1 //===-- lib/CodeGen/GlobalISel/GICombinerHelper.cpp -----------------------===// 2 // 3 // Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions. 4 // See https://llvm.org/LICENSE.txt for license information. 5 // SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception 6 // 7 //===----------------------------------------------------------------------===// 8 #include "llvm/CodeGen/GlobalISel/CombinerHelper.h" 9 #include "llvm/CodeGen/GlobalISel/Combiner.h" 10 #include "llvm/CodeGen/GlobalISel/GISelChangeObserver.h" 11 #include "llvm/CodeGen/GlobalISel/GISelKnownBits.h" 12 #include "llvm/CodeGen/GlobalISel/LegalizerInfo.h" 13 #include "llvm/CodeGen/GlobalISel/MIPatternMatch.h" 14 #include "llvm/CodeGen/GlobalISel/MachineIRBuilder.h" 15 #include "llvm/CodeGen/GlobalISel/Utils.h" 16 #include "llvm/CodeGen/MachineDominators.h" 17 #include "llvm/CodeGen/MachineFrameInfo.h" 18 #include "llvm/CodeGen/MachineInstr.h" 19 #include "llvm/CodeGen/MachineRegisterInfo.h" 20 #include "llvm/CodeGen/TargetInstrInfo.h" 21 #include "llvm/CodeGen/TargetLowering.h" 22 #include "llvm/Support/MathExtras.h" 23 #include "llvm/Target/TargetMachine.h" 24 25 #define DEBUG_TYPE "gi-combiner" 26 27 using namespace llvm; 28 using namespace MIPatternMatch; 29 30 // Option to allow testing of the combiner while no targets know about indexed 31 // addressing. 32 static cl::opt<bool> 33 ForceLegalIndexing("force-legal-indexing", cl::Hidden, cl::init(false), 34 cl::desc("Force all indexed operations to be " 35 "legal for the GlobalISel combiner")); 36 37 38 CombinerHelper::CombinerHelper(GISelChangeObserver &Observer, 39 MachineIRBuilder &B, GISelKnownBits *KB, 40 MachineDominatorTree *MDT, 41 const LegalizerInfo *LI) 42 : Builder(B), MRI(Builder.getMF().getRegInfo()), Observer(Observer), 43 KB(KB), MDT(MDT), LI(LI) { 44 (void)this->KB; 45 } 46 47 void CombinerHelper::replaceRegWith(MachineRegisterInfo &MRI, Register FromReg, 48 Register ToReg) const { 49 Observer.changingAllUsesOfReg(MRI, FromReg); 50 51 if (MRI.constrainRegAttrs(ToReg, FromReg)) 52 MRI.replaceRegWith(FromReg, ToReg); 53 else 54 Builder.buildCopy(ToReg, FromReg); 55 56 Observer.finishedChangingAllUsesOfReg(); 57 } 58 59 void CombinerHelper::replaceRegOpWith(MachineRegisterInfo &MRI, 60 MachineOperand &FromRegOp, 61 Register ToReg) const { 62 assert(FromRegOp.getParent() && "Expected an operand in an MI"); 63 Observer.changingInstr(*FromRegOp.getParent()); 64 65 FromRegOp.setReg(ToReg); 66 67 Observer.changedInstr(*FromRegOp.getParent()); 68 } 69 70 bool CombinerHelper::tryCombineCopy(MachineInstr &MI) { 71 if (matchCombineCopy(MI)) { 72 applyCombineCopy(MI); 73 return true; 74 } 75 return false; 76 } 77 bool CombinerHelper::matchCombineCopy(MachineInstr &MI) { 78 if (MI.getOpcode() != TargetOpcode::COPY) 79 return false; 80 Register DstReg = MI.getOperand(0).getReg(); 81 Register SrcReg = MI.getOperand(1).getReg(); 82 return canReplaceReg(DstReg, SrcReg, MRI); 83 } 84 void CombinerHelper::applyCombineCopy(MachineInstr &MI) { 85 Register DstReg = MI.getOperand(0).getReg(); 86 Register SrcReg = MI.getOperand(1).getReg(); 87 MI.eraseFromParent(); 88 replaceRegWith(MRI, DstReg, SrcReg); 89 } 90 91 bool CombinerHelper::tryCombineConcatVectors(MachineInstr &MI) { 92 bool IsUndef = false; 93 SmallVector<Register, 4> Ops; 94 if (matchCombineConcatVectors(MI, IsUndef, Ops)) { 95 applyCombineConcatVectors(MI, IsUndef, Ops); 96 return true; 97 } 98 return false; 99 } 100 101 bool CombinerHelper::matchCombineConcatVectors(MachineInstr &MI, bool &IsUndef, 102 SmallVectorImpl<Register> &Ops) { 103 assert(MI.getOpcode() == TargetOpcode::G_CONCAT_VECTORS && 104 "Invalid instruction"); 105 IsUndef = true; 106 MachineInstr *Undef = nullptr; 107 108 // Walk over all the operands of concat vectors and check if they are 109 // build_vector themselves or undef. 110 // Then collect their operands in Ops. 111 for (const MachineOperand &MO : MI.uses()) { 112 Register Reg = MO.getReg(); 113 MachineInstr *Def = MRI.getVRegDef(Reg); 114 assert(Def && "Operand not defined"); 115 switch (Def->getOpcode()) { 116 case TargetOpcode::G_BUILD_VECTOR: 117 IsUndef = false; 118 // Remember the operands of the build_vector to fold 119 // them into the yet-to-build flattened concat vectors. 120 for (const MachineOperand &BuildVecMO : Def->uses()) 121 Ops.push_back(BuildVecMO.getReg()); 122 break; 123 case TargetOpcode::G_IMPLICIT_DEF: { 124 LLT OpType = MRI.getType(Reg); 125 // Keep one undef value for all the undef operands. 126 if (!Undef) { 127 Builder.setInsertPt(*MI.getParent(), MI); 128 Undef = Builder.buildUndef(OpType.getScalarType()); 129 } 130 assert(MRI.getType(Undef->getOperand(0).getReg()) == 131 OpType.getScalarType() && 132 "All undefs should have the same type"); 133 // Break the undef vector in as many scalar elements as needed 134 // for the flattening. 135 for (unsigned EltIdx = 0, EltEnd = OpType.getNumElements(); 136 EltIdx != EltEnd; ++EltIdx) 137 Ops.push_back(Undef->getOperand(0).getReg()); 138 break; 139 } 140 default: 141 return false; 142 } 143 } 144 return true; 145 } 146 void CombinerHelper::applyCombineConcatVectors( 147 MachineInstr &MI, bool IsUndef, const ArrayRef<Register> Ops) { 148 // We determined that the concat_vectors can be flatten. 149 // Generate the flattened build_vector. 150 Register DstReg = MI.getOperand(0).getReg(); 151 Builder.setInsertPt(*MI.getParent(), MI); 152 Register NewDstReg = MRI.cloneVirtualRegister(DstReg); 153 154 // Note: IsUndef is sort of redundant. We could have determine it by 155 // checking that at all Ops are undef. Alternatively, we could have 156 // generate a build_vector of undefs and rely on another combine to 157 // clean that up. For now, given we already gather this information 158 // in tryCombineConcatVectors, just save compile time and issue the 159 // right thing. 160 if (IsUndef) 161 Builder.buildUndef(NewDstReg); 162 else 163 Builder.buildBuildVector(NewDstReg, Ops); 164 MI.eraseFromParent(); 165 replaceRegWith(MRI, DstReg, NewDstReg); 166 } 167 168 bool CombinerHelper::tryCombineShuffleVector(MachineInstr &MI) { 169 SmallVector<Register, 4> Ops; 170 if (matchCombineShuffleVector(MI, Ops)) { 171 applyCombineShuffleVector(MI, Ops); 172 return true; 173 } 174 return false; 175 } 176 177 bool CombinerHelper::matchCombineShuffleVector(MachineInstr &MI, 178 SmallVectorImpl<Register> &Ops) { 179 assert(MI.getOpcode() == TargetOpcode::G_SHUFFLE_VECTOR && 180 "Invalid instruction kind"); 181 LLT DstType = MRI.getType(MI.getOperand(0).getReg()); 182 Register Src1 = MI.getOperand(1).getReg(); 183 LLT SrcType = MRI.getType(Src1); 184 // As bizarre as it may look, shuffle vector can actually produce 185 // scalar! This is because at the IR level a <1 x ty> shuffle 186 // vector is perfectly valid. 187 unsigned DstNumElts = DstType.isVector() ? DstType.getNumElements() : 1; 188 unsigned SrcNumElts = SrcType.isVector() ? SrcType.getNumElements() : 1; 189 190 // If the resulting vector is smaller than the size of the source 191 // vectors being concatenated, we won't be able to replace the 192 // shuffle vector into a concat_vectors. 193 // 194 // Note: We may still be able to produce a concat_vectors fed by 195 // extract_vector_elt and so on. It is less clear that would 196 // be better though, so don't bother for now. 197 // 198 // If the destination is a scalar, the size of the sources doesn't 199 // matter. we will lower the shuffle to a plain copy. This will 200 // work only if the source and destination have the same size. But 201 // that's covered by the next condition. 202 // 203 // TODO: If the size between the source and destination don't match 204 // we could still emit an extract vector element in that case. 205 if (DstNumElts < 2 * SrcNumElts && DstNumElts != 1) 206 return false; 207 208 // Check that the shuffle mask can be broken evenly between the 209 // different sources. 210 if (DstNumElts % SrcNumElts != 0) 211 return false; 212 213 // Mask length is a multiple of the source vector length. 214 // Check if the shuffle is some kind of concatenation of the input 215 // vectors. 216 unsigned NumConcat = DstNumElts / SrcNumElts; 217 SmallVector<int, 8> ConcatSrcs(NumConcat, -1); 218 ArrayRef<int> Mask = MI.getOperand(3).getShuffleMask(); 219 for (unsigned i = 0; i != DstNumElts; ++i) { 220 int Idx = Mask[i]; 221 // Undef value. 222 if (Idx < 0) 223 continue; 224 // Ensure the indices in each SrcType sized piece are sequential and that 225 // the same source is used for the whole piece. 226 if ((Idx % SrcNumElts != (i % SrcNumElts)) || 227 (ConcatSrcs[i / SrcNumElts] >= 0 && 228 ConcatSrcs[i / SrcNumElts] != (int)(Idx / SrcNumElts))) 229 return false; 230 // Remember which source this index came from. 231 ConcatSrcs[i / SrcNumElts] = Idx / SrcNumElts; 232 } 233 234 // The shuffle is concatenating multiple vectors together. 235 // Collect the different operands for that. 236 Register UndefReg; 237 Register Src2 = MI.getOperand(2).getReg(); 238 for (auto Src : ConcatSrcs) { 239 if (Src < 0) { 240 if (!UndefReg) { 241 Builder.setInsertPt(*MI.getParent(), MI); 242 UndefReg = Builder.buildUndef(SrcType).getReg(0); 243 } 244 Ops.push_back(UndefReg); 245 } else if (Src == 0) 246 Ops.push_back(Src1); 247 else 248 Ops.push_back(Src2); 249 } 250 return true; 251 } 252 253 void CombinerHelper::applyCombineShuffleVector(MachineInstr &MI, 254 const ArrayRef<Register> Ops) { 255 Register DstReg = MI.getOperand(0).getReg(); 256 Builder.setInsertPt(*MI.getParent(), MI); 257 Register NewDstReg = MRI.cloneVirtualRegister(DstReg); 258 259 if (Ops.size() == 1) 260 Builder.buildCopy(NewDstReg, Ops[0]); 261 else 262 Builder.buildMerge(NewDstReg, Ops); 263 264 MI.eraseFromParent(); 265 replaceRegWith(MRI, DstReg, NewDstReg); 266 } 267 268 namespace { 269 270 /// Select a preference between two uses. CurrentUse is the current preference 271 /// while *ForCandidate is attributes of the candidate under consideration. 272 PreferredTuple ChoosePreferredUse(PreferredTuple &CurrentUse, 273 const LLT TyForCandidate, 274 unsigned OpcodeForCandidate, 275 MachineInstr *MIForCandidate) { 276 if (!CurrentUse.Ty.isValid()) { 277 if (CurrentUse.ExtendOpcode == OpcodeForCandidate || 278 CurrentUse.ExtendOpcode == TargetOpcode::G_ANYEXT) 279 return {TyForCandidate, OpcodeForCandidate, MIForCandidate}; 280 return CurrentUse; 281 } 282 283 // We permit the extend to hoist through basic blocks but this is only 284 // sensible if the target has extending loads. If you end up lowering back 285 // into a load and extend during the legalizer then the end result is 286 // hoisting the extend up to the load. 287 288 // Prefer defined extensions to undefined extensions as these are more 289 // likely to reduce the number of instructions. 290 if (OpcodeForCandidate == TargetOpcode::G_ANYEXT && 291 CurrentUse.ExtendOpcode != TargetOpcode::G_ANYEXT) 292 return CurrentUse; 293 else if (CurrentUse.ExtendOpcode == TargetOpcode::G_ANYEXT && 294 OpcodeForCandidate != TargetOpcode::G_ANYEXT) 295 return {TyForCandidate, OpcodeForCandidate, MIForCandidate}; 296 297 // Prefer sign extensions to zero extensions as sign-extensions tend to be 298 // more expensive. 299 if (CurrentUse.Ty == TyForCandidate) { 300 if (CurrentUse.ExtendOpcode == TargetOpcode::G_SEXT && 301 OpcodeForCandidate == TargetOpcode::G_ZEXT) 302 return CurrentUse; 303 else if (CurrentUse.ExtendOpcode == TargetOpcode::G_ZEXT && 304 OpcodeForCandidate == TargetOpcode::G_SEXT) 305 return {TyForCandidate, OpcodeForCandidate, MIForCandidate}; 306 } 307 308 // This is potentially target specific. We've chosen the largest type 309 // because G_TRUNC is usually free. One potential catch with this is that 310 // some targets have a reduced number of larger registers than smaller 311 // registers and this choice potentially increases the live-range for the 312 // larger value. 313 if (TyForCandidate.getSizeInBits() > CurrentUse.Ty.getSizeInBits()) { 314 return {TyForCandidate, OpcodeForCandidate, MIForCandidate}; 315 } 316 return CurrentUse; 317 } 318 319 /// Find a suitable place to insert some instructions and insert them. This 320 /// function accounts for special cases like inserting before a PHI node. 321 /// The current strategy for inserting before PHI's is to duplicate the 322 /// instructions for each predecessor. However, while that's ok for G_TRUNC 323 /// on most targets since it generally requires no code, other targets/cases may 324 /// want to try harder to find a dominating block. 325 static void InsertInsnsWithoutSideEffectsBeforeUse( 326 MachineIRBuilder &Builder, MachineInstr &DefMI, MachineOperand &UseMO, 327 std::function<void(MachineBasicBlock *, MachineBasicBlock::iterator, 328 MachineOperand &UseMO)> 329 Inserter) { 330 MachineInstr &UseMI = *UseMO.getParent(); 331 332 MachineBasicBlock *InsertBB = UseMI.getParent(); 333 334 // If the use is a PHI then we want the predecessor block instead. 335 if (UseMI.isPHI()) { 336 MachineOperand *PredBB = std::next(&UseMO); 337 InsertBB = PredBB->getMBB(); 338 } 339 340 // If the block is the same block as the def then we want to insert just after 341 // the def instead of at the start of the block. 342 if (InsertBB == DefMI.getParent()) { 343 MachineBasicBlock::iterator InsertPt = &DefMI; 344 Inserter(InsertBB, std::next(InsertPt), UseMO); 345 return; 346 } 347 348 // Otherwise we want the start of the BB 349 Inserter(InsertBB, InsertBB->getFirstNonPHI(), UseMO); 350 } 351 } // end anonymous namespace 352 353 bool CombinerHelper::tryCombineExtendingLoads(MachineInstr &MI) { 354 PreferredTuple Preferred; 355 if (matchCombineExtendingLoads(MI, Preferred)) { 356 applyCombineExtendingLoads(MI, Preferred); 357 return true; 358 } 359 return false; 360 } 361 362 bool CombinerHelper::matchCombineExtendingLoads(MachineInstr &MI, 363 PreferredTuple &Preferred) { 364 // We match the loads and follow the uses to the extend instead of matching 365 // the extends and following the def to the load. This is because the load 366 // must remain in the same position for correctness (unless we also add code 367 // to find a safe place to sink it) whereas the extend is freely movable. 368 // It also prevents us from duplicating the load for the volatile case or just 369 // for performance. 370 371 if (MI.getOpcode() != TargetOpcode::G_LOAD && 372 MI.getOpcode() != TargetOpcode::G_SEXTLOAD && 373 MI.getOpcode() != TargetOpcode::G_ZEXTLOAD) 374 return false; 375 376 auto &LoadValue = MI.getOperand(0); 377 assert(LoadValue.isReg() && "Result wasn't a register?"); 378 379 LLT LoadValueTy = MRI.getType(LoadValue.getReg()); 380 if (!LoadValueTy.isScalar()) 381 return false; 382 383 // Most architectures are going to legalize <s8 loads into at least a 1 byte 384 // load, and the MMOs can only describe memory accesses in multiples of bytes. 385 // If we try to perform extload combining on those, we can end up with 386 // %a(s8) = extload %ptr (load 1 byte from %ptr) 387 // ... which is an illegal extload instruction. 388 if (LoadValueTy.getSizeInBits() < 8) 389 return false; 390 391 // For non power-of-2 types, they will very likely be legalized into multiple 392 // loads. Don't bother trying to match them into extending loads. 393 if (!isPowerOf2_32(LoadValueTy.getSizeInBits())) 394 return false; 395 396 // Find the preferred type aside from the any-extends (unless it's the only 397 // one) and non-extending ops. We'll emit an extending load to that type and 398 // and emit a variant of (extend (trunc X)) for the others according to the 399 // relative type sizes. At the same time, pick an extend to use based on the 400 // extend involved in the chosen type. 401 unsigned PreferredOpcode = MI.getOpcode() == TargetOpcode::G_LOAD 402 ? TargetOpcode::G_ANYEXT 403 : MI.getOpcode() == TargetOpcode::G_SEXTLOAD 404 ? TargetOpcode::G_SEXT 405 : TargetOpcode::G_ZEXT; 406 Preferred = {LLT(), PreferredOpcode, nullptr}; 407 for (auto &UseMI : MRI.use_nodbg_instructions(LoadValue.getReg())) { 408 if (UseMI.getOpcode() == TargetOpcode::G_SEXT || 409 UseMI.getOpcode() == TargetOpcode::G_ZEXT || 410 (UseMI.getOpcode() == TargetOpcode::G_ANYEXT)) { 411 // Check for legality. 412 if (LI) { 413 LegalityQuery::MemDesc MMDesc; 414 const auto &MMO = **MI.memoperands_begin(); 415 MMDesc.SizeInBits = MMO.getSizeInBits(); 416 MMDesc.AlignInBits = MMO.getAlign().value() * 8; 417 MMDesc.Ordering = MMO.getOrdering(); 418 LLT UseTy = MRI.getType(UseMI.getOperand(0).getReg()); 419 LLT SrcTy = MRI.getType(MI.getOperand(1).getReg()); 420 if (LI->getAction({MI.getOpcode(), {UseTy, SrcTy}, {MMDesc}}).Action != 421 LegalizeActions::Legal) 422 continue; 423 } 424 Preferred = ChoosePreferredUse(Preferred, 425 MRI.getType(UseMI.getOperand(0).getReg()), 426 UseMI.getOpcode(), &UseMI); 427 } 428 } 429 430 // There were no extends 431 if (!Preferred.MI) 432 return false; 433 // It should be impossible to chose an extend without selecting a different 434 // type since by definition the result of an extend is larger. 435 assert(Preferred.Ty != LoadValueTy && "Extending to same type?"); 436 437 LLVM_DEBUG(dbgs() << "Preferred use is: " << *Preferred.MI); 438 return true; 439 } 440 441 void CombinerHelper::applyCombineExtendingLoads(MachineInstr &MI, 442 PreferredTuple &Preferred) { 443 // Rewrite the load to the chosen extending load. 444 Register ChosenDstReg = Preferred.MI->getOperand(0).getReg(); 445 446 // Inserter to insert a truncate back to the original type at a given point 447 // with some basic CSE to limit truncate duplication to one per BB. 448 DenseMap<MachineBasicBlock *, MachineInstr *> EmittedInsns; 449 auto InsertTruncAt = [&](MachineBasicBlock *InsertIntoBB, 450 MachineBasicBlock::iterator InsertBefore, 451 MachineOperand &UseMO) { 452 MachineInstr *PreviouslyEmitted = EmittedInsns.lookup(InsertIntoBB); 453 if (PreviouslyEmitted) { 454 Observer.changingInstr(*UseMO.getParent()); 455 UseMO.setReg(PreviouslyEmitted->getOperand(0).getReg()); 456 Observer.changedInstr(*UseMO.getParent()); 457 return; 458 } 459 460 Builder.setInsertPt(*InsertIntoBB, InsertBefore); 461 Register NewDstReg = MRI.cloneVirtualRegister(MI.getOperand(0).getReg()); 462 MachineInstr *NewMI = Builder.buildTrunc(NewDstReg, ChosenDstReg); 463 EmittedInsns[InsertIntoBB] = NewMI; 464 replaceRegOpWith(MRI, UseMO, NewDstReg); 465 }; 466 467 Observer.changingInstr(MI); 468 MI.setDesc( 469 Builder.getTII().get(Preferred.ExtendOpcode == TargetOpcode::G_SEXT 470 ? TargetOpcode::G_SEXTLOAD 471 : Preferred.ExtendOpcode == TargetOpcode::G_ZEXT 472 ? TargetOpcode::G_ZEXTLOAD 473 : TargetOpcode::G_LOAD)); 474 475 // Rewrite all the uses to fix up the types. 476 auto &LoadValue = MI.getOperand(0); 477 SmallVector<MachineOperand *, 4> Uses; 478 for (auto &UseMO : MRI.use_operands(LoadValue.getReg())) 479 Uses.push_back(&UseMO); 480 481 for (auto *UseMO : Uses) { 482 MachineInstr *UseMI = UseMO->getParent(); 483 484 // If the extend is compatible with the preferred extend then we should fix 485 // up the type and extend so that it uses the preferred use. 486 if (UseMI->getOpcode() == Preferred.ExtendOpcode || 487 UseMI->getOpcode() == TargetOpcode::G_ANYEXT) { 488 Register UseDstReg = UseMI->getOperand(0).getReg(); 489 MachineOperand &UseSrcMO = UseMI->getOperand(1); 490 const LLT UseDstTy = MRI.getType(UseDstReg); 491 if (UseDstReg != ChosenDstReg) { 492 if (Preferred.Ty == UseDstTy) { 493 // If the use has the same type as the preferred use, then merge 494 // the vregs and erase the extend. For example: 495 // %1:_(s8) = G_LOAD ... 496 // %2:_(s32) = G_SEXT %1(s8) 497 // %3:_(s32) = G_ANYEXT %1(s8) 498 // ... = ... %3(s32) 499 // rewrites to: 500 // %2:_(s32) = G_SEXTLOAD ... 501 // ... = ... %2(s32) 502 replaceRegWith(MRI, UseDstReg, ChosenDstReg); 503 Observer.erasingInstr(*UseMO->getParent()); 504 UseMO->getParent()->eraseFromParent(); 505 } else if (Preferred.Ty.getSizeInBits() < UseDstTy.getSizeInBits()) { 506 // If the preferred size is smaller, then keep the extend but extend 507 // from the result of the extending load. For example: 508 // %1:_(s8) = G_LOAD ... 509 // %2:_(s32) = G_SEXT %1(s8) 510 // %3:_(s64) = G_ANYEXT %1(s8) 511 // ... = ... %3(s64) 512 /// rewrites to: 513 // %2:_(s32) = G_SEXTLOAD ... 514 // %3:_(s64) = G_ANYEXT %2:_(s32) 515 // ... = ... %3(s64) 516 replaceRegOpWith(MRI, UseSrcMO, ChosenDstReg); 517 } else { 518 // If the preferred size is large, then insert a truncate. For 519 // example: 520 // %1:_(s8) = G_LOAD ... 521 // %2:_(s64) = G_SEXT %1(s8) 522 // %3:_(s32) = G_ZEXT %1(s8) 523 // ... = ... %3(s32) 524 /// rewrites to: 525 // %2:_(s64) = G_SEXTLOAD ... 526 // %4:_(s8) = G_TRUNC %2:_(s32) 527 // %3:_(s64) = G_ZEXT %2:_(s8) 528 // ... = ... %3(s64) 529 InsertInsnsWithoutSideEffectsBeforeUse(Builder, MI, *UseMO, 530 InsertTruncAt); 531 } 532 continue; 533 } 534 // The use is (one of) the uses of the preferred use we chose earlier. 535 // We're going to update the load to def this value later so just erase 536 // the old extend. 537 Observer.erasingInstr(*UseMO->getParent()); 538 UseMO->getParent()->eraseFromParent(); 539 continue; 540 } 541 542 // The use isn't an extend. Truncate back to the type we originally loaded. 543 // This is free on many targets. 544 InsertInsnsWithoutSideEffectsBeforeUse(Builder, MI, *UseMO, InsertTruncAt); 545 } 546 547 MI.getOperand(0).setReg(ChosenDstReg); 548 Observer.changedInstr(MI); 549 } 550 551 bool CombinerHelper::isPredecessor(const MachineInstr &DefMI, 552 const MachineInstr &UseMI) { 553 assert(!DefMI.isDebugInstr() && !UseMI.isDebugInstr() && 554 "shouldn't consider debug uses"); 555 assert(DefMI.getParent() == UseMI.getParent()); 556 if (&DefMI == &UseMI) 557 return false; 558 559 // Loop through the basic block until we find one of the instructions. 560 MachineBasicBlock::const_iterator I = DefMI.getParent()->begin(); 561 for (; &*I != &DefMI && &*I != &UseMI; ++I) 562 return &*I == &DefMI; 563 564 llvm_unreachable("Block must contain instructions"); 565 } 566 567 bool CombinerHelper::dominates(const MachineInstr &DefMI, 568 const MachineInstr &UseMI) { 569 assert(!DefMI.isDebugInstr() && !UseMI.isDebugInstr() && 570 "shouldn't consider debug uses"); 571 if (MDT) 572 return MDT->dominates(&DefMI, &UseMI); 573 else if (DefMI.getParent() != UseMI.getParent()) 574 return false; 575 576 return isPredecessor(DefMI, UseMI); 577 } 578 579 bool CombinerHelper::matchSextAlreadyExtended(MachineInstr &MI) { 580 assert(MI.getOpcode() == TargetOpcode::G_SEXT_INREG); 581 Register SrcReg = MI.getOperand(1).getReg(); 582 unsigned SrcSignBits = KB->computeNumSignBits(SrcReg); 583 unsigned NumSextBits = 584 MRI.getType(MI.getOperand(0).getReg()).getScalarSizeInBits() - 585 MI.getOperand(2).getImm(); 586 return SrcSignBits >= NumSextBits; 587 } 588 589 bool CombinerHelper::applySextAlreadyExtended(MachineInstr &MI) { 590 assert(MI.getOpcode() == TargetOpcode::G_SEXT_INREG); 591 MachineIRBuilder MIB(MI); 592 MIB.buildCopy(MI.getOperand(0).getReg(), MI.getOperand(1).getReg()); 593 MI.eraseFromParent(); 594 return true; 595 } 596 597 bool CombinerHelper::findPostIndexCandidate(MachineInstr &MI, Register &Addr, 598 Register &Base, Register &Offset) { 599 auto &MF = *MI.getParent()->getParent(); 600 const auto &TLI = *MF.getSubtarget().getTargetLowering(); 601 602 #ifndef NDEBUG 603 unsigned Opcode = MI.getOpcode(); 604 assert(Opcode == TargetOpcode::G_LOAD || Opcode == TargetOpcode::G_SEXTLOAD || 605 Opcode == TargetOpcode::G_ZEXTLOAD || Opcode == TargetOpcode::G_STORE); 606 #endif 607 608 Base = MI.getOperand(1).getReg(); 609 MachineInstr *BaseDef = MRI.getUniqueVRegDef(Base); 610 if (BaseDef && BaseDef->getOpcode() == TargetOpcode::G_FRAME_INDEX) 611 return false; 612 613 LLVM_DEBUG(dbgs() << "Searching for post-indexing opportunity for: " << MI); 614 615 for (auto &Use : MRI.use_nodbg_instructions(Base)) { 616 if (Use.getOpcode() != TargetOpcode::G_PTR_ADD) 617 continue; 618 619 Offset = Use.getOperand(2).getReg(); 620 if (!ForceLegalIndexing && 621 !TLI.isIndexingLegal(MI, Base, Offset, /*IsPre*/ false, MRI)) { 622 LLVM_DEBUG(dbgs() << " Ignoring candidate with illegal addrmode: " 623 << Use); 624 continue; 625 } 626 627 // Make sure the offset calculation is before the potentially indexed op. 628 // FIXME: we really care about dependency here. The offset calculation might 629 // be movable. 630 MachineInstr *OffsetDef = MRI.getUniqueVRegDef(Offset); 631 if (!OffsetDef || !dominates(*OffsetDef, MI)) { 632 LLVM_DEBUG(dbgs() << " Ignoring candidate with offset after mem-op: " 633 << Use); 634 continue; 635 } 636 637 // FIXME: check whether all uses of Base are load/store with foldable 638 // addressing modes. If so, using the normal addr-modes is better than 639 // forming an indexed one. 640 641 bool MemOpDominatesAddrUses = true; 642 for (auto &PtrAddUse : 643 MRI.use_nodbg_instructions(Use.getOperand(0).getReg())) { 644 if (!dominates(MI, PtrAddUse)) { 645 MemOpDominatesAddrUses = false; 646 break; 647 } 648 } 649 650 if (!MemOpDominatesAddrUses) { 651 LLVM_DEBUG( 652 dbgs() << " Ignoring candidate as memop does not dominate uses: " 653 << Use); 654 continue; 655 } 656 657 LLVM_DEBUG(dbgs() << " Found match: " << Use); 658 Addr = Use.getOperand(0).getReg(); 659 return true; 660 } 661 662 return false; 663 } 664 665 bool CombinerHelper::findPreIndexCandidate(MachineInstr &MI, Register &Addr, 666 Register &Base, Register &Offset) { 667 auto &MF = *MI.getParent()->getParent(); 668 const auto &TLI = *MF.getSubtarget().getTargetLowering(); 669 670 #ifndef NDEBUG 671 unsigned Opcode = MI.getOpcode(); 672 assert(Opcode == TargetOpcode::G_LOAD || Opcode == TargetOpcode::G_SEXTLOAD || 673 Opcode == TargetOpcode::G_ZEXTLOAD || Opcode == TargetOpcode::G_STORE); 674 #endif 675 676 Addr = MI.getOperand(1).getReg(); 677 MachineInstr *AddrDef = getOpcodeDef(TargetOpcode::G_PTR_ADD, Addr, MRI); 678 if (!AddrDef || MRI.hasOneNonDBGUse(Addr)) 679 return false; 680 681 Base = AddrDef->getOperand(1).getReg(); 682 Offset = AddrDef->getOperand(2).getReg(); 683 684 LLVM_DEBUG(dbgs() << "Found potential pre-indexed load_store: " << MI); 685 686 if (!ForceLegalIndexing && 687 !TLI.isIndexingLegal(MI, Base, Offset, /*IsPre*/ true, MRI)) { 688 LLVM_DEBUG(dbgs() << " Skipping, not legal for target"); 689 return false; 690 } 691 692 MachineInstr *BaseDef = getDefIgnoringCopies(Base, MRI); 693 if (BaseDef->getOpcode() == TargetOpcode::G_FRAME_INDEX) { 694 LLVM_DEBUG(dbgs() << " Skipping, frame index would need copy anyway."); 695 return false; 696 } 697 698 if (MI.getOpcode() == TargetOpcode::G_STORE) { 699 // Would require a copy. 700 if (Base == MI.getOperand(0).getReg()) { 701 LLVM_DEBUG(dbgs() << " Skipping, storing base so need copy anyway."); 702 return false; 703 } 704 705 // We're expecting one use of Addr in MI, but it could also be the 706 // value stored, which isn't actually dominated by the instruction. 707 if (MI.getOperand(0).getReg() == Addr) { 708 LLVM_DEBUG(dbgs() << " Skipping, does not dominate all addr uses"); 709 return false; 710 } 711 } 712 713 // FIXME: check whether all uses of the base pointer are constant PtrAdds. 714 // That might allow us to end base's liveness here by adjusting the constant. 715 716 for (auto &UseMI : MRI.use_nodbg_instructions(Addr)) { 717 if (!dominates(MI, UseMI)) { 718 LLVM_DEBUG(dbgs() << " Skipping, does not dominate all addr uses."); 719 return false; 720 } 721 } 722 723 return true; 724 } 725 726 bool CombinerHelper::tryCombineIndexedLoadStore(MachineInstr &MI) { 727 IndexedLoadStoreMatchInfo MatchInfo; 728 if (matchCombineIndexedLoadStore(MI, MatchInfo)) { 729 applyCombineIndexedLoadStore(MI, MatchInfo); 730 return true; 731 } 732 return false; 733 } 734 735 bool CombinerHelper::matchCombineIndexedLoadStore(MachineInstr &MI, IndexedLoadStoreMatchInfo &MatchInfo) { 736 unsigned Opcode = MI.getOpcode(); 737 if (Opcode != TargetOpcode::G_LOAD && Opcode != TargetOpcode::G_SEXTLOAD && 738 Opcode != TargetOpcode::G_ZEXTLOAD && Opcode != TargetOpcode::G_STORE) 739 return false; 740 741 MatchInfo.IsPre = findPreIndexCandidate(MI, MatchInfo.Addr, MatchInfo.Base, 742 MatchInfo.Offset); 743 if (!MatchInfo.IsPre && 744 !findPostIndexCandidate(MI, MatchInfo.Addr, MatchInfo.Base, 745 MatchInfo.Offset)) 746 return false; 747 748 return true; 749 } 750 751 void CombinerHelper::applyCombineIndexedLoadStore( 752 MachineInstr &MI, IndexedLoadStoreMatchInfo &MatchInfo) { 753 MachineInstr &AddrDef = *MRI.getUniqueVRegDef(MatchInfo.Addr); 754 MachineIRBuilder MIRBuilder(MI); 755 unsigned Opcode = MI.getOpcode(); 756 bool IsStore = Opcode == TargetOpcode::G_STORE; 757 unsigned NewOpcode; 758 switch (Opcode) { 759 case TargetOpcode::G_LOAD: 760 NewOpcode = TargetOpcode::G_INDEXED_LOAD; 761 break; 762 case TargetOpcode::G_SEXTLOAD: 763 NewOpcode = TargetOpcode::G_INDEXED_SEXTLOAD; 764 break; 765 case TargetOpcode::G_ZEXTLOAD: 766 NewOpcode = TargetOpcode::G_INDEXED_ZEXTLOAD; 767 break; 768 case TargetOpcode::G_STORE: 769 NewOpcode = TargetOpcode::G_INDEXED_STORE; 770 break; 771 default: 772 llvm_unreachable("Unknown load/store opcode"); 773 } 774 775 auto MIB = MIRBuilder.buildInstr(NewOpcode); 776 if (IsStore) { 777 MIB.addDef(MatchInfo.Addr); 778 MIB.addUse(MI.getOperand(0).getReg()); 779 } else { 780 MIB.addDef(MI.getOperand(0).getReg()); 781 MIB.addDef(MatchInfo.Addr); 782 } 783 784 MIB.addUse(MatchInfo.Base); 785 MIB.addUse(MatchInfo.Offset); 786 MIB.addImm(MatchInfo.IsPre); 787 MI.eraseFromParent(); 788 AddrDef.eraseFromParent(); 789 790 LLVM_DEBUG(dbgs() << " Combinined to indexed operation"); 791 } 792 793 bool CombinerHelper::matchElideBrByInvertingCond(MachineInstr &MI) { 794 if (MI.getOpcode() != TargetOpcode::G_BR) 795 return false; 796 797 // Try to match the following: 798 // bb1: 799 // %c(s32) = G_ICMP pred, %a, %b 800 // %c1(s1) = G_TRUNC %c(s32) 801 // G_BRCOND %c1, %bb2 802 // G_BR %bb3 803 // bb2: 804 // ... 805 // bb3: 806 807 // The above pattern does not have a fall through to the successor bb2, always 808 // resulting in a branch no matter which path is taken. Here we try to find 809 // and replace that pattern with conditional branch to bb3 and otherwise 810 // fallthrough to bb2. 811 812 MachineBasicBlock *MBB = MI.getParent(); 813 MachineBasicBlock::iterator BrIt(MI); 814 if (BrIt == MBB->begin()) 815 return false; 816 assert(std::next(BrIt) == MBB->end() && "expected G_BR to be a terminator"); 817 818 MachineInstr *BrCond = &*std::prev(BrIt); 819 if (BrCond->getOpcode() != TargetOpcode::G_BRCOND) 820 return false; 821 822 // Check that the next block is the conditional branch target. 823 if (!MBB->isLayoutSuccessor(BrCond->getOperand(1).getMBB())) 824 return false; 825 826 MachineInstr *CmpMI = MRI.getVRegDef(BrCond->getOperand(0).getReg()); 827 if (!CmpMI || CmpMI->getOpcode() != TargetOpcode::G_ICMP || 828 !MRI.hasOneNonDBGUse(CmpMI->getOperand(0).getReg())) 829 return false; 830 return true; 831 } 832 833 bool CombinerHelper::tryElideBrByInvertingCond(MachineInstr &MI) { 834 if (!matchElideBrByInvertingCond(MI)) 835 return false; 836 applyElideBrByInvertingCond(MI); 837 return true; 838 } 839 840 void CombinerHelper::applyElideBrByInvertingCond(MachineInstr &MI) { 841 MachineBasicBlock *BrTarget = MI.getOperand(0).getMBB(); 842 MachineBasicBlock::iterator BrIt(MI); 843 MachineInstr *BrCond = &*std::prev(BrIt); 844 MachineInstr *CmpMI = MRI.getVRegDef(BrCond->getOperand(0).getReg()); 845 846 CmpInst::Predicate InversePred = CmpInst::getInversePredicate( 847 (CmpInst::Predicate)CmpMI->getOperand(1).getPredicate()); 848 849 // Invert the G_ICMP condition. 850 Observer.changingInstr(*CmpMI); 851 CmpMI->getOperand(1).setPredicate(InversePred); 852 Observer.changedInstr(*CmpMI); 853 854 // Change the conditional branch target. 855 Observer.changingInstr(*BrCond); 856 BrCond->getOperand(1).setMBB(BrTarget); 857 Observer.changedInstr(*BrCond); 858 MI.eraseFromParent(); 859 } 860 861 static bool shouldLowerMemFuncForSize(const MachineFunction &MF) { 862 // On Darwin, -Os means optimize for size without hurting performance, so 863 // only really optimize for size when -Oz (MinSize) is used. 864 if (MF.getTarget().getTargetTriple().isOSDarwin()) 865 return MF.getFunction().hasMinSize(); 866 return MF.getFunction().hasOptSize(); 867 } 868 869 // Returns a list of types to use for memory op lowering in MemOps. A partial 870 // port of findOptimalMemOpLowering in TargetLowering. 871 static bool findGISelOptimalMemOpLowering(std::vector<LLT> &MemOps, 872 unsigned Limit, const MemOp &Op, 873 unsigned DstAS, unsigned SrcAS, 874 const AttributeList &FuncAttributes, 875 const TargetLowering &TLI) { 876 if (Op.isMemcpyWithFixedDstAlign() && Op.getSrcAlign() < Op.getDstAlign()) 877 return false; 878 879 LLT Ty = TLI.getOptimalMemOpLLT(Op, FuncAttributes); 880 881 if (Ty == LLT()) { 882 // Use the largest scalar type whose alignment constraints are satisfied. 883 // We only need to check DstAlign here as SrcAlign is always greater or 884 // equal to DstAlign (or zero). 885 Ty = LLT::scalar(64); 886 if (Op.isFixedDstAlign()) 887 while (Op.getDstAlign() < Ty.getSizeInBytes() && 888 !TLI.allowsMisalignedMemoryAccesses(Ty, DstAS, Op.getDstAlign())) 889 Ty = LLT::scalar(Ty.getSizeInBytes()); 890 assert(Ty.getSizeInBits() > 0 && "Could not find valid type"); 891 // FIXME: check for the largest legal type we can load/store to. 892 } 893 894 unsigned NumMemOps = 0; 895 uint64_t Size = Op.size(); 896 while (Size) { 897 unsigned TySize = Ty.getSizeInBytes(); 898 while (TySize > Size) { 899 // For now, only use non-vector load / store's for the left-over pieces. 900 LLT NewTy = Ty; 901 // FIXME: check for mem op safety and legality of the types. Not all of 902 // SDAGisms map cleanly to GISel concepts. 903 if (NewTy.isVector()) 904 NewTy = NewTy.getSizeInBits() > 64 ? LLT::scalar(64) : LLT::scalar(32); 905 NewTy = LLT::scalar(PowerOf2Floor(NewTy.getSizeInBits() - 1)); 906 unsigned NewTySize = NewTy.getSizeInBytes(); 907 assert(NewTySize > 0 && "Could not find appropriate type"); 908 909 // If the new LLT cannot cover all of the remaining bits, then consider 910 // issuing a (or a pair of) unaligned and overlapping load / store. 911 bool Fast; 912 // Need to get a VT equivalent for allowMisalignedMemoryAccesses(). 913 MVT VT = getMVTForLLT(Ty); 914 if (NumMemOps && Op.allowOverlap() && NewTySize < Size && 915 TLI.allowsMisalignedMemoryAccesses( 916 VT, DstAS, Op.isFixedDstAlign() ? Op.getDstAlign().value() : 0, 917 MachineMemOperand::MONone, &Fast) && 918 Fast) 919 TySize = Size; 920 else { 921 Ty = NewTy; 922 TySize = NewTySize; 923 } 924 } 925 926 if (++NumMemOps > Limit) 927 return false; 928 929 MemOps.push_back(Ty); 930 Size -= TySize; 931 } 932 933 return true; 934 } 935 936 static Type *getTypeForLLT(LLT Ty, LLVMContext &C) { 937 if (Ty.isVector()) 938 return FixedVectorType::get(IntegerType::get(C, Ty.getScalarSizeInBits()), 939 Ty.getNumElements()); 940 return IntegerType::get(C, Ty.getSizeInBits()); 941 } 942 943 // Get a vectorized representation of the memset value operand, GISel edition. 944 static Register getMemsetValue(Register Val, LLT Ty, MachineIRBuilder &MIB) { 945 MachineRegisterInfo &MRI = *MIB.getMRI(); 946 unsigned NumBits = Ty.getScalarSizeInBits(); 947 auto ValVRegAndVal = getConstantVRegValWithLookThrough(Val, MRI); 948 if (!Ty.isVector() && ValVRegAndVal) { 949 unsigned KnownVal = ValVRegAndVal->Value; 950 APInt Scalar = APInt(8, KnownVal); 951 APInt SplatVal = APInt::getSplat(NumBits, Scalar); 952 return MIB.buildConstant(Ty, SplatVal).getReg(0); 953 } 954 955 // Extend the byte value to the larger type, and then multiply by a magic 956 // value 0x010101... in order to replicate it across every byte. 957 // Unless it's zero, in which case just emit a larger G_CONSTANT 0. 958 if (ValVRegAndVal && ValVRegAndVal->Value == 0) { 959 return MIB.buildConstant(Ty, 0).getReg(0); 960 } 961 962 LLT ExtType = Ty.getScalarType(); 963 auto ZExt = MIB.buildZExtOrTrunc(ExtType, Val); 964 if (NumBits > 8) { 965 APInt Magic = APInt::getSplat(NumBits, APInt(8, 0x01)); 966 auto MagicMI = MIB.buildConstant(ExtType, Magic); 967 Val = MIB.buildMul(ExtType, ZExt, MagicMI).getReg(0); 968 } 969 970 // For vector types create a G_BUILD_VECTOR. 971 if (Ty.isVector()) 972 Val = MIB.buildSplatVector(Ty, Val).getReg(0); 973 974 return Val; 975 } 976 977 bool CombinerHelper::optimizeMemset(MachineInstr &MI, Register Dst, 978 Register Val, unsigned KnownLen, 979 Align Alignment, bool IsVolatile) { 980 auto &MF = *MI.getParent()->getParent(); 981 const auto &TLI = *MF.getSubtarget().getTargetLowering(); 982 auto &DL = MF.getDataLayout(); 983 LLVMContext &C = MF.getFunction().getContext(); 984 985 assert(KnownLen != 0 && "Have a zero length memset length!"); 986 987 bool DstAlignCanChange = false; 988 MachineFrameInfo &MFI = MF.getFrameInfo(); 989 bool OptSize = shouldLowerMemFuncForSize(MF); 990 991 MachineInstr *FIDef = getOpcodeDef(TargetOpcode::G_FRAME_INDEX, Dst, MRI); 992 if (FIDef && !MFI.isFixedObjectIndex(FIDef->getOperand(1).getIndex())) 993 DstAlignCanChange = true; 994 995 unsigned Limit = TLI.getMaxStoresPerMemset(OptSize); 996 std::vector<LLT> MemOps; 997 998 const auto &DstMMO = **MI.memoperands_begin(); 999 MachinePointerInfo DstPtrInfo = DstMMO.getPointerInfo(); 1000 1001 auto ValVRegAndVal = getConstantVRegValWithLookThrough(Val, MRI); 1002 bool IsZeroVal = ValVRegAndVal && ValVRegAndVal->Value == 0; 1003 1004 if (!findGISelOptimalMemOpLowering(MemOps, Limit, 1005 MemOp::Set(KnownLen, DstAlignCanChange, 1006 Alignment, 1007 /*IsZeroMemset=*/IsZeroVal, 1008 /*IsVolatile=*/IsVolatile), 1009 DstPtrInfo.getAddrSpace(), ~0u, 1010 MF.getFunction().getAttributes(), TLI)) 1011 return false; 1012 1013 if (DstAlignCanChange) { 1014 // Get an estimate of the type from the LLT. 1015 Type *IRTy = getTypeForLLT(MemOps[0], C); 1016 Align NewAlign = DL.getABITypeAlign(IRTy); 1017 if (NewAlign > Alignment) { 1018 Alignment = NewAlign; 1019 unsigned FI = FIDef->getOperand(1).getIndex(); 1020 // Give the stack frame object a larger alignment if needed. 1021 if (MFI.getObjectAlign(FI) < Alignment) 1022 MFI.setObjectAlignment(FI, Alignment); 1023 } 1024 } 1025 1026 MachineIRBuilder MIB(MI); 1027 // Find the largest store and generate the bit pattern for it. 1028 LLT LargestTy = MemOps[0]; 1029 for (unsigned i = 1; i < MemOps.size(); i++) 1030 if (MemOps[i].getSizeInBits() > LargestTy.getSizeInBits()) 1031 LargestTy = MemOps[i]; 1032 1033 // The memset stored value is always defined as an s8, so in order to make it 1034 // work with larger store types we need to repeat the bit pattern across the 1035 // wider type. 1036 Register MemSetValue = getMemsetValue(Val, LargestTy, MIB); 1037 1038 if (!MemSetValue) 1039 return false; 1040 1041 // Generate the stores. For each store type in the list, we generate the 1042 // matching store of that type to the destination address. 1043 LLT PtrTy = MRI.getType(Dst); 1044 unsigned DstOff = 0; 1045 unsigned Size = KnownLen; 1046 for (unsigned I = 0; I < MemOps.size(); I++) { 1047 LLT Ty = MemOps[I]; 1048 unsigned TySize = Ty.getSizeInBytes(); 1049 if (TySize > Size) { 1050 // Issuing an unaligned load / store pair that overlaps with the previous 1051 // pair. Adjust the offset accordingly. 1052 assert(I == MemOps.size() - 1 && I != 0); 1053 DstOff -= TySize - Size; 1054 } 1055 1056 // If this store is smaller than the largest store see whether we can get 1057 // the smaller value for free with a truncate. 1058 Register Value = MemSetValue; 1059 if (Ty.getSizeInBits() < LargestTy.getSizeInBits()) { 1060 MVT VT = getMVTForLLT(Ty); 1061 MVT LargestVT = getMVTForLLT(LargestTy); 1062 if (!LargestTy.isVector() && !Ty.isVector() && 1063 TLI.isTruncateFree(LargestVT, VT)) 1064 Value = MIB.buildTrunc(Ty, MemSetValue).getReg(0); 1065 else 1066 Value = getMemsetValue(Val, Ty, MIB); 1067 if (!Value) 1068 return false; 1069 } 1070 1071 auto *StoreMMO = 1072 MF.getMachineMemOperand(&DstMMO, DstOff, Ty.getSizeInBytes()); 1073 1074 Register Ptr = Dst; 1075 if (DstOff != 0) { 1076 auto Offset = 1077 MIB.buildConstant(LLT::scalar(PtrTy.getSizeInBits()), DstOff); 1078 Ptr = MIB.buildPtrAdd(PtrTy, Dst, Offset).getReg(0); 1079 } 1080 1081 MIB.buildStore(Value, Ptr, *StoreMMO); 1082 DstOff += Ty.getSizeInBytes(); 1083 Size -= TySize; 1084 } 1085 1086 MI.eraseFromParent(); 1087 return true; 1088 } 1089 1090 bool CombinerHelper::optimizeMemcpy(MachineInstr &MI, Register Dst, 1091 Register Src, unsigned KnownLen, 1092 Align DstAlign, Align SrcAlign, 1093 bool IsVolatile) { 1094 auto &MF = *MI.getParent()->getParent(); 1095 const auto &TLI = *MF.getSubtarget().getTargetLowering(); 1096 auto &DL = MF.getDataLayout(); 1097 LLVMContext &C = MF.getFunction().getContext(); 1098 1099 assert(KnownLen != 0 && "Have a zero length memcpy length!"); 1100 1101 bool DstAlignCanChange = false; 1102 MachineFrameInfo &MFI = MF.getFrameInfo(); 1103 bool OptSize = shouldLowerMemFuncForSize(MF); 1104 Align Alignment = commonAlignment(DstAlign, SrcAlign); 1105 1106 MachineInstr *FIDef = getOpcodeDef(TargetOpcode::G_FRAME_INDEX, Dst, MRI); 1107 if (FIDef && !MFI.isFixedObjectIndex(FIDef->getOperand(1).getIndex())) 1108 DstAlignCanChange = true; 1109 1110 // FIXME: infer better src pointer alignment like SelectionDAG does here. 1111 // FIXME: also use the equivalent of isMemSrcFromConstant and alwaysinlining 1112 // if the memcpy is in a tail call position. 1113 1114 unsigned Limit = TLI.getMaxStoresPerMemcpy(OptSize); 1115 std::vector<LLT> MemOps; 1116 1117 const auto &DstMMO = **MI.memoperands_begin(); 1118 const auto &SrcMMO = **std::next(MI.memoperands_begin()); 1119 MachinePointerInfo DstPtrInfo = DstMMO.getPointerInfo(); 1120 MachinePointerInfo SrcPtrInfo = SrcMMO.getPointerInfo(); 1121 1122 if (!findGISelOptimalMemOpLowering( 1123 MemOps, Limit, 1124 MemOp::Copy(KnownLen, DstAlignCanChange, Alignment, SrcAlign, 1125 IsVolatile), 1126 DstPtrInfo.getAddrSpace(), SrcPtrInfo.getAddrSpace(), 1127 MF.getFunction().getAttributes(), TLI)) 1128 return false; 1129 1130 if (DstAlignCanChange) { 1131 // Get an estimate of the type from the LLT. 1132 Type *IRTy = getTypeForLLT(MemOps[0], C); 1133 Align NewAlign = DL.getABITypeAlign(IRTy); 1134 1135 // Don't promote to an alignment that would require dynamic stack 1136 // realignment. 1137 const TargetRegisterInfo *TRI = MF.getSubtarget().getRegisterInfo(); 1138 if (!TRI->needsStackRealignment(MF)) 1139 while (NewAlign > Alignment && DL.exceedsNaturalStackAlignment(NewAlign)) 1140 NewAlign = NewAlign / 2; 1141 1142 if (NewAlign > Alignment) { 1143 Alignment = NewAlign; 1144 unsigned FI = FIDef->getOperand(1).getIndex(); 1145 // Give the stack frame object a larger alignment if needed. 1146 if (MFI.getObjectAlign(FI) < Alignment) 1147 MFI.setObjectAlignment(FI, Alignment); 1148 } 1149 } 1150 1151 LLVM_DEBUG(dbgs() << "Inlining memcpy: " << MI << " into loads & stores\n"); 1152 1153 MachineIRBuilder MIB(MI); 1154 // Now we need to emit a pair of load and stores for each of the types we've 1155 // collected. I.e. for each type, generate a load from the source pointer of 1156 // that type width, and then generate a corresponding store to the dest buffer 1157 // of that value loaded. This can result in a sequence of loads and stores 1158 // mixed types, depending on what the target specifies as good types to use. 1159 unsigned CurrOffset = 0; 1160 LLT PtrTy = MRI.getType(Src); 1161 unsigned Size = KnownLen; 1162 for (auto CopyTy : MemOps) { 1163 // Issuing an unaligned load / store pair that overlaps with the previous 1164 // pair. Adjust the offset accordingly. 1165 if (CopyTy.getSizeInBytes() > Size) 1166 CurrOffset -= CopyTy.getSizeInBytes() - Size; 1167 1168 // Construct MMOs for the accesses. 1169 auto *LoadMMO = 1170 MF.getMachineMemOperand(&SrcMMO, CurrOffset, CopyTy.getSizeInBytes()); 1171 auto *StoreMMO = 1172 MF.getMachineMemOperand(&DstMMO, CurrOffset, CopyTy.getSizeInBytes()); 1173 1174 // Create the load. 1175 Register LoadPtr = Src; 1176 Register Offset; 1177 if (CurrOffset != 0) { 1178 Offset = MIB.buildConstant(LLT::scalar(PtrTy.getSizeInBits()), CurrOffset) 1179 .getReg(0); 1180 LoadPtr = MIB.buildPtrAdd(PtrTy, Src, Offset).getReg(0); 1181 } 1182 auto LdVal = MIB.buildLoad(CopyTy, LoadPtr, *LoadMMO); 1183 1184 // Create the store. 1185 Register StorePtr = 1186 CurrOffset == 0 ? Dst : MIB.buildPtrAdd(PtrTy, Dst, Offset).getReg(0); 1187 MIB.buildStore(LdVal, StorePtr, *StoreMMO); 1188 CurrOffset += CopyTy.getSizeInBytes(); 1189 Size -= CopyTy.getSizeInBytes(); 1190 } 1191 1192 MI.eraseFromParent(); 1193 return true; 1194 } 1195 1196 bool CombinerHelper::optimizeMemmove(MachineInstr &MI, Register Dst, 1197 Register Src, unsigned KnownLen, 1198 Align DstAlign, Align SrcAlign, 1199 bool IsVolatile) { 1200 auto &MF = *MI.getParent()->getParent(); 1201 const auto &TLI = *MF.getSubtarget().getTargetLowering(); 1202 auto &DL = MF.getDataLayout(); 1203 LLVMContext &C = MF.getFunction().getContext(); 1204 1205 assert(KnownLen != 0 && "Have a zero length memmove length!"); 1206 1207 bool DstAlignCanChange = false; 1208 MachineFrameInfo &MFI = MF.getFrameInfo(); 1209 bool OptSize = shouldLowerMemFuncForSize(MF); 1210 Align Alignment = commonAlignment(DstAlign, SrcAlign); 1211 1212 MachineInstr *FIDef = getOpcodeDef(TargetOpcode::G_FRAME_INDEX, Dst, MRI); 1213 if (FIDef && !MFI.isFixedObjectIndex(FIDef->getOperand(1).getIndex())) 1214 DstAlignCanChange = true; 1215 1216 unsigned Limit = TLI.getMaxStoresPerMemmove(OptSize); 1217 std::vector<LLT> MemOps; 1218 1219 const auto &DstMMO = **MI.memoperands_begin(); 1220 const auto &SrcMMO = **std::next(MI.memoperands_begin()); 1221 MachinePointerInfo DstPtrInfo = DstMMO.getPointerInfo(); 1222 MachinePointerInfo SrcPtrInfo = SrcMMO.getPointerInfo(); 1223 1224 // FIXME: SelectionDAG always passes false for 'AllowOverlap', apparently due 1225 // to a bug in it's findOptimalMemOpLowering implementation. For now do the 1226 // same thing here. 1227 if (!findGISelOptimalMemOpLowering( 1228 MemOps, Limit, 1229 MemOp::Copy(KnownLen, DstAlignCanChange, Alignment, SrcAlign, 1230 /*IsVolatile*/ true), 1231 DstPtrInfo.getAddrSpace(), SrcPtrInfo.getAddrSpace(), 1232 MF.getFunction().getAttributes(), TLI)) 1233 return false; 1234 1235 if (DstAlignCanChange) { 1236 // Get an estimate of the type from the LLT. 1237 Type *IRTy = getTypeForLLT(MemOps[0], C); 1238 Align NewAlign = DL.getABITypeAlign(IRTy); 1239 1240 // Don't promote to an alignment that would require dynamic stack 1241 // realignment. 1242 const TargetRegisterInfo *TRI = MF.getSubtarget().getRegisterInfo(); 1243 if (!TRI->needsStackRealignment(MF)) 1244 while (NewAlign > Alignment && DL.exceedsNaturalStackAlignment(NewAlign)) 1245 NewAlign = NewAlign / 2; 1246 1247 if (NewAlign > Alignment) { 1248 Alignment = NewAlign; 1249 unsigned FI = FIDef->getOperand(1).getIndex(); 1250 // Give the stack frame object a larger alignment if needed. 1251 if (MFI.getObjectAlign(FI) < Alignment) 1252 MFI.setObjectAlignment(FI, Alignment); 1253 } 1254 } 1255 1256 LLVM_DEBUG(dbgs() << "Inlining memmove: " << MI << " into loads & stores\n"); 1257 1258 MachineIRBuilder MIB(MI); 1259 // Memmove requires that we perform the loads first before issuing the stores. 1260 // Apart from that, this loop is pretty much doing the same thing as the 1261 // memcpy codegen function. 1262 unsigned CurrOffset = 0; 1263 LLT PtrTy = MRI.getType(Src); 1264 SmallVector<Register, 16> LoadVals; 1265 for (auto CopyTy : MemOps) { 1266 // Construct MMO for the load. 1267 auto *LoadMMO = 1268 MF.getMachineMemOperand(&SrcMMO, CurrOffset, CopyTy.getSizeInBytes()); 1269 1270 // Create the load. 1271 Register LoadPtr = Src; 1272 if (CurrOffset != 0) { 1273 auto Offset = 1274 MIB.buildConstant(LLT::scalar(PtrTy.getSizeInBits()), CurrOffset); 1275 LoadPtr = MIB.buildPtrAdd(PtrTy, Src, Offset).getReg(0); 1276 } 1277 LoadVals.push_back(MIB.buildLoad(CopyTy, LoadPtr, *LoadMMO).getReg(0)); 1278 CurrOffset += CopyTy.getSizeInBytes(); 1279 } 1280 1281 CurrOffset = 0; 1282 for (unsigned I = 0; I < MemOps.size(); ++I) { 1283 LLT CopyTy = MemOps[I]; 1284 // Now store the values loaded. 1285 auto *StoreMMO = 1286 MF.getMachineMemOperand(&DstMMO, CurrOffset, CopyTy.getSizeInBytes()); 1287 1288 Register StorePtr = Dst; 1289 if (CurrOffset != 0) { 1290 auto Offset = 1291 MIB.buildConstant(LLT::scalar(PtrTy.getSizeInBits()), CurrOffset); 1292 StorePtr = MIB.buildPtrAdd(PtrTy, Dst, Offset).getReg(0); 1293 } 1294 MIB.buildStore(LoadVals[I], StorePtr, *StoreMMO); 1295 CurrOffset += CopyTy.getSizeInBytes(); 1296 } 1297 MI.eraseFromParent(); 1298 return true; 1299 } 1300 1301 bool CombinerHelper::tryCombineMemCpyFamily(MachineInstr &MI, unsigned MaxLen) { 1302 // This combine is fairly complex so it's not written with a separate 1303 // matcher function. 1304 assert(MI.getOpcode() == TargetOpcode::G_INTRINSIC_W_SIDE_EFFECTS); 1305 Intrinsic::ID ID = (Intrinsic::ID)MI.getIntrinsicID(); 1306 assert((ID == Intrinsic::memcpy || ID == Intrinsic::memmove || 1307 ID == Intrinsic::memset) && 1308 "Expected a memcpy like intrinsic"); 1309 1310 auto MMOIt = MI.memoperands_begin(); 1311 const MachineMemOperand *MemOp = *MMOIt; 1312 bool IsVolatile = MemOp->isVolatile(); 1313 // Don't try to optimize volatile. 1314 if (IsVolatile) 1315 return false; 1316 1317 Align DstAlign = MemOp->getBaseAlign(); 1318 Align SrcAlign; 1319 Register Dst = MI.getOperand(1).getReg(); 1320 Register Src = MI.getOperand(2).getReg(); 1321 Register Len = MI.getOperand(3).getReg(); 1322 1323 if (ID != Intrinsic::memset) { 1324 assert(MMOIt != MI.memoperands_end() && "Expected a second MMO on MI"); 1325 MemOp = *(++MMOIt); 1326 SrcAlign = MemOp->getBaseAlign(); 1327 } 1328 1329 // See if this is a constant length copy 1330 auto LenVRegAndVal = getConstantVRegValWithLookThrough(Len, MRI); 1331 if (!LenVRegAndVal) 1332 return false; // Leave it to the legalizer to lower it to a libcall. 1333 unsigned KnownLen = LenVRegAndVal->Value; 1334 1335 if (KnownLen == 0) { 1336 MI.eraseFromParent(); 1337 return true; 1338 } 1339 1340 if (MaxLen && KnownLen > MaxLen) 1341 return false; 1342 1343 if (ID == Intrinsic::memcpy) 1344 return optimizeMemcpy(MI, Dst, Src, KnownLen, DstAlign, SrcAlign, IsVolatile); 1345 if (ID == Intrinsic::memmove) 1346 return optimizeMemmove(MI, Dst, Src, KnownLen, DstAlign, SrcAlign, IsVolatile); 1347 if (ID == Intrinsic::memset) 1348 return optimizeMemset(MI, Dst, Src, KnownLen, DstAlign, IsVolatile); 1349 return false; 1350 } 1351 1352 bool CombinerHelper::matchPtrAddImmedChain(MachineInstr &MI, 1353 PtrAddChain &MatchInfo) { 1354 // We're trying to match the following pattern: 1355 // %t1 = G_PTR_ADD %base, G_CONSTANT imm1 1356 // %root = G_PTR_ADD %t1, G_CONSTANT imm2 1357 // --> 1358 // %root = G_PTR_ADD %base, G_CONSTANT (imm1 + imm2) 1359 1360 if (MI.getOpcode() != TargetOpcode::G_PTR_ADD) 1361 return false; 1362 1363 Register Add2 = MI.getOperand(1).getReg(); 1364 Register Imm1 = MI.getOperand(2).getReg(); 1365 auto MaybeImmVal = getConstantVRegValWithLookThrough(Imm1, MRI); 1366 if (!MaybeImmVal) 1367 return false; 1368 1369 MachineInstr *Add2Def = MRI.getUniqueVRegDef(Add2); 1370 if (!Add2Def || Add2Def->getOpcode() != TargetOpcode::G_PTR_ADD) 1371 return false; 1372 1373 Register Base = Add2Def->getOperand(1).getReg(); 1374 Register Imm2 = Add2Def->getOperand(2).getReg(); 1375 auto MaybeImm2Val = getConstantVRegValWithLookThrough(Imm2, MRI); 1376 if (!MaybeImm2Val) 1377 return false; 1378 1379 // Pass the combined immediate to the apply function. 1380 MatchInfo.Imm = MaybeImmVal->Value + MaybeImm2Val->Value; 1381 MatchInfo.Base = Base; 1382 return true; 1383 } 1384 1385 bool CombinerHelper::applyPtrAddImmedChain(MachineInstr &MI, 1386 PtrAddChain &MatchInfo) { 1387 assert(MI.getOpcode() == TargetOpcode::G_PTR_ADD && "Expected G_PTR_ADD"); 1388 MachineIRBuilder MIB(MI); 1389 LLT OffsetTy = MRI.getType(MI.getOperand(2).getReg()); 1390 auto NewOffset = MIB.buildConstant(OffsetTy, MatchInfo.Imm); 1391 Observer.changingInstr(MI); 1392 MI.getOperand(1).setReg(MatchInfo.Base); 1393 MI.getOperand(2).setReg(NewOffset.getReg(0)); 1394 Observer.changedInstr(MI); 1395 return true; 1396 } 1397 1398 bool CombinerHelper::matchCombineMulToShl(MachineInstr &MI, 1399 unsigned &ShiftVal) { 1400 assert(MI.getOpcode() == TargetOpcode::G_MUL && "Expected a G_MUL"); 1401 auto MaybeImmVal = 1402 getConstantVRegValWithLookThrough(MI.getOperand(2).getReg(), MRI); 1403 if (!MaybeImmVal || !isPowerOf2_64(MaybeImmVal->Value)) 1404 return false; 1405 ShiftVal = Log2_64(MaybeImmVal->Value); 1406 return true; 1407 } 1408 1409 bool CombinerHelper::applyCombineMulToShl(MachineInstr &MI, 1410 unsigned &ShiftVal) { 1411 assert(MI.getOpcode() == TargetOpcode::G_MUL && "Expected a G_MUL"); 1412 MachineIRBuilder MIB(MI); 1413 LLT ShiftTy = MRI.getType(MI.getOperand(0).getReg()); 1414 auto ShiftCst = MIB.buildConstant(ShiftTy, ShiftVal); 1415 Observer.changingInstr(MI); 1416 MI.setDesc(MIB.getTII().get(TargetOpcode::G_SHL)); 1417 MI.getOperand(2).setReg(ShiftCst.getReg(0)); 1418 Observer.changedInstr(MI); 1419 return true; 1420 } 1421 1422 bool CombinerHelper::matchCombineShiftToUnmerge(MachineInstr &MI, 1423 unsigned TargetShiftSize, 1424 unsigned &ShiftVal) { 1425 assert((MI.getOpcode() == TargetOpcode::G_SHL || 1426 MI.getOpcode() == TargetOpcode::G_LSHR || 1427 MI.getOpcode() == TargetOpcode::G_ASHR) && "Expected a shift"); 1428 1429 LLT Ty = MRI.getType(MI.getOperand(0).getReg()); 1430 if (Ty.isVector()) // TODO: 1431 return false; 1432 1433 // Don't narrow further than the requested size. 1434 unsigned Size = Ty.getSizeInBits(); 1435 if (Size <= TargetShiftSize) 1436 return false; 1437 1438 auto MaybeImmVal = 1439 getConstantVRegValWithLookThrough(MI.getOperand(2).getReg(), MRI); 1440 if (!MaybeImmVal) 1441 return false; 1442 1443 ShiftVal = MaybeImmVal->Value; 1444 return ShiftVal >= Size / 2 && ShiftVal < Size; 1445 } 1446 1447 bool CombinerHelper::applyCombineShiftToUnmerge(MachineInstr &MI, 1448 const unsigned &ShiftVal) { 1449 Register DstReg = MI.getOperand(0).getReg(); 1450 Register SrcReg = MI.getOperand(1).getReg(); 1451 LLT Ty = MRI.getType(SrcReg); 1452 unsigned Size = Ty.getSizeInBits(); 1453 unsigned HalfSize = Size / 2; 1454 assert(ShiftVal >= HalfSize); 1455 1456 LLT HalfTy = LLT::scalar(HalfSize); 1457 1458 Builder.setInstr(MI); 1459 auto Unmerge = Builder.buildUnmerge(HalfTy, SrcReg); 1460 unsigned NarrowShiftAmt = ShiftVal - HalfSize; 1461 1462 if (MI.getOpcode() == TargetOpcode::G_LSHR) { 1463 Register Narrowed = Unmerge.getReg(1); 1464 1465 // dst = G_LSHR s64:x, C for C >= 32 1466 // => 1467 // lo, hi = G_UNMERGE_VALUES x 1468 // dst = G_MERGE_VALUES (G_LSHR hi, C - 32), 0 1469 1470 if (NarrowShiftAmt != 0) { 1471 Narrowed = Builder.buildLShr(HalfTy, Narrowed, 1472 Builder.buildConstant(HalfTy, NarrowShiftAmt)).getReg(0); 1473 } 1474 1475 auto Zero = Builder.buildConstant(HalfTy, 0); 1476 Builder.buildMerge(DstReg, { Narrowed, Zero }); 1477 } else if (MI.getOpcode() == TargetOpcode::G_SHL) { 1478 Register Narrowed = Unmerge.getReg(0); 1479 // dst = G_SHL s64:x, C for C >= 32 1480 // => 1481 // lo, hi = G_UNMERGE_VALUES x 1482 // dst = G_MERGE_VALUES 0, (G_SHL hi, C - 32) 1483 if (NarrowShiftAmt != 0) { 1484 Narrowed = Builder.buildShl(HalfTy, Narrowed, 1485 Builder.buildConstant(HalfTy, NarrowShiftAmt)).getReg(0); 1486 } 1487 1488 auto Zero = Builder.buildConstant(HalfTy, 0); 1489 Builder.buildMerge(DstReg, { Zero, Narrowed }); 1490 } else { 1491 assert(MI.getOpcode() == TargetOpcode::G_ASHR); 1492 auto Hi = Builder.buildAShr( 1493 HalfTy, Unmerge.getReg(1), 1494 Builder.buildConstant(HalfTy, HalfSize - 1)); 1495 1496 if (ShiftVal == HalfSize) { 1497 // (G_ASHR i64:x, 32) -> 1498 // G_MERGE_VALUES hi_32(x), (G_ASHR hi_32(x), 31) 1499 Builder.buildMerge(DstReg, { Unmerge.getReg(1), Hi }); 1500 } else if (ShiftVal == Size - 1) { 1501 // Don't need a second shift. 1502 // (G_ASHR i64:x, 63) -> 1503 // %narrowed = (G_ASHR hi_32(x), 31) 1504 // G_MERGE_VALUES %narrowed, %narrowed 1505 Builder.buildMerge(DstReg, { Hi, Hi }); 1506 } else { 1507 auto Lo = Builder.buildAShr( 1508 HalfTy, Unmerge.getReg(1), 1509 Builder.buildConstant(HalfTy, ShiftVal - HalfSize)); 1510 1511 // (G_ASHR i64:x, C) ->, for C >= 32 1512 // G_MERGE_VALUES (G_ASHR hi_32(x), C - 32), (G_ASHR hi_32(x), 31) 1513 Builder.buildMerge(DstReg, { Lo, Hi }); 1514 } 1515 } 1516 1517 MI.eraseFromParent(); 1518 return true; 1519 } 1520 1521 bool CombinerHelper::tryCombineShiftToUnmerge(MachineInstr &MI, 1522 unsigned TargetShiftAmount) { 1523 unsigned ShiftAmt; 1524 if (matchCombineShiftToUnmerge(MI, TargetShiftAmount, ShiftAmt)) { 1525 applyCombineShiftToUnmerge(MI, ShiftAmt); 1526 return true; 1527 } 1528 1529 return false; 1530 } 1531 1532 bool CombinerHelper::matchAnyExplicitUseIsUndef(MachineInstr &MI) { 1533 return any_of(MI.explicit_uses(), [this](const MachineOperand &MO) { 1534 return MO.isReg() && 1535 getOpcodeDef(TargetOpcode::G_IMPLICIT_DEF, MO.getReg(), MRI); 1536 }); 1537 } 1538 1539 bool CombinerHelper::matchAllExplicitUsesAreUndef(MachineInstr &MI) { 1540 return all_of(MI.explicit_uses(), [this](const MachineOperand &MO) { 1541 return !MO.isReg() || 1542 getOpcodeDef(TargetOpcode::G_IMPLICIT_DEF, MO.getReg(), MRI); 1543 }); 1544 } 1545 1546 bool CombinerHelper::matchUndefShuffleVectorMask(MachineInstr &MI) { 1547 assert(MI.getOpcode() == TargetOpcode::G_SHUFFLE_VECTOR); 1548 ArrayRef<int> Mask = MI.getOperand(3).getShuffleMask(); 1549 return all_of(Mask, [](int Elt) { return Elt < 0; }); 1550 } 1551 1552 bool CombinerHelper::matchUndefStore(MachineInstr &MI) { 1553 assert(MI.getOpcode() == TargetOpcode::G_STORE); 1554 return getOpcodeDef(TargetOpcode::G_IMPLICIT_DEF, MI.getOperand(0).getReg(), 1555 MRI); 1556 } 1557 1558 bool CombinerHelper::eraseInst(MachineInstr &MI) { 1559 MI.eraseFromParent(); 1560 return true; 1561 } 1562 1563 bool CombinerHelper::matchEqualDefs(const MachineOperand &MOP1, 1564 const MachineOperand &MOP2) { 1565 if (!MOP1.isReg() || !MOP2.isReg()) 1566 return false; 1567 MachineInstr *I1 = getDefIgnoringCopies(MOP1.getReg(), MRI); 1568 if (!I1) 1569 return false; 1570 MachineInstr *I2 = getDefIgnoringCopies(MOP2.getReg(), MRI); 1571 if (!I2) 1572 return false; 1573 1574 // Handle a case like this: 1575 // 1576 // %0:_(s64), %1:_(s64) = G_UNMERGE_VALUES %2:_(<2 x s64>) 1577 // 1578 // Even though %0 and %1 are produced by the same instruction they are not 1579 // the same values. 1580 if (I1 == I2) 1581 return MOP1.getReg() == MOP2.getReg(); 1582 1583 // If we have an instruction which loads or stores, we can't guarantee that 1584 // it is identical. 1585 // 1586 // For example, we may have 1587 // 1588 // %x1 = G_LOAD %addr (load N from @somewhere) 1589 // ... 1590 // call @foo 1591 // ... 1592 // %x2 = G_LOAD %addr (load N from @somewhere) 1593 // ... 1594 // %or = G_OR %x1, %x2 1595 // 1596 // It's possible that @foo will modify whatever lives at the address we're 1597 // loading from. To be safe, let's just assume that all loads and stores 1598 // are different (unless we have something which is guaranteed to not 1599 // change.) 1600 if (I1->mayLoadOrStore() && !I1->isDereferenceableInvariantLoad(nullptr)) 1601 return false; 1602 1603 // Check for physical registers on the instructions first to avoid cases 1604 // like this: 1605 // 1606 // %a = COPY $physreg 1607 // ... 1608 // SOMETHING implicit-def $physreg 1609 // ... 1610 // %b = COPY $physreg 1611 // 1612 // These copies are not equivalent. 1613 if (any_of(I1->uses(), [](const MachineOperand &MO) { 1614 return MO.isReg() && MO.getReg().isPhysical(); 1615 })) { 1616 // Check if we have a case like this: 1617 // 1618 // %a = COPY $physreg 1619 // %b = COPY %a 1620 // 1621 // In this case, I1 and I2 will both be equal to %a = COPY $physreg. 1622 // From that, we know that they must have the same value, since they must 1623 // have come from the same COPY. 1624 return I1->isIdenticalTo(*I2); 1625 } 1626 1627 // We don't have any physical registers, so we don't necessarily need the 1628 // same vreg defs. 1629 // 1630 // On the off-chance that there's some target instruction feeding into the 1631 // instruction, let's use produceSameValue instead of isIdenticalTo. 1632 return Builder.getTII().produceSameValue(*I1, *I2, &MRI); 1633 } 1634 1635 bool CombinerHelper::matchConstantOp(const MachineOperand &MOP, int64_t C) { 1636 if (!MOP.isReg()) 1637 return false; 1638 // MIPatternMatch doesn't let us look through G_ZEXT etc. 1639 auto ValAndVReg = getConstantVRegValWithLookThrough(MOP.getReg(), MRI); 1640 return ValAndVReg && ValAndVReg->Value == C; 1641 } 1642 1643 bool CombinerHelper::replaceSingleDefInstWithOperand(MachineInstr &MI, 1644 unsigned OpIdx) { 1645 assert(MI.getNumExplicitDefs() == 1 && "Expected one explicit def?"); 1646 Register OldReg = MI.getOperand(0).getReg(); 1647 Register Replacement = MI.getOperand(OpIdx).getReg(); 1648 assert(canReplaceReg(OldReg, Replacement, MRI) && "Cannot replace register?"); 1649 MI.eraseFromParent(); 1650 replaceRegWith(MRI, OldReg, Replacement); 1651 return true; 1652 } 1653 1654 bool CombinerHelper::matchSelectSameVal(MachineInstr &MI) { 1655 assert(MI.getOpcode() == TargetOpcode::G_SELECT); 1656 // Match (cond ? x : x) 1657 return matchEqualDefs(MI.getOperand(2), MI.getOperand(3)) && 1658 canReplaceReg(MI.getOperand(0).getReg(), MI.getOperand(2).getReg(), 1659 MRI); 1660 } 1661 1662 bool CombinerHelper::matchBinOpSameVal(MachineInstr &MI) { 1663 return matchEqualDefs(MI.getOperand(1), MI.getOperand(2)) && 1664 canReplaceReg(MI.getOperand(0).getReg(), MI.getOperand(1).getReg(), 1665 MRI); 1666 } 1667 1668 bool CombinerHelper::matchOperandIsZero(MachineInstr &MI, unsigned OpIdx) { 1669 return matchConstantOp(MI.getOperand(OpIdx), 0) && 1670 canReplaceReg(MI.getOperand(0).getReg(), MI.getOperand(OpIdx).getReg(), 1671 MRI); 1672 } 1673 1674 bool CombinerHelper::replaceInstWithFConstant(MachineInstr &MI, double C) { 1675 assert(MI.getNumDefs() == 1 && "Expected only one def?"); 1676 Builder.setInstr(MI); 1677 Builder.buildFConstant(MI.getOperand(0), C); 1678 MI.eraseFromParent(); 1679 return true; 1680 } 1681 1682 bool CombinerHelper::replaceInstWithConstant(MachineInstr &MI, int64_t C) { 1683 assert(MI.getNumDefs() == 1 && "Expected only one def?"); 1684 Builder.setInstr(MI); 1685 Builder.buildConstant(MI.getOperand(0), C); 1686 MI.eraseFromParent(); 1687 return true; 1688 } 1689 1690 bool CombinerHelper::replaceInstWithUndef(MachineInstr &MI) { 1691 assert(MI.getNumDefs() == 1 && "Expected only one def?"); 1692 Builder.setInstr(MI); 1693 Builder.buildUndef(MI.getOperand(0)); 1694 MI.eraseFromParent(); 1695 return true; 1696 } 1697 1698 bool CombinerHelper::matchSimplifyAddToSub( 1699 MachineInstr &MI, std::tuple<Register, Register> &MatchInfo) { 1700 Register LHS = MI.getOperand(1).getReg(); 1701 Register RHS = MI.getOperand(2).getReg(); 1702 Register &NewLHS = std::get<0>(MatchInfo); 1703 Register &NewRHS = std::get<1>(MatchInfo); 1704 1705 // Helper lambda to check for opportunities for 1706 // ((0-A) + B) -> B - A 1707 // (A + (0-B)) -> A - B 1708 auto CheckFold = [&](Register &MaybeSub, Register &MaybeNewLHS) { 1709 int64_t Cst; 1710 if (!mi_match(MaybeSub, MRI, m_GSub(m_ICst(Cst), m_Reg(NewRHS))) || 1711 Cst != 0) 1712 return false; 1713 NewLHS = MaybeNewLHS; 1714 return true; 1715 }; 1716 1717 return CheckFold(LHS, RHS) || CheckFold(RHS, LHS); 1718 } 1719 1720 bool CombinerHelper::applySimplifyAddToSub( 1721 MachineInstr &MI, std::tuple<Register, Register> &MatchInfo) { 1722 Builder.setInstr(MI); 1723 Register SubLHS, SubRHS; 1724 std::tie(SubLHS, SubRHS) = MatchInfo; 1725 Builder.buildSub(MI.getOperand(0).getReg(), SubLHS, SubRHS); 1726 MI.eraseFromParent(); 1727 return true; 1728 } 1729 1730 bool CombinerHelper::tryCombine(MachineInstr &MI) { 1731 if (tryCombineCopy(MI)) 1732 return true; 1733 if (tryCombineExtendingLoads(MI)) 1734 return true; 1735 if (tryCombineIndexedLoadStore(MI)) 1736 return true; 1737 return false; 1738 } 1739