1 //===- LoopStrengthReduce.cpp - Strength Reduce IVs in Loops --------------===// 2 // 3 // The LLVM Compiler Infrastructure 4 // 5 // This file is distributed under the University of Illinois Open Source 6 // License. See LICENSE.TXT for details. 7 // 8 //===----------------------------------------------------------------------===// 9 // 10 // This transformation analyzes and transforms the induction variables (and 11 // computations derived from them) into forms suitable for efficient execution 12 // on the target. 13 // 14 // This pass performs a strength reduction on array references inside loops that 15 // have as one or more of their components the loop induction variable, it 16 // rewrites expressions to take advantage of scaled-index addressing modes 17 // available on the target, and it performs a variety of other optimizations 18 // related to loop induction variables. 19 // 20 // Terminology note: this code has a lot of handling for "post-increment" or 21 // "post-inc" users. This is not talking about post-increment addressing modes; 22 // it is instead talking about code like this: 23 // 24 // %i = phi [ 0, %entry ], [ %i.next, %latch ] 25 // ... 26 // %i.next = add %i, 1 27 // %c = icmp eq %i.next, %n 28 // 29 // The SCEV for %i is {0,+,1}<%L>. The SCEV for %i.next is {1,+,1}<%L>, however 30 // it's useful to think about these as the same register, with some uses using 31 // the value of the register before the add and some using it after. In this 32 // example, the icmp is a post-increment user, since it uses %i.next, which is 33 // the value of the induction variable after the increment. The other common 34 // case of post-increment users is users outside the loop. 35 // 36 // TODO: More sophistication in the way Formulae are generated and filtered. 37 // 38 // TODO: Handle multiple loops at a time. 39 // 40 // TODO: Should the addressing mode BaseGV be changed to a ConstantExpr instead 41 // of a GlobalValue? 42 // 43 // TODO: When truncation is free, truncate ICmp users' operands to make it a 44 // smaller encoding (on x86 at least). 45 // 46 // TODO: When a negated register is used by an add (such as in a list of 47 // multiple base registers, or as the increment expression in an addrec), 48 // we may not actually need both reg and (-1 * reg) in registers; the 49 // negation can be implemented by using a sub instead of an add. The 50 // lack of support for taking this into consideration when making 51 // register pressure decisions is partly worked around by the "Special" 52 // use kind. 53 // 54 //===----------------------------------------------------------------------===// 55 56 #include "llvm/Transforms/Scalar.h" 57 #include "llvm/ADT/DenseSet.h" 58 #include "llvm/ADT/Hashing.h" 59 #include "llvm/ADT/STLExtras.h" 60 #include "llvm/ADT/SetVector.h" 61 #include "llvm/ADT/SmallBitVector.h" 62 #include "llvm/Analysis/IVUsers.h" 63 #include "llvm/Analysis/LoopPass.h" 64 #include "llvm/Analysis/ScalarEvolutionExpander.h" 65 #include "llvm/Analysis/TargetTransformInfo.h" 66 #include "llvm/IR/Constants.h" 67 #include "llvm/IR/DerivedTypes.h" 68 #include "llvm/IR/Dominators.h" 69 #include "llvm/IR/Instructions.h" 70 #include "llvm/IR/IntrinsicInst.h" 71 #include "llvm/IR/Module.h" 72 #include "llvm/IR/ValueHandle.h" 73 #include "llvm/Support/CommandLine.h" 74 #include "llvm/Support/Debug.h" 75 #include "llvm/Support/raw_ostream.h" 76 #include "llvm/Transforms/Utils/BasicBlockUtils.h" 77 #include "llvm/Transforms/Utils/Local.h" 78 #include <algorithm> 79 using namespace llvm; 80 81 #define DEBUG_TYPE "loop-reduce" 82 83 /// MaxIVUsers is an arbitrary threshold that provides an early opportunitiy for 84 /// bail out. This threshold is far beyond the number of users that LSR can 85 /// conceivably solve, so it should not affect generated code, but catches the 86 /// worst cases before LSR burns too much compile time and stack space. 87 static const unsigned MaxIVUsers = 200; 88 89 // Temporary flag to cleanup congruent phis after LSR phi expansion. 90 // It's currently disabled until we can determine whether it's truly useful or 91 // not. The flag should be removed after the v3.0 release. 92 // This is now needed for ivchains. 93 static cl::opt<bool> EnablePhiElim( 94 "enable-lsr-phielim", cl::Hidden, cl::init(true), 95 cl::desc("Enable LSR phi elimination")); 96 97 #ifndef NDEBUG 98 // Stress test IV chain generation. 99 static cl::opt<bool> StressIVChain( 100 "stress-ivchain", cl::Hidden, cl::init(false), 101 cl::desc("Stress test LSR IV chains")); 102 #else 103 static bool StressIVChain = false; 104 #endif 105 106 namespace { 107 108 struct MemAccessTy { 109 /// Used in situations where the accessed memory type is unknown. 110 static const unsigned UnknownAddressSpace = ~0u; 111 112 Type *MemTy; 113 unsigned AddrSpace; 114 115 MemAccessTy() : MemTy(nullptr), AddrSpace(UnknownAddressSpace) {} 116 117 MemAccessTy(Type *Ty, unsigned AS) : 118 MemTy(Ty), AddrSpace(AS) {} 119 120 bool operator==(MemAccessTy Other) const { 121 return MemTy == Other.MemTy && AddrSpace == Other.AddrSpace; 122 } 123 124 bool operator!=(MemAccessTy Other) const { return !(*this == Other); } 125 126 static MemAccessTy getUnknown(LLVMContext &Ctx) { 127 return MemAccessTy(Type::getVoidTy(Ctx), UnknownAddressSpace); 128 } 129 }; 130 131 /// This class holds data which is used to order reuse candidates. 132 class RegSortData { 133 public: 134 /// This represents the set of LSRUse indices which reference 135 /// a particular register. 136 SmallBitVector UsedByIndices; 137 138 void print(raw_ostream &OS) const; 139 void dump() const; 140 }; 141 142 } 143 144 void RegSortData::print(raw_ostream &OS) const { 145 OS << "[NumUses=" << UsedByIndices.count() << ']'; 146 } 147 148 LLVM_DUMP_METHOD 149 void RegSortData::dump() const { 150 print(errs()); errs() << '\n'; 151 } 152 153 namespace { 154 155 /// Map register candidates to information about how they are used. 156 class RegUseTracker { 157 typedef DenseMap<const SCEV *, RegSortData> RegUsesTy; 158 159 RegUsesTy RegUsesMap; 160 SmallVector<const SCEV *, 16> RegSequence; 161 162 public: 163 void countRegister(const SCEV *Reg, size_t LUIdx); 164 void dropRegister(const SCEV *Reg, size_t LUIdx); 165 void swapAndDropUse(size_t LUIdx, size_t LastLUIdx); 166 167 bool isRegUsedByUsesOtherThan(const SCEV *Reg, size_t LUIdx) const; 168 169 const SmallBitVector &getUsedByIndices(const SCEV *Reg) const; 170 171 void clear(); 172 173 typedef SmallVectorImpl<const SCEV *>::iterator iterator; 174 typedef SmallVectorImpl<const SCEV *>::const_iterator const_iterator; 175 iterator begin() { return RegSequence.begin(); } 176 iterator end() { return RegSequence.end(); } 177 const_iterator begin() const { return RegSequence.begin(); } 178 const_iterator end() const { return RegSequence.end(); } 179 }; 180 181 } 182 183 void 184 RegUseTracker::countRegister(const SCEV *Reg, size_t LUIdx) { 185 std::pair<RegUsesTy::iterator, bool> Pair = 186 RegUsesMap.insert(std::make_pair(Reg, RegSortData())); 187 RegSortData &RSD = Pair.first->second; 188 if (Pair.second) 189 RegSequence.push_back(Reg); 190 RSD.UsedByIndices.resize(std::max(RSD.UsedByIndices.size(), LUIdx + 1)); 191 RSD.UsedByIndices.set(LUIdx); 192 } 193 194 void 195 RegUseTracker::dropRegister(const SCEV *Reg, size_t LUIdx) { 196 RegUsesTy::iterator It = RegUsesMap.find(Reg); 197 assert(It != RegUsesMap.end()); 198 RegSortData &RSD = It->second; 199 assert(RSD.UsedByIndices.size() > LUIdx); 200 RSD.UsedByIndices.reset(LUIdx); 201 } 202 203 void 204 RegUseTracker::swapAndDropUse(size_t LUIdx, size_t LastLUIdx) { 205 assert(LUIdx <= LastLUIdx); 206 207 // Update RegUses. The data structure is not optimized for this purpose; 208 // we must iterate through it and update each of the bit vectors. 209 for (auto &Pair : RegUsesMap) { 210 SmallBitVector &UsedByIndices = Pair.second.UsedByIndices; 211 if (LUIdx < UsedByIndices.size()) 212 UsedByIndices[LUIdx] = 213 LastLUIdx < UsedByIndices.size() ? UsedByIndices[LastLUIdx] : 0; 214 UsedByIndices.resize(std::min(UsedByIndices.size(), LastLUIdx)); 215 } 216 } 217 218 bool 219 RegUseTracker::isRegUsedByUsesOtherThan(const SCEV *Reg, size_t LUIdx) const { 220 RegUsesTy::const_iterator I = RegUsesMap.find(Reg); 221 if (I == RegUsesMap.end()) 222 return false; 223 const SmallBitVector &UsedByIndices = I->second.UsedByIndices; 224 int i = UsedByIndices.find_first(); 225 if (i == -1) return false; 226 if ((size_t)i != LUIdx) return true; 227 return UsedByIndices.find_next(i) != -1; 228 } 229 230 const SmallBitVector &RegUseTracker::getUsedByIndices(const SCEV *Reg) const { 231 RegUsesTy::const_iterator I = RegUsesMap.find(Reg); 232 assert(I != RegUsesMap.end() && "Unknown register!"); 233 return I->second.UsedByIndices; 234 } 235 236 void RegUseTracker::clear() { 237 RegUsesMap.clear(); 238 RegSequence.clear(); 239 } 240 241 namespace { 242 243 /// This class holds information that describes a formula for computing 244 /// satisfying a use. It may include broken-out immediates and scaled registers. 245 struct Formula { 246 /// Global base address used for complex addressing. 247 GlobalValue *BaseGV; 248 249 /// Base offset for complex addressing. 250 int64_t BaseOffset; 251 252 /// Whether any complex addressing has a base register. 253 bool HasBaseReg; 254 255 /// The scale of any complex addressing. 256 int64_t Scale; 257 258 /// The list of "base" registers for this use. When this is non-empty. The 259 /// canonical representation of a formula is 260 /// 1. BaseRegs.size > 1 implies ScaledReg != NULL and 261 /// 2. ScaledReg != NULL implies Scale != 1 || !BaseRegs.empty(). 262 /// #1 enforces that the scaled register is always used when at least two 263 /// registers are needed by the formula: e.g., reg1 + reg2 is reg1 + 1 * reg2. 264 /// #2 enforces that 1 * reg is reg. 265 /// This invariant can be temporarly broken while building a formula. 266 /// However, every formula inserted into the LSRInstance must be in canonical 267 /// form. 268 SmallVector<const SCEV *, 4> BaseRegs; 269 270 /// The 'scaled' register for this use. This should be non-null when Scale is 271 /// not zero. 272 const SCEV *ScaledReg; 273 274 /// An additional constant offset which added near the use. This requires a 275 /// temporary register, but the offset itself can live in an add immediate 276 /// field rather than a register. 277 int64_t UnfoldedOffset; 278 279 Formula() 280 : BaseGV(nullptr), BaseOffset(0), HasBaseReg(false), Scale(0), 281 ScaledReg(nullptr), UnfoldedOffset(0) {} 282 283 void initialMatch(const SCEV *S, Loop *L, ScalarEvolution &SE); 284 285 bool isCanonical() const; 286 287 void canonicalize(); 288 289 bool unscale(); 290 291 size_t getNumRegs() const; 292 Type *getType() const; 293 294 void deleteBaseReg(const SCEV *&S); 295 296 bool referencesReg(const SCEV *S) const; 297 bool hasRegsUsedByUsesOtherThan(size_t LUIdx, 298 const RegUseTracker &RegUses) const; 299 300 void print(raw_ostream &OS) const; 301 void dump() const; 302 }; 303 304 } 305 306 /// Recursion helper for initialMatch. 307 static void DoInitialMatch(const SCEV *S, Loop *L, 308 SmallVectorImpl<const SCEV *> &Good, 309 SmallVectorImpl<const SCEV *> &Bad, 310 ScalarEvolution &SE) { 311 // Collect expressions which properly dominate the loop header. 312 if (SE.properlyDominates(S, L->getHeader())) { 313 Good.push_back(S); 314 return; 315 } 316 317 // Look at add operands. 318 if (const SCEVAddExpr *Add = dyn_cast<SCEVAddExpr>(S)) { 319 for (const SCEV *S : Add->operands()) 320 DoInitialMatch(S, L, Good, Bad, SE); 321 return; 322 } 323 324 // Look at addrec operands. 325 if (const SCEVAddRecExpr *AR = dyn_cast<SCEVAddRecExpr>(S)) 326 if (!AR->getStart()->isZero()) { 327 DoInitialMatch(AR->getStart(), L, Good, Bad, SE); 328 DoInitialMatch(SE.getAddRecExpr(SE.getConstant(AR->getType(), 0), 329 AR->getStepRecurrence(SE), 330 // FIXME: AR->getNoWrapFlags() 331 AR->getLoop(), SCEV::FlagAnyWrap), 332 L, Good, Bad, SE); 333 return; 334 } 335 336 // Handle a multiplication by -1 (negation) if it didn't fold. 337 if (const SCEVMulExpr *Mul = dyn_cast<SCEVMulExpr>(S)) 338 if (Mul->getOperand(0)->isAllOnesValue()) { 339 SmallVector<const SCEV *, 4> Ops(Mul->op_begin()+1, Mul->op_end()); 340 const SCEV *NewMul = SE.getMulExpr(Ops); 341 342 SmallVector<const SCEV *, 4> MyGood; 343 SmallVector<const SCEV *, 4> MyBad; 344 DoInitialMatch(NewMul, L, MyGood, MyBad, SE); 345 const SCEV *NegOne = SE.getSCEV(ConstantInt::getAllOnesValue( 346 SE.getEffectiveSCEVType(NewMul->getType()))); 347 for (const SCEV *S : MyGood) 348 Good.push_back(SE.getMulExpr(NegOne, S)); 349 for (const SCEV *S : MyBad) 350 Bad.push_back(SE.getMulExpr(NegOne, S)); 351 return; 352 } 353 354 // Ok, we can't do anything interesting. Just stuff the whole thing into a 355 // register and hope for the best. 356 Bad.push_back(S); 357 } 358 359 /// Incorporate loop-variant parts of S into this Formula, attempting to keep 360 /// all loop-invariant and loop-computable values in a single base register. 361 void Formula::initialMatch(const SCEV *S, Loop *L, ScalarEvolution &SE) { 362 SmallVector<const SCEV *, 4> Good; 363 SmallVector<const SCEV *, 4> Bad; 364 DoInitialMatch(S, L, Good, Bad, SE); 365 if (!Good.empty()) { 366 const SCEV *Sum = SE.getAddExpr(Good); 367 if (!Sum->isZero()) 368 BaseRegs.push_back(Sum); 369 HasBaseReg = true; 370 } 371 if (!Bad.empty()) { 372 const SCEV *Sum = SE.getAddExpr(Bad); 373 if (!Sum->isZero()) 374 BaseRegs.push_back(Sum); 375 HasBaseReg = true; 376 } 377 canonicalize(); 378 } 379 380 /// \brief Check whether or not this formula statisfies the canonical 381 /// representation. 382 /// \see Formula::BaseRegs. 383 bool Formula::isCanonical() const { 384 if (ScaledReg) 385 return Scale != 1 || !BaseRegs.empty(); 386 return BaseRegs.size() <= 1; 387 } 388 389 /// \brief Helper method to morph a formula into its canonical representation. 390 /// \see Formula::BaseRegs. 391 /// Every formula having more than one base register, must use the ScaledReg 392 /// field. Otherwise, we would have to do special cases everywhere in LSR 393 /// to treat reg1 + reg2 + ... the same way as reg1 + 1*reg2 + ... 394 /// On the other hand, 1*reg should be canonicalized into reg. 395 void Formula::canonicalize() { 396 if (isCanonical()) 397 return; 398 // So far we did not need this case. This is easy to implement but it is 399 // useless to maintain dead code. Beside it could hurt compile time. 400 assert(!BaseRegs.empty() && "1*reg => reg, should not be needed."); 401 // Keep the invariant sum in BaseRegs and one of the variant sum in ScaledReg. 402 ScaledReg = BaseRegs.back(); 403 BaseRegs.pop_back(); 404 Scale = 1; 405 size_t BaseRegsSize = BaseRegs.size(); 406 size_t Try = 0; 407 // If ScaledReg is an invariant, try to find a variant expression. 408 while (Try < BaseRegsSize && !isa<SCEVAddRecExpr>(ScaledReg)) 409 std::swap(ScaledReg, BaseRegs[Try++]); 410 } 411 412 /// \brief Get rid of the scale in the formula. 413 /// In other words, this method morphes reg1 + 1*reg2 into reg1 + reg2. 414 /// \return true if it was possible to get rid of the scale, false otherwise. 415 /// \note After this operation the formula may not be in the canonical form. 416 bool Formula::unscale() { 417 if (Scale != 1) 418 return false; 419 Scale = 0; 420 BaseRegs.push_back(ScaledReg); 421 ScaledReg = nullptr; 422 return true; 423 } 424 425 /// Return the total number of register operands used by this formula. This does 426 /// not include register uses implied by non-constant addrec strides. 427 size_t Formula::getNumRegs() const { 428 return !!ScaledReg + BaseRegs.size(); 429 } 430 431 /// Return the type of this formula, if it has one, or null otherwise. This type 432 /// is meaningless except for the bit size. 433 Type *Formula::getType() const { 434 return !BaseRegs.empty() ? BaseRegs.front()->getType() : 435 ScaledReg ? ScaledReg->getType() : 436 BaseGV ? BaseGV->getType() : 437 nullptr; 438 } 439 440 /// Delete the given base reg from the BaseRegs list. 441 void Formula::deleteBaseReg(const SCEV *&S) { 442 if (&S != &BaseRegs.back()) 443 std::swap(S, BaseRegs.back()); 444 BaseRegs.pop_back(); 445 } 446 447 /// Test if this formula references the given register. 448 bool Formula::referencesReg(const SCEV *S) const { 449 return S == ScaledReg || 450 std::find(BaseRegs.begin(), BaseRegs.end(), S) != BaseRegs.end(); 451 } 452 453 /// Test whether this formula uses registers which are used by uses other than 454 /// the use with the given index. 455 bool Formula::hasRegsUsedByUsesOtherThan(size_t LUIdx, 456 const RegUseTracker &RegUses) const { 457 if (ScaledReg) 458 if (RegUses.isRegUsedByUsesOtherThan(ScaledReg, LUIdx)) 459 return true; 460 for (const SCEV *BaseReg : BaseRegs) 461 if (RegUses.isRegUsedByUsesOtherThan(BaseReg, LUIdx)) 462 return true; 463 return false; 464 } 465 466 void Formula::print(raw_ostream &OS) const { 467 bool First = true; 468 if (BaseGV) { 469 if (!First) OS << " + "; else First = false; 470 BaseGV->printAsOperand(OS, /*PrintType=*/false); 471 } 472 if (BaseOffset != 0) { 473 if (!First) OS << " + "; else First = false; 474 OS << BaseOffset; 475 } 476 for (const SCEV *BaseReg : BaseRegs) { 477 if (!First) OS << " + "; else First = false; 478 OS << "reg(" << *BaseReg << ')'; 479 } 480 if (HasBaseReg && BaseRegs.empty()) { 481 if (!First) OS << " + "; else First = false; 482 OS << "**error: HasBaseReg**"; 483 } else if (!HasBaseReg && !BaseRegs.empty()) { 484 if (!First) OS << " + "; else First = false; 485 OS << "**error: !HasBaseReg**"; 486 } 487 if (Scale != 0) { 488 if (!First) OS << " + "; else First = false; 489 OS << Scale << "*reg("; 490 if (ScaledReg) 491 OS << *ScaledReg; 492 else 493 OS << "<unknown>"; 494 OS << ')'; 495 } 496 if (UnfoldedOffset != 0) { 497 if (!First) OS << " + "; 498 OS << "imm(" << UnfoldedOffset << ')'; 499 } 500 } 501 502 LLVM_DUMP_METHOD 503 void Formula::dump() const { 504 print(errs()); errs() << '\n'; 505 } 506 507 /// Return true if the given addrec can be sign-extended without changing its 508 /// value. 509 static bool isAddRecSExtable(const SCEVAddRecExpr *AR, ScalarEvolution &SE) { 510 Type *WideTy = 511 IntegerType::get(SE.getContext(), SE.getTypeSizeInBits(AR->getType()) + 1); 512 return isa<SCEVAddRecExpr>(SE.getSignExtendExpr(AR, WideTy)); 513 } 514 515 /// Return true if the given add can be sign-extended without changing its 516 /// value. 517 static bool isAddSExtable(const SCEVAddExpr *A, ScalarEvolution &SE) { 518 Type *WideTy = 519 IntegerType::get(SE.getContext(), SE.getTypeSizeInBits(A->getType()) + 1); 520 return isa<SCEVAddExpr>(SE.getSignExtendExpr(A, WideTy)); 521 } 522 523 /// Return true if the given mul can be sign-extended without changing its 524 /// value. 525 static bool isMulSExtable(const SCEVMulExpr *M, ScalarEvolution &SE) { 526 Type *WideTy = 527 IntegerType::get(SE.getContext(), 528 SE.getTypeSizeInBits(M->getType()) * M->getNumOperands()); 529 return isa<SCEVMulExpr>(SE.getSignExtendExpr(M, WideTy)); 530 } 531 532 /// Return an expression for LHS /s RHS, if it can be determined and if the 533 /// remainder is known to be zero, or null otherwise. If IgnoreSignificantBits 534 /// is true, expressions like (X * Y) /s Y are simplified to Y, ignoring that 535 /// the multiplication may overflow, which is useful when the result will be 536 /// used in a context where the most significant bits are ignored. 537 static const SCEV *getExactSDiv(const SCEV *LHS, const SCEV *RHS, 538 ScalarEvolution &SE, 539 bool IgnoreSignificantBits = false) { 540 // Handle the trivial case, which works for any SCEV type. 541 if (LHS == RHS) 542 return SE.getConstant(LHS->getType(), 1); 543 544 // Handle a few RHS special cases. 545 const SCEVConstant *RC = dyn_cast<SCEVConstant>(RHS); 546 if (RC) { 547 const APInt &RA = RC->getAPInt(); 548 // Handle x /s -1 as x * -1, to give ScalarEvolution a chance to do 549 // some folding. 550 if (RA.isAllOnesValue()) 551 return SE.getMulExpr(LHS, RC); 552 // Handle x /s 1 as x. 553 if (RA == 1) 554 return LHS; 555 } 556 557 // Check for a division of a constant by a constant. 558 if (const SCEVConstant *C = dyn_cast<SCEVConstant>(LHS)) { 559 if (!RC) 560 return nullptr; 561 const APInt &LA = C->getAPInt(); 562 const APInt &RA = RC->getAPInt(); 563 if (LA.srem(RA) != 0) 564 return nullptr; 565 return SE.getConstant(LA.sdiv(RA)); 566 } 567 568 // Distribute the sdiv over addrec operands, if the addrec doesn't overflow. 569 if (const SCEVAddRecExpr *AR = dyn_cast<SCEVAddRecExpr>(LHS)) { 570 if (IgnoreSignificantBits || isAddRecSExtable(AR, SE)) { 571 const SCEV *Step = getExactSDiv(AR->getStepRecurrence(SE), RHS, SE, 572 IgnoreSignificantBits); 573 if (!Step) return nullptr; 574 const SCEV *Start = getExactSDiv(AR->getStart(), RHS, SE, 575 IgnoreSignificantBits); 576 if (!Start) return nullptr; 577 // FlagNW is independent of the start value, step direction, and is 578 // preserved with smaller magnitude steps. 579 // FIXME: AR->getNoWrapFlags(SCEV::FlagNW) 580 return SE.getAddRecExpr(Start, Step, AR->getLoop(), SCEV::FlagAnyWrap); 581 } 582 return nullptr; 583 } 584 585 // Distribute the sdiv over add operands, if the add doesn't overflow. 586 if (const SCEVAddExpr *Add = dyn_cast<SCEVAddExpr>(LHS)) { 587 if (IgnoreSignificantBits || isAddSExtable(Add, SE)) { 588 SmallVector<const SCEV *, 8> Ops; 589 for (const SCEV *S : Add->operands()) { 590 const SCEV *Op = getExactSDiv(S, RHS, SE, IgnoreSignificantBits); 591 if (!Op) return nullptr; 592 Ops.push_back(Op); 593 } 594 return SE.getAddExpr(Ops); 595 } 596 return nullptr; 597 } 598 599 // Check for a multiply operand that we can pull RHS out of. 600 if (const SCEVMulExpr *Mul = dyn_cast<SCEVMulExpr>(LHS)) { 601 if (IgnoreSignificantBits || isMulSExtable(Mul, SE)) { 602 SmallVector<const SCEV *, 4> Ops; 603 bool Found = false; 604 for (const SCEV *S : Mul->operands()) { 605 if (!Found) 606 if (const SCEV *Q = getExactSDiv(S, RHS, SE, 607 IgnoreSignificantBits)) { 608 S = Q; 609 Found = true; 610 } 611 Ops.push_back(S); 612 } 613 return Found ? SE.getMulExpr(Ops) : nullptr; 614 } 615 return nullptr; 616 } 617 618 // Otherwise we don't know. 619 return nullptr; 620 } 621 622 /// If S involves the addition of a constant integer value, return that integer 623 /// value, and mutate S to point to a new SCEV with that value excluded. 624 static int64_t ExtractImmediate(const SCEV *&S, ScalarEvolution &SE) { 625 if (const SCEVConstant *C = dyn_cast<SCEVConstant>(S)) { 626 if (C->getAPInt().getMinSignedBits() <= 64) { 627 S = SE.getConstant(C->getType(), 0); 628 return C->getValue()->getSExtValue(); 629 } 630 } else if (const SCEVAddExpr *Add = dyn_cast<SCEVAddExpr>(S)) { 631 SmallVector<const SCEV *, 8> NewOps(Add->op_begin(), Add->op_end()); 632 int64_t Result = ExtractImmediate(NewOps.front(), SE); 633 if (Result != 0) 634 S = SE.getAddExpr(NewOps); 635 return Result; 636 } else if (const SCEVAddRecExpr *AR = dyn_cast<SCEVAddRecExpr>(S)) { 637 SmallVector<const SCEV *, 8> NewOps(AR->op_begin(), AR->op_end()); 638 int64_t Result = ExtractImmediate(NewOps.front(), SE); 639 if (Result != 0) 640 S = SE.getAddRecExpr(NewOps, AR->getLoop(), 641 // FIXME: AR->getNoWrapFlags(SCEV::FlagNW) 642 SCEV::FlagAnyWrap); 643 return Result; 644 } 645 return 0; 646 } 647 648 /// If S involves the addition of a GlobalValue address, return that symbol, and 649 /// mutate S to point to a new SCEV with that value excluded. 650 static GlobalValue *ExtractSymbol(const SCEV *&S, ScalarEvolution &SE) { 651 if (const SCEVUnknown *U = dyn_cast<SCEVUnknown>(S)) { 652 if (GlobalValue *GV = dyn_cast<GlobalValue>(U->getValue())) { 653 S = SE.getConstant(GV->getType(), 0); 654 return GV; 655 } 656 } else if (const SCEVAddExpr *Add = dyn_cast<SCEVAddExpr>(S)) { 657 SmallVector<const SCEV *, 8> NewOps(Add->op_begin(), Add->op_end()); 658 GlobalValue *Result = ExtractSymbol(NewOps.back(), SE); 659 if (Result) 660 S = SE.getAddExpr(NewOps); 661 return Result; 662 } else if (const SCEVAddRecExpr *AR = dyn_cast<SCEVAddRecExpr>(S)) { 663 SmallVector<const SCEV *, 8> NewOps(AR->op_begin(), AR->op_end()); 664 GlobalValue *Result = ExtractSymbol(NewOps.front(), SE); 665 if (Result) 666 S = SE.getAddRecExpr(NewOps, AR->getLoop(), 667 // FIXME: AR->getNoWrapFlags(SCEV::FlagNW) 668 SCEV::FlagAnyWrap); 669 return Result; 670 } 671 return nullptr; 672 } 673 674 /// Returns true if the specified instruction is using the specified value as an 675 /// address. 676 static bool isAddressUse(Instruction *Inst, Value *OperandVal) { 677 bool isAddress = isa<LoadInst>(Inst); 678 if (StoreInst *SI = dyn_cast<StoreInst>(Inst)) { 679 if (SI->getOperand(1) == OperandVal) 680 isAddress = true; 681 } else if (IntrinsicInst *II = dyn_cast<IntrinsicInst>(Inst)) { 682 // Addressing modes can also be folded into prefetches and a variety 683 // of intrinsics. 684 switch (II->getIntrinsicID()) { 685 default: break; 686 case Intrinsic::prefetch: 687 case Intrinsic::x86_sse_storeu_ps: 688 case Intrinsic::x86_sse2_storeu_pd: 689 case Intrinsic::x86_sse2_storeu_dq: 690 case Intrinsic::x86_sse2_storel_dq: 691 if (II->getArgOperand(0) == OperandVal) 692 isAddress = true; 693 break; 694 } 695 } 696 return isAddress; 697 } 698 699 /// Return the type of the memory being accessed. 700 static MemAccessTy getAccessType(const Instruction *Inst) { 701 MemAccessTy AccessTy(Inst->getType(), MemAccessTy::UnknownAddressSpace); 702 if (const StoreInst *SI = dyn_cast<StoreInst>(Inst)) { 703 AccessTy.MemTy = SI->getOperand(0)->getType(); 704 AccessTy.AddrSpace = SI->getPointerAddressSpace(); 705 } else if (const LoadInst *LI = dyn_cast<LoadInst>(Inst)) { 706 AccessTy.AddrSpace = LI->getPointerAddressSpace(); 707 } else if (const IntrinsicInst *II = dyn_cast<IntrinsicInst>(Inst)) { 708 // Addressing modes can also be folded into prefetches and a variety 709 // of intrinsics. 710 switch (II->getIntrinsicID()) { 711 default: break; 712 case Intrinsic::x86_sse_storeu_ps: 713 case Intrinsic::x86_sse2_storeu_pd: 714 case Intrinsic::x86_sse2_storeu_dq: 715 case Intrinsic::x86_sse2_storel_dq: 716 AccessTy.MemTy = II->getArgOperand(0)->getType(); 717 break; 718 } 719 } 720 721 // All pointers have the same requirements, so canonicalize them to an 722 // arbitrary pointer type to minimize variation. 723 if (PointerType *PTy = dyn_cast<PointerType>(AccessTy.MemTy)) 724 AccessTy.MemTy = PointerType::get(IntegerType::get(PTy->getContext(), 1), 725 PTy->getAddressSpace()); 726 727 return AccessTy; 728 } 729 730 /// Return true if this AddRec is already a phi in its loop. 731 static bool isExistingPhi(const SCEVAddRecExpr *AR, ScalarEvolution &SE) { 732 for (BasicBlock::iterator I = AR->getLoop()->getHeader()->begin(); 733 PHINode *PN = dyn_cast<PHINode>(I); ++I) { 734 if (SE.isSCEVable(PN->getType()) && 735 (SE.getEffectiveSCEVType(PN->getType()) == 736 SE.getEffectiveSCEVType(AR->getType())) && 737 SE.getSCEV(PN) == AR) 738 return true; 739 } 740 return false; 741 } 742 743 /// Check if expanding this expression is likely to incur significant cost. This 744 /// is tricky because SCEV doesn't track which expressions are actually computed 745 /// by the current IR. 746 /// 747 /// We currently allow expansion of IV increments that involve adds, 748 /// multiplication by constants, and AddRecs from existing phis. 749 /// 750 /// TODO: Allow UDivExpr if we can find an existing IV increment that is an 751 /// obvious multiple of the UDivExpr. 752 static bool isHighCostExpansion(const SCEV *S, 753 SmallPtrSetImpl<const SCEV*> &Processed, 754 ScalarEvolution &SE) { 755 // Zero/One operand expressions 756 switch (S->getSCEVType()) { 757 case scUnknown: 758 case scConstant: 759 return false; 760 case scTruncate: 761 return isHighCostExpansion(cast<SCEVTruncateExpr>(S)->getOperand(), 762 Processed, SE); 763 case scZeroExtend: 764 return isHighCostExpansion(cast<SCEVZeroExtendExpr>(S)->getOperand(), 765 Processed, SE); 766 case scSignExtend: 767 return isHighCostExpansion(cast<SCEVSignExtendExpr>(S)->getOperand(), 768 Processed, SE); 769 } 770 771 if (!Processed.insert(S).second) 772 return false; 773 774 if (const SCEVAddExpr *Add = dyn_cast<SCEVAddExpr>(S)) { 775 for (const SCEV *S : Add->operands()) { 776 if (isHighCostExpansion(S, Processed, SE)) 777 return true; 778 } 779 return false; 780 } 781 782 if (const SCEVMulExpr *Mul = dyn_cast<SCEVMulExpr>(S)) { 783 if (Mul->getNumOperands() == 2) { 784 // Multiplication by a constant is ok 785 if (isa<SCEVConstant>(Mul->getOperand(0))) 786 return isHighCostExpansion(Mul->getOperand(1), Processed, SE); 787 788 // If we have the value of one operand, check if an existing 789 // multiplication already generates this expression. 790 if (const SCEVUnknown *U = dyn_cast<SCEVUnknown>(Mul->getOperand(1))) { 791 Value *UVal = U->getValue(); 792 for (User *UR : UVal->users()) { 793 // If U is a constant, it may be used by a ConstantExpr. 794 Instruction *UI = dyn_cast<Instruction>(UR); 795 if (UI && UI->getOpcode() == Instruction::Mul && 796 SE.isSCEVable(UI->getType())) { 797 return SE.getSCEV(UI) == Mul; 798 } 799 } 800 } 801 } 802 } 803 804 if (const SCEVAddRecExpr *AR = dyn_cast<SCEVAddRecExpr>(S)) { 805 if (isExistingPhi(AR, SE)) 806 return false; 807 } 808 809 // Fow now, consider any other type of expression (div/mul/min/max) high cost. 810 return true; 811 } 812 813 /// If any of the instructions is the specified set are trivially dead, delete 814 /// them and see if this makes any of their operands subsequently dead. 815 static bool 816 DeleteTriviallyDeadInstructions(SmallVectorImpl<WeakVH> &DeadInsts) { 817 bool Changed = false; 818 819 while (!DeadInsts.empty()) { 820 Value *V = DeadInsts.pop_back_val(); 821 Instruction *I = dyn_cast_or_null<Instruction>(V); 822 823 if (!I || !isInstructionTriviallyDead(I)) 824 continue; 825 826 for (Use &O : I->operands()) 827 if (Instruction *U = dyn_cast<Instruction>(O)) { 828 O = nullptr; 829 if (U->use_empty()) 830 DeadInsts.emplace_back(U); 831 } 832 833 I->eraseFromParent(); 834 Changed = true; 835 } 836 837 return Changed; 838 } 839 840 namespace { 841 class LSRUse; 842 } 843 844 /// \brief Check if the addressing mode defined by \p F is completely 845 /// folded in \p LU at isel time. 846 /// This includes address-mode folding and special icmp tricks. 847 /// This function returns true if \p LU can accommodate what \p F 848 /// defines and up to 1 base + 1 scaled + offset. 849 /// In other words, if \p F has several base registers, this function may 850 /// still return true. Therefore, users still need to account for 851 /// additional base registers and/or unfolded offsets to derive an 852 /// accurate cost model. 853 static bool isAMCompletelyFolded(const TargetTransformInfo &TTI, 854 const LSRUse &LU, const Formula &F); 855 // Get the cost of the scaling factor used in F for LU. 856 static unsigned getScalingFactorCost(const TargetTransformInfo &TTI, 857 const LSRUse &LU, const Formula &F); 858 859 namespace { 860 861 /// This class is used to measure and compare candidate formulae. 862 class Cost { 863 /// TODO: Some of these could be merged. Also, a lexical ordering 864 /// isn't always optimal. 865 unsigned NumRegs; 866 unsigned AddRecCost; 867 unsigned NumIVMuls; 868 unsigned NumBaseAdds; 869 unsigned ImmCost; 870 unsigned SetupCost; 871 unsigned ScaleCost; 872 873 public: 874 Cost() 875 : NumRegs(0), AddRecCost(0), NumIVMuls(0), NumBaseAdds(0), ImmCost(0), 876 SetupCost(0), ScaleCost(0) {} 877 878 bool operator<(const Cost &Other) const; 879 880 void Lose(); 881 882 #ifndef NDEBUG 883 // Once any of the metrics loses, they must all remain losers. 884 bool isValid() { 885 return ((NumRegs | AddRecCost | NumIVMuls | NumBaseAdds 886 | ImmCost | SetupCost | ScaleCost) != ~0u) 887 || ((NumRegs & AddRecCost & NumIVMuls & NumBaseAdds 888 & ImmCost & SetupCost & ScaleCost) == ~0u); 889 } 890 #endif 891 892 bool isLoser() { 893 assert(isValid() && "invalid cost"); 894 return NumRegs == ~0u; 895 } 896 897 void RateFormula(const TargetTransformInfo &TTI, 898 const Formula &F, 899 SmallPtrSetImpl<const SCEV *> &Regs, 900 const DenseSet<const SCEV *> &VisitedRegs, 901 const Loop *L, 902 const SmallVectorImpl<int64_t> &Offsets, 903 ScalarEvolution &SE, DominatorTree &DT, 904 const LSRUse &LU, 905 SmallPtrSetImpl<const SCEV *> *LoserRegs = nullptr); 906 907 void print(raw_ostream &OS) const; 908 void dump() const; 909 910 private: 911 void RateRegister(const SCEV *Reg, 912 SmallPtrSetImpl<const SCEV *> &Regs, 913 const Loop *L, 914 ScalarEvolution &SE, DominatorTree &DT); 915 void RatePrimaryRegister(const SCEV *Reg, 916 SmallPtrSetImpl<const SCEV *> &Regs, 917 const Loop *L, 918 ScalarEvolution &SE, DominatorTree &DT, 919 SmallPtrSetImpl<const SCEV *> *LoserRegs); 920 }; 921 922 } 923 924 /// Tally up interesting quantities from the given register. 925 void Cost::RateRegister(const SCEV *Reg, 926 SmallPtrSetImpl<const SCEV *> &Regs, 927 const Loop *L, 928 ScalarEvolution &SE, DominatorTree &DT) { 929 if (const SCEVAddRecExpr *AR = dyn_cast<SCEVAddRecExpr>(Reg)) { 930 // If this is an addrec for another loop, don't second-guess its addrec phi 931 // nodes. LSR isn't currently smart enough to reason about more than one 932 // loop at a time. LSR has already run on inner loops, will not run on outer 933 // loops, and cannot be expected to change sibling loops. 934 if (AR->getLoop() != L) { 935 // If the AddRec exists, consider it's register free and leave it alone. 936 if (isExistingPhi(AR, SE)) 937 return; 938 939 // Otherwise, do not consider this formula at all. 940 Lose(); 941 return; 942 } 943 AddRecCost += 1; /// TODO: This should be a function of the stride. 944 945 // Add the step value register, if it needs one. 946 // TODO: The non-affine case isn't precisely modeled here. 947 if (!AR->isAffine() || !isa<SCEVConstant>(AR->getOperand(1))) { 948 if (!Regs.count(AR->getOperand(1))) { 949 RateRegister(AR->getOperand(1), Regs, L, SE, DT); 950 if (isLoser()) 951 return; 952 } 953 } 954 } 955 ++NumRegs; 956 957 // Rough heuristic; favor registers which don't require extra setup 958 // instructions in the preheader. 959 if (!isa<SCEVUnknown>(Reg) && 960 !isa<SCEVConstant>(Reg) && 961 !(isa<SCEVAddRecExpr>(Reg) && 962 (isa<SCEVUnknown>(cast<SCEVAddRecExpr>(Reg)->getStart()) || 963 isa<SCEVConstant>(cast<SCEVAddRecExpr>(Reg)->getStart())))) 964 ++SetupCost; 965 966 NumIVMuls += isa<SCEVMulExpr>(Reg) && 967 SE.hasComputableLoopEvolution(Reg, L); 968 } 969 970 /// Record this register in the set. If we haven't seen it before, rate 971 /// it. Optional LoserRegs provides a way to declare any formula that refers to 972 /// one of those regs an instant loser. 973 void Cost::RatePrimaryRegister(const SCEV *Reg, 974 SmallPtrSetImpl<const SCEV *> &Regs, 975 const Loop *L, 976 ScalarEvolution &SE, DominatorTree &DT, 977 SmallPtrSetImpl<const SCEV *> *LoserRegs) { 978 if (LoserRegs && LoserRegs->count(Reg)) { 979 Lose(); 980 return; 981 } 982 if (Regs.insert(Reg).second) { 983 RateRegister(Reg, Regs, L, SE, DT); 984 if (LoserRegs && isLoser()) 985 LoserRegs->insert(Reg); 986 } 987 } 988 989 void Cost::RateFormula(const TargetTransformInfo &TTI, 990 const Formula &F, 991 SmallPtrSetImpl<const SCEV *> &Regs, 992 const DenseSet<const SCEV *> &VisitedRegs, 993 const Loop *L, 994 const SmallVectorImpl<int64_t> &Offsets, 995 ScalarEvolution &SE, DominatorTree &DT, 996 const LSRUse &LU, 997 SmallPtrSetImpl<const SCEV *> *LoserRegs) { 998 assert(F.isCanonical() && "Cost is accurate only for canonical formula"); 999 // Tally up the registers. 1000 if (const SCEV *ScaledReg = F.ScaledReg) { 1001 if (VisitedRegs.count(ScaledReg)) { 1002 Lose(); 1003 return; 1004 } 1005 RatePrimaryRegister(ScaledReg, Regs, L, SE, DT, LoserRegs); 1006 if (isLoser()) 1007 return; 1008 } 1009 for (const SCEV *BaseReg : F.BaseRegs) { 1010 if (VisitedRegs.count(BaseReg)) { 1011 Lose(); 1012 return; 1013 } 1014 RatePrimaryRegister(BaseReg, Regs, L, SE, DT, LoserRegs); 1015 if (isLoser()) 1016 return; 1017 } 1018 1019 // Determine how many (unfolded) adds we'll need inside the loop. 1020 size_t NumBaseParts = F.getNumRegs(); 1021 if (NumBaseParts > 1) 1022 // Do not count the base and a possible second register if the target 1023 // allows to fold 2 registers. 1024 NumBaseAdds += 1025 NumBaseParts - (1 + (F.Scale && isAMCompletelyFolded(TTI, LU, F))); 1026 NumBaseAdds += (F.UnfoldedOffset != 0); 1027 1028 // Accumulate non-free scaling amounts. 1029 ScaleCost += getScalingFactorCost(TTI, LU, F); 1030 1031 // Tally up the non-zero immediates. 1032 for (int64_t O : Offsets) { 1033 int64_t Offset = (uint64_t)O + F.BaseOffset; 1034 if (F.BaseGV) 1035 ImmCost += 64; // Handle symbolic values conservatively. 1036 // TODO: This should probably be the pointer size. 1037 else if (Offset != 0) 1038 ImmCost += APInt(64, Offset, true).getMinSignedBits(); 1039 } 1040 assert(isValid() && "invalid cost"); 1041 } 1042 1043 /// Set this cost to a losing value. 1044 void Cost::Lose() { 1045 NumRegs = ~0u; 1046 AddRecCost = ~0u; 1047 NumIVMuls = ~0u; 1048 NumBaseAdds = ~0u; 1049 ImmCost = ~0u; 1050 SetupCost = ~0u; 1051 ScaleCost = ~0u; 1052 } 1053 1054 /// Choose the lower cost. 1055 bool Cost::operator<(const Cost &Other) const { 1056 return std::tie(NumRegs, AddRecCost, NumIVMuls, NumBaseAdds, ScaleCost, 1057 ImmCost, SetupCost) < 1058 std::tie(Other.NumRegs, Other.AddRecCost, Other.NumIVMuls, 1059 Other.NumBaseAdds, Other.ScaleCost, Other.ImmCost, 1060 Other.SetupCost); 1061 } 1062 1063 void Cost::print(raw_ostream &OS) const { 1064 OS << NumRegs << " reg" << (NumRegs == 1 ? "" : "s"); 1065 if (AddRecCost != 0) 1066 OS << ", with addrec cost " << AddRecCost; 1067 if (NumIVMuls != 0) 1068 OS << ", plus " << NumIVMuls << " IV mul" << (NumIVMuls == 1 ? "" : "s"); 1069 if (NumBaseAdds != 0) 1070 OS << ", plus " << NumBaseAdds << " base add" 1071 << (NumBaseAdds == 1 ? "" : "s"); 1072 if (ScaleCost != 0) 1073 OS << ", plus " << ScaleCost << " scale cost"; 1074 if (ImmCost != 0) 1075 OS << ", plus " << ImmCost << " imm cost"; 1076 if (SetupCost != 0) 1077 OS << ", plus " << SetupCost << " setup cost"; 1078 } 1079 1080 LLVM_DUMP_METHOD 1081 void Cost::dump() const { 1082 print(errs()); errs() << '\n'; 1083 } 1084 1085 namespace { 1086 1087 /// An operand value in an instruction which is to be replaced with some 1088 /// equivalent, possibly strength-reduced, replacement. 1089 struct LSRFixup { 1090 /// The instruction which will be updated. 1091 Instruction *UserInst; 1092 1093 /// The operand of the instruction which will be replaced. The operand may be 1094 /// used more than once; every instance will be replaced. 1095 Value *OperandValToReplace; 1096 1097 /// If this user is to use the post-incremented value of an induction 1098 /// variable, this variable is non-null and holds the loop associated with the 1099 /// induction variable. 1100 PostIncLoopSet PostIncLoops; 1101 1102 /// The index of the LSRUse describing the expression which this fixup needs, 1103 /// minus an offset (below). 1104 size_t LUIdx; 1105 1106 /// A constant offset to be added to the LSRUse expression. This allows 1107 /// multiple fixups to share the same LSRUse with different offsets, for 1108 /// example in an unrolled loop. 1109 int64_t Offset; 1110 1111 bool isUseFullyOutsideLoop(const Loop *L) const; 1112 1113 LSRFixup(); 1114 1115 void print(raw_ostream &OS) const; 1116 void dump() const; 1117 }; 1118 1119 } 1120 1121 LSRFixup::LSRFixup() 1122 : UserInst(nullptr), OperandValToReplace(nullptr), LUIdx(~size_t(0)), 1123 Offset(0) {} 1124 1125 /// Test whether this fixup always uses its value outside of the given loop. 1126 bool LSRFixup::isUseFullyOutsideLoop(const Loop *L) const { 1127 // PHI nodes use their value in their incoming blocks. 1128 if (const PHINode *PN = dyn_cast<PHINode>(UserInst)) { 1129 for (unsigned i = 0, e = PN->getNumIncomingValues(); i != e; ++i) 1130 if (PN->getIncomingValue(i) == OperandValToReplace && 1131 L->contains(PN->getIncomingBlock(i))) 1132 return false; 1133 return true; 1134 } 1135 1136 return !L->contains(UserInst); 1137 } 1138 1139 void LSRFixup::print(raw_ostream &OS) const { 1140 OS << "UserInst="; 1141 // Store is common and interesting enough to be worth special-casing. 1142 if (StoreInst *Store = dyn_cast<StoreInst>(UserInst)) { 1143 OS << "store "; 1144 Store->getOperand(0)->printAsOperand(OS, /*PrintType=*/false); 1145 } else if (UserInst->getType()->isVoidTy()) 1146 OS << UserInst->getOpcodeName(); 1147 else 1148 UserInst->printAsOperand(OS, /*PrintType=*/false); 1149 1150 OS << ", OperandValToReplace="; 1151 OperandValToReplace->printAsOperand(OS, /*PrintType=*/false); 1152 1153 for (const Loop *PIL : PostIncLoops) { 1154 OS << ", PostIncLoop="; 1155 PIL->getHeader()->printAsOperand(OS, /*PrintType=*/false); 1156 } 1157 1158 if (LUIdx != ~size_t(0)) 1159 OS << ", LUIdx=" << LUIdx; 1160 1161 if (Offset != 0) 1162 OS << ", Offset=" << Offset; 1163 } 1164 1165 LLVM_DUMP_METHOD 1166 void LSRFixup::dump() const { 1167 print(errs()); errs() << '\n'; 1168 } 1169 1170 namespace { 1171 1172 /// A DenseMapInfo implementation for holding DenseMaps and DenseSets of sorted 1173 /// SmallVectors of const SCEV*. 1174 struct UniquifierDenseMapInfo { 1175 static SmallVector<const SCEV *, 4> getEmptyKey() { 1176 SmallVector<const SCEV *, 4> V; 1177 V.push_back(reinterpret_cast<const SCEV *>(-1)); 1178 return V; 1179 } 1180 1181 static SmallVector<const SCEV *, 4> getTombstoneKey() { 1182 SmallVector<const SCEV *, 4> V; 1183 V.push_back(reinterpret_cast<const SCEV *>(-2)); 1184 return V; 1185 } 1186 1187 static unsigned getHashValue(const SmallVector<const SCEV *, 4> &V) { 1188 return static_cast<unsigned>(hash_combine_range(V.begin(), V.end())); 1189 } 1190 1191 static bool isEqual(const SmallVector<const SCEV *, 4> &LHS, 1192 const SmallVector<const SCEV *, 4> &RHS) { 1193 return LHS == RHS; 1194 } 1195 }; 1196 1197 /// This class holds the state that LSR keeps for each use in IVUsers, as well 1198 /// as uses invented by LSR itself. It includes information about what kinds of 1199 /// things can be folded into the user, information about the user itself, and 1200 /// information about how the use may be satisfied. TODO: Represent multiple 1201 /// users of the same expression in common? 1202 class LSRUse { 1203 DenseSet<SmallVector<const SCEV *, 4>, UniquifierDenseMapInfo> Uniquifier; 1204 1205 public: 1206 /// An enum for a kind of use, indicating what types of scaled and immediate 1207 /// operands it might support. 1208 enum KindType { 1209 Basic, ///< A normal use, with no folding. 1210 Special, ///< A special case of basic, allowing -1 scales. 1211 Address, ///< An address use; folding according to TargetLowering 1212 ICmpZero ///< An equality icmp with both operands folded into one. 1213 // TODO: Add a generic icmp too? 1214 }; 1215 1216 typedef PointerIntPair<const SCEV *, 2, KindType> SCEVUseKindPair; 1217 1218 KindType Kind; 1219 MemAccessTy AccessTy; 1220 1221 SmallVector<int64_t, 8> Offsets; 1222 int64_t MinOffset; 1223 int64_t MaxOffset; 1224 1225 /// This records whether all of the fixups using this LSRUse are outside of 1226 /// the loop, in which case some special-case heuristics may be used. 1227 bool AllFixupsOutsideLoop; 1228 1229 /// RigidFormula is set to true to guarantee that this use will be associated 1230 /// with a single formula--the one that initially matched. Some SCEV 1231 /// expressions cannot be expanded. This allows LSR to consider the registers 1232 /// used by those expressions without the need to expand them later after 1233 /// changing the formula. 1234 bool RigidFormula; 1235 1236 /// This records the widest use type for any fixup using this 1237 /// LSRUse. FindUseWithSimilarFormula can't consider uses with different max 1238 /// fixup widths to be equivalent, because the narrower one may be relying on 1239 /// the implicit truncation to truncate away bogus bits. 1240 Type *WidestFixupType; 1241 1242 /// A list of ways to build a value that can satisfy this user. After the 1243 /// list is populated, one of these is selected heuristically and used to 1244 /// formulate a replacement for OperandValToReplace in UserInst. 1245 SmallVector<Formula, 12> Formulae; 1246 1247 /// The set of register candidates used by all formulae in this LSRUse. 1248 SmallPtrSet<const SCEV *, 4> Regs; 1249 1250 LSRUse(KindType K, MemAccessTy AT) 1251 : Kind(K), AccessTy(AT), MinOffset(INT64_MAX), MaxOffset(INT64_MIN), 1252 AllFixupsOutsideLoop(true), RigidFormula(false), 1253 WidestFixupType(nullptr) {} 1254 1255 bool HasFormulaWithSameRegs(const Formula &F) const; 1256 bool InsertFormula(const Formula &F); 1257 void DeleteFormula(Formula &F); 1258 void RecomputeRegs(size_t LUIdx, RegUseTracker &Reguses); 1259 1260 void print(raw_ostream &OS) const; 1261 void dump() const; 1262 }; 1263 1264 } 1265 1266 /// Test whether this use as a formula which has the same registers as the given 1267 /// formula. 1268 bool LSRUse::HasFormulaWithSameRegs(const Formula &F) const { 1269 SmallVector<const SCEV *, 4> Key = F.BaseRegs; 1270 if (F.ScaledReg) Key.push_back(F.ScaledReg); 1271 // Unstable sort by host order ok, because this is only used for uniquifying. 1272 std::sort(Key.begin(), Key.end()); 1273 return Uniquifier.count(Key); 1274 } 1275 1276 /// If the given formula has not yet been inserted, add it to the list, and 1277 /// return true. Return false otherwise. The formula must be in canonical form. 1278 bool LSRUse::InsertFormula(const Formula &F) { 1279 assert(F.isCanonical() && "Invalid canonical representation"); 1280 1281 if (!Formulae.empty() && RigidFormula) 1282 return false; 1283 1284 SmallVector<const SCEV *, 4> Key = F.BaseRegs; 1285 if (F.ScaledReg) Key.push_back(F.ScaledReg); 1286 // Unstable sort by host order ok, because this is only used for uniquifying. 1287 std::sort(Key.begin(), Key.end()); 1288 1289 if (!Uniquifier.insert(Key).second) 1290 return false; 1291 1292 // Using a register to hold the value of 0 is not profitable. 1293 assert((!F.ScaledReg || !F.ScaledReg->isZero()) && 1294 "Zero allocated in a scaled register!"); 1295 #ifndef NDEBUG 1296 for (const SCEV *BaseReg : F.BaseRegs) 1297 assert(!BaseReg->isZero() && "Zero allocated in a base register!"); 1298 #endif 1299 1300 // Add the formula to the list. 1301 Formulae.push_back(F); 1302 1303 // Record registers now being used by this use. 1304 Regs.insert(F.BaseRegs.begin(), F.BaseRegs.end()); 1305 if (F.ScaledReg) 1306 Regs.insert(F.ScaledReg); 1307 1308 return true; 1309 } 1310 1311 /// Remove the given formula from this use's list. 1312 void LSRUse::DeleteFormula(Formula &F) { 1313 if (&F != &Formulae.back()) 1314 std::swap(F, Formulae.back()); 1315 Formulae.pop_back(); 1316 } 1317 1318 /// Recompute the Regs field, and update RegUses. 1319 void LSRUse::RecomputeRegs(size_t LUIdx, RegUseTracker &RegUses) { 1320 // Now that we've filtered out some formulae, recompute the Regs set. 1321 SmallPtrSet<const SCEV *, 4> OldRegs = std::move(Regs); 1322 Regs.clear(); 1323 for (const Formula &F : Formulae) { 1324 if (F.ScaledReg) Regs.insert(F.ScaledReg); 1325 Regs.insert(F.BaseRegs.begin(), F.BaseRegs.end()); 1326 } 1327 1328 // Update the RegTracker. 1329 for (const SCEV *S : OldRegs) 1330 if (!Regs.count(S)) 1331 RegUses.dropRegister(S, LUIdx); 1332 } 1333 1334 void LSRUse::print(raw_ostream &OS) const { 1335 OS << "LSR Use: Kind="; 1336 switch (Kind) { 1337 case Basic: OS << "Basic"; break; 1338 case Special: OS << "Special"; break; 1339 case ICmpZero: OS << "ICmpZero"; break; 1340 case Address: 1341 OS << "Address of "; 1342 if (AccessTy.MemTy->isPointerTy()) 1343 OS << "pointer"; // the full pointer type could be really verbose 1344 else { 1345 OS << *AccessTy.MemTy; 1346 } 1347 1348 OS << " in addrspace(" << AccessTy.AddrSpace << ')'; 1349 } 1350 1351 OS << ", Offsets={"; 1352 bool NeedComma = false; 1353 for (int64_t O : Offsets) { 1354 if (NeedComma) OS << ','; 1355 OS << O; 1356 NeedComma = true; 1357 } 1358 OS << '}'; 1359 1360 if (AllFixupsOutsideLoop) 1361 OS << ", all-fixups-outside-loop"; 1362 1363 if (WidestFixupType) 1364 OS << ", widest fixup type: " << *WidestFixupType; 1365 } 1366 1367 LLVM_DUMP_METHOD 1368 void LSRUse::dump() const { 1369 print(errs()); errs() << '\n'; 1370 } 1371 1372 static bool isAMCompletelyFolded(const TargetTransformInfo &TTI, 1373 LSRUse::KindType Kind, MemAccessTy AccessTy, 1374 GlobalValue *BaseGV, int64_t BaseOffset, 1375 bool HasBaseReg, int64_t Scale) { 1376 switch (Kind) { 1377 case LSRUse::Address: 1378 return TTI.isLegalAddressingMode(AccessTy.MemTy, BaseGV, BaseOffset, 1379 HasBaseReg, Scale, AccessTy.AddrSpace); 1380 1381 case LSRUse::ICmpZero: 1382 // There's not even a target hook for querying whether it would be legal to 1383 // fold a GV into an ICmp. 1384 if (BaseGV) 1385 return false; 1386 1387 // ICmp only has two operands; don't allow more than two non-trivial parts. 1388 if (Scale != 0 && HasBaseReg && BaseOffset != 0) 1389 return false; 1390 1391 // ICmp only supports no scale or a -1 scale, as we can "fold" a -1 scale by 1392 // putting the scaled register in the other operand of the icmp. 1393 if (Scale != 0 && Scale != -1) 1394 return false; 1395 1396 // If we have low-level target information, ask the target if it can fold an 1397 // integer immediate on an icmp. 1398 if (BaseOffset != 0) { 1399 // We have one of: 1400 // ICmpZero BaseReg + BaseOffset => ICmp BaseReg, -BaseOffset 1401 // ICmpZero -1*ScaleReg + BaseOffset => ICmp ScaleReg, BaseOffset 1402 // Offs is the ICmp immediate. 1403 if (Scale == 0) 1404 // The cast does the right thing with INT64_MIN. 1405 BaseOffset = -(uint64_t)BaseOffset; 1406 return TTI.isLegalICmpImmediate(BaseOffset); 1407 } 1408 1409 // ICmpZero BaseReg + -1*ScaleReg => ICmp BaseReg, ScaleReg 1410 return true; 1411 1412 case LSRUse::Basic: 1413 // Only handle single-register values. 1414 return !BaseGV && Scale == 0 && BaseOffset == 0; 1415 1416 case LSRUse::Special: 1417 // Special case Basic to handle -1 scales. 1418 return !BaseGV && (Scale == 0 || Scale == -1) && BaseOffset == 0; 1419 } 1420 1421 llvm_unreachable("Invalid LSRUse Kind!"); 1422 } 1423 1424 static bool isAMCompletelyFolded(const TargetTransformInfo &TTI, 1425 int64_t MinOffset, int64_t MaxOffset, 1426 LSRUse::KindType Kind, MemAccessTy AccessTy, 1427 GlobalValue *BaseGV, int64_t BaseOffset, 1428 bool HasBaseReg, int64_t Scale) { 1429 // Check for overflow. 1430 if (((int64_t)((uint64_t)BaseOffset + MinOffset) > BaseOffset) != 1431 (MinOffset > 0)) 1432 return false; 1433 MinOffset = (uint64_t)BaseOffset + MinOffset; 1434 if (((int64_t)((uint64_t)BaseOffset + MaxOffset) > BaseOffset) != 1435 (MaxOffset > 0)) 1436 return false; 1437 MaxOffset = (uint64_t)BaseOffset + MaxOffset; 1438 1439 return isAMCompletelyFolded(TTI, Kind, AccessTy, BaseGV, MinOffset, 1440 HasBaseReg, Scale) && 1441 isAMCompletelyFolded(TTI, Kind, AccessTy, BaseGV, MaxOffset, 1442 HasBaseReg, Scale); 1443 } 1444 1445 static bool isAMCompletelyFolded(const TargetTransformInfo &TTI, 1446 int64_t MinOffset, int64_t MaxOffset, 1447 LSRUse::KindType Kind, MemAccessTy AccessTy, 1448 const Formula &F) { 1449 // For the purpose of isAMCompletelyFolded either having a canonical formula 1450 // or a scale not equal to zero is correct. 1451 // Problems may arise from non canonical formulae having a scale == 0. 1452 // Strictly speaking it would best to just rely on canonical formulae. 1453 // However, when we generate the scaled formulae, we first check that the 1454 // scaling factor is profitable before computing the actual ScaledReg for 1455 // compile time sake. 1456 assert((F.isCanonical() || F.Scale != 0)); 1457 return isAMCompletelyFolded(TTI, MinOffset, MaxOffset, Kind, AccessTy, 1458 F.BaseGV, F.BaseOffset, F.HasBaseReg, F.Scale); 1459 } 1460 1461 /// Test whether we know how to expand the current formula. 1462 static bool isLegalUse(const TargetTransformInfo &TTI, int64_t MinOffset, 1463 int64_t MaxOffset, LSRUse::KindType Kind, 1464 MemAccessTy AccessTy, GlobalValue *BaseGV, 1465 int64_t BaseOffset, bool HasBaseReg, int64_t Scale) { 1466 // We know how to expand completely foldable formulae. 1467 return isAMCompletelyFolded(TTI, MinOffset, MaxOffset, Kind, AccessTy, BaseGV, 1468 BaseOffset, HasBaseReg, Scale) || 1469 // Or formulae that use a base register produced by a sum of base 1470 // registers. 1471 (Scale == 1 && 1472 isAMCompletelyFolded(TTI, MinOffset, MaxOffset, Kind, AccessTy, 1473 BaseGV, BaseOffset, true, 0)); 1474 } 1475 1476 static bool isLegalUse(const TargetTransformInfo &TTI, int64_t MinOffset, 1477 int64_t MaxOffset, LSRUse::KindType Kind, 1478 MemAccessTy AccessTy, const Formula &F) { 1479 return isLegalUse(TTI, MinOffset, MaxOffset, Kind, AccessTy, F.BaseGV, 1480 F.BaseOffset, F.HasBaseReg, F.Scale); 1481 } 1482 1483 static bool isAMCompletelyFolded(const TargetTransformInfo &TTI, 1484 const LSRUse &LU, const Formula &F) { 1485 return isAMCompletelyFolded(TTI, LU.MinOffset, LU.MaxOffset, LU.Kind, 1486 LU.AccessTy, F.BaseGV, F.BaseOffset, F.HasBaseReg, 1487 F.Scale); 1488 } 1489 1490 static unsigned getScalingFactorCost(const TargetTransformInfo &TTI, 1491 const LSRUse &LU, const Formula &F) { 1492 if (!F.Scale) 1493 return 0; 1494 1495 // If the use is not completely folded in that instruction, we will have to 1496 // pay an extra cost only for scale != 1. 1497 if (!isAMCompletelyFolded(TTI, LU.MinOffset, LU.MaxOffset, LU.Kind, 1498 LU.AccessTy, F)) 1499 return F.Scale != 1; 1500 1501 switch (LU.Kind) { 1502 case LSRUse::Address: { 1503 // Check the scaling factor cost with both the min and max offsets. 1504 int ScaleCostMinOffset = TTI.getScalingFactorCost( 1505 LU.AccessTy.MemTy, F.BaseGV, F.BaseOffset + LU.MinOffset, F.HasBaseReg, 1506 F.Scale, LU.AccessTy.AddrSpace); 1507 int ScaleCostMaxOffset = TTI.getScalingFactorCost( 1508 LU.AccessTy.MemTy, F.BaseGV, F.BaseOffset + LU.MaxOffset, F.HasBaseReg, 1509 F.Scale, LU.AccessTy.AddrSpace); 1510 1511 assert(ScaleCostMinOffset >= 0 && ScaleCostMaxOffset >= 0 && 1512 "Legal addressing mode has an illegal cost!"); 1513 return std::max(ScaleCostMinOffset, ScaleCostMaxOffset); 1514 } 1515 case LSRUse::ICmpZero: 1516 case LSRUse::Basic: 1517 case LSRUse::Special: 1518 // The use is completely folded, i.e., everything is folded into the 1519 // instruction. 1520 return 0; 1521 } 1522 1523 llvm_unreachable("Invalid LSRUse Kind!"); 1524 } 1525 1526 static bool isAlwaysFoldable(const TargetTransformInfo &TTI, 1527 LSRUse::KindType Kind, MemAccessTy AccessTy, 1528 GlobalValue *BaseGV, int64_t BaseOffset, 1529 bool HasBaseReg) { 1530 // Fast-path: zero is always foldable. 1531 if (BaseOffset == 0 && !BaseGV) return true; 1532 1533 // Conservatively, create an address with an immediate and a 1534 // base and a scale. 1535 int64_t Scale = Kind == LSRUse::ICmpZero ? -1 : 1; 1536 1537 // Canonicalize a scale of 1 to a base register if the formula doesn't 1538 // already have a base register. 1539 if (!HasBaseReg && Scale == 1) { 1540 Scale = 0; 1541 HasBaseReg = true; 1542 } 1543 1544 return isAMCompletelyFolded(TTI, Kind, AccessTy, BaseGV, BaseOffset, 1545 HasBaseReg, Scale); 1546 } 1547 1548 static bool isAlwaysFoldable(const TargetTransformInfo &TTI, 1549 ScalarEvolution &SE, int64_t MinOffset, 1550 int64_t MaxOffset, LSRUse::KindType Kind, 1551 MemAccessTy AccessTy, const SCEV *S, 1552 bool HasBaseReg) { 1553 // Fast-path: zero is always foldable. 1554 if (S->isZero()) return true; 1555 1556 // Conservatively, create an address with an immediate and a 1557 // base and a scale. 1558 int64_t BaseOffset = ExtractImmediate(S, SE); 1559 GlobalValue *BaseGV = ExtractSymbol(S, SE); 1560 1561 // If there's anything else involved, it's not foldable. 1562 if (!S->isZero()) return false; 1563 1564 // Fast-path: zero is always foldable. 1565 if (BaseOffset == 0 && !BaseGV) return true; 1566 1567 // Conservatively, create an address with an immediate and a 1568 // base and a scale. 1569 int64_t Scale = Kind == LSRUse::ICmpZero ? -1 : 1; 1570 1571 return isAMCompletelyFolded(TTI, MinOffset, MaxOffset, Kind, AccessTy, BaseGV, 1572 BaseOffset, HasBaseReg, Scale); 1573 } 1574 1575 namespace { 1576 1577 /// An individual increment in a Chain of IV increments. Relate an IV user to 1578 /// an expression that computes the IV it uses from the IV used by the previous 1579 /// link in the Chain. 1580 /// 1581 /// For the head of a chain, IncExpr holds the absolute SCEV expression for the 1582 /// original IVOperand. The head of the chain's IVOperand is only valid during 1583 /// chain collection, before LSR replaces IV users. During chain generation, 1584 /// IncExpr can be used to find the new IVOperand that computes the same 1585 /// expression. 1586 struct IVInc { 1587 Instruction *UserInst; 1588 Value* IVOperand; 1589 const SCEV *IncExpr; 1590 1591 IVInc(Instruction *U, Value *O, const SCEV *E): 1592 UserInst(U), IVOperand(O), IncExpr(E) {} 1593 }; 1594 1595 // The list of IV increments in program order. We typically add the head of a 1596 // chain without finding subsequent links. 1597 struct IVChain { 1598 SmallVector<IVInc,1> Incs; 1599 const SCEV *ExprBase; 1600 1601 IVChain() : ExprBase(nullptr) {} 1602 1603 IVChain(const IVInc &Head, const SCEV *Base) 1604 : Incs(1, Head), ExprBase(Base) {} 1605 1606 typedef SmallVectorImpl<IVInc>::const_iterator const_iterator; 1607 1608 // Return the first increment in the chain. 1609 const_iterator begin() const { 1610 assert(!Incs.empty()); 1611 return std::next(Incs.begin()); 1612 } 1613 const_iterator end() const { 1614 return Incs.end(); 1615 } 1616 1617 // Returns true if this chain contains any increments. 1618 bool hasIncs() const { return Incs.size() >= 2; } 1619 1620 // Add an IVInc to the end of this chain. 1621 void add(const IVInc &X) { Incs.push_back(X); } 1622 1623 // Returns the last UserInst in the chain. 1624 Instruction *tailUserInst() const { return Incs.back().UserInst; } 1625 1626 // Returns true if IncExpr can be profitably added to this chain. 1627 bool isProfitableIncrement(const SCEV *OperExpr, 1628 const SCEV *IncExpr, 1629 ScalarEvolution&); 1630 }; 1631 1632 /// Helper for CollectChains to track multiple IV increment uses. Distinguish 1633 /// between FarUsers that definitely cross IV increments and NearUsers that may 1634 /// be used between IV increments. 1635 struct ChainUsers { 1636 SmallPtrSet<Instruction*, 4> FarUsers; 1637 SmallPtrSet<Instruction*, 4> NearUsers; 1638 }; 1639 1640 /// This class holds state for the main loop strength reduction logic. 1641 class LSRInstance { 1642 IVUsers &IU; 1643 ScalarEvolution &SE; 1644 DominatorTree &DT; 1645 LoopInfo &LI; 1646 const TargetTransformInfo &TTI; 1647 Loop *const L; 1648 bool Changed; 1649 1650 /// This is the insert position that the current loop's induction variable 1651 /// increment should be placed. In simple loops, this is the latch block's 1652 /// terminator. But in more complicated cases, this is a position which will 1653 /// dominate all the in-loop post-increment users. 1654 Instruction *IVIncInsertPos; 1655 1656 /// Interesting factors between use strides. 1657 SmallSetVector<int64_t, 8> Factors; 1658 1659 /// Interesting use types, to facilitate truncation reuse. 1660 SmallSetVector<Type *, 4> Types; 1661 1662 /// The list of operands which are to be replaced. 1663 SmallVector<LSRFixup, 16> Fixups; 1664 1665 /// The list of interesting uses. 1666 SmallVector<LSRUse, 16> Uses; 1667 1668 /// Track which uses use which register candidates. 1669 RegUseTracker RegUses; 1670 1671 // Limit the number of chains to avoid quadratic behavior. We don't expect to 1672 // have more than a few IV increment chains in a loop. Missing a Chain falls 1673 // back to normal LSR behavior for those uses. 1674 static const unsigned MaxChains = 8; 1675 1676 /// IV users can form a chain of IV increments. 1677 SmallVector<IVChain, MaxChains> IVChainVec; 1678 1679 /// IV users that belong to profitable IVChains. 1680 SmallPtrSet<Use*, MaxChains> IVIncSet; 1681 1682 void OptimizeShadowIV(); 1683 bool FindIVUserForCond(ICmpInst *Cond, IVStrideUse *&CondUse); 1684 ICmpInst *OptimizeMax(ICmpInst *Cond, IVStrideUse* &CondUse); 1685 void OptimizeLoopTermCond(); 1686 1687 void ChainInstruction(Instruction *UserInst, Instruction *IVOper, 1688 SmallVectorImpl<ChainUsers> &ChainUsersVec); 1689 void FinalizeChain(IVChain &Chain); 1690 void CollectChains(); 1691 void GenerateIVChain(const IVChain &Chain, SCEVExpander &Rewriter, 1692 SmallVectorImpl<WeakVH> &DeadInsts); 1693 1694 void CollectInterestingTypesAndFactors(); 1695 void CollectFixupsAndInitialFormulae(); 1696 1697 LSRFixup &getNewFixup() { 1698 Fixups.push_back(LSRFixup()); 1699 return Fixups.back(); 1700 } 1701 1702 // Support for sharing of LSRUses between LSRFixups. 1703 typedef DenseMap<LSRUse::SCEVUseKindPair, size_t> UseMapTy; 1704 UseMapTy UseMap; 1705 1706 bool reconcileNewOffset(LSRUse &LU, int64_t NewOffset, bool HasBaseReg, 1707 LSRUse::KindType Kind, MemAccessTy AccessTy); 1708 1709 std::pair<size_t, int64_t> getUse(const SCEV *&Expr, LSRUse::KindType Kind, 1710 MemAccessTy AccessTy); 1711 1712 void DeleteUse(LSRUse &LU, size_t LUIdx); 1713 1714 LSRUse *FindUseWithSimilarFormula(const Formula &F, const LSRUse &OrigLU); 1715 1716 void InsertInitialFormula(const SCEV *S, LSRUse &LU, size_t LUIdx); 1717 void InsertSupplementalFormula(const SCEV *S, LSRUse &LU, size_t LUIdx); 1718 void CountRegisters(const Formula &F, size_t LUIdx); 1719 bool InsertFormula(LSRUse &LU, unsigned LUIdx, const Formula &F); 1720 1721 void CollectLoopInvariantFixupsAndFormulae(); 1722 1723 void GenerateReassociations(LSRUse &LU, unsigned LUIdx, Formula Base, 1724 unsigned Depth = 0); 1725 1726 void GenerateReassociationsImpl(LSRUse &LU, unsigned LUIdx, 1727 const Formula &Base, unsigned Depth, 1728 size_t Idx, bool IsScaledReg = false); 1729 void GenerateCombinations(LSRUse &LU, unsigned LUIdx, Formula Base); 1730 void GenerateSymbolicOffsetsImpl(LSRUse &LU, unsigned LUIdx, 1731 const Formula &Base, size_t Idx, 1732 bool IsScaledReg = false); 1733 void GenerateSymbolicOffsets(LSRUse &LU, unsigned LUIdx, Formula Base); 1734 void GenerateConstantOffsetsImpl(LSRUse &LU, unsigned LUIdx, 1735 const Formula &Base, 1736 const SmallVectorImpl<int64_t> &Worklist, 1737 size_t Idx, bool IsScaledReg = false); 1738 void GenerateConstantOffsets(LSRUse &LU, unsigned LUIdx, Formula Base); 1739 void GenerateICmpZeroScales(LSRUse &LU, unsigned LUIdx, Formula Base); 1740 void GenerateScales(LSRUse &LU, unsigned LUIdx, Formula Base); 1741 void GenerateTruncates(LSRUse &LU, unsigned LUIdx, Formula Base); 1742 void GenerateCrossUseConstantOffsets(); 1743 void GenerateAllReuseFormulae(); 1744 1745 void FilterOutUndesirableDedicatedRegisters(); 1746 1747 size_t EstimateSearchSpaceComplexity() const; 1748 void NarrowSearchSpaceByDetectingSupersets(); 1749 void NarrowSearchSpaceByCollapsingUnrolledCode(); 1750 void NarrowSearchSpaceByRefilteringUndesirableDedicatedRegisters(); 1751 void NarrowSearchSpaceByPickingWinnerRegs(); 1752 void NarrowSearchSpaceUsingHeuristics(); 1753 1754 void SolveRecurse(SmallVectorImpl<const Formula *> &Solution, 1755 Cost &SolutionCost, 1756 SmallVectorImpl<const Formula *> &Workspace, 1757 const Cost &CurCost, 1758 const SmallPtrSet<const SCEV *, 16> &CurRegs, 1759 DenseSet<const SCEV *> &VisitedRegs) const; 1760 void Solve(SmallVectorImpl<const Formula *> &Solution) const; 1761 1762 BasicBlock::iterator 1763 HoistInsertPosition(BasicBlock::iterator IP, 1764 const SmallVectorImpl<Instruction *> &Inputs) const; 1765 BasicBlock::iterator 1766 AdjustInsertPositionForExpand(BasicBlock::iterator IP, 1767 const LSRFixup &LF, 1768 const LSRUse &LU, 1769 SCEVExpander &Rewriter) const; 1770 1771 Value *Expand(const LSRFixup &LF, 1772 const Formula &F, 1773 BasicBlock::iterator IP, 1774 SCEVExpander &Rewriter, 1775 SmallVectorImpl<WeakVH> &DeadInsts) const; 1776 void RewriteForPHI(PHINode *PN, const LSRFixup &LF, 1777 const Formula &F, 1778 SCEVExpander &Rewriter, 1779 SmallVectorImpl<WeakVH> &DeadInsts) const; 1780 void Rewrite(const LSRFixup &LF, 1781 const Formula &F, 1782 SCEVExpander &Rewriter, 1783 SmallVectorImpl<WeakVH> &DeadInsts) const; 1784 void ImplementSolution(const SmallVectorImpl<const Formula *> &Solution); 1785 1786 public: 1787 LSRInstance(Loop *L, IVUsers &IU, ScalarEvolution &SE, DominatorTree &DT, 1788 LoopInfo &LI, const TargetTransformInfo &TTI); 1789 1790 bool getChanged() const { return Changed; } 1791 1792 void print_factors_and_types(raw_ostream &OS) const; 1793 void print_fixups(raw_ostream &OS) const; 1794 void print_uses(raw_ostream &OS) const; 1795 void print(raw_ostream &OS) const; 1796 void dump() const; 1797 }; 1798 1799 } 1800 1801 /// If IV is used in a int-to-float cast inside the loop then try to eliminate 1802 /// the cast operation. 1803 void LSRInstance::OptimizeShadowIV() { 1804 const SCEV *BackedgeTakenCount = SE.getBackedgeTakenCount(L); 1805 if (isa<SCEVCouldNotCompute>(BackedgeTakenCount)) 1806 return; 1807 1808 for (IVUsers::const_iterator UI = IU.begin(), E = IU.end(); 1809 UI != E; /* empty */) { 1810 IVUsers::const_iterator CandidateUI = UI; 1811 ++UI; 1812 Instruction *ShadowUse = CandidateUI->getUser(); 1813 Type *DestTy = nullptr; 1814 bool IsSigned = false; 1815 1816 /* If shadow use is a int->float cast then insert a second IV 1817 to eliminate this cast. 1818 1819 for (unsigned i = 0; i < n; ++i) 1820 foo((double)i); 1821 1822 is transformed into 1823 1824 double d = 0.0; 1825 for (unsigned i = 0; i < n; ++i, ++d) 1826 foo(d); 1827 */ 1828 if (UIToFPInst *UCast = dyn_cast<UIToFPInst>(CandidateUI->getUser())) { 1829 IsSigned = false; 1830 DestTy = UCast->getDestTy(); 1831 } 1832 else if (SIToFPInst *SCast = dyn_cast<SIToFPInst>(CandidateUI->getUser())) { 1833 IsSigned = true; 1834 DestTy = SCast->getDestTy(); 1835 } 1836 if (!DestTy) continue; 1837 1838 // If target does not support DestTy natively then do not apply 1839 // this transformation. 1840 if (!TTI.isTypeLegal(DestTy)) continue; 1841 1842 PHINode *PH = dyn_cast<PHINode>(ShadowUse->getOperand(0)); 1843 if (!PH) continue; 1844 if (PH->getNumIncomingValues() != 2) continue; 1845 1846 Type *SrcTy = PH->getType(); 1847 int Mantissa = DestTy->getFPMantissaWidth(); 1848 if (Mantissa == -1) continue; 1849 if ((int)SE.getTypeSizeInBits(SrcTy) > Mantissa) 1850 continue; 1851 1852 unsigned Entry, Latch; 1853 if (PH->getIncomingBlock(0) == L->getLoopPreheader()) { 1854 Entry = 0; 1855 Latch = 1; 1856 } else { 1857 Entry = 1; 1858 Latch = 0; 1859 } 1860 1861 ConstantInt *Init = dyn_cast<ConstantInt>(PH->getIncomingValue(Entry)); 1862 if (!Init) continue; 1863 Constant *NewInit = ConstantFP::get(DestTy, IsSigned ? 1864 (double)Init->getSExtValue() : 1865 (double)Init->getZExtValue()); 1866 1867 BinaryOperator *Incr = 1868 dyn_cast<BinaryOperator>(PH->getIncomingValue(Latch)); 1869 if (!Incr) continue; 1870 if (Incr->getOpcode() != Instruction::Add 1871 && Incr->getOpcode() != Instruction::Sub) 1872 continue; 1873 1874 /* Initialize new IV, double d = 0.0 in above example. */ 1875 ConstantInt *C = nullptr; 1876 if (Incr->getOperand(0) == PH) 1877 C = dyn_cast<ConstantInt>(Incr->getOperand(1)); 1878 else if (Incr->getOperand(1) == PH) 1879 C = dyn_cast<ConstantInt>(Incr->getOperand(0)); 1880 else 1881 continue; 1882 1883 if (!C) continue; 1884 1885 // Ignore negative constants, as the code below doesn't handle them 1886 // correctly. TODO: Remove this restriction. 1887 if (!C->getValue().isStrictlyPositive()) continue; 1888 1889 /* Add new PHINode. */ 1890 PHINode *NewPH = PHINode::Create(DestTy, 2, "IV.S.", PH); 1891 1892 /* create new increment. '++d' in above example. */ 1893 Constant *CFP = ConstantFP::get(DestTy, C->getZExtValue()); 1894 BinaryOperator *NewIncr = 1895 BinaryOperator::Create(Incr->getOpcode() == Instruction::Add ? 1896 Instruction::FAdd : Instruction::FSub, 1897 NewPH, CFP, "IV.S.next.", Incr); 1898 1899 NewPH->addIncoming(NewInit, PH->getIncomingBlock(Entry)); 1900 NewPH->addIncoming(NewIncr, PH->getIncomingBlock(Latch)); 1901 1902 /* Remove cast operation */ 1903 ShadowUse->replaceAllUsesWith(NewPH); 1904 ShadowUse->eraseFromParent(); 1905 Changed = true; 1906 break; 1907 } 1908 } 1909 1910 /// If Cond has an operand that is an expression of an IV, set the IV user and 1911 /// stride information and return true, otherwise return false. 1912 bool LSRInstance::FindIVUserForCond(ICmpInst *Cond, IVStrideUse *&CondUse) { 1913 for (IVStrideUse &U : IU) 1914 if (U.getUser() == Cond) { 1915 // NOTE: we could handle setcc instructions with multiple uses here, but 1916 // InstCombine does it as well for simple uses, it's not clear that it 1917 // occurs enough in real life to handle. 1918 CondUse = &U; 1919 return true; 1920 } 1921 return false; 1922 } 1923 1924 /// Rewrite the loop's terminating condition if it uses a max computation. 1925 /// 1926 /// This is a narrow solution to a specific, but acute, problem. For loops 1927 /// like this: 1928 /// 1929 /// i = 0; 1930 /// do { 1931 /// p[i] = 0.0; 1932 /// } while (++i < n); 1933 /// 1934 /// the trip count isn't just 'n', because 'n' might not be positive. And 1935 /// unfortunately this can come up even for loops where the user didn't use 1936 /// a C do-while loop. For example, seemingly well-behaved top-test loops 1937 /// will commonly be lowered like this: 1938 // 1939 /// if (n > 0) { 1940 /// i = 0; 1941 /// do { 1942 /// p[i] = 0.0; 1943 /// } while (++i < n); 1944 /// } 1945 /// 1946 /// and then it's possible for subsequent optimization to obscure the if 1947 /// test in such a way that indvars can't find it. 1948 /// 1949 /// When indvars can't find the if test in loops like this, it creates a 1950 /// max expression, which allows it to give the loop a canonical 1951 /// induction variable: 1952 /// 1953 /// i = 0; 1954 /// max = n < 1 ? 1 : n; 1955 /// do { 1956 /// p[i] = 0.0; 1957 /// } while (++i != max); 1958 /// 1959 /// Canonical induction variables are necessary because the loop passes 1960 /// are designed around them. The most obvious example of this is the 1961 /// LoopInfo analysis, which doesn't remember trip count values. It 1962 /// expects to be able to rediscover the trip count each time it is 1963 /// needed, and it does this using a simple analysis that only succeeds if 1964 /// the loop has a canonical induction variable. 1965 /// 1966 /// However, when it comes time to generate code, the maximum operation 1967 /// can be quite costly, especially if it's inside of an outer loop. 1968 /// 1969 /// This function solves this problem by detecting this type of loop and 1970 /// rewriting their conditions from ICMP_NE back to ICMP_SLT, and deleting 1971 /// the instructions for the maximum computation. 1972 /// 1973 ICmpInst *LSRInstance::OptimizeMax(ICmpInst *Cond, IVStrideUse* &CondUse) { 1974 // Check that the loop matches the pattern we're looking for. 1975 if (Cond->getPredicate() != CmpInst::ICMP_EQ && 1976 Cond->getPredicate() != CmpInst::ICMP_NE) 1977 return Cond; 1978 1979 SelectInst *Sel = dyn_cast<SelectInst>(Cond->getOperand(1)); 1980 if (!Sel || !Sel->hasOneUse()) return Cond; 1981 1982 const SCEV *BackedgeTakenCount = SE.getBackedgeTakenCount(L); 1983 if (isa<SCEVCouldNotCompute>(BackedgeTakenCount)) 1984 return Cond; 1985 const SCEV *One = SE.getConstant(BackedgeTakenCount->getType(), 1); 1986 1987 // Add one to the backedge-taken count to get the trip count. 1988 const SCEV *IterationCount = SE.getAddExpr(One, BackedgeTakenCount); 1989 if (IterationCount != SE.getSCEV(Sel)) return Cond; 1990 1991 // Check for a max calculation that matches the pattern. There's no check 1992 // for ICMP_ULE here because the comparison would be with zero, which 1993 // isn't interesting. 1994 CmpInst::Predicate Pred = ICmpInst::BAD_ICMP_PREDICATE; 1995 const SCEVNAryExpr *Max = nullptr; 1996 if (const SCEVSMaxExpr *S = dyn_cast<SCEVSMaxExpr>(BackedgeTakenCount)) { 1997 Pred = ICmpInst::ICMP_SLE; 1998 Max = S; 1999 } else if (const SCEVSMaxExpr *S = dyn_cast<SCEVSMaxExpr>(IterationCount)) { 2000 Pred = ICmpInst::ICMP_SLT; 2001 Max = S; 2002 } else if (const SCEVUMaxExpr *U = dyn_cast<SCEVUMaxExpr>(IterationCount)) { 2003 Pred = ICmpInst::ICMP_ULT; 2004 Max = U; 2005 } else { 2006 // No match; bail. 2007 return Cond; 2008 } 2009 2010 // To handle a max with more than two operands, this optimization would 2011 // require additional checking and setup. 2012 if (Max->getNumOperands() != 2) 2013 return Cond; 2014 2015 const SCEV *MaxLHS = Max->getOperand(0); 2016 const SCEV *MaxRHS = Max->getOperand(1); 2017 2018 // ScalarEvolution canonicalizes constants to the left. For < and >, look 2019 // for a comparison with 1. For <= and >=, a comparison with zero. 2020 if (!MaxLHS || 2021 (ICmpInst::isTrueWhenEqual(Pred) ? !MaxLHS->isZero() : (MaxLHS != One))) 2022 return Cond; 2023 2024 // Check the relevant induction variable for conformance to 2025 // the pattern. 2026 const SCEV *IV = SE.getSCEV(Cond->getOperand(0)); 2027 const SCEVAddRecExpr *AR = dyn_cast<SCEVAddRecExpr>(IV); 2028 if (!AR || !AR->isAffine() || 2029 AR->getStart() != One || 2030 AR->getStepRecurrence(SE) != One) 2031 return Cond; 2032 2033 assert(AR->getLoop() == L && 2034 "Loop condition operand is an addrec in a different loop!"); 2035 2036 // Check the right operand of the select, and remember it, as it will 2037 // be used in the new comparison instruction. 2038 Value *NewRHS = nullptr; 2039 if (ICmpInst::isTrueWhenEqual(Pred)) { 2040 // Look for n+1, and grab n. 2041 if (AddOperator *BO = dyn_cast<AddOperator>(Sel->getOperand(1))) 2042 if (ConstantInt *BO1 = dyn_cast<ConstantInt>(BO->getOperand(1))) 2043 if (BO1->isOne() && SE.getSCEV(BO->getOperand(0)) == MaxRHS) 2044 NewRHS = BO->getOperand(0); 2045 if (AddOperator *BO = dyn_cast<AddOperator>(Sel->getOperand(2))) 2046 if (ConstantInt *BO1 = dyn_cast<ConstantInt>(BO->getOperand(1))) 2047 if (BO1->isOne() && SE.getSCEV(BO->getOperand(0)) == MaxRHS) 2048 NewRHS = BO->getOperand(0); 2049 if (!NewRHS) 2050 return Cond; 2051 } else if (SE.getSCEV(Sel->getOperand(1)) == MaxRHS) 2052 NewRHS = Sel->getOperand(1); 2053 else if (SE.getSCEV(Sel->getOperand(2)) == MaxRHS) 2054 NewRHS = Sel->getOperand(2); 2055 else if (const SCEVUnknown *SU = dyn_cast<SCEVUnknown>(MaxRHS)) 2056 NewRHS = SU->getValue(); 2057 else 2058 // Max doesn't match expected pattern. 2059 return Cond; 2060 2061 // Determine the new comparison opcode. It may be signed or unsigned, 2062 // and the original comparison may be either equality or inequality. 2063 if (Cond->getPredicate() == CmpInst::ICMP_EQ) 2064 Pred = CmpInst::getInversePredicate(Pred); 2065 2066 // Ok, everything looks ok to change the condition into an SLT or SGE and 2067 // delete the max calculation. 2068 ICmpInst *NewCond = 2069 new ICmpInst(Cond, Pred, Cond->getOperand(0), NewRHS, "scmp"); 2070 2071 // Delete the max calculation instructions. 2072 Cond->replaceAllUsesWith(NewCond); 2073 CondUse->setUser(NewCond); 2074 Instruction *Cmp = cast<Instruction>(Sel->getOperand(0)); 2075 Cond->eraseFromParent(); 2076 Sel->eraseFromParent(); 2077 if (Cmp->use_empty()) 2078 Cmp->eraseFromParent(); 2079 return NewCond; 2080 } 2081 2082 /// Change loop terminating condition to use the postinc iv when possible. 2083 void 2084 LSRInstance::OptimizeLoopTermCond() { 2085 SmallPtrSet<Instruction *, 4> PostIncs; 2086 2087 BasicBlock *LatchBlock = L->getLoopLatch(); 2088 SmallVector<BasicBlock*, 8> ExitingBlocks; 2089 L->getExitingBlocks(ExitingBlocks); 2090 2091 for (BasicBlock *ExitingBlock : ExitingBlocks) { 2092 2093 // Get the terminating condition for the loop if possible. If we 2094 // can, we want to change it to use a post-incremented version of its 2095 // induction variable, to allow coalescing the live ranges for the IV into 2096 // one register value. 2097 2098 BranchInst *TermBr = dyn_cast<BranchInst>(ExitingBlock->getTerminator()); 2099 if (!TermBr) 2100 continue; 2101 // FIXME: Overly conservative, termination condition could be an 'or' etc.. 2102 if (TermBr->isUnconditional() || !isa<ICmpInst>(TermBr->getCondition())) 2103 continue; 2104 2105 // Search IVUsesByStride to find Cond's IVUse if there is one. 2106 IVStrideUse *CondUse = nullptr; 2107 ICmpInst *Cond = cast<ICmpInst>(TermBr->getCondition()); 2108 if (!FindIVUserForCond(Cond, CondUse)) 2109 continue; 2110 2111 // If the trip count is computed in terms of a max (due to ScalarEvolution 2112 // being unable to find a sufficient guard, for example), change the loop 2113 // comparison to use SLT or ULT instead of NE. 2114 // One consequence of doing this now is that it disrupts the count-down 2115 // optimization. That's not always a bad thing though, because in such 2116 // cases it may still be worthwhile to avoid a max. 2117 Cond = OptimizeMax(Cond, CondUse); 2118 2119 // If this exiting block dominates the latch block, it may also use 2120 // the post-inc value if it won't be shared with other uses. 2121 // Check for dominance. 2122 if (!DT.dominates(ExitingBlock, LatchBlock)) 2123 continue; 2124 2125 // Conservatively avoid trying to use the post-inc value in non-latch 2126 // exits if there may be pre-inc users in intervening blocks. 2127 if (LatchBlock != ExitingBlock) 2128 for (IVUsers::const_iterator UI = IU.begin(), E = IU.end(); UI != E; ++UI) 2129 // Test if the use is reachable from the exiting block. This dominator 2130 // query is a conservative approximation of reachability. 2131 if (&*UI != CondUse && 2132 !DT.properlyDominates(UI->getUser()->getParent(), ExitingBlock)) { 2133 // Conservatively assume there may be reuse if the quotient of their 2134 // strides could be a legal scale. 2135 const SCEV *A = IU.getStride(*CondUse, L); 2136 const SCEV *B = IU.getStride(*UI, L); 2137 if (!A || !B) continue; 2138 if (SE.getTypeSizeInBits(A->getType()) != 2139 SE.getTypeSizeInBits(B->getType())) { 2140 if (SE.getTypeSizeInBits(A->getType()) > 2141 SE.getTypeSizeInBits(B->getType())) 2142 B = SE.getSignExtendExpr(B, A->getType()); 2143 else 2144 A = SE.getSignExtendExpr(A, B->getType()); 2145 } 2146 if (const SCEVConstant *D = 2147 dyn_cast_or_null<SCEVConstant>(getExactSDiv(B, A, SE))) { 2148 const ConstantInt *C = D->getValue(); 2149 // Stride of one or negative one can have reuse with non-addresses. 2150 if (C->isOne() || C->isAllOnesValue()) 2151 goto decline_post_inc; 2152 // Avoid weird situations. 2153 if (C->getValue().getMinSignedBits() >= 64 || 2154 C->getValue().isMinSignedValue()) 2155 goto decline_post_inc; 2156 // Check for possible scaled-address reuse. 2157 MemAccessTy AccessTy = getAccessType(UI->getUser()); 2158 int64_t Scale = C->getSExtValue(); 2159 if (TTI.isLegalAddressingMode(AccessTy.MemTy, /*BaseGV=*/nullptr, 2160 /*BaseOffset=*/0, 2161 /*HasBaseReg=*/false, Scale, 2162 AccessTy.AddrSpace)) 2163 goto decline_post_inc; 2164 Scale = -Scale; 2165 if (TTI.isLegalAddressingMode(AccessTy.MemTy, /*BaseGV=*/nullptr, 2166 /*BaseOffset=*/0, 2167 /*HasBaseReg=*/false, Scale, 2168 AccessTy.AddrSpace)) 2169 goto decline_post_inc; 2170 } 2171 } 2172 2173 DEBUG(dbgs() << " Change loop exiting icmp to use postinc iv: " 2174 << *Cond << '\n'); 2175 2176 // It's possible for the setcc instruction to be anywhere in the loop, and 2177 // possible for it to have multiple users. If it is not immediately before 2178 // the exiting block branch, move it. 2179 if (&*++BasicBlock::iterator(Cond) != TermBr) { 2180 if (Cond->hasOneUse()) { 2181 Cond->moveBefore(TermBr); 2182 } else { 2183 // Clone the terminating condition and insert into the loopend. 2184 ICmpInst *OldCond = Cond; 2185 Cond = cast<ICmpInst>(Cond->clone()); 2186 Cond->setName(L->getHeader()->getName() + ".termcond"); 2187 ExitingBlock->getInstList().insert(TermBr->getIterator(), Cond); 2188 2189 // Clone the IVUse, as the old use still exists! 2190 CondUse = &IU.AddUser(Cond, CondUse->getOperandValToReplace()); 2191 TermBr->replaceUsesOfWith(OldCond, Cond); 2192 } 2193 } 2194 2195 // If we get to here, we know that we can transform the setcc instruction to 2196 // use the post-incremented version of the IV, allowing us to coalesce the 2197 // live ranges for the IV correctly. 2198 CondUse->transformToPostInc(L); 2199 Changed = true; 2200 2201 PostIncs.insert(Cond); 2202 decline_post_inc:; 2203 } 2204 2205 // Determine an insertion point for the loop induction variable increment. It 2206 // must dominate all the post-inc comparisons we just set up, and it must 2207 // dominate the loop latch edge. 2208 IVIncInsertPos = L->getLoopLatch()->getTerminator(); 2209 for (Instruction *Inst : PostIncs) { 2210 BasicBlock *BB = 2211 DT.findNearestCommonDominator(IVIncInsertPos->getParent(), 2212 Inst->getParent()); 2213 if (BB == Inst->getParent()) 2214 IVIncInsertPos = Inst; 2215 else if (BB != IVIncInsertPos->getParent()) 2216 IVIncInsertPos = BB->getTerminator(); 2217 } 2218 } 2219 2220 /// Determine if the given use can accommodate a fixup at the given offset and 2221 /// other details. If so, update the use and return true. 2222 bool LSRInstance::reconcileNewOffset(LSRUse &LU, int64_t NewOffset, 2223 bool HasBaseReg, LSRUse::KindType Kind, 2224 MemAccessTy AccessTy) { 2225 int64_t NewMinOffset = LU.MinOffset; 2226 int64_t NewMaxOffset = LU.MaxOffset; 2227 MemAccessTy NewAccessTy = AccessTy; 2228 2229 // Check for a mismatched kind. It's tempting to collapse mismatched kinds to 2230 // something conservative, however this can pessimize in the case that one of 2231 // the uses will have all its uses outside the loop, for example. 2232 if (LU.Kind != Kind) 2233 return false; 2234 2235 // Check for a mismatched access type, and fall back conservatively as needed. 2236 // TODO: Be less conservative when the type is similar and can use the same 2237 // addressing modes. 2238 if (Kind == LSRUse::Address) { 2239 if (AccessTy != LU.AccessTy) 2240 NewAccessTy = MemAccessTy::getUnknown(AccessTy.MemTy->getContext()); 2241 } 2242 2243 // Conservatively assume HasBaseReg is true for now. 2244 if (NewOffset < LU.MinOffset) { 2245 if (!isAlwaysFoldable(TTI, Kind, NewAccessTy, /*BaseGV=*/nullptr, 2246 LU.MaxOffset - NewOffset, HasBaseReg)) 2247 return false; 2248 NewMinOffset = NewOffset; 2249 } else if (NewOffset > LU.MaxOffset) { 2250 if (!isAlwaysFoldable(TTI, Kind, NewAccessTy, /*BaseGV=*/nullptr, 2251 NewOffset - LU.MinOffset, HasBaseReg)) 2252 return false; 2253 NewMaxOffset = NewOffset; 2254 } 2255 2256 // Update the use. 2257 LU.MinOffset = NewMinOffset; 2258 LU.MaxOffset = NewMaxOffset; 2259 LU.AccessTy = NewAccessTy; 2260 if (NewOffset != LU.Offsets.back()) 2261 LU.Offsets.push_back(NewOffset); 2262 return true; 2263 } 2264 2265 /// Return an LSRUse index and an offset value for a fixup which needs the given 2266 /// expression, with the given kind and optional access type. Either reuse an 2267 /// existing use or create a new one, as needed. 2268 std::pair<size_t, int64_t> LSRInstance::getUse(const SCEV *&Expr, 2269 LSRUse::KindType Kind, 2270 MemAccessTy AccessTy) { 2271 const SCEV *Copy = Expr; 2272 int64_t Offset = ExtractImmediate(Expr, SE); 2273 2274 // Basic uses can't accept any offset, for example. 2275 if (!isAlwaysFoldable(TTI, Kind, AccessTy, /*BaseGV=*/ nullptr, 2276 Offset, /*HasBaseReg=*/ true)) { 2277 Expr = Copy; 2278 Offset = 0; 2279 } 2280 2281 std::pair<UseMapTy::iterator, bool> P = 2282 UseMap.insert(std::make_pair(LSRUse::SCEVUseKindPair(Expr, Kind), 0)); 2283 if (!P.second) { 2284 // A use already existed with this base. 2285 size_t LUIdx = P.first->second; 2286 LSRUse &LU = Uses[LUIdx]; 2287 if (reconcileNewOffset(LU, Offset, /*HasBaseReg=*/true, Kind, AccessTy)) 2288 // Reuse this use. 2289 return std::make_pair(LUIdx, Offset); 2290 } 2291 2292 // Create a new use. 2293 size_t LUIdx = Uses.size(); 2294 P.first->second = LUIdx; 2295 Uses.push_back(LSRUse(Kind, AccessTy)); 2296 LSRUse &LU = Uses[LUIdx]; 2297 2298 // We don't need to track redundant offsets, but we don't need to go out 2299 // of our way here to avoid them. 2300 if (LU.Offsets.empty() || Offset != LU.Offsets.back()) 2301 LU.Offsets.push_back(Offset); 2302 2303 LU.MinOffset = Offset; 2304 LU.MaxOffset = Offset; 2305 return std::make_pair(LUIdx, Offset); 2306 } 2307 2308 /// Delete the given use from the Uses list. 2309 void LSRInstance::DeleteUse(LSRUse &LU, size_t LUIdx) { 2310 if (&LU != &Uses.back()) 2311 std::swap(LU, Uses.back()); 2312 Uses.pop_back(); 2313 2314 // Update RegUses. 2315 RegUses.swapAndDropUse(LUIdx, Uses.size()); 2316 } 2317 2318 /// Look for a use distinct from OrigLU which is has a formula that has the same 2319 /// registers as the given formula. 2320 LSRUse * 2321 LSRInstance::FindUseWithSimilarFormula(const Formula &OrigF, 2322 const LSRUse &OrigLU) { 2323 // Search all uses for the formula. This could be more clever. 2324 for (size_t LUIdx = 0, NumUses = Uses.size(); LUIdx != NumUses; ++LUIdx) { 2325 LSRUse &LU = Uses[LUIdx]; 2326 // Check whether this use is close enough to OrigLU, to see whether it's 2327 // worthwhile looking through its formulae. 2328 // Ignore ICmpZero uses because they may contain formulae generated by 2329 // GenerateICmpZeroScales, in which case adding fixup offsets may 2330 // be invalid. 2331 if (&LU != &OrigLU && 2332 LU.Kind != LSRUse::ICmpZero && 2333 LU.Kind == OrigLU.Kind && OrigLU.AccessTy == LU.AccessTy && 2334 LU.WidestFixupType == OrigLU.WidestFixupType && 2335 LU.HasFormulaWithSameRegs(OrigF)) { 2336 // Scan through this use's formulae. 2337 for (const Formula &F : LU.Formulae) { 2338 // Check to see if this formula has the same registers and symbols 2339 // as OrigF. 2340 if (F.BaseRegs == OrigF.BaseRegs && 2341 F.ScaledReg == OrigF.ScaledReg && 2342 F.BaseGV == OrigF.BaseGV && 2343 F.Scale == OrigF.Scale && 2344 F.UnfoldedOffset == OrigF.UnfoldedOffset) { 2345 if (F.BaseOffset == 0) 2346 return &LU; 2347 // This is the formula where all the registers and symbols matched; 2348 // there aren't going to be any others. Since we declined it, we 2349 // can skip the rest of the formulae and proceed to the next LSRUse. 2350 break; 2351 } 2352 } 2353 } 2354 } 2355 2356 // Nothing looked good. 2357 return nullptr; 2358 } 2359 2360 void LSRInstance::CollectInterestingTypesAndFactors() { 2361 SmallSetVector<const SCEV *, 4> Strides; 2362 2363 // Collect interesting types and strides. 2364 SmallVector<const SCEV *, 4> Worklist; 2365 for (const IVStrideUse &U : IU) { 2366 const SCEV *Expr = IU.getExpr(U); 2367 2368 // Collect interesting types. 2369 Types.insert(SE.getEffectiveSCEVType(Expr->getType())); 2370 2371 // Add strides for mentioned loops. 2372 Worklist.push_back(Expr); 2373 do { 2374 const SCEV *S = Worklist.pop_back_val(); 2375 if (const SCEVAddRecExpr *AR = dyn_cast<SCEVAddRecExpr>(S)) { 2376 if (AR->getLoop() == L) 2377 Strides.insert(AR->getStepRecurrence(SE)); 2378 Worklist.push_back(AR->getStart()); 2379 } else if (const SCEVAddExpr *Add = dyn_cast<SCEVAddExpr>(S)) { 2380 Worklist.append(Add->op_begin(), Add->op_end()); 2381 } 2382 } while (!Worklist.empty()); 2383 } 2384 2385 // Compute interesting factors from the set of interesting strides. 2386 for (SmallSetVector<const SCEV *, 4>::const_iterator 2387 I = Strides.begin(), E = Strides.end(); I != E; ++I) 2388 for (SmallSetVector<const SCEV *, 4>::const_iterator NewStrideIter = 2389 std::next(I); NewStrideIter != E; ++NewStrideIter) { 2390 const SCEV *OldStride = *I; 2391 const SCEV *NewStride = *NewStrideIter; 2392 2393 if (SE.getTypeSizeInBits(OldStride->getType()) != 2394 SE.getTypeSizeInBits(NewStride->getType())) { 2395 if (SE.getTypeSizeInBits(OldStride->getType()) > 2396 SE.getTypeSizeInBits(NewStride->getType())) 2397 NewStride = SE.getSignExtendExpr(NewStride, OldStride->getType()); 2398 else 2399 OldStride = SE.getSignExtendExpr(OldStride, NewStride->getType()); 2400 } 2401 if (const SCEVConstant *Factor = 2402 dyn_cast_or_null<SCEVConstant>(getExactSDiv(NewStride, OldStride, 2403 SE, true))) { 2404 if (Factor->getAPInt().getMinSignedBits() <= 64) 2405 Factors.insert(Factor->getAPInt().getSExtValue()); 2406 } else if (const SCEVConstant *Factor = 2407 dyn_cast_or_null<SCEVConstant>(getExactSDiv(OldStride, 2408 NewStride, 2409 SE, true))) { 2410 if (Factor->getAPInt().getMinSignedBits() <= 64) 2411 Factors.insert(Factor->getAPInt().getSExtValue()); 2412 } 2413 } 2414 2415 // If all uses use the same type, don't bother looking for truncation-based 2416 // reuse. 2417 if (Types.size() == 1) 2418 Types.clear(); 2419 2420 DEBUG(print_factors_and_types(dbgs())); 2421 } 2422 2423 /// Helper for CollectChains that finds an IV operand (computed by an AddRec in 2424 /// this loop) within [OI,OE) or returns OE. If IVUsers mapped Instructions to 2425 /// IVStrideUses, we could partially skip this. 2426 static User::op_iterator 2427 findIVOperand(User::op_iterator OI, User::op_iterator OE, 2428 Loop *L, ScalarEvolution &SE) { 2429 for(; OI != OE; ++OI) { 2430 if (Instruction *Oper = dyn_cast<Instruction>(*OI)) { 2431 if (!SE.isSCEVable(Oper->getType())) 2432 continue; 2433 2434 if (const SCEVAddRecExpr *AR = 2435 dyn_cast<SCEVAddRecExpr>(SE.getSCEV(Oper))) { 2436 if (AR->getLoop() == L) 2437 break; 2438 } 2439 } 2440 } 2441 return OI; 2442 } 2443 2444 /// IVChain logic must consistenctly peek base TruncInst operands, so wrap it in 2445 /// a convenient helper. 2446 static Value *getWideOperand(Value *Oper) { 2447 if (TruncInst *Trunc = dyn_cast<TruncInst>(Oper)) 2448 return Trunc->getOperand(0); 2449 return Oper; 2450 } 2451 2452 /// Return true if we allow an IV chain to include both types. 2453 static bool isCompatibleIVType(Value *LVal, Value *RVal) { 2454 Type *LType = LVal->getType(); 2455 Type *RType = RVal->getType(); 2456 return (LType == RType) || (LType->isPointerTy() && RType->isPointerTy()); 2457 } 2458 2459 /// Return an approximation of this SCEV expression's "base", or NULL for any 2460 /// constant. Returning the expression itself is conservative. Returning a 2461 /// deeper subexpression is more precise and valid as long as it isn't less 2462 /// complex than another subexpression. For expressions involving multiple 2463 /// unscaled values, we need to return the pointer-type SCEVUnknown. This avoids 2464 /// forming chains across objects, such as: PrevOper==a[i], IVOper==b[i], 2465 /// IVInc==b-a. 2466 /// 2467 /// Since SCEVUnknown is the rightmost type, and pointers are the rightmost 2468 /// SCEVUnknown, we simply return the rightmost SCEV operand. 2469 static const SCEV *getExprBase(const SCEV *S) { 2470 switch (S->getSCEVType()) { 2471 default: // uncluding scUnknown. 2472 return S; 2473 case scConstant: 2474 return nullptr; 2475 case scTruncate: 2476 return getExprBase(cast<SCEVTruncateExpr>(S)->getOperand()); 2477 case scZeroExtend: 2478 return getExprBase(cast<SCEVZeroExtendExpr>(S)->getOperand()); 2479 case scSignExtend: 2480 return getExprBase(cast<SCEVSignExtendExpr>(S)->getOperand()); 2481 case scAddExpr: { 2482 // Skip over scaled operands (scMulExpr) to follow add operands as long as 2483 // there's nothing more complex. 2484 // FIXME: not sure if we want to recognize negation. 2485 const SCEVAddExpr *Add = cast<SCEVAddExpr>(S); 2486 for (std::reverse_iterator<SCEVAddExpr::op_iterator> I(Add->op_end()), 2487 E(Add->op_begin()); I != E; ++I) { 2488 const SCEV *SubExpr = *I; 2489 if (SubExpr->getSCEVType() == scAddExpr) 2490 return getExprBase(SubExpr); 2491 2492 if (SubExpr->getSCEVType() != scMulExpr) 2493 return SubExpr; 2494 } 2495 return S; // all operands are scaled, be conservative. 2496 } 2497 case scAddRecExpr: 2498 return getExprBase(cast<SCEVAddRecExpr>(S)->getStart()); 2499 } 2500 } 2501 2502 /// Return true if the chain increment is profitable to expand into a loop 2503 /// invariant value, which may require its own register. A profitable chain 2504 /// increment will be an offset relative to the same base. We allow such offsets 2505 /// to potentially be used as chain increment as long as it's not obviously 2506 /// expensive to expand using real instructions. 2507 bool IVChain::isProfitableIncrement(const SCEV *OperExpr, 2508 const SCEV *IncExpr, 2509 ScalarEvolution &SE) { 2510 // Aggressively form chains when -stress-ivchain. 2511 if (StressIVChain) 2512 return true; 2513 2514 // Do not replace a constant offset from IV head with a nonconstant IV 2515 // increment. 2516 if (!isa<SCEVConstant>(IncExpr)) { 2517 const SCEV *HeadExpr = SE.getSCEV(getWideOperand(Incs[0].IVOperand)); 2518 if (isa<SCEVConstant>(SE.getMinusSCEV(OperExpr, HeadExpr))) 2519 return 0; 2520 } 2521 2522 SmallPtrSet<const SCEV*, 8> Processed; 2523 return !isHighCostExpansion(IncExpr, Processed, SE); 2524 } 2525 2526 /// Return true if the number of registers needed for the chain is estimated to 2527 /// be less than the number required for the individual IV users. First prohibit 2528 /// any IV users that keep the IV live across increments (the Users set should 2529 /// be empty). Next count the number and type of increments in the chain. 2530 /// 2531 /// Chaining IVs can lead to considerable code bloat if ISEL doesn't 2532 /// effectively use postinc addressing modes. Only consider it profitable it the 2533 /// increments can be computed in fewer registers when chained. 2534 /// 2535 /// TODO: Consider IVInc free if it's already used in another chains. 2536 static bool 2537 isProfitableChain(IVChain &Chain, SmallPtrSetImpl<Instruction*> &Users, 2538 ScalarEvolution &SE, const TargetTransformInfo &TTI) { 2539 if (StressIVChain) 2540 return true; 2541 2542 if (!Chain.hasIncs()) 2543 return false; 2544 2545 if (!Users.empty()) { 2546 DEBUG(dbgs() << "Chain: " << *Chain.Incs[0].UserInst << " users:\n"; 2547 for (Instruction *Inst : Users) { 2548 dbgs() << " " << *Inst << "\n"; 2549 }); 2550 return false; 2551 } 2552 assert(!Chain.Incs.empty() && "empty IV chains are not allowed"); 2553 2554 // The chain itself may require a register, so intialize cost to 1. 2555 int cost = 1; 2556 2557 // A complete chain likely eliminates the need for keeping the original IV in 2558 // a register. LSR does not currently know how to form a complete chain unless 2559 // the header phi already exists. 2560 if (isa<PHINode>(Chain.tailUserInst()) 2561 && SE.getSCEV(Chain.tailUserInst()) == Chain.Incs[0].IncExpr) { 2562 --cost; 2563 } 2564 const SCEV *LastIncExpr = nullptr; 2565 unsigned NumConstIncrements = 0; 2566 unsigned NumVarIncrements = 0; 2567 unsigned NumReusedIncrements = 0; 2568 for (const IVInc &Inc : Chain) { 2569 if (Inc.IncExpr->isZero()) 2570 continue; 2571 2572 // Incrementing by zero or some constant is neutral. We assume constants can 2573 // be folded into an addressing mode or an add's immediate operand. 2574 if (isa<SCEVConstant>(Inc.IncExpr)) { 2575 ++NumConstIncrements; 2576 continue; 2577 } 2578 2579 if (Inc.IncExpr == LastIncExpr) 2580 ++NumReusedIncrements; 2581 else 2582 ++NumVarIncrements; 2583 2584 LastIncExpr = Inc.IncExpr; 2585 } 2586 // An IV chain with a single increment is handled by LSR's postinc 2587 // uses. However, a chain with multiple increments requires keeping the IV's 2588 // value live longer than it needs to be if chained. 2589 if (NumConstIncrements > 1) 2590 --cost; 2591 2592 // Materializing increment expressions in the preheader that didn't exist in 2593 // the original code may cost a register. For example, sign-extended array 2594 // indices can produce ridiculous increments like this: 2595 // IV + ((sext i32 (2 * %s) to i64) + (-1 * (sext i32 %s to i64))) 2596 cost += NumVarIncrements; 2597 2598 // Reusing variable increments likely saves a register to hold the multiple of 2599 // the stride. 2600 cost -= NumReusedIncrements; 2601 2602 DEBUG(dbgs() << "Chain: " << *Chain.Incs[0].UserInst << " Cost: " << cost 2603 << "\n"); 2604 2605 return cost < 0; 2606 } 2607 2608 /// Add this IV user to an existing chain or make it the head of a new chain. 2609 void LSRInstance::ChainInstruction(Instruction *UserInst, Instruction *IVOper, 2610 SmallVectorImpl<ChainUsers> &ChainUsersVec) { 2611 // When IVs are used as types of varying widths, they are generally converted 2612 // to a wider type with some uses remaining narrow under a (free) trunc. 2613 Value *const NextIV = getWideOperand(IVOper); 2614 const SCEV *const OperExpr = SE.getSCEV(NextIV); 2615 const SCEV *const OperExprBase = getExprBase(OperExpr); 2616 2617 // Visit all existing chains. Check if its IVOper can be computed as a 2618 // profitable loop invariant increment from the last link in the Chain. 2619 unsigned ChainIdx = 0, NChains = IVChainVec.size(); 2620 const SCEV *LastIncExpr = nullptr; 2621 for (; ChainIdx < NChains; ++ChainIdx) { 2622 IVChain &Chain = IVChainVec[ChainIdx]; 2623 2624 // Prune the solution space aggressively by checking that both IV operands 2625 // are expressions that operate on the same unscaled SCEVUnknown. This 2626 // "base" will be canceled by the subsequent getMinusSCEV call. Checking 2627 // first avoids creating extra SCEV expressions. 2628 if (!StressIVChain && Chain.ExprBase != OperExprBase) 2629 continue; 2630 2631 Value *PrevIV = getWideOperand(Chain.Incs.back().IVOperand); 2632 if (!isCompatibleIVType(PrevIV, NextIV)) 2633 continue; 2634 2635 // A phi node terminates a chain. 2636 if (isa<PHINode>(UserInst) && isa<PHINode>(Chain.tailUserInst())) 2637 continue; 2638 2639 // The increment must be loop-invariant so it can be kept in a register. 2640 const SCEV *PrevExpr = SE.getSCEV(PrevIV); 2641 const SCEV *IncExpr = SE.getMinusSCEV(OperExpr, PrevExpr); 2642 if (!SE.isLoopInvariant(IncExpr, L)) 2643 continue; 2644 2645 if (Chain.isProfitableIncrement(OperExpr, IncExpr, SE)) { 2646 LastIncExpr = IncExpr; 2647 break; 2648 } 2649 } 2650 // If we haven't found a chain, create a new one, unless we hit the max. Don't 2651 // bother for phi nodes, because they must be last in the chain. 2652 if (ChainIdx == NChains) { 2653 if (isa<PHINode>(UserInst)) 2654 return; 2655 if (NChains >= MaxChains && !StressIVChain) { 2656 DEBUG(dbgs() << "IV Chain Limit\n"); 2657 return; 2658 } 2659 LastIncExpr = OperExpr; 2660 // IVUsers may have skipped over sign/zero extensions. We don't currently 2661 // attempt to form chains involving extensions unless they can be hoisted 2662 // into this loop's AddRec. 2663 if (!isa<SCEVAddRecExpr>(LastIncExpr)) 2664 return; 2665 ++NChains; 2666 IVChainVec.push_back(IVChain(IVInc(UserInst, IVOper, LastIncExpr), 2667 OperExprBase)); 2668 ChainUsersVec.resize(NChains); 2669 DEBUG(dbgs() << "IV Chain#" << ChainIdx << " Head: (" << *UserInst 2670 << ") IV=" << *LastIncExpr << "\n"); 2671 } else { 2672 DEBUG(dbgs() << "IV Chain#" << ChainIdx << " Inc: (" << *UserInst 2673 << ") IV+" << *LastIncExpr << "\n"); 2674 // Add this IV user to the end of the chain. 2675 IVChainVec[ChainIdx].add(IVInc(UserInst, IVOper, LastIncExpr)); 2676 } 2677 IVChain &Chain = IVChainVec[ChainIdx]; 2678 2679 SmallPtrSet<Instruction*,4> &NearUsers = ChainUsersVec[ChainIdx].NearUsers; 2680 // This chain's NearUsers become FarUsers. 2681 if (!LastIncExpr->isZero()) { 2682 ChainUsersVec[ChainIdx].FarUsers.insert(NearUsers.begin(), 2683 NearUsers.end()); 2684 NearUsers.clear(); 2685 } 2686 2687 // All other uses of IVOperand become near uses of the chain. 2688 // We currently ignore intermediate values within SCEV expressions, assuming 2689 // they will eventually be used be the current chain, or can be computed 2690 // from one of the chain increments. To be more precise we could 2691 // transitively follow its user and only add leaf IV users to the set. 2692 for (User *U : IVOper->users()) { 2693 Instruction *OtherUse = dyn_cast<Instruction>(U); 2694 if (!OtherUse) 2695 continue; 2696 // Uses in the chain will no longer be uses if the chain is formed. 2697 // Include the head of the chain in this iteration (not Chain.begin()). 2698 IVChain::const_iterator IncIter = Chain.Incs.begin(); 2699 IVChain::const_iterator IncEnd = Chain.Incs.end(); 2700 for( ; IncIter != IncEnd; ++IncIter) { 2701 if (IncIter->UserInst == OtherUse) 2702 break; 2703 } 2704 if (IncIter != IncEnd) 2705 continue; 2706 2707 if (SE.isSCEVable(OtherUse->getType()) 2708 && !isa<SCEVUnknown>(SE.getSCEV(OtherUse)) 2709 && IU.isIVUserOrOperand(OtherUse)) { 2710 continue; 2711 } 2712 NearUsers.insert(OtherUse); 2713 } 2714 2715 // Since this user is part of the chain, it's no longer considered a use 2716 // of the chain. 2717 ChainUsersVec[ChainIdx].FarUsers.erase(UserInst); 2718 } 2719 2720 /// Populate the vector of Chains. 2721 /// 2722 /// This decreases ILP at the architecture level. Targets with ample registers, 2723 /// multiple memory ports, and no register renaming probably don't want 2724 /// this. However, such targets should probably disable LSR altogether. 2725 /// 2726 /// The job of LSR is to make a reasonable choice of induction variables across 2727 /// the loop. Subsequent passes can easily "unchain" computation exposing more 2728 /// ILP *within the loop* if the target wants it. 2729 /// 2730 /// Finding the best IV chain is potentially a scheduling problem. Since LSR 2731 /// will not reorder memory operations, it will recognize this as a chain, but 2732 /// will generate redundant IV increments. Ideally this would be corrected later 2733 /// by a smart scheduler: 2734 /// = A[i] 2735 /// = A[i+x] 2736 /// A[i] = 2737 /// A[i+x] = 2738 /// 2739 /// TODO: Walk the entire domtree within this loop, not just the path to the 2740 /// loop latch. This will discover chains on side paths, but requires 2741 /// maintaining multiple copies of the Chains state. 2742 void LSRInstance::CollectChains() { 2743 DEBUG(dbgs() << "Collecting IV Chains.\n"); 2744 SmallVector<ChainUsers, 8> ChainUsersVec; 2745 2746 SmallVector<BasicBlock *,8> LatchPath; 2747 BasicBlock *LoopHeader = L->getHeader(); 2748 for (DomTreeNode *Rung = DT.getNode(L->getLoopLatch()); 2749 Rung->getBlock() != LoopHeader; Rung = Rung->getIDom()) { 2750 LatchPath.push_back(Rung->getBlock()); 2751 } 2752 LatchPath.push_back(LoopHeader); 2753 2754 // Walk the instruction stream from the loop header to the loop latch. 2755 for (SmallVectorImpl<BasicBlock *>::reverse_iterator 2756 BBIter = LatchPath.rbegin(), BBEnd = LatchPath.rend(); 2757 BBIter != BBEnd; ++BBIter) { 2758 for (BasicBlock::iterator I = (*BBIter)->begin(), E = (*BBIter)->end(); 2759 I != E; ++I) { 2760 // Skip instructions that weren't seen by IVUsers analysis. 2761 if (isa<PHINode>(I) || !IU.isIVUserOrOperand(&*I)) 2762 continue; 2763 2764 // Ignore users that are part of a SCEV expression. This way we only 2765 // consider leaf IV Users. This effectively rediscovers a portion of 2766 // IVUsers analysis but in program order this time. 2767 if (SE.isSCEVable(I->getType()) && !isa<SCEVUnknown>(SE.getSCEV(&*I))) 2768 continue; 2769 2770 // Remove this instruction from any NearUsers set it may be in. 2771 for (unsigned ChainIdx = 0, NChains = IVChainVec.size(); 2772 ChainIdx < NChains; ++ChainIdx) { 2773 ChainUsersVec[ChainIdx].NearUsers.erase(&*I); 2774 } 2775 // Search for operands that can be chained. 2776 SmallPtrSet<Instruction*, 4> UniqueOperands; 2777 User::op_iterator IVOpEnd = I->op_end(); 2778 User::op_iterator IVOpIter = findIVOperand(I->op_begin(), IVOpEnd, L, SE); 2779 while (IVOpIter != IVOpEnd) { 2780 Instruction *IVOpInst = cast<Instruction>(*IVOpIter); 2781 if (UniqueOperands.insert(IVOpInst).second) 2782 ChainInstruction(&*I, IVOpInst, ChainUsersVec); 2783 IVOpIter = findIVOperand(std::next(IVOpIter), IVOpEnd, L, SE); 2784 } 2785 } // Continue walking down the instructions. 2786 } // Continue walking down the domtree. 2787 // Visit phi backedges to determine if the chain can generate the IV postinc. 2788 for (BasicBlock::iterator I = L->getHeader()->begin(); 2789 PHINode *PN = dyn_cast<PHINode>(I); ++I) { 2790 if (!SE.isSCEVable(PN->getType())) 2791 continue; 2792 2793 Instruction *IncV = 2794 dyn_cast<Instruction>(PN->getIncomingValueForBlock(L->getLoopLatch())); 2795 if (IncV) 2796 ChainInstruction(PN, IncV, ChainUsersVec); 2797 } 2798 // Remove any unprofitable chains. 2799 unsigned ChainIdx = 0; 2800 for (unsigned UsersIdx = 0, NChains = IVChainVec.size(); 2801 UsersIdx < NChains; ++UsersIdx) { 2802 if (!isProfitableChain(IVChainVec[UsersIdx], 2803 ChainUsersVec[UsersIdx].FarUsers, SE, TTI)) 2804 continue; 2805 // Preserve the chain at UsesIdx. 2806 if (ChainIdx != UsersIdx) 2807 IVChainVec[ChainIdx] = IVChainVec[UsersIdx]; 2808 FinalizeChain(IVChainVec[ChainIdx]); 2809 ++ChainIdx; 2810 } 2811 IVChainVec.resize(ChainIdx); 2812 } 2813 2814 void LSRInstance::FinalizeChain(IVChain &Chain) { 2815 assert(!Chain.Incs.empty() && "empty IV chains are not allowed"); 2816 DEBUG(dbgs() << "Final Chain: " << *Chain.Incs[0].UserInst << "\n"); 2817 2818 for (const IVInc &Inc : Chain) { 2819 DEBUG(dbgs() << " Inc: " << Inc.UserInst << "\n"); 2820 auto UseI = std::find(Inc.UserInst->op_begin(), Inc.UserInst->op_end(), 2821 Inc.IVOperand); 2822 assert(UseI != Inc.UserInst->op_end() && "cannot find IV operand"); 2823 IVIncSet.insert(UseI); 2824 } 2825 } 2826 2827 /// Return true if the IVInc can be folded into an addressing mode. 2828 static bool canFoldIVIncExpr(const SCEV *IncExpr, Instruction *UserInst, 2829 Value *Operand, const TargetTransformInfo &TTI) { 2830 const SCEVConstant *IncConst = dyn_cast<SCEVConstant>(IncExpr); 2831 if (!IncConst || !isAddressUse(UserInst, Operand)) 2832 return false; 2833 2834 if (IncConst->getAPInt().getMinSignedBits() > 64) 2835 return false; 2836 2837 MemAccessTy AccessTy = getAccessType(UserInst); 2838 int64_t IncOffset = IncConst->getValue()->getSExtValue(); 2839 if (!isAlwaysFoldable(TTI, LSRUse::Address, AccessTy, /*BaseGV=*/nullptr, 2840 IncOffset, /*HaseBaseReg=*/false)) 2841 return false; 2842 2843 return true; 2844 } 2845 2846 /// Generate an add or subtract for each IVInc in a chain to materialize the IV 2847 /// user's operand from the previous IV user's operand. 2848 void LSRInstance::GenerateIVChain(const IVChain &Chain, SCEVExpander &Rewriter, 2849 SmallVectorImpl<WeakVH> &DeadInsts) { 2850 // Find the new IVOperand for the head of the chain. It may have been replaced 2851 // by LSR. 2852 const IVInc &Head = Chain.Incs[0]; 2853 User::op_iterator IVOpEnd = Head.UserInst->op_end(); 2854 // findIVOperand returns IVOpEnd if it can no longer find a valid IV user. 2855 User::op_iterator IVOpIter = findIVOperand(Head.UserInst->op_begin(), 2856 IVOpEnd, L, SE); 2857 Value *IVSrc = nullptr; 2858 while (IVOpIter != IVOpEnd) { 2859 IVSrc = getWideOperand(*IVOpIter); 2860 2861 // If this operand computes the expression that the chain needs, we may use 2862 // it. (Check this after setting IVSrc which is used below.) 2863 // 2864 // Note that if Head.IncExpr is wider than IVSrc, then this phi is too 2865 // narrow for the chain, so we can no longer use it. We do allow using a 2866 // wider phi, assuming the LSR checked for free truncation. In that case we 2867 // should already have a truncate on this operand such that 2868 // getSCEV(IVSrc) == IncExpr. 2869 if (SE.getSCEV(*IVOpIter) == Head.IncExpr 2870 || SE.getSCEV(IVSrc) == Head.IncExpr) { 2871 break; 2872 } 2873 IVOpIter = findIVOperand(std::next(IVOpIter), IVOpEnd, L, SE); 2874 } 2875 if (IVOpIter == IVOpEnd) { 2876 // Gracefully give up on this chain. 2877 DEBUG(dbgs() << "Concealed chain head: " << *Head.UserInst << "\n"); 2878 return; 2879 } 2880 2881 DEBUG(dbgs() << "Generate chain at: " << *IVSrc << "\n"); 2882 Type *IVTy = IVSrc->getType(); 2883 Type *IntTy = SE.getEffectiveSCEVType(IVTy); 2884 const SCEV *LeftOverExpr = nullptr; 2885 for (const IVInc &Inc : Chain) { 2886 Instruction *InsertPt = Inc.UserInst; 2887 if (isa<PHINode>(InsertPt)) 2888 InsertPt = L->getLoopLatch()->getTerminator(); 2889 2890 // IVOper will replace the current IV User's operand. IVSrc is the IV 2891 // value currently held in a register. 2892 Value *IVOper = IVSrc; 2893 if (!Inc.IncExpr->isZero()) { 2894 // IncExpr was the result of subtraction of two narrow values, so must 2895 // be signed. 2896 const SCEV *IncExpr = SE.getNoopOrSignExtend(Inc.IncExpr, IntTy); 2897 LeftOverExpr = LeftOverExpr ? 2898 SE.getAddExpr(LeftOverExpr, IncExpr) : IncExpr; 2899 } 2900 if (LeftOverExpr && !LeftOverExpr->isZero()) { 2901 // Expand the IV increment. 2902 Rewriter.clearPostInc(); 2903 Value *IncV = Rewriter.expandCodeFor(LeftOverExpr, IntTy, InsertPt); 2904 const SCEV *IVOperExpr = SE.getAddExpr(SE.getUnknown(IVSrc), 2905 SE.getUnknown(IncV)); 2906 IVOper = Rewriter.expandCodeFor(IVOperExpr, IVTy, InsertPt); 2907 2908 // If an IV increment can't be folded, use it as the next IV value. 2909 if (!canFoldIVIncExpr(LeftOverExpr, Inc.UserInst, Inc.IVOperand, TTI)) { 2910 assert(IVTy == IVOper->getType() && "inconsistent IV increment type"); 2911 IVSrc = IVOper; 2912 LeftOverExpr = nullptr; 2913 } 2914 } 2915 Type *OperTy = Inc.IVOperand->getType(); 2916 if (IVTy != OperTy) { 2917 assert(SE.getTypeSizeInBits(IVTy) >= SE.getTypeSizeInBits(OperTy) && 2918 "cannot extend a chained IV"); 2919 IRBuilder<> Builder(InsertPt); 2920 IVOper = Builder.CreateTruncOrBitCast(IVOper, OperTy, "lsr.chain"); 2921 } 2922 Inc.UserInst->replaceUsesOfWith(Inc.IVOperand, IVOper); 2923 DeadInsts.emplace_back(Inc.IVOperand); 2924 } 2925 // If LSR created a new, wider phi, we may also replace its postinc. We only 2926 // do this if we also found a wide value for the head of the chain. 2927 if (isa<PHINode>(Chain.tailUserInst())) { 2928 for (BasicBlock::iterator I = L->getHeader()->begin(); 2929 PHINode *Phi = dyn_cast<PHINode>(I); ++I) { 2930 if (!isCompatibleIVType(Phi, IVSrc)) 2931 continue; 2932 Instruction *PostIncV = dyn_cast<Instruction>( 2933 Phi->getIncomingValueForBlock(L->getLoopLatch())); 2934 if (!PostIncV || (SE.getSCEV(PostIncV) != SE.getSCEV(IVSrc))) 2935 continue; 2936 Value *IVOper = IVSrc; 2937 Type *PostIncTy = PostIncV->getType(); 2938 if (IVTy != PostIncTy) { 2939 assert(PostIncTy->isPointerTy() && "mixing int/ptr IV types"); 2940 IRBuilder<> Builder(L->getLoopLatch()->getTerminator()); 2941 Builder.SetCurrentDebugLocation(PostIncV->getDebugLoc()); 2942 IVOper = Builder.CreatePointerCast(IVSrc, PostIncTy, "lsr.chain"); 2943 } 2944 Phi->replaceUsesOfWith(PostIncV, IVOper); 2945 DeadInsts.emplace_back(PostIncV); 2946 } 2947 } 2948 } 2949 2950 void LSRInstance::CollectFixupsAndInitialFormulae() { 2951 for (const IVStrideUse &U : IU) { 2952 Instruction *UserInst = U.getUser(); 2953 // Skip IV users that are part of profitable IV Chains. 2954 User::op_iterator UseI = std::find(UserInst->op_begin(), UserInst->op_end(), 2955 U.getOperandValToReplace()); 2956 assert(UseI != UserInst->op_end() && "cannot find IV operand"); 2957 if (IVIncSet.count(UseI)) 2958 continue; 2959 2960 // Record the uses. 2961 LSRFixup &LF = getNewFixup(); 2962 LF.UserInst = UserInst; 2963 LF.OperandValToReplace = U.getOperandValToReplace(); 2964 LF.PostIncLoops = U.getPostIncLoops(); 2965 2966 LSRUse::KindType Kind = LSRUse::Basic; 2967 MemAccessTy AccessTy; 2968 if (isAddressUse(LF.UserInst, LF.OperandValToReplace)) { 2969 Kind = LSRUse::Address; 2970 AccessTy = getAccessType(LF.UserInst); 2971 } 2972 2973 const SCEV *S = IU.getExpr(U); 2974 2975 // Equality (== and !=) ICmps are special. We can rewrite (i == N) as 2976 // (N - i == 0), and this allows (N - i) to be the expression that we work 2977 // with rather than just N or i, so we can consider the register 2978 // requirements for both N and i at the same time. Limiting this code to 2979 // equality icmps is not a problem because all interesting loops use 2980 // equality icmps, thanks to IndVarSimplify. 2981 if (ICmpInst *CI = dyn_cast<ICmpInst>(LF.UserInst)) 2982 if (CI->isEquality()) { 2983 // Swap the operands if needed to put the OperandValToReplace on the 2984 // left, for consistency. 2985 Value *NV = CI->getOperand(1); 2986 if (NV == LF.OperandValToReplace) { 2987 CI->setOperand(1, CI->getOperand(0)); 2988 CI->setOperand(0, NV); 2989 NV = CI->getOperand(1); 2990 Changed = true; 2991 } 2992 2993 // x == y --> x - y == 0 2994 const SCEV *N = SE.getSCEV(NV); 2995 if (SE.isLoopInvariant(N, L) && isSafeToExpand(N, SE)) { 2996 // S is normalized, so normalize N before folding it into S 2997 // to keep the result normalized. 2998 N = TransformForPostIncUse(Normalize, N, CI, nullptr, 2999 LF.PostIncLoops, SE, DT); 3000 Kind = LSRUse::ICmpZero; 3001 S = SE.getMinusSCEV(N, S); 3002 } 3003 3004 // -1 and the negations of all interesting strides (except the negation 3005 // of -1) are now also interesting. 3006 for (size_t i = 0, e = Factors.size(); i != e; ++i) 3007 if (Factors[i] != -1) 3008 Factors.insert(-(uint64_t)Factors[i]); 3009 Factors.insert(-1); 3010 } 3011 3012 // Set up the initial formula for this use. 3013 std::pair<size_t, int64_t> P = getUse(S, Kind, AccessTy); 3014 LF.LUIdx = P.first; 3015 LF.Offset = P.second; 3016 LSRUse &LU = Uses[LF.LUIdx]; 3017 LU.AllFixupsOutsideLoop &= LF.isUseFullyOutsideLoop(L); 3018 if (!LU.WidestFixupType || 3019 SE.getTypeSizeInBits(LU.WidestFixupType) < 3020 SE.getTypeSizeInBits(LF.OperandValToReplace->getType())) 3021 LU.WidestFixupType = LF.OperandValToReplace->getType(); 3022 3023 // If this is the first use of this LSRUse, give it a formula. 3024 if (LU.Formulae.empty()) { 3025 InsertInitialFormula(S, LU, LF.LUIdx); 3026 CountRegisters(LU.Formulae.back(), LF.LUIdx); 3027 } 3028 } 3029 3030 DEBUG(print_fixups(dbgs())); 3031 } 3032 3033 /// Insert a formula for the given expression into the given use, separating out 3034 /// loop-variant portions from loop-invariant and loop-computable portions. 3035 void 3036 LSRInstance::InsertInitialFormula(const SCEV *S, LSRUse &LU, size_t LUIdx) { 3037 // Mark uses whose expressions cannot be expanded. 3038 if (!isSafeToExpand(S, SE)) 3039 LU.RigidFormula = true; 3040 3041 Formula F; 3042 F.initialMatch(S, L, SE); 3043 bool Inserted = InsertFormula(LU, LUIdx, F); 3044 assert(Inserted && "Initial formula already exists!"); (void)Inserted; 3045 } 3046 3047 /// Insert a simple single-register formula for the given expression into the 3048 /// given use. 3049 void 3050 LSRInstance::InsertSupplementalFormula(const SCEV *S, 3051 LSRUse &LU, size_t LUIdx) { 3052 Formula F; 3053 F.BaseRegs.push_back(S); 3054 F.HasBaseReg = true; 3055 bool Inserted = InsertFormula(LU, LUIdx, F); 3056 assert(Inserted && "Supplemental formula already exists!"); (void)Inserted; 3057 } 3058 3059 /// Note which registers are used by the given formula, updating RegUses. 3060 void LSRInstance::CountRegisters(const Formula &F, size_t LUIdx) { 3061 if (F.ScaledReg) 3062 RegUses.countRegister(F.ScaledReg, LUIdx); 3063 for (const SCEV *BaseReg : F.BaseRegs) 3064 RegUses.countRegister(BaseReg, LUIdx); 3065 } 3066 3067 /// If the given formula has not yet been inserted, add it to the list, and 3068 /// return true. Return false otherwise. 3069 bool LSRInstance::InsertFormula(LSRUse &LU, unsigned LUIdx, const Formula &F) { 3070 // Do not insert formula that we will not be able to expand. 3071 assert(isLegalUse(TTI, LU.MinOffset, LU.MaxOffset, LU.Kind, LU.AccessTy, F) && 3072 "Formula is illegal"); 3073 if (!LU.InsertFormula(F)) 3074 return false; 3075 3076 CountRegisters(F, LUIdx); 3077 return true; 3078 } 3079 3080 /// Check for other uses of loop-invariant values which we're tracking. These 3081 /// other uses will pin these values in registers, making them less profitable 3082 /// for elimination. 3083 /// TODO: This currently misses non-constant addrec step registers. 3084 /// TODO: Should this give more weight to users inside the loop? 3085 void 3086 LSRInstance::CollectLoopInvariantFixupsAndFormulae() { 3087 SmallVector<const SCEV *, 8> Worklist(RegUses.begin(), RegUses.end()); 3088 SmallPtrSet<const SCEV *, 32> Visited; 3089 3090 while (!Worklist.empty()) { 3091 const SCEV *S = Worklist.pop_back_val(); 3092 3093 // Don't process the same SCEV twice 3094 if (!Visited.insert(S).second) 3095 continue; 3096 3097 if (const SCEVNAryExpr *N = dyn_cast<SCEVNAryExpr>(S)) 3098 Worklist.append(N->op_begin(), N->op_end()); 3099 else if (const SCEVCastExpr *C = dyn_cast<SCEVCastExpr>(S)) 3100 Worklist.push_back(C->getOperand()); 3101 else if (const SCEVUDivExpr *D = dyn_cast<SCEVUDivExpr>(S)) { 3102 Worklist.push_back(D->getLHS()); 3103 Worklist.push_back(D->getRHS()); 3104 } else if (const SCEVUnknown *US = dyn_cast<SCEVUnknown>(S)) { 3105 const Value *V = US->getValue(); 3106 if (const Instruction *Inst = dyn_cast<Instruction>(V)) { 3107 // Look for instructions defined outside the loop. 3108 if (L->contains(Inst)) continue; 3109 } else if (isa<UndefValue>(V)) 3110 // Undef doesn't have a live range, so it doesn't matter. 3111 continue; 3112 for (const Use &U : V->uses()) { 3113 const Instruction *UserInst = dyn_cast<Instruction>(U.getUser()); 3114 // Ignore non-instructions. 3115 if (!UserInst) 3116 continue; 3117 // Ignore instructions in other functions (as can happen with 3118 // Constants). 3119 if (UserInst->getParent()->getParent() != L->getHeader()->getParent()) 3120 continue; 3121 // Ignore instructions not dominated by the loop. 3122 const BasicBlock *UseBB = !isa<PHINode>(UserInst) ? 3123 UserInst->getParent() : 3124 cast<PHINode>(UserInst)->getIncomingBlock( 3125 PHINode::getIncomingValueNumForOperand(U.getOperandNo())); 3126 if (!DT.dominates(L->getHeader(), UseBB)) 3127 continue; 3128 // Don't bother if the instruction is in a BB which ends in an EHPad. 3129 if (UseBB->getTerminator()->isEHPad()) 3130 continue; 3131 // Ignore uses which are part of other SCEV expressions, to avoid 3132 // analyzing them multiple times. 3133 if (SE.isSCEVable(UserInst->getType())) { 3134 const SCEV *UserS = SE.getSCEV(const_cast<Instruction *>(UserInst)); 3135 // If the user is a no-op, look through to its uses. 3136 if (!isa<SCEVUnknown>(UserS)) 3137 continue; 3138 if (UserS == US) { 3139 Worklist.push_back( 3140 SE.getUnknown(const_cast<Instruction *>(UserInst))); 3141 continue; 3142 } 3143 } 3144 // Ignore icmp instructions which are already being analyzed. 3145 if (const ICmpInst *ICI = dyn_cast<ICmpInst>(UserInst)) { 3146 unsigned OtherIdx = !U.getOperandNo(); 3147 Value *OtherOp = const_cast<Value *>(ICI->getOperand(OtherIdx)); 3148 if (SE.hasComputableLoopEvolution(SE.getSCEV(OtherOp), L)) 3149 continue; 3150 } 3151 3152 LSRFixup &LF = getNewFixup(); 3153 LF.UserInst = const_cast<Instruction *>(UserInst); 3154 LF.OperandValToReplace = U; 3155 std::pair<size_t, int64_t> P = getUse( 3156 S, LSRUse::Basic, MemAccessTy()); 3157 LF.LUIdx = P.first; 3158 LF.Offset = P.second; 3159 LSRUse &LU = Uses[LF.LUIdx]; 3160 LU.AllFixupsOutsideLoop &= LF.isUseFullyOutsideLoop(L); 3161 if (!LU.WidestFixupType || 3162 SE.getTypeSizeInBits(LU.WidestFixupType) < 3163 SE.getTypeSizeInBits(LF.OperandValToReplace->getType())) 3164 LU.WidestFixupType = LF.OperandValToReplace->getType(); 3165 InsertSupplementalFormula(US, LU, LF.LUIdx); 3166 CountRegisters(LU.Formulae.back(), Uses.size() - 1); 3167 break; 3168 } 3169 } 3170 } 3171 } 3172 3173 /// Split S into subexpressions which can be pulled out into separate 3174 /// registers. If C is non-null, multiply each subexpression by C. 3175 /// 3176 /// Return remainder expression after factoring the subexpressions captured by 3177 /// Ops. If Ops is complete, return NULL. 3178 static const SCEV *CollectSubexprs(const SCEV *S, const SCEVConstant *C, 3179 SmallVectorImpl<const SCEV *> &Ops, 3180 const Loop *L, 3181 ScalarEvolution &SE, 3182 unsigned Depth = 0) { 3183 // Arbitrarily cap recursion to protect compile time. 3184 if (Depth >= 3) 3185 return S; 3186 3187 if (const SCEVAddExpr *Add = dyn_cast<SCEVAddExpr>(S)) { 3188 // Break out add operands. 3189 for (const SCEV *S : Add->operands()) { 3190 const SCEV *Remainder = CollectSubexprs(S, C, Ops, L, SE, Depth+1); 3191 if (Remainder) 3192 Ops.push_back(C ? SE.getMulExpr(C, Remainder) : Remainder); 3193 } 3194 return nullptr; 3195 } else if (const SCEVAddRecExpr *AR = dyn_cast<SCEVAddRecExpr>(S)) { 3196 // Split a non-zero base out of an addrec. 3197 if (AR->getStart()->isZero()) 3198 return S; 3199 3200 const SCEV *Remainder = CollectSubexprs(AR->getStart(), 3201 C, Ops, L, SE, Depth+1); 3202 // Split the non-zero AddRec unless it is part of a nested recurrence that 3203 // does not pertain to this loop. 3204 if (Remainder && (AR->getLoop() == L || !isa<SCEVAddRecExpr>(Remainder))) { 3205 Ops.push_back(C ? SE.getMulExpr(C, Remainder) : Remainder); 3206 Remainder = nullptr; 3207 } 3208 if (Remainder != AR->getStart()) { 3209 if (!Remainder) 3210 Remainder = SE.getConstant(AR->getType(), 0); 3211 return SE.getAddRecExpr(Remainder, 3212 AR->getStepRecurrence(SE), 3213 AR->getLoop(), 3214 //FIXME: AR->getNoWrapFlags(SCEV::FlagNW) 3215 SCEV::FlagAnyWrap); 3216 } 3217 } else if (const SCEVMulExpr *Mul = dyn_cast<SCEVMulExpr>(S)) { 3218 // Break (C * (a + b + c)) into C*a + C*b + C*c. 3219 if (Mul->getNumOperands() != 2) 3220 return S; 3221 if (const SCEVConstant *Op0 = 3222 dyn_cast<SCEVConstant>(Mul->getOperand(0))) { 3223 C = C ? cast<SCEVConstant>(SE.getMulExpr(C, Op0)) : Op0; 3224 const SCEV *Remainder = 3225 CollectSubexprs(Mul->getOperand(1), C, Ops, L, SE, Depth+1); 3226 if (Remainder) 3227 Ops.push_back(SE.getMulExpr(C, Remainder)); 3228 return nullptr; 3229 } 3230 } 3231 return S; 3232 } 3233 3234 /// \brief Helper function for LSRInstance::GenerateReassociations. 3235 void LSRInstance::GenerateReassociationsImpl(LSRUse &LU, unsigned LUIdx, 3236 const Formula &Base, 3237 unsigned Depth, size_t Idx, 3238 bool IsScaledReg) { 3239 const SCEV *BaseReg = IsScaledReg ? Base.ScaledReg : Base.BaseRegs[Idx]; 3240 SmallVector<const SCEV *, 8> AddOps; 3241 const SCEV *Remainder = CollectSubexprs(BaseReg, nullptr, AddOps, L, SE); 3242 if (Remainder) 3243 AddOps.push_back(Remainder); 3244 3245 if (AddOps.size() == 1) 3246 return; 3247 3248 for (SmallVectorImpl<const SCEV *>::const_iterator J = AddOps.begin(), 3249 JE = AddOps.end(); 3250 J != JE; ++J) { 3251 3252 // Loop-variant "unknown" values are uninteresting; we won't be able to 3253 // do anything meaningful with them. 3254 if (isa<SCEVUnknown>(*J) && !SE.isLoopInvariant(*J, L)) 3255 continue; 3256 3257 // Don't pull a constant into a register if the constant could be folded 3258 // into an immediate field. 3259 if (isAlwaysFoldable(TTI, SE, LU.MinOffset, LU.MaxOffset, LU.Kind, 3260 LU.AccessTy, *J, Base.getNumRegs() > 1)) 3261 continue; 3262 3263 // Collect all operands except *J. 3264 SmallVector<const SCEV *, 8> InnerAddOps( 3265 ((const SmallVector<const SCEV *, 8> &)AddOps).begin(), J); 3266 InnerAddOps.append(std::next(J), 3267 ((const SmallVector<const SCEV *, 8> &)AddOps).end()); 3268 3269 // Don't leave just a constant behind in a register if the constant could 3270 // be folded into an immediate field. 3271 if (InnerAddOps.size() == 1 && 3272 isAlwaysFoldable(TTI, SE, LU.MinOffset, LU.MaxOffset, LU.Kind, 3273 LU.AccessTy, InnerAddOps[0], Base.getNumRegs() > 1)) 3274 continue; 3275 3276 const SCEV *InnerSum = SE.getAddExpr(InnerAddOps); 3277 if (InnerSum->isZero()) 3278 continue; 3279 Formula F = Base; 3280 3281 // Add the remaining pieces of the add back into the new formula. 3282 const SCEVConstant *InnerSumSC = dyn_cast<SCEVConstant>(InnerSum); 3283 if (InnerSumSC && SE.getTypeSizeInBits(InnerSumSC->getType()) <= 64 && 3284 TTI.isLegalAddImmediate((uint64_t)F.UnfoldedOffset + 3285 InnerSumSC->getValue()->getZExtValue())) { 3286 F.UnfoldedOffset = 3287 (uint64_t)F.UnfoldedOffset + InnerSumSC->getValue()->getZExtValue(); 3288 if (IsScaledReg) 3289 F.ScaledReg = nullptr; 3290 else 3291 F.BaseRegs.erase(F.BaseRegs.begin() + Idx); 3292 } else if (IsScaledReg) 3293 F.ScaledReg = InnerSum; 3294 else 3295 F.BaseRegs[Idx] = InnerSum; 3296 3297 // Add J as its own register, or an unfolded immediate. 3298 const SCEVConstant *SC = dyn_cast<SCEVConstant>(*J); 3299 if (SC && SE.getTypeSizeInBits(SC->getType()) <= 64 && 3300 TTI.isLegalAddImmediate((uint64_t)F.UnfoldedOffset + 3301 SC->getValue()->getZExtValue())) 3302 F.UnfoldedOffset = 3303 (uint64_t)F.UnfoldedOffset + SC->getValue()->getZExtValue(); 3304 else 3305 F.BaseRegs.push_back(*J); 3306 // We may have changed the number of register in base regs, adjust the 3307 // formula accordingly. 3308 F.canonicalize(); 3309 3310 if (InsertFormula(LU, LUIdx, F)) 3311 // If that formula hadn't been seen before, recurse to find more like 3312 // it. 3313 GenerateReassociations(LU, LUIdx, LU.Formulae.back(), Depth + 1); 3314 } 3315 } 3316 3317 /// Split out subexpressions from adds and the bases of addrecs. 3318 void LSRInstance::GenerateReassociations(LSRUse &LU, unsigned LUIdx, 3319 Formula Base, unsigned Depth) { 3320 assert(Base.isCanonical() && "Input must be in the canonical form"); 3321 // Arbitrarily cap recursion to protect compile time. 3322 if (Depth >= 3) 3323 return; 3324 3325 for (size_t i = 0, e = Base.BaseRegs.size(); i != e; ++i) 3326 GenerateReassociationsImpl(LU, LUIdx, Base, Depth, i); 3327 3328 if (Base.Scale == 1) 3329 GenerateReassociationsImpl(LU, LUIdx, Base, Depth, 3330 /* Idx */ -1, /* IsScaledReg */ true); 3331 } 3332 3333 /// Generate a formula consisting of all of the loop-dominating registers added 3334 /// into a single register. 3335 void LSRInstance::GenerateCombinations(LSRUse &LU, unsigned LUIdx, 3336 Formula Base) { 3337 // This method is only interesting on a plurality of registers. 3338 if (Base.BaseRegs.size() + (Base.Scale == 1) <= 1) 3339 return; 3340 3341 // Flatten the representation, i.e., reg1 + 1*reg2 => reg1 + reg2, before 3342 // processing the formula. 3343 Base.unscale(); 3344 Formula F = Base; 3345 F.BaseRegs.clear(); 3346 SmallVector<const SCEV *, 4> Ops; 3347 for (const SCEV *BaseReg : Base.BaseRegs) { 3348 if (SE.properlyDominates(BaseReg, L->getHeader()) && 3349 !SE.hasComputableLoopEvolution(BaseReg, L)) 3350 Ops.push_back(BaseReg); 3351 else 3352 F.BaseRegs.push_back(BaseReg); 3353 } 3354 if (Ops.size() > 1) { 3355 const SCEV *Sum = SE.getAddExpr(Ops); 3356 // TODO: If Sum is zero, it probably means ScalarEvolution missed an 3357 // opportunity to fold something. For now, just ignore such cases 3358 // rather than proceed with zero in a register. 3359 if (!Sum->isZero()) { 3360 F.BaseRegs.push_back(Sum); 3361 F.canonicalize(); 3362 (void)InsertFormula(LU, LUIdx, F); 3363 } 3364 } 3365 } 3366 3367 /// \brief Helper function for LSRInstance::GenerateSymbolicOffsets. 3368 void LSRInstance::GenerateSymbolicOffsetsImpl(LSRUse &LU, unsigned LUIdx, 3369 const Formula &Base, size_t Idx, 3370 bool IsScaledReg) { 3371 const SCEV *G = IsScaledReg ? Base.ScaledReg : Base.BaseRegs[Idx]; 3372 GlobalValue *GV = ExtractSymbol(G, SE); 3373 if (G->isZero() || !GV) 3374 return; 3375 Formula F = Base; 3376 F.BaseGV = GV; 3377 if (!isLegalUse(TTI, LU.MinOffset, LU.MaxOffset, LU.Kind, LU.AccessTy, F)) 3378 return; 3379 if (IsScaledReg) 3380 F.ScaledReg = G; 3381 else 3382 F.BaseRegs[Idx] = G; 3383 (void)InsertFormula(LU, LUIdx, F); 3384 } 3385 3386 /// Generate reuse formulae using symbolic offsets. 3387 void LSRInstance::GenerateSymbolicOffsets(LSRUse &LU, unsigned LUIdx, 3388 Formula Base) { 3389 // We can't add a symbolic offset if the address already contains one. 3390 if (Base.BaseGV) return; 3391 3392 for (size_t i = 0, e = Base.BaseRegs.size(); i != e; ++i) 3393 GenerateSymbolicOffsetsImpl(LU, LUIdx, Base, i); 3394 if (Base.Scale == 1) 3395 GenerateSymbolicOffsetsImpl(LU, LUIdx, Base, /* Idx */ -1, 3396 /* IsScaledReg */ true); 3397 } 3398 3399 /// \brief Helper function for LSRInstance::GenerateConstantOffsets. 3400 void LSRInstance::GenerateConstantOffsetsImpl( 3401 LSRUse &LU, unsigned LUIdx, const Formula &Base, 3402 const SmallVectorImpl<int64_t> &Worklist, size_t Idx, bool IsScaledReg) { 3403 const SCEV *G = IsScaledReg ? Base.ScaledReg : Base.BaseRegs[Idx]; 3404 for (int64_t Offset : Worklist) { 3405 Formula F = Base; 3406 F.BaseOffset = (uint64_t)Base.BaseOffset - Offset; 3407 if (isLegalUse(TTI, LU.MinOffset - Offset, LU.MaxOffset - Offset, LU.Kind, 3408 LU.AccessTy, F)) { 3409 // Add the offset to the base register. 3410 const SCEV *NewG = SE.getAddExpr(SE.getConstant(G->getType(), Offset), G); 3411 // If it cancelled out, drop the base register, otherwise update it. 3412 if (NewG->isZero()) { 3413 if (IsScaledReg) { 3414 F.Scale = 0; 3415 F.ScaledReg = nullptr; 3416 } else 3417 F.deleteBaseReg(F.BaseRegs[Idx]); 3418 F.canonicalize(); 3419 } else if (IsScaledReg) 3420 F.ScaledReg = NewG; 3421 else 3422 F.BaseRegs[Idx] = NewG; 3423 3424 (void)InsertFormula(LU, LUIdx, F); 3425 } 3426 } 3427 3428 int64_t Imm = ExtractImmediate(G, SE); 3429 if (G->isZero() || Imm == 0) 3430 return; 3431 Formula F = Base; 3432 F.BaseOffset = (uint64_t)F.BaseOffset + Imm; 3433 if (!isLegalUse(TTI, LU.MinOffset, LU.MaxOffset, LU.Kind, LU.AccessTy, F)) 3434 return; 3435 if (IsScaledReg) 3436 F.ScaledReg = G; 3437 else 3438 F.BaseRegs[Idx] = G; 3439 (void)InsertFormula(LU, LUIdx, F); 3440 } 3441 3442 /// GenerateConstantOffsets - Generate reuse formulae using symbolic offsets. 3443 void LSRInstance::GenerateConstantOffsets(LSRUse &LU, unsigned LUIdx, 3444 Formula Base) { 3445 // TODO: For now, just add the min and max offset, because it usually isn't 3446 // worthwhile looking at everything inbetween. 3447 SmallVector<int64_t, 2> Worklist; 3448 Worklist.push_back(LU.MinOffset); 3449 if (LU.MaxOffset != LU.MinOffset) 3450 Worklist.push_back(LU.MaxOffset); 3451 3452 for (size_t i = 0, e = Base.BaseRegs.size(); i != e; ++i) 3453 GenerateConstantOffsetsImpl(LU, LUIdx, Base, Worklist, i); 3454 if (Base.Scale == 1) 3455 GenerateConstantOffsetsImpl(LU, LUIdx, Base, Worklist, /* Idx */ -1, 3456 /* IsScaledReg */ true); 3457 } 3458 3459 /// For ICmpZero, check to see if we can scale up the comparison. For example, x 3460 /// == y -> x*c == y*c. 3461 void LSRInstance::GenerateICmpZeroScales(LSRUse &LU, unsigned LUIdx, 3462 Formula Base) { 3463 if (LU.Kind != LSRUse::ICmpZero) return; 3464 3465 // Determine the integer type for the base formula. 3466 Type *IntTy = Base.getType(); 3467 if (!IntTy) return; 3468 if (SE.getTypeSizeInBits(IntTy) > 64) return; 3469 3470 // Don't do this if there is more than one offset. 3471 if (LU.MinOffset != LU.MaxOffset) return; 3472 3473 assert(!Base.BaseGV && "ICmpZero use is not legal!"); 3474 3475 // Check each interesting stride. 3476 for (int64_t Factor : Factors) { 3477 // Check that the multiplication doesn't overflow. 3478 if (Base.BaseOffset == INT64_MIN && Factor == -1) 3479 continue; 3480 int64_t NewBaseOffset = (uint64_t)Base.BaseOffset * Factor; 3481 if (NewBaseOffset / Factor != Base.BaseOffset) 3482 continue; 3483 // If the offset will be truncated at this use, check that it is in bounds. 3484 if (!IntTy->isPointerTy() && 3485 !ConstantInt::isValueValidForType(IntTy, NewBaseOffset)) 3486 continue; 3487 3488 // Check that multiplying with the use offset doesn't overflow. 3489 int64_t Offset = LU.MinOffset; 3490 if (Offset == INT64_MIN && Factor == -1) 3491 continue; 3492 Offset = (uint64_t)Offset * Factor; 3493 if (Offset / Factor != LU.MinOffset) 3494 continue; 3495 // If the offset will be truncated at this use, check that it is in bounds. 3496 if (!IntTy->isPointerTy() && 3497 !ConstantInt::isValueValidForType(IntTy, Offset)) 3498 continue; 3499 3500 Formula F = Base; 3501 F.BaseOffset = NewBaseOffset; 3502 3503 // Check that this scale is legal. 3504 if (!isLegalUse(TTI, Offset, Offset, LU.Kind, LU.AccessTy, F)) 3505 continue; 3506 3507 // Compensate for the use having MinOffset built into it. 3508 F.BaseOffset = (uint64_t)F.BaseOffset + Offset - LU.MinOffset; 3509 3510 const SCEV *FactorS = SE.getConstant(IntTy, Factor); 3511 3512 // Check that multiplying with each base register doesn't overflow. 3513 for (size_t i = 0, e = F.BaseRegs.size(); i != e; ++i) { 3514 F.BaseRegs[i] = SE.getMulExpr(F.BaseRegs[i], FactorS); 3515 if (getExactSDiv(F.BaseRegs[i], FactorS, SE) != Base.BaseRegs[i]) 3516 goto next; 3517 } 3518 3519 // Check that multiplying with the scaled register doesn't overflow. 3520 if (F.ScaledReg) { 3521 F.ScaledReg = SE.getMulExpr(F.ScaledReg, FactorS); 3522 if (getExactSDiv(F.ScaledReg, FactorS, SE) != Base.ScaledReg) 3523 continue; 3524 } 3525 3526 // Check that multiplying with the unfolded offset doesn't overflow. 3527 if (F.UnfoldedOffset != 0) { 3528 if (F.UnfoldedOffset == INT64_MIN && Factor == -1) 3529 continue; 3530 F.UnfoldedOffset = (uint64_t)F.UnfoldedOffset * Factor; 3531 if (F.UnfoldedOffset / Factor != Base.UnfoldedOffset) 3532 continue; 3533 // If the offset will be truncated, check that it is in bounds. 3534 if (!IntTy->isPointerTy() && 3535 !ConstantInt::isValueValidForType(IntTy, F.UnfoldedOffset)) 3536 continue; 3537 } 3538 3539 // If we make it here and it's legal, add it. 3540 (void)InsertFormula(LU, LUIdx, F); 3541 next:; 3542 } 3543 } 3544 3545 /// Generate stride factor reuse formulae by making use of scaled-offset address 3546 /// modes, for example. 3547 void LSRInstance::GenerateScales(LSRUse &LU, unsigned LUIdx, Formula Base) { 3548 // Determine the integer type for the base formula. 3549 Type *IntTy = Base.getType(); 3550 if (!IntTy) return; 3551 3552 // If this Formula already has a scaled register, we can't add another one. 3553 // Try to unscale the formula to generate a better scale. 3554 if (Base.Scale != 0 && !Base.unscale()) 3555 return; 3556 3557 assert(Base.Scale == 0 && "unscale did not did its job!"); 3558 3559 // Check each interesting stride. 3560 for (int64_t Factor : Factors) { 3561 Base.Scale = Factor; 3562 Base.HasBaseReg = Base.BaseRegs.size() > 1; 3563 // Check whether this scale is going to be legal. 3564 if (!isLegalUse(TTI, LU.MinOffset, LU.MaxOffset, LU.Kind, LU.AccessTy, 3565 Base)) { 3566 // As a special-case, handle special out-of-loop Basic users specially. 3567 // TODO: Reconsider this special case. 3568 if (LU.Kind == LSRUse::Basic && 3569 isLegalUse(TTI, LU.MinOffset, LU.MaxOffset, LSRUse::Special, 3570 LU.AccessTy, Base) && 3571 LU.AllFixupsOutsideLoop) 3572 LU.Kind = LSRUse::Special; 3573 else 3574 continue; 3575 } 3576 // For an ICmpZero, negating a solitary base register won't lead to 3577 // new solutions. 3578 if (LU.Kind == LSRUse::ICmpZero && 3579 !Base.HasBaseReg && Base.BaseOffset == 0 && !Base.BaseGV) 3580 continue; 3581 // For each addrec base reg, apply the scale, if possible. 3582 for (size_t i = 0, e = Base.BaseRegs.size(); i != e; ++i) 3583 if (const SCEVAddRecExpr *AR = 3584 dyn_cast<SCEVAddRecExpr>(Base.BaseRegs[i])) { 3585 const SCEV *FactorS = SE.getConstant(IntTy, Factor); 3586 if (FactorS->isZero()) 3587 continue; 3588 // Divide out the factor, ignoring high bits, since we'll be 3589 // scaling the value back up in the end. 3590 if (const SCEV *Quotient = getExactSDiv(AR, FactorS, SE, true)) { 3591 // TODO: This could be optimized to avoid all the copying. 3592 Formula F = Base; 3593 F.ScaledReg = Quotient; 3594 F.deleteBaseReg(F.BaseRegs[i]); 3595 // The canonical representation of 1*reg is reg, which is already in 3596 // Base. In that case, do not try to insert the formula, it will be 3597 // rejected anyway. 3598 if (F.Scale == 1 && F.BaseRegs.empty()) 3599 continue; 3600 (void)InsertFormula(LU, LUIdx, F); 3601 } 3602 } 3603 } 3604 } 3605 3606 /// Generate reuse formulae from different IV types. 3607 void LSRInstance::GenerateTruncates(LSRUse &LU, unsigned LUIdx, Formula Base) { 3608 // Don't bother truncating symbolic values. 3609 if (Base.BaseGV) return; 3610 3611 // Determine the integer type for the base formula. 3612 Type *DstTy = Base.getType(); 3613 if (!DstTy) return; 3614 DstTy = SE.getEffectiveSCEVType(DstTy); 3615 3616 for (Type *SrcTy : Types) { 3617 if (SrcTy != DstTy && TTI.isTruncateFree(SrcTy, DstTy)) { 3618 Formula F = Base; 3619 3620 if (F.ScaledReg) F.ScaledReg = SE.getAnyExtendExpr(F.ScaledReg, SrcTy); 3621 for (const SCEV *&BaseReg : F.BaseRegs) 3622 BaseReg = SE.getAnyExtendExpr(BaseReg, SrcTy); 3623 3624 // TODO: This assumes we've done basic processing on all uses and 3625 // have an idea what the register usage is. 3626 if (!F.hasRegsUsedByUsesOtherThan(LUIdx, RegUses)) 3627 continue; 3628 3629 (void)InsertFormula(LU, LUIdx, F); 3630 } 3631 } 3632 } 3633 3634 namespace { 3635 3636 /// Helper class for GenerateCrossUseConstantOffsets. It's used to defer 3637 /// modifications so that the search phase doesn't have to worry about the data 3638 /// structures moving underneath it. 3639 struct WorkItem { 3640 size_t LUIdx; 3641 int64_t Imm; 3642 const SCEV *OrigReg; 3643 3644 WorkItem(size_t LI, int64_t I, const SCEV *R) 3645 : LUIdx(LI), Imm(I), OrigReg(R) {} 3646 3647 void print(raw_ostream &OS) const; 3648 void dump() const; 3649 }; 3650 3651 } 3652 3653 void WorkItem::print(raw_ostream &OS) const { 3654 OS << "in formulae referencing " << *OrigReg << " in use " << LUIdx 3655 << " , add offset " << Imm; 3656 } 3657 3658 LLVM_DUMP_METHOD 3659 void WorkItem::dump() const { 3660 print(errs()); errs() << '\n'; 3661 } 3662 3663 /// Look for registers which are a constant distance apart and try to form reuse 3664 /// opportunities between them. 3665 void LSRInstance::GenerateCrossUseConstantOffsets() { 3666 // Group the registers by their value without any added constant offset. 3667 typedef std::map<int64_t, const SCEV *> ImmMapTy; 3668 DenseMap<const SCEV *, ImmMapTy> Map; 3669 DenseMap<const SCEV *, SmallBitVector> UsedByIndicesMap; 3670 SmallVector<const SCEV *, 8> Sequence; 3671 for (const SCEV *Use : RegUses) { 3672 const SCEV *Reg = Use; // Make a copy for ExtractImmediate to modify. 3673 int64_t Imm = ExtractImmediate(Reg, SE); 3674 auto Pair = Map.insert(std::make_pair(Reg, ImmMapTy())); 3675 if (Pair.second) 3676 Sequence.push_back(Reg); 3677 Pair.first->second.insert(std::make_pair(Imm, Use)); 3678 UsedByIndicesMap[Reg] |= RegUses.getUsedByIndices(Use); 3679 } 3680 3681 // Now examine each set of registers with the same base value. Build up 3682 // a list of work to do and do the work in a separate step so that we're 3683 // not adding formulae and register counts while we're searching. 3684 SmallVector<WorkItem, 32> WorkItems; 3685 SmallSet<std::pair<size_t, int64_t>, 32> UniqueItems; 3686 for (const SCEV *Reg : Sequence) { 3687 const ImmMapTy &Imms = Map.find(Reg)->second; 3688 3689 // It's not worthwhile looking for reuse if there's only one offset. 3690 if (Imms.size() == 1) 3691 continue; 3692 3693 DEBUG(dbgs() << "Generating cross-use offsets for " << *Reg << ':'; 3694 for (const auto &Entry : Imms) 3695 dbgs() << ' ' << Entry.first; 3696 dbgs() << '\n'); 3697 3698 // Examine each offset. 3699 for (ImmMapTy::const_iterator J = Imms.begin(), JE = Imms.end(); 3700 J != JE; ++J) { 3701 const SCEV *OrigReg = J->second; 3702 3703 int64_t JImm = J->first; 3704 const SmallBitVector &UsedByIndices = RegUses.getUsedByIndices(OrigReg); 3705 3706 if (!isa<SCEVConstant>(OrigReg) && 3707 UsedByIndicesMap[Reg].count() == 1) { 3708 DEBUG(dbgs() << "Skipping cross-use reuse for " << *OrigReg << '\n'); 3709 continue; 3710 } 3711 3712 // Conservatively examine offsets between this orig reg a few selected 3713 // other orig regs. 3714 ImmMapTy::const_iterator OtherImms[] = { 3715 Imms.begin(), std::prev(Imms.end()), 3716 Imms.lower_bound((Imms.begin()->first + std::prev(Imms.end())->first) / 3717 2) 3718 }; 3719 for (size_t i = 0, e = array_lengthof(OtherImms); i != e; ++i) { 3720 ImmMapTy::const_iterator M = OtherImms[i]; 3721 if (M == J || M == JE) continue; 3722 3723 // Compute the difference between the two. 3724 int64_t Imm = (uint64_t)JImm - M->first; 3725 for (int LUIdx = UsedByIndices.find_first(); LUIdx != -1; 3726 LUIdx = UsedByIndices.find_next(LUIdx)) 3727 // Make a memo of this use, offset, and register tuple. 3728 if (UniqueItems.insert(std::make_pair(LUIdx, Imm)).second) 3729 WorkItems.push_back(WorkItem(LUIdx, Imm, OrigReg)); 3730 } 3731 } 3732 } 3733 3734 Map.clear(); 3735 Sequence.clear(); 3736 UsedByIndicesMap.clear(); 3737 UniqueItems.clear(); 3738 3739 // Now iterate through the worklist and add new formulae. 3740 for (const WorkItem &WI : WorkItems) { 3741 size_t LUIdx = WI.LUIdx; 3742 LSRUse &LU = Uses[LUIdx]; 3743 int64_t Imm = WI.Imm; 3744 const SCEV *OrigReg = WI.OrigReg; 3745 3746 Type *IntTy = SE.getEffectiveSCEVType(OrigReg->getType()); 3747 const SCEV *NegImmS = SE.getSCEV(ConstantInt::get(IntTy, -(uint64_t)Imm)); 3748 unsigned BitWidth = SE.getTypeSizeInBits(IntTy); 3749 3750 // TODO: Use a more targeted data structure. 3751 for (size_t L = 0, LE = LU.Formulae.size(); L != LE; ++L) { 3752 Formula F = LU.Formulae[L]; 3753 // FIXME: The code for the scaled and unscaled registers looks 3754 // very similar but slightly different. Investigate if they 3755 // could be merged. That way, we would not have to unscale the 3756 // Formula. 3757 F.unscale(); 3758 // Use the immediate in the scaled register. 3759 if (F.ScaledReg == OrigReg) { 3760 int64_t Offset = (uint64_t)F.BaseOffset + Imm * (uint64_t)F.Scale; 3761 // Don't create 50 + reg(-50). 3762 if (F.referencesReg(SE.getSCEV( 3763 ConstantInt::get(IntTy, -(uint64_t)Offset)))) 3764 continue; 3765 Formula NewF = F; 3766 NewF.BaseOffset = Offset; 3767 if (!isLegalUse(TTI, LU.MinOffset, LU.MaxOffset, LU.Kind, LU.AccessTy, 3768 NewF)) 3769 continue; 3770 NewF.ScaledReg = SE.getAddExpr(NegImmS, NewF.ScaledReg); 3771 3772 // If the new scale is a constant in a register, and adding the constant 3773 // value to the immediate would produce a value closer to zero than the 3774 // immediate itself, then the formula isn't worthwhile. 3775 if (const SCEVConstant *C = dyn_cast<SCEVConstant>(NewF.ScaledReg)) 3776 if (C->getValue()->isNegative() != (NewF.BaseOffset < 0) && 3777 (C->getAPInt().abs() * APInt(BitWidth, F.Scale)) 3778 .ule(std::abs(NewF.BaseOffset))) 3779 continue; 3780 3781 // OK, looks good. 3782 NewF.canonicalize(); 3783 (void)InsertFormula(LU, LUIdx, NewF); 3784 } else { 3785 // Use the immediate in a base register. 3786 for (size_t N = 0, NE = F.BaseRegs.size(); N != NE; ++N) { 3787 const SCEV *BaseReg = F.BaseRegs[N]; 3788 if (BaseReg != OrigReg) 3789 continue; 3790 Formula NewF = F; 3791 NewF.BaseOffset = (uint64_t)NewF.BaseOffset + Imm; 3792 if (!isLegalUse(TTI, LU.MinOffset, LU.MaxOffset, 3793 LU.Kind, LU.AccessTy, NewF)) { 3794 if (!TTI.isLegalAddImmediate((uint64_t)NewF.UnfoldedOffset + Imm)) 3795 continue; 3796 NewF = F; 3797 NewF.UnfoldedOffset = (uint64_t)NewF.UnfoldedOffset + Imm; 3798 } 3799 NewF.BaseRegs[N] = SE.getAddExpr(NegImmS, BaseReg); 3800 3801 // If the new formula has a constant in a register, and adding the 3802 // constant value to the immediate would produce a value closer to 3803 // zero than the immediate itself, then the formula isn't worthwhile. 3804 for (const SCEV *NewReg : NewF.BaseRegs) 3805 if (const SCEVConstant *C = dyn_cast<SCEVConstant>(NewReg)) 3806 if ((C->getAPInt() + NewF.BaseOffset) 3807 .abs() 3808 .slt(std::abs(NewF.BaseOffset)) && 3809 (C->getAPInt() + NewF.BaseOffset).countTrailingZeros() >= 3810 countTrailingZeros<uint64_t>(NewF.BaseOffset)) 3811 goto skip_formula; 3812 3813 // Ok, looks good. 3814 NewF.canonicalize(); 3815 (void)InsertFormula(LU, LUIdx, NewF); 3816 break; 3817 skip_formula:; 3818 } 3819 } 3820 } 3821 } 3822 } 3823 3824 /// Generate formulae for each use. 3825 void 3826 LSRInstance::GenerateAllReuseFormulae() { 3827 // This is split into multiple loops so that hasRegsUsedByUsesOtherThan 3828 // queries are more precise. 3829 for (size_t LUIdx = 0, NumUses = Uses.size(); LUIdx != NumUses; ++LUIdx) { 3830 LSRUse &LU = Uses[LUIdx]; 3831 for (size_t i = 0, f = LU.Formulae.size(); i != f; ++i) 3832 GenerateReassociations(LU, LUIdx, LU.Formulae[i]); 3833 for (size_t i = 0, f = LU.Formulae.size(); i != f; ++i) 3834 GenerateCombinations(LU, LUIdx, LU.Formulae[i]); 3835 } 3836 for (size_t LUIdx = 0, NumUses = Uses.size(); LUIdx != NumUses; ++LUIdx) { 3837 LSRUse &LU = Uses[LUIdx]; 3838 for (size_t i = 0, f = LU.Formulae.size(); i != f; ++i) 3839 GenerateSymbolicOffsets(LU, LUIdx, LU.Formulae[i]); 3840 for (size_t i = 0, f = LU.Formulae.size(); i != f; ++i) 3841 GenerateConstantOffsets(LU, LUIdx, LU.Formulae[i]); 3842 for (size_t i = 0, f = LU.Formulae.size(); i != f; ++i) 3843 GenerateICmpZeroScales(LU, LUIdx, LU.Formulae[i]); 3844 for (size_t i = 0, f = LU.Formulae.size(); i != f; ++i) 3845 GenerateScales(LU, LUIdx, LU.Formulae[i]); 3846 } 3847 for (size_t LUIdx = 0, NumUses = Uses.size(); LUIdx != NumUses; ++LUIdx) { 3848 LSRUse &LU = Uses[LUIdx]; 3849 for (size_t i = 0, f = LU.Formulae.size(); i != f; ++i) 3850 GenerateTruncates(LU, LUIdx, LU.Formulae[i]); 3851 } 3852 3853 GenerateCrossUseConstantOffsets(); 3854 3855 DEBUG(dbgs() << "\n" 3856 "After generating reuse formulae:\n"; 3857 print_uses(dbgs())); 3858 } 3859 3860 /// If there are multiple formulae with the same set of registers used 3861 /// by other uses, pick the best one and delete the others. 3862 void LSRInstance::FilterOutUndesirableDedicatedRegisters() { 3863 DenseSet<const SCEV *> VisitedRegs; 3864 SmallPtrSet<const SCEV *, 16> Regs; 3865 SmallPtrSet<const SCEV *, 16> LoserRegs; 3866 #ifndef NDEBUG 3867 bool ChangedFormulae = false; 3868 #endif 3869 3870 // Collect the best formula for each unique set of shared registers. This 3871 // is reset for each use. 3872 typedef DenseMap<SmallVector<const SCEV *, 4>, size_t, UniquifierDenseMapInfo> 3873 BestFormulaeTy; 3874 BestFormulaeTy BestFormulae; 3875 3876 for (size_t LUIdx = 0, NumUses = Uses.size(); LUIdx != NumUses; ++LUIdx) { 3877 LSRUse &LU = Uses[LUIdx]; 3878 DEBUG(dbgs() << "Filtering for use "; LU.print(dbgs()); dbgs() << '\n'); 3879 3880 bool Any = false; 3881 for (size_t FIdx = 0, NumForms = LU.Formulae.size(); 3882 FIdx != NumForms; ++FIdx) { 3883 Formula &F = LU.Formulae[FIdx]; 3884 3885 // Some formulas are instant losers. For example, they may depend on 3886 // nonexistent AddRecs from other loops. These need to be filtered 3887 // immediately, otherwise heuristics could choose them over others leading 3888 // to an unsatisfactory solution. Passing LoserRegs into RateFormula here 3889 // avoids the need to recompute this information across formulae using the 3890 // same bad AddRec. Passing LoserRegs is also essential unless we remove 3891 // the corresponding bad register from the Regs set. 3892 Cost CostF; 3893 Regs.clear(); 3894 CostF.RateFormula(TTI, F, Regs, VisitedRegs, L, LU.Offsets, SE, DT, LU, 3895 &LoserRegs); 3896 if (CostF.isLoser()) { 3897 // During initial formula generation, undesirable formulae are generated 3898 // by uses within other loops that have some non-trivial address mode or 3899 // use the postinc form of the IV. LSR needs to provide these formulae 3900 // as the basis of rediscovering the desired formula that uses an AddRec 3901 // corresponding to the existing phi. Once all formulae have been 3902 // generated, these initial losers may be pruned. 3903 DEBUG(dbgs() << " Filtering loser "; F.print(dbgs()); 3904 dbgs() << "\n"); 3905 } 3906 else { 3907 SmallVector<const SCEV *, 4> Key; 3908 for (const SCEV *Reg : F.BaseRegs) { 3909 if (RegUses.isRegUsedByUsesOtherThan(Reg, LUIdx)) 3910 Key.push_back(Reg); 3911 } 3912 if (F.ScaledReg && 3913 RegUses.isRegUsedByUsesOtherThan(F.ScaledReg, LUIdx)) 3914 Key.push_back(F.ScaledReg); 3915 // Unstable sort by host order ok, because this is only used for 3916 // uniquifying. 3917 std::sort(Key.begin(), Key.end()); 3918 3919 std::pair<BestFormulaeTy::const_iterator, bool> P = 3920 BestFormulae.insert(std::make_pair(Key, FIdx)); 3921 if (P.second) 3922 continue; 3923 3924 Formula &Best = LU.Formulae[P.first->second]; 3925 3926 Cost CostBest; 3927 Regs.clear(); 3928 CostBest.RateFormula(TTI, Best, Regs, VisitedRegs, L, LU.Offsets, SE, 3929 DT, LU); 3930 if (CostF < CostBest) 3931 std::swap(F, Best); 3932 DEBUG(dbgs() << " Filtering out formula "; F.print(dbgs()); 3933 dbgs() << "\n" 3934 " in favor of formula "; Best.print(dbgs()); 3935 dbgs() << '\n'); 3936 } 3937 #ifndef NDEBUG 3938 ChangedFormulae = true; 3939 #endif 3940 LU.DeleteFormula(F); 3941 --FIdx; 3942 --NumForms; 3943 Any = true; 3944 } 3945 3946 // Now that we've filtered out some formulae, recompute the Regs set. 3947 if (Any) 3948 LU.RecomputeRegs(LUIdx, RegUses); 3949 3950 // Reset this to prepare for the next use. 3951 BestFormulae.clear(); 3952 } 3953 3954 DEBUG(if (ChangedFormulae) { 3955 dbgs() << "\n" 3956 "After filtering out undesirable candidates:\n"; 3957 print_uses(dbgs()); 3958 }); 3959 } 3960 3961 // This is a rough guess that seems to work fairly well. 3962 static const size_t ComplexityLimit = UINT16_MAX; 3963 3964 /// Estimate the worst-case number of solutions the solver might have to 3965 /// consider. It almost never considers this many solutions because it prune the 3966 /// search space, but the pruning isn't always sufficient. 3967 size_t LSRInstance::EstimateSearchSpaceComplexity() const { 3968 size_t Power = 1; 3969 for (const LSRUse &LU : Uses) { 3970 size_t FSize = LU.Formulae.size(); 3971 if (FSize >= ComplexityLimit) { 3972 Power = ComplexityLimit; 3973 break; 3974 } 3975 Power *= FSize; 3976 if (Power >= ComplexityLimit) 3977 break; 3978 } 3979 return Power; 3980 } 3981 3982 /// When one formula uses a superset of the registers of another formula, it 3983 /// won't help reduce register pressure (though it may not necessarily hurt 3984 /// register pressure); remove it to simplify the system. 3985 void LSRInstance::NarrowSearchSpaceByDetectingSupersets() { 3986 if (EstimateSearchSpaceComplexity() >= ComplexityLimit) { 3987 DEBUG(dbgs() << "The search space is too complex.\n"); 3988 3989 DEBUG(dbgs() << "Narrowing the search space by eliminating formulae " 3990 "which use a superset of registers used by other " 3991 "formulae.\n"); 3992 3993 for (size_t LUIdx = 0, NumUses = Uses.size(); LUIdx != NumUses; ++LUIdx) { 3994 LSRUse &LU = Uses[LUIdx]; 3995 bool Any = false; 3996 for (size_t i = 0, e = LU.Formulae.size(); i != e; ++i) { 3997 Formula &F = LU.Formulae[i]; 3998 // Look for a formula with a constant or GV in a register. If the use 3999 // also has a formula with that same value in an immediate field, 4000 // delete the one that uses a register. 4001 for (SmallVectorImpl<const SCEV *>::const_iterator 4002 I = F.BaseRegs.begin(), E = F.BaseRegs.end(); I != E; ++I) { 4003 if (const SCEVConstant *C = dyn_cast<SCEVConstant>(*I)) { 4004 Formula NewF = F; 4005 NewF.BaseOffset += C->getValue()->getSExtValue(); 4006 NewF.BaseRegs.erase(NewF.BaseRegs.begin() + 4007 (I - F.BaseRegs.begin())); 4008 if (LU.HasFormulaWithSameRegs(NewF)) { 4009 DEBUG(dbgs() << " Deleting "; F.print(dbgs()); dbgs() << '\n'); 4010 LU.DeleteFormula(F); 4011 --i; 4012 --e; 4013 Any = true; 4014 break; 4015 } 4016 } else if (const SCEVUnknown *U = dyn_cast<SCEVUnknown>(*I)) { 4017 if (GlobalValue *GV = dyn_cast<GlobalValue>(U->getValue())) 4018 if (!F.BaseGV) { 4019 Formula NewF = F; 4020 NewF.BaseGV = GV; 4021 NewF.BaseRegs.erase(NewF.BaseRegs.begin() + 4022 (I - F.BaseRegs.begin())); 4023 if (LU.HasFormulaWithSameRegs(NewF)) { 4024 DEBUG(dbgs() << " Deleting "; F.print(dbgs()); 4025 dbgs() << '\n'); 4026 LU.DeleteFormula(F); 4027 --i; 4028 --e; 4029 Any = true; 4030 break; 4031 } 4032 } 4033 } 4034 } 4035 } 4036 if (Any) 4037 LU.RecomputeRegs(LUIdx, RegUses); 4038 } 4039 4040 DEBUG(dbgs() << "After pre-selection:\n"; 4041 print_uses(dbgs())); 4042 } 4043 } 4044 4045 /// When there are many registers for expressions like A, A+1, A+2, etc., 4046 /// allocate a single register for them. 4047 void LSRInstance::NarrowSearchSpaceByCollapsingUnrolledCode() { 4048 if (EstimateSearchSpaceComplexity() < ComplexityLimit) 4049 return; 4050 4051 DEBUG(dbgs() << "The search space is too complex.\n" 4052 "Narrowing the search space by assuming that uses separated " 4053 "by a constant offset will use the same registers.\n"); 4054 4055 // This is especially useful for unrolled loops. 4056 4057 for (size_t LUIdx = 0, NumUses = Uses.size(); LUIdx != NumUses; ++LUIdx) { 4058 LSRUse &LU = Uses[LUIdx]; 4059 for (const Formula &F : LU.Formulae) { 4060 if (F.BaseOffset == 0 || (F.Scale != 0 && F.Scale != 1)) 4061 continue; 4062 4063 LSRUse *LUThatHas = FindUseWithSimilarFormula(F, LU); 4064 if (!LUThatHas) 4065 continue; 4066 4067 if (!reconcileNewOffset(*LUThatHas, F.BaseOffset, /*HasBaseReg=*/ false, 4068 LU.Kind, LU.AccessTy)) 4069 continue; 4070 4071 DEBUG(dbgs() << " Deleting use "; LU.print(dbgs()); dbgs() << '\n'); 4072 4073 LUThatHas->AllFixupsOutsideLoop &= LU.AllFixupsOutsideLoop; 4074 4075 // Update the relocs to reference the new use. 4076 for (LSRFixup &Fixup : Fixups) { 4077 if (Fixup.LUIdx == LUIdx) { 4078 Fixup.LUIdx = LUThatHas - &Uses.front(); 4079 Fixup.Offset += F.BaseOffset; 4080 // Add the new offset to LUThatHas' offset list. 4081 if (LUThatHas->Offsets.back() != Fixup.Offset) { 4082 LUThatHas->Offsets.push_back(Fixup.Offset); 4083 if (Fixup.Offset > LUThatHas->MaxOffset) 4084 LUThatHas->MaxOffset = Fixup.Offset; 4085 if (Fixup.Offset < LUThatHas->MinOffset) 4086 LUThatHas->MinOffset = Fixup.Offset; 4087 } 4088 DEBUG(dbgs() << "New fixup has offset " << Fixup.Offset << '\n'); 4089 } 4090 if (Fixup.LUIdx == NumUses-1) 4091 Fixup.LUIdx = LUIdx; 4092 } 4093 4094 // Delete formulae from the new use which are no longer legal. 4095 bool Any = false; 4096 for (size_t i = 0, e = LUThatHas->Formulae.size(); i != e; ++i) { 4097 Formula &F = LUThatHas->Formulae[i]; 4098 if (!isLegalUse(TTI, LUThatHas->MinOffset, LUThatHas->MaxOffset, 4099 LUThatHas->Kind, LUThatHas->AccessTy, F)) { 4100 DEBUG(dbgs() << " Deleting "; F.print(dbgs()); 4101 dbgs() << '\n'); 4102 LUThatHas->DeleteFormula(F); 4103 --i; 4104 --e; 4105 Any = true; 4106 } 4107 } 4108 4109 if (Any) 4110 LUThatHas->RecomputeRegs(LUThatHas - &Uses.front(), RegUses); 4111 4112 // Delete the old use. 4113 DeleteUse(LU, LUIdx); 4114 --LUIdx; 4115 --NumUses; 4116 break; 4117 } 4118 } 4119 4120 DEBUG(dbgs() << "After pre-selection:\n"; print_uses(dbgs())); 4121 } 4122 4123 /// Call FilterOutUndesirableDedicatedRegisters again, if necessary, now that 4124 /// we've done more filtering, as it may be able to find more formulae to 4125 /// eliminate. 4126 void LSRInstance::NarrowSearchSpaceByRefilteringUndesirableDedicatedRegisters(){ 4127 if (EstimateSearchSpaceComplexity() >= ComplexityLimit) { 4128 DEBUG(dbgs() << "The search space is too complex.\n"); 4129 4130 DEBUG(dbgs() << "Narrowing the search space by re-filtering out " 4131 "undesirable dedicated registers.\n"); 4132 4133 FilterOutUndesirableDedicatedRegisters(); 4134 4135 DEBUG(dbgs() << "After pre-selection:\n"; 4136 print_uses(dbgs())); 4137 } 4138 } 4139 4140 /// Pick a register which seems likely to be profitable, and then in any use 4141 /// which has any reference to that register, delete all formulae which do not 4142 /// reference that register. 4143 void LSRInstance::NarrowSearchSpaceByPickingWinnerRegs() { 4144 // With all other options exhausted, loop until the system is simple 4145 // enough to handle. 4146 SmallPtrSet<const SCEV *, 4> Taken; 4147 while (EstimateSearchSpaceComplexity() >= ComplexityLimit) { 4148 // Ok, we have too many of formulae on our hands to conveniently handle. 4149 // Use a rough heuristic to thin out the list. 4150 DEBUG(dbgs() << "The search space is too complex.\n"); 4151 4152 // Pick the register which is used by the most LSRUses, which is likely 4153 // to be a good reuse register candidate. 4154 const SCEV *Best = nullptr; 4155 unsigned BestNum = 0; 4156 for (const SCEV *Reg : RegUses) { 4157 if (Taken.count(Reg)) 4158 continue; 4159 if (!Best) 4160 Best = Reg; 4161 else { 4162 unsigned Count = RegUses.getUsedByIndices(Reg).count(); 4163 if (Count > BestNum) { 4164 Best = Reg; 4165 BestNum = Count; 4166 } 4167 } 4168 } 4169 4170 DEBUG(dbgs() << "Narrowing the search space by assuming " << *Best 4171 << " will yield profitable reuse.\n"); 4172 Taken.insert(Best); 4173 4174 // In any use with formulae which references this register, delete formulae 4175 // which don't reference it. 4176 for (size_t LUIdx = 0, NumUses = Uses.size(); LUIdx != NumUses; ++LUIdx) { 4177 LSRUse &LU = Uses[LUIdx]; 4178 if (!LU.Regs.count(Best)) continue; 4179 4180 bool Any = false; 4181 for (size_t i = 0, e = LU.Formulae.size(); i != e; ++i) { 4182 Formula &F = LU.Formulae[i]; 4183 if (!F.referencesReg(Best)) { 4184 DEBUG(dbgs() << " Deleting "; F.print(dbgs()); dbgs() << '\n'); 4185 LU.DeleteFormula(F); 4186 --e; 4187 --i; 4188 Any = true; 4189 assert(e != 0 && "Use has no formulae left! Is Regs inconsistent?"); 4190 continue; 4191 } 4192 } 4193 4194 if (Any) 4195 LU.RecomputeRegs(LUIdx, RegUses); 4196 } 4197 4198 DEBUG(dbgs() << "After pre-selection:\n"; 4199 print_uses(dbgs())); 4200 } 4201 } 4202 4203 /// If there are an extraordinary number of formulae to choose from, use some 4204 /// rough heuristics to prune down the number of formulae. This keeps the main 4205 /// solver from taking an extraordinary amount of time in some worst-case 4206 /// scenarios. 4207 void LSRInstance::NarrowSearchSpaceUsingHeuristics() { 4208 NarrowSearchSpaceByDetectingSupersets(); 4209 NarrowSearchSpaceByCollapsingUnrolledCode(); 4210 NarrowSearchSpaceByRefilteringUndesirableDedicatedRegisters(); 4211 NarrowSearchSpaceByPickingWinnerRegs(); 4212 } 4213 4214 /// This is the recursive solver. 4215 void LSRInstance::SolveRecurse(SmallVectorImpl<const Formula *> &Solution, 4216 Cost &SolutionCost, 4217 SmallVectorImpl<const Formula *> &Workspace, 4218 const Cost &CurCost, 4219 const SmallPtrSet<const SCEV *, 16> &CurRegs, 4220 DenseSet<const SCEV *> &VisitedRegs) const { 4221 // Some ideas: 4222 // - prune more: 4223 // - use more aggressive filtering 4224 // - sort the formula so that the most profitable solutions are found first 4225 // - sort the uses too 4226 // - search faster: 4227 // - don't compute a cost, and then compare. compare while computing a cost 4228 // and bail early. 4229 // - track register sets with SmallBitVector 4230 4231 const LSRUse &LU = Uses[Workspace.size()]; 4232 4233 // If this use references any register that's already a part of the 4234 // in-progress solution, consider it a requirement that a formula must 4235 // reference that register in order to be considered. This prunes out 4236 // unprofitable searching. 4237 SmallSetVector<const SCEV *, 4> ReqRegs; 4238 for (const SCEV *S : CurRegs) 4239 if (LU.Regs.count(S)) 4240 ReqRegs.insert(S); 4241 4242 SmallPtrSet<const SCEV *, 16> NewRegs; 4243 Cost NewCost; 4244 for (const Formula &F : LU.Formulae) { 4245 // Ignore formulae which may not be ideal in terms of register reuse of 4246 // ReqRegs. The formula should use all required registers before 4247 // introducing new ones. 4248 int NumReqRegsToFind = std::min(F.getNumRegs(), ReqRegs.size()); 4249 for (const SCEV *Reg : ReqRegs) { 4250 if ((F.ScaledReg && F.ScaledReg == Reg) || 4251 std::find(F.BaseRegs.begin(), F.BaseRegs.end(), Reg) != 4252 F.BaseRegs.end()) { 4253 --NumReqRegsToFind; 4254 if (NumReqRegsToFind == 0) 4255 break; 4256 } 4257 } 4258 if (NumReqRegsToFind != 0) { 4259 // If none of the formulae satisfied the required registers, then we could 4260 // clear ReqRegs and try again. Currently, we simply give up in this case. 4261 continue; 4262 } 4263 4264 // Evaluate the cost of the current formula. If it's already worse than 4265 // the current best, prune the search at that point. 4266 NewCost = CurCost; 4267 NewRegs = CurRegs; 4268 NewCost.RateFormula(TTI, F, NewRegs, VisitedRegs, L, LU.Offsets, SE, DT, 4269 LU); 4270 if (NewCost < SolutionCost) { 4271 Workspace.push_back(&F); 4272 if (Workspace.size() != Uses.size()) { 4273 SolveRecurse(Solution, SolutionCost, Workspace, NewCost, 4274 NewRegs, VisitedRegs); 4275 if (F.getNumRegs() == 1 && Workspace.size() == 1) 4276 VisitedRegs.insert(F.ScaledReg ? F.ScaledReg : F.BaseRegs[0]); 4277 } else { 4278 DEBUG(dbgs() << "New best at "; NewCost.print(dbgs()); 4279 dbgs() << ".\n Regs:"; 4280 for (const SCEV *S : NewRegs) 4281 dbgs() << ' ' << *S; 4282 dbgs() << '\n'); 4283 4284 SolutionCost = NewCost; 4285 Solution = Workspace; 4286 } 4287 Workspace.pop_back(); 4288 } 4289 } 4290 } 4291 4292 /// Choose one formula from each use. Return the results in the given Solution 4293 /// vector. 4294 void LSRInstance::Solve(SmallVectorImpl<const Formula *> &Solution) const { 4295 SmallVector<const Formula *, 8> Workspace; 4296 Cost SolutionCost; 4297 SolutionCost.Lose(); 4298 Cost CurCost; 4299 SmallPtrSet<const SCEV *, 16> CurRegs; 4300 DenseSet<const SCEV *> VisitedRegs; 4301 Workspace.reserve(Uses.size()); 4302 4303 // SolveRecurse does all the work. 4304 SolveRecurse(Solution, SolutionCost, Workspace, CurCost, 4305 CurRegs, VisitedRegs); 4306 if (Solution.empty()) { 4307 DEBUG(dbgs() << "\nNo Satisfactory Solution\n"); 4308 return; 4309 } 4310 4311 // Ok, we've now made all our decisions. 4312 DEBUG(dbgs() << "\n" 4313 "The chosen solution requires "; SolutionCost.print(dbgs()); 4314 dbgs() << ":\n"; 4315 for (size_t i = 0, e = Uses.size(); i != e; ++i) { 4316 dbgs() << " "; 4317 Uses[i].print(dbgs()); 4318 dbgs() << "\n" 4319 " "; 4320 Solution[i]->print(dbgs()); 4321 dbgs() << '\n'; 4322 }); 4323 4324 assert(Solution.size() == Uses.size() && "Malformed solution!"); 4325 } 4326 4327 /// Helper for AdjustInsertPositionForExpand. Climb up the dominator tree far as 4328 /// we can go while still being dominated by the input positions. This helps 4329 /// canonicalize the insert position, which encourages sharing. 4330 BasicBlock::iterator 4331 LSRInstance::HoistInsertPosition(BasicBlock::iterator IP, 4332 const SmallVectorImpl<Instruction *> &Inputs) 4333 const { 4334 for (;;) { 4335 const Loop *IPLoop = LI.getLoopFor(IP->getParent()); 4336 unsigned IPLoopDepth = IPLoop ? IPLoop->getLoopDepth() : 0; 4337 4338 BasicBlock *IDom; 4339 for (DomTreeNode *Rung = DT.getNode(IP->getParent()); ; ) { 4340 if (!Rung) return IP; 4341 Rung = Rung->getIDom(); 4342 if (!Rung) return IP; 4343 IDom = Rung->getBlock(); 4344 4345 // Don't climb into a loop though. 4346 const Loop *IDomLoop = LI.getLoopFor(IDom); 4347 unsigned IDomDepth = IDomLoop ? IDomLoop->getLoopDepth() : 0; 4348 if (IDomDepth <= IPLoopDepth && 4349 (IDomDepth != IPLoopDepth || IDomLoop == IPLoop)) 4350 break; 4351 } 4352 4353 bool AllDominate = true; 4354 Instruction *BetterPos = nullptr; 4355 Instruction *Tentative = IDom->getTerminator(); 4356 for (Instruction *Inst : Inputs) { 4357 if (Inst == Tentative || !DT.dominates(Inst, Tentative)) { 4358 AllDominate = false; 4359 break; 4360 } 4361 // Attempt to find an insert position in the middle of the block, 4362 // instead of at the end, so that it can be used for other expansions. 4363 if (IDom == Inst->getParent() && 4364 (!BetterPos || !DT.dominates(Inst, BetterPos))) 4365 BetterPos = &*std::next(BasicBlock::iterator(Inst)); 4366 } 4367 if (!AllDominate) 4368 break; 4369 if (BetterPos) 4370 IP = BetterPos->getIterator(); 4371 else 4372 IP = Tentative->getIterator(); 4373 } 4374 4375 return IP; 4376 } 4377 4378 /// Determine an input position which will be dominated by the operands and 4379 /// which will dominate the result. 4380 BasicBlock::iterator 4381 LSRInstance::AdjustInsertPositionForExpand(BasicBlock::iterator LowestIP, 4382 const LSRFixup &LF, 4383 const LSRUse &LU, 4384 SCEVExpander &Rewriter) const { 4385 // Collect some instructions which must be dominated by the 4386 // expanding replacement. These must be dominated by any operands that 4387 // will be required in the expansion. 4388 SmallVector<Instruction *, 4> Inputs; 4389 if (Instruction *I = dyn_cast<Instruction>(LF.OperandValToReplace)) 4390 Inputs.push_back(I); 4391 if (LU.Kind == LSRUse::ICmpZero) 4392 if (Instruction *I = 4393 dyn_cast<Instruction>(cast<ICmpInst>(LF.UserInst)->getOperand(1))) 4394 Inputs.push_back(I); 4395 if (LF.PostIncLoops.count(L)) { 4396 if (LF.isUseFullyOutsideLoop(L)) 4397 Inputs.push_back(L->getLoopLatch()->getTerminator()); 4398 else 4399 Inputs.push_back(IVIncInsertPos); 4400 } 4401 // The expansion must also be dominated by the increment positions of any 4402 // loops it for which it is using post-inc mode. 4403 for (const Loop *PIL : LF.PostIncLoops) { 4404 if (PIL == L) continue; 4405 4406 // Be dominated by the loop exit. 4407 SmallVector<BasicBlock *, 4> ExitingBlocks; 4408 PIL->getExitingBlocks(ExitingBlocks); 4409 if (!ExitingBlocks.empty()) { 4410 BasicBlock *BB = ExitingBlocks[0]; 4411 for (unsigned i = 1, e = ExitingBlocks.size(); i != e; ++i) 4412 BB = DT.findNearestCommonDominator(BB, ExitingBlocks[i]); 4413 Inputs.push_back(BB->getTerminator()); 4414 } 4415 } 4416 4417 assert(!isa<PHINode>(LowestIP) && !LowestIP->isEHPad() 4418 && !isa<DbgInfoIntrinsic>(LowestIP) && 4419 "Insertion point must be a normal instruction"); 4420 4421 // Then, climb up the immediate dominator tree as far as we can go while 4422 // still being dominated by the input positions. 4423 BasicBlock::iterator IP = HoistInsertPosition(LowestIP, Inputs); 4424 4425 // Don't insert instructions before PHI nodes. 4426 while (isa<PHINode>(IP)) ++IP; 4427 4428 // Ignore landingpad instructions. 4429 while (!isa<TerminatorInst>(IP) && IP->isEHPad()) ++IP; 4430 4431 // Ignore debug intrinsics. 4432 while (isa<DbgInfoIntrinsic>(IP)) ++IP; 4433 4434 // Set IP below instructions recently inserted by SCEVExpander. This keeps the 4435 // IP consistent across expansions and allows the previously inserted 4436 // instructions to be reused by subsequent expansion. 4437 while (Rewriter.isInsertedInstruction(&*IP) && IP != LowestIP) 4438 ++IP; 4439 4440 return IP; 4441 } 4442 4443 /// Emit instructions for the leading candidate expression for this LSRUse (this 4444 /// is called "expanding"). 4445 Value *LSRInstance::Expand(const LSRFixup &LF, 4446 const Formula &F, 4447 BasicBlock::iterator IP, 4448 SCEVExpander &Rewriter, 4449 SmallVectorImpl<WeakVH> &DeadInsts) const { 4450 const LSRUse &LU = Uses[LF.LUIdx]; 4451 if (LU.RigidFormula) 4452 return LF.OperandValToReplace; 4453 4454 // Determine an input position which will be dominated by the operands and 4455 // which will dominate the result. 4456 IP = AdjustInsertPositionForExpand(IP, LF, LU, Rewriter); 4457 4458 // Inform the Rewriter if we have a post-increment use, so that it can 4459 // perform an advantageous expansion. 4460 Rewriter.setPostInc(LF.PostIncLoops); 4461 4462 // This is the type that the user actually needs. 4463 Type *OpTy = LF.OperandValToReplace->getType(); 4464 // This will be the type that we'll initially expand to. 4465 Type *Ty = F.getType(); 4466 if (!Ty) 4467 // No type known; just expand directly to the ultimate type. 4468 Ty = OpTy; 4469 else if (SE.getEffectiveSCEVType(Ty) == SE.getEffectiveSCEVType(OpTy)) 4470 // Expand directly to the ultimate type if it's the right size. 4471 Ty = OpTy; 4472 // This is the type to do integer arithmetic in. 4473 Type *IntTy = SE.getEffectiveSCEVType(Ty); 4474 4475 // Build up a list of operands to add together to form the full base. 4476 SmallVector<const SCEV *, 8> Ops; 4477 4478 // Expand the BaseRegs portion. 4479 for (const SCEV *Reg : F.BaseRegs) { 4480 assert(!Reg->isZero() && "Zero allocated in a base register!"); 4481 4482 // If we're expanding for a post-inc user, make the post-inc adjustment. 4483 PostIncLoopSet &Loops = const_cast<PostIncLoopSet &>(LF.PostIncLoops); 4484 Reg = TransformForPostIncUse(Denormalize, Reg, 4485 LF.UserInst, LF.OperandValToReplace, 4486 Loops, SE, DT); 4487 4488 Ops.push_back(SE.getUnknown(Rewriter.expandCodeFor(Reg, nullptr, &*IP))); 4489 } 4490 4491 // Expand the ScaledReg portion. 4492 Value *ICmpScaledV = nullptr; 4493 if (F.Scale != 0) { 4494 const SCEV *ScaledS = F.ScaledReg; 4495 4496 // If we're expanding for a post-inc user, make the post-inc adjustment. 4497 PostIncLoopSet &Loops = const_cast<PostIncLoopSet &>(LF.PostIncLoops); 4498 ScaledS = TransformForPostIncUse(Denormalize, ScaledS, 4499 LF.UserInst, LF.OperandValToReplace, 4500 Loops, SE, DT); 4501 4502 if (LU.Kind == LSRUse::ICmpZero) { 4503 // Expand ScaleReg as if it was part of the base regs. 4504 if (F.Scale == 1) 4505 Ops.push_back( 4506 SE.getUnknown(Rewriter.expandCodeFor(ScaledS, nullptr, &*IP))); 4507 else { 4508 // An interesting way of "folding" with an icmp is to use a negated 4509 // scale, which we'll implement by inserting it into the other operand 4510 // of the icmp. 4511 assert(F.Scale == -1 && 4512 "The only scale supported by ICmpZero uses is -1!"); 4513 ICmpScaledV = Rewriter.expandCodeFor(ScaledS, nullptr, &*IP); 4514 } 4515 } else { 4516 // Otherwise just expand the scaled register and an explicit scale, 4517 // which is expected to be matched as part of the address. 4518 4519 // Flush the operand list to suppress SCEVExpander hoisting address modes. 4520 // Unless the addressing mode will not be folded. 4521 if (!Ops.empty() && LU.Kind == LSRUse::Address && 4522 isAMCompletelyFolded(TTI, LU, F)) { 4523 Value *FullV = Rewriter.expandCodeFor(SE.getAddExpr(Ops), Ty, &*IP); 4524 Ops.clear(); 4525 Ops.push_back(SE.getUnknown(FullV)); 4526 } 4527 ScaledS = SE.getUnknown(Rewriter.expandCodeFor(ScaledS, nullptr, &*IP)); 4528 if (F.Scale != 1) 4529 ScaledS = 4530 SE.getMulExpr(ScaledS, SE.getConstant(ScaledS->getType(), F.Scale)); 4531 Ops.push_back(ScaledS); 4532 } 4533 } 4534 4535 // Expand the GV portion. 4536 if (F.BaseGV) { 4537 // Flush the operand list to suppress SCEVExpander hoisting. 4538 if (!Ops.empty()) { 4539 Value *FullV = Rewriter.expandCodeFor(SE.getAddExpr(Ops), Ty, &*IP); 4540 Ops.clear(); 4541 Ops.push_back(SE.getUnknown(FullV)); 4542 } 4543 Ops.push_back(SE.getUnknown(F.BaseGV)); 4544 } 4545 4546 // Flush the operand list to suppress SCEVExpander hoisting of both folded and 4547 // unfolded offsets. LSR assumes they both live next to their uses. 4548 if (!Ops.empty()) { 4549 Value *FullV = Rewriter.expandCodeFor(SE.getAddExpr(Ops), Ty, &*IP); 4550 Ops.clear(); 4551 Ops.push_back(SE.getUnknown(FullV)); 4552 } 4553 4554 // Expand the immediate portion. 4555 int64_t Offset = (uint64_t)F.BaseOffset + LF.Offset; 4556 if (Offset != 0) { 4557 if (LU.Kind == LSRUse::ICmpZero) { 4558 // The other interesting way of "folding" with an ICmpZero is to use a 4559 // negated immediate. 4560 if (!ICmpScaledV) 4561 ICmpScaledV = ConstantInt::get(IntTy, -(uint64_t)Offset); 4562 else { 4563 Ops.push_back(SE.getUnknown(ICmpScaledV)); 4564 ICmpScaledV = ConstantInt::get(IntTy, Offset); 4565 } 4566 } else { 4567 // Just add the immediate values. These again are expected to be matched 4568 // as part of the address. 4569 Ops.push_back(SE.getUnknown(ConstantInt::getSigned(IntTy, Offset))); 4570 } 4571 } 4572 4573 // Expand the unfolded offset portion. 4574 int64_t UnfoldedOffset = F.UnfoldedOffset; 4575 if (UnfoldedOffset != 0) { 4576 // Just add the immediate values. 4577 Ops.push_back(SE.getUnknown(ConstantInt::getSigned(IntTy, 4578 UnfoldedOffset))); 4579 } 4580 4581 // Emit instructions summing all the operands. 4582 const SCEV *FullS = Ops.empty() ? 4583 SE.getConstant(IntTy, 0) : 4584 SE.getAddExpr(Ops); 4585 Value *FullV = Rewriter.expandCodeFor(FullS, Ty, &*IP); 4586 4587 // We're done expanding now, so reset the rewriter. 4588 Rewriter.clearPostInc(); 4589 4590 // An ICmpZero Formula represents an ICmp which we're handling as a 4591 // comparison against zero. Now that we've expanded an expression for that 4592 // form, update the ICmp's other operand. 4593 if (LU.Kind == LSRUse::ICmpZero) { 4594 ICmpInst *CI = cast<ICmpInst>(LF.UserInst); 4595 DeadInsts.emplace_back(CI->getOperand(1)); 4596 assert(!F.BaseGV && "ICmp does not support folding a global value and " 4597 "a scale at the same time!"); 4598 if (F.Scale == -1) { 4599 if (ICmpScaledV->getType() != OpTy) { 4600 Instruction *Cast = 4601 CastInst::Create(CastInst::getCastOpcode(ICmpScaledV, false, 4602 OpTy, false), 4603 ICmpScaledV, OpTy, "tmp", CI); 4604 ICmpScaledV = Cast; 4605 } 4606 CI->setOperand(1, ICmpScaledV); 4607 } else { 4608 // A scale of 1 means that the scale has been expanded as part of the 4609 // base regs. 4610 assert((F.Scale == 0 || F.Scale == 1) && 4611 "ICmp does not support folding a global value and " 4612 "a scale at the same time!"); 4613 Constant *C = ConstantInt::getSigned(SE.getEffectiveSCEVType(OpTy), 4614 -(uint64_t)Offset); 4615 if (C->getType() != OpTy) 4616 C = ConstantExpr::getCast(CastInst::getCastOpcode(C, false, 4617 OpTy, false), 4618 C, OpTy); 4619 4620 CI->setOperand(1, C); 4621 } 4622 } 4623 4624 return FullV; 4625 } 4626 4627 /// Helper for Rewrite. PHI nodes are special because the use of their operands 4628 /// effectively happens in their predecessor blocks, so the expression may need 4629 /// to be expanded in multiple places. 4630 void LSRInstance::RewriteForPHI(PHINode *PN, 4631 const LSRFixup &LF, 4632 const Formula &F, 4633 SCEVExpander &Rewriter, 4634 SmallVectorImpl<WeakVH> &DeadInsts) const { 4635 DenseMap<BasicBlock *, Value *> Inserted; 4636 for (unsigned i = 0, e = PN->getNumIncomingValues(); i != e; ++i) 4637 if (PN->getIncomingValue(i) == LF.OperandValToReplace) { 4638 BasicBlock *BB = PN->getIncomingBlock(i); 4639 4640 // If this is a critical edge, split the edge so that we do not insert 4641 // the code on all predecessor/successor paths. We do this unless this 4642 // is the canonical backedge for this loop, which complicates post-inc 4643 // users. 4644 if (e != 1 && BB->getTerminator()->getNumSuccessors() > 1 && 4645 !isa<IndirectBrInst>(BB->getTerminator())) { 4646 BasicBlock *Parent = PN->getParent(); 4647 Loop *PNLoop = LI.getLoopFor(Parent); 4648 if (!PNLoop || Parent != PNLoop->getHeader()) { 4649 // Split the critical edge. 4650 BasicBlock *NewBB = nullptr; 4651 if (!Parent->isLandingPad()) { 4652 NewBB = SplitCriticalEdge(BB, Parent, 4653 CriticalEdgeSplittingOptions(&DT, &LI) 4654 .setMergeIdenticalEdges() 4655 .setDontDeleteUselessPHIs()); 4656 } else { 4657 SmallVector<BasicBlock*, 2> NewBBs; 4658 SplitLandingPadPredecessors(Parent, BB, "", "", NewBBs, &DT, &LI); 4659 NewBB = NewBBs[0]; 4660 } 4661 // If NewBB==NULL, then SplitCriticalEdge refused to split because all 4662 // phi predecessors are identical. The simple thing to do is skip 4663 // splitting in this case rather than complicate the API. 4664 if (NewBB) { 4665 // If PN is outside of the loop and BB is in the loop, we want to 4666 // move the block to be immediately before the PHI block, not 4667 // immediately after BB. 4668 if (L->contains(BB) && !L->contains(PN)) 4669 NewBB->moveBefore(PN->getParent()); 4670 4671 // Splitting the edge can reduce the number of PHI entries we have. 4672 e = PN->getNumIncomingValues(); 4673 BB = NewBB; 4674 i = PN->getBasicBlockIndex(BB); 4675 } 4676 } 4677 } 4678 4679 std::pair<DenseMap<BasicBlock *, Value *>::iterator, bool> Pair = 4680 Inserted.insert(std::make_pair(BB, static_cast<Value *>(nullptr))); 4681 if (!Pair.second) 4682 PN->setIncomingValue(i, Pair.first->second); 4683 else { 4684 Value *FullV = Expand(LF, F, BB->getTerminator()->getIterator(), 4685 Rewriter, DeadInsts); 4686 4687 // If this is reuse-by-noop-cast, insert the noop cast. 4688 Type *OpTy = LF.OperandValToReplace->getType(); 4689 if (FullV->getType() != OpTy) 4690 FullV = 4691 CastInst::Create(CastInst::getCastOpcode(FullV, false, 4692 OpTy, false), 4693 FullV, LF.OperandValToReplace->getType(), 4694 "tmp", BB->getTerminator()); 4695 4696 PN->setIncomingValue(i, FullV); 4697 Pair.first->second = FullV; 4698 } 4699 } 4700 } 4701 4702 /// Emit instructions for the leading candidate expression for this LSRUse (this 4703 /// is called "expanding"), and update the UserInst to reference the newly 4704 /// expanded value. 4705 void LSRInstance::Rewrite(const LSRFixup &LF, 4706 const Formula &F, 4707 SCEVExpander &Rewriter, 4708 SmallVectorImpl<WeakVH> &DeadInsts) const { 4709 // First, find an insertion point that dominates UserInst. For PHI nodes, 4710 // find the nearest block which dominates all the relevant uses. 4711 if (PHINode *PN = dyn_cast<PHINode>(LF.UserInst)) { 4712 RewriteForPHI(PN, LF, F, Rewriter, DeadInsts); 4713 } else { 4714 Value *FullV = 4715 Expand(LF, F, LF.UserInst->getIterator(), Rewriter, DeadInsts); 4716 4717 // If this is reuse-by-noop-cast, insert the noop cast. 4718 Type *OpTy = LF.OperandValToReplace->getType(); 4719 if (FullV->getType() != OpTy) { 4720 Instruction *Cast = 4721 CastInst::Create(CastInst::getCastOpcode(FullV, false, OpTy, false), 4722 FullV, OpTy, "tmp", LF.UserInst); 4723 FullV = Cast; 4724 } 4725 4726 // Update the user. ICmpZero is handled specially here (for now) because 4727 // Expand may have updated one of the operands of the icmp already, and 4728 // its new value may happen to be equal to LF.OperandValToReplace, in 4729 // which case doing replaceUsesOfWith leads to replacing both operands 4730 // with the same value. TODO: Reorganize this. 4731 if (Uses[LF.LUIdx].Kind == LSRUse::ICmpZero) 4732 LF.UserInst->setOperand(0, FullV); 4733 else 4734 LF.UserInst->replaceUsesOfWith(LF.OperandValToReplace, FullV); 4735 } 4736 4737 DeadInsts.emplace_back(LF.OperandValToReplace); 4738 } 4739 4740 /// Rewrite all the fixup locations with new values, following the chosen 4741 /// solution. 4742 void LSRInstance::ImplementSolution( 4743 const SmallVectorImpl<const Formula *> &Solution) { 4744 // Keep track of instructions we may have made dead, so that 4745 // we can remove them after we are done working. 4746 SmallVector<WeakVH, 16> DeadInsts; 4747 4748 SCEVExpander Rewriter(SE, L->getHeader()->getModule()->getDataLayout(), 4749 "lsr"); 4750 #ifndef NDEBUG 4751 Rewriter.setDebugType(DEBUG_TYPE); 4752 #endif 4753 Rewriter.disableCanonicalMode(); 4754 Rewriter.enableLSRMode(); 4755 Rewriter.setIVIncInsertPos(L, IVIncInsertPos); 4756 4757 // Mark phi nodes that terminate chains so the expander tries to reuse them. 4758 for (const IVChain &Chain : IVChainVec) { 4759 if (PHINode *PN = dyn_cast<PHINode>(Chain.tailUserInst())) 4760 Rewriter.setChainedPhi(PN); 4761 } 4762 4763 // Expand the new value definitions and update the users. 4764 for (const LSRFixup &Fixup : Fixups) { 4765 Rewrite(Fixup, *Solution[Fixup.LUIdx], Rewriter, DeadInsts); 4766 4767 Changed = true; 4768 } 4769 4770 for (const IVChain &Chain : IVChainVec) { 4771 GenerateIVChain(Chain, Rewriter, DeadInsts); 4772 Changed = true; 4773 } 4774 // Clean up after ourselves. This must be done before deleting any 4775 // instructions. 4776 Rewriter.clear(); 4777 4778 Changed |= DeleteTriviallyDeadInstructions(DeadInsts); 4779 } 4780 4781 LSRInstance::LSRInstance(Loop *L, IVUsers &IU, ScalarEvolution &SE, 4782 DominatorTree &DT, LoopInfo &LI, 4783 const TargetTransformInfo &TTI) 4784 : IU(IU), SE(SE), DT(DT), LI(LI), TTI(TTI), L(L), Changed(false), 4785 IVIncInsertPos(nullptr) { 4786 // If LoopSimplify form is not available, stay out of trouble. 4787 if (!L->isLoopSimplifyForm()) 4788 return; 4789 4790 // If there's no interesting work to be done, bail early. 4791 if (IU.empty()) return; 4792 4793 // If there's too much analysis to be done, bail early. We won't be able to 4794 // model the problem anyway. 4795 unsigned NumUsers = 0; 4796 for (const IVStrideUse &U : IU) { 4797 if (++NumUsers > MaxIVUsers) { 4798 (void)U; 4799 DEBUG(dbgs() << "LSR skipping loop, too many IV Users in " << U << "\n"); 4800 return; 4801 } 4802 } 4803 4804 #ifndef NDEBUG 4805 // All dominating loops must have preheaders, or SCEVExpander may not be able 4806 // to materialize an AddRecExpr whose Start is an outer AddRecExpr. 4807 // 4808 // IVUsers analysis should only create users that are dominated by simple loop 4809 // headers. Since this loop should dominate all of its users, its user list 4810 // should be empty if this loop itself is not within a simple loop nest. 4811 for (DomTreeNode *Rung = DT.getNode(L->getLoopPreheader()); 4812 Rung; Rung = Rung->getIDom()) { 4813 BasicBlock *BB = Rung->getBlock(); 4814 const Loop *DomLoop = LI.getLoopFor(BB); 4815 if (DomLoop && DomLoop->getHeader() == BB) { 4816 assert(DomLoop->getLoopPreheader() && "LSR needs a simplified loop nest"); 4817 } 4818 } 4819 #endif // DEBUG 4820 4821 DEBUG(dbgs() << "\nLSR on loop "; 4822 L->getHeader()->printAsOperand(dbgs(), /*PrintType=*/false); 4823 dbgs() << ":\n"); 4824 4825 // First, perform some low-level loop optimizations. 4826 OptimizeShadowIV(); 4827 OptimizeLoopTermCond(); 4828 4829 // If loop preparation eliminates all interesting IV users, bail. 4830 if (IU.empty()) return; 4831 4832 // Skip nested loops until we can model them better with formulae. 4833 if (!L->empty()) { 4834 DEBUG(dbgs() << "LSR skipping outer loop " << *L << "\n"); 4835 return; 4836 } 4837 4838 // Start collecting data and preparing for the solver. 4839 CollectChains(); 4840 CollectInterestingTypesAndFactors(); 4841 CollectFixupsAndInitialFormulae(); 4842 CollectLoopInvariantFixupsAndFormulae(); 4843 4844 assert(!Uses.empty() && "IVUsers reported at least one use"); 4845 DEBUG(dbgs() << "LSR found " << Uses.size() << " uses:\n"; 4846 print_uses(dbgs())); 4847 4848 // Now use the reuse data to generate a bunch of interesting ways 4849 // to formulate the values needed for the uses. 4850 GenerateAllReuseFormulae(); 4851 4852 FilterOutUndesirableDedicatedRegisters(); 4853 NarrowSearchSpaceUsingHeuristics(); 4854 4855 SmallVector<const Formula *, 8> Solution; 4856 Solve(Solution); 4857 4858 // Release memory that is no longer needed. 4859 Factors.clear(); 4860 Types.clear(); 4861 RegUses.clear(); 4862 4863 if (Solution.empty()) 4864 return; 4865 4866 #ifndef NDEBUG 4867 // Formulae should be legal. 4868 for (const LSRUse &LU : Uses) { 4869 for (const Formula &F : LU.Formulae) 4870 assert(isLegalUse(TTI, LU.MinOffset, LU.MaxOffset, LU.Kind, LU.AccessTy, 4871 F) && "Illegal formula generated!"); 4872 }; 4873 #endif 4874 4875 // Now that we've decided what we want, make it so. 4876 ImplementSolution(Solution); 4877 } 4878 4879 void LSRInstance::print_factors_and_types(raw_ostream &OS) const { 4880 if (Factors.empty() && Types.empty()) return; 4881 4882 OS << "LSR has identified the following interesting factors and types: "; 4883 bool First = true; 4884 4885 for (int64_t Factor : Factors) { 4886 if (!First) OS << ", "; 4887 First = false; 4888 OS << '*' << Factor; 4889 } 4890 4891 for (Type *Ty : Types) { 4892 if (!First) OS << ", "; 4893 First = false; 4894 OS << '(' << *Ty << ')'; 4895 } 4896 OS << '\n'; 4897 } 4898 4899 void LSRInstance::print_fixups(raw_ostream &OS) const { 4900 OS << "LSR is examining the following fixup sites:\n"; 4901 for (const LSRFixup &LF : Fixups) { 4902 dbgs() << " "; 4903 LF.print(OS); 4904 OS << '\n'; 4905 } 4906 } 4907 4908 void LSRInstance::print_uses(raw_ostream &OS) const { 4909 OS << "LSR is examining the following uses:\n"; 4910 for (const LSRUse &LU : Uses) { 4911 dbgs() << " "; 4912 LU.print(OS); 4913 OS << '\n'; 4914 for (const Formula &F : LU.Formulae) { 4915 OS << " "; 4916 F.print(OS); 4917 OS << '\n'; 4918 } 4919 } 4920 } 4921 4922 void LSRInstance::print(raw_ostream &OS) const { 4923 print_factors_and_types(OS); 4924 print_fixups(OS); 4925 print_uses(OS); 4926 } 4927 4928 LLVM_DUMP_METHOD 4929 void LSRInstance::dump() const { 4930 print(errs()); errs() << '\n'; 4931 } 4932 4933 namespace { 4934 4935 class LoopStrengthReduce : public LoopPass { 4936 public: 4937 static char ID; // Pass ID, replacement for typeid 4938 LoopStrengthReduce(); 4939 4940 private: 4941 bool runOnLoop(Loop *L, LPPassManager &LPM) override; 4942 void getAnalysisUsage(AnalysisUsage &AU) const override; 4943 }; 4944 4945 } 4946 4947 char LoopStrengthReduce::ID = 0; 4948 INITIALIZE_PASS_BEGIN(LoopStrengthReduce, "loop-reduce", 4949 "Loop Strength Reduction", false, false) 4950 INITIALIZE_PASS_DEPENDENCY(TargetTransformInfoWrapperPass) 4951 INITIALIZE_PASS_DEPENDENCY(DominatorTreeWrapperPass) 4952 INITIALIZE_PASS_DEPENDENCY(ScalarEvolutionWrapperPass) 4953 INITIALIZE_PASS_DEPENDENCY(IVUsers) 4954 INITIALIZE_PASS_DEPENDENCY(LoopInfoWrapperPass) 4955 INITIALIZE_PASS_DEPENDENCY(LoopSimplify) 4956 INITIALIZE_PASS_END(LoopStrengthReduce, "loop-reduce", 4957 "Loop Strength Reduction", false, false) 4958 4959 4960 Pass *llvm::createLoopStrengthReducePass() { 4961 return new LoopStrengthReduce(); 4962 } 4963 4964 LoopStrengthReduce::LoopStrengthReduce() : LoopPass(ID) { 4965 initializeLoopStrengthReducePass(*PassRegistry::getPassRegistry()); 4966 } 4967 4968 void LoopStrengthReduce::getAnalysisUsage(AnalysisUsage &AU) const { 4969 // We split critical edges, so we change the CFG. However, we do update 4970 // many analyses if they are around. 4971 AU.addPreservedID(LoopSimplifyID); 4972 4973 AU.addRequired<LoopInfoWrapperPass>(); 4974 AU.addPreserved<LoopInfoWrapperPass>(); 4975 AU.addRequiredID(LoopSimplifyID); 4976 AU.addRequired<DominatorTreeWrapperPass>(); 4977 AU.addPreserved<DominatorTreeWrapperPass>(); 4978 AU.addRequired<ScalarEvolutionWrapperPass>(); 4979 AU.addPreserved<ScalarEvolutionWrapperPass>(); 4980 // Requiring LoopSimplify a second time here prevents IVUsers from running 4981 // twice, since LoopSimplify was invalidated by running ScalarEvolution. 4982 AU.addRequiredID(LoopSimplifyID); 4983 AU.addRequired<IVUsers>(); 4984 AU.addPreserved<IVUsers>(); 4985 AU.addRequired<TargetTransformInfoWrapperPass>(); 4986 } 4987 4988 bool LoopStrengthReduce::runOnLoop(Loop *L, LPPassManager & /*LPM*/) { 4989 if (skipOptnoneFunction(L)) 4990 return false; 4991 4992 auto &IU = getAnalysis<IVUsers>(); 4993 auto &SE = getAnalysis<ScalarEvolutionWrapperPass>().getSE(); 4994 auto &DT = getAnalysis<DominatorTreeWrapperPass>().getDomTree(); 4995 auto &LI = getAnalysis<LoopInfoWrapperPass>().getLoopInfo(); 4996 const auto &TTI = getAnalysis<TargetTransformInfoWrapperPass>().getTTI( 4997 *L->getHeader()->getParent()); 4998 bool Changed = false; 4999 5000 // Run the main LSR transformation. 5001 Changed |= LSRInstance(L, IU, SE, DT, LI, TTI).getChanged(); 5002 5003 // Remove any extra phis created by processing inner loops. 5004 Changed |= DeleteDeadPHIs(L->getHeader()); 5005 if (EnablePhiElim && L->isLoopSimplifyForm()) { 5006 SmallVector<WeakVH, 16> DeadInsts; 5007 const DataLayout &DL = L->getHeader()->getModule()->getDataLayout(); 5008 SCEVExpander Rewriter(getAnalysis<ScalarEvolutionWrapperPass>().getSE(), DL, 5009 "lsr"); 5010 #ifndef NDEBUG 5011 Rewriter.setDebugType(DEBUG_TYPE); 5012 #endif 5013 unsigned numFolded = Rewriter.replaceCongruentIVs( 5014 L, &getAnalysis<DominatorTreeWrapperPass>().getDomTree(), DeadInsts, 5015 &getAnalysis<TargetTransformInfoWrapperPass>().getTTI( 5016 *L->getHeader()->getParent())); 5017 if (numFolded) { 5018 Changed = true; 5019 DeleteTriviallyDeadInstructions(DeadInsts); 5020 DeleteDeadPHIs(L->getHeader()); 5021 } 5022 } 5023 return Changed; 5024 } 5025