXT: store double** [[C]], double*** [[C_ADDR]], align 4 10984 // CHECK11-NEXT: [[TMP0:%.*]] = load i32*, i32** [[N_ADDR]], align 4 10985 // CHECK11-NEXT: [[TMP1:%.*]] = load double**, double*** [[A_ADDR]], align 4 10986 // CHECK11-NEXT: [[TMP2:%.*]] = load double**, double*** [[B_ADDR]], align 4 10987 // CHECK11-NEXT: [[TMP3:%.*]] = load double**, double*** [[C_ADDR]], align 4 10988 // CHECK11-NEXT: [[TMP4:%.*]] = load i32, i32* [[TMP0]], align 4 10989 // CHECK11-NEXT: store i32 [[TMP4]], i32* [[DOTCAPTURE_EXPR_]], align 4 10990 // CHECK11-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 10991 // CHECK11-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP5]], 0 10992 // CHECK11-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 10993 // CHECK11-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1 10994 // CHECK11-NEXT: store i32 [[SUB2]], i32* [[DOTCAPTURE_EXPR_1]], align 4 10995 // CHECK11-NEXT: store i32 0, i32* [[I]], align 4 10996 // CHECK11-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 10997 // CHECK11-NEXT: [[CMP:%.*]] = icmp slt i32 0, [[TMP6]] 10998 // CHECK11-NEXT: br i1 [[CMP]], label [[OMP_PRECOND_THEN:%.*]], label [[OMP_PRECOND_END:%.*]] 10999 // CHECK11: omp.precond.then: 11000 // CHECK11-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4 11001 // CHECK11-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 11002 // CHECK11-NEXT: store i32 [[TMP7]], i32* [[DOTOMP_UB]], align 4 11003 // CHECK11-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTPREVIOUS_LB__ADDR]], align 4 11004 // CHECK11-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTPREVIOUS_UB__ADDR]], align 4 11005 // CHECK11-NEXT: store i32 [[TMP8]], i32* [[DOTOMP_LB]], align 4 11006 // CHECK11-NEXT: store i32 [[TMP9]], i32* [[DOTOMP_UB]], align 4 11007 // CHECK11-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4 11008 // CHECK11-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4 11009 // CHECK11-NEXT: [[TMP10:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 4 11010 // CHECK11-NEXT: [[TMP11:%.*]] = load i32, i32* [[TMP10]], align 4 11011 // CHECK11-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB2]], i32 [[TMP11]], i32 34, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1) 11012 // CHECK11-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 11013 // CHECK11-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 11014 // CHECK11-NEXT: [[CMP4:%.*]] = icmp sgt i32 [[TMP12]], [[TMP13]] 11015 // CHECK11-NEXT: br i1 [[CMP4]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] 11016 // CHECK11: cond.true: 11017 // CHECK11-NEXT: [[TMP14:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 11018 // CHECK11-NEXT: br label [[COND_END:%.*]] 11019 // CHECK11: cond.false: 11020 // CHECK11-NEXT: [[TMP15:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 11021 // CHECK11-NEXT: br label [[COND_END]] 11022 // CHECK11: cond.end: 11023 // CHECK11-NEXT: [[COND:%.*]] = phi i32 [ [[TMP14]], [[COND_TRUE]] ], [ [[TMP15]], [[COND_FALSE]] ] 11024 // CHECK11-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4 11025 // CHECK11-NEXT: [[TMP16:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4 11026 // CHECK11-NEXT: store i32 [[TMP16]], i32* [[DOTOMP_IV]], align 4 11027 // CHECK11-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] 11028 // CHECK11: omp.inner.for.cond: 11029 // CHECK11-NEXT: [[TMP17:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !36 11030 // CHECK11-NEXT: [[TMP18:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4, !llvm.access.group !36 11031 // CHECK11-NEXT: [[CMP5:%.*]] = icmp sle i32 [[TMP17]], [[TMP18]] 11032 // CHECK11-NEXT: br i1 [[CMP5]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] 11033 // CHECK11: omp.inner.for.body: 11034 // CHECK11-NEXT: [[TMP19:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !36 11035 // CHECK11-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP19]], 1 11036 // CHECK11-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] 11037 // CHECK11-NEXT: store i32 [[ADD]], i32* [[I3]], align 4, !llvm.access.group !36 11038 // CHECK11-NEXT: [[TMP20:%.*]] = load double*, double** [[TMP2]], align 4, !llvm.access.group !36 11039 // CHECK11-NEXT: [[TMP21:%.*]] = load i32, i32* [[I3]], align 4, !llvm.access.group !36 11040 // CHECK11-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds double, double* [[TMP20]], i32 [[TMP21]] 11041 // CHECK11-NEXT: [[TMP22:%.*]] = load double, double* [[ARRAYIDX]], align 4, !llvm.access.group !36 11042 // CHECK11-NEXT: [[TMP23:%.*]] = load double*, double** [[TMP3]], align 4, !llvm.access.group !36 11043 // CHECK11-NEXT: [[TMP24:%.*]] = load i32, i32* [[I3]], align 4, !llvm.access.group !36 11044 // CHECK11-NEXT: [[ARRAYIDX6:%.*]] = getelementptr inbounds double, double* [[TMP23]], i32 [[TMP24]] 11045 // CHECK11-NEXT: [[TMP25:%.*]] = load double, double* [[ARRAYIDX6]], align 4, !llvm.access.group !36 11046 // CHECK11-NEXT: [[ADD7:%.*]] = fadd double [[TMP22]], [[TMP25]] 11047 // CHECK11-NEXT: [[TMP26:%.*]] = load double*, double** [[TMP1]], align 4, !llvm.access.group !36 11048 // CHECK11-NEXT: [[TMP27:%.*]] = load i32, i32* [[I3]], align 4, !llvm.access.group !36 11049 // CHECK11-NEXT: [[ARRAYIDX8:%.*]] = getelementptr inbounds double, double* [[TMP26]], i32 [[TMP27]] 11050 // CHECK11-NEXT: store double [[ADD7]], double* [[ARRAYIDX8]], align 4, !llvm.access.group !36 11051 // CHECK11-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] 11052 // CHECK11: omp.body.continue: 11053 // CHECK11-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] 11054 // CHECK11: omp.inner.for.inc: 11055 // CHECK11-NEXT: [[TMP28:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !36 11056 // CHECK11-NEXT: [[ADD9:%.*]] = add nsw i32 [[TMP28]], 1 11057 // CHECK11-NEXT: store i32 [[ADD9]], i32* [[DOTOMP_IV]], align 4, !llvm.access.group !36 11058 // CHECK11-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP37:![0-9]+]] 11059 // CHECK11: omp.inner.for.end: 11060 // CHECK11-NEXT: br label [[OMP_LOOP_EXIT:%.*]] 11061 // CHECK11: omp.loop.exit: 11062 // CHECK11-NEXT: [[TMP29:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 4 11063 // CHECK11-NEXT: [[TMP30:%.*]] = load i32, i32* [[TMP29]], align 4 11064 // CHECK11-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP30]]) 11065 // CHECK11-NEXT: [[TMP31:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4 11066 // CHECK11-NEXT: [[TMP32:%.*]] = icmp ne i32 [[TMP31]], 0 11067 // CHECK11-NEXT: br i1 [[TMP32]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]] 11068 // CHECK11: .omp.final.then: 11069 // CHECK11-NEXT: [[TMP33:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 11070 // CHECK11-NEXT: [[SUB10:%.*]] = sub nsw i32 [[TMP33]], 0 11071 // CHECK11-NEXT: [[DIV11:%.*]] = sdiv i32 [[SUB10]], 1 11072 // CHECK11-NEXT: [[MUL12:%.*]] = mul nsw i32 [[DIV11]], 1 11073 // CHECK11-NEXT: [[ADD13:%.*]] = add nsw i32 0, [[MUL12]] 11074 // CHECK11-NEXT: store i32 [[ADD13]], i32* [[I3]], align 4 11075 // CHECK11-NEXT: br label [[DOTOMP_FINAL_DONE]] 11076 // CHECK11: .omp.final.done: 11077 // CHECK11-NEXT: br label [[OMP_PRECOND_END]] 11078 // CHECK11: omp.precond.end: 11079 // CHECK11-NEXT: ret void 11080 // 11081 // 11082 // CHECK11-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l477 11083 // CHECK11-SAME: (i32 noundef [[N:%.*]], double* noundef [[A:%.*]], double* noundef [[B:%.*]], double* noundef [[C:%.*]]) #[[ATTR1]] { 11084 // CHECK11-NEXT: entry: 11085 // CHECK11-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4 11086 // CHECK11-NEXT: [[A_ADDR:%.*]] = alloca double*, align 4 11087 // CHECK11-NEXT: [[B_ADDR:%.*]] = alloca double*, align 4 11088 // CHECK11-NEXT: [[C_ADDR:%.*]] = alloca double*, align 4 11089 // CHECK11-NEXT: store i32 [[N]], i32* [[N_ADDR]], align 4 11090 // CHECK11-NEXT: store double* [[A]], double** [[A_ADDR]], align 4 11091 // CHECK11-NEXT: store double* [[B]], double** [[B_ADDR]], align 4 11092 // CHECK11-NEXT: store double* [[C]], double** [[C_ADDR]], align 4 11093 // CHECK11-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 4, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32*, double**, double**, double**)* @.omp_outlined..10 to void (i32*, i32*, ...)*), i32* [[N_ADDR]], double** [[A_ADDR]], double** [[B_ADDR]], double** [[C_ADDR]]) 11094 // CHECK11-NEXT: ret void 11095 // 11096 // 11097 // CHECK11-LABEL: define {{[^@]+}}@.omp_outlined..10 11098 // CHECK11-SAME: (i32* noalias noundef [[DOTGLOBAL_TID_:%.*]], i32* noalias noundef [[DOTBOUND_TID_:%.*]], i32* noundef nonnull align 4 dereferenceable(4) [[N:%.*]], double** noundef nonnull align 4 dereferenceable(4) [[A:%.*]], double** noundef nonnull align 4 dereferenceable(4) [[B:%.*]], double** noundef nonnull align 4 dereferenceable(4) [[C:%.*]]) #[[ATTR1]] { 11099 // CHECK11-NEXT: entry: 11100 // CHECK11-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4 11101 // CHECK11-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4 11102 // CHECK11-NEXT: [[N_ADDR:%.*]] = alloca i32*, align 4 11103 // CHECK11-NEXT: [[A_ADDR:%.*]] = alloca double**, align 4 11104 // CHECK11-NEXT: [[B_ADDR:%.*]] = alloca double**, align 4 11105 // CHECK11-NEXT: [[C_ADDR:%.*]] = alloca double**, align 4 11106 // CHECK11-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 11107 // CHECK11-NEXT: [[TMP:%.*]] = alloca i32, align 4 11108 // CHECK11-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 11109 // CHECK11-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 11110 // CHECK11-NEXT: [[I:%.*]] = alloca i32, align 4 11111 // CHECK11-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4 11112 // CHECK11-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4 11113 // CHECK11-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 11114 // CHECK11-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 11115 // CHECK11-NEXT: [[I3:%.*]] = alloca i32, align 4 11116 // CHECK11-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 4 11117 // CHECK11-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 4 11118 // CHECK11-NEXT: store i32* [[N]], i32** [[N_ADDR]], align 4 11119 // CHECK11-NEXT: store double** [[A]], double*** [[A_ADDR]], align 4 11120 // CHECK11-NEXT: store double** [[B]], double*** [[B_ADDR]], align 4 11121 // CHECK11-NEXT: store double** [[C]], double*** [[C_ADDR]], align 4 11122 // CHECK11-NEXT: [[TMP0:%.*]] = load i32*, i32** [[N_ADDR]], align 4 11123 // CHECK11-NEXT: [[TMP1:%.*]] = load double**, double*** [[A_ADDR]], align 4 11124 // CHECK11-NEXT: [[TMP2:%.*]] = load double**, double*** [[B_ADDR]], align 4 11125 // CHECK11-NEXT: [[TMP3:%.*]] = load double**, double*** [[C_ADDR]], align 4 11126 // CHECK11-NEXT: [[TMP4:%.*]] = load i32, i32* [[TMP0]], align 4 11127 // CHECK11-NEXT: store i32 [[TMP4]], i32* [[DOTCAPTURE_EXPR_]], align 4 11128 // CHECK11-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 11129 // CHECK11-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP5]], 0 11130 // CHECK11-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 11131 // CHECK11-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1 11132 // CHECK11-NEXT: store i32 [[SUB2]], i32* [[DOTCAPTURE_EXPR_1]], align 4 11133 // CHECK11-NEXT: store i32 0, i32* [[I]], align 4 11134 // CHECK11-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 11135 // CHECK11-NEXT: [[CMP:%.*]] = icmp slt i32 0, [[TMP6]] 11136 // CHECK11-NEXT: br i1 [[CMP]], label [[OMP_PRECOND_THEN:%.*]], label [[OMP_PRECOND_END:%.*]] 11137 // CHECK11: omp.precond.then: 11138 // CHECK11-NEXT: store i32 0, i32* [[DOTOMP_COMB_LB]], align 4 11139 // CHECK11-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 11140 // CHECK11-NEXT: store i32 [[TMP7]], i32* [[DOTOMP_COMB_UB]], align 4 11141 // CHECK11-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4 11142 // CHECK11-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4 11143 // CHECK11-NEXT: [[TMP8:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 4 11144 // CHECK11-NEXT: [[TMP9:%.*]] = load i32, i32* [[TMP8]], align 4 11145 // CHECK11-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1]], i32 [[TMP9]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_COMB_LB]], i32* [[DOTOMP_COMB_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1) 11146 // CHECK11-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 11147 // CHECK11-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 11148 // CHECK11-NEXT: [[CMP4:%.*]] = icmp sgt i32 [[TMP10]], [[TMP11]] 11149 // CHECK11-NEXT: br i1 [[CMP4]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] 11150 // CHECK11: cond.true: 11151 // CHECK11-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 11152 // CHECK11-NEXT: br label [[COND_END:%.*]] 11153 // CHECK11: cond.false: 11154 // CHECK11-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 11155 // CHECK11-NEXT: br label [[COND_END]] 11156 // CHECK11: cond.end: 11157 // CHECK11-NEXT: [[COND:%.*]] = phi i32 [ [[TMP12]], [[COND_TRUE]] ], [ [[TMP13]], [[COND_FALSE]] ] 11158 // CHECK11-NEXT: store i32 [[COND]], i32* [[DOTOMP_COMB_UB]], align 4 11159 // CHECK11-NEXT: [[TMP14:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4 11160 // CHECK11-NEXT: store i32 [[TMP14]], i32* [[DOTOMP_IV]], align 4 11161 // CHECK11-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] 11162 // CHECK11: omp.inner.for.cond: 11163 // CHECK11-NEXT: [[TMP15:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !39 11164 // CHECK11-NEXT: [[TMP16:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4, !llvm.access.group !39 11165 // CHECK11-NEXT: [[CMP5:%.*]] = icmp sle i32 [[TMP15]], [[TMP16]] 11166 // CHECK11-NEXT: br i1 [[CMP5]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] 11167 // CHECK11: omp.inner.for.body: 11168 // CHECK11-NEXT: [[TMP17:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4, !llvm.access.group !39 11169 // CHECK11-NEXT: [[TMP18:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4, !llvm.access.group !39 11170 // CHECK11-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 6, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32, i32, i32*, double**, double**, double**)* @.omp_outlined..11 to void (i32*, i32*, ...)*), i32 [[TMP17]], i32 [[TMP18]], i32* [[TMP0]], double** [[TMP1]], double** [[TMP2]], double** [[TMP3]]), !llvm.access.group !39 11171 // CHECK11-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] 11172 // CHECK11: omp.inner.for.inc: 11173 // CHECK11-NEXT: [[TMP19:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !39 11174 // CHECK11-NEXT: [[TMP20:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4, !llvm.access.group !39 11175 // CHECK11-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP19]], [[TMP20]] 11176 // CHECK11-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4, !llvm.access.group !39 11177 // CHECK11-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP40:![0-9]+]] 11178 // CHECK11: omp.inner.for.end: 11179 // CHECK11-NEXT: br label [[OMP_LOOP_EXIT:%.*]] 11180 // CHECK11: omp.loop.exit: 11181 // CHECK11-NEXT: [[TMP21:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 4 11182 // CHECK11-NEXT: [[TMP22:%.*]] = load i32, i32* [[TMP21]], align 4 11183 // CHECK11-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP22]]) 11184 // CHECK11-NEXT: [[TMP23:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4 11185 // CHECK11-NEXT: [[TMP24:%.*]] = icmp ne i32 [[TMP23]], 0 11186 // CHECK11-NEXT: br i1 [[TMP24]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]] 11187 // CHECK11: .omp.final.then: 11188 // CHECK11-NEXT: [[TMP25:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 11189 // CHECK11-NEXT: [[SUB6:%.*]] = sub nsw i32 [[TMP25]], 0 11190 // CHECK11-NEXT: [[DIV7:%.*]] = sdiv i32 [[SUB6]], 1 11191 // CHECK11-NEXT: [[MUL:%.*]] = mul nsw i32 [[DIV7]], 1 11192 // CHECK11-NEXT: [[ADD8:%.*]] = add nsw i32 0, [[MUL]] 11193 // CHECK11-NEXT: store i32 [[ADD8]], i32* [[I3]], align 4 11194 // CHECK11-NEXT: br label [[DOTOMP_FINAL_DONE]] 11195 // CHECK11: .omp.final.done: 11196 // CHECK11-NEXT: br label [[OMP_PRECOND_END]] 11197 // CHECK11: omp.precond.end: 11198 // CHECK11-NEXT: ret void 11199 // 11200 // 11201 // CHECK11-LABEL: define {{[^@]+}}@.omp_outlined..11 11202 // CHECK11-SAME: (i32* noalias noundef [[DOTGLOBAL_TID_:%.*]], i32* noalias noundef [[DOTBOUND_TID_:%.*]], i32 noundef [[DOTPREVIOUS_LB_:%.*]], i32 noundef [[DOTPREVIOUS_UB_:%.*]], i32* noundef nonnull align 4 dereferenceable(4) [[N:%.*]], double** noundef nonnull align 4 dereferenceable(4) [[A:%.*]], double** noundef nonnull align 4 dereferenceable(4) [[B:%.*]], double** noundef nonnull align 4 dereferenceable(4) [[C:%.*]]) #[[ATTR1]] { 11203 // CHECK11-NEXT: entry: 11204 // CHECK11-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4 11205 // CHECK11-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4 11206 // CHECK11-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i32, align 4 11207 // CHECK11-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i32, align 4 11208 // CHECK11-NEXT: [[N_ADDR:%.*]] = alloca i32*, align 4 11209 // CHECK11-NEXT: [[A_ADDR:%.*]] = alloca double**, align 4 11210 // CHECK11-NEXT: [[B_ADDR:%.*]] = alloca double**, align 4 11211 // CHECK11-NEXT: [[C_ADDR:%.*]] = alloca double**, align 4 11212 // CHECK11-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 11213 // CHECK11-NEXT: [[TMP:%.*]] = alloca i32, align 4 11214 // CHECK11-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 11215 // CHECK11-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 11216 // CHECK11-NEXT: [[I:%.*]] = alloca i32, align 4 11217 // CHECK11-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 11218 // CHECK11-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4 11219 // CHECK11-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 11220 // CHECK11-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 11221 // CHECK11-NEXT: [[I3:%.*]] = alloca i32, align 4 11222 // CHECK11-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 4 11223 // CHECK11-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 4 11224 // CHECK11-NEXT: store i32 [[DOTPREVIOUS_LB_]], i32* [[DOTPREVIOUS_LB__ADDR]], align 4 11225 // CHECK11-NEXT: store i32 [[DOTPREVIOUS_UB_]], i32* [[DOTPREVIOUS_UB__ADDR]], align 4 11226 // CHECK11-NEXT: store i32* [[N]], i32** [[N_ADDR]], align 4 11227 // CHECK11-NEXT: store double** [[A]], double*** [[A_ADDR]], align 4 11228 // CHECK11-NEXT: store double** [[B]], double*** [[B_ADDR]], align 4 11229 // CHECK11-NEXT: store double** [[C]], double*** [[C_ADDR]], align 4 11230 // CHECK11-NEXT: [[TMP0:%.*]] = load i32*, i32** [[N_ADDR]], align 4 11231 // CHECK11-NEXT: [[TMP1:%.*]] = load double**, double*** [[A_ADDR]], align 4 11232 // CHECK11-NEXT: [[TMP2:%.*]] = load double**, double*** [[B_ADDR]], align 4 11233 // CHECK11-NEXT: [[TMP3:%.*]] = load double**, double*** [[C_ADDR]], align 4 11234 // CHECK11-NEXT: [[TMP4:%.*]] = load i32, i32* [[TMP0]], align 4 11235 // CHECK11-NEXT: store i32 [[TMP4]], i32* [[DOTCAPTURE_EXPR_]], align 4 11236 // CHECK11-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 11237 // CHECK11-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP5]], 0 11238 // CHECK11-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 11239 // CHECK11-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1 11240 // CHECK11-NEXT: store i32 [[SUB2]], i32* [[DOTCAPTURE_EXPR_1]], align 4 11241 // CHECK11-NEXT: store i32 0, i32* [[I]], align 4 11242 // CHECK11-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 11243 // CHECK11-NEXT: [[CMP:%.*]] = icmp slt i32 0, [[TMP6]] 11244 // CHECK11-NEXT: br i1 [[CMP]], label [[OMP_PRECOND_THEN:%.*]], label [[OMP_PRECOND_END:%.*]] 11245 // CHECK11: omp.precond.then: 11246 // CHECK11-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4 11247 // CHECK11-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 11248 // CHECK11-NEXT: store i32 [[TMP7]], i32* [[DOTOMP_UB]], align 4 11249 // CHECK11-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTPREVIOUS_LB__ADDR]], align 4 11250 // CHECK11-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTPREVIOUS_UB__ADDR]], align 4 11251 // CHECK11-NEXT: store i32 [[TMP8]], i32* [[DOTOMP_LB]], align 4 11252 // CHECK11-NEXT: store i32 [[TMP9]], i32* [[DOTOMP_UB]], align 4 11253 // CHECK11-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4 11254 // CHECK11-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4 11255 // CHECK11-NEXT: [[TMP10:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 4 11256 // CHECK11-NEXT: [[TMP11:%.*]] = load i32, i32* [[TMP10]], align 4 11257 // CHECK11-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB2]], i32 [[TMP11]], i32 34, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1) 11258 // CHECK11-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 11259 // CHECK11-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 11260 // CHECK11-NEXT: [[CMP4:%.*]] = icmp sgt i32 [[TMP12]], [[TMP13]] 11261 // CHECK11-NEXT: br i1 [[CMP4]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] 11262 // CHECK11: cond.true: 11263 // CHECK11-NEXT: [[TMP14:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 11264 // CHECK11-NEXT: br label [[COND_END:%.*]] 11265 // CHECK11: cond.false: 11266 // CHECK11-NEXT: [[TMP15:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 11267 // CHECK11-NEXT: br label [[COND_END]] 11268 // CHECK11: cond.end: 11269 // CHECK11-NEXT: [[COND:%.*]] = phi i32 [ [[TMP14]], [[COND_TRUE]] ], [ [[TMP15]], [[COND_FALSE]] ] 11270 // CHECK11-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4 11271 // CHECK11-NEXT: [[TMP16:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4 11272 // CHECK11-NEXT: store i32 [[TMP16]], i32* [[DOTOMP_IV]], align 4 11273 // CHECK11-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] 11274 // CHECK11: omp.inner.for.cond: 11275 // CHECK11-NEXT: [[TMP17:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !42 11276 // CHECK11-NEXT: [[TMP18:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4, !llvm.access.group !42 11277 // CHECK11-NEXT: [[CMP5:%.*]] = icmp sle i32 [[TMP17]], [[TMP18]] 11278 // CHECK11-NEXT: br i1 [[CMP5]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] 11279 // CHECK11: omp.inner.for.body: 11280 // CHECK11-NEXT: [[TMP19:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !42 11281 // CHECK11-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP19]], 1 11282 // CHECK11-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] 11283 // CHECK11-NEXT: store i32 [[ADD]], i32* [[I3]], align 4, !llvm.access.group !42 11284 // CHECK11-NEXT: [[TMP20:%.*]] = load double*, double** [[TMP2]], align 4, !llvm.access.group !42 11285 // CHECK11-NEXT: [[TMP21:%.*]] = load i32, i32* [[I3]], align 4, !llvm.access.group !42 11286 // CHECK11-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds double, double* [[TMP20]], i32 [[TMP21]] 11287 // CHECK11-NEXT: [[TMP22:%.*]] = load double, double* [[ARRAYIDX]], align 4, !llvm.access.group !42 11288 // CHECK11-NEXT: [[TMP23:%.*]] = load double*, double** [[TMP3]], align 4, !llvm.access.group !42 11289 // CHECK11-NEXT: [[TMP24:%.*]] = load i32, i32* [[I3]], align 4, !llvm.access.group !42 11290 // CHECK11-NEXT: [[ARRAYIDX6:%.*]] = getelementptr inbounds double, double* [[TMP23]], i32 [[TMP24]] 11291 // CHECK11-NEXT: [[TMP25:%.*]] = load double, double* [[ARRAYIDX6]], align 4, !llvm.access.group !42 11292 // CHECK11-NEXT: [[ADD7:%.*]] = fadd double [[TMP22]], [[TMP25]] 11293 // CHECK11-NEXT: [[TMP26:%.*]] = load double*, double** [[TMP1]], align 4, !llvm.access.group !42 11294 // CHECK11-NEXT: [[TMP27:%.*]] = load i32, i32* [[I3]], align 4, !llvm.access.group !42 11295 // CHECK11-NEXT: [[ARRAYIDX8:%.*]] = getelementptr inbounds double, double* [[TMP26]], i32 [[TMP27]] 11296 // CHECK11-NEXT: store double [[ADD7]], double* [[ARRAYIDX8]], align 4, !llvm.access.group !42 11297 // CHECK11-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] 11298 // CHECK11: omp.body.continue: 11299 // CHECK11-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] 11300 // CHECK11: omp.inner.for.inc: 11301 // CHECK11-NEXT: [[TMP28:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !42 11302 // CHECK11-NEXT: [[ADD9:%.*]] = add nsw i32 [[TMP28]], 1 11303 // CHECK11-NEXT: store i32 [[ADD9]], i32* [[DOTOMP_IV]], align 4, !llvm.access.group !42 11304 // CHECK11-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP43:![0-9]+]] 11305 // CHECK11: omp.inner.for.end: 11306 // CHECK11-NEXT: br label [[OMP_LOOP_EXIT:%.*]] 11307 // CHECK11: omp.loop.exit: 11308 // CHECK11-NEXT: [[TMP29:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 4 11309 // CHECK11-NEXT: [[TMP30:%.*]] = load i32, i32* [[TMP29]], align 4 11310 // CHECK11-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP30]]) 11311 // CHECK11-NEXT: [[TMP31:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4 11312 // CHECK11-NEXT: [[TMP32:%.*]] = icmp ne i32 [[TMP31]], 0 11313 // CHECK11-NEXT: br i1 [[TMP32]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]] 11314 // CHECK11: .omp.final.then: 11315 // CHECK11-NEXT: [[TMP33:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 11316 // CHECK11-NEXT: [[SUB10:%.*]] = sub nsw i32 [[TMP33]], 0 11317 // CHECK11-NEXT: [[DIV11:%.*]] = sdiv i32 [[SUB10]], 1 11318 // CHECK11-NEXT: [[MUL12:%.*]] = mul nsw i32 [[DIV11]], 1 11319 // CHECK11-NEXT: [[ADD13:%.*]] = add nsw i32 0, [[MUL12]] 11320 // CHECK11-NEXT: store i32 [[ADD13]], i32* [[I3]], align 4 11321 // CHECK11-NEXT: br label [[DOTOMP_FINAL_DONE]] 11322 // CHECK11: .omp.final.done: 11323 // CHECK11-NEXT: br label [[OMP_PRECOND_END]] 11324 // CHECK11: omp.precond.end: 11325 // CHECK11-NEXT: ret void 11326 // 11327 // 11328 // CHECK11-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l505 11329 // CHECK11-SAME: (i32 noundef [[CH:%.*]], i32 noundef [[N:%.*]], double* noundef [[A:%.*]], double* noundef [[B:%.*]], double* noundef [[C:%.*]]) #[[ATTR1]] { 11330 // CHECK11-NEXT: entry: 11331 // CHECK11-NEXT: [[CH_ADDR:%.*]] = alloca i32, align 4 11332 // CHECK11-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4 11333 // CHECK11-NEXT: [[A_ADDR:%.*]] = alloca double*, align 4 11334 // CHECK11-NEXT: [[B_ADDR:%.*]] = alloca double*, align 4 11335 // CHECK11-NEXT: [[C_ADDR:%.*]] = alloca double*, align 4 11336 // CHECK11-NEXT: store i32 [[CH]], i32* [[CH_ADDR]], align 4 11337 // CHECK11-NEXT: store i32 [[N]], i32* [[N_ADDR]], align 4 11338 // CHECK11-NEXT: store double* [[A]], double** [[A_ADDR]], align 4 11339 // CHECK11-NEXT: store double* [[B]], double** [[B_ADDR]], align 4 11340 // CHECK11-NEXT: store double* [[C]], double** [[C_ADDR]], align 4 11341 // CHECK11-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 5, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32*, i32*, double**, double**, double**)* @.omp_outlined..14 to void (i32*, i32*, ...)*), i32* [[CH_ADDR]], i32* [[N_ADDR]], double** [[A_ADDR]], double** [[B_ADDR]], double** [[C_ADDR]]) 11342 // CHECK11-NEXT: ret void 11343 // 11344 // 11345 // CHECK11-LABEL: define {{[^@]+}}@.omp_outlined..14 11346 // CHECK11-SAME: (i32* noalias noundef [[DOTGLOBAL_TID_:%.*]], i32* noalias noundef [[DOTBOUND_TID_:%.*]], i32* noundef nonnull align 4 dereferenceable(4) [[CH:%.*]], i32* noundef nonnull align 4 dereferenceable(4) [[N:%.*]], double** noundef nonnull align 4 dereferenceable(4) [[A:%.*]], double** noundef nonnull align 4 dereferenceable(4) [[B:%.*]], double** noundef nonnull align 4 dereferenceable(4) [[C:%.*]]) #[[ATTR1]] { 11347 // CHECK11-NEXT: entry: 11348 // CHECK11-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4 11349 // CHECK11-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4 11350 // CHECK11-NEXT: [[CH_ADDR:%.*]] = alloca i32*, align 4 11351 // CHECK11-NEXT: [[N_ADDR:%.*]] = alloca i32*, align 4 11352 // CHECK11-NEXT: [[A_ADDR:%.*]] = alloca double**, align 4 11353 // CHECK11-NEXT: [[B_ADDR:%.*]] = alloca double**, align 4 11354 // CHECK11-NEXT: [[C_ADDR:%.*]] = alloca double**, align 4 11355 // CHECK11-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 11356 // CHECK11-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 11357 // CHECK11-NEXT: [[TMP:%.*]] = alloca i32, align 4 11358 // CHECK11-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 11359 // CHECK11-NEXT: [[DOTCAPTURE_EXPR_2:%.*]] = alloca i32, align 4 11360 // CHECK11-NEXT: [[I:%.*]] = alloca i32, align 4 11361 // CHECK11-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4 11362 // CHECK11-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4 11363 // CHECK11-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 11364 // CHECK11-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 11365 // CHECK11-NEXT: [[I4:%.*]] = alloca i32, align 4 11366 // CHECK11-NEXT: [[DOTCAPTURE_EXPR__CASTED:%.*]] = alloca i32, align 4 11367 // CHECK11-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 4 11368 // CHECK11-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 4 11369 // CHECK11-NEXT: store i32* [[CH]], i32** [[CH_ADDR]], align 4 11370 // CHECK11-NEXT: store i32* [[N]], i32** [[N_ADDR]], align 4 11371 // CHECK11-NEXT: store double** [[A]], double*** [[A_ADDR]], align 4 11372 // CHECK11-NEXT: store double** [[B]], double*** [[B_ADDR]], align 4 11373 // CHECK11-NEXT: store double** [[C]], double*** [[C_ADDR]], align 4 11374 // CHECK11-NEXT: [[TMP0:%.*]] = load i32*, i32** [[CH_ADDR]], align 4 11375 // CHECK11-NEXT: [[TMP1:%.*]] = load i32*, i32** [[N_ADDR]], align 4 11376 // CHECK11-NEXT: [[TMP2:%.*]] = load double**, double*** [[A_ADDR]], align 4 11377 // CHECK11-NEXT: [[TMP3:%.*]] = load double**, double*** [[B_ADDR]], align 4 11378 // CHECK11-NEXT: [[TMP4:%.*]] = load double**, double*** [[C_ADDR]], align 4 11379 // CHECK11-NEXT: [[TMP5:%.*]] = load i32, i32* [[TMP0]], align 4 11380 // CHECK11-NEXT: store i32 [[TMP5]], i32* [[DOTCAPTURE_EXPR_]], align 4 11381 // CHECK11-NEXT: [[TMP6:%.*]] = load i32, i32* [[TMP1]], align 4 11382 // CHECK11-NEXT: store i32 [[TMP6]], i32* [[DOTCAPTURE_EXPR_1]], align 4 11383 // CHECK11-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 11384 // CHECK11-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP7]], 0 11385 // CHECK11-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 11386 // CHECK11-NEXT: [[SUB3:%.*]] = sub nsw i32 [[DIV]], 1 11387 // CHECK11-NEXT: store i32 [[SUB3]], i32* [[DOTCAPTURE_EXPR_2]], align 4 11388 // CHECK11-NEXT: store i32 0, i32* [[I]], align 4 11389 // CHECK11-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 11390 // CHECK11-NEXT: [[CMP:%.*]] = icmp slt i32 0, [[TMP8]] 11391 // CHECK11-NEXT: br i1 [[CMP]], label [[OMP_PRECOND_THEN:%.*]], label [[OMP_PRECOND_END:%.*]] 11392 // CHECK11: omp.precond.then: 11393 // CHECK11-NEXT: store i32 0, i32* [[DOTOMP_COMB_LB]], align 4 11394 // CHECK11-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_2]], align 4 11395 // CHECK11-NEXT: store i32 [[TMP9]], i32* [[DOTOMP_COMB_UB]], align 4 11396 // CHECK11-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4 11397 // CHECK11-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4 11398 // CHECK11-NEXT: [[TMP10:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 4 11399 // CHECK11-NEXT: [[TMP11:%.*]] = load i32, i32* [[TMP10]], align 4 11400 // CHECK11-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1]], i32 [[TMP11]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_COMB_LB]], i32* [[DOTOMP_COMB_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1) 11401 // CHECK11-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 11402 // CHECK11-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_2]], align 4 11403 // CHECK11-NEXT: [[CMP5:%.*]] = icmp sgt i32 [[TMP12]], [[TMP13]] 11404 // CHECK11-NEXT: br i1 [[CMP5]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] 11405 // CHECK11: cond.true: 11406 // CHECK11-NEXT: [[TMP14:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_2]], align 4 11407 // CHECK11-NEXT: br label [[COND_END:%.*]] 11408 // CHECK11: cond.false: 11409 // CHECK11-NEXT: [[TMP15:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 11410 // CHECK11-NEXT: br label [[COND_END]] 11411 // CHECK11: cond.end: 11412 // CHECK11-NEXT: [[COND:%.*]] = phi i32 [ [[TMP14]], [[COND_TRUE]] ], [ [[TMP15]], [[COND_FALSE]] ] 11413 // CHECK11-NEXT: store i32 [[COND]], i32* [[DOTOMP_COMB_UB]], align 4 11414 // CHECK11-NEXT: [[TMP16:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4 11415 // CHECK11-NEXT: store i32 [[TMP16]], i32* [[DOTOMP_IV]], align 4 11416 // CHECK11-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] 11417 // CHECK11: omp.inner.for.cond: 11418 // CHECK11-NEXT: [[TMP17:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !45 11419 // CHECK11-NEXT: [[TMP18:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4, !llvm.access.group !45 11420 // CHECK11-NEXT: [[CMP6:%.*]] = icmp sle i32 [[TMP17]], [[TMP18]] 11421 // CHECK11-NEXT: br i1 [[CMP6]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] 11422 // CHECK11: omp.inner.for.body: 11423 // CHECK11-NEXT: [[TMP19:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4, !llvm.access.group !45 11424 // CHECK11-NEXT: [[TMP20:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4, !llvm.access.group !45 11425 // CHECK11-NEXT: [[TMP21:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4, !llvm.access.group !45 11426 // CHECK11-NEXT: store i32 [[TMP21]], i32* [[DOTCAPTURE_EXPR__CASTED]], align 4, !llvm.access.group !45 11427 // CHECK11-NEXT: [[TMP22:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR__CASTED]], align 4, !llvm.access.group !45 11428 // CHECK11-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 7, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32, i32, i32*, double**, double**, double**, i32)* @.omp_outlined..15 to void (i32*, i32*, ...)*), i32 [[TMP19]], i32 [[TMP20]], i32* [[TMP1]], double** [[TMP2]], double** [[TMP3]], double** [[TMP4]], i32 [[TMP22]]), !llvm.access.group !45 11429 // CHECK11-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] 11430 // CHECK11: omp.inner.for.inc: 11431 // CHECK11-NEXT: [[TMP23:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !45 11432 // CHECK11-NEXT: [[TMP24:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4, !llvm.access.group !45 11433 // CHECK11-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP23]], [[TMP24]] 11434 // CHECK11-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4, !llvm.access.group !45 11435 // CHECK11-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP46:![0-9]+]] 11436 // CHECK11: omp.inner.for.end: 11437 // CHECK11-NEXT: br label [[OMP_LOOP_EXIT:%.*]] 11438 // CHECK11: omp.loop.exit: 11439 // CHECK11-NEXT: [[TMP25:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 4 11440 // CHECK11-NEXT: [[TMP26:%.*]] = load i32, i32* [[TMP25]], align 4 11441 // CHECK11-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP26]]) 11442 // CHECK11-NEXT: [[TMP27:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4 11443 // CHECK11-NEXT: [[TMP28:%.*]] = icmp ne i32 [[TMP27]], 0 11444 // CHECK11-NEXT: br i1 [[TMP28]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]] 11445 // CHECK11: .omp.final.then: 11446 // CHECK11-NEXT: [[TMP29:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 11447 // CHECK11-NEXT: [[SUB7:%.*]] = sub nsw i32 [[TMP29]], 0 11448 // CHECK11-NEXT: [[DIV8:%.*]] = sdiv i32 [[SUB7]], 1 11449 // CHECK11-NEXT: [[MUL:%.*]] = mul nsw i32 [[DIV8]], 1 11450 // CHECK11-NEXT: [[ADD9:%.*]] = add nsw i32 0, [[MUL]] 11451 // CHECK11-NEXT: store i32 [[ADD9]], i32* [[I4]], align 4 11452 // CHECK11-NEXT: br label [[DOTOMP_FINAL_DONE]] 11453 // CHECK11: .omp.final.done: 11454 // CHECK11-NEXT: br label [[OMP_PRECOND_END]] 11455 // CHECK11: omp.precond.end: 11456 // CHECK11-NEXT: ret void 11457 // 11458 // 11459 // CHECK11-LABEL: define {{[^@]+}}@.omp_outlined..15 11460 // CHECK11-SAME: (i32* noalias noundef [[DOTGLOBAL_TID_:%.*]], i32* noalias noundef [[DOTBOUND_TID_:%.*]], i32 noundef [[DOTPREVIOUS_LB_:%.*]], i32 noundef [[DOTPREVIOUS_UB_:%.*]], i32* noundef nonnull align 4 dereferenceable(4) [[N:%.*]], double** noundef nonnull align 4 dereferenceable(4) [[A:%.*]], double** noundef nonnull align 4 dereferenceable(4) [[B:%.*]], double** noundef nonnull align 4 dereferenceable(4) [[C:%.*]], i32 noundef [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR1]] { 11461 // CHECK11-NEXT: entry: 11462 // CHECK11-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4 11463 // CHECK11-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4 11464 // CHECK11-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i32, align 4 11465 // CHECK11-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i32, align 4 11466 // CHECK11-NEXT: [[N_ADDR:%.*]] = alloca i32*, align 4 11467 // CHECK11-NEXT: [[A_ADDR:%.*]] = alloca double**, align 4 11468 // CHECK11-NEXT: [[B_ADDR:%.*]] = alloca double**, align 4 11469 // CHECK11-NEXT: [[C_ADDR:%.*]] = alloca double**, align 4 11470 // CHECK11-NEXT: [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i32, align 4 11471 // CHECK11-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 11472 // CHECK11-NEXT: [[TMP:%.*]] = alloca i32, align 4 11473 // CHECK11-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 11474 // CHECK11-NEXT: [[DOTCAPTURE_EXPR_2:%.*]] = alloca i32, align 4 11475 // CHECK11-NEXT: [[I:%.*]] = alloca i32, align 4 11476 // CHECK11-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 11477 // CHECK11-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4 11478 // CHECK11-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 11479 // CHECK11-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 11480 // CHECK11-NEXT: [[I4:%.*]] = alloca i32, align 4 11481 // CHECK11-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 4 11482 // CHECK11-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 4 11483 // CHECK11-NEXT: store i32 [[DOTPREVIOUS_LB_]], i32* [[DOTPREVIOUS_LB__ADDR]], align 4 11484 // CHECK11-NEXT: store i32 [[DOTPREVIOUS_UB_]], i32* [[DOTPREVIOUS_UB__ADDR]], align 4 11485 // CHECK11-NEXT: store i32* [[N]], i32** [[N_ADDR]], align 4 11486 // CHECK11-NEXT: store double** [[A]], double*** [[A_ADDR]], align 4 11487 // CHECK11-NEXT: store double** [[B]], double*** [[B_ADDR]], align 4 11488 // CHECK11-NEXT: store double** [[C]], double*** [[C_ADDR]], align 4 11489 // CHECK11-NEXT: store i32 [[DOTCAPTURE_EXPR_]], i32* [[DOTCAPTURE_EXPR__ADDR]], align 4 11490 // CHECK11-NEXT: [[TMP0:%.*]] = load i32*, i32** [[N_ADDR]], align 4 11491 // CHECK11-NEXT: [[TMP1:%.*]] = load double**, double*** [[A_ADDR]], align 4 11492 // CHECK11-NEXT: [[TMP2:%.*]] = load double**, double*** [[B_ADDR]], align 4 11493 // CHECK11-NEXT: [[TMP3:%.*]] = load double**, double*** [[C_ADDR]], align 4 11494 // CHECK11-NEXT: [[TMP4:%.*]] = load i32, i32* [[TMP0]], align 4 11495 // CHECK11-NEXT: store i32 [[TMP4]], i32* [[DOTCAPTURE_EXPR_1]], align 4 11496 // CHECK11-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 11497 // CHECK11-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP5]], 0 11498 // CHECK11-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 11499 // CHECK11-NEXT: [[SUB3:%.*]] = sub nsw i32 [[DIV]], 1 11500 // CHECK11-NEXT: store i32 [[SUB3]], i32* [[DOTCAPTURE_EXPR_2]], align 4 11501 // CHECK11-NEXT: store i32 0, i32* [[I]], align 4 11502 // CHECK11-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 11503 // CHECK11-NEXT: [[CMP:%.*]] = icmp slt i32 0, [[TMP6]] 11504 // CHECK11-NEXT: br i1 [[CMP]], label [[OMP_PRECOND_THEN:%.*]], label [[OMP_PRECOND_END:%.*]] 11505 // CHECK11: omp.precond.then: 11506 // CHECK11-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4 11507 // CHECK11-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_2]], align 4 11508 // CHECK11-NEXT: store i32 [[TMP7]], i32* [[DOTOMP_UB]], align 4 11509 // CHECK11-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTPREVIOUS_LB__ADDR]], align 4 11510 // CHECK11-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTPREVIOUS_UB__ADDR]], align 4 11511 // CHECK11-NEXT: store i32 [[TMP8]], i32* [[DOTOMP_LB]], align 4 11512 // CHECK11-NEXT: store i32 [[TMP9]], i32* [[DOTOMP_UB]], align 4 11513 // CHECK11-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4 11514 // CHECK11-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4 11515 // CHECK11-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR__ADDR]], align 4 11516 // CHECK11-NEXT: [[TMP11:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 4 11517 // CHECK11-NEXT: [[TMP12:%.*]] = load i32, i32* [[TMP11]], align 4 11518 // CHECK11-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB2]], i32 [[TMP12]], i32 33, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 [[TMP10]]) 11519 // CHECK11-NEXT: br label [[OMP_DISPATCH_COND:%.*]] 11520 // CHECK11: omp.dispatch.cond: 11521 // CHECK11-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 11522 // CHECK11-NEXT: [[TMP14:%.*]] = load i32, i32* [[DOTPREVIOUS_UB__ADDR]], align 4 11523 // CHECK11-NEXT: [[CMP5:%.*]] = icmp sgt i32 [[TMP13]], [[TMP14]] 11524 // CHECK11-NEXT: br i1 [[CMP5]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] 11525 // CHECK11: cond.true: 11526 // CHECK11-NEXT: [[TMP15:%.*]] = load i32, i32* [[DOTPREVIOUS_UB__ADDR]], align 4 11527 // CHECK11-NEXT: br label [[COND_END:%.*]] 11528 // CHECK11: cond.false: 11529 // CHECK11-NEXT: [[TMP16:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 11530 // CHECK11-NEXT: br label [[COND_END]] 11531 // CHECK11: cond.end: 11532 // CHECK11-NEXT: [[COND:%.*]] = phi i32 [ [[TMP15]], [[COND_TRUE]] ], [ [[TMP16]], [[COND_FALSE]] ] 11533 // CHECK11-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4 11534 // CHECK11-NEXT: [[TMP17:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4 11535 // CHECK11-NEXT: store i32 [[TMP17]], i32* [[DOTOMP_IV]], align 4 11536 // CHECK11-NEXT: [[TMP18:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 11537 // CHECK11-NEXT: [[TMP19:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 11538 // CHECK11-NEXT: [[CMP6:%.*]] = icmp sle i32 [[TMP18]], [[TMP19]] 11539 // CHECK11-NEXT: br i1 [[CMP6]], label [[OMP_DISPATCH_BODY:%.*]], label [[OMP_DISPATCH_END:%.*]] 11540 // CHECK11: omp.dispatch.body: 11541 // CHECK11-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] 11542 // CHECK11: omp.inner.for.cond: 11543 // CHECK11-NEXT: [[TMP20:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !48 11544 // CHECK11-NEXT: [[TMP21:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4, !llvm.access.group !48 11545 // CHECK11-NEXT: [[CMP7:%.*]] = icmp sle i32 [[TMP20]], [[TMP21]] 11546 // CHECK11-NEXT: br i1 [[CMP7]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] 11547 // CHECK11: omp.inner.for.body: 11548 // CHECK11-NEXT: [[TMP22:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !48 11549 // CHECK11-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP22]], 1 11550 // CHECK11-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] 11551 // CHECK11-NEXT: store i32 [[ADD]], i32* [[I4]], align 4, !llvm.access.group !48 11552 // CHECK11-NEXT: [[TMP23:%.*]] = load double*, double** [[TMP2]], align 4, !llvm.access.group !48 11553 // CHECK11-NEXT: [[TMP24:%.*]] = load i32, i32* [[I4]], align 4, !llvm.access.group !48 11554 // CHECK11-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds double, double* [[TMP23]], i32 [[TMP24]] 11555 // CHECK11-NEXT: [[TMP25:%.*]] = load double, double* [[ARRAYIDX]], align 4, !llvm.access.group !48 11556 // CHECK11-NEXT: [[TMP26:%.*]] = load double*, double** [[TMP3]], align 4, !llvm.access.group !48 11557 // CHECK11-NEXT: [[TMP27:%.*]] = load i32, i32* [[I4]], align 4, !llvm.access.group !48 11558 // CHECK11-NEXT: [[ARRAYIDX8:%.*]] = getelementptr inbounds double, double* [[TMP26]], i32 [[TMP27]] 11559 // CHECK11-NEXT: [[TMP28:%.*]] = load double, double* [[ARRAYIDX8]], align 4, !llvm.access.group !48 11560 // CHECK11-NEXT: [[ADD9:%.*]] = fadd double [[TMP25]], [[TMP28]] 11561 // CHECK11-NEXT: [[TMP29:%.*]] = load double*, double** [[TMP1]], align 4, !llvm.access.group !48 11562 // CHECK11-NEXT: [[TMP30:%.*]] = load i32, i32* [[I4]], align 4, !llvm.access.group !48 11563 // CHECK11-NEXT: [[ARRAYIDX10:%.*]] = getelementptr inbounds double, double* [[TMP29]], i32 [[TMP30]] 11564 // CHECK11-NEXT: store double [[ADD9]], double* [[ARRAYIDX10]], align 4, !llvm.access.group !48 11565 // CHECK11-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] 11566 // CHECK11: omp.body.continue: 11567 // CHECK11-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] 11568 // CHECK11: omp.inner.for.inc: 11569 // CHECK11-NEXT: [[TMP31:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !48 11570 // CHECK11-NEXT: [[ADD11:%.*]] = add nsw i32 [[TMP31]], 1 11571 // CHECK11-NEXT: store i32 [[ADD11]], i32* [[DOTOMP_IV]], align 4, !llvm.access.group !48 11572 // CHECK11-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP49:![0-9]+]] 11573 // CHECK11: omp.inner.for.end: 11574 // CHECK11-NEXT: br label [[OMP_DISPATCH_INC:%.*]] 11575 // CHECK11: omp.dispatch.inc: 11576 // CHECK11-NEXT: [[TMP32:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4 11577 // CHECK11-NEXT: [[TMP33:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4 11578 // CHECK11-NEXT: [[ADD12:%.*]] = add nsw i32 [[TMP32]], [[TMP33]] 11579 // CHECK11-NEXT: store i32 [[ADD12]], i32* [[DOTOMP_LB]], align 4 11580 // CHECK11-NEXT: [[TMP34:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 11581 // CHECK11-NEXT: [[TMP35:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4 11582 // CHECK11-NEXT: [[ADD13:%.*]] = add nsw i32 [[TMP34]], [[TMP35]] 11583 // CHECK11-NEXT: store i32 [[ADD13]], i32* [[DOTOMP_UB]], align 4 11584 // CHECK11-NEXT: br label [[OMP_DISPATCH_COND]] 11585 // CHECK11: omp.dispatch.end: 11586 // CHECK11-NEXT: [[TMP36:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 4 11587 // CHECK11-NEXT: [[TMP37:%.*]] = load i32, i32* [[TMP36]], align 4 11588 // CHECK11-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP37]]) 11589 // CHECK11-NEXT: [[TMP38:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4 11590 // CHECK11-NEXT: [[TMP39:%.*]] = icmp ne i32 [[TMP38]], 0 11591 // CHECK11-NEXT: br i1 [[TMP39]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]] 11592 // CHECK11: .omp.final.then: 11593 // CHECK11-NEXT: [[TMP40:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 11594 // CHECK11-NEXT: [[SUB14:%.*]] = sub nsw i32 [[TMP40]], 0 11595 // CHECK11-NEXT: [[DIV15:%.*]] = sdiv i32 [[SUB14]], 1 11596 // CHECK11-NEXT: [[MUL16:%.*]] = mul nsw i32 [[DIV15]], 1 11597 // CHECK11-NEXT: [[ADD17:%.*]] = add nsw i32 0, [[MUL16]] 11598 // CHECK11-NEXT: store i32 [[ADD17]], i32* [[I4]], align 4 11599 // CHECK11-NEXT: br label [[DOTOMP_FINAL_DONE]] 11600 // CHECK11: .omp.final.done: 11601 // CHECK11-NEXT: br label [[OMP_PRECOND_END]] 11602 // CHECK11: omp.precond.end: 11603 // CHECK11-NEXT: ret void 11604 // 11605 // 11606 // CHECK11-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l535 11607 // CHECK11-SAME: (i32 noundef [[N:%.*]], double* noundef [[A:%.*]], double* noundef [[B:%.*]], double* noundef [[C:%.*]]) #[[ATTR1]] { 11608 // CHECK11-NEXT: entry: 11609 // CHECK11-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4 11610 // CHECK11-NEXT: [[A_ADDR:%.*]] = alloca double*, align 4 11611 // CHECK11-NEXT: [[B_ADDR:%.*]] = alloca double*, align 4 11612 // CHECK11-NEXT: [[C_ADDR:%.*]] = alloca double*, align 4 11613 // CHECK11-NEXT: store i32 [[N]], i32* [[N_ADDR]], align 4 11614 // CHECK11-NEXT: store double* [[A]], double** [[A_ADDR]], align 4 11615 // CHECK11-NEXT: store double* [[B]], double** [[B_ADDR]], align 4 11616 // CHECK11-NEXT: store double* [[C]], double** [[C_ADDR]], align 4 11617 // CHECK11-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 4, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32*, double**, double**, double**)* @.omp_outlined..18 to void (i32*, i32*, ...)*), i32* [[N_ADDR]], double** [[A_ADDR]], double** [[B_ADDR]], double** [[C_ADDR]]) 11618 // CHECK11-NEXT: ret void 11619 // 11620 // 11621 // CHECK11-LABEL: define {{[^@]+}}@.omp_outlined..18 11622 // CHECK11-SAME: (i32* noalias noundef [[DOTGLOBAL_TID_:%.*]], i32* noalias noundef [[DOTBOUND_TID_:%.*]], i32* noundef nonnull align 4 dereferenceable(4) [[N:%.*]], double** noundef nonnull align 4 dereferenceable(4) [[A:%.*]], double** noundef nonnull align 4 dereferenceable(4) [[B:%.*]], double** noundef nonnull align 4 dereferenceable(4) [[C:%.*]]) #[[ATTR1]] { 11623 // CHECK11-NEXT: entry: 11624 // CHECK11-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4 11625 // CHECK11-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4 11626 // CHECK11-NEXT: [[N_ADDR:%.*]] = alloca i32*, align 4 11627 // CHECK11-NEXT: [[A_ADDR:%.*]] = alloca double**, align 4 11628 // CHECK11-NEXT: [[B_ADDR:%.*]] = alloca double**, align 4 11629 // CHECK11-NEXT: [[C_ADDR:%.*]] = alloca double**, align 4 11630 // CHECK11-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 11631 // CHECK11-NEXT: [[TMP:%.*]] = alloca i32, align 4 11632 // CHECK11-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 11633 // CHECK11-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 11634 // CHECK11-NEXT: [[I:%.*]] = alloca i32, align 4 11635 // CHECK11-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4 11636 // CHECK11-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4 11637 // CHECK11-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 11638 // CHECK11-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 11639 // CHECK11-NEXT: [[I3:%.*]] = alloca i32, align 4 11640 // CHECK11-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 4 11641 // CHECK11-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 4 11642 // CHECK11-NEXT: store i32* [[N]], i32** [[N_ADDR]], align 4 11643 // CHECK11-NEXT: store double** [[A]], double*** [[A_ADDR]], align 4 11644 // CHECK11-NEXT: store double** [[B]], double*** [[B_ADDR]], align 4 11645 // CHECK11-NEXT: store double** [[C]], double*** [[C_ADDR]], align 4 11646 // CHECK11-NEXT: [[TMP0:%.*]] = load i32*, i32** [[N_ADDR]], align 4 11647 // CHECK11-NEXT: [[TMP1:%.*]] = load double**, double*** [[A_ADDR]], align 4 11648 // CHECK11-NEXT: [[TMP2:%.*]] = load double**, double*** [[B_ADDR]], align 4 11649 // CHECK11-NEXT: [[TMP3:%.*]] = load double**, double*** [[C_ADDR]], align 4 11650 // CHECK11-NEXT: [[TMP4:%.*]] = load i32, i32* [[TMP0]], align 4 11651 // CHECK11-NEXT: store i32 [[TMP4]], i32* [[DOTCAPTURE_EXPR_]], align 4 11652 // CHECK11-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 11653 // CHECK11-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP5]], 0 11654 // CHECK11-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 11655 // CHECK11-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1 11656 // CHECK11-NEXT: store i32 [[SUB2]], i32* [[DOTCAPTURE_EXPR_1]], align 4 11657 // CHECK11-NEXT: store i32 0, i32* [[I]], align 4 11658 // CHECK11-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 11659 // CHECK11-NEXT: [[CMP:%.*]] = icmp slt i32 0, [[TMP6]] 11660 // CHECK11-NEXT: br i1 [[CMP]], label [[OMP_PRECOND_THEN:%.*]], label [[OMP_PRECOND_END:%.*]] 11661 // CHECK11: omp.precond.then: 11662 // CHECK11-NEXT: store i32 0, i32* [[DOTOMP_COMB_LB]], align 4 11663 // CHECK11-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 11664 // CHECK11-NEXT: store i32 [[TMP7]], i32* [[DOTOMP_COMB_UB]], align 4 11665 // CHECK11-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4 11666 // CHECK11-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4 11667 // CHECK11-NEXT: [[TMP8:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 4 11668 // CHECK11-NEXT: [[TMP9:%.*]] = load i32, i32* [[TMP8]], align 4 11669 // CHECK11-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1]], i32 [[TMP9]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_COMB_LB]], i32* [[DOTOMP_COMB_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1) 11670 // CHECK11-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 11671 // CHECK11-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 11672 // CHECK11-NEXT: [[CMP4:%.*]] = icmp sgt i32 [[TMP10]], [[TMP11]] 11673 // CHECK11-NEXT: br i1 [[CMP4]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] 11674 // CHECK11: cond.true: 11675 // CHECK11-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 11676 // CHECK11-NEXT: br label [[COND_END:%.*]] 11677 // CHECK11: cond.false: 11678 // CHECK11-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 11679 // CHECK11-NEXT: br label [[COND_END]] 11680 // CHECK11: cond.end: 11681 // CHECK11-NEXT: [[COND:%.*]] = phi i32 [ [[TMP12]], [[COND_TRUE]] ], [ [[TMP13]], [[COND_FALSE]] ] 11682 // CHECK11-NEXT: store i32 [[COND]], i32* [[DOTOMP_COMB_UB]], align 4 11683 // CHECK11-NEXT: [[TMP14:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4 11684 // CHECK11-NEXT: store i32 [[TMP14]], i32* [[DOTOMP_IV]], align 4 11685 // CHECK11-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] 11686 // CHECK11: omp.inner.for.cond: 11687 // CHECK11-NEXT: [[TMP15:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !51 11688 // CHECK11-NEXT: [[TMP16:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4, !llvm.access.group !51 11689 // CHECK11-NEXT: [[CMP5:%.*]] = icmp sle i32 [[TMP15]], [[TMP16]] 11690 // CHECK11-NEXT: br i1 [[CMP5]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] 11691 // CHECK11: omp.inner.for.body: 11692 // CHECK11-NEXT: [[TMP17:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4, !llvm.access.group !51 11693 // CHECK11-NEXT: [[TMP18:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4, !llvm.access.group !51 11694 // CHECK11-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 6, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32, i32, i32*, double**, double**, double**)* @.omp_outlined..19 to void (i32*, i32*, ...)*), i32 [[TMP17]], i32 [[TMP18]], i32* [[TMP0]], double** [[TMP1]], double** [[TMP2]], double** [[TMP3]]), !llvm.access.group !51 11695 // CHECK11-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] 11696 // CHECK11: omp.inner.for.inc: 11697 // CHECK11-NEXT: [[TMP19:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !51 11698 // CHECK11-NEXT: [[TMP20:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4, !llvm.access.group !51 11699 // CHECK11-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP19]], [[TMP20]] 11700 // CHECK11-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4, !llvm.access.group !51 11701 // CHECK11-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP52:![0-9]+]] 11702 // CHECK11: omp.inner.for.end: 11703 // CHECK11-NEXT: br label [[OMP_LOOP_EXIT:%.*]] 11704 // CHECK11: omp.loop.exit: 11705 // CHECK11-NEXT: [[TMP21:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 4 11706 // CHECK11-NEXT: [[TMP22:%.*]] = load i32, i32* [[TMP21]], align 4 11707 // CHECK11-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP22]]) 11708 // CHECK11-NEXT: [[TMP23:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4 11709 // CHECK11-NEXT: [[TMP24:%.*]] = icmp ne i32 [[TMP23]], 0 11710 // CHECK11-NEXT: br i1 [[TMP24]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]] 11711 // CHECK11: .omp.final.then: 11712 // CHECK11-NEXT: [[TMP25:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 11713 // CHECK11-NEXT: [[SUB6:%.*]] = sub nsw i32 [[TMP25]], 0 11714 // CHECK11-NEXT: [[DIV7:%.*]] = sdiv i32 [[SUB6]], 1 11715 // CHECK11-NEXT: [[MUL:%.*]] = mul nsw i32 [[DIV7]], 1 11716 // CHECK11-NEXT: [[ADD8:%.*]] = add nsw i32 0, [[MUL]] 11717 // CHECK11-NEXT: store i32 [[ADD8]], i32* [[I3]], align 4 11718 // CHECK11-NEXT: br label [[DOTOMP_FINAL_DONE]] 11719 // CHECK11: .omp.final.done: 11720 // CHECK11-NEXT: br label [[OMP_PRECOND_END]] 11721 // CHECK11: omp.precond.end: 11722 // CHECK11-NEXT: ret void 11723 // 11724 // 11725 // CHECK11-LABEL: define {{[^@]+}}@.omp_outlined..19 11726 // CHECK11-SAME: (i32* noalias noundef [[DOTGLOBAL_TID_:%.*]], i32* noalias noundef [[DOTBOUND_TID_:%.*]], i32 noundef [[DOTPREVIOUS_LB_:%.*]], i32 noundef [[DOTPREVIOUS_UB_:%.*]], i32* noundef nonnull align 4 dereferenceable(4) [[N:%.*]], double** noundef nonnull align 4 dereferenceable(4) [[A:%.*]], double** noundef nonnull align 4 dereferenceable(4) [[B:%.*]], double** noundef nonnull align 4 dereferenceable(4) [[C:%.*]]) #[[ATTR1]] { 11727 // CHECK11-NEXT: entry: 11728 // CHECK11-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4 11729 // CHECK11-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4 11730 // CHECK11-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i32, align 4 11731 // CHECK11-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i32, align 4 11732 // CHECK11-NEXT: [[N_ADDR:%.*]] = alloca i32*, align 4 11733 // CHECK11-NEXT: [[A_ADDR:%.*]] = alloca double**, align 4 11734 // CHECK11-NEXT: [[B_ADDR:%.*]] = alloca double**, align 4 11735 // CHECK11-NEXT: [[C_ADDR:%.*]] = alloca double**, align 4 11736 // CHECK11-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 11737 // CHECK11-NEXT: [[TMP:%.*]] = alloca i32, align 4 11738 // CHECK11-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 11739 // CHECK11-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 11740 // CHECK11-NEXT: [[I:%.*]] = alloca i32, align 4 11741 // CHECK11-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 11742 // CHECK11-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4 11743 // CHECK11-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 11744 // CHECK11-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 11745 // CHECK11-NEXT: [[I3:%.*]] = alloca i32, align 4 11746 // CHECK11-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 4 11747 // CHECK11-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 4 11748 // CHECK11-NEXT: store i32 [[DOTPREVIOUS_LB_]], i32* [[DOTPREVIOUS_LB__ADDR]], align 4 11749 // CHECK11-NEXT: store i32 [[DOTPREVIOUS_UB_]], i32* [[DOTPREVIOUS_UB__ADDR]], align 4 11750 // CHECK11-NEXT: store i32* [[N]], i32** [[N_ADDR]], align 4 11751 // CHECK11-NEXT: store double** [[A]], double*** [[A_ADDR]], align 4 11752 // CHECK11-NEXT: store double** [[B]], double*** [[B_ADDR]], align 4 11753 // CHECK11-NEXT: store double** [[C]], double*** [[C_ADDR]], align 4 11754 // CHECK11-NEXT: [[TMP0:%.*]] = load i32*, i32** [[N_ADDR]], align 4 11755 // CHECK11-NEXT: [[TMP1:%.*]] = load double**, double*** [[A_ADDR]], align 4 11756 // CHECK11-NEXT: [[TMP2:%.*]] = load double**, double*** [[B_ADDR]], align 4 11757 // CHECK11-NEXT: [[TMP3:%.*]] = load double**, double*** [[C_ADDR]], align 4 11758 // CHECK11-NEXT: [[TMP4:%.*]] = load i32, i32* [[TMP0]], align 4 11759 // CHECK11-NEXT: store i32 [[TMP4]], i32* [[DOTCAPTURE_EXPR_]], align 4 11760 // CHECK11-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 11761 // CHECK11-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP5]], 0 11762 // CHECK11-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 11763 // CHECK11-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1 11764 // CHECK11-NEXT: store i32 [[SUB2]], i32* [[DOTCAPTURE_EXPR_1]], align 4 11765 // CHECK11-NEXT: store i32 0, i32* [[I]], align 4 11766 // CHECK11-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 11767 // CHECK11-NEXT: [[CMP:%.*]] = icmp slt i32 0, [[TMP6]] 11768 // CHECK11-NEXT: br i1 [[CMP]], label [[OMP_PRECOND_THEN:%.*]], label [[OMP_PRECOND_END:%.*]] 11769 // CHECK11: omp.precond.then: 11770 // CHECK11-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4 11771 // CHECK11-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 11772 // CHECK11-NEXT: store i32 [[TMP7]], i32* [[DOTOMP_UB]], align 4 11773 // CHECK11-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTPREVIOUS_LB__ADDR]], align 4 11774 // CHECK11-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTPREVIOUS_UB__ADDR]], align 4 11775 // CHECK11-NEXT: store i32 [[TMP8]], i32* [[DOTOMP_LB]], align 4 11776 // CHECK11-NEXT: store i32 [[TMP9]], i32* [[DOTOMP_UB]], align 4 11777 // CHECK11-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4 11778 // CHECK11-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4 11779 // CHECK11-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4 11780 // CHECK11-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 11781 // CHECK11-NEXT: [[TMP12:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 4 11782 // CHECK11-NEXT: [[TMP13:%.*]] = load i32, i32* [[TMP12]], align 4 11783 // CHECK11-NEXT: call void @__kmpc_dispatch_init_4(%struct.ident_t* @[[GLOB3]], i32 [[TMP13]], i32 35, i32 [[TMP10]], i32 [[TMP11]], i32 1, i32 1) 11784 // CHECK11-NEXT: br label [[OMP_DISPATCH_COND:%.*]] 11785 // CHECK11: omp.dispatch.cond: 11786 // CHECK11-NEXT: [[TMP14:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 4 11787 // CHECK11-NEXT: [[TMP15:%.*]] = load i32, i32* [[TMP14]], align 4 11788 // CHECK11-NEXT: [[TMP16:%.*]] = call i32 @__kmpc_dispatch_next_4(%struct.ident_t* @[[GLOB3]], i32 [[TMP15]], i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]]) 11789 // CHECK11-NEXT: [[TOBOOL:%.*]] = icmp ne i32 [[TMP16]], 0 11790 // CHECK11-NEXT: br i1 [[TOBOOL]], label [[OMP_DISPATCH_BODY:%.*]], label [[OMP_DISPATCH_END:%.*]] 11791 // CHECK11: omp.dispatch.body: 11792 // CHECK11-NEXT: [[TMP17:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4 11793 // CHECK11-NEXT: store i32 [[TMP17]], i32* [[DOTOMP_IV]], align 4 11794 // CHECK11-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] 11795 // CHECK11: omp.inner.for.cond: 11796 // CHECK11-NEXT: [[TMP18:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !54 11797 // CHECK11-NEXT: [[TMP19:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4, !llvm.access.group !54 11798 // CHECK11-NEXT: [[CMP4:%.*]] = icmp sle i32 [[TMP18]], [[TMP19]] 11799 // CHECK11-NEXT: br i1 [[CMP4]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] 11800 // CHECK11: omp.inner.for.body: 11801 // CHECK11-NEXT: [[TMP20:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !54 11802 // CHECK11-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP20]], 1 11803 // CHECK11-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] 11804 // CHECK11-NEXT: store i32 [[ADD]], i32* [[I3]], align 4, !llvm.access.group !54 11805 // CHECK11-NEXT: [[TMP21:%.*]] = load double*, double** [[TMP2]], align 4, !llvm.access.group !54 11806 // CHECK11-NEXT: [[TMP22:%.*]] = load i32, i32* [[I3]], align 4, !llvm.access.group !54 11807 // CHECK11-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds double, double* [[TMP21]], i32 [[TMP22]] 11808 // CHECK11-NEXT: [[TMP23:%.*]] = load double, double* [[ARRAYIDX]], align 4, !llvm.access.group !54 11809 // CHECK11-NEXT: [[TMP24:%.*]] = load double*, double** [[TMP3]], align 4, !llvm.access.group !54 11810 // CHECK11-NEXT: [[TMP25:%.*]] = load i32, i32* [[I3]], align 4, !llvm.access.group !54 11811 // CHECK11-NEXT: [[ARRAYIDX5:%.*]] = getelementptr inbounds double, double* [[TMP24]], i32 [[TMP25]] 11812 // CHECK11-NEXT: [[TMP26:%.*]] = load double, double* [[ARRAYIDX5]], align 4, !llvm.access.group !54 11813 // CHECK11-NEXT: [[ADD6:%.*]] = fadd double [[TMP23]], [[TMP26]] 11814 // CHECK11-NEXT: [[TMP27:%.*]] = load double*, double** [[TMP1]], align 4, !llvm.access.group !54 11815 // CHECK11-NEXT: [[TMP28:%.*]] = load i32, i32* [[I3]], align 4, !llvm.access.group !54 11816 // CHECK11-NEXT: [[ARRAYIDX7:%.*]] = getelementptr inbounds double, double* [[TMP27]], i32 [[TMP28]] 11817 // CHECK11-NEXT: store double [[ADD6]], double* [[ARRAYIDX7]], align 4, !llvm.access.group !54 11818 // CHECK11-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] 11819 // CHECK11: omp.body.continue: 11820 // CHECK11-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] 11821 // CHECK11: omp.inner.for.inc: 11822 // CHECK11-NEXT: [[TMP29:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !54 11823 // CHECK11-NEXT: [[ADD8:%.*]] = add nsw i32 [[TMP29]], 1 11824 // CHECK11-NEXT: store i32 [[ADD8]], i32* [[DOTOMP_IV]], align 4, !llvm.access.group !54 11825 // CHECK11-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP55:![0-9]+]] 11826 // CHECK11: omp.inner.for.end: 11827 // CHECK11-NEXT: br label [[OMP_DISPATCH_INC:%.*]] 11828 // CHECK11: omp.dispatch.inc: 11829 // CHECK11-NEXT: br label [[OMP_DISPATCH_COND]] 11830 // CHECK11: omp.dispatch.end: 11831 // CHECK11-NEXT: [[TMP30:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4 11832 // CHECK11-NEXT: [[TMP31:%.*]] = icmp ne i32 [[TMP30]], 0 11833 // CHECK11-NEXT: br i1 [[TMP31]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]] 11834 // CHECK11: .omp.final.then: 11835 // CHECK11-NEXT: [[TMP32:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 11836 // CHECK11-NEXT: [[SUB9:%.*]] = sub nsw i32 [[TMP32]], 0 11837 // CHECK11-NEXT: [[DIV10:%.*]] = sdiv i32 [[SUB9]], 1 11838 // CHECK11-NEXT: [[MUL11:%.*]] = mul nsw i32 [[DIV10]], 1 11839 // CHECK11-NEXT: [[ADD12:%.*]] = add nsw i32 0, [[MUL11]] 11840 // CHECK11-NEXT: store i32 [[ADD12]], i32* [[I3]], align 4 11841 // CHECK11-NEXT: br label [[DOTOMP_FINAL_DONE]] 11842 // CHECK11: .omp.final.done: 11843 // CHECK11-NEXT: br label [[OMP_PRECOND_END]] 11844 // CHECK11: omp.precond.end: 11845 // CHECK11-NEXT: ret void 11846 // 11847 // 11848 // CHECK11-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l561 11849 // CHECK11-SAME: (i32 noundef [[CH:%.*]], i32 noundef [[N:%.*]], double* noundef [[A:%.*]], double* noundef [[B:%.*]], double* noundef [[C:%.*]]) #[[ATTR1]] { 11850 // CHECK11-NEXT: entry: 11851 // CHECK11-NEXT: [[CH_ADDR:%.*]] = alloca i32, align 4 11852 // CHECK11-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4 11853 // CHECK11-NEXT: [[A_ADDR:%.*]] = alloca double*, align 4 11854 // CHECK11-NEXT: [[B_ADDR:%.*]] = alloca double*, align 4 11855 // CHECK11-NEXT: [[C_ADDR:%.*]] = alloca double*, align 4 11856 // CHECK11-NEXT: store i32 [[CH]], i32* [[CH_ADDR]], align 4 11857 // CHECK11-NEXT: store i32 [[N]], i32* [[N_ADDR]], align 4 11858 // CHECK11-NEXT: store double* [[A]], double** [[A_ADDR]], align 4 11859 // CHECK11-NEXT: store double* [[B]], double** [[B_ADDR]], align 4 11860 // CHECK11-NEXT: store double* [[C]], double** [[C_ADDR]], align 4 11861 // CHECK11-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 5, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32*, i32*, double**, double**, double**)* @.omp_outlined..22 to void (i32*, i32*, ...)*), i32* [[CH_ADDR]], i32* [[N_ADDR]], double** [[A_ADDR]], double** [[B_ADDR]], double** [[C_ADDR]]) 11862 // CHECK11-NEXT: ret void 11863 // 11864 // 11865 // CHECK11-LABEL: define {{[^@]+}}@.omp_outlined..22 11866 // CHECK11-SAME: (i32* noalias noundef [[DOTGLOBAL_TID_:%.*]], i32* noalias noundef [[DOTBOUND_TID_:%.*]], i32* noundef nonnull align 4 dereferenceable(4) [[CH:%.*]], i32* noundef nonnull align 4 dereferenceable(4) [[N:%.*]], double** noundef nonnull align 4 dereferenceable(4) [[A:%.*]], double** noundef nonnull align 4 dereferenceable(4) [[B:%.*]], double** noundef nonnull align 4 dereferenceable(4) [[C:%.*]]) #[[ATTR1]] { 11867 // CHECK11-NEXT: entry: 11868 // CHECK11-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4 11869 // CHECK11-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4 11870 // CHECK11-NEXT: [[CH_ADDR:%.*]] = alloca i32*, align 4 11871 // CHECK11-NEXT: [[N_ADDR:%.*]] = alloca i32*, align 4 11872 // CHECK11-NEXT: [[A_ADDR:%.*]] = alloca double**, align 4 11873 // CHECK11-NEXT: [[B_ADDR:%.*]] = alloca double**, align 4 11874 // CHECK11-NEXT: [[C_ADDR:%.*]] = alloca double**, align 4 11875 // CHECK11-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 11876 // CHECK11-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 11877 // CHECK11-NEXT: [[TMP:%.*]] = alloca i32, align 4 11878 // CHECK11-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 11879 // CHECK11-NEXT: [[DOTCAPTURE_EXPR_2:%.*]] = alloca i32, align 4 11880 // CHECK11-NEXT: [[I:%.*]] = alloca i32, align 4 11881 // CHECK11-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4 11882 // CHECK11-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4 11883 // CHECK11-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 11884 // CHECK11-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 11885 // CHECK11-NEXT: [[I4:%.*]] = alloca i32, align 4 11886 // CHECK11-NEXT: [[DOTCAPTURE_EXPR__CASTED:%.*]] = alloca i32, align 4 11887 // CHECK11-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 4 11888 // CHECK11-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 4 11889 // CHECK11-NEXT: store i32* [[CH]], i32** [[CH_ADDR]], align 4 11890 // CHECK11-NEXT: store i32* [[N]], i32** [[N_ADDR]], align 4 11891 // CHECK11-NEXT: store double** [[A]], double*** [[A_ADDR]], align 4 11892 // CHECK11-NEXT: store double** [[B]], double*** [[B_ADDR]], align 4 11893 // CHECK11-NEXT: store double** [[C]], double*** [[C_ADDR]], align 4 11894 // CHECK11-NEXT: [[TMP0:%.*]] = load i32*, i32** [[CH_ADDR]], align 4 11895 // CHECK11-NEXT: [[TMP1:%.*]] = load i32*, i32** [[N_ADDR]], align 4 11896 // CHECK11-NEXT: [[TMP2:%.*]] = load double**, double*** [[A_ADDR]], align 4 11897 // CHECK11-NEXT: [[TMP3:%.*]] = load double**, double*** [[B_ADDR]], align 4 11898 // CHECK11-NEXT: [[TMP4:%.*]] = load double**, double*** [[C_ADDR]], align 4 11899 // CHECK11-NEXT: [[TMP5:%.*]] = load i32, i32* [[TMP0]], align 4 11900 // CHECK11-NEXT: store i32 [[TMP5]], i32* [[DOTCAPTURE_EXPR_]], align 4 11901 // CHECK11-NEXT: [[TMP6:%.*]] = load i32, i32* [[TMP1]], align 4 11902 // CHECK11-NEXT: store i32 [[TMP6]], i32* [[DOTCAPTURE_EXPR_1]], align 4 11903 // CHECK11-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 11904 // CHECK11-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP7]], 0 11905 // CHECK11-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 11906 // CHECK11-NEXT: [[SUB3:%.*]] = sub nsw i32 [[DIV]], 1 11907 // CHECK11-NEXT: store i32 [[SUB3]], i32* [[DOTCAPTURE_EXPR_2]], align 4 11908 // CHECK11-NEXT: store i32 0, i32* [[I]], align 4 11909 // CHECK11-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 11910 // CHECK11-NEXT: [[CMP:%.*]] = icmp slt i32 0, [[TMP8]] 11911 // CHECK11-NEXT: br i1 [[CMP]], label [[OMP_PRECOND_THEN:%.*]], label [[OMP_PRECOND_END:%.*]] 11912 // CHECK11: omp.precond.then: 11913 // CHECK11-NEXT: store i32 0, i32* [[DOTOMP_COMB_LB]], align 4 11914 // CHECK11-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_2]], align 4 11915 // CHECK11-NEXT: store i32 [[TMP9]], i32* [[DOTOMP_COMB_UB]], align 4 11916 // CHECK11-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4 11917 // CHECK11-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4 11918 // CHECK11-NEXT: [[TMP10:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 4 11919 // CHECK11-NEXT: [[TMP11:%.*]] = load i32, i32* [[TMP10]], align 4 11920 // CHECK11-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1]], i32 [[TMP11]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_COMB_LB]], i32* [[DOTOMP_COMB_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1) 11921 // CHECK11-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 11922 // CHECK11-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_2]], align 4 11923 // CHECK11-NEXT: [[CMP5:%.*]] = icmp sgt i32 [[TMP12]], [[TMP13]] 11924 // CHECK11-NEXT: br i1 [[CMP5]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] 11925 // CHECK11: cond.true: 11926 // CHECK11-NEXT: [[TMP14:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_2]], align 4 11927 // CHECK11-NEXT: br label [[COND_END:%.*]] 11928 // CHECK11: cond.false: 11929 // CHECK11-NEXT: [[TMP15:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 11930 // CHECK11-NEXT: br label [[COND_END]] 11931 // CHECK11: cond.end: 11932 // CHECK11-NEXT: [[COND:%.*]] = phi i32 [ [[TMP14]], [[COND_TRUE]] ], [ [[TMP15]], [[COND_FALSE]] ] 11933 // CHECK11-NEXT: store i32 [[COND]], i32* [[DOTOMP_COMB_UB]], align 4 11934 // CHECK11-NEXT: [[TMP16:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4 11935 // CHECK11-NEXT: store i32 [[TMP16]], i32* [[DOTOMP_IV]], align 4 11936 // CHECK11-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] 11937 // CHECK11: omp.inner.for.cond: 11938 // CHECK11-NEXT: [[TMP17:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !57 11939 // CHECK11-NEXT: [[TMP18:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4, !llvm.access.group !57 11940 // CHECK11-NEXT: [[CMP6:%.*]] = icmp sle i32 [[TMP17]], [[TMP18]] 11941 // CHECK11-NEXT: br i1 [[CMP6]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] 11942 // CHECK11: omp.inner.for.body: 11943 // CHECK11-NEXT: [[TMP19:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4, !llvm.access.group !57 11944 // CHECK11-NEXT: [[TMP20:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4, !llvm.access.group !57 11945 // CHECK11-NEXT: [[TMP21:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4, !llvm.access.group !57 11946 // CHECK11-NEXT: store i32 [[TMP21]], i32* [[DOTCAPTURE_EXPR__CASTED]], align 4, !llvm.access.group !57 11947 // CHECK11-NEXT: [[TMP22:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR__CASTED]], align 4, !llvm.access.group !57 11948 // CHECK11-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 7, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32, i32, i32*, double**, double**, double**, i32)* @.omp_outlined..23 to void (i32*, i32*, ...)*), i32 [[TMP19]], i32 [[TMP20]], i32* [[TMP1]], double** [[TMP2]], double** [[TMP3]], double** [[TMP4]], i32 [[TMP22]]), !llvm.access.group !57 11949 // CHECK11-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] 11950 // CHECK11: omp.inner.for.inc: 11951 // CHECK11-NEXT: [[TMP23:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !57 11952 // CHECK11-NEXT: [[TMP24:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4, !llvm.access.group !57 11953 // CHECK11-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP23]], [[TMP24]] 11954 // CHECK11-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4, !llvm.access.group !57 11955 // CHECK11-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP58:![0-9]+]] 11956 // CHECK11: omp.inner.for.end: 11957 // CHECK11-NEXT: br label [[OMP_LOOP_EXIT:%.*]] 11958 // CHECK11: omp.loop.exit: 11959 // CHECK11-NEXT: [[TMP25:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 4 11960 // CHECK11-NEXT: [[TMP26:%.*]] = load i32, i32* [[TMP25]], align 4 11961 // CHECK11-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP26]]) 11962 // CHECK11-NEXT: [[TMP27:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4 11963 // CHECK11-NEXT: [[TMP28:%.*]] = icmp ne i32 [[TMP27]], 0 11964 // CHECK11-NEXT: br i1 [[TMP28]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]] 11965 // CHECK11: .omp.final.then: 11966 // CHECK11-NEXT: [[TMP29:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 11967 // CHECK11-NEXT: [[SUB7:%.*]] = sub nsw i32 [[TMP29]], 0 11968 // CHECK11-NEXT: [[DIV8:%.*]] = sdiv i32 [[SUB7]], 1 11969 // CHECK11-NEXT: [[MUL:%.*]] = mul nsw i32 [[DIV8]], 1 11970 // CHECK11-NEXT: [[ADD9:%.*]] = add nsw i32 0, [[MUL]] 11971 // CHECK11-NEXT: store i32 [[ADD9]], i32* [[I4]], align 4 11972 // CHECK11-NEXT: br label [[DOTOMP_FINAL_DONE]] 11973 // CHECK11: .omp.final.done: 11974 // CHECK11-NEXT: br label [[OMP_PRECOND_END]] 11975 // CHECK11: omp.precond.end: 11976 // CHECK11-NEXT: ret void 11977 // 11978 // 11979 // CHECK11-LABEL: define {{[^@]+}}@.omp_outlined..23 11980 // CHECK11-SAME: (i32* noalias noundef [[DOTGLOBAL_TID_:%.*]], i32* noalias noundef [[DOTBOUND_TID_:%.*]], i32 noundef [[DOTPREVIOUS_LB_:%.*]], i32 noundef [[DOTPREVIOUS_UB_:%.*]], i32* noundef nonnull align 4 dereferenceable(4) [[N:%.*]], double** noundef nonnull align 4 dereferenceable(4) [[A:%.*]], double** noundef nonnull align 4 dereferenceable(4) [[B:%.*]], double** noundef nonnull align 4 dereferenceable(4) [[C:%.*]], i32 noundef [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR1]] { 11981 // CHECK11-NEXT: entry: 11982 // CHECK11-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4 11983 // CHECK11-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4 11984 // CHECK11-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i32, align 4 11985 // CHECK11-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i32, align 4 11986 // CHECK11-NEXT: [[N_ADDR:%.*]] = alloca i32*, align 4 11987 // CHECK11-NEXT: [[A_ADDR:%.*]] = alloca double**, align 4 11988 // CHECK11-NEXT: [[B_ADDR:%.*]] = alloca double**, align 4 11989 // CHECK11-NEXT: [[C_ADDR:%.*]] = alloca double**, align 4 11990 // CHECK11-NEXT: [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i32, align 4 11991 // CHECK11-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 11992 // CHECK11-NEXT: [[TMP:%.*]] = alloca i32, align 4 11993 // CHECK11-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 11994 // CHECK11-NEXT: [[DOTCAPTURE_EXPR_2:%.*]] = alloca i32, align 4 11995 // CHECK11-NEXT: [[I:%.*]] = alloca i32, align 4 11996 // CHECK11-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 11997 // CHECK11-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4 11998 // CHECK11-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 11999 // CHECK11-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 12000 // CHECK11-NEXT: [[I4:%.*]] = alloca i32, align 4 12001 // CHECK11-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 4 12002 // CHECK11-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 4 12003 // CHECK11-NEXT: store i32 [[DOTPREVIOUS_LB_]], i32* [[DOTPREVIOUS_LB__ADDR]], align 4 12004 // CHECK11-NEXT: store i32 [[DOTPREVIOUS_UB_]], i32* [[DOTPREVIOUS_UB__ADDR]], align 4 12005 // CHECK11-NEXT: store i32* [[N]], i32** [[N_ADDR]], align 4 12006 // CHECK11-NEXT: store double** [[A]], double*** [[A_ADDR]], align 4 12007 // CHECK11-NEXT: store double** [[B]], double*** [[B_ADDR]], align 4 12008 // CHECK11-NEXT: store double** [[C]], double*** [[C_ADDR]], align 4 12009 // CHECK11-NEXT: store i32 [[DOTCAPTURE_EXPR_]], i32* [[DOTCAPTURE_EXPR__ADDR]], align 4 12010 // CHECK11-NEXT: [[TMP0:%.*]] = load i32*, i32** [[N_ADDR]], align 4 12011 // CHECK11-NEXT: [[TMP1:%.*]] = load double**, double*** [[A_ADDR]], align 4 12012 // CHECK11-NEXT: [[TMP2:%.*]] = load double**, double*** [[B_ADDR]], align 4 12013 // CHECK11-NEXT: [[TMP3:%.*]] = load double**, double*** [[C_ADDR]], align 4 12014 // CHECK11-NEXT: [[TMP4:%.*]] = load i32, i32* [[TMP0]], align 4 12015 // CHECK11-NEXT: store i32 [[TMP4]], i32* [[DOTCAPTURE_EXPR_1]], align 4 12016 // CHECK11-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 12017 // CHECK11-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP5]], 0 12018 // CHECK11-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 12019 // CHECK11-NEXT: [[SUB3:%.*]] = sub nsw i32 [[DIV]], 1 12020 // CHECK11-NEXT: store i32 [[SUB3]], i32* [[DOTCAPTURE_EXPR_2]], align 4 12021 // CHECK11-NEXT: store i32 0, i32* [[I]], align 4 12022 // CHECK11-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 12023 // CHECK11-NEXT: [[CMP:%.*]] = icmp slt i32 0, [[TMP6]] 12024 // CHECK11-NEXT: br i1 [[CMP]], label [[OMP_PRECOND_THEN:%.*]], label [[OMP_PRECOND_END:%.*]] 12025 // CHECK11: omp.precond.then: 12026 // CHECK11-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4 12027 // CHECK11-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_2]], align 4 12028 // CHECK11-NEXT: store i32 [[TMP7]], i32* [[DOTOMP_UB]], align 4 12029 // CHECK11-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTPREVIOUS_LB__ADDR]], align 4 12030 // CHECK11-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTPREVIOUS_UB__ADDR]], align 4 12031 // CHECK11-NEXT: store i32 [[TMP8]], i32* [[DOTOMP_LB]], align 4 12032 // CHECK11-NEXT: store i32 [[TMP9]], i32* [[DOTOMP_UB]], align 4 12033 // CHECK11-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4 12034 // CHECK11-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4 12035 // CHECK11-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR__ADDR]], align 4 12036 // CHECK11-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4 12037 // CHECK11-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 12038 // CHECK11-NEXT: [[TMP13:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 4 12039 // CHECK11-NEXT: [[TMP14:%.*]] = load i32, i32* [[TMP13]], align 4 12040 // CHECK11-NEXT: call void @__kmpc_dispatch_init_4(%struct.ident_t* @[[GLOB3]], i32 [[TMP14]], i32 35, i32 [[TMP11]], i32 [[TMP12]], i32 1, i32 [[TMP10]]) 12041 // CHECK11-NEXT: br label [[OMP_DISPATCH_COND:%.*]] 12042 // CHECK11: omp.dispatch.cond: 12043 // CHECK11-NEXT: [[TMP15:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 4 12044 // CHECK11-NEXT: [[TMP16:%.*]] = load i32, i32* [[TMP15]], align 4 12045 // CHECK11-NEXT: [[TMP17:%.*]] = call i32 @__kmpc_dispatch_next_4(%struct.ident_t* @[[GLOB3]], i32 [[TMP16]], i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]]) 12046 // CHECK11-NEXT: [[TOBOOL:%.*]] = icmp ne i32 [[TMP17]], 0 12047 // CHECK11-NEXT: br i1 [[TOBOOL]], label [[OMP_DISPATCH_BODY:%.*]], label [[OMP_DISPATCH_END:%.*]] 12048 // CHECK11: omp.dispatch.body: 12049 // CHECK11-NEXT: [[TMP18:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4 12050 // CHECK11-NEXT: store i32 [[TMP18]], i32* [[DOTOMP_IV]], align 4 12051 // CHECK11-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] 12052 // CHECK11: omp.inner.for.cond: 12053 // CHECK11-NEXT: [[TMP19:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !60 12054 // CHECK11-NEXT: [[TMP20:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4, !llvm.access.group !60 12055 // CHECK11-NEXT: [[CMP5:%.*]] = icmp sle i32 [[TMP19]], [[TMP20]] 12056 // CHECK11-NEXT: br i1 [[CMP5]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] 12057 // CHECK11: omp.inner.for.body: 12058 // CHECK11-NEXT: [[TMP21:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !60 12059 // CHECK11-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP21]], 1 12060 // CHECK11-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] 12061 // CHECK11-NEXT: store i32 [[ADD]], i32* [[I4]], align 4, !llvm.access.group !60 12062 // CHECK11-NEXT: [[TMP22:%.*]] = load double*, double** [[TMP2]], align 4, !llvm.access.group !60 12063 // CHECK11-NEXT: [[TMP23:%.*]] = load i32, i32* [[I4]], align 4, !llvm.access.group !60 12064 // CHECK11-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds double, double* [[TMP22]], i32 [[TMP23]] 12065 // CHECK11-NEXT: [[TMP24:%.*]] = load double, double* [[ARRAYIDX]], align 4, !llvm.access.group !60 12066 // CHECK11-NEXT: [[TMP25:%.*]] = load double*, double** [[TMP3]], align 4, !llvm.access.group !60 12067 // CHECK11-NEXT: [[TMP26:%.*]] = load i32, i32* [[I4]], align 4, !llvm.access.group !60 12068 // CHECK11-NEXT: [[ARRAYIDX6:%.*]] = getelementptr inbounds double, double* [[TMP25]], i32 [[TMP26]] 12069 // CHECK11-NEXT: [[TMP27:%.*]] = load double, double* [[ARRAYIDX6]], align 4, !llvm.access.group !60 12070 // CHECK11-NEXT: [[ADD7:%.*]] = fadd double [[TMP24]], [[TMP27]] 12071 // CHECK11-NEXT: [[TMP28:%.*]] = load double*, double** [[TMP1]], align 4, !llvm.access.group !60 12072 // CHECK11-NEXT: [[TMP29:%.*]] = load i32, i32* [[I4]], align 4, !llvm.access.group !60 12073 // CHECK11-NEXT: [[ARRAYIDX8:%.*]] = getelementptr inbounds double, double* [[TMP28]], i32 [[TMP29]] 12074 // CHECK11-NEXT: store double [[ADD7]], double* [[ARRAYIDX8]], align 4, !llvm.access.group !60 12075 // CHECK11-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] 12076 // CHECK11: omp.body.continue: 12077 // CHECK11-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] 12078 // CHECK11: omp.inner.for.inc: 12079 // CHECK11-NEXT: [[TMP30:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !60 12080 // CHECK11-NEXT: [[ADD9:%.*]] = add nsw i32 [[TMP30]], 1 12081 // CHECK11-NEXT: store i32 [[ADD9]], i32* [[DOTOMP_IV]], align 4, !llvm.access.group !60 12082 // CHECK11-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP61:![0-9]+]] 12083 // CHECK11: omp.inner.for.end: 12084 // CHECK11-NEXT: br label [[OMP_DISPATCH_INC:%.*]] 12085 // CHECK11: omp.dispatch.inc: 12086 // CHECK11-NEXT: br label [[OMP_DISPATCH_COND]] 12087 // CHECK11: omp.dispatch.end: 12088 // CHECK11-NEXT: [[TMP31:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4 12089 // CHECK11-NEXT: [[TMP32:%.*]] = icmp ne i32 [[TMP31]], 0 12090 // CHECK11-NEXT: br i1 [[TMP32]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]] 12091 // CHECK11: .omp.final.then: 12092 // CHECK11-NEXT: [[TMP33:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 12093 // CHECK11-NEXT: [[SUB10:%.*]] = sub nsw i32 [[TMP33]], 0 12094 // CHECK11-NEXT: [[DIV11:%.*]] = sdiv i32 [[SUB10]], 1 12095 // CHECK11-NEXT: [[MUL12:%.*]] = mul nsw i32 [[DIV11]], 1 12096 // CHECK11-NEXT: [[ADD13:%.*]] = add nsw i32 0, [[MUL12]] 12097 // CHECK11-NEXT: store i32 [[ADD13]], i32* [[I4]], align 4 12098 // CHECK11-NEXT: br label [[DOTOMP_FINAL_DONE]] 12099 // CHECK11: .omp.final.done: 12100 // CHECK11-NEXT: br label [[OMP_PRECOND_END]] 12101 // CHECK11: omp.precond.end: 12102 // CHECK11-NEXT: ret void 12103 // 12104 // 12105 // CHECK11-LABEL: define {{[^@]+}}@_Z5tmainIiET_v 12106 // CHECK11-SAME: () #[[ATTR3:[0-9]+]] comdat { 12107 // CHECK11-NEXT: entry: 12108 // CHECK11-NEXT: [[A:%.*]] = alloca i32*, align 4 12109 // CHECK11-NEXT: [[B:%.*]] = alloca i32*, align 4 12110 // CHECK11-NEXT: [[C:%.*]] = alloca i32*, align 4 12111 // CHECK11-NEXT: [[N:%.*]] = alloca i32, align 4 12112 // CHECK11-NEXT: [[CH:%.*]] = alloca i32, align 4 12113 // CHECK11-NEXT: [[N_CASTED:%.*]] = alloca i32, align 4 12114 // CHECK11-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [4 x i8*], align 4 12115 // CHECK11-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [4 x i8*], align 4 12116 // CHECK11-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [4 x i8*], align 4 12117 // CHECK11-NEXT: [[TMP:%.*]] = alloca i32, align 4 12118 // CHECK11-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 12119 // CHECK11-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 12120 // CHECK11-NEXT: [[N_CASTED3:%.*]] = alloca i32, align 4 12121 // CHECK11-NEXT: [[DOTOFFLOAD_BASEPTRS4:%.*]] = alloca [4 x i8*], align 4 12122 // CHECK11-NEXT: [[DOTOFFLOAD_PTRS5:%.*]] = alloca [4 x i8*], align 4 12123 // CHECK11-NEXT: [[DOTOFFLOAD_MAPPERS6:%.*]] = alloca [4 x i8*], align 4 12124 // CHECK11-NEXT: [[_TMP7:%.*]] = alloca i32, align 4 12125 // CHECK11-NEXT: [[DOTCAPTURE_EXPR_8:%.*]] = alloca i32, align 4 12126 // CHECK11-NEXT: [[DOTCAPTURE_EXPR_9:%.*]] = alloca i32, align 4 12127 // CHECK11-NEXT: [[CH_CASTED:%.*]] = alloca i32, align 4 12128 // CHECK11-NEXT: [[N_CASTED17:%.*]] = alloca i32, align 4 12129 // CHECK11-NEXT: [[DOTOFFLOAD_BASEPTRS18:%.*]] = alloca [5 x i8*], align 4 12130 // CHECK11-NEXT: [[DOTOFFLOAD_PTRS19:%.*]] = alloca [5 x i8*], align 4 12131 // CHECK11-NEXT: [[DOTOFFLOAD_MAPPERS20:%.*]] = alloca [5 x i8*], align 4 12132 // CHECK11-NEXT: [[_TMP21:%.*]] = alloca i32, align 4 12133 // CHECK11-NEXT: [[DOTCAPTURE_EXPR_22:%.*]] = alloca i32, align 4 12134 // CHECK11-NEXT: [[DOTCAPTURE_EXPR_23:%.*]] = alloca i32, align 4 12135 // CHECK11-NEXT: [[N_CASTED31:%.*]] = alloca i32, align 4 12136 // CHECK11-NEXT: [[DOTOFFLOAD_BASEPTRS32:%.*]] = alloca [4 x i8*], align 4 12137 // CHECK11-NEXT: [[DOTOFFLOAD_PTRS33:%.*]] = alloca [4 x i8*], align 4 12138 // CHECK11-NEXT: [[DOTOFFLOAD_MAPPERS34:%.*]] = alloca [4 x i8*], align 4 12139 // CHECK11-NEXT: [[_TMP35:%.*]] = alloca i32, align 4 12140 // CHECK11-NEXT: [[DOTCAPTURE_EXPR_36:%.*]] = alloca i32, align 4 12141 // CHECK11-NEXT: [[DOTCAPTURE_EXPR_37:%.*]] = alloca i32, align 4 12142 // CHECK11-NEXT: [[CH_CASTED45:%.*]] = alloca i32, align 4 12143 // CHECK11-NEXT: [[N_CASTED46:%.*]] = alloca i32, align 4 12144 // CHECK11-NEXT: [[DOTOFFLOAD_BASEPTRS47:%.*]] = alloca [5 x i8*], align 4 12145 // CHECK11-NEXT: [[DOTOFFLOAD_PTRS48:%.*]] = alloca [5 x i8*], align 4 12146 // CHECK11-NEXT: [[DOTOFFLOAD_MAPPERS49:%.*]] = alloca [5 x i8*], align 4 12147 // CHECK11-NEXT: [[_TMP50:%.*]] = alloca i32, align 4 12148 // CHECK11-NEXT: [[DOTCAPTURE_EXPR_51:%.*]] = alloca i32, align 4 12149 // CHECK11-NEXT: [[DOTCAPTURE_EXPR_52:%.*]] = alloca i32, align 4 12150 // CHECK11-NEXT: [[N_CASTED60:%.*]] = alloca i32, align 4 12151 // CHECK11-NEXT: [[DOTOFFLOAD_BASEPTRS61:%.*]] = alloca [4 x i8*], align 4 12152 // CHECK11-NEXT: [[DOTOFFLOAD_PTRS62:%.*]] = alloca [4 x i8*], align 4 12153 // CHECK11-NEXT: [[DOTOFFLOAD_MAPPERS63:%.*]] = alloca [4 x i8*], align 4 12154 // CHECK11-NEXT: [[_TMP64:%.*]] = alloca i32, align 4 12155 // CHECK11-NEXT: [[DOTCAPTURE_EXPR_65:%.*]] = alloca i32, align 4 12156 // CHECK11-NEXT: [[DOTCAPTURE_EXPR_66:%.*]] = alloca i32, align 4 12157 // CHECK11-NEXT: [[CH_CASTED74:%.*]] = alloca i32, align 4 12158 // CHECK11-NEXT: [[N_CASTED75:%.*]] = alloca i32, align 4 12159 // CHECK11-NEXT: [[DOTOFFLOAD_BASEPTRS76:%.*]] = alloca [5 x i8*], align 4 12160 // CHECK11-NEXT: [[DOTOFFLOAD_PTRS77:%.*]] = alloca [5 x i8*], align 4 12161 // CHECK11-NEXT: [[DOTOFFLOAD_MAPPERS78:%.*]] = alloca [5 x i8*], align 4 12162 // CHECK11-NEXT: [[_TMP79:%.*]] = alloca i32, align 4 12163 // CHECK11-NEXT: [[DOTCAPTURE_EXPR_80:%.*]] = alloca i32, align 4 12164 // CHECK11-NEXT: [[DOTCAPTURE_EXPR_81:%.*]] = alloca i32, align 4 12165 // CHECK11-NEXT: store i32 10000, i32* [[N]], align 4 12166 // CHECK11-NEXT: store i32 100, i32* [[CH]], align 4 12167 // CHECK11-NEXT: [[TMP0:%.*]] = load i32, i32* [[N]], align 4 12168 // CHECK11-NEXT: store i32 [[TMP0]], i32* [[N_CASTED]], align 4 12169 // CHECK11-NEXT: [[TMP1:%.*]] = load i32, i32* [[N_CASTED]], align 4 12170 // CHECK11-NEXT: [[TMP2:%.*]] = load i32*, i32** [[A]], align 4 12171 // CHECK11-NEXT: [[TMP3:%.*]] = load i32*, i32** [[B]], align 4 12172 // CHECK11-NEXT: [[TMP4:%.*]] = load i32*, i32** [[C]], align 4 12173 // CHECK11-NEXT: [[TMP5:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 12174 // CHECK11-NEXT: [[TMP6:%.*]] = bitcast i8** [[TMP5]] to i32* 12175 // CHECK11-NEXT: store i32 [[TMP1]], i32* [[TMP6]], align 4 12176 // CHECK11-NEXT: [[TMP7:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 12177 // CHECK11-NEXT: [[TMP8:%.*]] = bitcast i8** [[TMP7]] to i32* 12178 // CHECK11-NEXT: store i32 [[TMP1]], i32* [[TMP8]], align 4 12179 // CHECK11-NEXT: [[TMP9:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0 12180 // CHECK11-NEXT: store i8* null, i8** [[TMP9]], align 4 12181 // CHECK11-NEXT: [[TMP10:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1 12182 // CHECK11-NEXT: [[TMP11:%.*]] = bitcast i8** [[TMP10]] to i32** 12183 // CHECK11-NEXT: store i32* [[TMP2]], i32** [[TMP11]], align 4 12184 // CHECK11-NEXT: [[TMP12:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1 12185 // CHECK11-NEXT: [[TMP13:%.*]] = bitcast i8** [[TMP12]] to i32** 12186 // CHECK11-NEXT: store i32* [[TMP2]], i32** [[TMP13]], align 4 12187 // CHECK11-NEXT: [[TMP14:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 1 12188 // CHECK11-NEXT: store i8* null, i8** [[TMP14]], align 4 12189 // CHECK11-NEXT: [[TMP15:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2 12190 // CHECK11-NEXT: [[TMP16:%.*]] = bitcast i8** [[TMP15]] to i32** 12191 // CHECK11-NEXT: store i32* [[TMP3]], i32** [[TMP16]], align 4 12192 // CHECK11-NEXT: [[TMP17:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2 12193 // CHECK11-NEXT: [[TMP18:%.*]] = bitcast i8** [[TMP17]] to i32** 12194 // CHECK11-NEXT: store i32* [[TMP3]], i32** [[TMP18]], align 4 12195 // CHECK11-NEXT: [[TMP19:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2 12196 // CHECK11-NEXT: store i8* null, i8** [[TMP19]], align 4 12197 // CHECK11-NEXT: [[TMP20:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 3 12198 // CHECK11-NEXT: [[TMP21:%.*]] = bitcast i8** [[TMP20]] to i32** 12199 // CHECK11-NEXT: store i32* [[TMP4]], i32** [[TMP21]], align 4 12200 // CHECK11-NEXT: [[TMP22:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 3 12201 // CHECK11-NEXT: [[TMP23:%.*]] = bitcast i8** [[TMP22]] to i32** 12202 // CHECK11-NEXT: store i32* [[TMP4]], i32** [[TMP23]], align 4 12203 // CHECK11-NEXT: [[TMP24:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 3 12204 // CHECK11-NEXT: store i8* null, i8** [[TMP24]], align 4 12205 // CHECK11-NEXT: [[TMP25:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 12206 // CHECK11-NEXT: [[TMP26:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 12207 // CHECK11-NEXT: [[TMP27:%.*]] = load i32, i32* [[N]], align 4 12208 // CHECK11-NEXT: store i32 [[TMP27]], i32* [[DOTCAPTURE_EXPR_]], align 4 12209 // CHECK11-NEXT: [[TMP28:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 12210 // CHECK11-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP28]], 0 12211 // CHECK11-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 12212 // CHECK11-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1 12213 // CHECK11-NEXT: store i32 [[SUB2]], i32* [[DOTCAPTURE_EXPR_1]], align 4 12214 // CHECK11-NEXT: [[TMP29:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 12215 // CHECK11-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP29]], 1 12216 // CHECK11-NEXT: [[TMP30:%.*]] = zext i32 [[ADD]] to i64 12217 // CHECK11-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8 12218 // CHECK11-NEXT: [[TMP31:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS]], i32 0, i32 0 12219 // CHECK11-NEXT: store i32 1, i32* [[TMP31]], align 4 12220 // CHECK11-NEXT: [[TMP32:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS]], i32 0, i32 1 12221 // CHECK11-NEXT: store i32 4, i32* [[TMP32]], align 4 12222 // CHECK11-NEXT: [[TMP33:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS]], i32 0, i32 2 12223 // CHECK11-NEXT: store i8** [[TMP25]], i8*** [[TMP33]], align 4 12224 // CHECK11-NEXT: [[TMP34:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS]], i32 0, i32 3 12225 // CHECK11-NEXT: store i8** [[TMP26]], i8*** [[TMP34]], align 4 12226 // CHECK11-NEXT: [[TMP35:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS]], i32 0, i32 4 12227 // CHECK11-NEXT: store i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_sizes.28, i32 0, i32 0), i64** [[TMP35]], align 4 12228 // CHECK11-NEXT: [[TMP36:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS]], i32 0, i32 5 12229 // CHECK11-NEXT: store i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_maptypes.29, i32 0, i32 0), i64** [[TMP36]], align 4 12230 // CHECK11-NEXT: [[TMP37:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS]], i32 0, i32 6 12231 // CHECK11-NEXT: store i8** null, i8*** [[TMP37]], align 4 12232 // CHECK11-NEXT: [[TMP38:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS]], i32 0, i32 7 12233 // CHECK11-NEXT: store i8** null, i8*** [[TMP38]], align 4 12234 // CHECK11-NEXT: [[TMP39:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS]], i32 0, i32 8 12235 // CHECK11-NEXT: store i64 [[TMP30]], i64* [[TMP39]], align 8 12236 // CHECK11-NEXT: [[TMP40:%.*]] = call i32 @__tgt_target_kernel(%struct.ident_t* @[[GLOB3]], i64 -1, i32 0, i32 1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiET_v_l42.region_id, %struct.__tgt_kernel_arguments* [[KERNEL_ARGS]]) 12237 // CHECK11-NEXT: [[TMP41:%.*]] = icmp ne i32 [[TMP40]], 0 12238 // CHECK11-NEXT: br i1 [[TMP41]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] 12239 // CHECK11: omp_offload.failed: 12240 // CHECK11-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiET_v_l42(i32 [[TMP1]], i32* [[TMP2]], i32* [[TMP3]], i32* [[TMP4]]) #[[ATTR2]] 12241 // CHECK11-NEXT: br label [[OMP_OFFLOAD_CONT]] 12242 // CHECK11: omp_offload.cont: 12243 // CHECK11-NEXT: [[TMP42:%.*]] = load i32, i32* [[N]], align 4 12244 // CHECK11-NEXT: store i32 [[TMP42]], i32* [[N_CASTED3]], align 4 12245 // CHECK11-NEXT: [[TMP43:%.*]] = load i32, i32* [[N_CASTED3]], align 4 12246 // CHECK11-NEXT: [[TMP44:%.*]] = load i32*, i32** [[A]], align 4 12247 // CHECK11-NEXT: [[TMP45:%.*]] = load i32*, i32** [[B]], align 4 12248 // CHECK11-NEXT: [[TMP46:%.*]] = load i32*, i32** [[C]], align 4 12249 // CHECK11-NEXT: [[TMP47:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS4]], i32 0, i32 0 12250 // CHECK11-NEXT: [[TMP48:%.*]] = bitcast i8** [[TMP47]] to i32* 12251 // CHECK11-NEXT: store i32 [[TMP43]], i32* [[TMP48]], align 4 12252 // CHECK11-NEXT: [[TMP49:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS5]], i32 0, i32 0 12253 // CHECK11-NEXT: [[TMP50:%.*]] = bitcast i8** [[TMP49]] to i32* 12254 // CHECK11-NEXT: store i32 [[TMP43]], i32* [[TMP50]], align 4 12255 // CHECK11-NEXT: [[TMP51:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS6]], i32 0, i32 0 12256 // CHECK11-NEXT: store i8* null, i8** [[TMP51]], align 4 12257 // CHECK11-NEXT: [[TMP52:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS4]], i32 0, i32 1 12258 // CHECK11-NEXT: [[TMP53:%.*]] = bitcast i8** [[TMP52]] to i32** 12259 // CHECK11-NEXT: store i32* [[TMP44]], i32** [[TMP53]], align 4 12260 // CHECK11-NEXT: [[TMP54:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS5]], i32 0, i32 1 12261 // CHECK11-NEXT: [[TMP55:%.*]] = bitcast i8** [[TMP54]] to i32** 12262 // CHECK11-NEXT: store i32* [[TMP44]], i32** [[TMP55]], align 4 12263 // CHECK11-NEXT: [[TMP56:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS6]], i32 0, i32 1 12264 // CHECK11-NEXT: store i8* null, i8** [[TMP56]], align 4 12265 // CHECK11-NEXT: [[TMP57:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS4]], i32 0, i32 2 12266 // CHECK11-NEXT: [[TMP58:%.*]] = bitcast i8** [[TMP57]] to i32** 12267 // CHECK11-NEXT: store i32* [[TMP45]], i32** [[TMP58]], align 4 12268 // CHECK11-NEXT: [[TMP59:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS5]], i32 0, i32 2 12269 // CHECK11-NEXT: [[TMP60:%.*]] = bitcast i8** [[TMP59]] to i32** 12270 // CHECK11-NEXT: store i32* [[TMP45]], i32** [[TMP60]], align 4 12271 // CHECK11-NEXT: [[TMP61:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS6]], i32 0, i32 2 12272 // CHECK11-NEXT: store i8* null, i8** [[TMP61]], align 4 12273 // CHECK11-NEXT: [[TMP62:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS4]], i32 0, i32 3 12274 // CHECK11-NEXT: [[TMP63:%.*]] = bitcast i8** [[TMP62]] to i32** 12275 // CHECK11-NEXT: store i32* [[TMP46]], i32** [[TMP63]], align 4 12276 // CHECK11-NEXT: [[TMP64:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS5]], i32 0, i32 3 12277 // CHECK11-NEXT: [[TMP65:%.*]] = bitcast i8** [[TMP64]] to i32** 12278 // CHECK11-NEXT: store i32* [[TMP46]], i32** [[TMP65]], align 4 12279 // CHECK11-NEXT: [[TMP66:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS6]], i32 0, i32 3 12280 // CHECK11-NEXT: store i8* null, i8** [[TMP66]], align 4 12281 // CHECK11-NEXT: [[TMP67:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS4]], i32 0, i32 0 12282 // CHECK11-NEXT: [[TMP68:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS5]], i32 0, i32 0 12283 // CHECK11-NEXT: [[TMP69:%.*]] = load i32, i32* [[N]], align 4 12284 // CHECK11-NEXT: store i32 [[TMP69]], i32* [[DOTCAPTURE_EXPR_8]], align 4 12285 // CHECK11-NEXT: [[TMP70:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_8]], align 4 12286 // CHECK11-NEXT: [[SUB10:%.*]] = sub nsw i32 [[TMP70]], 0 12287 // CHECK11-NEXT: [[DIV11:%.*]] = sdiv i32 [[SUB10]], 1 12288 // CHECK11-NEXT: [[SUB12:%.*]] = sub nsw i32 [[DIV11]], 1 12289 // CHECK11-NEXT: store i32 [[SUB12]], i32* [[DOTCAPTURE_EXPR_9]], align 4 12290 // CHECK11-NEXT: [[TMP71:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_9]], align 4 12291 // CHECK11-NEXT: [[ADD13:%.*]] = add nsw i32 [[TMP71]], 1 12292 // CHECK11-NEXT: [[TMP72:%.*]] = zext i32 [[ADD13]] to i64 12293 // CHECK11-NEXT: [[KERNEL_ARGS14:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS]], align 8 12294 // CHECK11-NEXT: [[TMP73:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS14]], i32 0, i32 0 12295 // CHECK11-NEXT: store i32 1, i32* [[TMP73]], align 4 12296 // CHECK11-NEXT: [[TMP74:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS14]], i32 0, i32 1 12297 // CHECK11-NEXT: store i32 4, i32* [[TMP74]], align 4 12298 // CHECK11-NEXT: [[TMP75:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS14]], i32 0, i32 2 12299 // CHECK11-NEXT: store i8** [[TMP67]], i8*** [[TMP75]], align 4 12300 // CHECK11-NEXT: [[TMP76:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS14]], i32 0, i32 3 12301 // CHECK11-NEXT: store i8** [[TMP68]], i8*** [[TMP76]], align 4 12302 // CHECK11-NEXT: [[TMP77:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS14]], i32 0, i32 4 12303 // CHECK11-NEXT: store i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_sizes.32, i32 0, i32 0), i64** [[TMP77]], align 4 12304 // CHECK11-NEXT: [[TMP78:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS14]], i32 0, i32 5 12305 // CHECK11-NEXT: store i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_maptypes.33, i32 0, i32 0), i64** [[TMP78]], align 4 12306 // CHECK11-NEXT: [[TMP79:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS14]], i32 0, i32 6 12307 // CHECK11-NEXT: store i8** null, i8*** [[TMP79]], align 4 12308 // CHECK11-NEXT: [[TMP80:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS14]], i32 0, i32 7 12309 // CHECK11-NEXT: store i8** null, i8*** [[TMP80]], align 4 12310 // CHECK11-NEXT: [[TMP81:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS14]], i32 0, i32 8 12311 // CHECK11-NEXT: store i64 [[TMP72]], i64* [[TMP81]], align 8 12312 // CHECK11-NEXT: [[TMP82:%.*]] = call i32 @__tgt_target_kernel(%struct.ident_t* @[[GLOB3]], i64 -1, i32 0, i32 1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiET_v_l50.region_id, %struct.__tgt_kernel_arguments* [[KERNEL_ARGS14]]) 12313 // CHECK11-NEXT: [[TMP83:%.*]] = icmp ne i32 [[TMP82]], 0 12314 // CHECK11-NEXT: br i1 [[TMP83]], label [[OMP_OFFLOAD_FAILED15:%.*]], label [[OMP_OFFLOAD_CONT16:%.*]] 12315 // CHECK11: omp_offload.failed15: 12316 // CHECK11-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiET_v_l50(i32 [[TMP43]], i32* [[TMP44]], i32* [[TMP45]], i32* [[TMP46]]) #[[ATTR2]] 12317 // CHECK11-NEXT: br label [[OMP_OFFLOAD_CONT16]] 12318 // CHECK11: omp_offload.cont16: 12319 // CHECK11-NEXT: [[TMP84:%.*]] = load i32, i32* [[CH]], align 4 12320 // CHECK11-NEXT: store i32 [[TMP84]], i32* [[CH_CASTED]], align 4 12321 // CHECK11-NEXT: [[TMP85:%.*]] = load i32, i32* [[CH_CASTED]], align 4 12322 // CHECK11-NEXT: [[TMP86:%.*]] = load i32, i32* [[N]], align 4 12323 // CHECK11-NEXT: store i32 [[TMP86]], i32* [[N_CASTED17]], align 4 12324 // CHECK11-NEXT: [[TMP87:%.*]] = load i32, i32* [[N_CASTED17]], align 4 12325 // CHECK11-NEXT: [[TMP88:%.*]] = load i32*, i32** [[A]], align 4 12326 // CHECK11-NEXT: [[TMP89:%.*]] = load i32*, i32** [[B]], align 4 12327 // CHECK11-NEXT: [[TMP90:%.*]] = load i32*, i32** [[C]], align 4 12328 // CHECK11-NEXT: [[TMP91:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS18]], i32 0, i32 0 12329 // CHECK11-NEXT: [[TMP92:%.*]] = bitcast i8** [[TMP91]] to i32* 12330 // CHECK11-NEXT: store i32 [[TMP85]], i32* [[TMP92]], align 4 12331 // CHECK11-NEXT: [[TMP93:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS19]], i32 0, i32 0 12332 // CHECK11-NEXT: [[TMP94:%.*]] = bitcast i8** [[TMP93]] to i32* 12333 // CHECK11-NEXT: store i32 [[TMP85]], i32* [[TMP94]], align 4 12334 // CHECK11-NEXT: [[TMP95:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS20]], i32 0, i32 0 12335 // CHECK11-NEXT: store i8* null, i8** [[TMP95]], align 4 12336 // CHECK11-NEXT: [[TMP96:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS18]], i32 0, i32 1 12337 // CHECK11-NEXT: [[TMP97:%.*]] = bitcast i8** [[TMP96]] to i32* 12338 // CHECK11-NEXT: store i32 [[TMP87]], i32* [[TMP97]], align 4 12339 // CHECK11-NEXT: [[TMP98:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS19]], i32 0, i32 1 12340 // CHECK11-NEXT: [[TMP99:%.*]] = bitcast i8** [[TMP98]] to i32* 12341 // CHECK11-NEXT: store i32 [[TMP87]], i32* [[TMP99]], align 4 12342 // CHECK11-NEXT: [[TMP100:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS20]], i32 0, i32 1 12343 // CHECK11-NEXT: store i8* null, i8** [[TMP100]], align 4 12344 // CHECK11-NEXT: [[TMP101:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS18]], i32 0, i32 2 12345 // CHECK11-NEXT: [[TMP102:%.*]] = bitcast i8** [[TMP101]] to i32** 12346 // CHECK11-NEXT: store i32* [[TMP88]], i32** [[TMP102]], align 4 12347 // CHECK11-NEXT: [[TMP103:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS19]], i32 0, i32 2 12348 // CHECK11-NEXT: [[TMP104:%.*]] = bitcast i8** [[TMP103]] to i32** 12349 // CHECK11-NEXT: store i32* [[TMP88]], i32** [[TMP104]], align 4 12350 // CHECK11-NEXT: [[TMP105:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS20]], i32 0, i32 2 12351 // CHECK11-NEXT: store i8* null, i8** [[TMP105]], align 4 12352 // CHECK11-NEXT: [[TMP106:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS18]], i32 0, i32 3 12353 // CHECK11-NEXT: [[TMP107:%.*]] = bitcast i8** [[TMP106]] to i32** 12354 // CHECK11-NEXT: store i32* [[TMP89]], i32** [[TMP107]], align 4 12355 // CHECK11-NEXT: [[TMP108:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS19]], i32 0, i32 3 12356 // CHECK11-NEXT: [[TMP109:%.*]] = bitcast i8** [[TMP108]] to i32** 12357 // CHECK11-NEXT: store i32* [[TMP89]], i32** [[TMP109]], align 4 12358 // CHECK11-NEXT: [[TMP110:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS20]], i32 0, i32 3 12359 // CHECK11-NEXT: store i8* null, i8** [[TMP110]], align 4 12360 // CHECK11-NEXT: [[TMP111:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS18]], i32 0, i32 4 12361 // CHECK11-NEXT: [[TMP112:%.*]] = bitcast i8** [[TMP111]] to i32** 12362 // CHECK11-NEXT: store i32* [[TMP90]], i32** [[TMP112]], align 4 12363 // CHECK11-NEXT: [[TMP113:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS19]], i32 0, i32 4 12364 // CHECK11-NEXT: [[TMP114:%.*]] = bitcast i8** [[TMP113]] to i32** 12365 // CHECK11-NEXT: store i32* [[TMP90]], i32** [[TMP114]], align 4 12366 // CHECK11-NEXT: [[TMP115:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS20]], i32 0, i32 4 12367 // CHECK11-NEXT: store i8* null, i8** [[TMP115]], align 4 12368 // CHECK11-NEXT: [[TMP116:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS18]], i32 0, i32 0 12369 // CHECK11-NEXT: [[TMP117:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS19]], i32 0, i32 0 12370 // CHECK11-NEXT: [[TMP118:%.*]] = load i32, i32* [[N]], align 4 12371 // CHECK11-NEXT: store i32 [[TMP118]], i32* [[DOTCAPTURE_EXPR_22]], align 4 12372 // CHECK11-NEXT: [[TMP119:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_22]], align 4 12373 // CHECK11-NEXT: [[SUB24:%.*]] = sub nsw i32 [[TMP119]], 0 12374 // CHECK11-NEXT: [[DIV25:%.*]] = sdiv i32 [[SUB24]], 1 12375 // CHECK11-NEXT: [[SUB26:%.*]] = sub nsw i32 [[DIV25]], 1 12376 // CHECK11-NEXT: store i32 [[SUB26]], i32* [[DOTCAPTURE_EXPR_23]], align 4 12377 // CHECK11-NEXT: [[TMP120:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_23]], align 4 12378 // CHECK11-NEXT: [[ADD27:%.*]] = add nsw i32 [[TMP120]], 1 12379 // CHECK11-NEXT: [[TMP121:%.*]] = zext i32 [[ADD27]] to i64 12380 // CHECK11-NEXT: [[KERNEL_ARGS28:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS]], align 8 12381 // CHECK11-NEXT: [[TMP122:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS28]], i32 0, i32 0 12382 // CHECK11-NEXT: store i32 1, i32* [[TMP122]], align 4 12383 // CHECK11-NEXT: [[TMP123:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS28]], i32 0, i32 1 12384 // CHECK11-NEXT: store i32 5, i32* [[TMP123]], align 4 12385 // CHECK11-NEXT: [[TMP124:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS28]], i32 0, i32 2 12386 // CHECK11-NEXT: store i8** [[TMP116]], i8*** [[TMP124]], align 4 12387 // CHECK11-NEXT: [[TMP125:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS28]], i32 0, i32 3 12388 // CHECK11-NEXT: store i8** [[TMP117]], i8*** [[TMP125]], align 4 12389 // CHECK11-NEXT: [[TMP126:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS28]], i32 0, i32 4 12390 // CHECK11-NEXT: store i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_sizes.36, i32 0, i32 0), i64** [[TMP126]], align 4 12391 // CHECK11-NEXT: [[TMP127:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS28]], i32 0, i32 5 12392 // CHECK11-NEXT: store i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_maptypes.37, i32 0, i32 0), i64** [[TMP127]], align 4 12393 // CHECK11-NEXT: [[TMP128:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS28]], i32 0, i32 6 12394 // CHECK11-NEXT: store i8** null, i8*** [[TMP128]], align 4 12395 // CHECK11-NEXT: [[TMP129:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS28]], i32 0, i32 7 12396 // CHECK11-NEXT: store i8** null, i8*** [[TMP129]], align 4 12397 // CHECK11-NEXT: [[TMP130:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS28]], i32 0, i32 8 12398 // CHECK11-NEXT: store i64 [[TMP121]], i64* [[TMP130]], align 8 12399 // CHECK11-NEXT: [[TMP131:%.*]] = call i32 @__tgt_target_kernel(%struct.ident_t* @[[GLOB3]], i64 -1, i32 0, i32 1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiET_v_l58.region_id, %struct.__tgt_kernel_arguments* [[KERNEL_ARGS28]]) 12400 // CHECK11-NEXT: [[TMP132:%.*]] = icmp ne i32 [[TMP131]], 0 12401 // CHECK11-NEXT: br i1 [[TMP132]], label [[OMP_OFFLOAD_FAILED29:%.*]], label [[OMP_OFFLOAD_CONT30:%.*]] 12402 // CHECK11: omp_offload.failed29: 12403 // CHECK11-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiET_v_l58(i32 [[TMP85]], i32 [[TMP87]], i32* [[TMP88]], i32* [[TMP89]], i32* [[TMP90]]) #[[ATTR2]] 12404 // CHECK11-NEXT: br label [[OMP_OFFLOAD_CONT30]] 12405 // CHECK11: omp_offload.cont30: 12406 // CHECK11-NEXT: [[TMP133:%.*]] = load i32, i32* [[N]], align 4 12407 // CHECK11-NEXT: store i32 [[TMP133]], i32* [[N_CASTED31]], align 4 12408 // CHECK11-NEXT: [[TMP134:%.*]] = load i32, i32* [[N_CASTED31]], align 4 12409 // CHECK11-NEXT: [[TMP135:%.*]] = load i32*, i32** [[A]], align 4 12410 // CHECK11-NEXT: [[TMP136:%.*]] = load i32*, i32** [[B]], align 4 12411 // CHECK11-NEXT: [[TMP137:%.*]] = load i32*, i32** [[C]], align 4 12412 // CHECK11-NEXT: [[TMP138:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS32]], i32 0, i32 0 12413 // CHECK11-NEXT: [[TMP139:%.*]] = bitcast i8** [[TMP138]] to i32* 12414 // CHECK11-NEXT: store i32 [[TMP134]], i32* [[TMP139]], align 4 12415 // CHECK11-NEXT: [[TMP140:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS33]], i32 0, i32 0 12416 // CHECK11-NEXT: [[TMP141:%.*]] = bitcast i8** [[TMP140]] to i32* 12417 // CHECK11-NEXT: store i32 [[TMP134]], i32* [[TMP141]], align 4 12418 // CHECK11-NEXT: [[TMP142:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS34]], i32 0, i32 0 12419 // CHECK11-NEXT: store i8* null, i8** [[TMP142]], align 4 12420 // CHECK11-NEXT: [[TMP143:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS32]], i32 0, i32 1 12421 // CHECK11-NEXT: [[TMP144:%.*]] = bitcast i8** [[TMP143]] to i32** 12422 // CHECK11-NEXT: store i32* [[TMP135]], i32** [[TMP144]], align 4 12423 // CHECK11-NEXT: [[TMP145:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS33]], i32 0, i32 1 12424 // CHECK11-NEXT: [[TMP146:%.*]] = bitcast i8** [[TMP145]] to i32** 12425 // CHECK11-NEXT: store i32* [[TMP135]], i32** [[TMP146]], align 4 12426 // CHECK11-NEXT: [[TMP147:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS34]], i32 0, i32 1 12427 // CHECK11-NEXT: store i8* null, i8** [[TMP147]], align 4 12428 // CHECK11-NEXT: [[TMP148:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS32]], i32 0, i32 2 12429 // CHECK11-NEXT: [[TMP149:%.*]] = bitcast i8** [[TMP148]] to i32** 12430 // CHECK11-NEXT: store i32* [[TMP136]], i32** [[TMP149]], align 4 12431 // CHECK11-NEXT: [[TMP150:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS33]], i32 0, i32 2 12432 // CHECK11-NEXT: [[TMP151:%.*]] = bitcast i8** [[TMP150]] to i32** 12433 // CHECK11-NEXT: store i32* [[TMP136]], i32** [[TMP151]], align 4 12434 // CHECK11-NEXT: [[TMP152:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS34]], i32 0, i32 2 12435 // CHECK11-NEXT: store i8* null, i8** [[TMP152]], align 4 12436 // CHECK11-NEXT: [[TMP153:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS32]], i32 0, i32 3 12437 // CHECK11-NEXT: [[TMP154:%.*]] = bitcast i8** [[TMP153]] to i32** 12438 // CHECK11-NEXT: store i32* [[TMP137]], i32** [[TMP154]], align 4 12439 // CHECK11-NEXT: [[TMP155:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS33]], i32 0, i32 3 12440 // CHECK11-NEXT: [[TMP156:%.*]] = bitcast i8** [[TMP155]] to i32** 12441 // CHECK11-NEXT: store i32* [[TMP137]], i32** [[TMP156]], align 4 12442 // CHECK11-NEXT: [[TMP157:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS34]], i32 0, i32 3 12443 // CHECK11-NEXT: store i8* null, i8** [[TMP157]], align 4 12444 // CHECK11-NEXT: [[TMP158:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS32]], i32 0, i32 0 12445 // CHECK11-NEXT: [[TMP159:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS33]], i32 0, i32 0 12446 // CHECK11-NEXT: [[TMP160:%.*]] = load i32, i32* [[N]], align 4 12447 // CHECK11-NEXT: store i32 [[TMP160]], i32* [[DOTCAPTURE_EXPR_36]], align 4 12448 // CHECK11-NEXT: [[TMP161:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_36]], align 4 12449 // CHECK11-NEXT: [[SUB38:%.*]] = sub nsw i32 [[TMP161]], 0 12450 // CHECK11-NEXT: [[DIV39:%.*]] = sdiv i32 [[SUB38]], 1 12451 // CHECK11-NEXT: [[SUB40:%.*]] = sub nsw i32 [[DIV39]], 1 12452 // CHECK11-NEXT: store i32 [[SUB40]], i32* [[DOTCAPTURE_EXPR_37]], align 4 12453 // CHECK11-NEXT: [[TMP162:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_37]], align 4 12454 // CHECK11-NEXT: [[ADD41:%.*]] = add nsw i32 [[TMP162]], 1 12455 // CHECK11-NEXT: [[TMP163:%.*]] = zext i32 [[ADD41]] to i64 12456 // CHECK11-NEXT: [[KERNEL_ARGS42:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS]], align 8 12457 // CHECK11-NEXT: [[TMP164:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS42]], i32 0, i32 0 12458 // CHECK11-NEXT: store i32 1, i32* [[TMP164]], align 4 12459 // CHECK11-NEXT: [[TMP165:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS42]], i32 0, i32 1 12460 // CHECK11-NEXT: store i32 4, i32* [[TMP165]], align 4 12461 // CHECK11-NEXT: [[TMP166:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS42]], i32 0, i32 2 12462 // CHECK11-NEXT: store i8** [[TMP158]], i8*** [[TMP166]], align 4 12463 // CHECK11-NEXT: [[TMP167:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS42]], i32 0, i32 3 12464 // CHECK11-NEXT: store i8** [[TMP159]], i8*** [[TMP167]], align 4 12465 // CHECK11-NEXT: [[TMP168:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS42]], i32 0, i32 4 12466 // CHECK11-NEXT: store i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_sizes.40, i32 0, i32 0), i64** [[TMP168]], align 4 12467 // CHECK11-NEXT: [[TMP169:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS42]], i32 0, i32 5 12468 // CHECK11-NEXT: store i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_maptypes.41, i32 0, i32 0), i64** [[TMP169]], align 4 12469 // CHECK11-NEXT: [[TMP170:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS42]], i32 0, i32 6 12470 // CHECK11-NEXT: store i8** null, i8*** [[TMP170]], align 4 12471 // CHECK11-NEXT: [[TMP171:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS42]], i32 0, i32 7 12472 // CHECK11-NEXT: store i8** null, i8*** [[TMP171]], align 4 12473 // CHECK11-NEXT: [[TMP172:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS42]], i32 0, i32 8 12474 // CHECK11-NEXT: store i64 [[TMP163]], i64* [[TMP172]], align 8 12475 // CHECK11-NEXT: [[TMP173:%.*]] = call i32 @__tgt_target_kernel(%struct.ident_t* @[[GLOB3]], i64 -1, i32 0, i32 1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiET_v_l66.region_id, %struct.__tgt_kernel_arguments* [[KERNEL_ARGS42]]) 12476 // CHECK11-NEXT: [[TMP174:%.*]] = icmp ne i32 [[TMP173]], 0 12477 // CHECK11-NEXT: br i1 [[TMP174]], label [[OMP_OFFLOAD_FAILED43:%.*]], label [[OMP_OFFLOAD_CONT44:%.*]] 12478 // CHECK11: omp_offload.failed43: 12479 // CHECK11-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiET_v_l66(i32 [[TMP134]], i32* [[TMP135]], i32* [[TMP136]], i32* [[TMP137]]) #[[ATTR2]] 12480 // CHECK11-NEXT: br label [[OMP_OFFLOAD_CONT44]] 12481 // CHECK11: omp_offload.cont44: 12482 // CHECK11-NEXT: [[TMP175:%.*]] = load i32, i32* [[CH]], align 4 12483 // CHECK11-NEXT: store i32 [[TMP175]], i32* [[CH_CASTED45]], align 4 12484 // CHECK11-NEXT: [[TMP176:%.*]] = load i32, i32* [[CH_CASTED45]], align 4 12485 // CHECK11-NEXT: [[TMP177:%.*]] = load i32, i32* [[N]], align 4 12486 // CHECK11-NEXT: store i32 [[TMP177]], i32* [[N_CASTED46]], align 4 12487 // CHECK11-NEXT: [[TMP178:%.*]] = load i32, i32* [[N_CASTED46]], align 4 12488 // CHECK11-NEXT: [[TMP179:%.*]] = load i32*, i32** [[A]], align 4 12489 // CHECK11-NEXT: [[TMP180:%.*]] = load i32*, i32** [[B]], align 4 12490 // CHECK11-NEXT: [[TMP181:%.*]] = load i32*, i32** [[C]], align 4 12491 // CHECK11-NEXT: [[TMP182:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS47]], i32 0, i32 0 12492 // CHECK11-NEXT: [[TMP183:%.*]] = bitcast i8** [[TMP182]] to i32* 12493 // CHECK11-NEXT: store i32 [[TMP176]], i32* [[TMP183]], align 4 12494 // CHECK11-NEXT: [[TMP184:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS48]], i32 0, i32 0 12495 // CHECK11-NEXT: [[TMP185:%.*]] = bitcast i8** [[TMP184]] to i32* 12496 // CHECK11-NEXT: store i32 [[TMP176]], i32* [[TMP185]], align 4 12497 // CHECK11-NEXT: [[TMP186:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS49]], i32 0, i32 0 12498 // CHECK11-NEXT: store i8* null, i8** [[TMP186]], align 4 12499 // CHECK11-NEXT: [[TMP187:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS47]], i32 0, i32 1 12500 // CHECK11-NEXT: [[TMP188:%.*]] = bitcast i8** [[TMP187]] to i32* 12501 // CHECK11-NEXT: store i32 [[TMP178]], i32* [[TMP188]], align 4 12502 // CHECK11-NEXT: [[TMP189:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS48]], i32 0, i32 1 12503 // CHECK11-NEXT: [[TMP190:%.*]] = bitcast i8** [[TMP189]] to i32* 12504 // CHECK11-NEXT: store i32 [[TMP178]], i32* [[TMP190]], align 4 12505 // CHECK11-NEXT: [[TMP191:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS49]], i32 0, i32 1 12506 // CHECK11-NEXT: store i8* null, i8** [[TMP191]], align 4 12507 // CHECK11-NEXT: [[TMP192:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS47]], i32 0, i32 2 12508 // CHECK11-NEXT: [[TMP193:%.*]] = bitcast i8** [[TMP192]] to i32** 12509 // CHECK11-NEXT: store i32* [[TMP179]], i32** [[TMP193]], align 4 12510 // CHECK11-NEXT: [[TMP194:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS48]], i32 0, i32 2 12511 // CHECK11-NEXT: [[TMP195:%.*]] = bitcast i8** [[TMP194]] to i32** 12512 // CHECK11-NEXT: store i32* [[TMP179]], i32** [[TMP195]], align 4 12513 // CHECK11-NEXT: [[TMP196:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS49]], i32 0, i32 2 12514 // CHECK11-NEXT: store i8* null, i8** [[TMP196]], align 4 12515 // CHECK11-NEXT: [[TMP197:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS47]], i32 0, i32 3 12516 // CHECK11-NEXT: [[TMP198:%.*]] = bitcast i8** [[TMP197]] to i32** 12517 // CHECK11-NEXT: store i32* [[TMP180]], i32** [[TMP198]], align 4 12518 // CHECK11-NEXT: [[TMP199:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS48]], i32 0, i32 3 12519 // CHECK11-NEXT: [[TMP200:%.*]] = bitcast i8** [[TMP199]] to i32** 12520 // CHECK11-NEXT: store i32* [[TMP180]], i32** [[TMP200]], align 4 12521 // CHECK11-NEXT: [[TMP201:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS49]], i32 0, i32 3 12522 // CHECK11-NEXT: store i8* null, i8** [[TMP201]], align 4 12523 // CHECK11-NEXT: [[TMP202:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS47]], i32 0, i32 4 12524 // CHECK11-NEXT: [[TMP203:%.*]] = bitcast i8** [[TMP202]] to i32** 12525 // CHECK11-NEXT: store i32* [[TMP181]], i32** [[TMP203]], align 4 12526 // CHECK11-NEXT: [[TMP204:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS48]], i32 0, i32 4 12527 // CHECK11-NEXT: [[TMP205:%.*]] = bitcast i8** [[TMP204]] to i32** 12528 // CHECK11-NEXT: store i32* [[TMP181]], i32** [[TMP205]], align 4 12529 // CHECK11-NEXT: [[TMP206:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS49]], i32 0, i32 4 12530 // CHECK11-NEXT: store i8* null, i8** [[TMP206]], align 4 12531 // CHECK11-NEXT: [[TMP207:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS47]], i32 0, i32 0 12532 // CHECK11-NEXT: [[TMP208:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS48]], i32 0, i32 0 12533 // CHECK11-NEXT: [[TMP209:%.*]] = load i32, i32* [[N]], align 4 12534 // CHECK11-NEXT: store i32 [[TMP209]], i32* [[DOTCAPTURE_EXPR_51]], align 4 12535 // CHECK11-NEXT: [[TMP210:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_51]], align 4 12536 // CHECK11-NEXT: [[SUB53:%.*]] = sub nsw i32 [[TMP210]], 0 12537 // CHECK11-NEXT: [[DIV54:%.*]] = sdiv i32 [[SUB53]], 1 12538 // CHECK11-NEXT: [[SUB55:%.*]] = sub nsw i32 [[DIV54]], 1 12539 // CHECK11-NEXT: store i32 [[SUB55]], i32* [[DOTCAPTURE_EXPR_52]], align 4 12540 // CHECK11-NEXT: [[TMP211:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_52]], align 4 12541 // CHECK11-NEXT: [[ADD56:%.*]] = add nsw i32 [[TMP211]], 1 12542 // CHECK11-NEXT: [[TMP212:%.*]] = zext i32 [[ADD56]] to i64 12543 // CHECK11-NEXT: [[KERNEL_ARGS57:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS]], align 8 12544 // CHECK11-NEXT: [[TMP213:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS57]], i32 0, i32 0 12545 // CHECK11-NEXT: store i32 1, i32* [[TMP213]], align 4 12546 // CHECK11-NEXT: [[TMP214:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS57]], i32 0, i32 1 12547 // CHECK11-NEXT: store i32 5, i32* [[TMP214]], align 4 12548 // CHECK11-NEXT: [[TMP215:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS57]], i32 0, i32 2 12549 // CHECK11-NEXT: store i8** [[TMP207]], i8*** [[TMP215]], align 4 12550 // CHECK11-NEXT: [[TMP216:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS57]], i32 0, i32 3 12551 // CHECK11-NEXT: store i8** [[TMP208]], i8*** [[TMP216]], align 4 12552 // CHECK11-NEXT: [[TMP217:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS57]], i32 0, i32 4 12553 // CHECK11-NEXT: store i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_sizes.44, i32 0, i32 0), i64** [[TMP217]], align 4 12554 // CHECK11-NEXT: [[TMP218:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS57]], i32 0, i32 5 12555 // CHECK11-NEXT: store i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_maptypes.45, i32 0, i32 0), i64** [[TMP218]], align 4 12556 // CHECK11-NEXT: [[TMP219:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS57]], i32 0, i32 6 12557 // CHECK11-NEXT: store i8** null, i8*** [[TMP219]], align 4 12558 // CHECK11-NEXT: [[TMP220:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS57]], i32 0, i32 7 12559 // CHECK11-NEXT: store i8** null, i8*** [[TMP220]], align 4 12560 // CHECK11-NEXT: [[TMP221:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS57]], i32 0, i32 8 12561 // CHECK11-NEXT: store i64 [[TMP212]], i64* [[TMP221]], align 8 12562 // CHECK11-NEXT: [[TMP222:%.*]] = call i32 @__tgt_target_kernel(%struct.ident_t* @[[GLOB3]], i64 -1, i32 0, i32 1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiET_v_l74.region_id, %struct.__tgt_kernel_arguments* [[KERNEL_ARGS57]]) 12563 // CHECK11-NEXT: [[TMP223:%.*]] = icmp ne i32 [[TMP222]], 0 12564 // CHECK11-NEXT: br i1 [[TMP223]], label [[OMP_OFFLOAD_FAILED58:%.*]], label [[OMP_OFFLOAD_CONT59:%.*]] 12565 // CHECK11: omp_offload.failed58: 12566 // CHECK11-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiET_v_l74(i32 [[TMP176]], i32 [[TMP178]], i32* [[TMP179]], i32* [[TMP180]], i32* [[TMP181]]) #[[ATTR2]] 12567 // CHECK11-NEXT: br label [[OMP_OFFLOAD_CONT59]] 12568 // CHECK11: omp_offload.cont59: 12569 // CHECK11-NEXT: [[TMP224:%.*]] = load i32, i32* [[N]], align 4 12570 // CHECK11-NEXT: store i32 [[TMP224]], i32* [[N_CASTED60]], align 4 12571 // CHECK11-NEXT: [[TMP225:%.*]] = load i32, i32* [[N_CASTED60]], align 4 12572 // CHECK11-NEXT: [[TMP226:%.*]] = load i32*, i32** [[A]], align 4 12573 // CHECK11-NEXT: [[TMP227:%.*]] = load i32*, i32** [[B]], align 4 12574 // CHECK11-NEXT: [[TMP228:%.*]] = load i32*, i32** [[C]], align 4 12575 // CHECK11-NEXT: [[TMP229:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS61]], i32 0, i32 0 12576 // CHECK11-NEXT: [[TMP230:%.*]] = bitcast i8** [[TMP229]] to i32* 12577 // CHECK11-NEXT: store i32 [[TMP225]], i32* [[TMP230]], align 4 12578 // CHECK11-NEXT: [[TMP231:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS62]], i32 0, i32 0 12579 // CHECK11-NEXT: [[TMP232:%.*]] = bitcast i8** [[TMP231]] to i32* 12580 // CHECK11-NEXT: store i32 [[TMP225]], i32* [[TMP232]], align 4 12581 // CHECK11-NEXT: [[TMP233:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS63]], i32 0, i32 0 12582 // CHECK11-NEXT: store i8* null, i8** [[TMP233]], align 4 12583 // CHECK11-NEXT: [[TMP234:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS61]], i32 0, i32 1 12584 // CHECK11-NEXT: [[TMP235:%.*]] = bitcast i8** [[TMP234]] to i32** 12585 // CHECK11-NEXT: store i32* [[TMP226]], i32** [[TMP235]], align 4 12586 // CHECK11-NEXT: [[TMP236:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS62]], i32 0, i32 1 12587 // CHECK11-NEXT: [[TMP237:%.*]] = bitcast i8** [[TMP236]] to i32** 12588 // CHECK11-NEXT: store i32* [[TMP226]], i32** [[TMP237]], align 4 12589 // CHECK11-NEXT: [[TMP238:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS63]], i32 0, i32 1 12590 // CHECK11-NEXT: store i8* null, i8** [[TMP238]], align 4 12591 // CHECK11-NEXT: [[TMP239:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS61]], i32 0, i32 2 12592 // CHECK11-NEXT: [[TMP240:%.*]] = bitcast i8** [[TMP239]] to i32** 12593 // CHECK11-NEXT: store i32* [[TMP227]], i32** [[TMP240]], align 4 12594 // CHECK11-NEXT: [[TMP241:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS62]], i32 0, i32 2 12595 // CHECK11-NEXT: [[TMP242:%.*]] = bitcast i8** [[TMP241]] to i32** 12596 // CHECK11-NEXT: store i32* [[TMP227]], i32** [[TMP242]], align 4 12597 // CHECK11-NEXT: [[TMP243:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS63]], i32 0, i32 2 12598 // CHECK11-NEXT: store i8* null, i8** [[TMP243]], align 4 12599 // CHECK11-NEXT: [[TMP244:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS61]], i32 0, i32 3 12600 // CHECK11-NEXT: [[TMP245:%.*]] = bitcast i8** [[TMP244]] to i32** 12601 // CHECK11-NEXT: store i32* [[TMP228]], i32** [[TMP245]], align 4 12602 // CHECK11-NEXT: [[TMP246:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS62]], i32 0, i32 3 12603 // CHECK11-NEXT: [[TMP247:%.*]] = bitcast i8** [[TMP246]] to i32** 12604 // CHECK11-NEXT: store i32* [[TMP228]], i32** [[TMP247]], align 4 12605 // CHECK11-NEXT: [[TMP248:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS63]], i32 0, i32 3 12606 // CHECK11-NEXT: store i8* null, i8** [[TMP248]], align 4 12607 // CHECK11-NEXT: [[TMP249:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS61]], i32 0, i32 0 12608 // CHECK11-NEXT: [[TMP250:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS62]], i32 0, i32 0 12609 // CHECK11-NEXT: [[TMP251:%.*]] = load i32, i32* [[N]], align 4 12610 // CHECK11-NEXT: store i32 [[TMP251]], i32* [[DOTCAPTURE_EXPR_65]], align 4 12611 // CHECK11-NEXT: [[TMP252:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_65]], align 4 12612 // CHECK11-NEXT: [[SUB67:%.*]] = sub nsw i32 [[TMP252]], 0 12613 // CHECK11-NEXT: [[DIV68:%.*]] = sdiv i32 [[SUB67]], 1 12614 // CHECK11-NEXT: [[SUB69:%.*]] = sub nsw i32 [[DIV68]], 1 12615 // CHECK11-NEXT: store i32 [[SUB69]], i32* [[DOTCAPTURE_EXPR_66]], align 4 12616 // CHECK11-NEXT: [[TMP253:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_66]], align 4 12617 // CHECK11-NEXT: [[ADD70:%.*]] = add nsw i32 [[TMP253]], 1 12618 // CHECK11-NEXT: [[TMP254:%.*]] = zext i32 [[ADD70]] to i64 12619 // CHECK11-NEXT: [[KERNEL_ARGS71:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS]], align 8 12620 // CHECK11-NEXT: [[TMP255:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS71]], i32 0, i32 0 12621 // CHECK11-NEXT: store i32 1, i32* [[TMP255]], align 4 12622 // CHECK11-NEXT: [[TMP256:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS71]], i32 0, i32 1 12623 // CHECK11-NEXT: store i32 4, i32* [[TMP256]], align 4 12624 // CHECK11-NEXT: [[TMP257:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS71]], i32 0, i32 2 12625 // CHECK11-NEXT: store i8** [[TMP249]], i8*** [[TMP257]], align 4 12626 // CHECK11-NEXT: [[TMP258:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS71]], i32 0, i32 3 12627 // CHECK11-NEXT: store i8** [[TMP250]], i8*** [[TMP258]], align 4 12628 // CHECK11-NEXT: [[TMP259:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS71]], i32 0, i32 4 12629 // CHECK11-NEXT: store i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_sizes.48, i32 0, i32 0), i64** [[TMP259]], align 4 12630 // CHECK11-NEXT: [[TMP260:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS71]], i32 0, i32 5 12631 // CHECK11-NEXT: store i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_maptypes.49, i32 0, i32 0), i64** [[TMP260]], align 4 12632 // CHECK11-NEXT: [[TMP261:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS71]], i32 0, i32 6 12633 // CHECK11-NEXT: store i8** null, i8*** [[TMP261]], align 4 12634 // CHECK11-NEXT: [[TMP262:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS71]], i32 0, i32 7 12635 // CHECK11-NEXT: store i8** null, i8*** [[TMP262]], align 4 12636 // CHECK11-NEXT: [[TMP263:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS71]], i32 0, i32 8 12637 // CHECK11-NEXT: store i64 [[TMP254]], i64* [[TMP263]], align 8 12638 // CHECK11-NEXT: [[TMP264:%.*]] = call i32 @__tgt_target_kernel(%struct.ident_t* @[[GLOB3]], i64 -1, i32 0, i32 1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiET_v_l82.region_id, %struct.__tgt_kernel_arguments* [[KERNEL_ARGS71]]) 12639 // CHECK11-NEXT: [[TMP265:%.*]] = icmp ne i32 [[TMP264]], 0 12640 // CHECK11-NEXT: br i1 [[TMP265]], label [[OMP_OFFLOAD_FAILED72:%.*]], label [[OMP_OFFLOAD_CONT73:%.*]] 12641 // CHECK11: omp_offload.failed72: 12642 // CHECK11-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiET_v_l82(i32 [[TMP225]], i32* [[TMP226]], i32* [[TMP227]], i32* [[TMP228]]) #[[ATTR2]] 12643 // CHECK11-NEXT: br label [[OMP_OFFLOAD_CONT73]] 12644 // CHECK11: omp_offload.cont73: 12645 // CHECK11-NEXT: [[TMP266:%.*]] = load i32, i32* [[CH]], align 4 12646 // CHECK11-NEXT: store i32 [[TMP266]], i32* [[CH_CASTED74]], align 4 12647 // CHECK11-NEXT: [[TMP267:%.*]] = load i32, i32* [[CH_CASTED74]], align 4 12648 // CHECK11-NEXT: [[TMP268:%.*]] = load i32, i32* [[N]], align 4 12649 // CHECK11-NEXT: store i32 [[TMP268]], i32* [[N_CASTED75]], align 4 12650 // CHECK11-NEXT: [[TMP269:%.*]] = load i32, i32* [[N_CASTED75]], align 4 12651 // CHECK11-NEXT: [[TMP270:%.*]] = load i32*, i32** [[A]], align 4 12652 // CHECK11-NEXT: [[TMP271:%.*]] = load i32*, i32** [[B]], align 4 12653 // CHECK11-NEXT: [[TMP272:%.*]] = load i32*, i32** [[C]], align 4 12654 // CHECK11-NEXT: [[TMP273:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS76]], i32 0, i32 0 12655 // CHECK11-NEXT: [[TMP274:%.*]] = bitcast i8** [[TMP273]] to i32* 12656 // CHECK11-NEXT: store i32 [[TMP267]], i32* [[TMP274]], align 4 12657 // CHECK11-NEXT: [[TMP275:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS77]], i32 0, i32 0 12658 // CHECK11-NEXT: [[TMP276:%.*]] = bitcast i8** [[TMP275]] to i32* 12659 // CHECK11-NEXT: store i32 [[TMP267]], i32* [[TMP276]], align 4 12660 // CHECK11-NEXT: [[TMP277:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS78]], i32 0, i32 0 12661 // CHECK11-NEXT: store i8* null, i8** [[TMP277]], align 4 12662 // CHECK11-NEXT: [[TMP278:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS76]], i32 0, i32 1 12663 // CHECK11-NEXT: [[TMP279:%.*]] = bitcast i8** [[TMP278]] to i32* 12664 // CHECK11-NEXT: store i32 [[TMP269]], i32* [[TMP279]], align 4 12665 // CHECK11-NEXT: [[TMP280:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS77]], i32 0, i32 1 12666 // CHECK11-NEXT: [[TMP281:%.*]] = bitcast i8** [[TMP280]] to i32* 12667 // CHECK11-NEXT: store i32 [[TMP269]], i32* [[TMP281]], align 4 12668 // CHECK11-NEXT: [[TMP282:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS78]], i32 0, i32 1 12669 // CHECK11-NEXT: store i8* null, i8** [[TMP282]], align 4 12670 // CHECK11-NEXT: [[TMP283:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS76]], i32 0, i32 2 12671 // CHECK11-NEXT: [[TMP284:%.*]] = bitcast i8** [[TMP283]] to i32** 12672 // CHECK11-NEXT: store i32* [[TMP270]], i32** [[TMP284]], align 4 12673 // CHECK11-NEXT: [[TMP285:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS77]], i32 0, i32 2 12674 // CHECK11-NEXT: [[TMP286:%.*]] = bitcast i8** [[TMP285]] to i32** 12675 // CHECK11-NEXT: store i32* [[TMP270]], i32** [[TMP286]], align 4 12676 // CHECK11-NEXT: [[TMP287:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS78]], i32 0, i32 2 12677 // CHECK11-NEXT: store i8* null, i8** [[TMP287]], align 4 12678 // CHECK11-NEXT: [[TMP288:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS76]], i32 0, i32 3 12679 // CHECK11-NEXT: [[TMP289:%.*]] = bitcast i8** [[TMP288]] to i32** 12680 // CHECK11-NEXT: store i32* [[TMP271]], i32** [[TMP289]], align 4 12681 // CHECK11-NEXT: [[TMP290:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS77]], i32 0, i32 3 12682 // CHECK11-NEXT: [[TMP291:%.*]] = bitcast i8** [[TMP290]] to i32** 12683 // CHECK11-NEXT: store i32* [[TMP271]], i32** [[TMP291]], align 4 12684 // CHECK11-NEXT: [[TMP292:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS78]], i32 0, i32 3 12685 // CHECK11-NEXT: store i8* null, i8** [[TMP292]], align 4 12686 // CHECK11-NEXT: [[TMP293:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS76]], i32 0, i32 4 12687 // CHECK11-NEXT: [[TMP294:%.*]] = bitcast i8** [[TMP293]] to i32** 12688 // CHECK11-NEXT: store i32* [[TMP272]], i32** [[TMP294]], align 4 12689 // CHECK11-NEXT: [[TMP295:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS77]], i32 0, i32 4 12690 // CHECK11-NEXT: [[TMP296:%.*]] = bitcast i8** [[TMP295]] to i32** 12691 // CHECK11-NEXT: store i32* [[TMP272]], i32** [[TMP296]], align 4 12692 // CHECK11-NEXT: [[TMP297:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS78]], i32 0, i32 4 12693 // CHECK11-NEXT: store i8* null, i8** [[TMP297]], align 4 12694 // CHECK11-NEXT: [[TMP298:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS76]], i32 0, i32 0 12695 // CHECK11-NEXT: [[TMP299:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS77]], i32 0, i32 0 12696 // CHECK11-NEXT: [[TMP300:%.*]] = load i32, i32* [[N]], align 4 12697 // CHECK11-NEXT: store i32 [[TMP300]], i32* [[DOTCAPTURE_EXPR_80]], align 4 12698 // CHECK11-NEXT: [[TMP301:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_80]], align 4 12699 // CHECK11-NEXT: [[SUB82:%.*]] = sub nsw i32 [[TMP301]], 0 12700 // CHECK11-NEXT: [[DIV83:%.*]] = sdiv i32 [[SUB82]], 1 12701 // CHECK11-NEXT: [[SUB84:%.*]] = sub nsw i32 [[DIV83]], 1 12702 // CHECK11-NEXT: store i32 [[SUB84]], i32* [[DOTCAPTURE_EXPR_81]], align 4 12703 // CHECK11-NEXT: [[TMP302:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_81]], align 4 12704 // CHECK11-NEXT: [[ADD85:%.*]] = add nsw i32 [[TMP302]], 1 12705 // CHECK11-NEXT: [[TMP303:%.*]] = zext i32 [[ADD85]] to i64 12706 // CHECK11-NEXT: [[KERNEL_ARGS86:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS]], align 8 12707 // CHECK11-NEXT: [[TMP304:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS86]], i32 0, i32 0 12708 // CHECK11-NEXT: store i32 1, i32* [[TMP304]], align 4 12709 // CHECK11-NEXT: [[TMP305:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS86]], i32 0, i32 1 12710 // CHECK11-NEXT: store i32 5, i32* [[TMP305]], align 4 12711 // CHECK11-NEXT: [[TMP306:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS86]], i32 0, i32 2 12712 // CHECK11-NEXT: store i8** [[TMP298]], i8*** [[TMP306]], align 4 12713 // CHECK11-NEXT: [[TMP307:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS86]], i32 0, i32 3 12714 // CHECK11-NEXT: store i8** [[TMP299]], i8*** [[TMP307]], align 4 12715 // CHECK11-NEXT: [[TMP308:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS86]], i32 0, i32 4 12716 // CHECK11-NEXT: store i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_sizes.52, i32 0, i32 0), i64** [[TMP308]], align 4 12717 // CHECK11-NEXT: [[TMP309:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS86]], i32 0, i32 5 12718 // CHECK11-NEXT: store i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_maptypes.53, i32 0, i32 0), i64** [[TMP309]], align 4 12719 // CHECK11-NEXT: [[TMP310:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS86]], i32 0, i32 6 12720 // CHECK11-NEXT: store i8** null, i8*** [[TMP310]], align 4 12721 // CHECK11-NEXT: [[TMP311:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS86]], i32 0, i32 7 12722 // CHECK11-NEXT: store i8** null, i8*** [[TMP311]], align 4 12723 // CHECK11-NEXT: [[TMP312:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS86]], i32 0, i32 8 12724 // CHECK11-NEXT: store i64 [[TMP303]], i64* [[TMP312]], align 8 12725 // CHECK11-NEXT: [[TMP313:%.*]] = call i32 @__tgt_target_kernel(%struct.ident_t* @[[GLOB3]], i64 -1, i32 0, i32 1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiET_v_l90.region_id, %struct.__tgt_kernel_arguments* [[KERNEL_ARGS86]]) 12726 // CHECK11-NEXT: [[TMP314:%.*]] = icmp ne i32 [[TMP313]], 0 12727 // CHECK11-NEXT: br i1 [[TMP314]], label [[OMP_OFFLOAD_FAILED87:%.*]], label [[OMP_OFFLOAD_CONT88:%.*]] 12728 // CHECK11: omp_offload.failed87: 12729 // CHECK11-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiET_v_l90(i32 [[TMP267]], i32 [[TMP269]], i32* [[TMP270]], i32* [[TMP271]], i32* [[TMP272]]) #[[ATTR2]] 12730 // CHECK11-NEXT: br label [[OMP_OFFLOAD_CONT88]] 12731 // CHECK11: omp_offload.cont88: 12732 // CHECK11-NEXT: ret i32 0 12733 // 12734 // 12735 // CHECK11-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiET_v_l42 12736 // CHECK11-SAME: (i32 noundef [[N:%.*]], i32* noundef [[A:%.*]], i32* noundef [[B:%.*]], i32* noundef [[C:%.*]]) #[[ATTR1]] { 12737 // CHECK11-NEXT: entry: 12738 // CHECK11-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4 12739 // CHECK11-NEXT: [[A_ADDR:%.*]] = alloca i32*, align 4 12740 // CHECK11-NEXT: [[B_ADDR:%.*]] = alloca i32*, align 4 12741 // CHECK11-NEXT: [[C_ADDR:%.*]] = alloca i32*, align 4 12742 // CHECK11-NEXT: store i32 [[N]], i32* [[N_ADDR]], align 4 12743 // CHECK11-NEXT: store i32* [[A]], i32** [[A_ADDR]], align 4 12744 // CHECK11-NEXT: store i32* [[B]], i32** [[B_ADDR]], align 4 12745 // CHECK11-NEXT: store i32* [[C]], i32** [[C_ADDR]], align 4 12746 // CHECK11-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 4, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32*, i32**, i32**, i32**)* @.omp_outlined..26 to void (i32*, i32*, ...)*), i32* [[N_ADDR]], i32** [[A_ADDR]], i32** [[B_ADDR]], i32** [[C_ADDR]]) 12747 // CHECK11-NEXT: ret void 12748 // 12749 // 12750 // CHECK11-LABEL: define {{[^@]+}}@.omp_outlined..26 12751 // CHECK11-SAME: (i32* noalias noundef [[DOTGLOBAL_TID_:%.*]], i32* noalias noundef [[DOTBOUND_TID_:%.*]], i32* noundef nonnull align 4 dereferenceable(4) [[N:%.*]], i32** noundef nonnull align 4 dereferenceable(4) [[A:%.*]], i32** noundef nonnull align 4 dereferenceable(4) [[B:%.*]], i32** noundef nonnull align 4 dereferenceable(4) [[C:%.*]]) #[[ATTR1]] { 12752 // CHECK11-NEXT: entry: 12753 // CHECK11-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4 12754 // CHECK11-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4 12755 // CHECK11-NEXT: [[N_ADDR:%.*]] = alloca i32*, align 4 12756 // CHECK11-NEXT: [[A_ADDR:%.*]] = alloca i32**, align 4 12757 // CHECK11-NEXT: [[B_ADDR:%.*]] = alloca i32**, align 4 12758 // CHECK11-NEXT: [[C_ADDR:%.*]] = alloca i32**, align 4 12759 // CHECK11-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 12760 // CHECK11-NEXT: [[TMP:%.*]] = alloca i32, align 4 12761 // CHECK11-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 12762 // CHECK11-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 12763 // CHECK11-NEXT: [[I:%.*]] = alloca i32, align 4 12764 // CHECK11-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4 12765 // CHECK11-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4 12766 // CHECK11-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 12767 // CHECK11-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 12768 // CHECK11-NEXT: [[I3:%.*]] = alloca i32, align 4 12769 // CHECK11-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 4 12770 // CHECK11-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 4 12771 // CHECK11-NEXT: store i32* [[N]], i32** [[N_ADDR]], align 4 12772 // CHECK11-NEXT: store i32** [[A]], i32*** [[A_ADDR]], align 4 12773 // CHECK11-NEXT: store i32** [[B]], i32*** [[B_ADDR]], align 4 12774 // CHECK11-NEXT: store i32** [[C]], i32*** [[C_ADDR]], align 4 12775 // CHECK11-NEXT: [[TMP0:%.*]] = load i32*, i32** [[N_ADDR]], align 4 12776 // CHECK11-NEXT: [[TMP1:%.*]] = load i32**, i32*** [[A_ADDR]], align 4 12777 // CHECK11-NEXT: [[TMP2:%.*]] = load i32**, i32*** [[B_ADDR]], align 4 12778 // CHECK11-NEXT: [[TMP3:%.*]] = load i32**, i32*** [[C_ADDR]], align 4 12779 // CHECK11-NEXT: [[TMP4:%.*]] = load i32, i32* [[TMP0]], align 4 12780 // CHECK11-NEXT: store i32 [[TMP4]], i32* [[DOTCAPTURE_EXPR_]], align 4 12781 // CHECK11-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 12782 // CHECK11-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP5]], 0 12783 // CHECK11-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 12784 // CHECK11-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1 12785 // CHECK11-NEXT: store i32 [[SUB2]], i32* [[DOTCAPTURE_EXPR_1]], align 4 12786 // CHECK11-NEXT: store i32 0, i32* [[I]], align 4 12787 // CHECK11-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 12788 // CHECK11-NEXT: [[CMP:%.*]] = icmp slt i32 0, [[TMP6]] 12789 // CHECK11-NEXT: br i1 [[CMP]], label [[OMP_PRECOND_THEN:%.*]], label [[OMP_PRECOND_END:%.*]] 12790 // CHECK11: omp.precond.then: 12791 // CHECK11-NEXT: store i32 0, i32* [[DOTOMP_COMB_LB]], align 4 12792 // CHECK11-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 12793 // CHECK11-NEXT: store i32 [[TMP7]], i32* [[DOTOMP_COMB_UB]], align 4 12794 // CHECK11-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4 12795 // CHECK11-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4 12796 // CHECK11-NEXT: [[TMP8:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 4 12797 // CHECK11-NEXT: [[TMP9:%.*]] = load i32, i32* [[TMP8]], align 4 12798 // CHECK11-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1]], i32 [[TMP9]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_COMB_LB]], i32* [[DOTOMP_COMB_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1) 12799 // CHECK11-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 12800 // CHECK11-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 12801 // CHECK11-NEXT: [[CMP4:%.*]] = icmp sgt i32 [[TMP10]], [[TMP11]] 12802 // CHECK11-NEXT: br i1 [[CMP4]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] 12803 // CHECK11: cond.true: 12804 // CHECK11-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 12805 // CHECK11-NEXT: br label [[COND_END:%.*]] 12806 // CHECK11: cond.false: 12807 // CHECK11-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 12808 // CHECK11-NEXT: br label [[COND_END]] 12809 // CHECK11: cond.end: 12810 // CHECK11-NEXT: [[COND:%.*]] = phi i32 [ [[TMP12]], [[COND_TRUE]] ], [ [[TMP13]], [[COND_FALSE]] ] 12811 // CHECK11-NEXT: store i32 [[COND]], i32* [[DOTOMP_COMB_UB]], align 4 12812 // CHECK11-NEXT: [[TMP14:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4 12813 // CHECK11-NEXT: store i32 [[TMP14]], i32* [[DOTOMP_IV]], align 4 12814 // CHECK11-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] 12815 // CHECK11: omp.inner.for.cond: 12816 // CHECK11-NEXT: [[TMP15:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !63 12817 // CHECK11-NEXT: [[TMP16:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4, !llvm.access.group !63 12818 // CHECK11-NEXT: [[CMP5:%.*]] = icmp sle i32 [[TMP15]], [[TMP16]] 12819 // CHECK11-NEXT: br i1 [[CMP5]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] 12820 // CHECK11: omp.inner.for.body: 12821 // CHECK11-NEXT: [[TMP17:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4, !llvm.access.group !63 12822 // CHECK11-NEXT: [[TMP18:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4, !llvm.access.group !63 12823 // CHECK11-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 6, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32, i32, i32*, i32**, i32**, i32**)* @.omp_outlined..27 to void (i32*, i32*, ...)*), i32 [[TMP17]], i32 [[TMP18]], i32* [[TMP0]], i32** [[TMP1]], i32** [[TMP2]], i32** [[TMP3]]), !llvm.access.group !63 12824 // CHECK11-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] 12825 // CHECK11: omp.inner.for.inc: 12826 // CHECK11-NEXT: [[TMP19:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !63 12827 // CHECK11-NEXT: [[TMP20:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4, !llvm.access.group !63 12828 // CHECK11-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP19]], [[TMP20]] 12829 // CHECK11-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4, !llvm.access.group !63 12830 // CHECK11-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP64:![0-9]+]] 12831 // CHECK11: omp.inner.for.end: 12832 // CHECK11-NEXT: br label [[OMP_LOOP_EXIT:%.*]] 12833 // CHECK11: omp.loop.exit: 12834 // CHECK11-NEXT: [[TMP21:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 4 12835 // CHECK11-NEXT: [[TMP22:%.*]] = load i32, i32* [[TMP21]], align 4 12836 // CHECK11-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP22]]) 12837 // CHECK11-NEXT: [[TMP23:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4 12838 // CHECK11-NEXT: [[TMP24:%.*]] = icmp ne i32 [[TMP23]], 0 12839 // CHECK11-NEXT: br i1 [[TMP24]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]] 12840 // CHECK11: .omp.final.then: 12841 // CHECK11-NEXT: [[TMP25:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 12842 // CHECK11-NEXT: [[SUB6:%.*]] = sub nsw i32 [[TMP25]], 0 12843 // CHECK11-NEXT: [[DIV7:%.*]] = sdiv i32 [[SUB6]], 1 12844 // CHECK11-NEXT: [[MUL:%.*]] = mul nsw i32 [[DIV7]], 1 12845 // CHECK11-NEXT: [[ADD8:%.*]] = add nsw i32 0, [[MUL]] 12846 // CHECK11-NEXT: store i32 [[ADD8]], i32* [[I3]], align 4 12847 // CHECK11-NEXT: br label [[DOTOMP_FINAL_DONE]] 12848 // CHECK11: .omp.final.done: 12849 // CHECK11-NEXT: br label [[OMP_PRECOND_END]] 12850 // CHECK11: omp.precond.end: 12851 // CHECK11-NEXT: ret void 12852 // 12853 // 12854 // CHECK11-LABEL: define {{[^@]+}}@.omp_outlined..27 12855 // CHECK11-SAME: (i32* noalias noundef [[DOTGLOBAL_TID_:%.*]], i32* noalias noundef [[DOTBOUND_TID_:%.*]], i32 noundef [[DOTPREVIOUS_LB_:%.*]], i32 noundef [[DOTPREVIOUS_UB_:%.*]], i32* noundef nonnull align 4 dereferenceable(4) [[N:%.*]], i32** noundef nonnull align 4 dereferenceable(4) [[A:%.*]], i32** noundef nonnull align 4 dereferenceable(4) [[B:%.*]], i32** noundef nonnull align 4 dereferenceable(4) [[C:%.*]]) #[[ATTR1]] { 12856 // CHECK11-NEXT: entry: 12857 // CHECK11-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4 12858 // CHECK11-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4 12859 // CHECK11-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i32, align 4 12860 // CHECK11-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i32, align 4 12861 // CHECK11-NEXT: [[N_ADDR:%.*]] = alloca i32*, align 4 12862 // CHECK11-NEXT: [[A_ADDR:%.*]] = alloca i32**, align 4 12863 // CHECK11-NEXT: [[B_ADDR:%.*]] = alloca i32**, align 4 12864 // CHECK11-NEXT: [[C_ADDR:%.*]] = alloca i32**, align 4 12865 // CHECK11-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 12866 // CHECK11-NEXT: [[TMP:%.*]] = alloca i32, align 4 12867 // CHECK11-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 12868 // CHECK11-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 12869 // CHECK11-NEXT: [[I:%.*]] = alloca i32, align 4 12870 // CHECK11-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 12871 // CHECK11-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4 12872 // CHECK11-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 12873 // CHECK11-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 12874 // CHECK11-NEXT: [[I3:%.*]] = alloca i32, align 4 12875 // CHECK11-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 4 12876 // CHECK11-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 4 12877 // CHECK11-NEXT: store i32 [[DOTPREVIOUS_LB_]], i32* [[DOTPREVIOUS_LB__ADDR]], align 4 12878 // CHECK11-NEXT: store i32 [[DOTPREVIOUS_UB_]], i32* [[DOTPREVIOUS_UB__ADDR]], align 4 12879 // CHECK11-NEXT: store i32* [[N]], i32** [[N_ADDR]], align 4 12880 // CHECK11-NEXT: store i32** [[A]], i32*** [[A_ADDR]], align 4 12881 // CHECK11-NEXT: store i32** [[B]], i32*** [[B_ADDR]], align 4 12882 // CHECK11-NEXT: store i32** [[C]], i32*** [[C_ADDR]], align 4 12883 // CHECK11-NEXT: [[TMP0:%.*]] = load i32*, i32** [[N_ADDR]], align 4 12884 // CHECK11-NEXT: [[TMP1:%.*]] = load i32**, i32*** [[A_ADDR]], align 4 12885 // CHECK11-NEXT: [[TMP2:%.*]] = load i32**, i32*** [[B_ADDR]], align 4 12886 // CHECK11-NEXT: [[TMP3:%.*]] = load i32**, i32*** [[C_ADDR]], align 4 12887 // CHECK11-NEXT: [[TMP4:%.*]] = load i32, i32* [[TMP0]], align 4 12888 // CHECK11-NEXT: store i32 [[TMP4]], i32* [[DOTCAPTURE_EXPR_]], align 4 12889 // CHECK11-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 12890 // CHECK11-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP5]], 0 12891 // CHECK11-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 12892 // CHECK11-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1 12893 // CHECK11-NEXT: store i32 [[SUB2]], i32* [[DOTCAPTURE_EXPR_1]], align 4 12894 // CHECK11-NEXT: store i32 0, i32* [[I]], align 4 12895 // CHECK11-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 12896 // CHECK11-NEXT: [[CMP:%.*]] = icmp slt i32 0, [[TMP6]] 12897 // CHECK11-NEXT: br i1 [[CMP]], label [[OMP_PRECOND_THEN:%.*]], label [[OMP_PRECOND_END:%.*]] 12898 // CHECK11: omp.precond.then: 12899 // CHECK11-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4 12900 // CHECK11-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 12901 // CHECK11-NEXT: store i32 [[TMP7]], i32* [[DOTOMP_UB]], align 4 12902 // CHECK11-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTPREVIOUS_LB__ADDR]], align 4 12903 // CHECK11-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTPREVIOUS_UB__ADDR]], align 4 12904 // CHECK11-NEXT: store i32 [[TMP8]], i32* [[DOTOMP_LB]], align 4 12905 // CHECK11-NEXT: store i32 [[TMP9]], i32* [[DOTOMP_UB]], align 4 12906 // CHECK11-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4 12907 // CHECK11-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4 12908 // CHECK11-NEXT: [[TMP10:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 4 12909 // CHECK11-NEXT: [[TMP11:%.*]] = load i32, i32* [[TMP10]], align 4 12910 // CHECK11-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB2]], i32 [[TMP11]], i32 34, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1) 12911 // CHECK11-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 12912 // CHECK11-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 12913 // CHECK11-NEXT: [[CMP4:%.*]] = icmp sgt i32 [[TMP12]], [[TMP13]] 12914 // CHECK11-NEXT: br i1 [[CMP4]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] 12915 // CHECK11: cond.true: 12916 // CHECK11-NEXT: [[TMP14:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 12917 // CHECK11-NEXT: br label [[COND_END:%.*]] 12918 // CHECK11: cond.false: 12919 // CHECK11-NEXT: [[TMP15:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 12920 // CHECK11-NEXT: br label [[COND_END]] 12921 // CHECK11: cond.end: 12922 // CHECK11-NEXT: [[COND:%.*]] = phi i32 [ [[TMP14]], [[COND_TRUE]] ], [ [[TMP15]], [[COND_FALSE]] ] 12923 // CHECK11-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4 12924 // CHECK11-NEXT: [[TMP16:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4 12925 // CHECK11-NEXT: store i32 [[TMP16]], i32* [[DOTOMP_IV]], align 4 12926 // CHECK11-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] 12927 // CHECK11: omp.inner.for.cond: 12928 // CHECK11-NEXT: [[TMP17:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !66 12929 // CHECK11-NEXT: [[TMP18:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4, !llvm.access.group !66 12930 // CHECK11-NEXT: [[CMP5:%.*]] = icmp sle i32 [[TMP17]], [[TMP18]] 12931 // CHECK11-NEXT: br i1 [[CMP5]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] 12932 // CHECK11: omp.inner.for.body: 12933 // CHECK11-NEXT: [[TMP19:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !66 12934 // CHECK11-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP19]], 1 12935 // CHECK11-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] 12936 // CHECK11-NEXT: store i32 [[ADD]], i32* [[I3]], align 4, !llvm.access.group !66 12937 // CHECK11-NEXT: [[TMP20:%.*]] = load i32*, i32** [[TMP2]], align 4, !llvm.access.group !66 12938 // CHECK11-NEXT: [[TMP21:%.*]] = load i32, i32* [[I3]], align 4, !llvm.access.group !66 12939 // CHECK11-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i32, i32* [[TMP20]], i32 [[TMP21]] 12940 // CHECK11-NEXT: [[TMP22:%.*]] = load i32, i32* [[ARRAYIDX]], align 4, !llvm.access.group !66 12941 // CHECK11-NEXT: [[TMP23:%.*]] = load i32*, i32** [[TMP3]], align 4, !llvm.access.group !66 12942 // CHECK11-NEXT: [[TMP24:%.*]] = load i32, i32* [[I3]], align 4, !llvm.access.group !66 12943 // CHECK11-NEXT: [[ARRAYIDX6:%.*]] = getelementptr inbounds i32, i32* [[TMP23]], i32 [[TMP24]] 12944 // CHECK11-NEXT: [[TMP25:%.*]] = load i32, i32* [[ARRAYIDX6]], align 4, !llvm.access.group !66 12945 // CHECK11-NEXT: [[ADD7:%.*]] = add nsw i32 [[TMP22]], [[TMP25]] 12946 // CHECK11-NEXT: [[TMP26:%.*]] = load i32*, i32** [[TMP1]], align 4, !llvm.access.group !66 12947 // CHECK11-NEXT: [[TMP27:%.*]] = load i32, i32* [[I3]], align 4, !llvm.access.group !66 12948 // CHECK11-NEXT: [[ARRAYIDX8:%.*]] = getelementptr inbounds i32, i32* [[TMP26]], i32 [[TMP27]] 12949 // CHECK11-NEXT: store i32 [[ADD7]], i32* [[ARRAYIDX8]], align 4, !llvm.access.group !66 12950 // CHECK11-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] 12951 // CHECK11: omp.body.continue: 12952 // CHECK11-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] 12953 // CHECK11: omp.inner.for.inc: 12954 // CHECK11-NEXT: [[TMP28:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !66 12955 // CHECK11-NEXT: [[ADD9:%.*]] = add nsw i32 [[TMP28]], 1 12956 // CHECK11-NEXT: store i32 [[ADD9]], i32* [[DOTOMP_IV]], align 4, !llvm.access.group !66 12957 // CHECK11-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP67:![0-9]+]] 12958 // CHECK11: omp.inner.for.end: 12959 // CHECK11-NEXT: br label [[OMP_LOOP_EXIT:%.*]] 12960 // CHECK11: omp.loop.exit: 12961 // CHECK11-NEXT: [[TMP29:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 4 12962 // CHECK11-NEXT: [[TMP30:%.*]] = load i32, i32* [[TMP29]], align 4 12963 // CHECK11-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP30]]) 12964 // CHECK11-NEXT: [[TMP31:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4 12965 // CHECK11-NEXT: [[TMP32:%.*]] = icmp ne i32 [[TMP31]], 0 12966 // CHECK11-NEXT: br i1 [[TMP32]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]] 12967 // CHECK11: .omp.final.then: 12968 // CHECK11-NEXT: [[TMP33:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 12969 // CHECK11-NEXT: [[SUB10:%.*]] = sub nsw i32 [[TMP33]], 0 12970 // CHECK11-NEXT: [[DIV11:%.*]] = sdiv i32 [[SUB10]], 1 12971 // CHECK11-NEXT: [[MUL12:%.*]] = mul nsw i32 [[DIV11]], 1 12972 // CHECK11-NEXT: [[ADD13:%.*]] = add nsw i32 0, [[MUL12]] 12973 // CHECK11-NEXT: store i32 [[ADD13]], i32* [[I3]], align 4 12974 // CHECK11-NEXT: br label [[DOTOMP_FINAL_DONE]] 12975 // CHECK11: .omp.final.done: 12976 // CHECK11-NEXT: br label [[OMP_PRECOND_END]] 12977 // CHECK11: omp.precond.end: 12978 // CHECK11-NEXT: ret void 12979 // 12980 // 12981 // CHECK11-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiET_v_l50 12982 // CHECK11-SAME: (i32 noundef [[N:%.*]], i32* noundef [[A:%.*]], i32* noundef [[B:%.*]], i32* noundef [[C:%.*]]) #[[ATTR1]] { 12983 // CHECK11-NEXT: entry: 12984 // CHECK11-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4 12985 // CHECK11-NEXT: [[A_ADDR:%.*]] = alloca i32*, align 4 12986 // CHECK11-NEXT: [[B_ADDR:%.*]] = alloca i32*, align 4 12987 // CHECK11-NEXT: [[C_ADDR:%.*]] = alloca i32*, align 4 12988 // CHECK11-NEXT: store i32 [[N]], i32* [[N_ADDR]], align 4 12989 // CHECK11-NEXT: store i32* [[A]], i32** [[A_ADDR]], align 4 12990 // CHECK11-NEXT: store i32* [[B]], i32** [[B_ADDR]], align 4 12991 // CHECK11-NEXT: store i32* [[C]], i32** [[C_ADDR]], align 4 12992 // CHECK11-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 4, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32*, i32**, i32**, i32**)* @.omp_outlined..30 to void (i32*, i32*, ...)*), i32* [[N_ADDR]], i32** [[A_ADDR]], i32** [[B_ADDR]], i32** [[C_ADDR]]) 12993 // CHECK11-NEXT: ret void 12994 // 12995 // 12996 // CHECK11-LABEL: define {{[^@]+}}@.omp_outlined..30 12997 // CHECK11-SAME: (i32* noalias noundef [[DOTGLOBAL_TID_:%.*]], i32* noalias noundef [[DOTBOUND_TID_:%.*]], i32* noundef nonnull align 4 dereferenceable(4) [[N:%.*]], i32** noundef nonnull align 4 dereferenceable(4) [[A:%.*]], i32** noundef nonnull align 4 dereferenceable(4) [[B:%.*]], i32** noundef nonnull align 4 dereferenceable(4) [[C:%.*]]) #[[ATTR1]] { 12998 // CHECK11-NEXT: entry: 12999 // CHECK11-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4 13000 // CHECK11-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4 13001 // CHECK11-NEXT: [[N_ADDR:%.*]] = alloca i32*, align 4 13002 // CHECK11-NEXT: [[A_ADDR:%.*]] = alloca i32**, align 4 13003 // CHECK11-NEXT: [[B_ADDR:%.*]] = alloca i32**, align 4 13004 // CHECK11-NEXT: [[C_ADDR:%.*]] = alloca i32**, align 4 13005 // CHECK11-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 13006 // CHECK11-NEXT: [[TMP:%.*]] = alloca i32, align 4 13007 // CHECK11-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 13008 // CHECK11-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 13009 // CHECK11-NEXT: [[I:%.*]] = alloca i32, align 4 13010 // CHECK11-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4 13011 // CHECK11-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4 13012 // CHECK11-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 13013 // CHECK11-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 13014 // CHECK11-NEXT: [[I3:%.*]] = alloca i32, align 4 13015 // CHECK11-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 4 13016 // CHECK11-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 4 13017 // CHECK11-NEXT: store i32* [[N]], i32** [[N_ADDR]], align 4 13018 // CHECK11-NEXT: store i32** [[A]], i32*** [[A_ADDR]], align 4 13019 // CHECK11-NEXT: store i32** [[B]], i32*** [[B_ADDR]], align 4 13020 // CHECK11-NEXT: store i32** [[C]], i32*** [[C_ADDR]], align 4 13021 // CHECK11-NEXT: [[TMP0:%.*]] = load i32*, i32** [[N_ADDR]], align 4 13022 // CHECK11-NEXT: [[TMP1:%.*]] = load i32**, i32*** [[A_ADDR]], align 4 13023 // CHECK11-NEXT: [[TMP2:%.*]] = load i32**, i32*** [[B_ADDR]], align 4 13024 // CHECK11-NEXT: [[TMP3:%.*]] = load i32**, i32*** [[C_ADDR]], align 4 13025 // CHECK11-NEXT: [[TMP4:%.*]] = load i32, i32* [[TMP0]], align 4 13026 // CHECK11-NEXT: store i32 [[TMP4]], i32* [[DOTCAPTURE_EXPR_]], align 4 13027 // CHECK11-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 13028 // CHECK11-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP5]], 0 13029 // CHECK11-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 13030 // CHECK11-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1 13031 // CHECK11-NEXT: store i32 [[SUB2]], i32* [[DOTCAPTURE_EXPR_1]], align 4 13032 // CHECK11-NEXT: store i32 0, i32* [[I]], align 4 13033 // CHECK11-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 13034 // CHECK11-NEXT: [[CMP:%.*]] = icmp slt i32 0, [[TMP6]] 13035 // CHECK11-NEXT: br i1 [[CMP]], label [[OMP_PRECOND_THEN:%.*]], label [[OMP_PRECOND_END:%.*]] 13036 // CHECK11: omp.precond.then: 13037 // CHECK11-NEXT: store i32 0, i32* [[DOTOMP_COMB_LB]], align 4 13038 // CHECK11-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 13039 // CHECK11-NEXT: store i32 [[TMP7]], i32* [[DOTOMP_COMB_UB]], align 4 13040 // CHECK11-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4 13041 // CHECK11-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4 13042 // CHECK11-NEXT: [[TMP8:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 4 13043 // CHECK11-NEXT: [[TMP9:%.*]] = load i32, i32* [[TMP8]], align 4 13044 // CHECK11-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1]], i32 [[TMP9]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_COMB_LB]], i32* [[DOTOMP_COMB_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1) 13045 // CHECK11-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 13046 // CHECK11-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 13047 // CHECK11-NEXT: [[CMP4:%.*]] = icmp sgt i32 [[TMP10]], [[TMP11]] 13048 // CHECK11-NEXT: br i1 [[CMP4]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] 13049 // CHECK11: cond.true: 13050 // CHECK11-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 13051 // CHECK11-NEXT: br label [[COND_END:%.*]] 13052 // CHECK11: cond.false: 13053 // CHECK11-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 13054 // CHECK11-NEXT: br label [[COND_END]] 13055 // CHECK11: cond.end: 13056 // CHECK11-NEXT: [[COND:%.*]] = phi i32 [ [[TMP12]], [[COND_TRUE]] ], [ [[TMP13]], [[COND_FALSE]] ] 13057 // CHECK11-NEXT: store i32 [[COND]], i32* [[DOTOMP_COMB_UB]], align 4 13058 // CHECK11-NEXT: [[TMP14:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4 13059 // CHECK11-NEXT: store i32 [[TMP14]], i32* [[DOTOMP_IV]], align 4 13060 // CHECK11-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] 13061 // CHECK11: omp.inner.for.cond: 13062 // CHECK11-NEXT: [[TMP15:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !69 13063 // CHECK11-NEXT: [[TMP16:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4, !llvm.access.group !69 13064 // CHECK11-NEXT: [[CMP5:%.*]] = icmp sle i32 [[TMP15]], [[TMP16]] 13065 // CHECK11-NEXT: br i1 [[CMP5]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] 13066 // CHECK11: omp.inner.for.body: 13067 // CHECK11-NEXT: [[TMP17:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4, !llvm.access.group !69 13068 // CHECK11-NEXT: [[TMP18:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4, !llvm.access.group !69 13069 // CHECK11-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 6, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32, i32, i32*, i32**, i32**, i32**)* @.omp_outlined..31 to void (i32*, i32*, ...)*), i32 [[TMP17]], i32 [[TMP18]], i32* [[TMP0]], i32** [[TMP1]], i32** [[TMP2]], i32** [[TMP3]]), !llvm.access.group !69 13070 // CHECK11-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] 13071 // CHECK11: omp.inner.for.inc: 13072 // CHECK11-NEXT: [[TMP19:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !69 13073 // CHECK11-NEXT: [[TMP20:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4, !llvm.access.group !69 13074 // CHECK11-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP19]], [[TMP20]] 13075 // CHECK11-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4, !llvm.access.group !69 13076 // CHECK11-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP70:![0-9]+]] 13077 // CHECK11: omp.inner.for.end: 13078 // CHECK11-NEXT: br label [[OMP_LOOP_EXIT:%.*]] 13079 // CHECK11: omp.loop.exit: 13080 // CHECK11-NEXT: [[TMP21:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 4 13081 // CHECK11-NEXT: [[TMP22:%.*]] = load i32, i32* [[TMP21]], align 4 13082 // CHECK11-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP22]]) 13083 // CHECK11-NEXT: [[TMP23:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4 13084 // CHECK11-NEXT: [[TMP24:%.*]] = icmp ne i32 [[TMP23]], 0 13085 // CHECK11-NEXT: br i1 [[TMP24]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]] 13086 // CHECK11: .omp.final.then: 13087 // CHECK11-NEXT: [[TMP25:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 13088 // CHECK11-NEXT: [[SUB6:%.*]] = sub nsw i32 [[TMP25]], 0 13089 // CHECK11-NEXT: [[DIV7:%.*]] = sdiv i32 [[SUB6]], 1 13090 // CHECK11-NEXT: [[MUL:%.*]] = mul nsw i32 [[DIV7]], 1 13091 // CHECK11-NEXT: [[ADD8:%.*]] = add nsw i32 0, [[MUL]] 13092 // CHECK11-NEXT: store i32 [[ADD8]], i32* [[I3]], align 4 13093 // CHECK11-NEXT: br label [[DOTOMP_FINAL_DONE]] 13094 // CHECK11: .omp.final.done: 13095 // CHECK11-NEXT: br label [[OMP_PRECOND_END]] 13096 // CHECK11: omp.precond.end: 13097 // CHECK11-NEXT: ret void 13098 // 13099 // 13100 // CHECK11-LABEL: define {{[^@]+}}@.omp_outlined..31 13101 // CHECK11-SAME: (i32* noalias noundef [[DOTGLOBAL_TID_:%.*]], i32* noalias noundef [[DOTBOUND_TID_:%.*]], i32 noundef [[DOTPREVIOUS_LB_:%.*]], i32 noundef [[DOTPREVIOUS_UB_:%.*]], i32* noundef nonnull align 4 dereferenceable(4) [[N:%.*]], i32** noundef nonnull align 4 dereferenceable(4) [[A:%.*]], i32** noundef nonnull align 4 dereferenceable(4) [[B:%.*]], i32** noundef nonnull align 4 dereferenceable(4) [[C:%.*]]) #[[ATTR1]] { 13102 // CHECK11-NEXT: entry: 13103 // CHECK11-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4 13104 // CHECK11-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4 13105 // CHECK11-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i32, align 4 13106 // CHECK11-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i32, align 4 13107 // CHECK11-NEXT: [[N_ADDR:%.*]] = alloca i32*, align 4 13108 // CHECK11-NEXT: [[A_ADDR:%.*]] = alloca i32**, align 4 13109 // CHECK11-NEXT: [[B_ADDR:%.*]] = alloca i32**, align 4 13110 // CHECK11-NEXT: [[C_ADDR:%.*]] = alloca i32**, align 4 13111 // CHECK11-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 13112 // CHECK11-NEXT: [[TMP:%.*]] = alloca i32, align 4 13113 // CHECK11-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 13114 // CHECK11-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 13115 // CHECK11-NEXT: [[I:%.*]] = alloca i32, align 4 13116 // CHECK11-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 13117 // CHECK11-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4 13118 // CHECK11-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 13119 // CHECK11-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 13120 // CHECK11-NEXT: [[I3:%.*]] = alloca i32, align 4 13121 // CHECK11-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 4 13122 // CHECK11-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 4 13123 // CHECK11-NEXT: store i32 [[DOTPREVIOUS_LB_]], i32* [[DOTPREVIOUS_LB__ADDR]], align 4 13124 // CHECK11-NEXT: store i32 [[DOTPREVIOUS_UB_]], i32* [[DOTPREVIOUS_UB__ADDR]], align 4 13125 // CHECK11-NEXT: store i32* [[N]], i32** [[N_ADDR]], align 4 13126 // CHECK11-NEXT: store i32** [[A]], i32*** [[A_ADDR]], align 4 13127 // CHECK11-NEXT: store i32** [[B]], i32*** [[B_ADDR]], align 4 13128 // CHECK11-NEXT: store i32** [[C]], i32*** [[C_ADDR]], align 4 13129 // CHECK11-NEXT: [[TMP0:%.*]] = load i32*, i32** [[N_ADDR]], align 4 13130 // CHECK11-NEXT: [[TMP1:%.*]] = load i32**, i32*** [[A_ADDR]], align 4 13131 // CHECK11-NEXT: [[TMP2:%.*]] = load i32**, i32*** [[B_ADDR]], align 4 13132 // CHECK11-NEXT: [[TMP3:%.*]] = load i32**, i32*** [[C_ADDR]], align 4 13133 // CHECK11-NEXT: [[TMP4:%.*]] = load i32, i32* [[TMP0]], align 4 13134 // CHECK11-NEXT: store i32 [[TMP4]], i32* [[DOTCAPTURE_EXPR_]], align 4 13135 // CHECK11-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 13136 // CHECK11-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP5]], 0 13137 // CHECK11-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 13138 // CHECK11-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1 13139 // CHECK11-NEXT: store i32 [[SUB2]], i32* [[DOTCAPTURE_EXPR_1]], align 4 13140 // CHECK11-NEXT: store i32 0, i32* [[I]], align 4 13141 // CHECK11-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 13142 // CHECK11-NEXT: [[CMP:%.*]] = icmp slt i32 0, [[TMP6]] 13143 // CHECK11-NEXT: br i1 [[CMP]], label [[OMP_PRECOND_THEN:%.*]], label [[OMP_PRECOND_END:%.*]] 13144 // CHECK11: omp.precond.then: 13145 // CHECK11-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4 13146 // CHECK11-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 13147 // CHECK11-NEXT: store i32 [[TMP7]], i32* [[DOTOMP_UB]], align 4 13148 // CHECK11-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTPREVIOUS_LB__ADDR]], align 4 13149 // CHECK11-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTPREVIOUS_UB__ADDR]], align 4 13150 // CHECK11-NEXT: store i32 [[TMP8]], i32* [[DOTOMP_LB]], align 4 13151 // CHECK11-NEXT: store i32 [[TMP9]], i32* [[DOTOMP_UB]], align 4 13152 // CHECK11-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4 13153 // CHECK11-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4 13154 // CHECK11-NEXT: [[TMP10:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 4 13155 // CHECK11-NEXT: [[TMP11:%.*]] = load i32, i32* [[TMP10]], align 4 13156 // CHECK11-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB2]], i32 [[TMP11]], i32 34, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1) 13157 // CHECK11-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 13158 // CHECK11-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 13159 // CHECK11-NEXT: [[CMP4:%.*]] = icmp sgt i32 [[TMP12]], [[TMP13]] 13160 // CHECK11-NEXT: br i1 [[CMP4]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] 13161 // CHECK11: cond.true: 13162 // CHECK11-NEXT: [[TMP14:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 13163 // CHECK11-NEXT: br label [[COND_END:%.*]] 13164 // CHECK11: cond.false: 13165 // CHECK11-NEXT: [[TMP15:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 13166 // CHECK11-NEXT: br label [[COND_END]] 13167 // CHECK11: cond.end: 13168 // CHECK11-NEXT: [[COND:%.*]] = phi i32 [ [[TMP14]], [[COND_TRUE]] ], [ [[TMP15]], [[COND_FALSE]] ] 13169 // CHECK11-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4 13170 // CHECK11-NEXT: [[TMP16:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4 13171 // CHECK11-NEXT: store i32 [[TMP16]], i32* [[DOTOMP_IV]], align 4 13172 // CHECK11-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] 13173 // CHECK11: omp.inner.for.cond: 13174 // CHECK11-NEXT: [[TMP17:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !72 13175 // CHECK11-NEXT: [[TMP18:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4, !llvm.access.group !72 13176 // CHECK11-NEXT: [[CMP5:%.*]] = icmp sle i32 [[TMP17]], [[TMP18]] 13177 // CHECK11-NEXT: br i1 [[CMP5]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] 13178 // CHECK11: omp.inner.for.body: 13179 // CHECK11-NEXT: [[TMP19:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !72 13180 // CHECK11-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP19]], 1 13181 // CHECK11-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] 13182 // CHECK11-NEXT: store i32 [[ADD]], i32* [[I3]], align 4, !llvm.access.group !72 13183 // CHECK11-NEXT: [[TMP20:%.*]] = load i32*, i32** [[TMP2]], align 4, !llvm.access.group !72 13184 // CHECK11-NEXT: [[TMP21:%.*]] = load i32, i32* [[I3]], align 4, !llvm.access.group !72 13185 // CHECK11-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i32, i32* [[TMP20]], i32 [[TMP21]] 13186 // CHECK11-NEXT: [[TMP22:%.*]] = load i32, i32* [[ARRAYIDX]], align 4, !llvm.access.group !72 13187 // CHECK11-NEXT: [[TMP23:%.*]] = load i32*, i32** [[TMP3]], align 4, !llvm.access.group !72 13188 // CHECK11-NEXT: [[TMP24:%.*]] = load i32, i32* [[I3]], align 4, !llvm.access.group !72 13189 // CHECK11-NEXT: [[ARRAYIDX6:%.*]] = getelementptr inbounds i32, i32* [[TMP23]], i32 [[TMP24]] 13190 // CHECK11-NEXT: [[TMP25:%.*]] = load i32, i32* [[ARRAYIDX6]], align 4, !llvm.access.group !72 13191 // CHECK11-NEXT: [[ADD7:%.*]] = add nsw i32 [[TMP22]], [[TMP25]] 13192 // CHECK11-NEXT: [[TMP26:%.*]] = load i32*, i32** [[TMP1]], align 4, !llvm.access.group !72 13193 // CHECK11-NEXT: [[TMP27:%.*]] = load i32, i32* [[I3]], align 4, !llvm.access.group !72 13194 // CHECK11-NEXT: [[ARRAYIDX8:%.*]] = getelementptr inbounds i32, i32* [[TMP26]], i32 [[TMP27]] 13195 // CHECK11-NEXT: store i32 [[ADD7]], i32* [[ARRAYIDX8]], align 4, !llvm.access.group !72 13196 // CHECK11-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] 13197 // CHECK11: omp.body.continue: 13198 // CHECK11-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] 13199 // CHECK11: omp.inner.for.inc: 13200 // CHECK11-NEXT: [[TMP28:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !72 13201 // CHECK11-NEXT: [[ADD9:%.*]] = add nsw i32 [[TMP28]], 1 13202 // CHECK11-NEXT: store i32 [[ADD9]], i32* [[DOTOMP_IV]], align 4, !llvm.access.group !72 13203 // CHECK11-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP73:![0-9]+]] 13204 // CHECK11: omp.inner.for.end: 13205 // CHECK11-NEXT: br label [[OMP_LOOP_EXIT:%.*]] 13206 // CHECK11: omp.loop.exit: 13207 // CHECK11-NEXT: [[TMP29:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 4 13208 // CHECK11-NEXT: [[TMP30:%.*]] = load i32, i32* [[TMP29]], align 4 13209 // CHECK11-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP30]]) 13210 // CHECK11-NEXT: [[TMP31:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4 13211 // CHECK11-NEXT: [[TMP32:%.*]] = icmp ne i32 [[TMP31]], 0 13212 // CHECK11-NEXT: br i1 [[TMP32]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]] 13213 // CHECK11: .omp.final.then: 13214 // CHECK11-NEXT: [[TMP33:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 13215 // CHECK11-NEXT: [[SUB10:%.*]] = sub nsw i32 [[TMP33]], 0 13216 // CHECK11-NEXT: [[DIV11:%.*]] = sdiv i32 [[SUB10]], 1 13217 // CHECK11-NEXT: [[MUL12:%.*]] = mul nsw i32 [[DIV11]], 1 13218 // CHECK11-NEXT: [[ADD13:%.*]] = add nsw i32 0, [[MUL12]] 13219 // CHECK11-NEXT: store i32 [[ADD13]], i32* [[I3]], align 4 13220 // CHECK11-NEXT: br label [[DOTOMP_FINAL_DONE]] 13221 // CHECK11: .omp.final.done: 13222 // CHECK11-NEXT: br label [[OMP_PRECOND_END]] 13223 // CHECK11: omp.precond.end: 13224 // CHECK11-NEXT: ret void 13225 // 13226 // 13227 // CHECK11-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiET_v_l58 13228 // CHECK11-SAME: (i32 noundef [[CH:%.*]], i32 noundef [[N:%.*]], i32* noundef [[A:%.*]], i32* noundef [[B:%.*]], i32* noundef [[C:%.*]]) #[[ATTR1]] { 13229 // CHECK11-NEXT: entry: 13230 // CHECK11-NEXT: [[CH_ADDR:%.*]] = alloca i32, align 4 13231 // CHECK11-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4 13232 // CHECK11-NEXT: [[A_ADDR:%.*]] = alloca i32*, align 4 13233 // CHECK11-NEXT: [[B_ADDR:%.*]] = alloca i32*, align 4 13234 // CHECK11-NEXT: [[C_ADDR:%.*]] = alloca i32*, align 4 13235 // CHECK11-NEXT: store i32 [[CH]], i32* [[CH_ADDR]], align 4 13236 // CHECK11-NEXT: store i32 [[N]], i32* [[N_ADDR]], align 4 13237 // CHECK11-NEXT: store i32* [[A]], i32** [[A_ADDR]], align 4 13238 // CHECK11-NEXT: store i32* [[B]], i32** [[B_ADDR]], align 4 13239 // CHECK11-NEXT: store i32* [[C]], i32** [[C_ADDR]], align 4 13240 // CHECK11-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 5, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32*, i32*, i32**, i32**, i32**)* @.omp_outlined..34 to void (i32*, i32*, ...)*), i32* [[CH_ADDR]], i32* [[N_ADDR]], i32** [[A_ADDR]], i32** [[B_ADDR]], i32** [[C_ADDR]]) 13241 // CHECK11-NEXT: ret void 13242 // 13243 // 13244 // CHECK11-LABEL: define {{[^@]+}}@.omp_outlined..34 13245 // CHECK11-SAME: (i32* noalias noundef [[DOTGLOBAL_TID_:%.*]], i32* noalias noundef [[DOTBOUND_TID_:%.*]], i32* noundef nonnull align 4 dereferenceable(4) [[CH:%.*]], i32* noundef nonnull align 4 dereferenceable(4) [[N:%.*]], i32** noundef nonnull align 4 dereferenceable(4) [[A:%.*]], i32** noundef nonnull align 4 dereferenceable(4) [[B:%.*]], i32** noundef nonnull align 4 dereferenceable(4) [[C:%.*]]) #[[ATTR1]] { 13246 // CHECK11-NEXT: entry: 13247 // CHECK11-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4 13248 // CHECK11-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4 13249 // CHECK11-NEXT: [[CH_ADDR:%.*]] = alloca i32*, align 4 13250 // CHECK11-NEXT: [[N_ADDR:%.*]] = alloca i32*, align 4 13251 // CHECK11-NEXT: [[A_ADDR:%.*]] = alloca i32**, align 4 13252 // CHECK11-NEXT: [[B_ADDR:%.*]] = alloca i32**, align 4 13253 // CHECK11-NEXT: [[C_ADDR:%.*]] = alloca i32**, align 4 13254 // CHECK11-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 13255 // CHECK11-NEXT: [[TMP:%.*]] = alloca i32, align 4 13256 // CHECK11-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 13257 // CHECK11-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 13258 // CHECK11-NEXT: [[I:%.*]] = alloca i32, align 4 13259 // CHECK11-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4 13260 // CHECK11-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4 13261 // CHECK11-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 13262 // CHECK11-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 13263 // CHECK11-NEXT: [[I3:%.*]] = alloca i32, align 4 13264 // CHECK11-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 4 13265 // CHECK11-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 4 13266 // CHECK11-NEXT: store i32* [[CH]], i32** [[CH_ADDR]], align 4 13267 // CHECK11-NEXT: store i32* [[N]], i32** [[N_ADDR]], align 4 13268 // CHECK11-NEXT: store i32** [[A]], i32*** [[A_ADDR]], align 4 13269 // CHECK11-NEXT: store i32** [[B]], i32*** [[B_ADDR]], align 4 13270 // CHECK11-NEXT: store i32** [[C]], i32*** [[C_ADDR]], align 4 13271 // CHECK11-NEXT: [[TMP0:%.*]] = load i32*, i32** [[CH_ADDR]], align 4 13272 // CHECK11-NEXT: [[TMP1:%.*]] = load i32*, i32** [[N_ADDR]], align 4 13273 // CHECK11-NEXT: [[TMP2:%.*]] = load i32**, i32*** [[A_ADDR]], align 4 13274 // CHECK11-NEXT: [[TMP3:%.*]] = load i32**, i32*** [[B_ADDR]], align 4 13275 // CHECK11-NEXT: [[TMP4:%.*]] = load i32**, i32*** [[C_ADDR]], align 4 13276 // CHECK11-NEXT: [[TMP5:%.*]] = load i32, i32* [[TMP1]], align 4 13277 // CHECK11-NEXT: store i32 [[TMP5]], i32* [[DOTCAPTURE_EXPR_]], align 4 13278 // CHECK11-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 13279 // CHECK11-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP6]], 0 13280 // CHECK11-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 13281 // CHECK11-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1 13282 // CHECK11-NEXT: store i32 [[SUB2]], i32* [[DOTCAPTURE_EXPR_1]], align 4 13283 // CHECK11-NEXT: store i32 0, i32* [[I]], align 4 13284 // CHECK11-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 13285 // CHECK11-NEXT: [[CMP:%.*]] = icmp slt i32 0, [[TMP7]] 13286 // CHECK11-NEXT: br i1 [[CMP]], label [[OMP_PRECOND_THEN:%.*]], label [[OMP_PRECOND_END:%.*]] 13287 // CHECK11: omp.precond.then: 13288 // CHECK11-NEXT: store i32 0, i32* [[DOTOMP_COMB_LB]], align 4 13289 // CHECK11-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 13290 // CHECK11-NEXT: store i32 [[TMP8]], i32* [[DOTOMP_COMB_UB]], align 4 13291 // CHECK11-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4 13292 // CHECK11-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4 13293 // CHECK11-NEXT: [[TMP9:%.*]] = load i32, i32* [[TMP0]], align 4 13294 // CHECK11-NEXT: [[TMP10:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 4 13295 // CHECK11-NEXT: [[TMP11:%.*]] = load i32, i32* [[TMP10]], align 4 13296 // CHECK11-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1]], i32 [[TMP11]], i32 91, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_COMB_LB]], i32* [[DOTOMP_COMB_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 [[TMP9]]) 13297 // CHECK11-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 13298 // CHECK11-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 13299 // CHECK11-NEXT: [[CMP4:%.*]] = icmp sgt i32 [[TMP12]], [[TMP13]] 13300 // CHECK11-NEXT: br i1 [[CMP4]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] 13301 // CHECK11: cond.true: 13302 // CHECK11-NEXT: [[TMP14:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 13303 // CHECK11-NEXT: br label [[COND_END:%.*]] 13304 // CHECK11: cond.false: 13305 // CHECK11-NEXT: [[TMP15:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 13306 // CHECK11-NEXT: br label [[COND_END]] 13307 // CHECK11: cond.end: 13308 // CHECK11-NEXT: [[COND:%.*]] = phi i32 [ [[TMP14]], [[COND_TRUE]] ], [ [[TMP15]], [[COND_FALSE]] ] 13309 // CHECK11-NEXT: store i32 [[COND]], i32* [[DOTOMP_COMB_UB]], align 4 13310 // CHECK11-NEXT: [[TMP16:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4 13311 // CHECK11-NEXT: store i32 [[TMP16]], i32* [[DOTOMP_IV]], align 4 13312 // CHECK11-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] 13313 // CHECK11: omp.inner.for.cond: 13314 // CHECK11-NEXT: [[TMP17:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !75 13315 // CHECK11-NEXT: [[TMP18:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4, !llvm.access.group !75 13316 // CHECK11-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP18]], 1 13317 // CHECK11-NEXT: [[CMP5:%.*]] = icmp slt i32 [[TMP17]], [[ADD]] 13318 // CHECK11-NEXT: br i1 [[CMP5]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] 13319 // CHECK11: omp.inner.for.body: 13320 // CHECK11-NEXT: [[TMP19:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4, !llvm.access.group !75 13321 // CHECK11-NEXT: [[TMP20:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4, !llvm.access.group !75 13322 // CHECK11-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 6, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32, i32, i32*, i32**, i32**, i32**)* @.omp_outlined..35 to void (i32*, i32*, ...)*), i32 [[TMP19]], i32 [[TMP20]], i32* [[TMP1]], i32** [[TMP2]], i32** [[TMP3]], i32** [[TMP4]]), !llvm.access.group !75 13323 // CHECK11-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] 13324 // CHECK11: omp.inner.for.inc: 13325 // CHECK11-NEXT: [[TMP21:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !75 13326 // CHECK11-NEXT: [[TMP22:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4, !llvm.access.group !75 13327 // CHECK11-NEXT: [[ADD6:%.*]] = add nsw i32 [[TMP21]], [[TMP22]] 13328 // CHECK11-NEXT: store i32 [[ADD6]], i32* [[DOTOMP_IV]], align 4, !llvm.access.group !75 13329 // CHECK11-NEXT: [[TMP23:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4, !llvm.access.group !75 13330 // CHECK11-NEXT: [[TMP24:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4, !llvm.access.group !75 13331 // CHECK11-NEXT: [[ADD7:%.*]] = add nsw i32 [[TMP23]], [[TMP24]] 13332 // CHECK11-NEXT: store i32 [[ADD7]], i32* [[DOTOMP_COMB_LB]], align 4, !llvm.access.group !75 13333 // CHECK11-NEXT: [[TMP25:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4, !llvm.access.group !75 13334 // CHECK11-NEXT: [[TMP26:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4, !llvm.access.group !75 13335 // CHECK11-NEXT: [[ADD8:%.*]] = add nsw i32 [[TMP25]], [[TMP26]] 13336 // CHECK11-NEXT: store i32 [[ADD8]], i32* [[DOTOMP_COMB_UB]], align 4, !llvm.access.group !75 13337 // CHECK11-NEXT: [[TMP27:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4, !llvm.access.group !75 13338 // CHECK11-NEXT: [[TMP28:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4, !llvm.access.group !75 13339 // CHECK11-NEXT: [[CMP9:%.*]] = icmp sgt i32 [[TMP27]], [[TMP28]] 13340 // CHECK11-NEXT: br i1 [[CMP9]], label [[COND_TRUE10:%.*]], label [[COND_FALSE11:%.*]] 13341 // CHECK11: cond.true10: 13342 // CHECK11-NEXT: [[TMP29:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4, !llvm.access.group !75 13343 // CHECK11-NEXT: br label [[COND_END12:%.*]] 13344 // CHECK11: cond.false11: 13345 // CHECK11-NEXT: [[TMP30:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4, !llvm.access.group !75 13346 // CHECK11-NEXT: br label [[COND_END12]] 13347 // CHECK11: cond.end12: 13348 // CHECK11-NEXT: [[COND13:%.*]] = phi i32 [ [[TMP29]], [[COND_TRUE10]] ], [ [[TMP30]], [[COND_FALSE11]] ] 13349 // CHECK11-NEXT: store i32 [[COND13]], i32* [[DOTOMP_COMB_UB]], align 4, !llvm.access.group !75 13350 // CHECK11-NEXT: [[TMP31:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4, !llvm.access.group !75 13351 // CHECK11-NEXT: store i32 [[TMP31]], i32* [[DOTOMP_IV]], align 4, !llvm.access.group !75 13352 // CHECK11-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP76:![0-9]+]] 13353 // CHECK11: omp.inner.for.end: 13354 // CHECK11-NEXT: br label [[OMP_LOOP_EXIT:%.*]] 13355 // CHECK11: omp.loop.exit: 13356 // CHECK11-NEXT: [[TMP32:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 4 13357 // CHECK11-NEXT: [[TMP33:%.*]] = load i32, i32* [[TMP32]], align 4 13358 // CHECK11-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP33]]) 13359 // CHECK11-NEXT: [[TMP34:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4 13360 // CHECK11-NEXT: [[TMP35:%.*]] = icmp ne i32 [[TMP34]], 0 13361 // CHECK11-NEXT: br i1 [[TMP35]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]] 13362 // CHECK11: .omp.final.then: 13363 // CHECK11-NEXT: [[TMP36:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 13364 // CHECK11-NEXT: [[SUB14:%.*]] = sub nsw i32 [[TMP36]], 0 13365 // CHECK11-NEXT: [[DIV15:%.*]] = sdiv i32 [[SUB14]], 1 13366 // CHECK11-NEXT: [[MUL:%.*]] = mul nsw i32 [[DIV15]], 1 13367 // CHECK11-NEXT: [[ADD16:%.*]] = add nsw i32 0, [[MUL]] 13368 // CHECK11-NEXT: store i32 [[ADD16]], i32* [[I3]], align 4 13369 // CHECK11-NEXT: br label [[DOTOMP_FINAL_DONE]] 13370 // CHECK11: .omp.final.done: 13371 // CHECK11-NEXT: br label [[OMP_PRECOND_END]] 13372 // CHECK11: omp.precond.end: 13373 // CHECK11-NEXT: ret void 13374 // 13375 // 13376 // CHECK11-LABEL: define {{[^@]+}}@.omp_outlined..35 13377 // CHECK11-SAME: (i32* noalias noundef [[DOTGLOBAL_TID_:%.*]], i32* noalias noundef [[DOTBOUND_TID_:%.*]], i32 noundef [[DOTPREVIOUS_LB_:%.*]], i32 noundef [[DOTPREVIOUS_UB_:%.*]], i32* noundef nonnull align 4 dereferenceable(4) [[N:%.*]], i32** noundef nonnull align 4 dereferenceable(4) [[A:%.*]], i32** noundef nonnull align 4 dereferenceable(4) [[B:%.*]], i32** noundef nonnull align 4 dereferenceable(4) [[C:%.*]]) #[[ATTR1]] { 13378 // CHECK11-NEXT: entry: 13379 // CHECK11-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4 13380 // CHECK11-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4 13381 // CHECK11-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i32, align 4 13382 // CHECK11-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i32, align 4 13383 // CHECK11-NEXT: [[N_ADDR:%.*]] = alloca i32*, align 4 13384 // CHECK11-NEXT: [[A_ADDR:%.*]] = alloca i32**, align 4 13385 // CHECK11-NEXT: [[B_ADDR:%.*]] = alloca i32**, align 4 13386 // CHECK11-NEXT: [[C_ADDR:%.*]] = alloca i32**, align 4 13387 // CHECK11-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 13388 // CHECK11-NEXT: [[TMP:%.*]] = alloca i32, align 4 13389 // CHECK11-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 13390 // CHECK11-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 13391 // CHECK11-NEXT: [[I:%.*]] = alloca i32, align 4 13392 // CHECK11-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 13393 // CHECK11-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4 13394 // CHECK11-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 13395 // CHECK11-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 13396 // CHECK11-NEXT: [[I3:%.*]] = alloca i32, align 4 13397 // CHECK11-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 4 13398 // CHECK11-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 4 13399 // CHECK11-NEXT: store i32 [[DOTPREVIOUS_LB_]], i32* [[DOTPREVIOUS_LB__ADDR]], align 4 13400 // CHECK11-NEXT: store i32 [[DOTPREVIOUS_UB_]], i32* [[DOTPREVIOUS_UB__ADDR]], align 4 13401 // CHECK11-NEXT: store i32* [[N]], i32** [[N_ADDR]], align 4 13402 // CHECK11-NEXT: store i32** [[A]], i32*** [[A_ADDR]], align 4 13403 // CHECK11-NEXT: store i32** [[B]], i32*** [[B_ADDR]], align 4 13404 // CHECK11-NEXT: store i32** [[C]], i32*** [[C_ADDR]], align 4 13405 // CHECK11-NEXT: [[TMP0:%.*]] = load i32*, i32** [[N_ADDR]], align 4 13406 // CHECK11-NEXT: [[TMP1:%.*]] = load i32**, i32*** [[A_ADDR]], align 4 13407 // CHECK11-NEXT: [[TMP2:%.*]] = load i32**, i32*** [[B_ADDR]], align 4 13408 // CHECK11-NEXT: [[TMP3:%.*]] = load i32**, i32*** [[C_ADDR]], align 4 13409 // CHECK11-NEXT: [[TMP4:%.*]] = load i32, i32* [[TMP0]], align 4 13410 // CHECK11-NEXT: store i32 [[TMP4]], i32* [[DOTCAPTURE_EXPR_]], align 4 13411 // CHECK11-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 13412 // CHECK11-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP5]], 0 13413 // CHECK11-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 13414 // CHECK11-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1 13415 // CHECK11-NEXT: store i32 [[SUB2]], i32* [[DOTCAPTURE_EXPR_1]], align 4 13416 // CHECK11-NEXT: store i32 0, i32* [[I]], align 4 13417 // CHECK11-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 13418 // CHECK11-NEXT: [[CMP:%.*]] = icmp slt i32 0, [[TMP6]] 13419 // CHECK11-NEXT: br i1 [[CMP]], label [[OMP_PRECOND_THEN:%.*]], label [[OMP_PRECOND_END:%.*]] 13420 // CHECK11: omp.precond.then: 13421 // CHECK11-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4 13422 // CHECK11-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 13423 // CHECK11-NEXT: store i32 [[TMP7]], i32* [[DOTOMP_UB]], align 4 13424 // CHECK11-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTPREVIOUS_LB__ADDR]], align 4 13425 // CHECK11-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTPREVIOUS_UB__ADDR]], align 4 13426 // CHECK11-NEXT: store i32 [[TMP8]], i32* [[DOTOMP_LB]], align 4 13427 // CHECK11-NEXT: store i32 [[TMP9]], i32* [[DOTOMP_UB]], align 4 13428 // CHECK11-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4 13429 // CHECK11-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4 13430 // CHECK11-NEXT: [[TMP10:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 4 13431 // CHECK11-NEXT: [[TMP11:%.*]] = load i32, i32* [[TMP10]], align 4 13432 // CHECK11-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB2]], i32 [[TMP11]], i32 34, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1) 13433 // CHECK11-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 13434 // CHECK11-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 13435 // CHECK11-NEXT: [[CMP4:%.*]] = icmp sgt i32 [[TMP12]], [[TMP13]] 13436 // CHECK11-NEXT: br i1 [[CMP4]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] 13437 // CHECK11: cond.true: 13438 // CHECK11-NEXT: [[TMP14:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 13439 // CHECK11-NEXT: br label [[COND_END:%.*]] 13440 // CHECK11: cond.false: 13441 // CHECK11-NEXT: [[TMP15:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 13442 // CHECK11-NEXT: br label [[COND_END]] 13443 // CHECK11: cond.end: 13444 // CHECK11-NEXT: [[COND:%.*]] = phi i32 [ [[TMP14]], [[COND_TRUE]] ], [ [[TMP15]], [[COND_FALSE]] ] 13445 // CHECK11-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4 13446 // CHECK11-NEXT: [[TMP16:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4 13447 // CHECK11-NEXT: store i32 [[TMP16]], i32* [[DOTOMP_IV]], align 4 13448 // CHECK11-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] 13449 // CHECK11: omp.inner.for.cond: 13450 // CHECK11-NEXT: [[TMP17:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !78 13451 // CHECK11-NEXT: [[TMP18:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4, !llvm.access.group !78 13452 // CHECK11-NEXT: [[CMP5:%.*]] = icmp sle i32 [[TMP17]], [[TMP18]] 13453 // CHECK11-NEXT: br i1 [[CMP5]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] 13454 // CHECK11: omp.inner.for.body: 13455 // CHECK11-NEXT: [[TMP19:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !78 13456 // CHECK11-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP19]], 1 13457 // CHECK11-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] 13458 // CHECK11-NEXT: store i32 [[ADD]], i32* [[I3]], align 4, !llvm.access.group !78 13459 // CHECK11-NEXT: [[TMP20:%.*]] = load i32*, i32** [[TMP2]], align 4, !llvm.access.group !78 13460 // CHECK11-NEXT: [[TMP21:%.*]] = load i32, i32* [[I3]], align 4, !llvm.access.group !78 13461 // CHECK11-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i32, i32* [[TMP20]], i32 [[TMP21]] 13462 // CHECK11-NEXT: [[TMP22:%.*]] = load i32, i32* [[ARRAYIDX]], align 4, !llvm.access.group !78 13463 // CHECK11-NEXT: [[TMP23:%.*]] = load i32*, i32** [[TMP3]], align 4, !llvm.access.group !78 13464 // CHECK11-NEXT: [[TMP24:%.*]] = load i32, i32* [[I3]], align 4, !llvm.access.group !78 13465 // CHECK11-NEXT: [[ARRAYIDX6:%.*]] = getelementptr inbounds i32, i32* [[TMP23]], i32 [[TMP24]] 13466 // CHECK11-NEXT: [[TMP25:%.*]] = load i32, i32* [[ARRAYIDX6]], align 4, !llvm.access.group !78 13467 // CHECK11-NEXT: [[ADD7:%.*]] = add nsw i32 [[TMP22]], [[TMP25]] 13468 // CHECK11-NEXT: [[TMP26:%.*]] = load i32*, i32** [[TMP1]], align 4, !llvm.access.group !78 13469 // CHECK11-NEXT: [[TMP27:%.*]] = load i32, i32* [[I3]], align 4, !llvm.access.group !78 13470 // CHECK11-NEXT: [[ARRAYIDX8:%.*]] = getelementptr inbounds i32, i32* [[TMP26]], i32 [[TMP27]] 13471 // CHECK11-NEXT: store i32 [[ADD7]], i32* [[ARRAYIDX8]], align 4, !llvm.access.group !78 13472 // CHECK11-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] 13473 // CHECK11: omp.body.continue: 13474 // CHECK11-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] 13475 // CHECK11: omp.inner.for.inc: 13476 // CHECK11-NEXT: [[TMP28:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !78 13477 // CHECK11-NEXT: [[ADD9:%.*]] = add nsw i32 [[TMP28]], 1 13478 // CHECK11-NEXT: store i32 [[ADD9]], i32* [[DOTOMP_IV]], align 4, !llvm.access.group !78 13479 // CHECK11-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP79:![0-9]+]] 13480 // CHECK11: omp.inner.for.end: 13481 // CHECK11-NEXT: br label [[OMP_LOOP_EXIT:%.*]] 13482 // CHECK11: omp.loop.exit: 13483 // CHECK11-NEXT: [[TMP29:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 4 13484 // CHECK11-NEXT: [[TMP30:%.*]] = load i32, i32* [[TMP29]], align 4 13485 // CHECK11-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP30]]) 13486 // CHECK11-NEXT: [[TMP31:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4 13487 // CHECK11-NEXT: [[TMP32:%.*]] = icmp ne i32 [[TMP31]], 0 13488 // CHECK11-NEXT: br i1 [[TMP32]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]] 13489 // CHECK11: .omp.final.then: 13490 // CHECK11-NEXT: [[TMP33:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 13491 // CHECK11-NEXT: [[SUB10:%.*]] = sub nsw i32 [[TMP33]], 0 13492 // CHECK11-NEXT: [[DIV11:%.*]] = sdiv i32 [[SUB10]], 1 13493 // CHECK11-NEXT: [[MUL12:%.*]] = mul nsw i32 [[DIV11]], 1 13494 // CHECK11-NEXT: [[ADD13:%.*]] = add nsw i32 0, [[MUL12]] 13495 // CHECK11-NEXT: store i32 [[ADD13]], i32* [[I3]], align 4 13496 // CHECK11-NEXT: br label [[DOTOMP_FINAL_DONE]] 13497 // CHECK11: .omp.final.done: 13498 // CHECK11-NEXT: br label [[OMP_PRECOND_END]] 13499 // CHECK11: omp.precond.end: 13500 // CHECK11-NEXT: ret void 13501 // 13502 // 13503 // CHECK11-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiET_v_l66 13504 // CHECK11-SAME: (i32 noundef [[N:%.*]], i32* noundef [[A:%.*]], i32* noundef [[B:%.*]], i32* noundef [[C:%.*]]) #[[ATTR1]] { 13505 // CHECK11-NEXT: entry: 13506 // CHECK11-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4 13507 // CHECK11-NEXT: [[A_ADDR:%.*]] = alloca i32*, align 4 13508 // CHECK11-NEXT: [[B_ADDR:%.*]] = alloca i32*, align 4 13509 // CHECK11-NEXT: [[C_ADDR:%.*]] = alloca i32*, align 4 13510 // CHECK11-NEXT: store i32 [[N]], i32* [[N_ADDR]], align 4 13511 // CHECK11-NEXT: store i32* [[A]], i32** [[A_ADDR]], align 4 13512 // CHECK11-NEXT: store i32* [[B]], i32** [[B_ADDR]], align 4 13513 // CHECK11-NEXT: store i32* [[C]], i32** [[C_ADDR]], align 4 13514 // CHECK11-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 4, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32*, i32**, i32**, i32**)* @.omp_outlined..38 to void (i32*, i32*, ...)*), i32* [[N_ADDR]], i32** [[A_ADDR]], i32** [[B_ADDR]], i32** [[C_ADDR]]) 13515 // CHECK11-NEXT: ret void 13516 // 13517 // 13518 // CHECK11-LABEL: define {{[^@]+}}@.omp_outlined..38 13519 // CHECK11-SAME: (i32* noalias noundef [[DOTGLOBAL_TID_:%.*]], i32* noalias noundef [[DOTBOUND_TID_:%.*]], i32* noundef nonnull align 4 dereferenceable(4) [[N:%.*]], i32** noundef nonnull align 4 dereferenceable(4) [[A:%.*]], i32** noundef nonnull align 4 dereferenceable(4) [[B:%.*]], i32** noundef nonnull align 4 dereferenceable(4) [[C:%.*]]) #[[ATTR1]] { 13520 // CHECK11-NEXT: entry: 13521 // CHECK11-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4 13522 // CHECK11-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4 13523 // CHECK11-NEXT: [[N_ADDR:%.*]] = alloca i32*, align 4 13524 // CHECK11-NEXT: [[A_ADDR:%.*]] = alloca i32**, align 4 13525 // CHECK11-NEXT: [[B_ADDR:%.*]] = alloca i32**, align 4 13526 // CHECK11-NEXT: [[C_ADDR:%.*]] = alloca i32**, align 4 13527 // CHECK11-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 13528 // CHECK11-NEXT: [[TMP:%.*]] = alloca i32, align 4 13529 // CHECK11-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 13530 // CHECK11-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 13531 // CHECK11-NEXT: [[I:%.*]] = alloca i32, align 4 13532 // CHECK11-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4 13533 // CHECK11-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4 13534 // CHECK11-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 13535 // CHECK11-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 13536 // CHECK11-NEXT: [[I3:%.*]] = alloca i32, align 4 13537 // CHECK11-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 4 13538 // CHECK11-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 4 13539 // CHECK11-NEXT: store i32* [[N]], i32** [[N_ADDR]], align 4 13540 // CHECK11-NEXT: store i32** [[A]], i32*** [[A_ADDR]], align 4 13541 // CHECK11-NEXT: store i32** [[B]], i32*** [[B_ADDR]], align 4 13542 // CHECK11-NEXT: store i32** [[C]], i32*** [[C_ADDR]], align 4 13543 // CHECK11-NEXT: [[TMP0:%.*]] = load i32*, i32** [[N_ADDR]], align 4 13544 // CHECK11-NEXT: [[TMP1:%.*]] = load i32**, i32*** [[A_ADDR]], align 4 13545 // CHECK11-NEXT: [[TMP2:%.*]] = load i32**, i32*** [[B_ADDR]], align 4 13546 // CHECK11-NEXT: [[TMP3:%.*]] = load i32**, i32*** [[C_ADDR]], align 4 13547 // CHECK11-NEXT: [[TMP4:%.*]] = load i32, i32* [[TMP0]], align 4 13548 // CHECK11-NEXT: store i32 [[TMP4]], i32* [[DOTCAPTURE_EXPR_]], align 4 13549 // CHECK11-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 13550 // CHECK11-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP5]], 0 13551 // CHECK11-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 13552 // CHECK11-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1 13553 // CHECK11-NEXT: store i32 [[SUB2]], i32* [[DOTCAPTURE_EXPR_1]], align 4 13554 // CHECK11-NEXT: store i32 0, i32* [[I]], align 4 13555 // CHECK11-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 13556 // CHECK11-NEXT: [[CMP:%.*]] = icmp slt i32 0, [[TMP6]] 13557 // CHECK11-NEXT: br i1 [[CMP]], label [[OMP_PRECOND_THEN:%.*]], label [[OMP_PRECOND_END:%.*]] 13558 // CHECK11: omp.precond.then: 13559 // CHECK11-NEXT: store i32 0, i32* [[DOTOMP_COMB_LB]], align 4 13560 // CHECK11-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 13561 // CHECK11-NEXT: store i32 [[TMP7]], i32* [[DOTOMP_COMB_UB]], align 4 13562 // CHECK11-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4 13563 // CHECK11-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4 13564 // CHECK11-NEXT: [[TMP8:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 4 13565 // CHECK11-NEXT: [[TMP9:%.*]] = load i32, i32* [[TMP8]], align 4 13566 // CHECK11-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1]], i32 [[TMP9]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_COMB_LB]], i32* [[DOTOMP_COMB_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1) 13567 // CHECK11-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 13568 // CHECK11-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 13569 // CHECK11-NEXT: [[CMP4:%.*]] = icmp sgt i32 [[TMP10]], [[TMP11]] 13570 // CHECK11-NEXT: br i1 [[CMP4]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] 13571 // CHECK11: cond.true: 13572 // CHECK11-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 13573 // CHECK11-NEXT: br label [[COND_END:%.*]] 13574 // CHECK11: cond.false: 13575 // CHECK11-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 13576 // CHECK11-NEXT: br label [[COND_END]] 13577 // CHECK11: cond.end: 13578 // CHECK11-NEXT: [[COND:%.*]] = phi i32 [ [[TMP12]], [[COND_TRUE]] ], [ [[TMP13]], [[COND_FALSE]] ] 13579 // CHECK11-NEXT: store i32 [[COND]], i32* [[DOTOMP_COMB_UB]], align 4 13580 // CHECK11-NEXT: [[TMP14:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4 13581 // CHECK11-NEXT: store i32 [[TMP14]], i32* [[DOTOMP_IV]], align 4 13582 // CHECK11-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] 13583 // CHECK11: omp.inner.for.cond: 13584 // CHECK11-NEXT: [[TMP15:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !81 13585 // CHECK11-NEXT: [[TMP16:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4, !llvm.access.group !81 13586 // CHECK11-NEXT: [[CMP5:%.*]] = icmp sle i32 [[TMP15]], [[TMP16]] 13587 // CHECK11-NEXT: br i1 [[CMP5]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] 13588 // CHECK11: omp.inner.for.body: 13589 // CHECK11-NEXT: [[TMP17:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4, !llvm.access.group !81 13590 // CHECK11-NEXT: [[TMP18:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4, !llvm.access.group !81 13591 // CHECK11-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 6, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32, i32, i32*, i32**, i32**, i32**)* @.omp_outlined..39 to void (i32*, i32*, ...)*), i32 [[TMP17]], i32 [[TMP18]], i32* [[TMP0]], i32** [[TMP1]], i32** [[TMP2]], i32** [[TMP3]]), !llvm.access.group !81 13592 // CHECK11-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] 13593 // CHECK11: omp.inner.for.inc: 13594 // CHECK11-NEXT: [[TMP19:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !81 13595 // CHECK11-NEXT: [[TMP20:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4, !llvm.access.group !81 13596 // CHECK11-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP19]], [[TMP20]] 13597 // CHECK11-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4, !llvm.access.group !81 13598 // CHECK11-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP82:![0-9]+]] 13599 // CHECK11: omp.inner.for.end: 13600 // CHECK11-NEXT: br label [[OMP_LOOP_EXIT:%.*]] 13601 // CHECK11: omp.loop.exit: 13602 // CHECK11-NEXT: [[TMP21:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 4 13603 // CHECK11-NEXT: [[TMP22:%.*]] = load i32, i32* [[TMP21]], align 4 13604 // CHECK11-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP22]]) 13605 // CHECK11-NEXT: [[TMP23:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4 13606 // CHECK11-NEXT: [[TMP24:%.*]] = icmp ne i32 [[TMP23]], 0 13607 // CHECK11-NEXT: br i1 [[TMP24]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]] 13608 // CHECK11: .omp.final.then: 13609 // CHECK11-NEXT: [[TMP25:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 13610 // CHECK11-NEXT: [[SUB6:%.*]] = sub nsw i32 [[TMP25]], 0 13611 // CHECK11-NEXT: [[DIV7:%.*]] = sdiv i32 [[SUB6]], 1 13612 // CHECK11-NEXT: [[MUL:%.*]] = mul nsw i32 [[DIV7]], 1 13613 // CHECK11-NEXT: [[ADD8:%.*]] = add nsw i32 0, [[MUL]] 13614 // CHECK11-NEXT: store i32 [[ADD8]], i32* [[I3]], align 4 13615 // CHECK11-NEXT: br label [[DOTOMP_FINAL_DONE]] 13616 // CHECK11: .omp.final.done: 13617 // CHECK11-NEXT: br label [[OMP_PRECOND_END]] 13618 // CHECK11: omp.precond.end: 13619 // CHECK11-NEXT: ret void 13620 // 13621 // 13622 // CHECK11-LABEL: define {{[^@]+}}@.omp_outlined..39 13623 // CHECK11-SAME: (i32* noalias noundef [[DOTGLOBAL_TID_:%.*]], i32* noalias noundef [[DOTBOUND_TID_:%.*]], i32 noundef [[DOTPREVIOUS_LB_:%.*]], i32 noundef [[DOTPREVIOUS_UB_:%.*]], i32* noundef nonnull align 4 dereferenceable(4) [[N:%.*]], i32** noundef nonnull align 4 dereferenceable(4) [[A:%.*]], i32** noundef nonnull align 4 dereferenceable(4) [[B:%.*]], i32** noundef nonnull align 4 dereferenceable(4) [[C:%.*]]) #[[ATTR1]] { 13624 // CHECK11-NEXT: entry: 13625 // CHECK11-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4 13626 // CHECK11-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4 13627 // CHECK11-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i32, align 4 13628 // CHECK11-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i32, align 4 13629 // CHECK11-NEXT: [[N_ADDR:%.*]] = alloca i32*, align 4 13630 // CHECK11-NEXT: [[A_ADDR:%.*]] = alloca i32**, align 4 13631 // CHECK11-NEXT: [[B_ADDR:%.*]] = alloca i32**, align 4 13632 // CHECK11-NEXT: [[C_ADDR:%.*]] = alloca i32**, align 4 13633 // CHECK11-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 13634 // CHECK11-NEXT: [[TMP:%.*]] = alloca i32, align 4 13635 // CHECK11-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 13636 // CHECK11-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 13637 // CHECK11-NEXT: [[I:%.*]] = alloca i32, align 4 13638 // CHECK11-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 13639 // CHECK11-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4 13640 // CHECK11-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 13641 // CHECK11-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 13642 // CHECK11-NEXT: [[I3:%.*]] = alloca i32, align 4 13643 // CHECK11-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 4 13644 // CHECK11-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 4 13645 // CHECK11-NEXT: store i32 [[DOTPREVIOUS_LB_]], i32* [[DOTPREVIOUS_LB__ADDR]], align 4 13646 // CHECK11-NEXT: store i32 [[DOTPREVIOUS_UB_]], i32* [[DOTPREVIOUS_UB__ADDR]], align 4 13647 // CHECK11-NEXT: store i32* [[N]], i32** [[N_ADDR]], align 4 13648 // CHECK11-NEXT: store i32** [[A]], i32*** [[A_ADDR]], align 4 13649 // CHECK11-NEXT: store i32** [[B]], i32*** [[B_ADDR]], align 4 13650 // CHECK11-NEXT: store i32** [[C]], i32*** [[C_ADDR]], align 4 13651 // CHECK11-NEXT: [[TMP0:%.*]] = load i32*, i32** [[N_ADDR]], align 4 13652 // CHECK11-NEXT: [[TMP1:%.*]] = load i32**, i32*** [[A_ADDR]], align 4 13653 // CHECK11-NEXT: [[TMP2:%.*]] = load i32**, i32*** [[B_ADDR]], align 4 13654 // CHECK11-NEXT: [[TMP3:%.*]] = load i32**, i32*** [[C_ADDR]], align 4 13655 // CHECK11-NEXT: [[TMP4:%.*]] = load i32, i32* [[TMP0]], align 4 13656 // CHECK11-NEXT: store i32 [[TMP4]], i32* [[DOTCAPTURE_EXPR_]], align 4 13657 // CHECK11-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 13658 // CHECK11-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP5]], 0 13659 // CHECK11-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 13660 // CHECK11-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1 13661 // CHECK11-NEXT: store i32 [[SUB2]], i32* [[DOTCAPTURE_EXPR_1]], align 4 13662 // CHECK11-NEXT: store i32 0, i32* [[I]], align 4 13663 // CHECK11-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 13664 // CHECK11-NEXT: [[CMP:%.*]] = icmp slt i32 0, [[TMP6]] 13665 // CHECK11-NEXT: br i1 [[CMP]], label [[OMP_PRECOND_THEN:%.*]], label [[OMP_PRECOND_END:%.*]] 13666 // CHECK11: omp.precond.then: 13667 // CHECK11-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4 13668 // CHECK11-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 13669 // CHECK11-NEXT: store i32 [[TMP7]], i32* [[DOTOMP_UB]], align 4 13670 // CHECK11-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTPREVIOUS_LB__ADDR]], align 4 13671 // CHECK11-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTPREVIOUS_UB__ADDR]], align 4 13672 // CHECK11-NEXT: store i32 [[TMP8]], i32* [[DOTOMP_LB]], align 4 13673 // CHECK11-NEXT: store i32 [[TMP9]], i32* [[DOTOMP_UB]], align 4 13674 // CHECK11-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4 13675 // CHECK11-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4 13676 // CHECK11-NEXT: [[TMP10:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 4 13677 // CHECK11-NEXT: [[TMP11:%.*]] = load i32, i32* [[TMP10]], align 4 13678 // CHECK11-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB2]], i32 [[TMP11]], i32 34, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1) 13679 // CHECK11-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 13680 // CHECK11-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 13681 // CHECK11-NEXT: [[CMP4:%.*]] = icmp sgt i32 [[TMP12]], [[TMP13]] 13682 // CHECK11-NEXT: br i1 [[CMP4]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] 13683 // CHECK11: cond.true: 13684 // CHECK11-NEXT: [[TMP14:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 13685 // CHECK11-NEXT: br label [[COND_END:%.*]] 13686 // CHECK11: cond.false: 13687 // CHECK11-NEXT: [[TMP15:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 13688 // CHECK11-NEXT: br label [[COND_END]] 13689 // CHECK11: cond.end: 13690 // CHECK11-NEXT: [[COND:%.*]] = phi i32 [ [[TMP14]], [[COND_TRUE]] ], [ [[TMP15]], [[COND_FALSE]] ] 13691 // CHECK11-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4 13692 // CHECK11-NEXT: [[TMP16:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4 13693 // CHECK11-NEXT: store i32 [[TMP16]], i32* [[DOTOMP_IV]], align 4 13694 // CHECK11-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] 13695 // CHECK11: omp.inner.for.cond: 13696 // CHECK11-NEXT: [[TMP17:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !84 13697 // CHECK11-NEXT: [[TMP18:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4, !llvm.access.group !84 13698 // CHECK11-NEXT: [[CMP5:%.*]] = icmp sle i32 [[TMP17]], [[TMP18]] 13699 // CHECK11-NEXT: br i1 [[CMP5]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] 13700 // CHECK11: omp.inner.for.body: 13701 // CHECK11-NEXT: [[TMP19:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !84 13702 // CHECK11-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP19]], 1 13703 // CHECK11-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] 13704 // CHECK11-NEXT: store i32 [[ADD]], i32* [[I3]], align 4, !llvm.access.group !84 13705 // CHECK11-NEXT: [[TMP20:%.*]] = load i32*, i32** [[TMP2]], align 4, !llvm.access.group !84 13706 // CHECK11-NEXT: [[TMP21:%.*]] = load i32, i32* [[I3]], align 4, !llvm.access.group !84 13707 // CHECK11-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i32, i32* [[TMP20]], i32 [[TMP21]] 13708 // CHECK11-NEXT: [[TMP22:%.*]] = load i32, i32* [[ARRAYIDX]], align 4, !llvm.access.group !84 13709 // CHECK11-NEXT: [[TMP23:%.*]] = load i32*, i32** [[TMP3]], align 4, !llvm.access.group !84 13710 // CHECK11-NEXT: [[TMP24:%.*]] = load i32, i32* [[I3]], align 4, !llvm.access.group !84 13711 // CHECK11-NEXT: [[ARRAYIDX6:%.*]] = getelementptr inbounds i32, i32* [[TMP23]], i32 [[TMP24]] 13712 // CHECK11-NEXT: [[TMP25:%.*]] = load i32, i32* [[ARRAYIDX6]], align 4, !llvm.access.group !84 13713 // CHECK11-NEXT: [[ADD7:%.*]] = add nsw i32 [[TMP22]], [[TMP25]] 13714 // CHECK11-NEXT: [[TMP26:%.*]] = load i32*, i32** [[TMP1]], align 4, !llvm.access.group !84 13715 // CHECK11-NEXT: [[TMP27:%.*]] = load i32, i32* [[I3]], align 4, !llvm.access.group !84 13716 // CHECK11-NEXT: [[ARRAYIDX8:%.*]] = getelementptr inbounds i32, i32* [[TMP26]], i32 [[TMP27]] 13717 // CHECK11-NEXT: store i32 [[ADD7]], i32* [[ARRAYIDX8]], align 4, !llvm.access.group !84 13718 // CHECK11-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] 13719 // CHECK11: omp.body.continue: 13720 // CHECK11-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] 13721 // CHECK11: omp.inner.for.inc: 13722 // CHECK11-NEXT: [[TMP28:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !84 13723 // CHECK11-NEXT: [[ADD9:%.*]] = add nsw i32 [[TMP28]], 1 13724 // CHECK11-NEXT: store i32 [[ADD9]], i32* [[DOTOMP_IV]], align 4, !llvm.access.group !84 13725 // CHECK11-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP85:![0-9]+]] 13726 // CHECK11: omp.inner.for.end: 13727 // CHECK11-NEXT: br label [[OMP_LOOP_EXIT:%.*]] 13728 // CHECK11: omp.loop.exit: 13729 // CHECK11-NEXT: [[TMP29:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 4 13730 // CHECK11-NEXT: [[TMP30:%.*]] = load i32, i32* [[TMP29]], align 4 13731 // CHECK11-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP30]]) 13732 // CHECK11-NEXT: [[TMP31:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4 13733 // CHECK11-NEXT: [[TMP32:%.*]] = icmp ne i32 [[TMP31]], 0 13734 // CHECK11-NEXT: br i1 [[TMP32]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]] 13735 // CHECK11: .omp.final.then: 13736 // CHECK11-NEXT: [[TMP33:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 13737 // CHECK11-NEXT: [[SUB10:%.*]] = sub nsw i32 [[TMP33]], 0 13738 // CHECK11-NEXT: [[DIV11:%.*]] = sdiv i32 [[SUB10]], 1 13739 // CHECK11-NEXT: [[MUL12:%.*]] = mul nsw i32 [[DIV11]], 1 13740 // CHECK11-NEXT: [[ADD13:%.*]] = add nsw i32 0, [[MUL12]] 13741 // CHECK11-NEXT: store i32 [[ADD13]], i32* [[I3]], align 4 13742 // CHECK11-NEXT: br label [[DOTOMP_FINAL_DONE]] 13743 // CHECK11: .omp.final.done: 13744 // CHECK11-NEXT: br label [[OMP_PRECOND_END]] 13745 // CHECK11: omp.precond.end: 13746 // CHECK11-NEXT: ret void 13747 // 13748 // 13749 // CHECK11-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiET_v_l74 13750 // CHECK11-SAME: (i32 noundef [[CH:%.*]], i32 noundef [[N:%.*]], i32* noundef [[A:%.*]], i32* noundef [[B:%.*]], i32* noundef [[C:%.*]]) #[[ATTR1]] { 13751 // CHECK11-NEXT: entry: 13752 // CHECK11-NEXT: [[CH_ADDR:%.*]] = alloca i32, align 4 13753 // CHECK11-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4 13754 // CHECK11-NEXT: [[A_ADDR:%.*]] = alloca i32*, align 4 13755 // CHECK11-NEXT: [[B_ADDR:%.*]] = alloca i32*, align 4 13756 // CHECK11-NEXT: [[C_ADDR:%.*]] = alloca i32*, align 4 13757 // CHECK11-NEXT: store i32 [[CH]], i32* [[CH_ADDR]], align 4 13758 // CHECK11-NEXT: store i32 [[N]], i32* [[N_ADDR]], align 4 13759 // CHECK11-NEXT: store i32* [[A]], i32** [[A_ADDR]], align 4 13760 // CHECK11-NEXT: store i32* [[B]], i32** [[B_ADDR]], align 4 13761 // CHECK11-NEXT: store i32* [[C]], i32** [[C_ADDR]], align 4 13762 // CHECK11-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 5, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32*, i32*, i32**, i32**, i32**)* @.omp_outlined..42 to void (i32*, i32*, ...)*), i32* [[CH_ADDR]], i32* [[N_ADDR]], i32** [[A_ADDR]], i32** [[B_ADDR]], i32** [[C_ADDR]]) 13763 // CHECK11-NEXT: ret void 13764 // 13765 // 13766 // CHECK11-LABEL: define {{[^@]+}}@.omp_outlined..42 13767 // CHECK11-SAME: (i32* noalias noundef [[DOTGLOBAL_TID_:%.*]], i32* noalias noundef [[DOTBOUND_TID_:%.*]], i32* noundef nonnull align 4 dereferenceable(4) [[CH:%.*]], i32* noundef nonnull align 4 dereferenceable(4) [[N:%.*]], i32** noundef nonnull align 4 dereferenceable(4) [[A:%.*]], i32** noundef nonnull align 4 dereferenceable(4) [[B:%.*]], i32** noundef nonnull align 4 dereferenceable(4) [[C:%.*]]) #[[ATTR1]] { 13768 // CHECK11-NEXT: entry: 13769 // CHECK11-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4 13770 // CHECK11-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4 13771 // CHECK11-NEXT: [[CH_ADDR:%.*]] = alloca i32*, align 4 13772 // CHECK11-NEXT: [[N_ADDR:%.*]] = alloca i32*, align 4 13773 // CHECK11-NEXT: [[A_ADDR:%.*]] = alloca i32**, align 4 13774 // CHECK11-NEXT: [[B_ADDR:%.*]] = alloca i32**, align 4 13775 // CHECK11-NEXT: [[C_ADDR:%.*]] = alloca i32**, align 4 13776 // CHECK11-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 13777 // CHECK11-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 13778 // CHECK11-NEXT: [[TMP:%.*]] = alloca i32, align 4 13779 // CHECK11-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 13780 // CHECK11-NEXT: [[DOTCAPTURE_EXPR_2:%.*]] = alloca i32, align 4 13781 // CHECK11-NEXT: [[I:%.*]] = alloca i32, align 4 13782 // CHECK11-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4 13783 // CHECK11-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4 13784 // CHECK11-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 13785 // CHECK11-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 13786 // CHECK11-NEXT: [[I4:%.*]] = alloca i32, align 4 13787 // CHECK11-NEXT: [[DOTCAPTURE_EXPR__CASTED:%.*]] = alloca i32, align 4 13788 // CHECK11-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 4 13789 // CHECK11-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 4 13790 // CHECK11-NEXT: store i32* [[CH]], i32** [[CH_ADDR]], align 4 13791 // CHECK11-NEXT: store i32* [[N]], i32** [[N_ADDR]], align 4 13792 // CHECK11-NEXT: store i32** [[A]], i32*** [[A_ADDR]], align 4 13793 // CHECK11-NEXT: store i32** [[B]], i32*** [[B_ADDR]], align 4 13794 // CHECK11-NEXT: store i32** [[C]], i32*** [[C_ADDR]], align 4 13795 // CHECK11-NEXT: [[TMP0:%.*]] = load i32*, i32** [[CH_ADDR]], align 4 13796 // CHECK11-NEXT: [[TMP1:%.*]] = load i32*, i32** [[N_ADDR]], align 4 13797 // CHECK11-NEXT: [[TMP2:%.*]] = load i32**, i32*** [[A_ADDR]], align 4 13798 // CHECK11-NEXT: [[TMP3:%.*]] = load i32**, i32*** [[B_ADDR]], align 4 13799 // CHECK11-NEXT: [[TMP4:%.*]] = load i32**, i32*** [[C_ADDR]], align 4 13800 // CHECK11-NEXT: [[TMP5:%.*]] = load i32, i32* [[TMP0]], align 4 13801 // CHECK11-NEXT: store i32 [[TMP5]], i32* [[DOTCAPTURE_EXPR_]], align 4 13802 // CHECK11-NEXT: [[TMP6:%.*]] = load i32, i32* [[TMP1]], align 4 13803 // CHECK11-NEXT: store i32 [[TMP6]], i32* [[DOTCAPTURE_EXPR_1]], align 4 13804 // CHECK11-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 13805 // CHECK11-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP7]], 0 13806 // CHECK11-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 13807 // CHECK11-NEXT: [[SUB3:%.*]] = sub nsw i32 [[DIV]], 1 13808 // CHECK11-NEXT: store i32 [[SUB3]], i32* [[DOTCAPTURE_EXPR_2]], align 4 13809 // CHECK11-NEXT: store i32 0, i32* [[I]], align 4 13810 // CHECK11-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 13811 // CHECK11-NEXT: [[CMP:%.*]] = icmp slt i32 0, [[TMP8]] 13812 // CHECK11-NEXT: br i1 [[CMP]], label [[OMP_PRECOND_THEN:%.*]], label [[OMP_PRECOND_END:%.*]] 13813 // CHECK11: omp.precond.then: 13814 // CHECK11-NEXT: store i32 0, i32* [[DOTOMP_COMB_LB]], align 4 13815 // CHECK11-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_2]], align 4 13816 // CHECK11-NEXT: store i32 [[TMP9]], i32* [[DOTOMP_COMB_UB]], align 4 13817 // CHECK11-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4 13818 // CHECK11-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4 13819 // CHECK11-NEXT: [[TMP10:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 4 13820 // CHECK11-NEXT: [[TMP11:%.*]] = load i32, i32* [[TMP10]], align 4 13821 // CHECK11-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1]], i32 [[TMP11]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_COMB_LB]], i32* [[DOTOMP_COMB_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1) 13822 // CHECK11-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 13823 // CHECK11-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_2]], align 4 13824 // CHECK11-NEXT: [[CMP5:%.*]] = icmp sgt i32 [[TMP12]], [[TMP13]] 13825 // CHECK11-NEXT: br i1 [[CMP5]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] 13826 // CHECK11: cond.true: 13827 // CHECK11-NEXT: [[TMP14:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_2]], align 4 13828 // CHECK11-NEXT: br label [[COND_END:%.*]] 13829 // CHECK11: cond.false: 13830 // CHECK11-NEXT: [[TMP15:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 13831 // CHECK11-NEXT: br label [[COND_END]] 13832 // CHECK11: cond.end: 13833 // CHECK11-NEXT: [[COND:%.*]] = phi i32 [ [[TMP14]], [[COND_TRUE]] ], [ [[TMP15]], [[COND_FALSE]] ] 13834 // CHECK11-NEXT: store i32 [[COND]], i32* [[DOTOMP_COMB_UB]], align 4 13835 // CHECK11-NEXT: [[TMP16:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4 13836 // CHECK11-NEXT: store i32 [[TMP16]], i32* [[DOTOMP_IV]], align 4 13837 // CHECK11-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] 13838 // CHECK11: omp.inner.for.cond: 13839 // CHECK11-NEXT: [[TMP17:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !87 13840 // CHECK11-NEXT: [[TMP18:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4, !llvm.access.group !87 13841 // CHECK11-NEXT: [[CMP6:%.*]] = icmp sle i32 [[TMP17]], [[TMP18]] 13842 // CHECK11-NEXT: br i1 [[CMP6]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] 13843 // CHECK11: omp.inner.for.body: 13844 // CHECK11-NEXT: [[TMP19:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4, !llvm.access.group !87 13845 // CHECK11-NEXT: [[TMP20:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4, !llvm.access.group !87 13846 // CHECK11-NEXT: [[TMP21:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4, !llvm.access.group !87 13847 // CHECK11-NEXT: store i32 [[TMP21]], i32* [[DOTCAPTURE_EXPR__CASTED]], align 4, !llvm.access.group !87 13848 // CHECK11-NEXT: [[TMP22:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR__CASTED]], align 4, !llvm.access.group !87 13849 // CHECK11-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 7, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32, i32, i32*, i32**, i32**, i32**, i32)* @.omp_outlined..43 to void (i32*, i32*, ...)*), i32 [[TMP19]], i32 [[TMP20]], i32* [[TMP1]], i32** [[TMP2]], i32** [[TMP3]], i32** [[TMP4]], i32 [[TMP22]]), !llvm.access.group !87 13850 // CHECK11-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] 13851 // CHECK11: omp.inner.for.inc: 13852 // CHECK11-NEXT: [[TMP23:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !87 13853 // CHECK11-NEXT: [[TMP24:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4, !llvm.access.group !87 13854 // CHECK11-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP23]], [[TMP24]] 13855 // CHECK11-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4, !llvm.access.group !87 13856 // CHECK11-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP88:![0-9]+]] 13857 // CHECK11: omp.inner.for.end: 13858 // CHECK11-NEXT: br label [[OMP_LOOP_EXIT:%.*]] 13859 // CHECK11: omp.loop.exit: 13860 // CHECK11-NEXT: [[TMP25:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 4 13861 // CHECK11-NEXT: [[TMP26:%.*]] = load i32, i32* [[TMP25]], align 4 13862 // CHECK11-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP26]]) 13863 // CHECK11-NEXT: [[TMP27:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4 13864 // CHECK11-NEXT: [[TMP28:%.*]] = icmp ne i32 [[TMP27]], 0 13865 // CHECK11-NEXT: br i1 [[TMP28]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]] 13866 // CHECK11: .omp.final.then: 13867 // CHECK11-NEXT: [[TMP29:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 13868 // CHECK11-NEXT: [[SUB7:%.*]] = sub nsw i32 [[TMP29]], 0 13869 // CHECK11-NEXT: [[DIV8:%.*]] = sdiv i32 [[SUB7]], 1 13870 // CHECK11-NEXT: [[MUL:%.*]] = mul nsw i32 [[DIV8]], 1 13871 // CHECK11-NEXT: [[ADD9:%.*]] = add nsw i32 0, [[MUL]] 13872 // CHECK11-NEXT: store i32 [[ADD9]], i32* [[I4]], align 4 13873 // CHECK11-NEXT: br label [[DOTOMP_FINAL_DONE]] 13874 // CHECK11: .omp.final.done: 13875 // CHECK11-NEXT: br label [[OMP_PRECOND_END]] 13876 // CHECK11: omp.precond.end: 13877 // CHECK11-NEXT: ret void 13878 // 13879 // 13880 // CHECK11-LABEL: define {{[^@]+}}@.omp_outlined..43 13881 // CHECK11-SAME: (i32* noalias noundef [[DOTGLOBAL_TID_:%.*]], i32* noalias noundef [[DOTBOUND_TID_:%.*]], i32 noundef [[DOTPREVIOUS_LB_:%.*]], i32 noundef [[DOTPREVIOUS_UB_:%.*]], i32* noundef nonnull align 4 dereferenceable(4) [[N:%.*]], i32** noundef nonnull align 4 dereferenceable(4) [[A:%.*]], i32** noundef nonnull align 4 dereferenceable(4) [[B:%.*]], i32** noundef nonnull align 4 dereferenceable(4) [[C:%.*]], i32 noundef [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR1]] { 13882 // CHECK11-NEXT: entry: 13883 // CHECK11-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4 13884 // CHECK11-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4 13885 // CHECK11-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i32, align 4 13886 // CHECK11-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i32, align 4 13887 // CHECK11-NEXT: [[N_ADDR:%.*]] = alloca i32*, align 4 13888 // CHECK11-NEXT: [[A_ADDR:%.*]] = alloca i32**, align 4 13889 // CHECK11-NEXT: [[B_ADDR:%.*]] = alloca i32**, align 4 13890 // CHECK11-NEXT: [[C_ADDR:%.*]] = alloca i32**, align 4 13891 // CHECK11-NEXT: [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i32, align 4 13892 // CHECK11-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 13893 // CHECK11-NEXT: [[TMP:%.*]] = alloca i32, align 4 13894 // CHECK11-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 13895 // CHECK11-NEXT: [[DOTCAPTURE_EXPR_2:%.*]] = alloca i32, align 4 13896 // CHECK11-NEXT: [[I:%.*]] = alloca i32, align 4 13897 // CHECK11-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 13898 // CHECK11-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4 13899 // CHECK11-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 13900 // CHECK11-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 13901 // CHECK11-NEXT: [[I4:%.*]] = alloca i32, align 4 13902 // CHECK11-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 4 13903 // CHECK11-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 4 13904 // CHECK11-NEXT: store i32 [[DOTPREVIOUS_LB_]], i32* [[DOTPREVIOUS_LB__ADDR]], align 4 13905 // CHECK11-NEXT: store i32 [[DOTPREVIOUS_UB_]], i32* [[DOTPREVIOUS_UB__ADDR]], align 4 13906 // CHECK11-NEXT: store i32* [[N]], i32** [[N_ADDR]], align 4 13907 // CHECK11-NEXT: store i32** [[A]], i32*** [[A_ADDR]], align 4 13908 // CHECK11-NEXT: store i32** [[B]], i32*** [[B_ADDR]], align 4 13909 // CHECK11-NEXT: store i32** [[C]], i32*** [[C_ADDR]], align 4 13910 // CHECK11-NEXT: store i32 [[DOTCAPTURE_EXPR_]], i32* [[DOTCAPTURE_EXPR__ADDR]], align 4 13911 // CHECK11-NEXT: [[TMP0:%.*]] = load i32*, i32** [[N_ADDR]], align 4 13912 // CHECK11-NEXT: [[TMP1:%.*]] = load i32**, i32*** [[A_ADDR]], align 4 13913 // CHECK11-NEXT: [[TMP2:%.*]] = load i32**, i32*** [[B_ADDR]], align 4 13914 // CHECK11-NEXT: [[TMP3:%.*]] = load i32**, i32*** [[C_ADDR]], align 4 13915 // CHECK11-NEXT: [[TMP4:%.*]] = load i32, i32* [[TMP0]], align 4 13916 // CHECK11-NEXT: store i32 [[TMP4]], i32* [[DOTCAPTURE_EXPR_1]], align 4 13917 // CHECK11-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 13918 // CHECK11-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP5]], 0 13919 // CHECK11-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 13920 // CHECK11-NEXT: [[SUB3:%.*]] = sub nsw i32 [[DIV]], 1 13921 // CHECK11-NEXT: store i32 [[SUB3]], i32* [[DOTCAPTURE_EXPR_2]], align 4 13922 // CHECK11-NEXT: store i32 0, i32* [[I]], align 4 13923 // CHECK11-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 13924 // CHECK11-NEXT: [[CMP:%.*]] = icmp slt i32 0, [[TMP6]] 13925 // CHECK11-NEXT: br i1 [[CMP]], label [[OMP_PRECOND_THEN:%.*]], label [[OMP_PRECOND_END:%.*]] 13926 // CHECK11: omp.precond.then: 13927 // CHECK11-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4 13928 // CHECK11-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_2]], align 4 13929 // CHECK11-NEXT: store i32 [[TMP7]], i32* [[DOTOMP_UB]], align 4 13930 // CHECK11-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTPREVIOUS_LB__ADDR]], align 4 13931 // CHECK11-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTPREVIOUS_UB__ADDR]], align 4 13932 // CHECK11-NEXT: store i32 [[TMP8]], i32* [[DOTOMP_LB]], align 4 13933 // CHECK11-NEXT: store i32 [[TMP9]], i32* [[DOTOMP_UB]], align 4 13934 // CHECK11-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4 13935 // CHECK11-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4 13936 // CHECK11-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR__ADDR]], align 4 13937 // CHECK11-NEXT: [[TMP11:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 4 13938 // CHECK11-NEXT: [[TMP12:%.*]] = load i32, i32* [[TMP11]], align 4 13939 // CHECK11-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB2]], i32 [[TMP12]], i32 33, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 [[TMP10]]) 13940 // CHECK11-NEXT: br label [[OMP_DISPATCH_COND:%.*]] 13941 // CHECK11: omp.dispatch.cond: 13942 // CHECK11-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 13943 // CHECK11-NEXT: [[TMP14:%.*]] = load i32, i32* [[DOTPREVIOUS_UB__ADDR]], align 4 13944 // CHECK11-NEXT: [[CMP5:%.*]] = icmp sgt i32 [[TMP13]], [[TMP14]] 13945 // CHECK11-NEXT: br i1 [[CMP5]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] 13946 // CHECK11: cond.true: 13947 // CHECK11-NEXT: [[TMP15:%.*]] = load i32, i32* [[DOTPREVIOUS_UB__ADDR]], align 4 13948 // CHECK11-NEXT: br label [[COND_END:%.*]] 13949 // CHECK11: cond.false: 13950 // CHECK11-NEXT: [[TMP16:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 13951 // CHECK11-NEXT: br label [[COND_END]] 13952 // CHECK11: cond.end: 13953 // CHECK11-NEXT: [[COND:%.*]] = phi i32 [ [[TMP15]], [[COND_TRUE]] ], [ [[TMP16]], [[COND_FALSE]] ] 13954 // CHECK11-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4 13955 // CHECK11-NEXT: [[TMP17:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4 13956 // CHECK11-NEXT: store i32 [[TMP17]], i32* [[DOTOMP_IV]], align 4 13957 // CHECK11-NEXT: [[TMP18:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 13958 // CHECK11-NEXT: [[TMP19:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 13959 // CHECK11-NEXT: [[CMP6:%.*]] = icmp sle i32 [[TMP18]], [[TMP19]] 13960 // CHECK11-NEXT: br i1 [[CMP6]], label [[OMP_DISPATCH_BODY:%.*]], label [[OMP_DISPATCH_END:%.*]] 13961 // CHECK11: omp.dispatch.body: 13962 // CHECK11-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] 13963 // CHECK11: omp.inner.for.cond: 13964 // CHECK11-NEXT: [[TMP20:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !90 13965 // CHECK11-NEXT: [[TMP21:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4, !llvm.access.group !90 13966 // CHECK11-NEXT: [[CMP7:%.*]] = icmp sle i32 [[TMP20]], [[TMP21]] 13967 // CHECK11-NEXT: br i1 [[CMP7]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] 13968 // CHECK11: omp.inner.for.body: 13969 // CHECK11-NEXT: [[TMP22:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !90 13970 // CHECK11-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP22]], 1 13971 // CHECK11-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] 13972 // CHECK11-NEXT: store i32 [[ADD]], i32* [[I4]], align 4, !llvm.access.group !90 13973 // CHECK11-NEXT: [[TMP23:%.*]] = load i32*, i32** [[TMP2]], align 4, !llvm.access.group !90 13974 // CHECK11-NEXT: [[TMP24:%.*]] = load i32, i32* [[I4]], align 4, !llvm.access.group !90 13975 // CHECK11-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i32, i32* [[TMP23]], i32 [[TMP24]] 13976 // CHECK11-NEXT: [[TMP25:%.*]] = load i32, i32* [[ARRAYIDX]], align 4, !llvm.access.group !90 13977 // CHECK11-NEXT: [[TMP26:%.*]] = load i32*, i32** [[TMP3]], align 4, !llvm.access.group !90 13978 // CHECK11-NEXT: [[TMP27:%.*]] = load i32, i32* [[I4]], align 4, !llvm.access.group !90 13979 // CHECK11-NEXT: [[ARRAYIDX8:%.*]] = getelementptr inbounds i32, i32* [[TMP26]], i32 [[TMP27]] 13980 // CHECK11-NEXT: [[TMP28:%.*]] = load i32, i32* [[ARRAYIDX8]], align 4, !llvm.access.group !90 13981 // CHECK11-NEXT: [[ADD9:%.*]] = add nsw i32 [[TMP25]], [[TMP28]] 13982 // CHECK11-NEXT: [[TMP29:%.*]] = load i32*, i32** [[TMP1]], align 4, !llvm.access.group !90 13983 // CHECK11-NEXT: [[TMP30:%.*]] = load i32, i32* [[I4]], align 4, !llvm.access.group !90 13984 // CHECK11-NEXT: [[ARRAYIDX10:%.*]] = getelementptr inbounds i32, i32* [[TMP29]], i32 [[TMP30]] 13985 // CHECK11-NEXT: store i32 [[ADD9]], i32* [[ARRAYIDX10]], align 4, !llvm.access.group !90 13986 // CHECK11-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] 13987 // CHECK11: omp.body.continue: 13988 // CHECK11-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] 13989 // CHECK11: omp.inner.for.inc: 13990 // CHECK11-NEXT: [[TMP31:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !90 13991 // CHECK11-NEXT: [[ADD11:%.*]] = add nsw i32 [[TMP31]], 1 13992 // CHECK11-NEXT: store i32 [[ADD11]], i32* [[DOTOMP_IV]], align 4, !llvm.access.group !90 13993 // CHECK11-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP91:![0-9]+]] 13994 // CHECK11: omp.inner.for.end: 13995 // CHECK11-NEXT: br label [[OMP_DISPATCH_INC:%.*]] 13996 // CHECK11: omp.dispatch.inc: 13997 // CHECK11-NEXT: [[TMP32:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4 13998 // CHECK11-NEXT: [[TMP33:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4 13999 // CHECK11-NEXT: [[ADD12:%.*]] = add nsw i32 [[TMP32]], [[TMP33]] 14000 // CHECK11-NEXT: store i32 [[ADD12]], i32* [[DOTOMP_LB]], align 4 14001 // CHECK11-NEXT: [[TMP34:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 14002 // CHECK11-NEXT: [[TMP35:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4 14003 // CHECK11-NEXT: [[ADD13:%.*]] = add nsw i32 [[TMP34]], [[TMP35]] 14004 // CHECK11-NEXT: store i32 [[ADD13]], i32* [[DOTOMP_UB]], align 4 14005 // CHECK11-NEXT: br label [[OMP_DISPATCH_COND]] 14006 // CHECK11: omp.dispatch.end: 14007 // CHECK11-NEXT: [[TMP36:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 4 14008 // CHECK11-NEXT: [[TMP37:%.*]] = load i32, i32* [[TMP36]], align 4 14009 // CHECK11-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP37]]) 14010 // CHECK11-NEXT: [[TMP38:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4 14011 // CHECK11-NEXT: [[TMP39:%.*]] = icmp ne i32 [[TMP38]], 0 14012 // CHECK11-NEXT: br i1 [[TMP39]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]] 14013 // CHECK11: .omp.final.then: 14014 // CHECK11-NEXT: [[TMP40:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 14015 // CHECK11-NEXT: [[SUB14:%.*]] = sub nsw i32 [[TMP40]], 0 14016 // CHECK11-NEXT: [[DIV15:%.*]] = sdiv i32 [[SUB14]], 1 14017 // CHECK11-NEXT: [[MUL16:%.*]] = mul nsw i32 [[DIV15]], 1 14018 // CHECK11-NEXT: [[ADD17:%.*]] = add nsw i32 0, [[MUL16]] 14019 // CHECK11-NEXT: store i32 [[ADD17]], i32* [[I4]], align 4 14020 // CHECK11-NEXT: br label [[DOTOMP_FINAL_DONE]] 14021 // CHECK11: .omp.final.done: 14022 // CHECK11-NEXT: br label [[OMP_PRECOND_END]] 14023 // CHECK11: omp.precond.end: 14024 // CHECK11-NEXT: ret void 14025 // 14026 // 14027 // CHECK11-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiET_v_l82 14028 // CHECK11-SAME: (i32 noundef [[N:%.*]], i32* noundef [[A:%.*]], i32* noundef [[B:%.*]], i32* noundef [[C:%.*]]) #[[ATTR1]] { 14029 // CHECK11-NEXT: entry: 14030 // CHECK11-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4 14031 // CHECK11-NEXT: [[A_ADDR:%.*]] = alloca i32*, align 4 14032 // CHECK11-NEXT: [[B_ADDR:%.*]] = alloca i32*, align 4 14033 // CHECK11-NEXT: [[C_ADDR:%.*]] = alloca i32*, align 4 14034 // CHECK11-NEXT: store i32 [[N]], i32* [[N_ADDR]], align 4 14035 // CHECK11-NEXT: store i32* [[A]], i32** [[A_ADDR]], align 4 14036 // CHECK11-NEXT: store i32* [[B]], i32** [[B_ADDR]], align 4 14037 // CHECK11-NEXT: store i32* [[C]], i32** [[C_ADDR]], align 4 14038 // CHECK11-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 4, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32*, i32**, i32**, i32**)* @.omp_outlined..46 to void (i32*, i32*, ...)*), i32* [[N_ADDR]], i32** [[A_ADDR]], i32** [[B_ADDR]], i32** [[C_ADDR]]) 14039 // CHECK11-NEXT: ret void 14040 // 14041 // 14042 // CHECK11-LABEL: define {{[^@]+}}@.omp_outlined..46 14043 // CHECK11-SAME: (i32* noalias noundef [[DOTGLOBAL_TID_:%.*]], i32* noalias noundef [[DOTBOUND_TID_:%.*]], i32* noundef nonnull align 4 dereferenceable(4) [[N:%.*]], i32** noundef nonnull align 4 dereferenceable(4) [[A:%.*]], i32** noundef nonnull align 4 dereferenceable(4) [[B:%.*]], i32** noundef nonnull align 4 dereferenceable(4) [[C:%.*]]) #[[ATTR1]] { 14044 // CHECK11-NEXT: entry: 14045 // CHECK11-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4 14046 // CHECK11-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4 14047 // CHECK11-NEXT: [[N_ADDR:%.*]] = alloca i32*, align 4 14048 // CHECK11-NEXT: [[A_ADDR:%.*]] = alloca i32**, align 4 14049 // CHECK11-NEXT: [[B_ADDR:%.*]] = alloca i32**, align 4 14050 // CHECK11-NEXT: [[C_ADDR:%.*]] = alloca i32**, align 4 14051 // CHECK11-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 14052 // CHECK11-NEXT: [[TMP:%.*]] = alloca i32, align 4 14053 // CHECK11-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 14054 // CHECK11-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 14055 // CHECK11-NEXT: [[I:%.*]] = alloca i32, align 4 14056 // CHECK11-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4 14057 // CHECK11-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4 14058 // CHECK11-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 14059 // CHECK11-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 14060 // CHECK11-NEXT: [[I3:%.*]] = alloca i32, align 4 14061 // CHECK11-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 4 14062 // CHECK11-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 4 14063 // CHECK11-NEXT: store i32* [[N]], i32** [[N_ADDR]], align 4 14064 // CHECK11-NEXT: store i32** [[A]], i32*** [[A_ADDR]], align 4 14065 // CHECK11-NEXT: store i32** [[B]], i32*** [[B_ADDR]], align 4 14066 // CHECK11-NEXT: store i32** [[C]], i32*** [[C_ADDR]], align 4 14067 // CHECK11-NEXT: [[TMP0:%.*]] = load i32*, i32** [[N_ADDR]], align 4 14068 // CHECK11-NEXT: [[TMP1:%.*]] = load i32**, i32*** [[A_ADDR]], align 4 14069 // CHECK11-NEXT: [[TMP2:%.*]] = load i32**, i32*** [[B_ADDR]], align 4 14070 // CHECK11-NEXT: [[TMP3:%.*]] = load i32**, i32*** [[C_ADDR]], align 4 14071 // CHECK11-NEXT: [[TMP4:%.*]] = load i32, i32* [[TMP0]], align 4 14072 // CHECK11-NEXT: store i32 [[TMP4]], i32* [[DOTCAPTURE_EXPR_]], align 4 14073 // CHECK11-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 14074 // CHECK11-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP5]], 0 14075 // CHECK11-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 14076 // CHECK11-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1 14077 // CHECK11-NEXT: store i32 [[SUB2]], i32* [[DOTCAPTURE_EXPR_1]], align 4 14078 // CHECK11-NEXT: store i32 0, i32* [[I]], align 4 14079 // CHECK11-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 14080 // CHECK11-NEXT: [[CMP:%.*]] = icmp slt i32 0, [[TMP6]] 14081 // CHECK11-NEXT: br i1 [[CMP]], label [[OMP_PRECOND_THEN:%.*]], label [[OMP_PRECOND_END:%.*]] 14082 // CHECK11: omp.precond.then: 14083 // CHECK11-NEXT: store i32 0, i32* [[DOTOMP_COMB_LB]], align 4 14084 // CHECK11-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 14085 // CHECK11-NEXT: store i32 [[TMP7]], i32* [[DOTOMP_COMB_UB]], align 4 14086 // CHECK11-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4 14087 // CHECK11-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4 14088 // CHECK11-NEXT: [[TMP8:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 4 14089 // CHECK11-NEXT: [[TMP9:%.*]] = load i32, i32* [[TMP8]], align 4 14090 // CHECK11-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1]], i32 [[TMP9]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_COMB_LB]], i32* [[DOTOMP_COMB_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1) 14091 // CHECK11-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 14092 // CHECK11-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 14093 // CHECK11-NEXT: [[CMP4:%.*]] = icmp sgt i32 [[TMP10]], [[TMP11]] 14094 // CHECK11-NEXT: br i1 [[CMP4]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] 14095 // CHECK11: cond.true: 14096 // CHECK11-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 14097 // CHECK11-NEXT: br label [[COND_END:%.*]] 14098 // CHECK11: cond.false: 14099 // CHECK11-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 14100 // CHECK11-NEXT: br label [[COND_END]] 14101 // CHECK11: cond.end: 14102 // CHECK11-NEXT: [[COND:%.*]] = phi i32 [ [[TMP12]], [[COND_TRUE]] ], [ [[TMP13]], [[COND_FALSE]] ] 14103 // CHECK11-NEXT: store i32 [[COND]], i32* [[DOTOMP_COMB_UB]], align 4 14104 // CHECK11-NEXT: [[TMP14:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4 14105 // CHECK11-NEXT: store i32 [[TMP14]], i32* [[DOTOMP_IV]], align 4 14106 // CHECK11-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] 14107 // CHECK11: omp.inner.for.cond: 14108 // CHECK11-NEXT: [[TMP15:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !93 14109 // CHECK11-NEXT: [[TMP16:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4, !llvm.access.group !93 14110 // CHECK11-NEXT: [[CMP5:%.*]] = icmp sle i32 [[TMP15]], [[TMP16]] 14111 // CHECK11-NEXT: br i1 [[CMP5]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] 14112 // CHECK11: omp.inner.for.body: 14113 // CHECK11-NEXT: [[TMP17:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4, !llvm.access.group !93 14114 // CHECK11-NEXT: [[TMP18:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4, !llvm.access.group !93 14115 // CHECK11-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 6, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32, i32, i32*, i32**, i32**, i32**)* @.omp_outlined..47 to void (i32*, i32*, ...)*), i32 [[TMP17]], i32 [[TMP18]], i32* [[TMP0]], i32** [[TMP1]], i32** [[TMP2]], i32** [[TMP3]]), !llvm.access.group !93 14116 // CHECK11-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] 14117 // CHECK11: omp.inner.for.inc: 14118 // CHECK11-NEXT: [[TMP19:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !93 14119 // CHECK11-NEXT: [[TMP20:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4, !llvm.access.group !93 14120 // CHECK11-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP19]], [[TMP20]] 14121 // CHECK11-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4, !llvm.access.group !93 14122 // CHECK11-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP94:![0-9]+]] 14123 // CHECK11: omp.inner.for.end: 14124 // CHECK11-NEXT: br label [[OMP_LOOP_EXIT:%.*]] 14125 // CHECK11: omp.loop.exit: 14126 // CHECK11-NEXT: [[TMP21:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 4 14127 // CHECK11-NEXT: [[TMP22:%.*]] = load i32, i32* [[TMP21]], align 4 14128 // CHECK11-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP22]]) 14129 // CHECK11-NEXT: [[TMP23:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4 14130 // CHECK11-NEXT: [[TMP24:%.*]] = icmp ne i32 [[TMP23]], 0 14131 // CHECK11-NEXT: br i1 [[TMP24]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]] 14132 // CHECK11: .omp.final.then: 14133 // CHECK11-NEXT: [[TMP25:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 14134 // CHECK11-NEXT: [[SUB6:%.*]] = sub nsw i32 [[TMP25]], 0 14135 // CHECK11-NEXT: [[DIV7:%.*]] = sdiv i32 [[SUB6]], 1 14136 // CHECK11-NEXT: [[MUL:%.*]] = mul nsw i32 [[DIV7]], 1 14137 // CHECK11-NEXT: [[ADD8:%.*]] = add nsw i32 0, [[MUL]] 14138 // CHECK11-NEXT: store i32 [[ADD8]], i32* [[I3]], align 4 14139 // CHECK11-NEXT: br label [[DOTOMP_FINAL_DONE]] 14140 // CHECK11: .omp.final.done: 14141 // CHECK11-NEXT: br label [[OMP_PRECOND_END]] 14142 // CHECK11: omp.precond.end: 14143 // CHECK11-NEXT: ret void 14144 // 14145 // 14146 // CHECK11-LABEL: define {{[^@]+}}@.omp_outlined..47 14147 // CHECK11-SAME: (i32* noalias noundef [[DOTGLOBAL_TID_:%.*]], i32* noalias noundef [[DOTBOUND_TID_:%.*]], i32 noundef [[DOTPREVIOUS_LB_:%.*]], i32 noundef [[DOTPREVIOUS_UB_:%.*]], i32* noundef nonnull align 4 dereferenceable(4) [[N:%.*]], i32** noundef nonnull align 4 dereferenceable(4) [[A:%.*]], i32** noundef nonnull align 4 dereferenceable(4) [[B:%.*]], i32** noundef nonnull align 4 dereferenceable(4) [[C:%.*]]) #[[ATTR1]] { 14148 // CHECK11-NEXT: entry: 14149 // CHECK11-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4 14150 // CHECK11-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4 14151 // CHECK11-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i32, align 4 14152 // CHECK11-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i32, align 4 14153 // CHECK11-NEXT: [[N_ADDR:%.*]] = alloca i32*, align 4 14154 // CHECK11-NEXT: [[A_ADDR:%.*]] = alloca i32**, align 4 14155 // CHECK11-NEXT: [[B_ADDR:%.*]] = alloca i32**, align 4 14156 // CHECK11-NEXT: [[C_ADDR:%.*]] = alloca i32**, align 4 14157 // CHECK11-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 14158 // CHECK11-NEXT: [[TMP:%.*]] = alloca i32, align 4 14159 // CHECK11-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 14160 // CHECK11-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 14161 // CHECK11-NEXT: [[I:%.*]] = alloca i32, align 4 14162 // CHECK11-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 14163 // CHECK11-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4 14164 // CHECK11-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 14165 // CHECK11-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 14166 // CHECK11-NEXT: [[I3:%.*]] = alloca i32, align 4 14167 // CHECK11-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 4 14168 // CHECK11-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 4 14169 // CHECK11-NEXT: store i32 [[DOTPREVIOUS_LB_]], i32* [[DOTPREVIOUS_LB__ADDR]], align 4 14170 // CHECK11-NEXT: store i32 [[DOTPREVIOUS_UB_]], i32* [[DOTPREVIOUS_UB__ADDR]], align 4 14171 // CHECK11-NEXT: store i32* [[N]], i32** [[N_ADDR]], align 4 14172 // CHECK11-NEXT: store i32** [[A]], i32*** [[A_ADDR]], align 4 14173 // CHECK11-NEXT: store i32** [[B]], i32*** [[B_ADDR]], align 4 14174 // CHECK11-NEXT: store i32** [[C]], i32*** [[C_ADDR]], align 4 14175 // CHECK11-NEXT: [[TMP0:%.*]] = load i32*, i32** [[N_ADDR]], align 4 14176 // CHECK11-NEXT: [[TMP1:%.*]] = load i32**, i32*** [[A_ADDR]], align 4 14177 // CHECK11-NEXT: [[TMP2:%.*]] = load i32**, i32*** [[B_ADDR]], align 4 14178 // CHECK11-NEXT: [[TMP3:%.*]] = load i32**, i32*** [[C_ADDR]], align 4 14179 // CHECK11-NEXT: [[TMP4:%.*]] = load i32, i32* [[TMP0]], align 4 14180 // CHECK11-NEXT: store i32 [[TMP4]], i32* [[DOTCAPTURE_EXPR_]], align 4 14181 // CHECK11-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 14182 // CHECK11-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP5]], 0 14183 // CHECK11-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 14184 // CHECK11-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1 14185 // CHECK11-NEXT: store i32 [[SUB2]], i32* [[DOTCAPTURE_EXPR_1]], align 4 14186 // CHECK11-NEXT: store i32 0, i32* [[I]], align 4 14187 // CHECK11-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 14188 // CHECK11-NEXT: [[CMP:%.*]] = icmp slt i32 0, [[TMP6]] 14189 // CHECK11-NEXT: br i1 [[CMP]], label [[OMP_PRECOND_THEN:%.*]], label [[OMP_PRECOND_END:%.*]] 14190 // CHECK11: omp.precond.then: 14191 // CHECK11-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4 14192 // CHECK11-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 14193 // CHECK11-NEXT: store i32 [[TMP7]], i32* [[DOTOMP_UB]], align 4 14194 // CHECK11-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTPREVIOUS_LB__ADDR]], align 4 14195 // CHECK11-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTPREVIOUS_UB__ADDR]], align 4 14196 // CHECK11-NEXT: store i32 [[TMP8]], i32* [[DOTOMP_LB]], align 4 14197 // CHECK11-NEXT: store i32 [[TMP9]], i32* [[DOTOMP_UB]], align 4 14198 // CHECK11-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4 14199 // CHECK11-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4 14200 // CHECK11-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4 14201 // CHECK11-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 14202 // CHECK11-NEXT: [[TMP12:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 4 14203 // CHECK11-NEXT: [[TMP13:%.*]] = load i32, i32* [[TMP12]], align 4 14204 // CHECK11-NEXT: call void @__kmpc_dispatch_init_4(%struct.ident_t* @[[GLOB3]], i32 [[TMP13]], i32 35, i32 [[TMP10]], i32 [[TMP11]], i32 1, i32 1) 14205 // CHECK11-NEXT: br label [[OMP_DISPATCH_COND:%.*]] 14206 // CHECK11: omp.dispatch.cond: 14207 // CHECK11-NEXT: [[TMP14:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 4 14208 // CHECK11-NEXT: [[TMP15:%.*]] = load i32, i32* [[TMP14]], align 4 14209 // CHECK11-NEXT: [[TMP16:%.*]] = call i32 @__kmpc_dispatch_next_4(%struct.ident_t* @[[GLOB3]], i32 [[TMP15]], i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]]) 14210 // CHECK11-NEXT: [[TOBOOL:%.*]] = icmp ne i32 [[TMP16]], 0 14211 // CHECK11-NEXT: br i1 [[TOBOOL]], label [[OMP_DISPATCH_BODY:%.*]], label [[OMP_DISPATCH_END:%.*]] 14212 // CHECK11: omp.dispatch.body: 14213 // CHECK11-NEXT: [[TMP17:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4 14214 // CHECK11-NEXT: store i32 [[TMP17]], i32* [[DOTOMP_IV]], align 4 14215 // CHECK11-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] 14216 // CHECK11: omp.inner.for.cond: 14217 // CHECK11-NEXT: [[TMP18:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !96 14218 // CHECK11-NEXT: [[TMP19:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4, !llvm.access.group !96 14219 // CHECK11-NEXT: [[CMP4:%.*]] = icmp sle i32 [[TMP18]], [[TMP19]] 14220 // CHECK11-NEXT: br i1 [[CMP4]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] 14221 // CHECK11: omp.inner.for.body: 14222 // CHECK11-NEXT: [[TMP20:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !96 14223 // CHECK11-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP20]], 1 14224 // CHECK11-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] 14225 // CHECK11-NEXT: store i32 [[ADD]], i32* [[I3]], align 4, !llvm.access.group !96 14226 // CHECK11-NEXT: [[TMP21:%.*]] = load i32*, i32** [[TMP2]], align 4, !llvm.access.group !96 14227 // CHECK11-NEXT: [[TMP22:%.*]] = load i32, i32* [[I3]], align 4, !llvm.access.group !96 14228 // CHECK11-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i32, i32* [[TMP21]], i32 [[TMP22]] 14229 // CHECK11-NEXT: [[TMP23:%.*]] = load i32, i32* [[ARRAYIDX]], align 4, !llvm.access.group !96 14230 // CHECK11-NEXT: [[TMP24:%.*]] = load i32*, i32** [[TMP3]], align 4, !llvm.access.group !96 14231 // CHECK11-NEXT: [[TMP25:%.*]] = load i32, i32* [[I3]], align 4, !llvm.access.group !96 14232 // CHECK11-NEXT: [[ARRAYIDX5:%.*]] = getelementptr inbounds i32, i32* [[TMP24]], i32 [[TMP25]] 14233 // CHECK11-NEXT: [[TMP26:%.*]] = load i32, i32* [[ARRAYIDX5]], align 4, !llvm.access.group !96 14234 // CHECK11-NEXT: [[ADD6:%.*]] = add nsw i32 [[TMP23]], [[TMP26]] 14235 // CHECK11-NEXT: [[TMP27:%.*]] = load i32*, i32** [[TMP1]], align 4, !llvm.access.group !96 14236 // CHECK11-NEXT: [[TMP28:%.*]] = load i32, i32* [[I3]], align 4, !llvm.access.group !96 14237 // CHECK11-NEXT: [[ARRAYIDX7:%.*]] = getelementptr inbounds i32, i32* [[TMP27]], i32 [[TMP28]] 14238 // CHECK11-NEXT: store i32 [[ADD6]], i32* [[ARRAYIDX7]], align 4, !llvm.access.group !96 14239 // CHECK11-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] 14240 // CHECK11: omp.body.continue: 14241 // CHECK11-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] 14242 // CHECK11: omp.inner.for.inc: 14243 // CHECK11-NEXT: [[TMP29:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !96 14244 // CHECK11-NEXT: [[ADD8:%.*]] = add nsw i32 [[TMP29]], 1 14245 // CHECK11-NEXT: store i32 [[ADD8]], i32* [[DOTOMP_IV]], align 4, !llvm.access.group !96 14246 // CHECK11-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP97:![0-9]+]] 14247 // CHECK11: omp.inner.for.end: 14248 // CHECK11-NEXT: br label [[OMP_DISPATCH_INC:%.*]] 14249 // CHECK11: omp.dispatch.inc: 14250 // CHECK11-NEXT: br label [[OMP_DISPATCH_COND]] 14251 // CHECK11: omp.dispatch.end: 14252 // CHECK11-NEXT: [[TMP30:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4 14253 // CHECK11-NEXT: [[TMP31:%.*]] = icmp ne i32 [[TMP30]], 0 14254 // CHECK11-NEXT: br i1 [[TMP31]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]] 14255 // CHECK11: .omp.final.then: 14256 // CHECK11-NEXT: [[TMP32:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 14257 // CHECK11-NEXT: [[SUB9:%.*]] = sub nsw i32 [[TMP32]], 0 14258 // CHECK11-NEXT: [[DIV10:%.*]] = sdiv i32 [[SUB9]], 1 14259 // CHECK11-NEXT: [[MUL11:%.*]] = mul nsw i32 [[DIV10]], 1 14260 // CHECK11-NEXT: [[ADD12:%.*]] = add nsw i32 0, [[MUL11]] 14261 // CHECK11-NEXT: store i32 [[ADD12]], i32* [[I3]], align 4 14262 // CHECK11-NEXT: br label [[DOTOMP_FINAL_DONE]] 14263 // CHECK11: .omp.final.done: 14264 // CHECK11-NEXT: br label [[OMP_PRECOND_END]] 14265 // CHECK11: omp.precond.end: 14266 // CHECK11-NEXT: ret void 14267 // 14268 // 14269 // CHECK11-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiET_v_l90 14270 // CHECK11-SAME: (i32 noundef [[CH:%.*]], i32 noundef [[N:%.*]], i32* noundef [[A:%.*]], i32* noundef [[B:%.*]], i32* noundef [[C:%.*]]) #[[ATTR1]] { 14271 // CHECK11-NEXT: entry: 14272 // CHECK11-NEXT: [[CH_ADDR:%.*]] = alloca i32, align 4 14273 // CHECK11-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4 14274 // CHECK11-NEXT: [[A_ADDR:%.*]] = alloca i32*, align 4 14275 // CHECK11-NEXT: [[B_ADDR:%.*]] = alloca i32*, align 4 14276 // CHECK11-NEXT: [[C_ADDR:%.*]] = alloca i32*, align 4 14277 // CHECK11-NEXT: store i32 [[CH]], i32* [[CH_ADDR]], align 4 14278 // CHECK11-NEXT: store i32 [[N]], i32* [[N_ADDR]], align 4 14279 // CHECK11-NEXT: store i32* [[A]], i32** [[A_ADDR]], align 4 14280 // CHECK11-NEXT: store i32* [[B]], i32** [[B_ADDR]], align 4 14281 // CHECK11-NEXT: store i32* [[C]], i32** [[C_ADDR]], align 4 14282 // CHECK11-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 5, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32*, i32*, i32**, i32**, i32**)* @.omp_outlined..50 to void (i32*, i32*, ...)*), i32* [[CH_ADDR]], i32* [[N_ADDR]], i32** [[A_ADDR]], i32** [[B_ADDR]], i32** [[C_ADDR]]) 14283 // CHECK11-NEXT: ret void 14284 // 14285 // 14286 // CHECK11-LABEL: define {{[^@]+}}@.omp_outlined..50 14287 // CHECK11-SAME: (i32* noalias noundef [[DOTGLOBAL_TID_:%.*]], i32* noalias noundef [[DOTBOUND_TID_:%.*]], i32* noundef nonnull align 4 dereferenceable(4) [[CH:%.*]], i32* noundef nonnull align 4 dereferenceable(4) [[N:%.*]], i32** noundef nonnull align 4 dereferenceable(4) [[A:%.*]], i32** noundef nonnull align 4 dereferenceable(4) [[B:%.*]], i32** noundef nonnull align 4 dereferenceable(4) [[C:%.*]]) #[[ATTR1]] { 14288 // CHECK11-NEXT: entry: 14289 // CHECK11-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4 14290 // CHECK11-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4 14291 // CHECK11-NEXT: [[CH_ADDR:%.*]] = alloca i32*, align 4 14292 // CHECK11-NEXT: [[N_ADDR:%.*]] = alloca i32*, align 4 14293 // CHECK11-NEXT: [[A_ADDR:%.*]] = alloca i32**, align 4 14294 // CHECK11-NEXT: [[B_ADDR:%.*]] = alloca i32**, align 4 14295 // CHECK11-NEXT: [[C_ADDR:%.*]] = alloca i32**, align 4 14296 // CHECK11-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 14297 // CHECK11-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 14298 // CHECK11-NEXT: [[TMP:%.*]] = alloca i32, align 4 14299 // CHECK11-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 14300 // CHECK11-NEXT: [[DOTCAPTURE_EXPR_2:%.*]] = alloca i32, align 4 14301 // CHECK11-NEXT: [[I:%.*]] = alloca i32, align 4 14302 // CHECK11-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4 14303 // CHECK11-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4 14304 // CHECK11-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 14305 // CHECK11-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 14306 // CHECK11-NEXT: [[I4:%.*]] = alloca i32, align 4 14307 // CHECK11-NEXT: [[DOTCAPTURE_EXPR__CASTED:%.*]] = alloca i32, align 4 14308 // CHECK11-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 4 14309 // CHECK11-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 4 14310 // CHECK11-NEXT: store i32* [[CH]], i32** [[CH_ADDR]], align 4 14311 // CHECK11-NEXT: store i32* [[N]], i32** [[N_ADDR]], align 4 14312 // CHECK11-NEXT: store i32** [[A]], i32*** [[A_ADDR]], align 4 14313 // CHECK11-NEXT: store i32** [[B]], i32*** [[B_ADDR]], align 4 14314 // CHECK11-NEXT: store i32** [[C]], i32*** [[C_ADDR]], align 4 14315 // CHECK11-NEXT: [[TMP0:%.*]] = load i32*, i32** [[CH_ADDR]], align 4 14316 // CHECK11-NEXT: [[TMP1:%.*]] = load i32*, i32** [[N_ADDR]], align 4 14317 // CHECK11-NEXT: [[TMP2:%.*]] = load i32**, i32*** [[A_ADDR]], align 4 14318 // CHECK11-NEXT: [[TMP3:%.*]] = load i32**, i32*** [[B_ADDR]], align 4 14319 // CHECK11-NEXT: [[TMP4:%.*]] = load i32**, i32*** [[C_ADDR]], align 4 14320 // CHECK11-NEXT: [[TMP5:%.*]] = load i32, i32* [[TMP0]], align 4 14321 // CHECK11-NEXT: store i32 [[TMP5]], i32* [[DOTCAPTURE_EXPR_]], align 4 14322 // CHECK11-NEXT: [[TMP6:%.*]] = load i32, i32* [[TMP1]], align 4 14323 // CHECK11-NEXT: store i32 [[TMP6]], i32* [[DOTCAPTURE_EXPR_1]], align 4 14324 // CHECK11-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 14325 // CHECK11-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP7]], 0 14326 // CHECK11-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 14327 // CHECK11-NEXT: [[SUB3:%.*]] = sub nsw i32 [[DIV]], 1 14328 // CHECK11-NEXT: store i32 [[SUB3]], i32* [[DOTCAPTURE_EXPR_2]], align 4 14329 // CHECK11-NEXT: store i32 0, i32* [[I]], align 4 14330 // CHECK11-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 14331 // CHECK11-NEXT: [[CMP:%.*]] = icmp slt i32 0, [[TMP8]] 14332 // CHECK11-NEXT: br i1 [[CMP]], label [[OMP_PRECOND_THEN:%.*]], label [[OMP_PRECOND_END:%.*]] 14333 // CHECK11: omp.precond.then: 14334 // CHECK11-NEXT: store i32 0, i32* [[DOTOMP_COMB_LB]], align 4 14335 // CHECK11-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_2]], align 4 14336 // CHECK11-NEXT: store i32 [[TMP9]], i32* [[DOTOMP_COMB_UB]], align 4 14337 // CHECK11-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4 14338 // CHECK11-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4 14339 // CHECK11-NEXT: [[TMP10:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 4 14340 // CHECK11-NEXT: [[TMP11:%.*]] = load i32, i32* [[TMP10]], align 4 14341 // CHECK11-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1]], i32 [[TMP11]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_COMB_LB]], i32* [[DOTOMP_COMB_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1) 14342 // CHECK11-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 14343 // CHECK11-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_2]], align 4 14344 // CHECK11-NEXT: [[CMP5:%.*]] = icmp sgt i32 [[TMP12]], [[TMP13]] 14345 // CHECK11-NEXT: br i1 [[CMP5]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] 14346 // CHECK11: cond.true: 14347 // CHECK11-NEXT: [[TMP14:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_2]], align 4 14348 // CHECK11-NEXT: br label [[COND_END:%.*]] 14349 // CHECK11: cond.false: 14350 // CHECK11-NEXT: [[TMP15:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 14351 // CHECK11-NEXT: br label [[COND_END]] 14352 // CHECK11: cond.end: 14353 // CHECK11-NEXT: [[COND:%.*]] = phi i32 [ [[TMP14]], [[COND_TRUE]] ], [ [[TMP15]], [[COND_FALSE]] ] 14354 // CHECK11-NEXT: store i32 [[COND]], i32* [[DOTOMP_COMB_UB]], align 4 14355 // CHECK11-NEXT: [[TMP16:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4 14356 // CHECK11-NEXT: store i32 [[TMP16]], i32* [[DOTOMP_IV]], align 4 14357 // CHECK11-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] 14358 // CHECK11: omp.inner.for.cond: 14359 // CHECK11-NEXT: [[TMP17:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !99 14360 // CHECK11-NEXT: [[TMP18:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4, !llvm.access.group !99 14361 // CHECK11-NEXT: [[CMP6:%.*]] = icmp sle i32 [[TMP17]], [[TMP18]] 14362 // CHECK11-NEXT: br i1 [[CMP6]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] 14363 // CHECK11: omp.inner.for.body: 14364 // CHECK11-NEXT: [[TMP19:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4, !llvm.access.group !99 14365 // CHECK11-NEXT: [[TMP20:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4, !llvm.access.group !99 14366 // CHECK11-NEXT: [[TMP21:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4, !llvm.access.group !99 14367 // CHECK11-NEXT: store i32 [[TMP21]], i32* [[DOTCAPTURE_EXPR__CASTED]], align 4, !llvm.access.group !99 14368 // CHECK11-NEXT: [[TMP22:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR__CASTED]], align 4, !llvm.access.group !99 14369 // CHECK11-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 7, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32, i32, i32*, i32**, i32**, i32**, i32)* @.omp_outlined..51 to void (i32*, i32*, ...)*), i32 [[TMP19]], i32 [[TMP20]], i32* [[TMP1]], i32** [[TMP2]], i32** [[TMP3]], i32** [[TMP4]], i32 [[TMP22]]), !llvm.access.group !99 14370 // CHECK11-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] 14371 // CHECK11: omp.inner.for.inc: 14372 // CHECK11-NEXT: [[TMP23:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !99 14373 // CHECK11-NEXT: [[TMP24:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4, !llvm.access.group !99 14374 // CHECK11-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP23]], [[TMP24]] 14375 // CHECK11-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4, !llvm.access.group !99 14376 // CHECK11-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP100:![0-9]+]] 14377 // CHECK11: omp.inner.for.end: 14378 // CHECK11-NEXT: br label [[OMP_LOOP_EXIT:%.*]] 14379 // CHECK11: omp.loop.exit: 14380 // CHECK11-NEXT: [[TMP25:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 4 14381 // CHECK11-NEXT: [[TMP26:%.*]] = load i32, i32* [[TMP25]], align 4 14382 // CHECK11-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP26]]) 14383 // CHECK11-NEXT: [[TMP27:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4 14384 // CHECK11-NEXT: [[TMP28:%.*]] = icmp ne i32 [[TMP27]], 0 14385 // CHECK11-NEXT: br i1 [[TMP28]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]] 14386 // CHECK11: .omp.final.then: 14387 // CHECK11-NEXT: [[TMP29:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 14388 // CHECK11-NEXT: [[SUB7:%.*]] = sub nsw i32 [[TMP29]], 0 14389 // CHECK11-NEXT: [[DIV8:%.*]] = sdiv i32 [[SUB7]], 1 14390 // CHECK11-NEXT: [[MUL:%.*]] = mul nsw i32 [[DIV8]], 1 14391 // CHECK11-NEXT: [[ADD9:%.*]] = add nsw i32 0, [[MUL]] 14392 // CHECK11-NEXT: store i32 [[ADD9]], i32* [[I4]], align 4 14393 // CHECK11-NEXT: br label [[DOTOMP_FINAL_DONE]] 14394 // CHECK11: .omp.final.done: 14395 // CHECK11-NEXT: br label [[OMP_PRECOND_END]] 14396 // CHECK11: omp.precond.end: 14397 // CHECK11-NEXT: ret void 14398 // 14399 // 14400 // CHECK11-LABEL: define {{[^@]+}}@.omp_outlined..51 14401 // CHECK11-SAME: (i32* noalias noundef [[DOTGLOBAL_TID_:%.*]], i32* noalias noundef [[DOTBOUND_TID_:%.*]], i32 noundef [[DOTPREVIOUS_LB_:%.*]], i32 noundef [[DOTPREVIOUS_UB_:%.*]], i32* noundef nonnull align 4 dereferenceable(4) [[N:%.*]], i32** noundef nonnull align 4 dereferenceable(4) [[A:%.*]], i32** noundef nonnull align 4 dereferenceable(4) [[B:%.*]], i32** noundef nonnull align 4 dereferenceable(4) [[C:%.*]], i32 noundef [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR1]] { 14402 // CHECK11-NEXT: entry: 14403 // CHECK11-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4 14404 // CHECK11-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4 14405 // CHECK11-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i32, align 4 14406 // CHECK11-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i32, align 4 14407 // CHECK11-NEXT: [[N_ADDR:%.*]] = alloca i32*, align 4 14408 // CHECK11-NEXT: [[A_ADDR:%.*]] = alloca i32**, align 4 14409 // CHECK11-NEXT: [[B_ADDR:%.*]] = alloca i32**, align 4 14410 // CHECK11-NEXT: [[C_ADDR:%.*]] = alloca i32**, align 4 14411 // CHECK11-NEXT: [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i32, align 4 14412 // CHECK11-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 14413 // CHECK11-NEXT: [[TMP:%.*]] = alloca i32, align 4 14414 // CHECK11-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 14415 // CHECK11-NEXT: [[DOTCAPTURE_EXPR_2:%.*]] = alloca i32, align 4 14416 // CHECK11-NEXT: [[I:%.*]] = alloca i32, align 4 14417 // CHECK11-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 14418 // CHECK11-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4 14419 // CHECK11-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 14420 // CHECK11-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 14421 // CHECK11-NEXT: [[I4:%.*]] = alloca i32, align 4 14422 // CHECK11-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 4 14423 // CHECK11-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 4 14424 // CHECK11-NEXT: store i32 [[DOTPREVIOUS_LB_]], i32* [[DOTPREVIOUS_LB__ADDR]], align 4 14425 // CHECK11-NEXT: store i32 [[DOTPREVIOUS_UB_]], i32* [[DOTPREVIOUS_UB__ADDR]], align 4 14426 // CHECK11-NEXT: store i32* [[N]], i32** [[N_ADDR]], align 4 14427 // CHECK11-NEXT: store i32** [[A]], i32*** [[A_ADDR]], align 4 14428 // CHECK11-NEXT: store i32** [[B]], i32*** [[B_ADDR]], align 4 14429 // CHECK11-NEXT: store i32** [[C]], i32*** [[C_ADDR]], align 4 14430 // CHECK11-NEXT: store i32 [[DOTCAPTURE_EXPR_]], i32* [[DOTCAPTURE_EXPR__ADDR]], align 4 14431 // CHECK11-NEXT: [[TMP0:%.*]] = load i32*, i32** [[N_ADDR]], align 4 14432 // CHECK11-NEXT: [[TMP1:%.*]] = load i32**, i32*** [[A_ADDR]], align 4 14433 // CHECK11-NEXT: [[TMP2:%.*]] = load i32**, i32*** [[B_ADDR]], align 4 14434 // CHECK11-NEXT: [[TMP3:%.*]] = load i32**, i32*** [[C_ADDR]], align 4 14435 // CHECK11-NEXT: [[TMP4:%.*]] = load i32, i32* [[TMP0]], align 4 14436 // CHECK11-NEXT: store i32 [[TMP4]], i32* [[DOTCAPTURE_EXPR_1]], align 4 14437 // CHECK11-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 14438 // CHECK11-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP5]], 0 14439 // CHECK11-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 14440 // CHECK11-NEXT: [[SUB3:%.*]] = sub nsw i32 [[DIV]], 1 14441 // CHECK11-NEXT: store i32 [[SUB3]], i32* [[DOTCAPTURE_EXPR_2]], align 4 14442 // CHECK11-NEXT: store i32 0, i32* [[I]], align 4 14443 // CHECK11-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 14444 // CHECK11-NEXT: [[CMP:%.*]] = icmp slt i32 0, [[TMP6]] 14445 // CHECK11-NEXT: br i1 [[CMP]], label [[OMP_PRECOND_THEN:%.*]], label [[OMP_PRECOND_END:%.*]] 14446 // CHECK11: omp.precond.then: 14447 // CHECK11-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4 14448 // CHECK11-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_2]], align 4 14449 // CHECK11-NEXT: store i32 [[TMP7]], i32* [[DOTOMP_UB]], align 4 14450 // CHECK11-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTPREVIOUS_LB__ADDR]], align 4 14451 // CHECK11-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTPREVIOUS_UB__ADDR]], align 4 14452 // CHECK11-NEXT: store i32 [[TMP8]], i32* [[DOTOMP_LB]], align 4 14453 // CHECK11-NEXT: store i32 [[TMP9]], i32* [[DOTOMP_UB]], align 4 14454 // CHECK11-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4 14455 // CHECK11-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4 14456 // CHECK11-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR__ADDR]], align 4 14457 // CHECK11-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4 14458 // CHECK11-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 14459 // CHECK11-NEXT: [[TMP13:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 4 14460 // CHECK11-NEXT: [[TMP14:%.*]] = load i32, i32* [[TMP13]], align 4 14461 // CHECK11-NEXT: call void @__kmpc_dispatch_init_4(%struct.ident_t* @[[GLOB3]], i32 [[TMP14]], i32 35, i32 [[TMP11]], i32 [[TMP12]], i32 1, i32 [[TMP10]]) 14462 // CHECK11-NEXT: br label [[OMP_DISPATCH_COND:%.*]] 14463 // CHECK11: omp.dispatch.cond: 14464 // CHECK11-NEXT: [[TMP15:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 4 14465 // CHECK11-NEXT: [[TMP16:%.*]] = load i32, i32* [[TMP15]], align 4 14466 // CHECK11-NEXT: [[TMP17:%.*]] = call i32 @__kmpc_dispatch_next_4(%struct.ident_t* @[[GLOB3]], i32 [[TMP16]], i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]]) 14467 // CHECK11-NEXT: [[TOBOOL:%.*]] = icmp ne i32 [[TMP17]], 0 14468 // CHECK11-NEXT: br i1 [[TOBOOL]], label [[OMP_DISPATCH_BODY:%.*]], label [[OMP_DISPATCH_END:%.*]] 14469 // CHECK11: omp.dispatch.body: 14470 // CHECK11-NEXT: [[TMP18:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4 14471 // CHECK11-NEXT: store i32 [[TMP18]], i32* [[DOTOMP_IV]], align 4 14472 // CHECK11-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] 14473 // CHECK11: omp.inner.for.cond: 14474 // CHECK11-NEXT: [[TMP19:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !102 14475 // CHECK11-NEXT: [[TMP20:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4, !llvm.access.group !102 14476 // CHECK11-NEXT: [[CMP5:%.*]] = icmp sle i32 [[TMP19]], [[TMP20]] 14477 // CHECK11-NEXT: br i1 [[CMP5]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] 14478 // CHECK11: omp.inner.for.body: 14479 // CHECK11-NEXT: [[TMP21:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !102 14480 // CHECK11-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP21]], 1 14481 // CHECK11-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] 14482 // CHECK11-NEXT: store i32 [[ADD]], i32* [[I4]], align 4, !llvm.access.group !102 14483 // CHECK11-NEXT: [[TMP22:%.*]] = load i32*, i32** [[TMP2]], align 4, !llvm.access.group !102 14484 // CHECK11-NEXT: [[TMP23:%.*]] = load i32, i32* [[I4]], align 4, !llvm.access.group !102 14485 // CHECK11-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i32, i32* [[TMP22]], i32 [[TMP23]] 14486 // CHECK11-NEXT: [[TMP24:%.*]] = load i32, i32* [[ARRAYIDX]], align 4, !llvm.access.group !102 14487 // CHECK11-NEXT: [[TMP25:%.*]] = load i32*, i32** [[TMP3]], align 4, !llvm.access.group !102 14488 // CHECK11-NEXT: [[TMP26:%.*]] = load i32, i32* [[I4]], align 4, !llvm.access.group !102 14489 // CHECK11-NEXT: [[ARRAYIDX6:%.*]] = getelementptr inbounds i32, i32* [[TMP25]], i32 [[TMP26]] 14490 // CHECK11-NEXT: [[TMP27:%.*]] = load i32, i32* [[ARRAYIDX6]], align 4, !llvm.access.group !102 14491 // CHECK11-NEXT: [[ADD7:%.*]] = add nsw i32 [[TMP24]], [[TMP27]] 14492 // CHECK11-NEXT: [[TMP28:%.*]] = load i32*, i32** [[TMP1]], align 4, !llvm.access.group !102 14493 // CHECK11-NEXT: [[TMP29:%.*]] = load i32, i32* [[I4]], align 4, !llvm.access.group !102 14494 // CHECK11-NEXT: [[ARRAYIDX8:%.*]] = getelementptr inbounds i32, i32* [[TMP28]], i32 [[TMP29]] 14495 // CHECK11-NEXT: store i32 [[ADD7]], i32* [[ARRAYIDX8]], align 4, !llvm.access.group !102 14496 // CHECK11-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] 14497 // CHECK11: omp.body.continue: 14498 // CHECK11-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] 14499 // CHECK11: omp.inner.for.inc: 14500 // CHECK11-NEXT: [[TMP30:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !102 14501 // CHECK11-NEXT: [[ADD9:%.*]] = add nsw i32 [[TMP30]], 1 14502 // CHECK11-NEXT: store i32 [[ADD9]], i32* [[DOTOMP_IV]], align 4, !llvm.access.group !102 14503 // CHECK11-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP103:![0-9]+]] 14504 // CHECK11: omp.inner.for.end: 14505 // CHECK11-NEXT: br label [[OMP_DISPATCH_INC:%.*]] 14506 // CHECK11: omp.dispatch.inc: 14507 // CHECK11-NEXT: br label [[OMP_DISPATCH_COND]] 14508 // CHECK11: omp.dispatch.end: 14509 // CHECK11-NEXT: [[TMP31:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4 14510 // CHECK11-NEXT: [[TMP32:%.*]] = icmp ne i32 [[TMP31]], 0 14511 // CHECK11-NEXT: br i1 [[TMP32]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]] 14512 // CHECK11: .omp.final.then: 14513 // CHECK11-NEXT: [[TMP33:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 14514 // CHECK11-NEXT: [[SUB10:%.*]] = sub nsw i32 [[TMP33]], 0 14515 // CHECK11-NEXT: [[DIV11:%.*]] = sdiv i32 [[SUB10]], 1 14516 // CHECK11-NEXT: [[MUL12:%.*]] = mul nsw i32 [[DIV11]], 1 14517 // CHECK11-NEXT: [[ADD13:%.*]] = add nsw i32 0, [[MUL12]] 14518 // CHECK11-NEXT: store i32 [[ADD13]], i32* [[I4]], align 4 14519 // CHECK11-NEXT: br label [[DOTOMP_FINAL_DONE]] 14520 // CHECK11: .omp.final.done: 14521 // CHECK11-NEXT: br label [[OMP_PRECOND_END]] 14522 // CHECK11: omp.precond.end: 14523 // CHECK11-NEXT: ret void 14524 // 14525 // 14526 // CHECK11-LABEL: define {{[^@]+}}@.omp_offloading.requires_reg 14527 // CHECK11-SAME: () #[[ATTR4:[0-9]+]] { 14528 // CHECK11-NEXT: entry: 14529 // CHECK11-NEXT: call void @__tgt_register_requires(i64 1) 14530 // CHECK11-NEXT: ret void 14531 // 14532 // 14533 // CHECK13-LABEL: define {{[^@]+}}@main 14534 // CHECK13-SAME: () #[[ATTR0:[0-9]+]] { 14535 // CHECK13-NEXT: entry: 14536 // CHECK13-NEXT: [[RETVAL:%.*]] = alloca i32, align 4 14537 // CHECK13-NEXT: [[A:%.*]] = alloca double*, align 8 14538 // CHECK13-NEXT: [[B:%.*]] = alloca double*, align 8 14539 // CHECK13-NEXT: [[C:%.*]] = alloca double*, align 8 14540 // CHECK13-NEXT: [[N:%.*]] = alloca i32, align 4 14541 // CHECK13-NEXT: [[CH:%.*]] = alloca i32, align 4 14542 // CHECK13-NEXT: [[TMP:%.*]] = alloca i32, align 4 14543 // CHECK13-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 14544 // CHECK13-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 14545 // CHECK13-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 14546 // CHECK13-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4 14547 // CHECK13-NEXT: [[I:%.*]] = alloca i32, align 4 14548 // CHECK13-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 14549 // CHECK13-NEXT: [[I3:%.*]] = alloca i32, align 4 14550 // CHECK13-NEXT: [[_TMP15:%.*]] = alloca i32, align 4 14551 // CHECK13-NEXT: [[DOTCAPTURE_EXPR_16:%.*]] = alloca i32, align 4 14552 // CHECK13-NEXT: [[DOTCAPTURE_EXPR_17:%.*]] = alloca i32, align 4 14553 // CHECK13-NEXT: [[DOTOMP_LB21:%.*]] = alloca i32, align 4 14554 // CHECK13-NEXT: [[DOTOMP_UB22:%.*]] = alloca i32, align 4 14555 // CHECK13-NEXT: [[I23:%.*]] = alloca i32, align 4 14556 // CHECK13-NEXT: [[DOTOMP_IV26:%.*]] = alloca i32, align 4 14557 // CHECK13-NEXT: [[I27:%.*]] = alloca i32, align 4 14558 // CHECK13-NEXT: [[_TMP49:%.*]] = alloca i32, align 4 14559 // CHECK13-NEXT: [[DOTCAPTURE_EXPR_50:%.*]] = alloca i32, align 4 14560 // CHECK13-NEXT: [[DOTCAPTURE_EXPR_51:%.*]] = alloca i32, align 4 14561 // CHECK13-NEXT: [[DOTOMP_LB55:%.*]] = alloca i32, align 4 14562 // CHECK13-NEXT: [[DOTOMP_UB56:%.*]] = alloca i32, align 4 14563 // CHECK13-NEXT: [[I57:%.*]] = alloca i32, align 4 14564 // CHECK13-NEXT: [[DOTOMP_IV60:%.*]] = alloca i32, align 4 14565 // CHECK13-NEXT: [[I61:%.*]] = alloca i32, align 4 14566 // CHECK13-NEXT: [[_TMP83:%.*]] = alloca i32, align 4 14567 // CHECK13-NEXT: [[DOTCAPTURE_EXPR_84:%.*]] = alloca i32, align 4 14568 // CHECK13-NEXT: [[DOTCAPTURE_EXPR_85:%.*]] = alloca i32, align 4 14569 // CHECK13-NEXT: [[DOTOMP_LB89:%.*]] = alloca i32, align 4 14570 // CHECK13-NEXT: [[DOTOMP_UB90:%.*]] = alloca i32, align 4 14571 // CHECK13-NEXT: [[I91:%.*]] = alloca i32, align 4 14572 // CHECK13-NEXT: [[DOTOMP_IV94:%.*]] = alloca i32, align 4 14573 // CHECK13-NEXT: [[I95:%.*]] = alloca i32, align 4 14574 // CHECK13-NEXT: [[DOTCAPTURE_EXPR_117:%.*]] = alloca i32, align 4 14575 // CHECK13-NEXT: [[_TMP118:%.*]] = alloca i32, align 4 14576 // CHECK13-NEXT: [[DOTCAPTURE_EXPR_119:%.*]] = alloca i32, align 4 14577 // CHECK13-NEXT: [[DOTCAPTURE_EXPR_120:%.*]] = alloca i32, align 4 14578 // CHECK13-NEXT: [[DOTOMP_LB124:%.*]] = alloca i32, align 4 14579 // CHECK13-NEXT: [[DOTOMP_UB125:%.*]] = alloca i32, align 4 14580 // CHECK13-NEXT: [[I126:%.*]] = alloca i32, align 4 14581 // CHECK13-NEXT: [[DOTOMP_IV129:%.*]] = alloca i32, align 4 14582 // CHECK13-NEXT: [[I130:%.*]] = alloca i32, align 4 14583 // CHECK13-NEXT: [[_TMP152:%.*]] = alloca i32, align 4 14584 // CHECK13-NEXT: [[DOTCAPTURE_EXPR_153:%.*]] = alloca i32, align 4 14585 // CHECK13-NEXT: [[DOTCAPTURE_EXPR_154:%.*]] = alloca i32, align 4 14586 // CHECK13-NEXT: [[DOTOMP_LB158:%.*]] = alloca i32, align 4 14587 // CHECK13-NEXT: [[DOTOMP_UB159:%.*]] = alloca i32, align 4 14588 // CHECK13-NEXT: [[I160:%.*]] = alloca i32, align 4 14589 // CHECK13-NEXT: [[DOTOMP_IV163:%.*]] = alloca i32, align 4 14590 // CHECK13-NEXT: [[I164:%.*]] = alloca i32, align 4 14591 // CHECK13-NEXT: [[DOTCAPTURE_EXPR_186:%.*]] = alloca i32, align 4 14592 // CHECK13-NEXT: [[_TMP187:%.*]] = alloca i32, align 4 14593 // CHECK13-NEXT: [[DOTCAPTURE_EXPR_188:%.*]] = alloca i32, align 4 14594 // CHECK13-NEXT: [[DOTCAPTURE_EXPR_189:%.*]] = alloca i32, align 4 14595 // CHECK13-NEXT: [[DOTOMP_LB193:%.*]] = alloca i32, align 4 14596 // CHECK13-NEXT: [[DOTOMP_UB194:%.*]] = alloca i32, align 4 14597 // CHECK13-NEXT: [[I195:%.*]] = alloca i32, align 4 14598 // CHECK13-NEXT: [[DOTOMP_IV198:%.*]] = alloca i32, align 4 14599 // CHECK13-NEXT: [[I199:%.*]] = alloca i32, align 4 14600 // CHECK13-NEXT: store i32 0, i32* [[RETVAL]], align 4 14601 // CHECK13-NEXT: store i32 10000, i32* [[N]], align 4 14602 // CHECK13-NEXT: store i32 100, i32* [[CH]], align 4 14603 // CHECK13-NEXT: [[TMP0:%.*]] = load i32, i32* [[N]], align 4 14604 // CHECK13-NEXT: store i32 [[TMP0]], i32* [[DOTCAPTURE_EXPR_]], align 4 14605 // CHECK13-NEXT: [[TMP1:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 14606 // CHECK13-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP1]], 0 14607 // CHECK13-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 14608 // CHECK13-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1 14609 // CHECK13-NEXT: store i32 [[SUB2]], i32* [[DOTCAPTURE_EXPR_1]], align 4 14610 // CHECK13-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4 14611 // CHECK13-NEXT: [[TMP2:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 14612 // CHECK13-NEXT: store i32 [[TMP2]], i32* [[DOTOMP_UB]], align 4 14613 // CHECK13-NEXT: store i32 0, i32* [[I]], align 4 14614 // CHECK13-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 14615 // CHECK13-NEXT: [[CMP:%.*]] = icmp slt i32 0, [[TMP3]] 14616 // CHECK13-NEXT: br i1 [[CMP]], label [[SIMD_IF_THEN:%.*]], label [[SIMD_IF_END:%.*]] 14617 // CHECK13: simd.if.then: 14618 // CHECK13-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4 14619 // CHECK13-NEXT: store i32 [[TMP4]], i32* [[DOTOMP_IV]], align 4 14620 // CHECK13-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] 14621 // CHECK13: omp.inner.for.cond: 14622 // CHECK13-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !2 14623 // CHECK13-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4, !llvm.access.group !2 14624 // CHECK13-NEXT: [[CMP4:%.*]] = icmp sle i32 [[TMP5]], [[TMP6]] 14625 // CHECK13-NEXT: br i1 [[CMP4]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] 14626 // CHECK13: omp.inner.for.body: 14627 // CHECK13-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !2 14628 // CHECK13-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP7]], 1 14629 // CHECK13-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] 14630 // CHECK13-NEXT: store i32 [[ADD]], i32* [[I3]], align 4, !llvm.access.group !2 14631 // CHECK13-NEXT: [[TMP8:%.*]] = load double*, double** [[B]], align 8, !llvm.access.group !2 14632 // CHECK13-NEXT: [[TMP9:%.*]] = load i32, i32* [[I3]], align 4, !llvm.access.group !2 14633 // CHECK13-NEXT: [[IDXPROM:%.*]] = sext i32 [[TMP9]] to i64 14634 // CHECK13-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds double, double* [[TMP8]], i64 [[IDXPROM]] 14635 // CHECK13-NEXT: [[TMP10:%.*]] = load double, double* [[ARRAYIDX]], align 8, !llvm.access.group !2 14636 // CHECK13-NEXT: [[TMP11:%.*]] = load double*, double** [[C]], align 8, !llvm.access.group !2 14637 // CHECK13-NEXT: [[TMP12:%.*]] = load i32, i32* [[I3]], align 4, !llvm.access.group !2 14638 // CHECK13-NEXT: [[IDXPROM5:%.*]] = sext i32 [[TMP12]] to i64 14639 // CHECK13-NEXT: [[ARRAYIDX6:%.*]] = getelementptr inbounds double, double* [[TMP11]], i64 [[IDXPROM5]] 14640 // CHECK13-NEXT: [[TMP13:%.*]] = load double, double* [[ARRAYIDX6]], align 8, !llvm.access.group !2 14641 // CHECK13-NEXT: [[ADD7:%.*]] = fadd double [[TMP10]], [[TMP13]] 14642 // CHECK13-NEXT: [[TMP14:%.*]] = load double*, double** [[A]], align 8, !llvm.access.group !2 14643 // CHECK13-NEXT: [[TMP15:%.*]] = load i32, i32* [[I3]], align 4, !llvm.access.group !2 14644 // CHECK13-NEXT: [[IDXPROM8:%.*]] = sext i32 [[TMP15]] to i64 14645 // CHECK13-NEXT: [[ARRAYIDX9:%.*]] = getelementptr inbounds double, double* [[TMP14]], i64 [[IDXPROM8]] 14646 // CHECK13-NEXT: store double [[ADD7]], double* [[ARRAYIDX9]], align 8, !llvm.access.group !2 14647 // CHECK13-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] 14648 // CHECK13: omp.body.continue: 14649 // CHECK13-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] 14650 // CHECK13: omp.inner.for.inc: 14651 // CHECK13-NEXT: [[TMP16:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !2 14652 // CHECK13-NEXT: [[ADD10:%.*]] = add nsw i32 [[TMP16]], 1 14653 // CHECK13-NEXT: store i32 [[ADD10]], i32* [[DOTOMP_IV]], align 4, !llvm.access.group !2 14654 // CHECK13-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP3:![0-9]+]] 14655 // CHECK13: omp.inner.for.end: 14656 // CHECK13-NEXT: [[TMP17:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 14657 // CHECK13-NEXT: [[SUB11:%.*]] = sub nsw i32 [[TMP17]], 0 14658 // CHECK13-NEXT: [[DIV12:%.*]] = sdiv i32 [[SUB11]], 1 14659 // CHECK13-NEXT: [[MUL13:%.*]] = mul nsw i32 [[DIV12]], 1 14660 // CHECK13-NEXT: [[ADD14:%.*]] = add nsw i32 0, [[MUL13]] 14661 // CHECK13-NEXT: store i32 [[ADD14]], i32* [[I3]], align 4 14662 // CHECK13-NEXT: br label [[SIMD_IF_END]] 14663 // CHECK13: simd.if.end: 14664 // CHECK13-NEXT: [[TMP18:%.*]] = load i32, i32* [[N]], align 4 14665 // CHECK13-NEXT: store i32 [[TMP18]], i32* [[DOTCAPTURE_EXPR_16]], align 4 14666 // CHECK13-NEXT: [[TMP19:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_16]], align 4 14667 // CHECK13-NEXT: [[SUB18:%.*]] = sub nsw i32 [[TMP19]], 0 14668 // CHECK13-NEXT: [[DIV19:%.*]] = sdiv i32 [[SUB18]], 1 14669 // CHECK13-NEXT: [[SUB20:%.*]] = sub nsw i32 [[DIV19]], 1 14670 // CHECK13-NEXT: store i32 [[SUB20]], i32* [[DOTCAPTURE_EXPR_17]], align 4 14671 // CHECK13-NEXT: store i32 0, i32* [[DOTOMP_LB21]], align 4 14672 // CHECK13-NEXT: [[TMP20:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_17]], align 4 14673 // CHECK13-NEXT: store i32 [[TMP20]], i32* [[DOTOMP_UB22]], align 4 14674 // CHECK13-NEXT: store i32 0, i32* [[I23]], align 4 14675 // CHECK13-NEXT: [[TMP21:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_16]], align 4 14676 // CHECK13-NEXT: [[CMP24:%.*]] = icmp slt i32 0, [[TMP21]] 14677 // CHECK13-NEXT: br i1 [[CMP24]], label [[SIMD_IF_THEN25:%.*]], label [[SIMD_IF_END48:%.*]] 14678 // CHECK13: simd.if.then25: 14679 // CHECK13-NEXT: [[TMP22:%.*]] = load i32, i32* [[DOTOMP_LB21]], align 4 14680 // CHECK13-NEXT: store i32 [[TMP22]], i32* [[DOTOMP_IV26]], align 4 14681 // CHECK13-NEXT: br label [[OMP_INNER_FOR_COND28:%.*]] 14682 // CHECK13: omp.inner.for.cond28: 14683 // CHECK13-NEXT: [[TMP23:%.*]] = load i32, i32* [[DOTOMP_IV26]], align 4, !llvm.access.group !6 14684 // CHECK13-NEXT: [[TMP24:%.*]] = load i32, i32* [[DOTOMP_UB22]], align 4, !llvm.access.group !6 14685 // CHECK13-NEXT: [[CMP29:%.*]] = icmp sle i32 [[TMP23]], [[TMP24]] 14686 // CHECK13-NEXT: br i1 [[CMP29]], label [[OMP_INNER_FOR_BODY30:%.*]], label [[OMP_INNER_FOR_END43:%.*]] 14687 // CHECK13: omp.inner.for.body30: 14688 // CHECK13-NEXT: [[TMP25:%.*]] = load i32, i32* [[DOTOMP_IV26]], align 4, !llvm.access.group !6 14689 // CHECK13-NEXT: [[MUL31:%.*]] = mul nsw i32 [[TMP25]], 1 14690 // CHECK13-NEXT: [[ADD32:%.*]] = add nsw i32 0, [[MUL31]] 14691 // CHECK13-NEXT: store i32 [[ADD32]], i32* [[I27]], align 4, !llvm.access.group !6 14692 // CHECK13-NEXT: [[TMP26:%.*]] = load double*, double** [[B]], align 8, !llvm.access.group !6 14693 // CHECK13-NEXT: [[TMP27:%.*]] = load i32, i32* [[I27]], align 4, !llvm.access.group !6 14694 // CHECK13-NEXT: [[IDXPROM33:%.*]] = sext i32 [[TMP27]] to i64 14695 // CHECK13-NEXT: [[ARRAYIDX34:%.*]] = getelementptr inbounds double, double* [[TMP26]], i64 [[IDXPROM33]] 14696 // CHECK13-NEXT: [[TMP28:%.*]] = load double, double* [[ARRAYIDX34]], align 8, !llvm.access.group !6 14697 // CHECK13-NEXT: [[TMP29:%.*]] = load double*, double** [[C]], align 8, !llvm.access.group !6 14698 // CHECK13-NEXT: [[TMP30:%.*]] = load i32, i32* [[I27]], align 4, !llvm.access.group !6 14699 // CHECK13-NEXT: [[IDXPROM35:%.*]] = sext i32 [[TMP30]] to i64 14700 // CHECK13-NEXT: [[ARRAYIDX36:%.*]] = getelementptr inbounds double, double* [[TMP29]], i64 [[IDXPROM35]] 14701 // CHECK13-NEXT: [[TMP31:%.*]] = load double, double* [[ARRAYIDX36]], align 8, !llvm.access.group !6 14702 // CHECK13-NEXT: [[ADD37:%.*]] = fadd double [[TMP28]], [[TMP31]] 14703 // CHECK13-NEXT: [[TMP32:%.*]] = load double*, double** [[A]], align 8, !llvm.access.group !6 14704 // CHECK13-NEXT: [[TMP33:%.*]] = load i32, i32* [[I27]], align 4, !llvm.access.group !6 14705 // CHECK13-NEXT: [[IDXPROM38:%.*]] = sext i32 [[TMP33]] to i64 14706 // CHECK13-NEXT: [[ARRAYIDX39:%.*]] = getelementptr inbounds double, double* [[TMP32]], i64 [[IDXPROM38]] 14707 // CHECK13-NEXT: store double [[ADD37]], double* [[ARRAYIDX39]], align 8, !llvm.access.group !6 14708 // CHECK13-NEXT: br label [[OMP_BODY_CONTINUE40:%.*]] 14709 // CHECK13: omp.body.continue40: 14710 // CHECK13-NEXT: br label [[OMP_INNER_FOR_INC41:%.*]] 14711 // CHECK13: omp.inner.for.inc41: 14712 // CHECK13-NEXT: [[TMP34:%.*]] = load i32, i32* [[DOTOMP_IV26]], align 4, !llvm.access.group !6 14713 // CHECK13-NEXT: [[ADD42:%.*]] = add nsw i32 [[TMP34]], 1 14714 // CHECK13-NEXT: store i32 [[ADD42]], i32* [[DOTOMP_IV26]], align 4, !llvm.access.group !6 14715 // CHECK13-NEXT: br label [[OMP_INNER_FOR_COND28]], !llvm.loop [[LOOP7:![0-9]+]] 14716 // CHECK13: omp.inner.for.end43: 14717 // CHECK13-NEXT: [[TMP35:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_16]], align 4 14718 // CHECK13-NEXT: [[SUB44:%.*]] = sub nsw i32 [[TMP35]], 0 14719 // CHECK13-NEXT: [[DIV45:%.*]] = sdiv i32 [[SUB44]], 1 14720 // CHECK13-NEXT: [[MUL46:%.*]] = mul nsw i32 [[DIV45]], 1 14721 // CHECK13-NEXT: [[ADD47:%.*]] = add nsw i32 0, [[MUL46]] 14722 // CHECK13-NEXT: store i32 [[ADD47]], i32* [[I27]], align 4 14723 // CHECK13-NEXT: br label [[SIMD_IF_END48]] 14724 // CHECK13: simd.if.end48: 14725 // CHECK13-NEXT: [[TMP36:%.*]] = load i32, i32* [[N]], align 4 14726 // CHECK13-NEXT: store i32 [[TMP36]], i32* [[DOTCAPTURE_EXPR_50]], align 4 14727 // CHECK13-NEXT: [[TMP37:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_50]], align 4 14728 // CHECK13-NEXT: [[SUB52:%.*]] = sub nsw i32 [[TMP37]], 0 14729 // CHECK13-NEXT: [[DIV53:%.*]] = sdiv i32 [[SUB52]], 1 14730 // CHECK13-NEXT: [[SUB54:%.*]] = sub nsw i32 [[DIV53]], 1 14731 // CHECK13-NEXT: store i32 [[SUB54]], i32* [[DOTCAPTURE_EXPR_51]], align 4 14732 // CHECK13-NEXT: store i32 0, i32* [[DOTOMP_LB55]], align 4 14733 // CHECK13-NEXT: [[TMP38:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_51]], align 4 14734 // CHECK13-NEXT: store i32 [[TMP38]], i32* [[DOTOMP_UB56]], align 4 14735 // CHECK13-NEXT: store i32 0, i32* [[I57]], align 4 14736 // CHECK13-NEXT: [[TMP39:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_50]], align 4 14737 // CHECK13-NEXT: [[CMP58:%.*]] = icmp slt i32 0, [[TMP39]] 14738 // CHECK13-NEXT: br i1 [[CMP58]], label [[SIMD_IF_THEN59:%.*]], label [[SIMD_IF_END82:%.*]] 14739 // CHECK13: simd.if.then59: 14740 // CHECK13-NEXT: [[TMP40:%.*]] = load i32, i32* [[DOTOMP_LB55]], align 4 14741 // CHECK13-NEXT: store i32 [[TMP40]], i32* [[DOTOMP_IV60]], align 4 14742 // CHECK13-NEXT: br label [[OMP_INNER_FOR_COND62:%.*]] 14743 // CHECK13: omp.inner.for.cond62: 14744 // CHECK13-NEXT: [[TMP41:%.*]] = load i32, i32* [[DOTOMP_IV60]], align 4, !llvm.access.group !9 14745 // CHECK13-NEXT: [[TMP42:%.*]] = load i32, i32* [[DOTOMP_UB56]], align 4, !llvm.access.group !9 14746 // CHECK13-NEXT: [[CMP63:%.*]] = icmp sle i32 [[TMP41]], [[TMP42]] 14747 // CHECK13-NEXT: br i1 [[CMP63]], label [[OMP_INNER_FOR_BODY64:%.*]], label [[OMP_INNER_FOR_END77:%.*]] 14748 // CHECK13: omp.inner.for.body64: 14749 // CHECK13-NEXT: [[TMP43:%.*]] = load i32, i32* [[DOTOMP_IV60]], align 4, !llvm.access.group !9 14750 // CHECK13-NEXT: [[MUL65:%.*]] = mul nsw i32 [[TMP43]], 1 14751 // CHECK13-NEXT: [[ADD66:%.*]] = add nsw i32 0, [[MUL65]] 14752 // CHECK13-NEXT: store i32 [[ADD66]], i32* [[I61]], align 4, !llvm.access.group !9 14753 // CHECK13-NEXT: [[TMP44:%.*]] = load double*, double** [[B]], align 8, !llvm.access.group !9 14754 // CHECK13-NEXT: [[TMP45:%.*]] = load i32, i32* [[I61]], align 4, !llvm.access.group !9 14755 // CHECK13-NEXT: [[IDXPROM67:%.*]] = sext i32 [[TMP45]] to i64 14756 // CHECK13-NEXT: [[ARRAYIDX68:%.*]] = getelementptr inbounds double, double* [[TMP44]], i64 [[IDXPROM67]] 14757 // CHECK13-NEXT: [[TMP46:%.*]] = load double, double* [[ARRAYIDX68]], align 8, !llvm.access.group !9 14758 // CHECK13-NEXT: [[TMP47:%.*]] = load double*, double** [[C]], align 8, !llvm.access.group !9 14759 // CHECK13-NEXT: [[TMP48:%.*]] = load i32, i32* [[I61]], align 4, !llvm.access.group !9 14760 // CHECK13-NEXT: [[IDXPROM69:%.*]] = sext i32 [[TMP48]] to i64 14761 // CHECK13-NEXT: [[ARRAYIDX70:%.*]] = getelementptr inbounds double, double* [[TMP47]], i64 [[IDXPROM69]] 14762 // CHECK13-NEXT: [[TMP49:%.*]] = load double, double* [[ARRAYIDX70]], align 8, !llvm.access.group !9 14763 // CHECK13-NEXT: [[ADD71:%.*]] = fadd double [[TMP46]], [[TMP49]] 14764 // CHECK13-NEXT: [[TMP50:%.*]] = load double*, double** [[A]], align 8, !llvm.access.group !9 14765 // CHECK13-NEXT: [[TMP51:%.*]] = load i32, i32* [[I61]], align 4, !llvm.access.group !9 14766 // CHECK13-NEXT: [[IDXPROM72:%.*]] = sext i32 [[TMP51]] to i64 14767 // CHECK13-NEXT: [[ARRAYIDX73:%.*]] = getelementptr inbounds double, double* [[TMP50]], i64 [[IDXPROM72]] 14768 // CHECK13-NEXT: store double [[ADD71]], double* [[ARRAYIDX73]], align 8, !llvm.access.group !9 14769 // CHECK13-NEXT: br label [[OMP_BODY_CONTINUE74:%.*]] 14770 // CHECK13: omp.body.continue74: 14771 // CHECK13-NEXT: br label [[OMP_INNER_FOR_INC75:%.*]] 14772 // CHECK13: omp.inner.for.inc75: 14773 // CHECK13-NEXT: [[TMP52:%.*]] = load i32, i32* [[DOTOMP_IV60]], align 4, !llvm.access.group !9 14774 // CHECK13-NEXT: [[ADD76:%.*]] = add nsw i32 [[TMP52]], 1 14775 // CHECK13-NEXT: store i32 [[ADD76]], i32* [[DOTOMP_IV60]], align 4, !llvm.access.group !9 14776 // CHECK13-NEXT: br label [[OMP_INNER_FOR_COND62]], !llvm.loop [[LOOP10:![0-9]+]] 14777 // CHECK13: omp.inner.for.end77: 14778 // CHECK13-NEXT: [[TMP53:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_50]], align 4 14779 // CHECK13-NEXT: [[SUB78:%.*]] = sub nsw i32 [[TMP53]], 0 14780 // CHECK13-NEXT: [[DIV79:%.*]] = sdiv i32 [[SUB78]], 1 14781 // CHECK13-NEXT: [[MUL80:%.*]] = mul nsw i32 [[DIV79]], 1 14782 // CHECK13-NEXT: [[ADD81:%.*]] = add nsw i32 0, [[MUL80]] 14783 // CHECK13-NEXT: store i32 [[ADD81]], i32* [[I61]], align 4 14784 // CHECK13-NEXT: br label [[SIMD_IF_END82]] 14785 // CHECK13: simd.if.end82: 14786 // CHECK13-NEXT: [[TMP54:%.*]] = load i32, i32* [[N]], align 4 14787 // CHECK13-NEXT: store i32 [[TMP54]], i32* [[DOTCAPTURE_EXPR_84]], align 4 14788 // CHECK13-NEXT: [[TMP55:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_84]], align 4 14789 // CHECK13-NEXT: [[SUB86:%.*]] = sub nsw i32 [[TMP55]], 0 14790 // CHECK13-NEXT: [[DIV87:%.*]] = sdiv i32 [[SUB86]], 1 14791 // CHECK13-NEXT: [[SUB88:%.*]] = sub nsw i32 [[DIV87]], 1 14792 // CHECK13-NEXT: store i32 [[SUB88]], i32* [[DOTCAPTURE_EXPR_85]], align 4 14793 // CHECK13-NEXT: store i32 0, i32* [[DOTOMP_LB89]], align 4 14794 // CHECK13-NEXT: [[TMP56:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_85]], align 4 14795 // CHECK13-NEXT: store i32 [[TMP56]], i32* [[DOTOMP_UB90]], align 4 14796 // CHECK13-NEXT: store i32 0, i32* [[I91]], align 4 14797 // CHECK13-NEXT: [[TMP57:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_84]], align 4 14798 // CHECK13-NEXT: [[CMP92:%.*]] = icmp slt i32 0, [[TMP57]] 14799 // CHECK13-NEXT: br i1 [[CMP92]], label [[SIMD_IF_THEN93:%.*]], label [[SIMD_IF_END116:%.*]] 14800 // CHECK13: simd.if.then93: 14801 // CHECK13-NEXT: [[TMP58:%.*]] = load i32, i32* [[DOTOMP_LB89]], align 4 14802 // CHECK13-NEXT: store i32 [[TMP58]], i32* [[DOTOMP_IV94]], align 4 14803 // CHECK13-NEXT: br label [[OMP_INNER_FOR_COND96:%.*]] 14804 // CHECK13: omp.inner.for.cond96: 14805 // CHECK13-NEXT: [[TMP59:%.*]] = load i32, i32* [[DOTOMP_IV94]], align 4, !llvm.access.group !12 14806 // CHECK13-NEXT: [[TMP60:%.*]] = load i32, i32* [[DOTOMP_UB90]], align 4, !llvm.access.group !12 14807 // CHECK13-NEXT: [[CMP97:%.*]] = icmp sle i32 [[TMP59]], [[TMP60]] 14808 // CHECK13-NEXT: br i1 [[CMP97]], label [[OMP_INNER_FOR_BODY98:%.*]], label [[OMP_INNER_FOR_END111:%.*]] 14809 // CHECK13: omp.inner.for.body98: 14810 // CHECK13-NEXT: [[TMP61:%.*]] = load i32, i32* [[DOTOMP_IV94]], align 4, !llvm.access.group !12 14811 // CHECK13-NEXT: [[MUL99:%.*]] = mul nsw i32 [[TMP61]], 1 14812 // CHECK13-NEXT: [[ADD100:%.*]] = add nsw i32 0, [[MUL99]] 14813 // CHECK13-NEXT: store i32 [[ADD100]], i32* [[I95]], align 4, !llvm.access.group !12 14814 // CHECK13-NEXT: [[TMP62:%.*]] = load double*, double** [[B]], align 8, !llvm.access.group !12 14815 // CHECK13-NEXT: [[TMP63:%.*]] = load i32, i32* [[I95]], align 4, !llvm.access.group !12 14816 // CHECK13-NEXT: [[IDXPROM101:%.*]] = sext i32 [[TMP63]] to i64 14817 // CHECK13-NEXT: [[ARRAYIDX102:%.*]] = getelementptr inbounds double, double* [[TMP62]], i64 [[IDXPROM101]] 14818 // CHECK13-NEXT: [[TMP64:%.*]] = load double, double* [[ARRAYIDX102]], align 8, !llvm.access.group !12 14819 // CHECK13-NEXT: [[TMP65:%.*]] = load double*, double** [[C]], align 8, !llvm.access.group !12 14820 // CHECK13-NEXT: [[TMP66:%.*]] = load i32, i32* [[I95]], align 4, !llvm.access.group !12 14821 // CHECK13-NEXT: [[IDXPROM103:%.*]] = sext i32 [[TMP66]] to i64 14822 // CHECK13-NEXT: [[ARRAYIDX104:%.*]] = getelementptr inbounds double, double* [[TMP65]], i64 [[IDXPROM103]] 14823 // CHECK13-NEXT: [[TMP67:%.*]] = load double, double* [[ARRAYIDX104]], align 8, !llvm.access.group !12 14824 // CHECK13-NEXT: [[ADD105:%.*]] = fadd double [[TMP64]], [[TMP67]] 14825 // CHECK13-NEXT: [[TMP68:%.*]] = load double*, double** [[A]], align 8, !llvm.access.group !12 14826 // CHECK13-NEXT: [[TMP69:%.*]] = load i32, i32* [[I95]], align 4, !llvm.access.group !12 14827 // CHECK13-NEXT: [[IDXPROM106:%.*]] = sext i32 [[TMP69]] to i64 14828 // CHECK13-NEXT: [[ARRAYIDX107:%.*]] = getelementptr inbounds double, double* [[TMP68]], i64 [[IDXPROM106]] 14829 // CHECK13-NEXT: store double [[ADD105]], double* [[ARRAYIDX107]], align 8, !llvm.access.group !12 14830 // CHECK13-NEXT: br label [[OMP_BODY_CONTINUE108:%.*]] 14831 // CHECK13: omp.body.continue108: 14832 // CHECK13-NEXT: br label [[OMP_INNER_FOR_INC109:%.*]] 14833 // CHECK13: omp.inner.for.inc109: 14834 // CHECK13-NEXT: [[TMP70:%.*]] = load i32, i32* [[DOTOMP_IV94]], align 4, !llvm.access.group !12 14835 // CHECK13-NEXT: [[ADD110:%.*]] = add nsw i32 [[TMP70]], 1 14836 // CHECK13-NEXT: store i32 [[ADD110]], i32* [[DOTOMP_IV94]], align 4, !llvm.access.group !12 14837 // CHECK13-NEXT: br label [[OMP_INNER_FOR_COND96]], !llvm.loop [[LOOP13:![0-9]+]] 14838 // CHECK13: omp.inner.for.end111: 14839 // CHECK13-NEXT: [[TMP71:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_84]], align 4 14840 // CHECK13-NEXT: [[SUB112:%.*]] = sub nsw i32 [[TMP71]], 0 14841 // CHECK13-NEXT: [[DIV113:%.*]] = sdiv i32 [[SUB112]], 1 14842 // CHECK13-NEXT: [[MUL114:%.*]] = mul nsw i32 [[DIV113]], 1 14843 // CHECK13-NEXT: [[ADD115:%.*]] = add nsw i32 0, [[MUL114]] 14844 // CHECK13-NEXT: store i32 [[ADD115]], i32* [[I95]], align 4 14845 // CHECK13-NEXT: br label [[SIMD_IF_END116]] 14846 // CHECK13: simd.if.end116: 14847 // CHECK13-NEXT: [[TMP72:%.*]] = load i32, i32* [[CH]], align 4 14848 // CHECK13-NEXT: store i32 [[TMP72]], i32* [[DOTCAPTURE_EXPR_117]], align 4 14849 // CHECK13-NEXT: [[TMP73:%.*]] = load i32, i32* [[N]], align 4 14850 // CHECK13-NEXT: store i32 [[TMP73]], i32* [[DOTCAPTURE_EXPR_119]], align 4 14851 // CHECK13-NEXT: [[TMP74:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_119]], align 4 14852 // CHECK13-NEXT: [[SUB121:%.*]] = sub nsw i32 [[TMP74]], 0 14853 // CHECK13-NEXT: [[DIV122:%.*]] = sdiv i32 [[SUB121]], 1 14854 // CHECK13-NEXT: [[SUB123:%.*]] = sub nsw i32 [[DIV122]], 1 14855 // CHECK13-NEXT: store i32 [[SUB123]], i32* [[DOTCAPTURE_EXPR_120]], align 4 14856 // CHECK13-NEXT: store i32 0, i32* [[DOTOMP_LB124]], align 4 14857 // CHECK13-NEXT: [[TMP75:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_120]], align 4 14858 // CHECK13-NEXT: store i32 [[TMP75]], i32* [[DOTOMP_UB125]], align 4 14859 // CHECK13-NEXT: store i32 0, i32* [[I126]], align 4 14860 // CHECK13-NEXT: [[TMP76:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_119]], align 4 14861 // CHECK13-NEXT: [[CMP127:%.*]] = icmp slt i32 0, [[TMP76]] 14862 // CHECK13-NEXT: br i1 [[CMP127]], label [[SIMD_IF_THEN128:%.*]], label [[SIMD_IF_END151:%.*]] 14863 // CHECK13: simd.if.then128: 14864 // CHECK13-NEXT: [[TMP77:%.*]] = load i32, i32* [[DOTOMP_LB124]], align 4 14865 // CHECK13-NEXT: store i32 [[TMP77]], i32* [[DOTOMP_IV129]], align 4 14866 // CHECK13-NEXT: br label [[OMP_INNER_FOR_COND131:%.*]] 14867 // CHECK13: omp.inner.for.cond131: 14868 // CHECK13-NEXT: [[TMP78:%.*]] = load i32, i32* [[DOTOMP_IV129]], align 4, !llvm.access.group !15 14869 // CHECK13-NEXT: [[TMP79:%.*]] = load i32, i32* [[DOTOMP_UB125]], align 4, !llvm.access.group !15 14870 // CHECK13-NEXT: [[CMP132:%.*]] = icmp sle i32 [[TMP78]], [[TMP79]] 14871 // CHECK13-NEXT: br i1 [[CMP132]], label [[OMP_INNER_FOR_BODY133:%.*]], label [[OMP_INNER_FOR_END146:%.*]] 14872 // CHECK13: omp.inner.for.body133: 14873 // CHECK13-NEXT: [[TMP80:%.*]] = load i32, i32* [[DOTOMP_IV129]], align 4, !llvm.access.group !15 14874 // CHECK13-NEXT: [[MUL134:%.*]] = mul nsw i32 [[TMP80]], 1 14875 // CHECK13-NEXT: [[ADD135:%.*]] = add nsw i32 0, [[MUL134]] 14876 // CHECK13-NEXT: store i32 [[ADD135]], i32* [[I130]], align 4, !llvm.access.group !15 14877 // CHECK13-NEXT: [[TMP81:%.*]] = load double*, double** [[B]], align 8, !llvm.access.group !15 14878 // CHECK13-NEXT: [[TMP82:%.*]] = load i32, i32* [[I130]], align 4, !llvm.access.group !15 14879 // CHECK13-NEXT: [[IDXPROM136:%.*]] = sext i32 [[TMP82]] to i64 14880 // CHECK13-NEXT: [[ARRAYIDX137:%.*]] = getelementptr inbounds double, double* [[TMP81]], i64 [[IDXPROM136]] 14881 // CHECK13-NEXT: [[TMP83:%.*]] = load double, double* [[ARRAYIDX137]], align 8, !llvm.access.group !15 14882 // CHECK13-NEXT: [[TMP84:%.*]] = load double*, double** [[C]], align 8, !llvm.access.group !15 14883 // CHECK13-NEXT: [[TMP85:%.*]] = load i32, i32* [[I130]], align 4, !llvm.access.group !15 14884 // CHECK13-NEXT: [[IDXPROM138:%.*]] = sext i32 [[TMP85]] to i64 14885 // CHECK13-NEXT: [[ARRAYIDX139:%.*]] = getelementptr inbounds double, double* [[TMP84]], i64 [[IDXPROM138]] 14886 // CHECK13-NEXT: [[TMP86:%.*]] = load double, double* [[ARRAYIDX139]], align 8, !llvm.access.group !15 14887 // CHECK13-NEXT: [[ADD140:%.*]] = fadd double [[TMP83]], [[TMP86]] 14888 // CHECK13-NEXT: [[TMP87:%.*]] = load double*, double** [[A]], align 8, !llvm.access.group !15 14889 // CHECK13-NEXT: [[TMP88:%.*]] = load i32, i32* [[I130]], align 4, !llvm.access.group !15 14890 // CHECK13-NEXT: [[IDXPROM141:%.*]] = sext i32 [[TMP88]] to i64 14891 // CHECK13-NEXT: [[ARRAYIDX142:%.*]] = getelementptr inbounds double, double* [[TMP87]], i64 [[IDXPROM141]] 14892 // CHECK13-NEXT: store double [[ADD140]], double* [[ARRAYIDX142]], align 8, !llvm.access.group !15 14893 // CHECK13-NEXT: br label [[OMP_BODY_CONTINUE143:%.*]] 14894 // CHECK13: omp.body.continue143: 14895 // CHECK13-NEXT: br label [[OMP_INNER_FOR_INC144:%.*]] 14896 // CHECK13: omp.inner.for.inc144: 14897 // CHECK13-NEXT: [[TMP89:%.*]] = load i32, i32* [[DOTOMP_IV129]], align 4, !llvm.access.group !15 14898 // CHECK13-NEXT: [[ADD145:%.*]] = add nsw i32 [[TMP89]], 1 14899 // CHECK13-NEXT: store i32 [[ADD145]], i32* [[DOTOMP_IV129]], align 4, !llvm.access.group !15 14900 // CHECK13-NEXT: br label [[OMP_INNER_FOR_COND131]], !llvm.loop [[LOOP16:![0-9]+]] 14901 // CHECK13: omp.inner.for.end146: 14902 // CHECK13-NEXT: [[TMP90:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_119]], align 4 14903 // CHECK13-NEXT: [[SUB147:%.*]] = sub nsw i32 [[TMP90]], 0 14904 // CHECK13-NEXT: [[DIV148:%.*]] = sdiv i32 [[SUB147]], 1 14905 // CHECK13-NEXT: [[MUL149:%.*]] = mul nsw i32 [[DIV148]], 1 14906 // CHECK13-NEXT: [[ADD150:%.*]] = add nsw i32 0, [[MUL149]] 14907 // CHECK13-NEXT: store i32 [[ADD150]], i32* [[I130]], align 4 14908 // CHECK13-NEXT: br label [[SIMD_IF_END151]] 14909 // CHECK13: simd.if.end151: 14910 // CHECK13-NEXT: [[TMP91:%.*]] = load i32, i32* [[N]], align 4 14911 // CHECK13-NEXT: store i32 [[TMP91]], i32* [[DOTCAPTURE_EXPR_153]], align 4 14912 // CHECK13-NEXT: [[TMP92:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_153]], align 4 14913 // CHECK13-NEXT: [[SUB155:%.*]] = sub nsw i32 [[TMP92]], 0 14914 // CHECK13-NEXT: [[DIV156:%.*]] = sdiv i32 [[SUB155]], 1 14915 // CHECK13-NEXT: [[SUB157:%.*]] = sub nsw i32 [[DIV156]], 1 14916 // CHECK13-NEXT: store i32 [[SUB157]], i32* [[DOTCAPTURE_EXPR_154]], align 4 14917 // CHECK13-NEXT: store i32 0, i32* [[DOTOMP_LB158]], align 4 14918 // CHECK13-NEXT: [[TMP93:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_154]], align 4 14919 // CHECK13-NEXT: store i32 [[TMP93]], i32* [[DOTOMP_UB159]], align 4 14920 // CHECK13-NEXT: store i32 0, i32* [[I160]], align 4 14921 // CHECK13-NEXT: [[TMP94:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_153]], align 4 14922 // CHECK13-NEXT: [[CMP161:%.*]] = icmp slt i32 0, [[TMP94]] 14923 // CHECK13-NEXT: br i1 [[CMP161]], label [[SIMD_IF_THEN162:%.*]], label [[SIMD_IF_END185:%.*]] 14924 // CHECK13: simd.if.then162: 14925 // CHECK13-NEXT: [[TMP95:%.*]] = load i32, i32* [[DOTOMP_LB158]], align 4 14926 // CHECK13-NEXT: store i32 [[TMP95]], i32* [[DOTOMP_IV163]], align 4 14927 // CHECK13-NEXT: br label [[OMP_INNER_FOR_COND165:%.*]] 14928 // CHECK13: omp.inner.for.cond165: 14929 // CHECK13-NEXT: [[TMP96:%.*]] = load i32, i32* [[DOTOMP_IV163]], align 4, !llvm.access.group !18 14930 // CHECK13-NEXT: [[TMP97:%.*]] = load i32, i32* [[DOTOMP_UB159]], align 4, !llvm.access.group !18 14931 // CHECK13-NEXT: [[CMP166:%.*]] = icmp sle i32 [[TMP96]], [[TMP97]] 14932 // CHECK13-NEXT: br i1 [[CMP166]], label [[OMP_INNER_FOR_BODY167:%.*]], label [[OMP_INNER_FOR_END180:%.*]] 14933 // CHECK13: omp.inner.for.body167: 14934 // CHECK13-NEXT: [[TMP98:%.*]] = load i32, i32* [[DOTOMP_IV163]], align 4, !llvm.access.group !18 14935 // CHECK13-NEXT: [[MUL168:%.*]] = mul nsw i32 [[TMP98]], 1 14936 // CHECK13-NEXT: [[ADD169:%.*]] = add nsw i32 0, [[MUL168]] 14937 // CHECK13-NEXT: store i32 [[ADD169]], i32* [[I164]], align 4, !llvm.access.group !18 14938 // CHECK13-NEXT: [[TMP99:%.*]] = load double*, double** [[B]], align 8, !llvm.access.group !18 14939 // CHECK13-NEXT: [[TMP100:%.*]] = load i32, i32* [[I164]], align 4, !llvm.access.group !18 14940 // CHECK13-NEXT: [[IDXPROM170:%.*]] = sext i32 [[TMP100]] to i64 14941 // CHECK13-NEXT: [[ARRAYIDX171:%.*]] = getelementptr inbounds double, double* [[TMP99]], i64 [[IDXPROM170]] 14942 // CHECK13-NEXT: [[TMP101:%.*]] = load double, double* [[ARRAYIDX171]], align 8, !llvm.access.group !18 14943 // CHECK13-NEXT: [[TMP102:%.*]] = load double*, double** [[C]], align 8, !llvm.access.group !18 14944 // CHECK13-NEXT: [[TMP103:%.*]] = load i32, i32* [[I164]], align 4, !llvm.access.group !18 14945 // CHECK13-NEXT: [[IDXPROM172:%.*]] = sext i32 [[TMP103]] to i64 14946 // CHECK13-NEXT: [[ARRAYIDX173:%.*]] = getelementptr inbounds double, double* [[TMP102]], i64 [[IDXPROM172]] 14947 // CHECK13-NEXT: [[TMP104:%.*]] = load double, double* [[ARRAYIDX173]], align 8, !llvm.access.group !18 14948 // CHECK13-NEXT: [[ADD174:%.*]] = fadd double [[TMP101]], [[TMP104]] 14949 // CHECK13-NEXT: [[TMP105:%.*]] = load double*, double** [[A]], align 8, !llvm.access.group !18 14950 // CHECK13-NEXT: [[TMP106:%.*]] = load i32, i32* [[I164]], align 4, !llvm.access.group !18 14951 // CHECK13-NEXT: [[IDXPROM175:%.*]] = sext i32 [[TMP106]] to i64 14952 // CHECK13-NEXT: [[ARRAYIDX176:%.*]] = getelementptr inbounds double, double* [[TMP105]], i64 [[IDXPROM175]] 14953 // CHECK13-NEXT: store double [[ADD174]], double* [[ARRAYIDX176]], align 8, !llvm.access.group !18 14954 // CHECK13-NEXT: br label [[OMP_BODY_CONTINUE177:%.*]] 14955 // CHECK13: omp.body.continue177: 14956 // CHECK13-NEXT: br label [[OMP_INNER_FOR_INC178:%.*]] 14957 // CHECK13: omp.inner.for.inc178: 14958 // CHECK13-NEXT: [[TMP107:%.*]] = load i32, i32* [[DOTOMP_IV163]], align 4, !llvm.access.group !18 14959 // CHECK13-NEXT: [[ADD179:%.*]] = add nsw i32 [[TMP107]], 1 14960 // CHECK13-NEXT: store i32 [[ADD179]], i32* [[DOTOMP_IV163]], align 4, !llvm.access.group !18 14961 // CHECK13-NEXT: br label [[OMP_INNER_FOR_COND165]], !llvm.loop [[LOOP19:![0-9]+]] 14962 // CHECK13: omp.inner.for.end180: 14963 // CHECK13-NEXT: [[TMP108:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_153]], align 4 14964 // CHECK13-NEXT: [[SUB181:%.*]] = sub nsw i32 [[TMP108]], 0 14965 // CHECK13-NEXT: [[DIV182:%.*]] = sdiv i32 [[SUB181]], 1 14966 // CHECK13-NEXT: [[MUL183:%.*]] = mul nsw i32 [[DIV182]], 1 14967 // CHECK13-NEXT: [[ADD184:%.*]] = add nsw i32 0, [[MUL183]] 14968 // CHECK13-NEXT: store i32 [[ADD184]], i32* [[I164]], align 4 14969 // CHECK13-NEXT: br label [[SIMD_IF_END185]] 14970 // CHECK13: simd.if.end185: 14971 // CHECK13-NEXT: [[TMP109:%.*]] = load i32, i32* [[CH]], align 4 14972 // CHECK13-NEXT: store i32 [[TMP109]], i32* [[DOTCAPTURE_EXPR_186]], align 4 14973 // CHECK13-NEXT: [[TMP110:%.*]] = load i32, i32* [[N]], align 4 14974 // CHECK13-NEXT: store i32 [[TMP110]], i32* [[DOTCAPTURE_EXPR_188]], align 4 14975 // CHECK13-NEXT: [[TMP111:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_188]], align 4 14976 // CHECK13-NEXT: [[SUB190:%.*]] = sub nsw i32 [[TMP111]], 0 14977 // CHECK13-NEXT: [[DIV191:%.*]] = sdiv i32 [[SUB190]], 1 14978 // CHECK13-NEXT: [[SUB192:%.*]] = sub nsw i32 [[DIV191]], 1 14979 // CHECK13-NEXT: store i32 [[SUB192]], i32* [[DOTCAPTURE_EXPR_189]], align 4 14980 // CHECK13-NEXT: store i32 0, i32* [[DOTOMP_LB193]], align 4 14981 // CHECK13-NEXT: [[TMP112:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_189]], align 4 14982 // CHECK13-NEXT: store i32 [[TMP112]], i32* [[DOTOMP_UB194]], align 4 14983 // CHECK13-NEXT: store i32 0, i32* [[I195]], align 4 14984 // CHECK13-NEXT: [[TMP113:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_188]], align 4 14985 // CHECK13-NEXT: [[CMP196:%.*]] = icmp slt i32 0, [[TMP113]] 14986 // CHECK13-NEXT: br i1 [[CMP196]], label [[SIMD_IF_THEN197:%.*]], label [[SIMD_IF_END220:%.*]] 14987 // CHECK13: simd.if.then197: 14988 // CHECK13-NEXT: [[TMP114:%.*]] = load i32, i32* [[DOTOMP_LB193]], align 4 14989 // CHECK13-NEXT: store i32 [[TMP114]], i32* [[DOTOMP_IV198]], align 4 14990 // CHECK13-NEXT: br label [[OMP_INNER_FOR_COND200:%.*]] 14991 // CHECK13: omp.inner.for.cond200: 14992 // CHECK13-NEXT: [[TMP115:%.*]] = load i32, i32* [[DOTOMP_IV198]], align 4, !llvm.access.group !21 14993 // CHECK13-NEXT: [[TMP116:%.*]] = load i32, i32* [[DOTOMP_UB194]], align 4, !llvm.access.group !21 14994 // CHECK13-NEXT: [[CMP201:%.*]] = icmp sle i32 [[TMP115]], [[TMP116]] 14995 // CHECK13-NEXT: br i1 [[CMP201]], label [[OMP_INNER_FOR_BODY202:%.*]], label [[OMP_INNER_FOR_END215:%.*]] 14996 // CHECK13: omp.inner.for.body202: 14997 // CHECK13-NEXT: [[TMP117:%.*]] = load i32, i32* [[DOTOMP_IV198]], align 4, !llvm.access.group !21 14998 // CHECK13-NEXT: [[MUL203:%.*]] = mul nsw i32 [[TMP117]], 1 14999 // CHECK13-NEXT: [[ADD204:%.*]] = add nsw i32 0, [[MUL203]] 15000 // CHECK13-NEXT: store i32 [[ADD204]], i32* [[I199]], align 4, !llvm.access.group !21 15001 // CHECK13-NEXT: [[TMP118:%.*]] = load double*, double** [[B]], align 8, !llvm.access.group !21 15002 // CHECK13-NEXT: [[TMP119:%.*]] = load i32, i32* [[I199]], align 4, !llvm.access.group !21 15003 // CHECK13-NEXT: [[IDXPROM205:%.*]] = sext i32 [[TMP119]] to i64 15004 // CHECK13-NEXT: [[ARRAYIDX206:%.*]] = getelementptr inbounds double, double* [[TMP118]], i64 [[IDXPROM205]] 15005 // CHECK13-NEXT: [[TMP120:%.*]] = load double, double* [[ARRAYIDX206]], align 8, !llvm.access.group !21 15006 // CHECK13-NEXT: [[TMP121:%.*]] = load double*, double** [[C]], align 8, !llvm.access.group !21 15007 // CHECK13-NEXT: [[TMP122:%.*]] = load i32, i32* [[I199]], align 4, !llvm.access.group !21 15008 // CHECK13-NEXT: [[IDXPROM207:%.*]] = sext i32 [[TMP122]] to i64 15009 // CHECK13-NEXT: [[ARRAYIDX208:%.*]] = getelementptr inbounds double, double* [[TMP121]], i64 [[IDXPROM207]] 15010 // CHECK13-NEXT: [[TMP123:%.*]] = load double, double* [[ARRAYIDX208]], align 8, !llvm.access.group !21 15011 // CHECK13-NEXT: [[ADD209:%.*]] = fadd double [[TMP120]], [[TMP123]] 15012 // CHECK13-NEXT: [[TMP124:%.*]] = load double*, double** [[A]], align 8, !llvm.access.group !21 15013 // CHECK13-NEXT: [[TMP125:%.*]] = load i32, i32* [[I199]], align 4, !llvm.access.group !21 15014 // CHECK13-NEXT: [[IDXPROM210:%.*]] = sext i32 [[TMP125]] to i64 15015 // CHECK13-NEXT: [[ARRAYIDX211:%.*]] = getelementptr inbounds double, double* [[TMP124]], i64 [[IDXPROM210]] 15016 // CHECK13-NEXT: store double [[ADD209]], double* [[ARRAYIDX211]], align 8, !llvm.access.group !21 15017 // CHECK13-NEXT: br label [[OMP_BODY_CONTINUE212:%.*]] 15018 // CHECK13: omp.body.continue212: 15019 // CHECK13-NEXT: br label [[OMP_INNER_FOR_INC213:%.*]] 15020 // CHECK13: omp.inner.for.inc213: 15021 // CHECK13-NEXT: [[TMP126:%.*]] = load i32, i32* [[DOTOMP_IV198]], align 4, !llvm.access.group !21 15022 // CHECK13-NEXT: [[ADD214:%.*]] = add nsw i32 [[TMP126]], 1 15023 // CHECK13-NEXT: store i32 [[ADD214]], i32* [[DOTOMP_IV198]], align 4, !llvm.access.group !21 15024 // CHECK13-NEXT: br label [[OMP_INNER_FOR_COND200]], !llvm.loop [[LOOP22:![0-9]+]] 15025 // CHECK13: omp.inner.for.end215: 15026 // CHECK13-NEXT: [[TMP127:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_188]], align 4 15027 // CHECK13-NEXT: [[SUB216:%.*]] = sub nsw i32 [[TMP127]], 0 15028 // CHECK13-NEXT: [[DIV217:%.*]] = sdiv i32 [[SUB216]], 1 15029 // CHECK13-NEXT: [[MUL218:%.*]] = mul nsw i32 [[DIV217]], 1 15030 // CHECK13-NEXT: [[ADD219:%.*]] = add nsw i32 0, [[MUL218]] 15031 // CHECK13-NEXT: store i32 [[ADD219]], i32* [[I199]], align 4 15032 // CHECK13-NEXT: br label [[SIMD_IF_END220]] 15033 // CHECK13: simd.if.end220: 15034 // CHECK13-NEXT: [[CALL:%.*]] = call noundef signext i32 @_Z5tmainIiET_v() 15035 // CHECK13-NEXT: ret i32 [[CALL]] 15036 // 15037 // 15038 // CHECK13-LABEL: define {{[^@]+}}@_Z5tmainIiET_v 15039 // CHECK13-SAME: () #[[ATTR1:[0-9]+]] comdat { 15040 // CHECK13-NEXT: entry: 15041 // CHECK13-NEXT: [[A:%.*]] = alloca i32*, align 8 15042 // CHECK13-NEXT: [[B:%.*]] = alloca i32*, align 8 15043 // CHECK13-NEXT: [[C:%.*]] = alloca i32*, align 8 15044 // CHECK13-NEXT: [[N:%.*]] = alloca i32, align 4 15045 // CHECK13-NEXT: [[CH:%.*]] = alloca i32, align 4 15046 // CHECK13-NEXT: [[TMP:%.*]] = alloca i32, align 4 15047 // CHECK13-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 15048 // CHECK13-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 15049 // CHECK13-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 15050 // CHECK13-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4 15051 // CHECK13-NEXT: [[I:%.*]] = alloca i32, align 4 15052 // CHECK13-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 15053 // CHECK13-NEXT: [[I3:%.*]] = alloca i32, align 4 15054 // CHECK13-NEXT: [[_TMP15:%.*]] = alloca i32, align 4 15055 // CHECK13-NEXT: [[DOTCAPTURE_EXPR_16:%.*]] = alloca i32, align 4 15056 // CHECK13-NEXT: [[DOTCAPTURE_EXPR_17:%.*]] = alloca i32, align 4 15057 // CHECK13-NEXT: [[DOTOMP_LB21:%.*]] = alloca i32, align 4 15058 // CHECK13-NEXT: [[DOTOMP_UB22:%.*]] = alloca i32, align 4 15059 // CHECK13-NEXT: [[I23:%.*]] = alloca i32, align 4 15060 // CHECK13-NEXT: [[DOTOMP_IV26:%.*]] = alloca i32, align 4 15061 // CHECK13-NEXT: [[I27:%.*]] = alloca i32, align 4 15062 // CHECK13-NEXT: [[_TMP49:%.*]] = alloca i32, align 4 15063 // CHECK13-NEXT: [[DOTCAPTURE_EXPR_50:%.*]] = alloca i32, align 4 15064 // CHECK13-NEXT: [[DOTCAPTURE_EXPR_51:%.*]] = alloca i32, align 4 15065 // CHECK13-NEXT: [[DOTOMP_LB55:%.*]] = alloca i32, align 4 15066 // CHECK13-NEXT: [[DOTOMP_UB56:%.*]] = alloca i32, align 4 15067 // CHECK13-NEXT: [[I57:%.*]] = alloca i32, align 4 15068 // CHECK13-NEXT: [[DOTOMP_IV60:%.*]] = alloca i32, align 4 15069 // CHECK13-NEXT: [[I61:%.*]] = alloca i32, align 4 15070 // CHECK13-NEXT: [[_TMP83:%.*]] = alloca i32, align 4 15071 // CHECK13-NEXT: [[DOTCAPTURE_EXPR_84:%.*]] = alloca i32, align 4 15072 // CHECK13-NEXT: [[DOTCAPTURE_EXPR_85:%.*]] = alloca i32, align 4 15073 // CHECK13-NEXT: [[DOTOMP_LB89:%.*]] = alloca i32, align 4 15074 // CHECK13-NEXT: [[DOTOMP_UB90:%.*]] = alloca i32, align 4 15075 // CHECK13-NEXT: [[I91:%.*]] = alloca i32, align 4 15076 // CHECK13-NEXT: [[DOTOMP_IV94:%.*]] = alloca i32, align 4 15077 // CHECK13-NEXT: [[I95:%.*]] = alloca i32, align 4 15078 // CHECK13-NEXT: [[DOTCAPTURE_EXPR_117:%.*]] = alloca i32, align 4 15079 // CHECK13-NEXT: [[_TMP118:%.*]] = alloca i32, align 4 15080 // CHECK13-NEXT: [[DOTCAPTURE_EXPR_119:%.*]] = alloca i32, align 4 15081 // CHECK13-NEXT: [[DOTCAPTURE_EXPR_120:%.*]] = alloca i32, align 4 15082 // CHECK13-NEXT: [[DOTOMP_LB124:%.*]] = alloca i32, align 4 15083 // CHECK13-NEXT: [[DOTOMP_UB125:%.*]] = alloca i32, align 4 15084 // CHECK13-NEXT: [[I126:%.*]] = alloca i32, align 4 15085 // CHECK13-NEXT: [[DOTOMP_IV129:%.*]] = alloca i32, align 4 15086 // CHECK13-NEXT: [[I130:%.*]] = alloca i32, align 4 15087 // CHECK13-NEXT: [[_TMP152:%.*]] = alloca i32, align 4 15088 // CHECK13-NEXT: [[DOTCAPTURE_EXPR_153:%.*]] = alloca i32, align 4 15089 // CHECK13-NEXT: [[DOTCAPTURE_EXPR_154:%.*]] = alloca i32, align 4 15090 // CHECK13-NEXT: [[DOTOMP_LB158:%.*]] = alloca i32, align 4 15091 // CHECK13-NEXT: [[DOTOMP_UB159:%.*]] = alloca i32, align 4 15092 // CHECK13-NEXT: [[I160:%.*]] = alloca i32, align 4 15093 // CHECK13-NEXT: [[DOTOMP_IV163:%.*]] = alloca i32, align 4 15094 // CHECK13-NEXT: [[I164:%.*]] = alloca i32, align 4 15095 // CHECK13-NEXT: [[DOTCAPTURE_EXPR_186:%.*]] = alloca i32, align 4 15096 // CHECK13-NEXT: [[_TMP187:%.*]] = alloca i32, align 4 15097 // CHECK13-NEXT: [[DOTCAPTURE_EXPR_188:%.*]] = alloca i32, align 4 15098 // CHECK13-NEXT: [[DOTCAPTURE_EXPR_189:%.*]] = alloca i32, align 4 15099 // CHECK13-NEXT: [[DOTOMP_LB193:%.*]] = alloca i32, align 4 15100 // CHECK13-NEXT: [[DOTOMP_UB194:%.*]] = alloca i32, align 4 15101 // CHECK13-NEXT: [[I195:%.*]] = alloca i32, align 4 15102 // CHECK13-NEXT: [[DOTOMP_IV198:%.*]] = alloca i32, align 4 15103 // CHECK13-NEXT: [[I199:%.*]] = alloca i32, align 4 15104 // CHECK13-NEXT: store i32 10000, i32* [[N]], align 4 15105 // CHECK13-NEXT: store i32 100, i32* [[CH]], align 4 15106 // CHECK13-NEXT: [[TMP0:%.*]] = load i32, i32* [[N]], align 4 15107 // CHECK13-NEXT: store i32 [[TMP0]], i32* [[DOTCAPTURE_EXPR_]], align 4 15108 // CHECK13-NEXT: [[TMP1:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 15109 // CHECK13-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP1]], 0 15110 // CHECK13-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 15111 // CHECK13-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1 15112 // CHECK13-NEXT: store i32 [[SUB2]], i32* [[DOTCAPTURE_EXPR_1]], align 4 15113 // CHECK13-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4 15114 // CHECK13-NEXT: [[TMP2:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 15115 // CHECK13-NEXT: store i32 [[TMP2]], i32* [[DOTOMP_UB]], align 4 15116 // CHECK13-NEXT: store i32 0, i32* [[I]], align 4 15117 // CHECK13-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 15118 // CHECK13-NEXT: [[CMP:%.*]] = icmp slt i32 0, [[TMP3]] 15119 // CHECK13-NEXT: br i1 [[CMP]], label [[SIMD_IF_THEN:%.*]], label [[SIMD_IF_END:%.*]] 15120 // CHECK13: simd.if.then: 15121 // CHECK13-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4 15122 // CHECK13-NEXT: store i32 [[TMP4]], i32* [[DOTOMP_IV]], align 4 15123 // CHECK13-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] 15124 // CHECK13: omp.inner.for.cond: 15125 // CHECK13-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !24 15126 // CHECK13-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4, !llvm.access.group !24 15127 // CHECK13-NEXT: [[CMP4:%.*]] = icmp sle i32 [[TMP5]], [[TMP6]] 15128 // CHECK13-NEXT: br i1 [[CMP4]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] 15129 // CHECK13: omp.inner.for.body: 15130 // CHECK13-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !24 15131 // CHECK13-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP7]], 1 15132 // CHECK13-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] 15133 // CHECK13-NEXT: store i32 [[ADD]], i32* [[I3]], align 4, !llvm.access.group !24 15134 // CHECK13-NEXT: [[TMP8:%.*]] = load i32*, i32** [[B]], align 8, !llvm.access.group !24 15135 // CHECK13-NEXT: [[TMP9:%.*]] = load i32, i32* [[I3]], align 4, !llvm.access.group !24 15136 // CHECK13-NEXT: [[IDXPROM:%.*]] = sext i32 [[TMP9]] to i64 15137 // CHECK13-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i32, i32* [[TMP8]], i64 [[IDXPROM]] 15138 // CHECK13-NEXT: [[TMP10:%.*]] = load i32, i32* [[ARRAYIDX]], align 4, !llvm.access.group !24 15139 // CHECK13-NEXT: [[TMP11:%.*]] = load i32*, i32** [[C]], align 8, !llvm.access.group !24 15140 // CHECK13-NEXT: [[TMP12:%.*]] = load i32, i32* [[I3]], align 4, !llvm.access.group !24 15141 // CHECK13-NEXT: [[IDXPROM5:%.*]] = sext i32 [[TMP12]] to i64 15142 // CHECK13-NEXT: [[ARRAYIDX6:%.*]] = getelementptr inbounds i32, i32* [[TMP11]], i64 [[IDXPROM5]] 15143 // CHECK13-NEXT: [[TMP13:%.*]] = load i32, i32* [[ARRAYIDX6]], align 4, !llvm.access.group !24 15144 // CHECK13-NEXT: [[ADD7:%.*]] = add nsw i32 [[TMP10]], [[TMP13]] 15145 // CHECK13-NEXT: [[TMP14:%.*]] = load i32*, i32** [[A]], align 8, !llvm.access.group !24 15146 // CHECK13-NEXT: [[TMP15:%.*]] = load i32, i32* [[I3]], align 4, !llvm.access.group !24 15147 // CHECK13-NEXT: [[IDXPROM8:%.*]] = sext i32 [[TMP15]] to i64 15148 // CHECK13-NEXT: [[ARRAYIDX9:%.*]] = getelementptr inbounds i32, i32* [[TMP14]], i64 [[IDXPROM8]] 15149 // CHECK13-NEXT: store i32 [[ADD7]], i32* [[ARRAYIDX9]], align 4, !llvm.access.group !24 15150 // CHECK13-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] 15151 // CHECK13: omp.body.continue: 15152 // CHECK13-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] 15153 // CHECK13: omp.inner.for.inc: 15154 // CHECK13-NEXT: [[TMP16:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !24 15155 // CHECK13-NEXT: [[ADD10:%.*]] = add nsw i32 [[TMP16]], 1 15156 // CHECK13-NEXT: store i32 [[ADD10]], i32* [[DOTOMP_IV]], align 4, !llvm.access.group !24 15157 // CHECK13-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP25:![0-9]+]] 15158 // CHECK13: omp.inner.for.end: 15159 // CHECK13-NEXT: [[TMP17:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 15160 // CHECK13-NEXT: [[SUB11:%.*]] = sub nsw i32 [[TMP17]], 0 15161 // CHECK13-NEXT: [[DIV12:%.*]] = sdiv i32 [[SUB11]], 1 15162 // CHECK13-NEXT: [[MUL13:%.*]] = mul nsw i32 [[DIV12]], 1 15163 // CHECK13-NEXT: [[ADD14:%.*]] = add nsw i32 0, [[MUL13]] 15164 // CHECK13-NEXT: store i32 [[ADD14]], i32* [[I3]], align 4 15165 // CHECK13-NEXT: br label [[SIMD_IF_END]] 15166 // CHECK13: simd.if.end: 15167 // CHECK13-NEXT: [[TMP18:%.*]] = load i32, i32* [[N]], align 4 15168 // CHECK13-NEXT: store i32 [[TMP18]], i32* [[DOTCAPTURE_EXPR_16]], align 4 15169 // CHECK13-NEXT: [[TMP19:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_16]], align 4 15170 // CHECK13-NEXT: [[SUB18:%.*]] = sub nsw i32 [[TMP19]], 0 15171 // CHECK13-NEXT: [[DIV19:%.*]] = sdiv i32 [[SUB18]], 1 15172 // CHECK13-NEXT: [[SUB20:%.*]] = sub nsw i32 [[DIV19]], 1 15173 // CHECK13-NEXT: store i32 [[SUB20]], i32* [[DOTCAPTURE_EXPR_17]], align 4 15174 // CHECK13-NEXT: store i32 0, i32* [[DOTOMP_LB21]], align 4 15175 // CHECK13-NEXT: [[TMP20:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_17]], align 4 15176 // CHECK13-NEXT: store i32 [[TMP20]], i32* [[DOTOMP_UB22]], align 4 15177 // CHECK13-NEXT: store i32 0, i32* [[I23]], align 4 15178 // CHECK13-NEXT: [[TMP21:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_16]], align 4 15179 // CHECK13-NEXT: [[CMP24:%.*]] = icmp slt i32 0, [[TMP21]] 15180 // CHECK13-NEXT: br i1 [[CMP24]], label [[SIMD_IF_THEN25:%.*]], label [[SIMD_IF_END48:%.*]] 15181 // CHECK13: simd.if.then25: 15182 // CHECK13-NEXT: [[TMP22:%.*]] = load i32, i32* [[DOTOMP_LB21]], align 4 15183 // CHECK13-NEXT: store i32 [[TMP22]], i32* [[DOTOMP_IV26]], align 4 15184 // CHECK13-NEXT: br label [[OMP_INNER_FOR_COND28:%.*]] 15185 // CHECK13: omp.inner.for.cond28: 15186 // CHECK13-NEXT: [[TMP23:%.*]] = load i32, i32* [[DOTOMP_IV26]], align 4, !llvm.access.group !27 15187 // CHECK13-NEXT: [[TMP24:%.*]] = load i32, i32* [[DOTOMP_UB22]], align 4, !llvm.access.group !27 15188 // CHECK13-NEXT: [[CMP29:%.*]] = icmp sle i32 [[TMP23]], [[TMP24]] 15189 // CHECK13-NEXT: br i1 [[CMP29]], label [[OMP_INNER_FOR_BODY30:%.*]], label [[OMP_INNER_FOR_END43:%.*]] 15190 // CHECK13: omp.inner.for.body30: 15191 // CHECK13-NEXT: [[TMP25:%.*]] = load i32, i32* [[DOTOMP_IV26]], align 4, !llvm.access.group !27 15192 // CHECK13-NEXT: [[MUL31:%.*]] = mul nsw i32 [[TMP25]], 1 15193 // CHECK13-NEXT: [[ADD32:%.*]] = add nsw i32 0, [[MUL31]] 15194 // CHECK13-NEXT: store i32 [[ADD32]], i32* [[I27]], align 4, !llvm.access.group !27 15195 // CHECK13-NEXT: [[TMP26:%.*]] = load i32*, i32** [[B]], align 8, !llvm.access.group !27 15196 // CHECK13-NEXT: [[TMP27:%.*]] = load i32, i32* [[I27]], align 4, !llvm.access.group !27 15197 // CHECK13-NEXT: [[IDXPROM33:%.*]] = sext i32 [[TMP27]] to i64 15198 // CHECK13-NEXT: [[ARRAYIDX34:%.*]] = getelementptr inbounds i32, i32* [[TMP26]], i64 [[IDXPROM33]] 15199 // CHECK13-NEXT: [[TMP28:%.*]] = load i32, i32* [[ARRAYIDX34]], align 4, !llvm.access.group !27 15200 // CHECK13-NEXT: [[TMP29:%.*]] = load i32*, i32** [[C]], align 8, !llvm.access.group !27 15201 // CHECK13-NEXT: [[TMP30:%.*]] = load i32, i32* [[I27]], align 4, !llvm.access.group !27 15202 // CHECK13-NEXT: [[IDXPROM35:%.*]] = sext i32 [[TMP30]] to i64 15203 // CHECK13-NEXT: [[ARRAYIDX36:%.*]] = getelementptr inbounds i32, i32* [[TMP29]], i64 [[IDXPROM35]] 15204 // CHECK13-NEXT: [[TMP31:%.*]] = load i32, i32* [[ARRAYIDX36]], align 4, !llvm.access.group !27 15205 // CHECK13-NEXT: [[ADD37:%.*]] = add nsw i32 [[TMP28]], [[TMP31]] 15206 // CHECK13-NEXT: [[TMP32:%.*]] = load i32*, i32** [[A]], align 8, !llvm.access.group !27 15207 // CHECK13-NEXT: [[TMP33:%.*]] = load i32, i32* [[I27]], align 4, !llvm.access.group !27 15208 // CHECK13-NEXT: [[IDXPROM38:%.*]] = sext i32 [[TMP33]] to i64 15209 // CHECK13-NEXT: [[ARRAYIDX39:%.*]] = getelementptr inbounds i32, i32* [[TMP32]], i64 [[IDXPROM38]] 15210 // CHECK13-NEXT: store i32 [[ADD37]], i32* [[ARRAYIDX39]], align 4, !llvm.access.group !27 15211 // CHECK13-NEXT: br label [[OMP_BODY_CONTINUE40:%.*]] 15212 // CHECK13: omp.body.continue40: 15213 // CHECK13-NEXT: br label [[OMP_INNER_FOR_INC41:%.*]] 15214 // CHECK13: omp.inner.for.inc41: 15215 // CHECK13-NEXT: [[TMP34:%.*]] = load i32, i32* [[DOTOMP_IV26]], align 4, !llvm.access.group !27 15216 // CHECK13-NEXT: [[ADD42:%.*]] = add nsw i32 [[TMP34]], 1 15217 // CHECK13-NEXT: store i32 [[ADD42]], i32* [[DOTOMP_IV26]], align 4, !llvm.access.group !27 15218 // CHECK13-NEXT: br label [[OMP_INNER_FOR_COND28]], !llvm.loop [[LOOP28:![0-9]+]] 15219 // CHECK13: omp.inner.for.end43: 15220 // CHECK13-NEXT: [[TMP35:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_16]], align 4 15221 // CHECK13-NEXT: [[SUB44:%.*]] = sub nsw i32 [[TMP35]], 0 15222 // CHECK13-NEXT: [[DIV45:%.*]] = sdiv i32 [[SUB44]], 1 15223 // CHECK13-NEXT: [[MUL46:%.*]] = mul nsw i32 [[DIV45]], 1 15224 // CHECK13-NEXT: [[ADD47:%.*]] = add nsw i32 0, [[MUL46]] 15225 // CHECK13-NEXT: store i32 [[ADD47]], i32* [[I27]], align 4 15226 // CHECK13-NEXT: br label [[SIMD_IF_END48]] 15227 // CHECK13: simd.if.end48: 15228 // CHECK13-NEXT: [[TMP36:%.*]] = load i32, i32* [[N]], align 4 15229 // CHECK13-NEXT: store i32 [[TMP36]], i32* [[DOTCAPTURE_EXPR_50]], align 4 15230 // CHECK13-NEXT: [[TMP37:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_50]], align 4 15231 // CHECK13-NEXT: [[SUB52:%.*]] = sub nsw i32 [[TMP37]], 0 15232 // CHECK13-NEXT: [[DIV53:%.*]] = sdiv i32 [[SUB52]], 1 15233 // CHECK13-NEXT: [[SUB54:%.*]] = sub nsw i32 [[DIV53]], 1 15234 // CHECK13-NEXT: store i32 [[SUB54]], i32* [[DOTCAPTURE_EXPR_51]], align 4 15235 // CHECK13-NEXT: store i32 0, i32* [[DOTOMP_LB55]], align 4 15236 // CHECK13-NEXT: [[TMP38:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_51]], align 4 15237 // CHECK13-NEXT: store i32 [[TMP38]], i32* [[DOTOMP_UB56]], align 4 15238 // CHECK13-NEXT: store i32 0, i32* [[I57]], align 4 15239 // CHECK13-NEXT: [[TMP39:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_50]], align 4 15240 // CHECK13-NEXT: [[CMP58:%.*]] = icmp slt i32 0, [[TMP39]] 15241 // CHECK13-NEXT: br i1 [[CMP58]], label [[SIMD_IF_THEN59:%.*]], label [[SIMD_IF_END82:%.*]] 15242 // CHECK13: simd.if.then59: 15243 // CHECK13-NEXT: [[TMP40:%.*]] = load i32, i32* [[DOTOMP_LB55]], align 4 15244 // CHECK13-NEXT: store i32 [[TMP40]], i32* [[DOTOMP_IV60]], align 4 15245 // CHECK13-NEXT: br label [[OMP_INNER_FOR_COND62:%.*]] 15246 // CHECK13: omp.inner.for.cond62: 15247 // CHECK13-NEXT: [[TMP41:%.*]] = load i32, i32* [[DOTOMP_IV60]], align 4, !llvm.access.group !30 15248 // CHECK13-NEXT: [[TMP42:%.*]] = load i32, i32* [[DOTOMP_UB56]], align 4, !llvm.access.group !30 15249 // CHECK13-NEXT: [[CMP63:%.*]] = icmp sle i32 [[TMP41]], [[TMP42]] 15250 // CHECK13-NEXT: br i1 [[CMP63]], label [[OMP_INNER_FOR_BODY64:%.*]], label [[OMP_INNER_FOR_END77:%.*]] 15251 // CHECK13: omp.inner.for.body64: 15252 // CHECK13-NEXT: [[TMP43:%.*]] = load i32, i32* [[DOTOMP_IV60]], align 4, !llvm.access.group !30 15253 // CHECK13-NEXT: [[MUL65:%.*]] = mul nsw i32 [[TMP43]], 1 15254 // CHECK13-NEXT: [[ADD66:%.*]] = add nsw i32 0, [[MUL65]] 15255 // CHECK13-NEXT: store i32 [[ADD66]], i32* [[I61]], align 4, !llvm.access.group !30 15256 // CHECK13-NEXT: [[TMP44:%.*]] = load i32*, i32** [[B]], align 8, !llvm.access.group !30 15257 // CHECK13-NEXT: [[TMP45:%.*]] = load i32, i32* [[I61]], align 4, !llvm.access.group !30 15258 // CHECK13-NEXT: [[IDXPROM67:%.*]] = sext i32 [[TMP45]] to i64 15259 // CHECK13-NEXT: [[ARRAYIDX68:%.*]] = getelementptr inbounds i32, i32* [[TMP44]], i64 [[IDXPROM67]] 15260 // CHECK13-NEXT: [[TMP46:%.*]] = load i32, i32* [[ARRAYIDX68]], align 4, !llvm.access.group !30 15261 // CHECK13-NEXT: [[TMP47:%.*]] = load i32*, i32** [[C]], align 8, !llvm.access.group !30 15262 // CHECK13-NEXT: [[TMP48:%.*]] = load i32, i32* [[I61]], align 4, !llvm.access.group !30 15263 // CHECK13-NEXT: [[IDXPROM69:%.*]] = sext i32 [[TMP48]] to i64 15264 // CHECK13-NEXT: [[ARRAYIDX70:%.*]] = getelementptr inbounds i32, i32* [[TMP47]], i64 [[IDXPROM69]] 15265 // CHECK13-NEXT: [[TMP49:%.*]] = load i32, i32* [[ARRAYIDX70]], align 4, !llvm.access.group !30 15266 // CHECK13-NEXT: [[ADD71:%.*]] = add nsw i32 [[TMP46]], [[TMP49]] 15267 // CHECK13-NEXT: [[TMP50:%.*]] = load i32*, i32** [[A]], align 8, !llvm.access.group !30 15268 // CHECK13-NEXT: [[TMP51:%.*]] = load i32, i32* [[I61]], align 4, !llvm.access.group !30 15269 // CHECK13-NEXT: [[IDXPROM72:%.*]] = sext i32 [[TMP51]] to i64 15270 // CHECK13-NEXT: [[ARRAYIDX73:%.*]] = getelementptr inbounds i32, i32* [[TMP50]], i64 [[IDXPROM72]] 15271 // CHECK13-NEXT: store i32 [[ADD71]], i32* [[ARRAYIDX73]], align 4, !llvm.access.group !30 15272 // CHECK13-NEXT: br label [[OMP_BODY_CONTINUE74:%.*]] 15273 // CHECK13: omp.body.continue74: 15274 // CHECK13-NEXT: br label [[OMP_INNER_FOR_INC75:%.*]] 15275 // CHECK13: omp.inner.for.inc75: 15276 // CHECK13-NEXT: [[TMP52:%.*]] = load i32, i32* [[DOTOMP_IV60]], align 4, !llvm.access.group !30 15277 // CHECK13-NEXT: [[ADD76:%.*]] = add nsw i32 [[TMP52]], 1 15278 // CHECK13-NEXT: store i32 [[ADD76]], i32* [[DOTOMP_IV60]], align 4, !llvm.access.group !30 15279 // CHECK13-NEXT: br label [[OMP_INNER_FOR_COND62]], !llvm.loop [[LOOP31:![0-9]+]] 15280 // CHECK13: omp.inner.for.end77: 15281 // CHECK13-NEXT: [[TMP53:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_50]], align 4 15282 // CHECK13-NEXT: [[SUB78:%.*]] = sub nsw i32 [[TMP53]], 0 15283 // CHECK13-NEXT: [[DIV79:%.*]] = sdiv i32 [[SUB78]], 1 15284 // CHECK13-NEXT: [[MUL80:%.*]] = mul nsw i32 [[DIV79]], 1 15285 // CHECK13-NEXT: [[ADD81:%.*]] = add nsw i32 0, [[MUL80]] 15286 // CHECK13-NEXT: store i32 [[ADD81]], i32* [[I61]], align 4 15287 // CHECK13-NEXT: br label [[SIMD_IF_END82]] 15288 // CHECK13: simd.if.end82: 15289 // CHECK13-NEXT: [[TMP54:%.*]] = load i32, i32* [[N]], align 4 15290 // CHECK13-NEXT: store i32 [[TMP54]], i32* [[DOTCAPTURE_EXPR_84]], align 4 15291 // CHECK13-NEXT: [[TMP55:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_84]], align 4 15292 // CHECK13-NEXT: [[SUB86:%.*]] = sub nsw i32 [[TMP55]], 0 15293 // CHECK13-NEXT: [[DIV87:%.*]] = sdiv i32 [[SUB86]], 1 15294 // CHECK13-NEXT: [[SUB88:%.*]] = sub nsw i32 [[DIV87]], 1 15295 // CHECK13-NEXT: store i32 [[SUB88]], i32* [[DOTCAPTURE_EXPR_85]], align 4 15296 // CHECK13-NEXT: store i32 0, i32* [[DOTOMP_LB89]], align 4 15297 // CHECK13-NEXT: [[TMP56:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_85]], align 4 15298 // CHECK13-NEXT: store i32 [[TMP56]], i32* [[DOTOMP_UB90]], align 4 15299 // CHECK13-NEXT: store i32 0, i32* [[I91]], align 4 15300 // CHECK13-NEXT: [[TMP57:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_84]], align 4 15301 // CHECK13-NEXT: [[CMP92:%.*]] = icmp slt i32 0, [[TMP57]] 15302 // CHECK13-NEXT: br i1 [[CMP92]], label [[SIMD_IF_THEN93:%.*]], label [[SIMD_IF_END116:%.*]] 15303 // CHECK13: simd.if.then93: 15304 // CHECK13-NEXT: [[TMP58:%.*]] = load i32, i32* [[DOTOMP_LB89]], align 4 15305 // CHECK13-NEXT: store i32 [[TMP58]], i32* [[DOTOMP_IV94]], align 4 15306 // CHECK13-NEXT: br label [[OMP_INNER_FOR_COND96:%.*]] 15307 // CHECK13: omp.inner.for.cond96: 15308 // CHECK13-NEXT: [[TMP59:%.*]] = load i32, i32* [[DOTOMP_IV94]], align 4, !llvm.access.group !33 15309 // CHECK13-NEXT: [[TMP60:%.*]] = load i32, i32* [[DOTOMP_UB90]], align 4, !llvm.access.group !33 15310 // CHECK13-NEXT: [[CMP97:%.*]] = icmp sle i32 [[TMP59]], [[TMP60]] 15311 // CHECK13-NEXT: br i1 [[CMP97]], label [[OMP_INNER_FOR_BODY98:%.*]], label [[OMP_INNER_FOR_END111:%.*]] 15312 // CHECK13: omp.inner.for.body98: 15313 // CHECK13-NEXT: [[TMP61:%.*]] = load i32, i32* [[DOTOMP_IV94]], align 4, !llvm.access.group !33 15314 // CHECK13-NEXT: [[MUL99:%.*]] = mul nsw i32 [[TMP61]], 1 15315 // CHECK13-NEXT: [[ADD100:%.*]] = add nsw i32 0, [[MUL99]] 15316 // CHECK13-NEXT: store i32 [[ADD100]], i32* [[I95]], align 4, !llvm.access.group !33 15317 // CHECK13-NEXT: [[TMP62:%.*]] = load i32*, i32** [[B]], align 8, !llvm.access.group !33 15318 // CHECK13-NEXT: [[TMP63:%.*]] = load i32, i32* [[I95]], align 4, !llvm.access.group !33 15319 // CHECK13-NEXT: [[IDXPROM101:%.*]] = sext i32 [[TMP63]] to i64 15320 // CHECK13-NEXT: [[ARRAYIDX102:%.*]] = getelementptr inbounds i32, i32* [[TMP62]], i64 [[IDXPROM101]] 15321 // CHECK13-NEXT: [[TMP64:%.*]] = load i32, i32* [[ARRAYIDX102]], align 4, !llvm.access.group !33 15322 // CHECK13-NEXT: [[TMP65:%.*]] = load i32*, i32** [[C]], align 8, !llvm.access.group !33 15323 // CHECK13-NEXT: [[TMP66:%.*]] = load i32, i32* [[I95]], align 4, !llvm.access.group !33 15324 // CHECK13-NEXT: [[IDXPROM103:%.*]] = sext i32 [[TMP66]] to i64 15325 // CHECK13-NEXT: [[ARRAYIDX104:%.*]] = getelementptr inbounds i32, i32* [[TMP65]], i64 [[IDXPROM103]] 15326 // CHECK13-NEXT: [[TMP67:%.*]] = load i32, i32* [[ARRAYIDX104]], align 4, !llvm.access.group !33 15327 // CHECK13-NEXT: [[ADD105:%.*]] = add nsw i32 [[TMP64]], [[TMP67]] 15328 // CHECK13-NEXT: [[TMP68:%.*]] = load i32*, i32** [[A]], align 8, !llvm.access.group !33 15329 // CHECK13-NEXT: [[TMP69:%.*]] = load i32, i32* [[I95]], align 4, !llvm.access.group !33 15330 // CHECK13-NEXT: [[IDXPROM106:%.*]] = sext i32 [[TMP69]] to i64 15331 // CHECK13-NEXT: [[ARRAYIDX107:%.*]] = getelementptr inbounds i32, i32* [[TMP68]], i64 [[IDXPROM106]] 15332 // CHECK13-NEXT: store i32 [[ADD105]], i32* [[ARRAYIDX107]], align 4, !llvm.access.group !33 15333 // CHECK13-NEXT: br label [[OMP_BODY_CONTINUE108:%.*]] 15334 // CHECK13: omp.body.continue108: 15335 // CHECK13-NEXT: br label [[OMP_INNER_FOR_INC109:%.*]] 15336 // CHECK13: omp.inner.for.inc109: 15337 // CHECK13-NEXT: [[TMP70:%.*]] = load i32, i32* [[DOTOMP_IV94]], align 4, !llvm.access.group !33 15338 // CHECK13-NEXT: [[ADD110:%.*]] = add nsw i32 [[TMP70]], 1 15339 // CHECK13-NEXT: store i32 [[ADD110]], i32* [[DOTOMP_IV94]], align 4, !llvm.access.group !33 15340 // CHECK13-NEXT: br label [[OMP_INNER_FOR_COND96]], !llvm.loop [[LOOP34:![0-9]+]] 15341 // CHECK13: omp.inner.for.end111: 15342 // CHECK13-NEXT: [[TMP71:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_84]], align 4 15343 // CHECK13-NEXT: [[SUB112:%.*]] = sub nsw i32 [[TMP71]], 0 15344 // CHECK13-NEXT: [[DIV113:%.*]] = sdiv i32 [[SUB112]], 1 15345 // CHECK13-NEXT: [[MUL114:%.*]] = mul nsw i32 [[DIV113]], 1 15346 // CHECK13-NEXT: [[ADD115:%.*]] = add nsw i32 0, [[MUL114]] 15347 // CHECK13-NEXT: store i32 [[ADD115]], i32* [[I95]], align 4 15348 // CHECK13-NEXT: br label [[SIMD_IF_END116]] 15349 // CHECK13: simd.if.end116: 15350 // CHECK13-NEXT: [[TMP72:%.*]] = load i32, i32* [[CH]], align 4 15351 // CHECK13-NEXT: store i32 [[TMP72]], i32* [[DOTCAPTURE_EXPR_117]], align 4 15352 // CHECK13-NEXT: [[TMP73:%.*]] = load i32, i32* [[N]], align 4 15353 // CHECK13-NEXT: store i32 [[TMP73]], i32* [[DOTCAPTURE_EXPR_119]], align 4 15354 // CHECK13-NEXT: [[TMP74:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_119]], align 4 15355 // CHECK13-NEXT: [[SUB121:%.*]] = sub nsw i32 [[TMP74]], 0 15356 // CHECK13-NEXT: [[DIV122:%.*]] = sdiv i32 [[SUB121]], 1 15357 // CHECK13-NEXT: [[SUB123:%.*]] = sub nsw i32 [[DIV122]], 1 15358 // CHECK13-NEXT: store i32 [[SUB123]], i32* [[DOTCAPTURE_EXPR_120]], align 4 15359 // CHECK13-NEXT: store i32 0, i32* [[DOTOMP_LB124]], align 4 15360 // CHECK13-NEXT: [[TMP75:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_120]], align 4 15361 // CHECK13-NEXT: store i32 [[TMP75]], i32* [[DOTOMP_UB125]], align 4 15362 // CHECK13-NEXT: store i32 0, i32* [[I126]], align 4 15363 // CHECK13-NEXT: [[TMP76:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_119]], align 4 15364 // CHECK13-NEXT: [[CMP127:%.*]] = icmp slt i32 0, [[TMP76]] 15365 // CHECK13-NEXT: br i1 [[CMP127]], label [[SIMD_IF_THEN128:%.*]], label [[SIMD_IF_END151:%.*]] 15366 // CHECK13: simd.if.then128: 15367 // CHECK13-NEXT: [[TMP77:%.*]] = load i32, i32* [[DOTOMP_LB124]], align 4 15368 // CHECK13-NEXT: store i32 [[TMP77]], i32* [[DOTOMP_IV129]], align 4 15369 // CHECK13-NEXT: br label [[OMP_INNER_FOR_COND131:%.*]] 15370 // CHECK13: omp.inner.for.cond131: 15371 // CHECK13-NEXT: [[TMP78:%.*]] = load i32, i32* [[DOTOMP_IV129]], align 4, !llvm.access.group !36 15372 // CHECK13-NEXT: [[TMP79:%.*]] = load i32, i32* [[DOTOMP_UB125]], align 4, !llvm.access.group !36 15373 // CHECK13-NEXT: [[CMP132:%.*]] = icmp sle i32 [[TMP78]], [[TMP79]] 15374 // CHECK13-NEXT: br i1 [[CMP132]], label [[OMP_INNER_FOR_BODY133:%.*]], label [[OMP_INNER_FOR_END146:%.*]] 15375 // CHECK13: omp.inner.for.body133: 15376 // CHECK13-NEXT: [[TMP80:%.*]] = load i32, i32* [[DOTOMP_IV129]], align 4, !llvm.access.group !36 15377 // CHECK13-NEXT: [[MUL134:%.*]] = mul nsw i32 [[TMP80]], 1 15378 // CHECK13-NEXT: [[ADD135:%.*]] = add nsw i32 0, [[MUL134]] 15379 // CHECK13-NEXT: store i32 [[ADD135]], i32* [[I130]], align 4, !llvm.access.group !36 15380 // CHECK13-NEXT: [[TMP81:%.*]] = load i32*, i32** [[B]], align 8, !llvm.access.group !36 15381 // CHECK13-NEXT: [[TMP82:%.*]] = load i32, i32* [[I130]], align 4, !llvm.access.group !36 15382 // CHECK13-NEXT: [[IDXPROM136:%.*]] = sext i32 [[TMP82]] to i64 15383 // CHECK13-NEXT: [[ARRAYIDX137:%.*]] = getelementptr inbounds i32, i32* [[TMP81]], i64 [[IDXPROM136]] 15384 // CHECK13-NEXT: [[TMP83:%.*]] = load i32, i32* [[ARRAYIDX137]], align 4, !llvm.access.group !36 15385 // CHECK13-NEXT: [[TMP84:%.*]] = load i32*, i32** [[C]], align 8, !llvm.access.group !36 15386 // CHECK13-NEXT: [[TMP85:%.*]] = load i32, i32* [[I130]], align 4, !llvm.access.group !36 15387 // CHECK13-NEXT: [[IDXPROM138:%.*]] = sext i32 [[TMP85]] to i64 15388 // CHECK13-NEXT: [[ARRAYIDX139:%.*]] = getelementptr inbounds i32, i32* [[TMP84]], i64 [[IDXPROM138]] 15389 // CHECK13-NEXT: [[TMP86:%.*]] = load i32, i32* [[ARRAYIDX139]], align 4, !llvm.access.group !36 15390 // CHECK13-NEXT: [[ADD140:%.*]] = add nsw i32 [[TMP83]], [[TMP86]] 15391 // CHECK13-NEXT: [[TMP87:%.*]] = load i32*, i32** [[A]], align 8, !llvm.access.group !36 15392 // CHECK13-NEXT: [[TMP88:%.*]] = load i32, i32* [[I130]], align 4, !llvm.access.group !36 15393 // CHECK13-NEXT: [[IDXPROM141:%.*]] = sext i32 [[TMP88]] to i64 15394 // CHECK13-NEXT: [[ARRAYIDX142:%.*]] = getelementptr inbounds i32, i32* [[TMP87]], i64 [[IDXPROM141]] 15395 // CHECK13-NEXT: store i32 [[ADD140]], i32* [[ARRAYIDX142]], align 4, !llvm.access.group !36 15396 // CHECK13-NEXT: br label [[OMP_BODY_CONTINUE143:%.*]] 15397 // CHECK13: omp.body.continue143: 15398 // CHECK13-NEXT: br label [[OMP_INNER_FOR_INC144:%.*]] 15399 // CHECK13: omp.inner.for.inc144: 15400 // CHECK13-NEXT: [[TMP89:%.*]] = load i32, i32* [[DOTOMP_IV129]], align 4, !llvm.access.group !36 15401 // CHECK13-NEXT: [[ADD145:%.*]] = add nsw i32 [[TMP89]], 1 15402 // CHECK13-NEXT: store i32 [[ADD145]], i32* [[DOTOMP_IV129]], align 4, !llvm.access.group !36 15403 // CHECK13-NEXT: br label [[OMP_INNER_FOR_COND131]], !llvm.loop [[LOOP37:![0-9]+]] 15404 // CHECK13: omp.inner.for.end146: 15405 // CHECK13-NEXT: [[TMP90:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_119]], align 4 15406 // CHECK13-NEXT: [[SUB147:%.*]] = sub nsw i32 [[TMP90]], 0 15407 // CHECK13-NEXT: [[DIV148:%.*]] = sdiv i32 [[SUB147]], 1 15408 // CHECK13-NEXT: [[MUL149:%.*]] = mul nsw i32 [[DIV148]], 1 15409 // CHECK13-NEXT: [[ADD150:%.*]] = add nsw i32 0, [[MUL149]] 15410 // CHECK13-NEXT: store i32 [[ADD150]], i32* [[I130]], align 4 15411 // CHECK13-NEXT: br label [[SIMD_IF_END151]] 15412 // CHECK13: simd.if.end151: 15413 // CHECK13-NEXT: [[TMP91:%.*]] = load i32, i32* [[N]], align 4 15414 // CHECK13-NEXT: store i32 [[TMP91]], i32* [[DOTCAPTURE_EXPR_153]], align 4 15415 // CHECK13-NEXT: [[TMP92:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_153]], align 4 15416 // CHECK13-NEXT: [[SUB155:%.*]] = sub nsw i32 [[TMP92]], 0 15417 // CHECK13-NEXT: [[DIV156:%.*]] = sdiv i32 [[SUB155]], 1 15418 // CHECK13-NEXT: [[SUB157:%.*]] = sub nsw i32 [[DIV156]], 1 15419 // CHECK13-NEXT: store i32 [[SUB157]], i32* [[DOTCAPTURE_EXPR_154]], align 4 15420 // CHECK13-NEXT: store i32 0, i32* [[DOTOMP_LB158]], align 4 15421 // CHECK13-NEXT: [[TMP93:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_154]], align 4 15422 // CHECK13-NEXT: store i32 [[TMP93]], i32* [[DOTOMP_UB159]], align 4 15423 // CHECK13-NEXT: store i32 0, i32* [[I160]], align 4 15424 // CHECK13-NEXT: [[TMP94:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_153]], align 4 15425 // CHECK13-NEXT: [[CMP161:%.*]] = icmp slt i32 0, [[TMP94]] 15426 // CHECK13-NEXT: br i1 [[CMP161]], label [[SIMD_IF_THEN162:%.*]], label [[SIMD_IF_END185:%.*]] 15427 // CHECK13: simd.if.then162: 15428 // CHECK13-NEXT: [[TMP95:%.*]] = load i32, i32* [[DOTOMP_LB158]], align 4 15429 // CHECK13-NEXT: store i32 [[TMP95]], i32* [[DOTOMP_IV163]], align 4 15430 // CHECK13-NEXT: br label [[OMP_INNER_FOR_COND165:%.*]] 15431 // CHECK13: omp.inner.for.cond165: 15432 // CHECK13-NEXT: [[TMP96:%.*]] = load i32, i32* [[DOTOMP_IV163]], align 4, !llvm.access.group !39 15433 // CHECK13-NEXT: [[TMP97:%.*]] = load i32, i32* [[DOTOMP_UB159]], align 4, !llvm.access.group !39 15434 // CHECK13-NEXT: [[CMP166:%.*]] = icmp sle i32 [[TMP96]], [[TMP97]] 15435 // CHECK13-NEXT: br i1 [[CMP166]], label [[OMP_INNER_FOR_BODY167:%.*]], label [[OMP_INNER_FOR_END180:%.*]] 15436 // CHECK13: omp.inner.for.body167: 15437 // CHECK13-NEXT: [[TMP98:%.*]] = load i32, i32* [[DOTOMP_IV163]], align 4, !llvm.access.group !39 15438 // CHECK13-NEXT: [[MUL168:%.*]] = mul nsw i32 [[TMP98]], 1 15439 // CHECK13-NEXT: [[ADD169:%.*]] = add nsw i32 0, [[MUL168]] 15440 // CHECK13-NEXT: store i32 [[ADD169]], i32* [[I164]], align 4, !llvm.access.group !39 15441 // CHECK13-NEXT: [[TMP99:%.*]] = load i32*, i32** [[B]], align 8, !llvm.access.group !39 15442 // CHECK13-NEXT: [[TMP100:%.*]] = load i32, i32* [[I164]], align 4, !llvm.access.group !39 15443 // CHECK13-NEXT: [[IDXPROM170:%.*]] = sext i32 [[TMP100]] to i64 15444 // CHECK13-NEXT: [[ARRAYIDX171:%.*]] = getelementptr inbounds i32, i32* [[TMP99]], i64 [[IDXPROM170]] 15445 // CHECK13-NEXT: [[TMP101:%.*]] = load i32, i32* [[ARRAYIDX171]], align 4, !llvm.access.group !39 15446 // CHECK13-NEXT: [[TMP102:%.*]] = load i32*, i32** [[C]], align 8, !llvm.access.group !39 15447 // CHECK13-NEXT: [[TMP103:%.*]] = load i32, i32* [[I164]], align 4, !llvm.access.group !39 15448 // CHECK13-NEXT: [[IDXPROM172:%.*]] = sext i32 [[TMP103]] to i64 15449 // CHECK13-NEXT: [[ARRAYIDX173:%.*]] = getelementptr inbounds i32, i32* [[TMP102]], i64 [[IDXPROM172]] 15450 // CHECK13-NEXT: [[TMP104:%.*]] = load i32, i32* [[ARRAYIDX173]], align 4, !llvm.access.group !39 15451 // CHECK13-NEXT: [[ADD174:%.*]] = add nsw i32 [[TMP101]], [[TMP104]] 15452 // CHECK13-NEXT: [[TMP105:%.*]] = load i32*, i32** [[A]], align 8, !llvm.access.group !39 15453 // CHECK13-NEXT: [[TMP106:%.*]] = load i32, i32* [[I164]], align 4, !llvm.access.group !39 15454 // CHECK13-NEXT: [[IDXPROM175:%.*]] = sext i32 [[TMP106]] to i64 15455 // CHECK13-NEXT: [[ARRAYIDX176:%.*]] = getelementptr inbounds i32, i32* [[TMP105]], i64 [[IDXPROM175]] 15456 // CHECK13-NEXT: store i32 [[ADD174]], i32* [[ARRAYIDX176]], align 4, !llvm.access.group !39 15457 // CHECK13-NEXT: br label [[OMP_BODY_CONTINUE177:%.*]] 15458 // CHECK13: omp.body.continue177: 15459 // CHECK13-NEXT: br label [[OMP_INNER_FOR_INC178:%.*]] 15460 // CHECK13: omp.inner.for.inc178: 15461 // CHECK13-NEXT: [[TMP107:%.*]] = load i32, i32* [[DOTOMP_IV163]], align 4, !llvm.access.group !39 15462 // CHECK13-NEXT: [[ADD179:%.*]] = add nsw i32 [[TMP107]], 1 15463 // CHECK13-NEXT: store i32 [[ADD179]], i32* [[DOTOMP_IV163]], align 4, !llvm.access.group !39 15464 // CHECK13-NEXT: br label [[OMP_INNER_FOR_COND165]], !llvm.loop [[LOOP40:![0-9]+]] 15465 // CHECK13: omp.inner.for.end180: 15466 // CHECK13-NEXT: [[TMP108:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_153]], align 4 15467 // CHECK13-NEXT: [[SUB181:%.*]] = sub nsw i32 [[TMP108]], 0 15468 // CHECK13-NEXT: [[DIV182:%.*]] = sdiv i32 [[SUB181]], 1 15469 // CHECK13-NEXT: [[MUL183:%.*]] = mul nsw i32 [[DIV182]], 1 15470 // CHECK13-NEXT: [[ADD184:%.*]] = add nsw i32 0, [[MUL183]] 15471 // CHECK13-NEXT: store i32 [[ADD184]], i32* [[I164]], align 4 15472 // CHECK13-NEXT: br label [[SIMD_IF_END185]] 15473 // CHECK13: simd.if.end185: 15474 // CHECK13-NEXT: [[TMP109:%.*]] = load i32, i32* [[CH]], align 4 15475 // CHECK13-NEXT: store i32 [[TMP109]], i32* [[DOTCAPTURE_EXPR_186]], align 4 15476 // CHECK13-NEXT: [[TMP110:%.*]] = load i32, i32* [[N]], align 4 15477 // CHECK13-NEXT: store i32 [[TMP110]], i32* [[DOTCAPTURE_EXPR_188]], align 4 15478 // CHECK13-NEXT: [[TMP111:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_188]], align 4 15479 // CHECK13-NEXT: [[SUB190:%.*]] = sub nsw i32 [[TMP111]], 0 15480 // CHECK13-NEXT: [[DIV191:%.*]] = sdiv i32 [[SUB190]], 1 15481 // CHECK13-NEXT: [[SUB192:%.*]] = sub nsw i32 [[DIV191]], 1 15482 // CHECK13-NEXT: store i32 [[SUB192]], i32* [[DOTCAPTURE_EXPR_189]], align 4 15483 // CHECK13-NEXT: store i32 0, i32* [[DOTOMP_LB193]], align 4 15484 // CHECK13-NEXT: [[TMP112:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_189]], align 4 15485 // CHECK13-NEXT: store i32 [[TMP112]], i32* [[DOTOMP_UB194]], align 4 15486 // CHECK13-NEXT: store i32 0, i32* [[I195]], align 4 15487 // CHECK13-NEXT: [[TMP113:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_188]], align 4 15488 // CHECK13-NEXT: [[CMP196:%.*]] = icmp slt i32 0, [[TMP113]] 15489 // CHECK13-NEXT: br i1 [[CMP196]], label [[SIMD_IF_THEN197:%.*]], label [[SIMD_IF_END220:%.*]] 15490 // CHECK13: simd.if.then197: 15491 // CHECK13-NEXT: [[TMP114:%.*]] = load i32, i32* [[DOTOMP_LB193]], align 4 15492 // CHECK13-NEXT: store i32 [[TMP114]], i32* [[DOTOMP_IV198]], align 4 15493 // CHECK13-NEXT: br label [[OMP_INNER_FOR_COND200:%.*]] 15494 // CHECK13: omp.inner.for.cond200: 15495 // CHECK13-NEXT: [[TMP115:%.*]] = load i32, i32* [[DOTOMP_IV198]], align 4, !llvm.access.group !42 15496 // CHECK13-NEXT: [[TMP116:%.*]] = load i32, i32* [[DOTOMP_UB194]], align 4, !llvm.access.group !42 15497 // CHECK13-NEXT: [[CMP201:%.*]] = icmp sle i32 [[TMP115]], [[TMP116]] 15498 // CHECK13-NEXT: br i1 [[CMP201]], label [[OMP_INNER_FOR_BODY202:%.*]], label [[OMP_INNER_FOR_END215:%.*]] 15499 // CHECK13: omp.inner.for.body202: 15500 // CHECK13-NEXT: [[TMP117:%.*]] = load i32, i32* [[DOTOMP_IV198]], align 4, !llvm.access.group !42 15501 // CHECK13-NEXT: [[MUL203:%.*]] = mul nsw i32 [[TMP117]], 1 15502 // CHECK13-NEXT: [[ADD204:%.*]] = add nsw i32 0, [[MUL203]] 15503 // CHECK13-NEXT: store i32 [[ADD204]], i32* [[I199]], align 4, !llvm.access.group !42 15504 // CHECK13-NEXT: [[TMP118:%.*]] = load i32*, i32** [[B]], align 8, !llvm.access.group !42 15505 // CHECK13-NEXT: [[TMP119:%.*]] = load i32, i32* [[I199]], align 4, !llvm.access.group !42 15506 // CHECK13-NEXT: [[IDXPROM205:%.*]] = sext i32 [[TMP119]] to i64 15507 // CHECK13-NEXT: [[ARRAYIDX206:%.*]] = getelementptr inbounds i32, i32* [[TMP118]], i64 [[IDXPROM205]] 15508 // CHECK13-NEXT: [[TMP120:%.*]] = load i32, i32* [[ARRAYIDX206]], align 4, !llvm.access.group !42 15509 // CHECK13-NEXT: [[TMP121:%.*]] = load i32*, i32** [[C]], align 8, !llvm.access.group !42 15510 // CHECK13-NEXT: [[TMP122:%.*]] = load i32, i32* [[I199]], align 4, !llvm.access.group !42 15511 // CHECK13-NEXT: [[IDXPROM207:%.*]] = sext i32 [[TMP122]] to i64 15512 // CHECK13-NEXT: [[ARRAYIDX208:%.*]] = getelementptr inbounds i32, i32* [[TMP121]], i64 [[IDXPROM207]] 15513 // CHECK13-NEXT: [[TMP123:%.*]] = load i32, i32* [[ARRAYIDX208]], align 4, !llvm.access.group !42 15514 // CHECK13-NEXT: [[ADD209:%.*]] = add nsw i32 [[TMP120]], [[TMP123]] 15515 // CHECK13-NEXT: [[TMP124:%.*]] = load i32*, i32** [[A]], align 8, !llvm.access.group !42 15516 // CHECK13-NEXT: [[TMP125:%.*]] = load i32, i32* [[I199]], align 4, !llvm.access.group !42 15517 // CHECK13-NEXT: [[IDXPROM210:%.*]] = sext i32 [[TMP125]] to i64 15518 // CHECK13-NEXT: [[ARRAYIDX211:%.*]] = getelementptr inbounds i32, i32* [[TMP124]], i64 [[IDXPROM210]] 15519 // CHECK13-NEXT: store i32 [[ADD209]], i32* [[ARRAYIDX211]], align 4, !llvm.access.group !42 15520 // CHECK13-NEXT: br label [[OMP_BODY_CONTINUE212:%.*]] 15521 // CHECK13: omp.body.continue212: 15522 // CHECK13-NEXT: br label [[OMP_INNER_FOR_INC213:%.*]] 15523 // CHECK13: omp.inner.for.inc213: 15524 // CHECK13-NEXT: [[TMP126:%.*]] = load i32, i32* [[DOTOMP_IV198]], align 4, !llvm.access.group !42 15525 // CHECK13-NEXT: [[ADD214:%.*]] = add nsw i32 [[TMP126]], 1 15526 // CHECK13-NEXT: store i32 [[ADD214]], i32* [[DOTOMP_IV198]], align 4, !llvm.access.group !42 15527 // CHECK13-NEXT: br label [[OMP_INNER_FOR_COND200]], !llvm.loop [[LOOP43:![0-9]+]] 15528 // CHECK13: omp.inner.for.end215: 15529 // CHECK13-NEXT: [[TMP127:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_188]], align 4 15530 // CHECK13-NEXT: [[SUB216:%.*]] = sub nsw i32 [[TMP127]], 0 15531 // CHECK13-NEXT: [[DIV217:%.*]] = sdiv i32 [[SUB216]], 1 15532 // CHECK13-NEXT: [[MUL218:%.*]] = mul nsw i32 [[DIV217]], 1 15533 // CHECK13-NEXT: [[ADD219:%.*]] = add nsw i32 0, [[MUL218]] 15534 // CHECK13-NEXT: store i32 [[ADD219]], i32* [[I199]], align 4 15535 // CHECK13-NEXT: br label [[SIMD_IF_END220]] 15536 // CHECK13: simd.if.end220: 15537 // CHECK13-NEXT: ret i32 0 15538 // 15539 // 15540 // CHECK15-LABEL: define {{[^@]+}}@main 15541 // CHECK15-SAME: () #[[ATTR0:[0-9]+]] { 15542 // CHECK15-NEXT: entry: 15543 // CHECK15-NEXT: [[RETVAL:%.*]] = alloca i32, align 4 15544 // CHECK15-NEXT: [[A:%.*]] = alloca double*, align 4 15545 // CHECK15-NEXT: [[B:%.*]] = alloca double*, align 4 15546 // CHECK15-NEXT: [[C:%.*]] = alloca double*, align 4 15547 // CHECK15-NEXT: [[N:%.*]] = alloca i32, align 4 15548 // CHECK15-NEXT: [[CH:%.*]] = alloca i32, align 4 15549 // CHECK15-NEXT: [[TMP:%.*]] = alloca i32, align 4 15550 // CHECK15-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 15551 // CHECK15-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 15552 // CHECK15-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 15553 // CHECK15-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4 15554 // CHECK15-NEXT: [[I:%.*]] = alloca i32, align 4 15555 // CHECK15-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 15556 // CHECK15-NEXT: [[I3:%.*]] = alloca i32, align 4 15557 // CHECK15-NEXT: [[_TMP13:%.*]] = alloca i32, align 4 15558 // CHECK15-NEXT: [[DOTCAPTURE_EXPR_14:%.*]] = alloca i32, align 4 15559 // CHECK15-NEXT: [[DOTCAPTURE_EXPR_15:%.*]] = alloca i32, align 4 15560 // CHECK15-NEXT: [[DOTOMP_LB19:%.*]] = alloca i32, align 4 15561 // CHECK15-NEXT: [[DOTOMP_UB20:%.*]] = alloca i32, align 4 15562 // CHECK15-NEXT: [[I21:%.*]] = alloca i32, align 4 15563 // CHECK15-NEXT: [[DOTOMP_IV24:%.*]] = alloca i32, align 4 15564 // CHECK15-NEXT: [[I25:%.*]] = alloca i32, align 4 15565 // CHECK15-NEXT: [[_TMP44:%.*]] = alloca i32, align 4 15566 // CHECK15-NEXT: [[DOTCAPTURE_EXPR_45:%.*]] = alloca i32, align 4 15567 // CHECK15-NEXT: [[DOTCAPTURE_EXPR_46:%.*]] = alloca i32, align 4 15568 // CHECK15-NEXT: [[DOTOMP_LB50:%.*]] = alloca i32, align 4 15569 // CHECK15-NEXT: [[DOTOMP_UB51:%.*]] = alloca i32, align 4 15570 // CHECK15-NEXT: [[I52:%.*]] = alloca i32, align 4 15571 // CHECK15-NEXT: [[DOTOMP_IV55:%.*]] = alloca i32, align 4 15572 // CHECK15-NEXT: [[I56:%.*]] = alloca i32, align 4 15573 // CHECK15-NEXT: [[_TMP75:%.*]] = alloca i32, align 4 15574 // CHECK15-NEXT: [[DOTCAPTURE_EXPR_76:%.*]] = alloca i32, align 4 15575 // CHECK15-NEXT: [[DOTCAPTURE_EXPR_77:%.*]] = alloca i32, align 4 15576 // CHECK15-NEXT: [[DOTOMP_LB81:%.*]] = alloca i32, align 4 15577 // CHECK15-NEXT: [[DOTOMP_UB82:%.*]] = alloca i32, align 4 15578 // CHECK15-NEXT: [[I83:%.*]] = alloca i32, align 4 15579 // CHECK15-NEXT: [[DOTOMP_IV86:%.*]] = alloca i32, align 4 15580 // CHECK15-NEXT: [[I87:%.*]] = alloca i32, align 4 15581 // CHECK15-NEXT: [[DOTCAPTURE_EXPR_106:%.*]] = alloca i32, align 4 15582 // CHECK15-NEXT: [[_TMP107:%.*]] = alloca i32, align 4 15583 // CHECK15-NEXT: [[DOTCAPTURE_EXPR_108:%.*]] = alloca i32, align 4 15584 // CHECK15-NEXT: [[DOTCAPTURE_EXPR_109:%.*]] = alloca i32, align 4 15585 // CHECK15-NEXT: [[DOTOMP_LB113:%.*]] = alloca i32, align 4 15586 // CHECK15-NEXT: [[DOTOMP_UB114:%.*]] = alloca i32, align 4 15587 // CHECK15-NEXT: [[I115:%.*]] = alloca i32, align 4 15588 // CHECK15-NEXT: [[DOTOMP_IV118:%.*]] = alloca i32, align 4 15589 // CHECK15-NEXT: [[I119:%.*]] = alloca i32, align 4 15590 // CHECK15-NEXT: [[_TMP138:%.*]] = alloca i32, align 4 15591 // CHECK15-NEXT: [[DOTCAPTURE_EXPR_139:%.*]] = alloca i32, align 4 15592 // CHECK15-NEXT: [[DOTCAPTURE_EXPR_140:%.*]] = alloca i32, align 4 15593 // CHECK15-NEXT: [[DOTOMP_LB144:%.*]] = alloca i32, align 4 15594 // CHECK15-NEXT: [[DOTOMP_UB145:%.*]] = alloca i32, align 4 15595 // CHECK15-NEXT: [[I146:%.*]] = alloca i32, align 4 15596 // CHECK15-NEXT: [[DOTOMP_IV149:%.*]] = alloca i32, align 4 15597 // CHECK15-NEXT: [[I150:%.*]] = alloca i32, align 4 15598 // CHECK15-NEXT: [[DOTCAPTURE_EXPR_169:%.*]] = alloca i32, align 4 15599 // CHECK15-NEXT: [[_TMP170:%.*]] = alloca i32, align 4 15600 // CHECK15-NEXT: [[DOTCAPTURE_EXPR_171:%.*]] = alloca i32, align 4 15601 // CHECK15-NEXT: [[DOTCAPTURE_EXPR_172:%.*]] = alloca i32, align 4 15602 // CHECK15-NEXT: [[DOTOMP_LB176:%.*]] = alloca i32, align 4 15603 // CHECK15-NEXT: [[DOTOMP_UB177:%.*]] = alloca i32, align 4 15604 // CHECK15-NEXT: [[I178:%.*]] = alloca i32, align 4 15605 // CHECK15-NEXT: [[DOTOMP_IV181:%.*]] = alloca i32, align 4 15606 // CHECK15-NEXT: [[I182:%.*]] = alloca i32, align 4 15607 // CHECK15-NEXT: store i32 0, i32* [[RETVAL]], align 4 15608 // CHECK15-NEXT: store i32 10000, i32* [[N]], align 4 15609 // CHECK15-NEXT: store i32 100, i32* [[CH]], align 4 15610 // CHECK15-NEXT: [[TMP0:%.*]] = load i32, i32* [[N]], align 4 15611 // CHECK15-NEXT: store i32 [[TMP0]], i32* [[DOTCAPTURE_EXPR_]], align 4 15612 // CHECK15-NEXT: [[TMP1:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 15613 // CHECK15-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP1]], 0 15614 // CHECK15-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 15615 // CHECK15-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1 15616 // CHECK15-NEXT: store i32 [[SUB2]], i32* [[DOTCAPTURE_EXPR_1]], align 4 15617 // CHECK15-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4 15618 // CHECK15-NEXT: [[TMP2:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 15619 // CHECK15-NEXT: store i32 [[TMP2]], i32* [[DOTOMP_UB]], align 4 15620 // CHECK15-NEXT: store i32 0, i32* [[I]], align 4 15621 // CHECK15-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 15622 // CHECK15-NEXT: [[CMP:%.*]] = icmp slt i32 0, [[TMP3]] 15623 // CHECK15-NEXT: br i1 [[CMP]], label [[SIMD_IF_THEN:%.*]], label [[SIMD_IF_END:%.*]] 15624 // CHECK15: simd.if.then: 15625 // CHECK15-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4 15626 // CHECK15-NEXT: store i32 [[TMP4]], i32* [[DOTOMP_IV]], align 4 15627 // CHECK15-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] 15628 // CHECK15: omp.inner.for.cond: 15629 // CHECK15-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !3 15630 // CHECK15-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4, !llvm.access.group !3 15631 // CHECK15-NEXT: [[CMP4:%.*]] = icmp sle i32 [[TMP5]], [[TMP6]] 15632 // CHECK15-NEXT: br i1 [[CMP4]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] 15633 // CHECK15: omp.inner.for.body: 15634 // CHECK15-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !3 15635 // CHECK15-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP7]], 1 15636 // CHECK15-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] 15637 // CHECK15-NEXT: store i32 [[ADD]], i32* [[I3]], align 4, !llvm.access.group !3 15638 // CHECK15-NEXT: [[TMP8:%.*]] = load double*, double** [[B]], align 4, !llvm.access.group !3 15639 // CHECK15-NEXT: [[TMP9:%.*]] = load i32, i32* [[I3]], align 4, !llvm.access.group !3 15640 // CHECK15-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds double, double* [[TMP8]], i32 [[TMP9]] 15641 // CHECK15-NEXT: [[TMP10:%.*]] = load double, double* [[ARRAYIDX]], align 4, !llvm.access.group !3 15642 // CHECK15-NEXT: [[TMP11:%.*]] = load double*, double** [[C]], align 4, !llvm.access.group !3 15643 // CHECK15-NEXT: [[TMP12:%.*]] = load i32, i32* [[I3]], align 4, !llvm.access.group !3 15644 // CHECK15-NEXT: [[ARRAYIDX5:%.*]] = getelementptr inbounds double, double* [[TMP11]], i32 [[TMP12]] 15645 // CHECK15-NEXT: [[TMP13:%.*]] = load double, double* [[ARRAYIDX5]], align 4, !llvm.access.group !3 15646 // CHECK15-NEXT: [[ADD6:%.*]] = fadd double [[TMP10]], [[TMP13]] 15647 // CHECK15-NEXT: [[TMP14:%.*]] = load double*, double** [[A]], align 4, !llvm.access.group !3 15648 // CHECK15-NEXT: [[TMP15:%.*]] = load i32, i32* [[I3]], align 4, !llvm.access.group !3 15649 // CHECK15-NEXT: [[ARRAYIDX7:%.*]] = getelementptr inbounds double, double* [[TMP14]], i32 [[TMP15]] 15650 // CHECK15-NEXT: store double [[ADD6]], double* [[ARRAYIDX7]], align 4, !llvm.access.group !3 15651 // CHECK15-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] 15652 // CHECK15: omp.body.continue: 15653 // CHECK15-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] 15654 // CHECK15: omp.inner.for.inc: 15655 // CHECK15-NEXT: [[TMP16:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !3 15656 // CHECK15-NEXT: [[ADD8:%.*]] = add nsw i32 [[TMP16]], 1 15657 // CHECK15-NEXT: store i32 [[ADD8]], i32* [[DOTOMP_IV]], align 4, !llvm.access.group !3 15658 // CHECK15-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP4:![0-9]+]] 15659 // CHECK15: omp.inner.for.end: 15660 // CHECK15-NEXT: [[TMP17:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 15661 // CHECK15-NEXT: [[SUB9:%.*]] = sub nsw i32 [[TMP17]], 0 15662 // CHECK15-NEXT: [[DIV10:%.*]] = sdiv i32 [[SUB9]], 1 15663 // CHECK15-NEXT: [[MUL11:%.*]] = mul nsw i32 [[DIV10]], 1 15664 // CHECK15-NEXT: [[ADD12:%.*]] = add nsw i32 0, [[MUL11]] 15665 // CHECK15-NEXT: store i32 [[ADD12]], i32* [[I3]], align 4 15666 // CHECK15-NEXT: br label [[SIMD_IF_END]] 15667 // CHECK15: simd.if.end: 15668 // CHECK15-NEXT: [[TMP18:%.*]] = load i32, i32* [[N]], align 4 15669 // CHECK15-NEXT: store i32 [[TMP18]], i32* [[DOTCAPTURE_EXPR_14]], align 4 15670 // CHECK15-NEXT: [[TMP19:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_14]], align 4 15671 // CHECK15-NEXT: [[SUB16:%.*]] = sub nsw i32 [[TMP19]], 0 15672 // CHECK15-NEXT: [[DIV17:%.*]] = sdiv i32 [[SUB16]], 1 15673 // CHECK15-NEXT: [[SUB18:%.*]] = sub nsw i32 [[DIV17]], 1 15674 // CHECK15-NEXT: store i32 [[SUB18]], i32* [[DOTCAPTURE_EXPR_15]], align 4 15675 // CHECK15-NEXT: store i32 0, i32* [[DOTOMP_LB19]], align 4 15676 // CHECK15-NEXT: [[TMP20:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_15]], align 4 15677 // CHECK15-NEXT: store i32 [[TMP20]], i32* [[DOTOMP_UB20]], align 4 15678 // CHECK15-NEXT: store i32 0, i32* [[I21]], align 4 15679 // CHECK15-NEXT: [[TMP21:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_14]], align 4 15680 // CHECK15-NEXT: [[CMP22:%.*]] = icmp slt i32 0, [[TMP21]] 15681 // CHECK15-NEXT: br i1 [[CMP22]], label [[SIMD_IF_THEN23:%.*]], label [[SIMD_IF_END43:%.*]] 15682 // CHECK15: simd.if.then23: 15683 // CHECK15-NEXT: [[TMP22:%.*]] = load i32, i32* [[DOTOMP_LB19]], align 4 15684 // CHECK15-NEXT: store i32 [[TMP22]], i32* [[DOTOMP_IV24]], align 4 15685 // CHECK15-NEXT: br label [[OMP_INNER_FOR_COND26:%.*]] 15686 // CHECK15: omp.inner.for.cond26: 15687 // CHECK15-NEXT: [[TMP23:%.*]] = load i32, i32* [[DOTOMP_IV24]], align 4, !llvm.access.group !7 15688 // CHECK15-NEXT: [[TMP24:%.*]] = load i32, i32* [[DOTOMP_UB20]], align 4, !llvm.access.group !7 15689 // CHECK15-NEXT: [[CMP27:%.*]] = icmp sle i32 [[TMP23]], [[TMP24]] 15690 // CHECK15-NEXT: br i1 [[CMP27]], label [[OMP_INNER_FOR_BODY28:%.*]], label [[OMP_INNER_FOR_END38:%.*]] 15691 // CHECK15: omp.inner.for.body28: 15692 // CHECK15-NEXT: [[TMP25:%.*]] = load i32, i32* [[DOTOMP_IV24]], align 4, !llvm.access.group !7 15693 // CHECK15-NEXT: [[MUL29:%.*]] = mul nsw i32 [[TMP25]], 1 15694 // CHECK15-NEXT: [[ADD30:%.*]] = add nsw i32 0, [[MUL29]] 15695 // CHECK15-NEXT: store i32 [[ADD30]], i32* [[I25]], align 4, !llvm.access.group !7 15696 // CHECK15-NEXT: [[TMP26:%.*]] = load double*, double** [[B]], align 4, !llvm.access.group !7 15697 // CHECK15-NEXT: [[TMP27:%.*]] = load i32, i32* [[I25]], align 4, !llvm.access.group !7 15698 // CHECK15-NEXT: [[ARRAYIDX31:%.*]] = getelementptr inbounds double, double* [[TMP26]], i32 [[TMP27]] 15699 // CHECK15-NEXT: [[TMP28:%.*]] = load double, double* [[ARRAYIDX31]], align 4, !llvm.access.group !7 15700 // CHECK15-NEXT: [[TMP29:%.*]] = load double*, double** [[C]], align 4, !llvm.access.group !7 15701 // CHECK15-NEXT: [[TMP30:%.*]] = load i32, i32* [[I25]], align 4, !llvm.access.group !7 15702 // CHECK15-NEXT: [[ARRAYIDX32:%.*]] = getelementptr inbounds double, double* [[TMP29]], i32 [[TMP30]] 15703 // CHECK15-NEXT: [[TMP31:%.*]] = load double, double* [[ARRAYIDX32]], align 4, !llvm.access.group !7 15704 // CHECK15-NEXT: [[ADD33:%.*]] = fadd double [[TMP28]], [[TMP31]] 15705 // CHECK15-NEXT: [[TMP32:%.*]] = load double*, double** [[A]], align 4, !llvm.access.group !7 15706 // CHECK15-NEXT: [[TMP33:%.*]] = load i32, i32* [[I25]], align 4, !llvm.access.group !7 15707 // CHECK15-NEXT: [[ARRAYIDX34:%.*]] = getelementptr inbounds double, double* [[TMP32]], i32 [[TMP33]] 15708 // CHECK15-NEXT: store double [[ADD33]], double* [[ARRAYIDX34]], align 4, !llvm.access.group !7 15709 // CHECK15-NEXT: br label [[OMP_BODY_CONTINUE35:%.*]] 15710 // CHECK15: omp.body.continue35: 15711 // CHECK15-NEXT: br label [[OMP_INNER_FOR_INC36:%.*]] 15712 // CHECK15: omp.inner.for.inc36: 15713 // CHECK15-NEXT: [[TMP34:%.*]] = load i32, i32* [[DOTOMP_IV24]], align 4, !llvm.access.group !7 15714 // CHECK15-NEXT: [[ADD37:%.*]] = add nsw i32 [[TMP34]], 1 15715 // CHECK15-NEXT: store i32 [[ADD37]], i32* [[DOTOMP_IV24]], align 4, !llvm.access.group !7 15716 // CHECK15-NEXT: br label [[OMP_INNER_FOR_COND26]], !llvm.loop [[LOOP8:![0-9]+]] 15717 // CHECK15: omp.inner.for.end38: 15718 // CHECK15-NEXT: [[TMP35:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_14]], align 4 15719 // CHECK15-NEXT: [[SUB39:%.*]] = sub nsw i32 [[TMP35]], 0 15720 // CHECK15-NEXT: [[DIV40:%.*]] = sdiv i32 [[SUB39]], 1 15721 // CHECK15-NEXT: [[MUL41:%.*]] = mul nsw i32 [[DIV40]], 1 15722 // CHECK15-NEXT: [[ADD42:%.*]] = add nsw i32 0, [[MUL41]] 15723 // CHECK15-NEXT: store i32 [[ADD42]], i32* [[I25]], align 4 15724 // CHECK15-NEXT: br label [[SIMD_IF_END43]] 15725 // CHECK15: simd.if.end43: 15726 // CHECK15-NEXT: [[TMP36:%.*]] = load i32, i32* [[N]], align 4 15727 // CHECK15-NEXT: store i32 [[TMP36]], i32* [[DOTCAPTURE_EXPR_45]], align 4 15728 // CHECK15-NEXT: [[TMP37:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_45]], align 4 15729 // CHECK15-NEXT: [[SUB47:%.*]] = sub nsw i32 [[TMP37]], 0 15730 // CHECK15-NEXT: [[DIV48:%.*]] = sdiv i32 [[SUB47]], 1 15731 // CHECK15-NEXT: [[SUB49:%.*]] = sub nsw i32 [[DIV48]], 1 15732 // CHECK15-NEXT: store i32 [[SUB49]], i32* [[DOTCAPTURE_EXPR_46]], align 4 15733 // CHECK15-NEXT: store i32 0, i32* [[DOTOMP_LB50]], align 4 15734 // CHECK15-NEXT: [[TMP38:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_46]], align 4 15735 // CHECK15-NEXT: store i32 [[TMP38]], i32* [[DOTOMP_UB51]], align 4 15736 // CHECK15-NEXT: store i32 0, i32* [[I52]], align 4 15737 // CHECK15-NEXT: [[TMP39:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_45]], align 4 15738 // CHECK15-NEXT: [[CMP53:%.*]] = icmp slt i32 0, [[TMP39]] 15739 // CHECK15-NEXT: br i1 [[CMP53]], label [[SIMD_IF_THEN54:%.*]], label [[SIMD_IF_END74:%.*]] 15740 // CHECK15: simd.if.then54: 15741 // CHECK15-NEXT: [[TMP40:%.*]] = load i32, i32* [[DOTOMP_LB50]], align 4 15742 // CHECK15-NEXT: store i32 [[TMP40]], i32* [[DOTOMP_IV55]], align 4 15743 // CHECK15-NEXT: br label [[OMP_INNER_FOR_COND57:%.*]] 15744 // CHECK15: omp.inner.for.cond57: 15745 // CHECK15-NEXT: [[TMP41:%.*]] = load i32, i32* [[DOTOMP_IV55]], align 4, !llvm.access.group !10 15746 // CHECK15-NEXT: [[TMP42:%.*]] = load i32, i32* [[DOTOMP_UB51]], align 4, !llvm.access.group !10 15747 // CHECK15-NEXT: [[CMP58:%.*]] = icmp sle i32 [[TMP41]], [[TMP42]] 15748 // CHECK15-NEXT: br i1 [[CMP58]], label [[OMP_INNER_FOR_BODY59:%.*]], label [[OMP_INNER_FOR_END69:%.*]] 15749 // CHECK15: omp.inner.for.body59: 15750 // CHECK15-NEXT: [[TMP43:%.*]] = load i32, i32* [[DOTOMP_IV55]], align 4, !llvm.access.group !10 15751 // CHECK15-NEXT: [[MUL60:%.*]] = mul nsw i32 [[TMP43]], 1 15752 // CHECK15-NEXT: [[ADD61:%.*]] = add nsw i32 0, [[MUL60]] 15753 // CHECK15-NEXT: store i32 [[ADD61]], i32* [[I56]], align 4, !llvm.access.group !10 15754 // CHECK15-NEXT: [[TMP44:%.*]] = load double*, double** [[B]], align 4, !llvm.access.group !10 15755 // CHECK15-NEXT: [[TMP45:%.*]] = load i32, i32* [[I56]], align 4, !llvm.access.group !10 15756 // CHECK15-NEXT: [[ARRAYIDX62:%.*]] = getelementptr inbounds double, double* [[TMP44]], i32 [[TMP45]] 15757 // CHECK15-NEXT: [[TMP46:%.*]] = load double, double* [[ARRAYIDX62]], align 4, !llvm.access.group !10 15758 // CHECK15-NEXT: [[TMP47:%.*]] = load double*, double** [[C]], align 4, !llvm.access.group !10 15759 // CHECK15-NEXT: [[TMP48:%.*]] = load i32, i32* [[I56]], align 4, !llvm.access.group !10 15760 // CHECK15-NEXT: [[ARRAYIDX63:%.*]] = getelementptr inbounds double, double* [[TMP47]], i32 [[TMP48]] 15761 // CHECK15-NEXT: [[TMP49:%.*]] = load double, double* [[ARRAYIDX63]], align 4, !llvm.access.group !10 15762 // CHECK15-NEXT: [[ADD64:%.*]] = fadd double [[TMP46]], [[TMP49]] 15763 // CHECK15-NEXT: [[TMP50:%.*]] = load double*, double** [[A]], align 4, !llvm.access.group !10 15764 // CHECK15-NEXT: [[TMP51:%.*]] = load i32, i32* [[I56]], align 4, !llvm.access.group !10 15765 // CHECK15-NEXT: [[ARRAYIDX65:%.*]] = getelementptr inbounds double, double* [[TMP50]], i32 [[TMP51]] 15766 // CHECK15-NEXT: store double [[ADD64]], double* [[ARRAYIDX65]], align 4, !llvm.access.group !10 15767 // CHECK15-NEXT: br label [[OMP_BODY_CONTINUE66:%.*]] 15768 // CHECK15: omp.body.continue66: 15769 // CHECK15-NEXT: br label [[OMP_INNER_FOR_INC67:%.*]] 15770 // CHECK15: omp.inner.for.inc67: 15771 // CHECK15-NEXT: [[TMP52:%.*]] = load i32, i32* [[DOTOMP_IV55]], align 4, !llvm.access.group !10 15772 // CHECK15-NEXT: [[ADD68:%.*]] = add nsw i32 [[TMP52]], 1 15773 // CHECK15-NEXT: store i32 [[ADD68]], i32* [[DOTOMP_IV55]], align 4, !llvm.access.group !10 15774 // CHECK15-NEXT: br label [[OMP_INNER_FOR_COND57]], !llvm.loop [[LOOP11:![0-9]+]] 15775 // CHECK15: omp.inner.for.end69: 15776 // CHECK15-NEXT: [[TMP53:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_45]], align 4 15777 // CHECK15-NEXT: [[SUB70:%.*]] = sub nsw i32 [[TMP53]], 0 15778 // CHECK15-NEXT: [[DIV71:%.*]] = sdiv i32 [[SUB70]], 1 15779 // CHECK15-NEXT: [[MUL72:%.*]] = mul nsw i32 [[DIV71]], 1 15780 // CHECK15-NEXT: [[ADD73:%.*]] = add nsw i32 0, [[MUL72]] 15781 // CHECK15-NEXT: store i32 [[ADD73]], i32* [[I56]], align 4 15782 // CHECK15-NEXT: br label [[SIMD_IF_END74]] 15783 // CHECK15: simd.if.end74: 15784 // CHECK15-NEXT: [[TMP54:%.*]] = load i32, i32* [[N]], align 4 15785 // CHECK15-NEXT: store i32 [[TMP54]], i32* [[DOTCAPTURE_EXPR_76]], align 4 15786 // CHECK15-NEXT: [[TMP55:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_76]], align 4 15787 // CHECK15-NEXT: [[SUB78:%.*]] = sub nsw i32 [[TMP55]], 0 15788 // CHECK15-NEXT: [[DIV79:%.*]] = sdiv i32 [[SUB78]], 1 15789 // CHECK15-NEXT: [[SUB80:%.*]] = sub nsw i32 [[DIV79]], 1 15790 // CHECK15-NEXT: store i32 [[SUB80]], i32* [[DOTCAPTURE_EXPR_77]], align 4 15791 // CHECK15-NEXT: store i32 0, i32* [[DOTOMP_LB81]], align 4 15792 // CHECK15-NEXT: [[TMP56:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_77]], align 4 15793 // CHECK15-NEXT: store i32 [[TMP56]], i32* [[DOTOMP_UB82]], align 4 15794 // CHECK15-NEXT: store i32 0, i32* [[I83]], align 4 15795 // CHECK15-NEXT: [[TMP57:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_76]], align 4 15796 // CHECK15-NEXT: [[CMP84:%.*]] = icmp slt i32 0, [[TMP57]] 15797 // CHECK15-NEXT: br i1 [[CMP84]], label [[SIMD_IF_THEN85:%.*]], label [[SIMD_IF_END105:%.*]] 15798 // CHECK15: simd.if.then85: 15799 // CHECK15-NEXT: [[TMP58:%.*]] = load i32, i32* [[DOTOMP_LB81]], align 4 15800 // CHECK15-NEXT: store i32 [[TMP58]], i32* [[DOTOMP_IV86]], align 4 15801 // CHECK15-NEXT: br label [[OMP_INNER_FOR_COND88:%.*]] 15802 // CHECK15: omp.inner.for.cond88: 15803 // CHECK15-NEXT: [[TMP59:%.*]] = load i32, i32* [[DOTOMP_IV86]], align 4, !llvm.access.group !13 15804 // CHECK15-NEXT: [[TMP60:%.*]] = load i32, i32* [[DOTOMP_UB82]], align 4, !llvm.access.group !13 15805 // CHECK15-NEXT: [[CMP89:%.*]] = icmp sle i32 [[TMP59]], [[TMP60]] 15806 // CHECK15-NEXT: br i1 [[CMP89]], label [[OMP_INNER_FOR_BODY90:%.*]], label [[OMP_INNER_FOR_END100:%.*]] 15807 // CHECK15: omp.inner.for.body90: 15808 // CHECK15-NEXT: [[TMP61:%.*]] = load i32, i32* [[DOTOMP_IV86]], align 4, !llvm.access.group !13 15809 // CHECK15-NEXT: [[MUL91:%.*]] = mul nsw i32 [[TMP61]], 1 15810 // CHECK15-NEXT: [[ADD92:%.*]] = add nsw i32 0, [[MUL91]] 15811 // CHECK15-NEXT: store i32 [[ADD92]], i32* [[I87]], align 4, !llvm.access.group !13 15812 // CHECK15-NEXT: [[TMP62:%.*]] = load double*, double** [[B]], align 4, !llvm.access.group !13 15813 // CHECK15-NEXT: [[TMP63:%.*]] = load i32, i32* [[I87]], align 4, !llvm.access.group !13 15814 // CHECK15-NEXT: [[ARRAYIDX93:%.*]] = getelementptr inbounds double, double* [[TMP62]], i32 [[TMP63]] 15815 // CHECK15-NEXT: [[TMP64:%.*]] = load double, double* [[ARRAYIDX93]], align 4, !llvm.access.group !13 15816 // CHECK15-NEXT: [[TMP65:%.*]] = load double*, double** [[C]], align 4, !llvm.access.group !13 15817 // CHECK15-NEXT: [[TMP66:%.*]] = load i32, i32* [[I87]], align 4, !llvm.access.group !13 15818 // CHECK15-NEXT: [[ARRAYIDX94:%.*]] = getelementptr inbounds double, double* [[TMP65]], i32 [[TMP66]] 15819 // CHECK15-NEXT: [[TMP67:%.*]] = load double, double* [[ARRAYIDX94]], align 4, !llvm.access.group !13 15820 // CHECK15-NEXT: [[ADD95:%.*]] = fadd double [[TMP64]], [[TMP67]] 15821 // CHECK15-NEXT: [[TMP68:%.*]] = load double*, double** [[A]], align 4, !llvm.access.group !13 15822 // CHECK15-NEXT: [[TMP69:%.*]] = load i32, i32* [[I87]], align 4, !llvm.access.group !13 15823 // CHECK15-NEXT: [[ARRAYIDX96:%.*]] = getelementptr inbounds double, double* [[TMP68]], i32 [[TMP69]] 15824 // CHECK15-NEXT: store double [[ADD95]], double* [[ARRAYIDX96]], align 4, !llvm.access.group !13 15825 // CHECK15-NEXT: br label [[OMP_BODY_CONTINUE97:%.*]] 15826 // CHECK15: omp.body.continue97: 15827 // CHECK15-NEXT: br label [[OMP_INNER_FOR_INC98:%.*]] 15828 // CHECK15: omp.inner.for.inc98: 15829 // CHECK15-NEXT: [[TMP70:%.*]] = load i32, i32* [[DOTOMP_IV86]], align 4, !llvm.access.group !13 15830 // CHECK15-NEXT: [[ADD99:%.*]] = add nsw i32 [[TMP70]], 1 15831 // CHECK15-NEXT: store i32 [[ADD99]], i32* [[DOTOMP_IV86]], align 4, !llvm.access.group !13 15832 // CHECK15-NEXT: br label [[OMP_INNER_FOR_COND88]], !llvm.loop [[LOOP14:![0-9]+]] 15833 // CHECK15: omp.inner.for.end100: 15834 // CHECK15-NEXT: [[TMP71:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_76]], align 4 15835 // CHECK15-NEXT: [[SUB101:%.*]] = sub nsw i32 [[TMP71]], 0 15836 // CHECK15-NEXT: [[DIV102:%.*]] = sdiv i32 [[SUB101]], 1 15837 // CHECK15-NEXT: [[MUL103:%.*]] = mul nsw i32 [[DIV102]], 1 15838 // CHECK15-NEXT: [[ADD104:%.*]] = add nsw i32 0, [[MUL103]] 15839 // CHECK15-NEXT: store i32 [[ADD104]], i32* [[I87]], align 4 15840 // CHECK15-NEXT: br label [[SIMD_IF_END105]] 15841 // CHECK15: simd.if.end105: 15842 // CHECK15-NEXT: [[TMP72:%.*]] = load i32, i32* [[CH]], align 4 15843 // CHECK15-NEXT: store i32 [[TMP72]], i32* [[DOTCAPTURE_EXPR_106]], align 4 15844 // CHECK15-NEXT: [[TMP73:%.*]] = load i32, i32* [[N]], align 4 15845 // CHECK15-NEXT: store i32 [[TMP73]], i32* [[DOTCAPTURE_EXPR_108]], align 4 15846 // CHECK15-NEXT: [[TMP74:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_108]], align 4 15847 // CHECK15-NEXT: [[SUB110:%.*]] = sub nsw i32 [[TMP74]], 0 15848 // CHECK15-NEXT: [[DIV111:%.*]] = sdiv i32 [[SUB110]], 1 15849 // CHECK15-NEXT: [[SUB112:%.*]] = sub nsw i32 [[DIV111]], 1 15850 // CHECK15-NEXT: store i32 [[SUB112]], i32* [[DOTCAPTURE_EXPR_109]], align 4 15851 // CHECK15-NEXT: store i32 0, i32* [[DOTOMP_LB113]], align 4 15852 // CHECK15-NEXT: [[TMP75:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_109]], align 4 15853 // CHECK15-NEXT: store i32 [[TMP75]], i32* [[DOTOMP_UB114]], align 4 15854 // CHECK15-NEXT: store i32 0, i32* [[I115]], align 4 15855 // CHECK15-NEXT: [[TMP76:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_108]], align 4 15856 // CHECK15-NEXT: [[CMP116:%.*]] = icmp slt i32 0, [[TMP76]] 15857 // CHECK15-NEXT: br i1 [[CMP116]], label [[SIMD_IF_THEN117:%.*]], label [[SIMD_IF_END137:%.*]] 15858 // CHECK15: simd.if.then117: 15859 // CHECK15-NEXT: [[TMP77:%.*]] = load i32, i32* [[DOTOMP_LB113]], align 4 15860 // CHECK15-NEXT: store i32 [[TMP77]], i32* [[DOTOMP_IV118]], align 4 15861 // CHECK15-NEXT: br label [[OMP_INNER_FOR_COND120:%.*]] 15862 // CHECK15: omp.inner.for.cond120: 15863 // CHECK15-NEXT: [[TMP78:%.*]] = load i32, i32* [[DOTOMP_IV118]], align 4, !llvm.access.group !16 15864 // CHECK15-NEXT: [[TMP79:%.*]] = load i32, i32* [[DOTOMP_UB114]], align 4, !llvm.access.group !16 15865 // CHECK15-NEXT: [[CMP121:%.*]] = icmp sle i32 [[TMP78]], [[TMP79]] 15866 // CHECK15-NEXT: br i1 [[CMP121]], label [[OMP_INNER_FOR_BODY122:%.*]], label [[OMP_INNER_FOR_END132:%.*]] 15867 // CHECK15: omp.inner.for.body122: 15868 // CHECK15-NEXT: [[TMP80:%.*]] = load i32, i32* [[DOTOMP_IV118]], align 4, !llvm.access.group !16 15869 // CHECK15-NEXT: [[MUL123:%.*]] = mul nsw i32 [[TMP80]], 1 15870 // CHECK15-NEXT: [[ADD124:%.*]] = add nsw i32 0, [[MUL123]] 15871 // CHECK15-NEXT: store i32 [[ADD124]], i32* [[I119]], align 4, !llvm.access.group !16 15872 // CHECK15-NEXT: [[TMP81:%.*]] = load double*, double** [[B]], align 4, !llvm.access.group !16 15873 // CHECK15-NEXT: [[TMP82:%.*]] = load i32, i32* [[I119]], align 4, !llvm.access.group !16 15874 // CHECK15-NEXT: [[ARRAYIDX125:%.*]] = getelementptr inbounds double, double* [[TMP81]], i32 [[TMP82]] 15875 // CHECK15-NEXT: [[TMP83:%.*]] = load double, double* [[ARRAYIDX125]], align 4, !llvm.access.group !16 15876 // CHECK15-NEXT: [[TMP84:%.*]] = load double*, double** [[C]], align 4, !llvm.access.group !16 15877 // CHECK15-NEXT: [[TMP85:%.*]] = load i32, i32* [[I119]], align 4, !llvm.access.group !16 15878 // CHECK15-NEXT: [[ARRAYIDX126:%.*]] = getelementptr inbounds double, double* [[TMP84]], i32 [[TMP85]] 15879 // CHECK15-NEXT: [[TMP86:%.*]] = load double, double* [[ARRAYIDX126]], align 4, !llvm.access.group !16 15880 // CHECK15-NEXT: [[ADD127:%.*]] = fadd double [[TMP83]], [[TMP86]] 15881 // CHECK15-NEXT: [[TMP87:%.*]] = load double*, double** [[A]], align 4, !llvm.access.group !16 15882 // CHECK15-NEXT: [[TMP88:%.*]] = load i32, i32* [[I119]], align 4, !llvm.access.group !16 15883 // CHECK15-NEXT: [[ARRAYIDX128:%.*]] = getelementptr inbounds double, double* [[TMP87]], i32 [[TMP88]] 15884 // CHECK15-NEXT: store double [[ADD127]], double* [[ARRAYIDX128]], align 4, !llvm.access.group !16 15885 // CHECK15-NEXT: br label [[OMP_BODY_CONTINUE129:%.*]] 15886 // CHECK15: omp.body.continue129: 15887 // CHECK15-NEXT: br label [[OMP_INNER_FOR_INC130:%.*]] 15888 // CHECK15: omp.inner.for.inc130: 15889 // CHECK15-NEXT: [[TMP89:%.*]] = load i32, i32* [[DOTOMP_IV118]], align 4, !llvm.access.group !16 15890 // CHECK15-NEXT: [[ADD131:%.*]] = add nsw i32 [[TMP89]], 1 15891 // CHECK15-NEXT: store i32 [[ADD131]], i32* [[DOTOMP_IV118]], align 4, !llvm.access.group !16 15892 // CHECK15-NEXT: br label [[OMP_INNER_FOR_COND120]], !llvm.loop [[LOOP17:![0-9]+]] 15893 // CHECK15: omp.inner.for.end132: 15894 // CHECK15-NEXT: [[TMP90:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_108]], align 4 15895 // CHECK15-NEXT: [[SUB133:%.*]] = sub nsw i32 [[TMP90]], 0 15896 // CHECK15-NEXT: [[DIV134:%.*]] = sdiv i32 [[SUB133]], 1 15897 // CHECK15-NEXT: [[MUL135:%.*]] = mul nsw i32 [[DIV134]], 1 15898 // CHECK15-NEXT: [[ADD136:%.*]] = add nsw i32 0, [[MUL135]] 15899 // CHECK15-NEXT: store i32 [[ADD136]], i32* [[I119]], align 4 15900 // CHECK15-NEXT: br label [[SIMD_IF_END137]] 15901 // CHECK15: simd.if.end137: 15902 // CHECK15-NEXT: [[TMP91:%.*]] = load i32, i32* [[N]], align 4 15903 // CHECK15-NEXT: store i32 [[TMP91]], i32* [[DOTCAPTURE_EXPR_139]], align 4 15904 // CHECK15-NEXT: [[TMP92:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_139]], align 4 15905 // CHECK15-NEXT: [[SUB141:%.*]] = sub nsw i32 [[TMP92]], 0 15906 // CHECK15-NEXT: [[DIV142:%.*]] = sdiv i32 [[SUB141]], 1 15907 // CHECK15-NEXT: [[SUB143:%.*]] = sub nsw i32 [[DIV142]], 1 15908 // CHECK15-NEXT: store i32 [[SUB143]], i32* [[DOTCAPTURE_EXPR_140]], align 4 15909 // CHECK15-NEXT: store i32 0, i32* [[DOTOMP_LB144]], align 4 15910 // CHECK15-NEXT: [[TMP93:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_140]], align 4 15911 // CHECK15-NEXT: store i32 [[TMP93]], i32* [[DOTOMP_UB145]], align 4 15912 // CHECK15-NEXT: store i32 0, i32* [[I146]], align 4 15913 // CHECK15-NEXT: [[TMP94:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_139]], align 4 15914 // CHECK15-NEXT: [[CMP147:%.*]] = icmp slt i32 0, [[TMP94]] 15915 // CHECK15-NEXT: br i1 [[CMP147]], label [[SIMD_IF_THEN148:%.*]], label [[SIMD_IF_END168:%.*]] 15916 // CHECK15: simd.if.then148: 15917 // CHECK15-NEXT: [[TMP95:%.*]] = load i32, i32* [[DOTOMP_LB144]], align 4 15918 // CHECK15-NEXT: store i32 [[TMP95]], i32* [[DOTOMP_IV149]], align 4 15919 // CHECK15-NEXT: br label [[OMP_INNER_FOR_COND151:%.*]] 15920 // CHECK15: omp.inner.for.cond151: 15921 // CHECK15-NEXT: [[TMP96:%.*]] = load i32, i32* [[DOTOMP_IV149]], align 4, !llvm.access.group !19 15922 // CHECK15-NEXT: [[TMP97:%.*]] = load i32, i32* [[DOTOMP_UB145]], align 4, !llvm.access.group !19 15923 // CHECK15-NEXT: [[CMP152:%.*]] = icmp sle i32 [[TMP96]], [[TMP97]] 15924 // CHECK15-NEXT: br i1 [[CMP152]], label [[OMP_INNER_FOR_BODY153:%.*]], label [[OMP_INNER_FOR_END163:%.*]] 15925 // CHECK15: omp.inner.for.body153: 15926 // CHECK15-NEXT: [[TMP98:%.*]] = load i32, i32* [[DOTOMP_IV149]], align 4, !llvm.access.group !19 15927 // CHECK15-NEXT: [[MUL154:%.*]] = mul nsw i32 [[TMP98]], 1 15928 // CHECK15-NEXT: [[ADD155:%.*]] = add nsw i32 0, [[MUL154]] 15929 // CHECK15-NEXT: store i32 [[ADD155]], i32* [[I150]], align 4, !llvm.access.group !19 15930 // CHECK15-NEXT: [[TMP99:%.*]] = load double*, double** [[B]], align 4, !llvm.access.group !19 15931 // CHECK15-NEXT: [[TMP100:%.*]] = load i32, i32* [[I150]], align 4, !llvm.access.group !19 15932 // CHECK15-NEXT: [[ARRAYIDX156:%.*]] = getelementptr inbounds double, double* [[TMP99]], i32 [[TMP100]] 15933 // CHECK15-NEXT: [[TMP101:%.*]] = load double, double* [[ARRAYIDX156]], align 4, !llvm.access.group !19 15934 // CHECK15-NEXT: [[TMP102:%.*]] = load double*, double** [[C]], align 4, !llvm.access.group !19 15935 // CHECK15-NEXT: [[TMP103:%.*]] = load i32, i32* [[I150]], align 4, !llvm.access.group !19 15936 // CHECK15-NEXT: [[ARRAYIDX157:%.*]] = getelementptr inbounds double, double* [[TMP102]], i32 [[TMP103]] 15937 // CHECK15-NEXT: [[TMP104:%.*]] = load double, double* [[ARRAYIDX157]], align 4, !llvm.access.group !19 15938 // CHECK15-NEXT: [[ADD158:%.*]] = fadd double [[TMP101]], [[TMP104]] 15939 // CHECK15-NEXT: [[TMP105:%.*]] = load double*, double** [[A]], align 4, !llvm.access.group !19 15940 // CHECK15-NEXT: [[TMP106:%.*]] = load i32, i32* [[I150]], align 4, !llvm.access.group !19 15941 // CHECK15-NEXT: [[ARRAYIDX159:%.*]] = getelementptr inbounds double, double* [[TMP105]], i32 [[TMP106]] 15942 // CHECK15-NEXT: store double [[ADD158]], double* [[ARRAYIDX159]], align 4, !llvm.access.group !19 15943 // CHECK15-NEXT: br label [[OMP_BODY_CONTINUE160:%.*]] 15944 // CHECK15: omp.body.continue160: 15945 // CHECK15-NEXT: br label [[OMP_INNER_FOR_INC161:%.*]] 15946 // CHECK15: omp.inner.for.inc161: 15947 // CHECK15-NEXT: [[TMP107:%.*]] = load i32, i32* [[DOTOMP_IV149]], align 4, !llvm.access.group !19 15948 // CHECK15-NEXT: [[ADD162:%.*]] = add nsw i32 [[TMP107]], 1 15949 // CHECK15-NEXT: store i32 [[ADD162]], i32* [[DOTOMP_IV149]], align 4, !llvm.access.group !19 15950 // CHECK15-NEXT: br label [[OMP_INNER_FOR_COND151]], !llvm.loop [[LOOP20:![0-9]+]] 15951 // CHECK15: omp.inner.for.end163: 15952 // CHECK15-NEXT: [[TMP108:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_139]], align 4 15953 // CHECK15-NEXT: [[SUB164:%.*]] = sub nsw i32 [[TMP108]], 0 15954 // CHECK15-NEXT: [[DIV165:%.*]] = sdiv i32 [[SUB164]], 1 15955 // CHECK15-NEXT: [[MUL166:%.*]] = mul nsw i32 [[DIV165]], 1 15956 // CHECK15-NEXT: [[ADD167:%.*]] = add nsw i32 0, [[MUL166]] 15957 // CHECK15-NEXT: store i32 [[ADD167]], i32* [[I150]], align 4 15958 // CHECK15-NEXT: br label [[SIMD_IF_END168]] 15959 // CHECK15: simd.if.end168: 15960 // CHECK15-NEXT: [[TMP109:%.*]] = load i32, i32* [[CH]], align 4 15961 // CHECK15-NEXT: store i32 [[TMP109]], i32* [[DOTCAPTURE_EXPR_169]], align 4 15962 // CHECK15-NEXT: [[TMP110:%.*]] = load i32, i32* [[N]], align 4 15963 // CHECK15-NEXT: store i32 [[TMP110]], i32* [[DOTCAPTURE_EXPR_171]], align 4 15964 // CHECK15-NEXT: [[TMP111:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_171]], align 4 15965 // CHECK15-NEXT: [[SUB173:%.*]] = sub nsw i32 [[TMP111]], 0 15966 // CHECK15-NEXT: [[DIV174:%.*]] = sdiv i32 [[SUB173]], 1 15967 // CHECK15-NEXT: [[SUB175:%.*]] = sub nsw i32 [[DIV174]], 1 15968 // CHECK15-NEXT: store i32 [[SUB175]], i32* [[DOTCAPTURE_EXPR_172]], align 4 15969 // CHECK15-NEXT: store i32 0, i32* [[DOTOMP_LB176]], align 4 15970 // CHECK15-NEXT: [[TMP112:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_172]], align 4 15971 // CHECK15-NEXT: store i32 [[TMP112]], i32* [[DOTOMP_UB177]], align 4 15972 // CHECK15-NEXT: store i32 0, i32* [[I178]], align 4 15973 // CHECK15-NEXT: [[TMP113:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_171]], align 4 15974 // CHECK15-NEXT: [[CMP179:%.*]] = icmp slt i32 0, [[TMP113]] 15975 // CHECK15-NEXT: br i1 [[CMP179]], label [[SIMD_IF_THEN180:%.*]], label [[SIMD_IF_END200:%.*]] 15976 // CHECK15: simd.if.then180: 15977 // CHECK15-NEXT: [[TMP114:%.*]] = load i32, i32* [[DOTOMP_LB176]], align 4 15978 // CHECK15-NEXT: store i32 [[TMP114]], i32* [[DOTOMP_IV181]], align 4 15979 // CHECK15-NEXT: br label [[OMP_INNER_FOR_COND183:%.*]] 15980 // CHECK15: omp.inner.for.cond183: 15981 // CHECK15-NEXT: [[TMP115:%.*]] = load i32, i32* [[DOTOMP_IV181]], align 4, !llvm.access.group !22 15982 // CHECK15-NEXT: [[TMP116:%.*]] = load i32, i32* [[DOTOMP_UB177]], align 4, !llvm.access.group !22 15983 // CHECK15-NEXT: [[CMP184:%.*]] = icmp sle i32 [[TMP115]], [[TMP116]] 15984 // CHECK15-NEXT: br i1 [[CMP184]], label [[OMP_INNER_FOR_BODY185:%.*]], label [[OMP_INNER_FOR_END195:%.*]] 15985 // CHECK15: omp.inner.for.body185: 15986 // CHECK15-NEXT: [[TMP117:%.*]] = load i32, i32* [[DOTOMP_IV181]], align 4, !llvm.access.group !22 15987 // CHECK15-NEXT: [[MUL186:%.*]] = mul nsw i32 [[TMP117]], 1 15988 // CHECK15-NEXT: [[ADD187:%.*]] = add nsw i32 0, [[MUL186]] 15989 // CHECK15-NEXT: store i32 [[ADD187]], i32* [[I182]], align 4, !llvm.access.group !22 15990 // CHECK15-NEXT: [[TMP118:%.*]] = load double*, double** [[B]], align 4, !llvm.access.group !22 15991 // CHECK15-NEXT: [[TMP119:%.*]] = load i32, i32* [[I182]], align 4, !llvm.access.group !22 15992 // CHECK15-NEXT: [[ARRAYIDX188:%.*]] = getelementptr inbounds double, double* [[TMP118]], i32 [[TMP119]] 15993 // CHECK15-NEXT: [[TMP120:%.*]] = load double, double* [[ARRAYIDX188]], align 4, !llvm.access.group !22 15994 // CHECK15-NEXT: [[TMP121:%.*]] = load double*, double** [[C]], align 4, !llvm.access.group !22 15995 // CHECK15-NEXT: [[TMP122:%.*]] = load i32, i32* [[I182]], align 4, !llvm.access.group !22 15996 // CHECK15-NEXT: [[ARRAYIDX189:%.*]] = getelementptr inbounds double, double* [[TMP121]], i32 [[TMP122]] 15997 // CHECK15-NEXT: [[TMP123:%.*]] = load double, double* [[ARRAYIDX189]], align 4, !llvm.access.group !22 15998 // CHECK15-NEXT: [[ADD190:%.*]] = fadd double [[TMP120]], [[TMP123]] 15999 // CHECK15-NEXT: [[TMP124:%.*]] = load double*, double** [[A]], align 4, !llvm.access.group !22 16000 // CHECK15-NEXT: [[TMP125:%.*]] = load i32, i32* [[I182]], align 4, !llvm.access.group !22 16001 // CHECK15-NEXT: [[ARRAYIDX191:%.*]] = getelementptr inbounds double, double* [[TMP124]], i32 [[TMP125]] 16002 // CHECK15-NEXT: store double [[ADD190]], double* [[ARRAYIDX191]], align 4, !llvm.access.group !22 16003 // CHECK15-NEXT: br label [[OMP_BODY_CONTINUE192:%.*]] 16004 // CHECK15: omp.body.continue192: 16005 // CHECK15-NEXT: br label [[OMP_INNER_FOR_INC193:%.*]] 16006 // CHECK15: omp.inner.for.inc193: 16007 // CHECK15-NEXT: [[TMP126:%.*]] = load i32, i32* [[DOTOMP_IV181]], align 4, !llvm.access.group !22 16008 // CHECK15-NEXT: [[ADD194:%.*]] = add nsw i32 [[TMP126]], 1 16009 // CHECK15-NEXT: store i32 [[ADD194]], i32* [[DOTOMP_IV181]], align 4, !llvm.access.group !22 16010 // CHECK15-NEXT: br label [[OMP_INNER_FOR_COND183]], !llvm.loop [[LOOP23:![0-9]+]] 16011 // CHECK15: omp.inner.for.end195: 16012 // CHECK15-NEXT: [[TMP127:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_171]], align 4 16013 // CHECK15-NEXT: [[SUB196:%.*]] = sub nsw i32 [[TMP127]], 0 16014 // CHECK15-NEXT: [[DIV197:%.*]] = sdiv i32 [[SUB196]], 1 16015 // CHECK15-NEXT: [[MUL198:%.*]] = mul nsw i32 [[DIV197]], 1 16016 // CHECK15-NEXT: [[ADD199:%.*]] = add nsw i32 0, [[MUL198]] 16017 // CHECK15-NEXT: store i32 [[ADD199]], i32* [[I182]], align 4 16018 // CHECK15-NEXT: br label [[SIMD_IF_END200]] 16019 // CHECK15: simd.if.end200: 16020 // CHECK15-NEXT: [[CALL:%.*]] = call noundef i32 @_Z5tmainIiET_v() 16021 // CHECK15-NEXT: ret i32 [[CALL]] 16022 // 16023 // 16024 // CHECK15-LABEL: define {{[^@]+}}@_Z5tmainIiET_v 16025 // CHECK15-SAME: () #[[ATTR1:[0-9]+]] comdat { 16026 // CHECK15-NEXT: entry: 16027 // CHECK15-NEXT: [[A:%.*]] = alloca i32*, align 4 16028 // CHECK15-NEXT: [[B:%.*]] = alloca i32*, align 4 16029 // CHECK15-NEXT: [[C:%.*]] = alloca i32*, align 4 16030 // CHECK15-NEXT: [[N:%.*]] = alloca i32, align 4 16031 // CHECK15-NEXT: [[CH:%.*]] = alloca i32, align 4 16032 // CHECK15-NEXT: [[TMP:%.*]] = alloca i32, align 4 16033 // CHECK15-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 16034 // CHECK15-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 16035 // CHECK15-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 16036 // CHECK15-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4 16037 // CHECK15-NEXT: [[I:%.*]] = alloca i32, align 4 16038 // CHECK15-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 16039 // CHECK15-NEXT: [[I3:%.*]] = alloca i32, align 4 16040 // CHECK15-NEXT: [[_TMP13:%.*]] = alloca i32, align 4 16041 // CHECK15-NEXT: [[DOTCAPTURE_EXPR_14:%.*]] = alloca i32, align 4 16042 // CHECK15-NEXT: [[DOTCAPTURE_EXPR_15:%.*]] = alloca i32, align 4 16043 // CHECK15-NEXT: [[DOTOMP_LB19:%.*]] = alloca i32, align 4 16044 // CHECK15-NEXT: [[DOTOMP_UB20:%.*]] = alloca i32, align 4 16045 // CHECK15-NEXT: [[I21:%.*]] = alloca i32, align 4 16046 // CHECK15-NEXT: [[DOTOMP_IV24:%.*]] = alloca i32, align 4 16047 // CHECK15-NEXT: [[I25:%.*]] = alloca i32, align 4 16048 // CHECK15-NEXT: [[_TMP44:%.*]] = alloca i32, align 4 16049 // CHECK15-NEXT: [[DOTCAPTURE_EXPR_45:%.*]] = alloca i32, align 4 16050 // CHECK15-NEXT: [[DOTCAPTURE_EXPR_46:%.*]] = alloca i32, align 4 16051 // CHECK15-NEXT: [[DOTOMP_LB50:%.*]] = alloca i32, align 4 16052 // CHECK15-NEXT: [[DOTOMP_UB51:%.*]] = alloca i32, align 4 16053 // CHECK15-NEXT: [[I52:%.*]] = alloca i32, align 4 16054 // CHECK15-NEXT: [[DOTOMP_IV55:%.*]] = alloca i32, align 4 16055 // CHECK15-NEXT: [[I56:%.*]] = alloca i32, align 4 16056 // CHECK15-NEXT: [[_TMP75:%.*]] = alloca i32, align 4 16057 // CHECK15-NEXT: [[DOTCAPTURE_EXPR_76:%.*]] = alloca i32, align 4 16058 // CHECK15-NEXT: [[DOTCAPTURE_EXPR_77:%.*]] = alloca i32, align 4 16059 // CHECK15-NEXT: [[DOTOMP_LB81:%.*]] = alloca i32, align 4 16060 // CHECK15-NEXT: [[DOTOMP_UB82:%.*]] = alloca i32, align 4 16061 // CHECK15-NEXT: [[I83:%.*]] = alloca i32, align 4 16062 // CHECK15-NEXT: [[DOTOMP_IV86:%.*]] = alloca i32, align 4 16063 // CHECK15-NEXT: [[I87:%.*]] = alloca i32, align 4 16064 // CHECK15-NEXT: [[DOTCAPTURE_EXPR_106:%.*]] = alloca i32, align 4 16065 // CHECK15-NEXT: [[_TMP107:%.*]] = alloca i32, align 4 16066 // CHECK15-NEXT: [[DOTCAPTURE_EXPR_108:%.*]] = alloca i32, align 4 16067 // CHECK15-NEXT: [[DOTCAPTURE_EXPR_109:%.*]] = alloca i32, align 4 16068 // CHECK15-NEXT: [[DOTOMP_LB113:%.*]] = alloca i32, align 4 16069 // CHECK15-NEXT: [[DOTOMP_UB114:%.*]] = alloca i32, align 4 16070 // CHECK15-NEXT: [[I115:%.*]] = alloca i32, align 4 16071 // CHECK15-NEXT: [[DOTOMP_IV118:%.*]] = alloca i32, align 4 16072 // CHECK15-NEXT: [[I119:%.*]] = alloca i32, align 4 16073 // CHECK15-NEXT: [[_TMP138:%.*]] = alloca i32, align 4 16074 // CHECK15-NEXT: [[DOTCAPTURE_EXPR_139:%.*]] = alloca i32, align 4 16075 // CHECK15-NEXT: [[DOTCAPTURE_EXPR_140:%.*]] = alloca i32, align 4 16076 // CHECK15-NEXT: [[DOTOMP_LB144:%.*]] = alloca i32, align 4 16077 // CHECK15-NEXT: [[DOTOMP_UB145:%.*]] = alloca i32, align 4 16078 // CHECK15-NEXT: [[I146:%.*]] = alloca i32, align 4 16079 // CHECK15-NEXT: [[DOTOMP_IV149:%.*]] = alloca i32, align 4 16080 // CHECK15-NEXT: [[I150:%.*]] = alloca i32, align 4 16081 // CHECK15-NEXT: [[DOTCAPTURE_EXPR_169:%.*]] = alloca i32, align 4 16082 // CHECK15-NEXT: [[_TMP170:%.*]] = alloca i32, align 4 16083 // CHECK15-NEXT: [[DOTCAPTURE_EXPR_171:%.*]] = alloca i32, align 4 16084 // CHECK15-NEXT: [[DOTCAPTURE_EXPR_172:%.*]] = alloca i32, align 4 16085 // CHECK15-NEXT: [[DOTOMP_LB176:%.*]] = alloca i32, align 4 16086 // CHECK15-NEXT: [[DOTOMP_UB177:%.*]] = alloca i32, align 4 16087 // CHECK15-NEXT: [[I178:%.*]] = alloca i32, align 4 16088 // CHECK15-NEXT: [[DOTOMP_IV181:%.*]] = alloca i32, align 4 16089 // CHECK15-NEXT: [[I182:%.*]] = alloca i32, align 4 16090 // CHECK15-NEXT: store i32 10000, i32* [[N]], align 4 16091 // CHECK15-NEXT: store i32 100, i32* [[CH]], align 4 16092 // CHECK15-NEXT: [[TMP0:%.*]] = load i32, i32* [[N]], align 4 16093 // CHECK15-NEXT: store i32 [[TMP0]], i32* [[DOTCAPTURE_EXPR_]], align 4 16094 // CHECK15-NEXT: [[TMP1:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 16095 // CHECK15-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP1]], 0 16096 // CHECK15-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 16097 // CHECK15-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1 16098 // CHECK15-NEXT: store i32 [[SUB2]], i32* [[DOTCAPTURE_EXPR_1]], align 4 16099 // CHECK15-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4 16100 // CHECK15-NEXT: [[TMP2:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 16101 // CHECK15-NEXT: store i32 [[TMP2]], i32* [[DOTOMP_UB]], align 4 16102 // CHECK15-NEXT: store i32 0, i32* [[I]], align 4 16103 // CHECK15-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 16104 // CHECK15-NEXT: [[CMP:%.*]] = icmp slt i32 0, [[TMP3]] 16105 // CHECK15-NEXT: br i1 [[CMP]], label [[SIMD_IF_THEN:%.*]], label [[SIMD_IF_END:%.*]] 16106 // CHECK15: simd.if.then: 16107 // CHECK15-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4 16108 // CHECK15-NEXT: store i32 [[TMP4]], i32* [[DOTOMP_IV]], align 4 16109 // CHECK15-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] 16110 // CHECK15: omp.inner.for.cond: 16111 // CHECK15-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !25 16112 // CHECK15-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4, !llvm.access.group !25 16113 // CHECK15-NEXT: [[CMP4:%.*]] = icmp sle i32 [[TMP5]], [[TMP6]] 16114 // CHECK15-NEXT: br i1 [[CMP4]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] 16115 // CHECK15: omp.inner.for.body: 16116 // CHECK15-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !25 16117 // CHECK15-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP7]], 1 16118 // CHECK15-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] 16119 // CHECK15-NEXT: store i32 [[ADD]], i32* [[I3]], align 4, !llvm.access.group !25 16120 // CHECK15-NEXT: [[TMP8:%.*]] = load i32*, i32** [[B]], align 4, !llvm.access.group !25 16121 // CHECK15-NEXT: [[TMP9:%.*]] = load i32, i32* [[I3]], align 4, !llvm.access.group !25 16122 // CHECK15-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i32, i32* [[TMP8]], i32 [[TMP9]] 16123 // CHECK15-NEXT: [[TMP10:%.*]] = load i32, i32* [[ARRAYIDX]], align 4, !llvm.access.group !25 16124 // CHECK15-NEXT: [[TMP11:%.*]] = load i32*, i32** [[C]], align 4, !llvm.access.group !25 16125 // CHECK15-NEXT: [[TMP12:%.*]] = load i32, i32* [[I3]], align 4, !llvm.access.group !25 16126 // CHECK15-NEXT: [[ARRAYIDX5:%.*]] = getelementptr inbounds i32, i32* [[TMP11]], i32 [[TMP12]] 16127 // CHECK15-NEXT: [[TMP13:%.*]] = load i32, i32* [[ARRAYIDX5]], align 4, !llvm.access.group !25 16128 // CHECK15-NEXT: [[ADD6:%.*]] = add nsw i32 [[TMP10]], [[TMP13]] 16129 // CHECK15-NEXT: [[TMP14:%.*]] = load i32*, i32** [[A]], align 4, !llvm.access.group !25 16130 // CHECK15-NEXT: [[TMP15:%.*]] = load i32, i32* [[I3]], align 4, !llvm.access.group !25 16131 // CHECK15-NEXT: [[ARRAYIDX7:%.*]] = getelementptr inbounds i32, i32* [[TMP14]], i32 [[TMP15]] 16132 // CHECK15-NEXT: store i32 [[ADD6]], i32* [[ARRAYIDX7]], align 4, !llvm.access.group !25 16133 // CHECK15-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] 16134 // CHECK15: omp.body.continue: 16135 // CHECK15-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] 16136 // CHECK15: omp.inner.for.inc: 16137 // CHECK15-NEXT: [[TMP16:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !25 16138 // CHECK15-NEXT: [[ADD8:%.*]] = add nsw i32 [[TMP16]], 1 16139 // CHECK15-NEXT: store i32 [[ADD8]], i32* [[DOTOMP_IV]], align 4, !llvm.access.group !25 16140 // CHECK15-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP26:![0-9]+]] 16141 // CHECK15: omp.inner.for.end: 16142 // CHECK15-NEXT: [[TMP17:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 16143 // CHECK15-NEXT: [[SUB9:%.*]] = sub nsw i32 [[TMP17]], 0 16144 // CHECK15-NEXT: [[DIV10:%.*]] = sdiv i32 [[SUB9]], 1 16145 // CHECK15-NEXT: [[MUL11:%.*]] = mul nsw i32 [[DIV10]], 1 16146 // CHECK15-NEXT: [[ADD12:%.*]] = add nsw i32 0, [[MUL11]] 16147 // CHECK15-NEXT: store i32 [[ADD12]], i32* [[I3]], align 4 16148 // CHECK15-NEXT: br label [[SIMD_IF_END]] 16149 // CHECK15: simd.if.end: 16150 // CHECK15-NEXT: [[TMP18:%.*]] = load i32, i32* [[N]], align 4 16151 // CHECK15-NEXT: store i32 [[TMP18]], i32* [[DOTCAPTURE_EXPR_14]], align 4 16152 // CHECK15-NEXT: [[TMP19:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_14]], align 4 16153 // CHECK15-NEXT: [[SUB16:%.*]] = sub nsw i32 [[TMP19]], 0 16154 // CHECK15-NEXT: [[DIV17:%.*]] = sdiv i32 [[SUB16]], 1 16155 // CHECK15-NEXT: [[SUB18:%.*]] = sub nsw i32 [[DIV17]], 1 16156 // CHECK15-NEXT: store i32 [[SUB18]], i32* [[DOTCAPTURE_EXPR_15]], align 4 16157 // CHECK15-NEXT: store i32 0, i32* [[DOTOMP_LB19]], align 4 16158 // CHECK15-NEXT: [[TMP20:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_15]], align 4 16159 // CHECK15-NEXT: store i32 [[TMP20]], i32* [[DOTOMP_UB20]], align 4 16160 // CHECK15-NEXT: store i32 0, i32* [[I21]], align 4 16161 // CHECK15-NEXT: [[TMP21:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_14]], align 4 16162 // CHECK15-NEXT: [[CMP22:%.*]] = icmp slt i32 0, [[TMP21]] 16163 // CHECK15-NEXT: br i1 [[CMP22]], label [[SIMD_IF_THEN23:%.*]], label [[SIMD_IF_END43:%.*]] 16164 // CHECK15: simd.if.then23: 16165 // CHECK15-NEXT: [[TMP22:%.*]] = load i32, i32* [[DOTOMP_LB19]], align 4 16166 // CHECK15-NEXT: store i32 [[TMP22]], i32* [[DOTOMP_IV24]], align 4 16167 // CHECK15-NEXT: br label [[OMP_INNER_FOR_COND26:%.*]] 16168 // CHECK15: omp.inner.for.cond26: 16169 // CHECK15-NEXT: [[TMP23:%.*]] = load i32, i32* [[DOTOMP_IV24]], align 4, !llvm.access.group !28 16170 // CHECK15-NEXT: [[TMP24:%.*]] = load i32, i32* [[DOTOMP_UB20]], align 4, !llvm.access.group !28 16171 // CHECK15-NEXT: [[CMP27:%.*]] = icmp sle i32 [[TMP23]], [[TMP24]] 16172 // CHECK15-NEXT: br i1 [[CMP27]], label [[OMP_INNER_FOR_BODY28:%.*]], label [[OMP_INNER_FOR_END38:%.*]] 16173 // CHECK15: omp.inner.for.body28: 16174 // CHECK15-NEXT: [[TMP25:%.*]] = load i32, i32* [[DOTOMP_IV24]], align 4, !llvm.access.group !28 16175 // CHECK15-NEXT: [[MUL29:%.*]] = mul nsw i32 [[TMP25]], 1 16176 // CHECK15-NEXT: [[ADD30:%.*]] = add nsw i32 0, [[MUL29]] 16177 // CHECK15-NEXT: store i32 [[ADD30]], i32* [[I25]], align 4, !llvm.access.group !28 16178 // CHECK15-NEXT: [[TMP26:%.*]] = load i32*, i32** [[B]], align 4, !llvm.access.group !28 16179 // CHECK15-NEXT: [[TMP27:%.*]] = load i32, i32* [[I25]], align 4, !llvm.access.group !28 16180 // CHECK15-NEXT: [[ARRAYIDX31:%.*]] = getelementptr inbounds i32, i32* [[TMP26]], i32 [[TMP27]] 16181 // CHECK15-NEXT: [[TMP28:%.*]] = load i32, i32* [[ARRAYIDX31]], align 4, !llvm.access.group !28 16182 // CHECK15-NEXT: [[TMP29:%.*]] = load i32*, i32** [[C]], align 4, !llvm.access.group !28 16183 // CHECK15-NEXT: [[TMP30:%.*]] = load i32, i32* [[I25]], align 4, !llvm.access.group !28 16184 // CHECK15-NEXT: [[ARRAYIDX32:%.*]] = getelementptr inbounds i32, i32* [[TMP29]], i32 [[TMP30]] 16185 // CHECK15-NEXT: [[TMP31:%.*]] = load i32, i32* [[ARRAYIDX32]], align 4, !llvm.access.group !28 16186 // CHECK15-NEXT: [[ADD33:%.*]] = add nsw i32 [[TMP28]], [[TMP31]] 16187 // CHECK15-NEXT: [[TMP32:%.*]] = load i32*, i32** [[A]], align 4, !llvm.access.group !28 16188 // CHECK15-NEXT: [[TMP33:%.*]] = load i32, i32* [[I25]], align 4, !llvm.access.group !28 16189 // CHECK15-NEXT: [[ARRAYIDX34:%.*]] = getelementptr inbounds i32, i32* [[TMP32]], i32 [[TMP33]] 16190 // CHECK15-NEXT: store i32 [[ADD33]], i32* [[ARRAYIDX34]], align 4, !llvm.access.group !28 16191 // CHECK15-NEXT: br label [[OMP_BODY_CONTINUE35:%.*]] 16192 // CHECK15: omp.body.continue35: 16193 // CHECK15-NEXT: br label [[OMP_INNER_FOR_INC36:%.*]] 16194 // CHECK15: omp.inner.for.inc36: 16195 // CHECK15-NEXT: [[TMP34:%.*]] = load i32, i32* [[DOTOMP_IV24]], align 4, !llvm.access.group !28 16196 // CHECK15-NEXT: [[ADD37:%.*]] = add nsw i32 [[TMP34]], 1 16197 // CHECK15-NEXT: store i32 [[ADD37]], i32* [[DOTOMP_IV24]], align 4, !llvm.access.group !28 16198 // CHECK15-NEXT: br label [[OMP_INNER_FOR_COND26]], !llvm.loop [[LOOP29:![0-9]+]] 16199 // CHECK15: omp.inner.for.end38: 16200 // CHECK15-NEXT: [[TMP35:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_14]], align 4 16201 // CHECK15-NEXT: [[SUB39:%.*]] = sub nsw i32 [[TMP35]], 0 16202 // CHECK15-NEXT: [[DIV40:%.*]] = sdiv i32 [[SUB39]], 1 16203 // CHECK15-NEXT: [[MUL41:%.*]] = mul nsw i32 [[DIV40]], 1 16204 // CHECK15-NEXT: [[ADD42:%.*]] = add nsw i32 0, [[MUL41]] 16205 // CHECK15-NEXT: store i32 [[ADD42]], i32* [[I25]], align 4 16206 // CHECK15-NEXT: br label [[SIMD_IF_END43]] 16207 // CHECK15: simd.if.end43: 16208 // CHECK15-NEXT: [[TMP36:%.*]] = load i32, i32* [[N]], align 4 16209 // CHECK15-NEXT: store i32 [[TMP36]], i32* [[DOTCAPTURE_EXPR_45]], align 4 16210 // CHECK15-NEXT: [[TMP37:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_45]], align 4 16211 // CHECK15-NEXT: [[SUB47:%.*]] = sub nsw i32 [[TMP37]], 0 16212 // CHECK15-NEXT: [[DIV48:%.*]] = sdiv i32 [[SUB47]], 1 16213 // CHECK15-NEXT: [[SUB49:%.*]] = sub nsw i32 [[DIV48]], 1 16214 // CHECK15-NEXT: store i32 [[SUB49]], i32* [[DOTCAPTURE_EXPR_46]], align 4 16215 // CHECK15-NEXT: store i32 0, i32* [[DOTOMP_LB50]], align 4 16216 // CHECK15-NEXT: [[TMP38:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_46]], align 4 16217 // CHECK15-NEXT: store i32 [[TMP38]], i32* [[DOTOMP_UB51]], align 4 16218 // CHECK15-NEXT: store i32 0, i32* [[I52]], align 4 16219 // CHECK15-NEXT: [[TMP39:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_45]], align 4 16220 // CHECK15-NEXT: [[CMP53:%.*]] = icmp slt i32 0, [[TMP39]] 16221 // CHECK15-NEXT: br i1 [[CMP53]], label [[SIMD_IF_THEN54:%.*]], label [[SIMD_IF_END74:%.*]] 16222 // CHECK15: simd.if.then54: 16223 // CHECK15-NEXT: [[TMP40:%.*]] = load i32, i32* [[DOTOMP_LB50]], align 4 16224 // CHECK15-NEXT: store i32 [[TMP40]], i32* [[DOTOMP_IV55]], align 4 16225 // CHECK15-NEXT: br label [[OMP_INNER_FOR_COND57:%.*]] 16226 // CHECK15: omp.inner.for.cond57: 16227 // CHECK15-NEXT: [[TMP41:%.*]] = load i32, i32* [[DOTOMP_IV55]], align 4, !llvm.access.group !31 16228 // CHECK15-NEXT: [[TMP42:%.*]] = load i32, i32* [[DOTOMP_UB51]], align 4, !llvm.access.group !31 16229 // CHECK15-NEXT: [[CMP58:%.*]] = icmp sle i32 [[TMP41]], [[TMP42]] 16230 // CHECK15-NEXT: br i1 [[CMP58]], label [[OMP_INNER_FOR_BODY59:%.*]], label [[OMP_INNER_FOR_END69:%.*]] 16231 // CHECK15: omp.inner.for.body59: 16232 // CHECK15-NEXT: [[TMP43:%.*]] = load i32, i32* [[DOTOMP_IV55]], align 4, !llvm.access.group !31 16233 // CHECK15-NEXT: [[MUL60:%.*]] = mul nsw i32 [[TMP43]], 1 16234 // CHECK15-NEXT: [[ADD61:%.*]] = add nsw i32 0, [[MUL60]] 16235 // CHECK15-NEXT: store i32 [[ADD61]], i32* [[I56]], align 4, !llvm.access.group !31 16236 // CHECK15-NEXT: [[TMP44:%.*]] = load i32*, i32** [[B]], align 4, !llvm.access.group !31 16237 // CHECK15-NEXT: [[TMP45:%.*]] = load i32, i32* [[I56]], align 4, !llvm.access.group !31 16238 // CHECK15-NEXT: [[ARRAYIDX62:%.*]] = getelementptr inbounds i32, i32* [[TMP44]], i32 [[TMP45]] 16239 // CHECK15-NEXT: [[TMP46:%.*]] = load i32, i32* [[ARRAYIDX62]], align 4, !llvm.access.group !31 16240 // CHECK15-NEXT: [[TMP47:%.*]] = load i32*, i32** [[C]], align 4, !llvm.access.group !31 16241 // CHECK15-NEXT: [[TMP48:%.*]] = load i32, i32* [[I56]], align 4, !llvm.access.group !31 16242 // CHECK15-NEXT: [[ARRAYIDX63:%.*]] = getelementptr inbounds i32, i32* [[TMP47]], i32 [[TMP48]] 16243 // CHECK15-NEXT: [[TMP49:%.*]] = load i32, i32* [[ARRAYIDX63]], align 4, !llvm.access.group !31 16244 // CHECK15-NEXT: [[ADD64:%.*]] = add nsw i32 [[TMP46]], [[TMP49]] 16245 // CHECK15-NEXT: [[TMP50:%.*]] = load i32*, i32** [[A]], align 4, !llvm.access.group !31 16246 // CHECK15-NEXT: [[TMP51:%.*]] = load i32, i32* [[I56]], align 4, !llvm.access.group !31 16247 // CHECK15-NEXT: [[ARRAYIDX65:%.*]] = getelementptr inbounds i32, i32* [[TMP50]], i32 [[TMP51]] 16248 // CHECK15-NEXT: store i32 [[ADD64]], i32* [[ARRAYIDX65]], align 4, !llvm.access.group !31 16249 // CHECK15-NEXT: br label [[OMP_BODY_CONTINUE66:%.*]] 16250 // CHECK15: omp.body.continue66: 16251 // CHECK15-NEXT: br label [[OMP_INNER_FOR_INC67:%.*]] 16252 // CHECK15: omp.inner.for.inc67: 16253 // CHECK15-NEXT: [[TMP52:%.*]] = load i32, i32* [[DOTOMP_IV55]], align 4, !llvm.access.group !31 16254 // CHECK15-NEXT: [[ADD68:%.*]] = add nsw i32 [[TMP52]], 1 16255 // CHECK15-NEXT: store i32 [[ADD68]], i32* [[DOTOMP_IV55]], align 4, !llvm.access.group !31 16256 // CHECK15-NEXT: br label [[OMP_INNER_FOR_COND57]], !llvm.loop [[LOOP32:![0-9]+]] 16257 // CHECK15: omp.inner.for.end69: 16258 // CHECK15-NEXT: [[TMP53:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_45]], align 4 16259 // CHECK15-NEXT: [[SUB70:%.*]] = sub nsw i32 [[TMP53]], 0 16260 // CHECK15-NEXT: [[DIV71:%.*]] = sdiv i32 [[SUB70]], 1 16261 // CHECK15-NEXT: [[MUL72:%.*]] = mul nsw i32 [[DIV71]], 1 16262 // CHECK15-NEXT: [[ADD73:%.*]] = add nsw i32 0, [[MUL72]] 16263 // CHECK15-NEXT: store i32 [[ADD73]], i32* [[I56]], align 4 16264 // CHECK15-NEXT: br label [[SIMD_IF_END74]] 16265 // CHECK15: simd.if.end74: 16266 // CHECK15-NEXT: [[TMP54:%.*]] = load i32, i32* [[N]], align 4 16267 // CHECK15-NEXT: store i32 [[TMP54]], i32* [[DOTCAPTURE_EXPR_76]], align 4 16268 // CHECK15-NEXT: [[TMP55:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_76]], align 4 16269 // CHECK15-NEXT: [[SUB78:%.*]] = sub nsw i32 [[TMP55]], 0 16270 // CHECK15-NEXT: [[DIV79:%.*]] = sdiv i32 [[SUB78]], 1 16271 // CHECK15-NEXT: [[SUB80:%.*]] = sub nsw i32 [[DIV79]], 1 16272 // CHECK15-NEXT: store i32 [[SUB80]], i32* [[DOTCAPTURE_EXPR_77]], align 4 16273 // CHECK15-NEXT: store i32 0, i32* [[DOTOMP_LB81]], align 4 16274 // CHECK15-NEXT: [[TMP56:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_77]], align 4 16275 // CHECK15-NEXT: store i32 [[TMP56]], i32* [[DOTOMP_UB82]], align 4 16276 // CHECK15-NEXT: store i32 0, i32* [[I83]], align 4 16277 // CHECK15-NEXT: [[TMP57:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_76]], align 4 16278 // CHECK15-NEXT: [[CMP84:%.*]] = icmp slt i32 0, [[TMP57]] 16279 // CHECK15-NEXT: br i1 [[CMP84]], label [[SIMD_IF_THEN85:%.*]], label [[SIMD_IF_END105:%.*]] 16280 // CHECK15: simd.if.then85: 16281 // CHECK15-NEXT: [[TMP58:%.*]] = load i32, i32* [[DOTOMP_LB81]], align 4 16282 // CHECK15-NEXT: store i32 [[TMP58]], i32* [[DOTOMP_IV86]], align 4 16283 // CHECK15-NEXT: br label [[OMP_INNER_FOR_COND88:%.*]] 16284 // CHECK15: omp.inner.for.cond88: 16285 // CHECK15-NEXT: [[TMP59:%.*]] = load i32, i32* [[DOTOMP_IV86]], align 4, !llvm.access.group !34 16286 // CHECK15-NEXT: [[TMP60:%.*]] = load i32, i32* [[DOTOMP_UB82]], align 4, !llvm.access.group !34 16287 // CHECK15-NEXT: [[CMP89:%.*]] = icmp sle i32 [[TMP59]], [[TMP60]] 16288 // CHECK15-NEXT: br i1 [[CMP89]], label [[OMP_INNER_FOR_BODY90:%.*]], label [[OMP_INNER_FOR_END100:%.*]] 16289 // CHECK15: omp.inner.for.body90: 16290 // CHECK15-NEXT: [[TMP61:%.*]] = load i32, i32* [[DOTOMP_IV86]], align 4, !llvm.access.group !34 16291 // CHECK15-NEXT: [[MUL91:%.*]] = mul nsw i32 [[TMP61]], 1 16292 // CHECK15-NEXT: [[ADD92:%.*]] = add nsw i32 0, [[MUL91]] 16293 // CHECK15-NEXT: store i32 [[ADD92]], i32* [[I87]], align 4, !llvm.access.group !34 16294 // CHECK15-NEXT: [[TMP62:%.*]] = load i32*, i32** [[B]], align 4, !llvm.access.group !34 16295 // CHECK15-NEXT: [[TMP63:%.*]] = load i32, i32* [[I87]], align 4, !llvm.access.group !34 16296 // CHECK15-NEXT: [[ARRAYIDX93:%.*]] = getelementptr inbounds i32, i32* [[TMP62]], i32 [[TMP63]] 16297 // CHECK15-NEXT: [[TMP64:%.*]] = load i32, i32* [[ARRAYIDX93]], align 4, !llvm.access.group !34 16298 // CHECK15-NEXT: [[TMP65:%.*]] = load i32*, i32** [[C]], align 4, !llvm.access.group !34 16299 // CHECK15-NEXT: [[TMP66:%.*]] = load i32, i32* [[I87]], align 4, !llvm.access.group !34 16300 // CHECK15-NEXT: [[ARRAYIDX94:%.*]] = getelementptr inbounds i32, i32* [[TMP65]], i32 [[TMP66]] 16301 // CHECK15-NEXT: [[TMP67:%.*]] = load i32, i32* [[ARRAYIDX94]], align 4, !llvm.access.group !34 16302 // CHECK15-NEXT: [[ADD95:%.*]] = add nsw i32 [[TMP64]], [[TMP67]] 16303 // CHECK15-NEXT: [[TMP68:%.*]] = load i32*, i32** [[A]], align 4, !llvm.access.group !34 16304 // CHECK15-NEXT: [[TMP69:%.*]] = load i32, i32* [[I87]], align 4, !llvm.access.group !34 16305 // CHECK15-NEXT: [[ARRAYIDX96:%.*]] = getelementptr inbounds i32, i32* [[TMP68]], i32 [[TMP69]] 16306 // CHECK15-NEXT: store i32 [[ADD95]], i32* [[ARRAYIDX96]], align 4, !llvm.access.group !34 16307 // CHECK15-NEXT: br label [[OMP_BODY_CONTINUE97:%.*]] 16308 // CHECK15: omp.body.continue97: 16309 // CHECK15-NEXT: br label [[OMP_INNER_FOR_INC98:%.*]] 16310 // CHECK15: omp.inner.for.inc98: 16311 // CHECK15-NEXT: [[TMP70:%.*]] = load i32, i32* [[DOTOMP_IV86]], align 4, !llvm.access.group !34 16312 // CHECK15-NEXT: [[ADD99:%.*]] = add nsw i32 [[TMP70]], 1 16313 // CHECK15-NEXT: store i32 [[ADD99]], i32* [[DOTOMP_IV86]], align 4, !llvm.access.group !34 16314 // CHECK15-NEXT: br label [[OMP_INNER_FOR_COND88]], !llvm.loop [[LOOP35:![0-9]+]] 16315 // CHECK15: omp.inner.for.end100: 16316 // CHECK15-NEXT: [[TMP71:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_76]], align 4 16317 // CHECK15-NEXT: [[SUB101:%.*]] = sub nsw i32 [[TMP71]], 0 16318 // CHECK15-NEXT: [[DIV102:%.*]] = sdiv i32 [[SUB101]], 1 16319 // CHECK15-NEXT: [[MUL103:%.*]] = mul nsw i32 [[DIV102]], 1 16320 // CHECK15-NEXT: [[ADD104:%.*]] = add nsw i32 0, [[MUL103]] 16321 // CHECK15-NEXT: store i32 [[ADD104]], i32* [[I87]], align 4 16322 // CHECK15-NEXT: br label [[SIMD_IF_END105]] 16323 // CHECK15: simd.if.end105: 16324 // CHECK15-NEXT: [[TMP72:%.*]] = load i32, i32* [[CH]], align 4 16325 // CHECK15-NEXT: store i32 [[TMP72]], i32* [[DOTCAPTURE_EXPR_106]], align 4 16326 // CHECK15-NEXT: [[TMP73:%.*]] = load i32, i32* [[N]], align 4 16327 // CHECK15-NEXT: store i32 [[TMP73]], i32* [[DOTCAPTURE_EXPR_108]], align 4 16328 // CHECK15-NEXT: [[TMP74:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_108]], align 4 16329 // CHECK15-NEXT: [[SUB110:%.*]] = sub nsw i32 [[TMP74]], 0 16330 // CHECK15-NEXT: [[DIV111:%.*]] = sdiv i32 [[SUB110]], 1 16331 // CHECK15-NEXT: [[SUB112:%.*]] = sub nsw i32 [[DIV111]], 1 16332 // CHECK15-NEXT: store i32 [[SUB112]], i32* [[DOTCAPTURE_EXPR_109]], align 4 16333 // CHECK15-NEXT: store i32 0, i32* [[DOTOMP_LB113]], align 4 16334 // CHECK15-NEXT: [[TMP75:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_109]], align 4 16335 // CHECK15-NEXT: store i32 [[TMP75]], i32* [[DOTOMP_UB114]], align 4 16336 // CHECK15-NEXT: store i32 0, i32* [[I115]], align 4 16337 // CHECK15-NEXT: [[TMP76:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_108]], align 4 16338 // CHECK15-NEXT: [[CMP116:%.*]] = icmp slt i32 0, [[TMP76]] 16339 // CHECK15-NEXT: br i1 [[CMP116]], label [[SIMD_IF_THEN117:%.*]], label [[SIMD_IF_END137:%.*]] 16340 // CHECK15: simd.if.then117: 16341 // CHECK15-NEXT: [[TMP77:%.*]] = load i32, i32* [[DOTOMP_LB113]], align 4 16342 // CHECK15-NEXT: store i32 [[TMP77]], i32* [[DOTOMP_IV118]], align 4 16343 // CHECK15-NEXT: br label [[OMP_INNER_FOR_COND120:%.*]] 16344 // CHECK15: omp.inner.for.cond120: 16345 // CHECK15-NEXT: [[TMP78:%.*]] = load i32, i32* [[DOTOMP_IV118]], align 4, !llvm.access.group !37 16346 // CHECK15-NEXT: [[TMP79:%.*]] = load i32, i32* [[DOTOMP_UB114]], align 4, !llvm.access.group !37 16347 // CHECK15-NEXT: [[CMP121:%.*]] = icmp sle i32 [[TMP78]], [[TMP79]] 16348 // CHECK15-NEXT: br i1 [[CMP121]], label [[OMP_INNER_FOR_BODY122:%.*]], label [[OMP_INNER_FOR_END132:%.*]] 16349 // CHECK15: omp.inner.for.body122: 16350 // CHECK15-NEXT: [[TMP80:%.*]] = load i32, i32* [[DOTOMP_IV118]], align 4, !llvm.access.group !37 16351 // CHECK15-NEXT: [[MUL123:%.*]] = mul nsw i32 [[TMP80]], 1 16352 // CHECK15-NEXT: [[ADD124:%.*]] = add nsw i32 0, [[MUL123]] 16353 // CHECK15-NEXT: store i32 [[ADD124]], i32* [[I119]], align 4, !llvm.access.group !37 16354 // CHECK15-NEXT: [[TMP81:%.*]] = load i32*, i32** [[B]], align 4, !llvm.access.group !37 16355 // CHECK15-NEXT: [[TMP82:%.*]] = load i32, i32* [[I119]], align 4, !llvm.access.group !37 16356 // CHECK15-NEXT: [[ARRAYIDX125:%.*]] = getelementptr inbounds i32, i32* [[TMP81]], i32 [[TMP82]] 16357 // CHECK15-NEXT: [[TMP83:%.*]] = load i32, i32* [[ARRAYIDX125]], align 4, !llvm.access.group !37 16358 // CHECK15-NEXT: [[TMP84:%.*]] = load i32*, i32** [[C]], align 4, !llvm.access.group !37 16359 // CHECK15-NEXT: [[TMP85:%.*]] = load i32, i32* [[I119]], align 4, !llvm.access.group !37 16360 // CHECK15-NEXT: [[ARRAYIDX126:%.*]] = getelementptr inbounds i32, i32* [[TMP84]], i32 [[TMP85]] 16361 // CHECK15-NEXT: [[TMP86:%.*]] = load i32, i32* [[ARRAYIDX126]], align 4, !llvm.access.group !37 16362 // CHECK15-NEXT: [[ADD127:%.*]] = add nsw i32 [[TMP83]], [[TMP86]] 16363 // CHECK15-NEXT: [[TMP87:%.*]] = load i32*, i32** [[A]], align 4, !llvm.access.group !37 16364 // CHECK15-NEXT: [[TMP88:%.*]] = load i32, i32* [[I119]], align 4, !llvm.access.group !37 16365 // CHECK15-NEXT: [[ARRAYIDX128:%.*]] = getelementptr inbounds i32, i32* [[TMP87]], i32 [[TMP88]] 16366 // CHECK15-NEXT: store i32 [[ADD127]], i32* [[ARRAYIDX128]], align 4, !llvm.access.group !37 16367 // CHECK15-NEXT: br label [[OMP_BODY_CONTINUE129:%.*]] 16368 // CHECK15: omp.body.continue129: 16369 // CHECK15-NEXT: br label [[OMP_INNER_FOR_INC130:%.*]] 16370 // CHECK15: omp.inner.for.inc130: 16371 // CHECK15-NEXT: [[TMP89:%.*]] = load i32, i32* [[DOTOMP_IV118]], align 4, !llvm.access.group !37 16372 // CHECK15-NEXT: [[ADD131:%.*]] = add nsw i32 [[TMP89]], 1 16373 // CHECK15-NEXT: store i32 [[ADD131]], i32* [[DOTOMP_IV118]], align 4, !llvm.access.group !37 16374 // CHECK15-NEXT: br label [[OMP_INNER_FOR_COND120]], !llvm.loop [[LOOP38:![0-9]+]] 16375 // CHECK15: omp.inner.for.end132: 16376 // CHECK15-NEXT: [[TMP90:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_108]], align 4 16377 // CHECK15-NEXT: [[SUB133:%.*]] = sub nsw i32 [[TMP90]], 0 16378 // CHECK15-NEXT: [[DIV134:%.*]] = sdiv i32 [[SUB133]], 1 16379 // CHECK15-NEXT: [[MUL135:%.*]] = mul nsw i32 [[DIV134]], 1 16380 // CHECK15-NEXT: [[ADD136:%.*]] = add nsw i32 0, [[MUL135]] 16381 // CHECK15-NEXT: store i32 [[ADD136]], i32* [[I119]], align 4 16382 // CHECK15-NEXT: br label [[SIMD_IF_END137]] 16383 // CHECK15: simd.if.end137: 16384 // CHECK15-NEXT: [[TMP91:%.*]] = load i32, i32* [[N]], align 4 16385 // CHECK15-NEXT: store i32 [[TMP91]], i32* [[DOTCAPTURE_EXPR_139]], align 4 16386 // CHECK15-NEXT: [[TMP92:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_139]], align 4 16387 // CHECK15-NEXT: [[SUB141:%.*]] = sub nsw i32 [[TMP92]], 0 16388 // CHECK15-NEXT: [[DIV142:%.*]] = sdiv i32 [[SUB141]], 1 16389 // CHECK15-NEXT: [[SUB143:%.*]] = sub nsw i32 [[DIV142]], 1 16390 // CHECK15-NEXT: store i32 [[SUB143]], i32* [[DOTCAPTURE_EXPR_140]], align 4 16391 // CHECK15-NEXT: store i32 0, i32* [[DOTOMP_LB144]], align 4 16392 // CHECK15-NEXT: [[TMP93:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_140]], align 4 16393 // CHECK15-NEXT: store i32 [[TMP93]], i32* [[DOTOMP_UB145]], align 4 16394 // CHECK15-NEXT: store i32 0, i32* [[I146]], align 4 16395 // CHECK15-NEXT: [[TMP94:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_139]], align 4 16396 // CHECK15-NEXT: [[CMP147:%.*]] = icmp slt i32 0, [[TMP94]] 16397 // CHECK15-NEXT: br i1 [[CMP147]], label [[SIMD_IF_THEN148:%.*]], label [[SIMD_IF_END168:%.*]] 16398 // CHECK15: simd.if.then148: 16399 // CHECK15-NEXT: [[TMP95:%.*]] = load i32, i32* [[DOTOMP_LB144]], align 4 16400 // CHECK15-NEXT: store i32 [[TMP95]], i32* [[DOTOMP_IV149]], align 4 16401 // CHECK15-NEXT: br label [[OMP_INNER_FOR_COND151:%.*]] 16402 // CHECK15: omp.inner.for.cond151: 16403 // CHECK15-NEXT: [[TMP96:%.*]] = load i32, i32* [[DOTOMP_IV149]], align 4, !llvm.access.group !40 16404 // CHECK15-NEXT: [[TMP97:%.*]] = load i32, i32* [[DOTOMP_UB145]], align 4, !llvm.access.group !40 16405 // CHECK15-NEXT: [[CMP152:%.*]] = icmp sle i32 [[TMP96]], [[TMP97]] 16406 // CHECK15-NEXT: br i1 [[CMP152]], label [[OMP_INNER_FOR_BODY153:%.*]], label [[OMP_INNER_FOR_END163:%.*]] 16407 // CHECK15: omp.inner.for.body153: 16408 // CHECK15-NEXT: [[TMP98:%.*]] = load i32, i32* [[DOTOMP_IV149]], align 4, !llvm.access.group !40 16409 // CHECK15-NEXT: [[MUL154:%.*]] = mul nsw i32 [[TMP98]], 1 16410 // CHECK15-NEXT: [[ADD155:%.*]] = add nsw i32 0, [[MUL154]] 16411 // CHECK15-NEXT: store i32 [[ADD155]], i32* [[I150]], align 4, !llvm.access.group !40 16412 // CHECK15-NEXT: [[TMP99:%.*]] = load i32*, i32** [[B]], align 4, !llvm.access.group !40 16413 // CHECK15-NEXT: [[TMP100:%.*]] = load i32, i32* [[I150]], align 4, !llvm.access.group !40 16414 // CHECK15-NEXT: [[ARRAYIDX156:%.*]] = getelementptr inbounds i32, i32* [[TMP99]], i32 [[TMP100]] 16415 // CHECK15-NEXT: [[TMP101:%.*]] = load i32, i32* [[ARRAYIDX156]], align 4, !llvm.access.group !40 16416 // CHECK15-NEXT: [[TMP102:%.*]] = load i32*, i32** [[C]], align 4, !llvm.access.group !40 16417 // CHECK15-NEXT: [[TMP103:%.*]] = load i32, i32* [[I150]], align 4, !llvm.access.group !40 16418 // CHECK15-NEXT: [[ARRAYIDX157:%.*]] = getelementptr inbounds i32, i32* [[TMP102]], i32 [[TMP103]] 16419 // CHECK15-NEXT: [[TMP104:%.*]] = load i32, i32* [[ARRAYIDX157]], align 4, !llvm.access.group !40 16420 // CHECK15-NEXT: [[ADD158:%.*]] = add nsw i32 [[TMP101]], [[TMP104]] 16421 // CHECK15-NEXT: [[TMP105:%.*]] = load i32*, i32** [[A]], align 4, !llvm.access.group !40 16422 // CHECK15-NEXT: [[TMP106:%.*]] = load i32, i32* [[I150]], align 4, !llvm.access.group !40 16423 // CHECK15-NEXT: [[ARRAYIDX159:%.*]] = getelementptr inbounds i32, i32* [[TMP105]], i32 [[TMP106]] 16424 // CHECK15-NEXT: store i32 [[ADD158]], i32* [[ARRAYIDX159]], align 4, !llvm.access.group !40 16425 // CHECK15-NEXT: br label [[OMP_BODY_CONTINUE160:%.*]] 16426 // CHECK15: omp.body.continue160: 16427 // CHECK15-NEXT: br label [[OMP_INNER_FOR_INC161:%.*]] 16428 // CHECK15: omp.inner.for.inc161: 16429 // CHECK15-NEXT: [[TMP107:%.*]] = load i32, i32* [[DOTOMP_IV149]], align 4, !llvm.access.group !40 16430 // CHECK15-NEXT: [[ADD162:%.*]] = add nsw i32 [[TMP107]], 1 16431 // CHECK15-NEXT: store i32 [[ADD162]], i32* [[DOTOMP_IV149]], align 4, !llvm.access.group !40 16432 // CHECK15-NEXT: br label [[OMP_INNER_FOR_COND151]], !llvm.loop [[LOOP41:![0-9]+]] 16433 // CHECK15: omp.inner.for.end163: 16434 // CHECK15-NEXT: [[TMP108:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_139]], align 4 16435 // CHECK15-NEXT: [[SUB164:%.*]] = sub nsw i32 [[TMP108]], 0 16436 // CHECK15-NEXT: [[DIV165:%.*]] = sdiv i32 [[SUB164]], 1 16437 // CHECK15-NEXT: [[MUL166:%.*]] = mul nsw i32 [[DIV165]], 1 16438 // CHECK15-NEXT: [[ADD167:%.*]] = add nsw i32 0, [[MUL166]] 16439 // CHECK15-NEXT: store i32 [[ADD167]], i32* [[I150]], align 4 16440 // CHECK15-NEXT: br label [[SIMD_IF_END168]] 16441 // CHECK15: simd.if.end168: 16442 // CHECK15-NEXT: [[TMP109:%.*]] = load i32, i32* [[CH]], align 4 16443 // CHECK15-NEXT: store i32 [[TMP109]], i32* [[DOTCAPTURE_EXPR_169]], align 4 16444 // CHECK15-NEXT: [[TMP110:%.*]] = load i32, i32* [[N]], align 4 16445 // CHECK15-NEXT: store i32 [[TMP110]], i32* [[DOTCAPTURE_EXPR_171]], align 4 16446 // CHECK15-NEXT: [[TMP111:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_171]], align 4 16447 // CHECK15-NEXT: [[SUB173:%.*]] = sub nsw i32 [[TMP111]], 0 16448 // CHECK15-NEXT: [[DIV174:%.*]] = sdiv i32 [[SUB173]], 1 16449 // CHECK15-NEXT: [[SUB175:%.*]] = sub nsw i32 [[DIV174]], 1 16450 // CHECK15-NEXT: store i32 [[SUB175]], i32* [[DOTCAPTURE_EXPR_172]], align 4 16451 // CHECK15-NEXT: store i32 0, i32* [[DOTOMP_LB176]], align 4 16452 // CHECK15-NEXT: [[TMP112:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_172]], align 4 16453 // CHECK15-NEXT: store i32 [[TMP112]], i32* [[DOTOMP_UB177]], align 4 16454 // CHECK15-NEXT: store i32 0, i32* [[I178]], align 4 16455 // CHECK15-NEXT: [[TMP113:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_171]], align 4 16456 // CHECK15-NEXT: [[CMP179:%.*]] = icmp slt i32 0, [[TMP113]] 16457 // CHECK15-NEXT: br i1 [[CMP179]], label [[SIMD_IF_THEN180:%.*]], label [[SIMD_IF_END200:%.*]] 16458 // CHECK15: simd.if.then180: 16459 // CHECK15-NEXT: [[TMP114:%.*]] = load i32, i32* [[DOTOMP_LB176]], align 4 16460 // CHECK15-NEXT: store i32 [[TMP114]], i32* [[DOTOMP_IV181]], align 4 16461 // CHECK15-NEXT: br label [[OMP_INNER_FOR_COND183:%.*]] 16462 // CHECK15: omp.inner.for.cond183: 16463 // CHECK15-NEXT: [[TMP115:%.*]] = load i32, i32* [[DOTOMP_IV181]], align 4, !llvm.access.group !43 16464 // CHECK15-NEXT: [[TMP116:%.*]] = load i32, i32* [[DOTOMP_UB177]], align 4, !llvm.access.group !43 16465 // CHECK15-NEXT: [[CMP184:%.*]] = icmp sle i32 [[TMP115]], [[TMP116]] 16466 // CHECK15-NEXT: br i1 [[CMP184]], label [[OMP_INNER_FOR_BODY185:%.*]], label [[OMP_INNER_FOR_END195:%.*]] 16467 // CHECK15: omp.inner.for.body185: 16468 // CHECK15-NEXT: [[TMP117:%.*]] = load i32, i32* [[DOTOMP_IV181]], align 4, !llvm.access.group !43 16469 // CHECK15-NEXT: [[MUL186:%.*]] = mul nsw i32 [[TMP117]], 1 16470 // CHECK15-NEXT: [[ADD187:%.*]] = add nsw i32 0, [[MUL186]] 16471 // CHECK15-NEXT: store i32 [[ADD187]], i32* [[I182]], align 4, !llvm.access.group !43 16472 // CHECK15-NEXT: [[TMP118:%.*]] = load i32*, i32** [[B]], align 4, !llvm.access.group !43 16473 // CHECK15-NEXT: [[TMP119:%.*]] = load i32, i32* [[I182]], align 4, !llvm.access.group !43 16474 // CHECK15-NEXT: [[ARRAYIDX188:%.*]] = getelementptr inbounds i32, i32* [[TMP118]], i32 [[TMP119]] 16475 // CHECK15-NEXT: [[TMP120:%.*]] = load i32, i32* [[ARRAYIDX188]], align 4, !llvm.access.group !43 16476 // CHECK15-NEXT: [[TMP121:%.*]] = load i32*, i32** [[C]], align 4, !llvm.access.group !43 16477 // CHECK15-NEXT: [[TMP122:%.*]] = load i32, i32* [[I182]], align 4, !llvm.access.group !43 16478 // CHECK15-NEXT: [[ARRAYIDX189:%.*]] = getelementptr inbounds i32, i32* [[TMP121]], i32 [[TMP122]] 16479 // CHECK15-NEXT: [[TMP123:%.*]] = load i32, i32* [[ARRAYIDX189]], align 4, !llvm.access.group !43 16480 // CHECK15-NEXT: [[ADD190:%.*]] = add nsw i32 [[TMP120]], [[TMP123]] 16481 // CHECK15-NEXT: [[TMP124:%.*]] = load i32*, i32** [[A]], align 4, !llvm.access.group !43 16482 // CHECK15-NEXT: [[TMP125:%.*]] = load i32, i32* [[I182]], align 4, !llvm.access.group !43 16483 // CHECK15-NEXT: [[ARRAYIDX191:%.*]] = getelementptr inbounds i32, i32* [[TMP124]], i32 [[TMP125]] 16484 // CHECK15-NEXT: store i32 [[ADD190]], i32* [[ARRAYIDX191]], align 4, !llvm.access.group !43 16485 // CHECK15-NEXT: br label [[OMP_BODY_CONTINUE192:%.*]] 16486 // CHECK15: omp.body.continue192: 16487 // CHECK15-NEXT: br label [[OMP_INNER_FOR_INC193:%.*]] 16488 // CHECK15: omp.inner.for.inc193: 16489 // CHECK15-NEXT: [[TMP126:%.*]] = load i32, i32* [[DOTOMP_IV181]], align 4, !llvm.access.group !43 16490 // CHECK15-NEXT: [[ADD194:%.*]] = add nsw i32 [[TMP126]], 1 16491 // CHECK15-NEXT: store i32 [[ADD194]], i32* [[DOTOMP_IV181]], align 4, !llvm.access.group !43 16492 // CHECK15-NEXT: br label [[OMP_INNER_FOR_COND183]], !llvm.loop [[LOOP44:![0-9]+]] 16493 // CHECK15: omp.inner.for.end195: 16494 // CHECK15-NEXT: [[TMP127:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_171]], align 4 16495 // CHECK15-NEXT: [[SUB196:%.*]] = sub nsw i32 [[TMP127]], 0 16496 // CHECK15-NEXT: [[DIV197:%.*]] = sdiv i32 [[SUB196]], 1 16497 // CHECK15-NEXT: [[MUL198:%.*]] = mul nsw i32 [[DIV197]], 1 16498 // CHECK15-NEXT: [[ADD199:%.*]] = add nsw i32 0, [[MUL198]] 16499 // CHECK15-NEXT: store i32 [[ADD199]], i32* [[I182]], align 4 16500 // CHECK15-NEXT: br label [[SIMD_IF_END200]] 16501 // CHECK15: simd.if.end200: 16502 // CHECK15-NEXT: ret i32 0 16503 // 16504