brintos

brintos / llvm-project-archived public Read only

0
0
Text · 25.6 KiB · b816f11 Raw
722 lines · cpp
1//===-- AArch64A57FPLoadBalancing.cpp - Balance FP ops statically on A57---===//2//3// Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.4// See https://llvm.org/LICENSE.txt for license information.5// SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception6//7//===----------------------------------------------------------------------===//8// For best-case performance on Cortex-A57, we should try to use a balanced9// mix of odd and even D-registers when performing a critical sequence of10// independent, non-quadword FP/ASIMD floating-point multiply or11// multiply-accumulate operations.12//13// This pass attempts to detect situations where the register allocation may14// adversely affect this load balancing and to change the registers used so as15// to better utilize the CPU.16//17// Ideally we'd just take each multiply or multiply-accumulate in turn and18// allocate it alternating even or odd registers. However, multiply-accumulates19// are most efficiently performed in the same functional unit as their20// accumulation operand. Therefore this pass tries to find maximal sequences21// ("Chains") of multiply-accumulates linked via their accumulation operand,22// and assign them all the same "color" (oddness/evenness).23//24// This optimization affects S-register and D-register floating point25// multiplies and FMADD/FMAs, as well as vector (floating point only) muls and26// FMADD/FMA. Q register instructions (and 128-bit vector instructions) are27// not affected.28//===----------------------------------------------------------------------===//29 30#include "AArch64.h"31#include "AArch64InstrInfo.h"32#include "AArch64Subtarget.h"33#include "llvm/ADT/EquivalenceClasses.h"34#include "llvm/CodeGen/MachineFunction.h"35#include "llvm/CodeGen/MachineFunctionPass.h"36#include "llvm/CodeGen/MachineInstr.h"37#include "llvm/CodeGen/MachineInstrBuilder.h"38#include "llvm/CodeGen/MachineRegisterInfo.h"39#include "llvm/CodeGen/RegisterClassInfo.h"40#include "llvm/CodeGen/RegisterScavenging.h"41#include "llvm/Support/CommandLine.h"42#include "llvm/Support/Debug.h"43#include "llvm/Support/raw_ostream.h"44using namespace llvm;45 46#define DEBUG_TYPE "aarch64-a57-fp-load-balancing"47 48// Enforce the algorithm to use the scavenged register even when the original49// destination register is the correct color. Used for testing.50static cl::opt<bool>51TransformAll("aarch64-a57-fp-load-balancing-force-all",52             cl::desc("Always modify dest registers regardless of color"),53             cl::init(false), cl::Hidden);54 55// Never use the balance information obtained from chains - return a specific56// color always. Used for testing.57static cl::opt<unsigned>58OverrideBalance("aarch64-a57-fp-load-balancing-override",59              cl::desc("Ignore balance information, always return "60                       "(1: Even, 2: Odd)."),61              cl::init(0), cl::Hidden);62 63//===----------------------------------------------------------------------===//64// Helper functions65 66// Is the instruction a type of multiply on 64-bit (or 32-bit) FPRs?67static bool isMul(MachineInstr *MI) {68  switch (MI->getOpcode()) {69  case AArch64::FMULSrr:70  case AArch64::FNMULSrr:71  case AArch64::FMULDrr:72  case AArch64::FNMULDrr:73    return true;74  default:75    return false;76  }77}78 79// Is the instruction a type of FP multiply-accumulate on 64-bit (or 32-bit) FPRs?80static bool isMla(MachineInstr *MI) {81  switch (MI->getOpcode()) {82  case AArch64::FMSUBSrrr:83  case AArch64::FMADDSrrr:84  case AArch64::FNMSUBSrrr:85  case AArch64::FNMADDSrrr:86  case AArch64::FMSUBDrrr:87  case AArch64::FMADDDrrr:88  case AArch64::FNMSUBDrrr:89  case AArch64::FNMADDDrrr:90    return true;91  default:92    return false;93  }94}95 96//===----------------------------------------------------------------------===//97 98namespace {99/// A "color", which is either even or odd. Yes, these aren't really colors100/// but the algorithm is conceptually doing two-color graph coloring.101enum class Color { Even, Odd };102#ifndef NDEBUG103static const char *ColorNames[2] = { "Even", "Odd" };104#endif105 106class Chain;107 108class AArch64A57FPLoadBalancing : public MachineFunctionPass {109  MachineRegisterInfo *MRI;110  const TargetRegisterInfo *TRI;111  RegisterClassInfo RCI;112 113public:114  static char ID;115  explicit AArch64A57FPLoadBalancing() : MachineFunctionPass(ID) {}116 117  bool runOnMachineFunction(MachineFunction &F) override;118 119  MachineFunctionProperties getRequiredProperties() const override {120    return MachineFunctionProperties().setNoVRegs();121  }122 123  StringRef getPassName() const override {124    return "A57 FP Anti-dependency breaker";125  }126 127  void getAnalysisUsage(AnalysisUsage &AU) const override {128    AU.setPreservesCFG();129    MachineFunctionPass::getAnalysisUsage(AU);130  }131 132private:133  bool runOnBasicBlock(MachineBasicBlock &MBB);134  bool colorChainSet(std::vector<Chain*> GV, MachineBasicBlock &MBB,135                     int &Balance);136  bool colorChain(Chain *G, Color C, MachineBasicBlock &MBB);137  int scavengeRegister(Chain *G, Color C, MachineBasicBlock &MBB);138  void scanInstruction(MachineInstr *MI, unsigned Idx,139                       std::map<unsigned, Chain*> &Active,140                       std::vector<std::unique_ptr<Chain>> &AllChains);141  void maybeKillChain(MachineOperand &MO, unsigned Idx,142                      std::map<unsigned, Chain*> &RegChains);143  Color getColor(unsigned Register);144  Chain *getAndEraseNext(Color PreferredColor, std::vector<Chain*> &L);145};146}147 148char AArch64A57FPLoadBalancing::ID = 0;149 150INITIALIZE_PASS_BEGIN(AArch64A57FPLoadBalancing, DEBUG_TYPE,151                      "AArch64 A57 FP Load-Balancing", false, false)152INITIALIZE_PASS_END(AArch64A57FPLoadBalancing, DEBUG_TYPE,153                    "AArch64 A57 FP Load-Balancing", false, false)154 155namespace {156/// A Chain is a sequence of instructions that are linked together by157/// an accumulation operand. For example:158///159///   fmul def d0, ?160///   fmla def d1, ?, ?, killed d0161///   fmla def d2, ?, ?, killed d1162///163/// There may be other instructions interleaved in the sequence that164/// do not belong to the chain. These other instructions must not use165/// the "chain" register at any point.166///167/// We currently only support chains where the "chain" operand is killed168/// at each link in the chain for simplicity.169/// A chain has three important instructions - Start, Last and Kill.170///   * The start instruction is the first instruction in the chain.171///   * Last is the final instruction in the chain.172///   * Kill may or may not be defined. If defined, Kill is the instruction173///     where the outgoing value of the Last instruction is killed.174///     This information is important as if we know the outgoing value is175///     killed with no intervening uses, we can safely change its register.176///177/// Without a kill instruction, we must assume the outgoing value escapes178/// beyond our model and either must not change its register or must179/// create a fixup FMOV to keep the old register value consistent.180///181class Chain {182public:183  /// The important (marker) instructions.184  MachineInstr *StartInst, *LastInst, *KillInst;185  /// The index, from the start of the basic block, that each marker186  /// appears. These are stored so we can do quick interval tests.187  unsigned StartInstIdx, LastInstIdx, KillInstIdx;188  /// All instructions in the chain.189  std::set<MachineInstr*> Insts;190  /// True if KillInst cannot be modified. If this is true,191  /// we cannot change LastInst's outgoing register.192  /// This will be true for tied values and regmasks.193  bool KillIsImmutable;194  /// The "color" of LastInst. This will be the preferred chain color,195  /// as changing intermediate nodes is easy but changing the last196  /// instruction can be more tricky.197  Color LastColor;198 199  Chain(MachineInstr *MI, unsigned Idx, Color C)200      : StartInst(MI), LastInst(MI), KillInst(nullptr),201        StartInstIdx(Idx), LastInstIdx(Idx), KillInstIdx(0),202        LastColor(C) {203    Insts.insert(MI);204  }205 206  /// Add a new instruction into the chain. The instruction's dest operand207  /// has the given color.208  void add(MachineInstr *MI, unsigned Idx, Color C) {209    LastInst = MI;210    LastInstIdx = Idx;211    LastColor = C;212    assert((KillInstIdx == 0 || LastInstIdx < KillInstIdx) &&213           "Chain: broken invariant. A Chain can only be killed after its last "214           "def");215 216    Insts.insert(MI);217  }218 219  /// Return true if MI is a member of the chain.220  bool contains(MachineInstr &MI) { return Insts.count(&MI) > 0; }221 222  /// Return the number of instructions in the chain.223  unsigned size() const {224    return Insts.size();225  }226 227  /// Inform the chain that its last active register (the dest register of228  /// LastInst) is killed by MI with no intervening uses or defs.229  void setKill(MachineInstr *MI, unsigned Idx, bool Immutable) {230    KillInst = MI;231    KillInstIdx = Idx;232    KillIsImmutable = Immutable;233    assert((KillInstIdx == 0 || LastInstIdx < KillInstIdx) &&234           "Chain: broken invariant. A Chain can only be killed after its last "235           "def");236  }237 238  /// Return the first instruction in the chain.239  MachineInstr *getStart() const { return StartInst; }240  /// Return the last instruction in the chain.241  MachineInstr *getLast() const { return LastInst; }242  /// Return the "kill" instruction (as set with setKill()) or NULL.243  MachineInstr *getKill() const { return KillInst; }244  /// Return an instruction that can be used as an iterator for the end245  /// of the chain. This is the maximum of KillInst (if set) and LastInst.246  MachineBasicBlock::iterator end() const {247    return ++MachineBasicBlock::iterator(KillInst ? KillInst : LastInst);248  }249  MachineBasicBlock::iterator begin() const { return getStart(); }250 251  /// Can the Kill instruction (assuming one exists) be modified?252  bool isKillImmutable() const { return KillIsImmutable; }253 254  /// Return the preferred color of this chain.255  Color getPreferredColor() {256    if (OverrideBalance != 0)257      return OverrideBalance == 1 ? Color::Even : Color::Odd;258    return LastColor;259  }260 261  /// Return true if this chain (StartInst..KillInst) overlaps with Other.262  bool rangeOverlapsWith(const Chain &Other) const {263    unsigned End = KillInst ? KillInstIdx : LastInstIdx;264    unsigned OtherEnd = Other.KillInst ?265      Other.KillInstIdx : Other.LastInstIdx;266 267    return StartInstIdx <= OtherEnd && Other.StartInstIdx <= End;268  }269 270  /// Return true if this chain starts before Other.271  bool startsBefore(const Chain *Other) const {272    return StartInstIdx < Other->StartInstIdx;273  }274 275  /// Return true if the group will require a fixup MOV at the end.276  bool requiresFixup() const {277    return (getKill() && isKillImmutable()) || !getKill();278  }279 280  /// Return a simple string representation of the chain.281  std::string str() const {282    std::string S;283    raw_string_ostream OS(S);284 285    OS << "{";286    StartInst->print(OS, /* SkipOpers= */true);287    OS << " -> ";288    LastInst->print(OS, /* SkipOpers= */true);289    if (KillInst) {290      OS << " (kill @ ";291      KillInst->print(OS, /* SkipOpers= */true);292      OS << ")";293    }294    OS << "}";295 296    return OS.str();297  }298 299};300 301} // end anonymous namespace302 303//===----------------------------------------------------------------------===//304 305bool AArch64A57FPLoadBalancing::runOnMachineFunction(MachineFunction &F) {306  if (skipFunction(F.getFunction()))307    return false;308 309  if (!F.getSubtarget<AArch64Subtarget>().balanceFPOps())310    return false;311 312  bool Changed = false;313  LLVM_DEBUG(dbgs() << "***** AArch64A57FPLoadBalancing *****\n");314 315  MRI = &F.getRegInfo();316  TRI = F.getRegInfo().getTargetRegisterInfo();317  RCI.runOnMachineFunction(F);318 319  for (auto &MBB : F) {320    Changed |= runOnBasicBlock(MBB);321  }322 323  return Changed;324}325 326bool AArch64A57FPLoadBalancing::runOnBasicBlock(MachineBasicBlock &MBB) {327  bool Changed = false;328  LLVM_DEBUG(dbgs() << "Running on MBB: " << MBB329                    << " - scanning instructions...\n");330 331  // First, scan the basic block producing a set of chains.332 333  // The currently "active" chains - chains that can be added to and haven't334  // been killed yet. This is keyed by register - all chains can only have one335  // "link" register between each inst in the chain.336  std::map<unsigned, Chain*> ActiveChains;337  std::vector<std::unique_ptr<Chain>> AllChains;338  unsigned Idx = 0;339  for (auto &MI : MBB)340    scanInstruction(&MI, Idx++, ActiveChains, AllChains);341 342  LLVM_DEBUG(dbgs() << "Scan complete, " << AllChains.size()343                    << " chains created.\n");344 345  // Group the chains into disjoint sets based on their liveness range. This is346  // a poor-man's version of graph coloring. Ideally we'd create an interference347  // graph and perform full-on graph coloring on that, but;348  //   (a) That's rather heavyweight for only two colors.349  //   (b) We expect multiple disjoint interference regions - in practice the live350  //       range of chains is quite small and they are clustered between loads351  //       and stores.352  EquivalenceClasses<Chain*> EC;353  for (auto &I : AllChains)354    EC.insert(I.get());355 356  for (auto &I : AllChains)357    for (auto &J : AllChains)358      if (I != J && I->rangeOverlapsWith(*J))359        EC.unionSets(I.get(), J.get());360  LLVM_DEBUG(dbgs() << "Created " << EC.getNumClasses() << " disjoint sets.\n");361 362  // Now we assume that every member of an equivalence class interferes363  // with every other member of that class, and with no members of other classes.364 365  // Convert the EquivalenceClasses to a simpler set of sets.366  std::vector<std::vector<Chain*> > V;367  for (const auto &E : EC) {368    if (!E->isLeader())369      continue;370    std::vector<Chain *> Cs(EC.member_begin(*E), EC.member_end());371    if (Cs.empty()) continue;372    V.push_back(std::move(Cs));373  }374 375  // Now we have a set of sets, order them by start address so376  // we can iterate over them sequentially.377  llvm::sort(V,378             [](const std::vector<Chain *> &A, const std::vector<Chain *> &B) {379               return A.front()->startsBefore(B.front());380             });381 382  // As we only have two colors, we can track the global (BB-level) balance of383  // odds versus evens. We aim to keep this near zero to keep both execution384  // units fed.385  // Positive means we're even-heavy, negative we're odd-heavy.386  //387  // FIXME: If chains have interdependencies, for example:388  //   mul r0, r1, r2389  //   mul r3, r0, r1390  // We do not model this and may color each one differently, assuming we'll391  // get ILP when we obviously can't. This hasn't been seen to be a problem392  // in practice so far, so we simplify the algorithm by ignoring it.393  int Parity = 0;394 395  for (auto &I : V)396    Changed |= colorChainSet(std::move(I), MBB, Parity);397 398  return Changed;399}400 401Chain *AArch64A57FPLoadBalancing::getAndEraseNext(Color PreferredColor,402                                                  std::vector<Chain*> &L) {403  if (L.empty())404    return nullptr;405 406  // We try and get the best candidate from L to color next, given that our407  // preferred color is "PreferredColor". L is ordered from larger to smaller408  // chains. It is beneficial to color the large chains before the small chains,409  // but if we can't find a chain of the maximum length with the preferred color,410  // we fuzz the size and look for slightly smaller chains before giving up and411  // returning a chain that must be recolored.412 413  // FIXME: Does this need to be configurable?414  const unsigned SizeFuzz = 1;415  unsigned MinSize = L.front()->size() - SizeFuzz;416  for (auto I = L.begin(), E = L.end(); I != E; ++I) {417    if ((*I)->size() <= MinSize) {418      // We've gone past the size limit. Return the previous item.419      Chain *Ch = *--I;420      L.erase(I);421      return Ch;422    }423 424    if ((*I)->getPreferredColor() == PreferredColor) {425      Chain *Ch = *I;426      L.erase(I);427      return Ch;428    }429  }430 431  // Bailout case - just return the first item.432  Chain *Ch = L.front();433  L.erase(L.begin());434  return Ch;435}436 437bool AArch64A57FPLoadBalancing::colorChainSet(std::vector<Chain*> GV,438                                              MachineBasicBlock &MBB,439                                              int &Parity) {440  bool Changed = false;441  LLVM_DEBUG(dbgs() << "colorChainSet(): #sets=" << GV.size() << "\n");442 443  // Sort by descending size order so that we allocate the most important444  // sets first.445  // Tie-break equivalent sizes by sorting chains requiring fixups before446  // those without fixups. The logic here is that we should look at the447  // chains that we cannot change before we look at those we can,448  // so the parity counter is updated and we know what color we should449  // change them to!450  // Final tie-break with instruction order so pass output is stable (i.e. not451  // dependent on malloc'd pointer values).452  llvm::sort(GV, [](const Chain *G1, const Chain *G2) {453    if (G1->size() != G2->size())454      return G1->size() > G2->size();455    if (G1->requiresFixup() != G2->requiresFixup())456      return G1->requiresFixup() > G2->requiresFixup();457    // Make sure startsBefore() produces a stable final order.458    assert((G1 == G2 || (G1->startsBefore(G2) ^ G2->startsBefore(G1))) &&459           "Starts before not total order!");460    return G1->startsBefore(G2);461  });462 463  Color PreferredColor = Parity < 0 ? Color::Even : Color::Odd;464  while (Chain *G = getAndEraseNext(PreferredColor, GV)) {465    // Start off by assuming we'll color to our own preferred color.466    Color C = PreferredColor;467    if (Parity == 0)468      // But if we really don't care, use the chain's preferred color.469      C = G->getPreferredColor();470 471    LLVM_DEBUG(dbgs() << " - Parity=" << Parity472                      << ", Color=" << ColorNames[(int)C] << "\n");473 474    // If we'll need a fixup FMOV, don't bother. Testing has shown that this475    // happens infrequently and when it does it has at least a 50% chance of476    // slowing code down instead of speeding it up.477    if (G->requiresFixup() && C != G->getPreferredColor()) {478      C = G->getPreferredColor();479      LLVM_DEBUG(dbgs() << " - " << G->str()480                        << " - not worthwhile changing; "481                           "color remains "482                        << ColorNames[(int)C] << "\n");483    }484 485    Changed |= colorChain(G, C, MBB);486 487    Parity += (C == Color::Even) ? G->size() : -G->size();488    PreferredColor = Parity < 0 ? Color::Even : Color::Odd;489  }490 491  return Changed;492}493 494int AArch64A57FPLoadBalancing::scavengeRegister(Chain *G, Color C,495                                                MachineBasicBlock &MBB) {496  // Can we find an appropriate register that is available throughout the life497  // of the chain? Simulate liveness backwards until the end of the chain.498  LiveRegUnits Units(*TRI);499  Units.addLiveOuts(MBB);500  MachineBasicBlock::iterator I = MBB.end();501  MachineBasicBlock::iterator ChainEnd = G->end();502  while (I != ChainEnd) {503    --I;504    Units.stepBackward(*I);505  }506 507  // Check which register units are alive throughout the chain.508  MachineBasicBlock::iterator ChainBegin = G->begin();509  assert(ChainBegin != ChainEnd && "Chain should contain instructions");510  do {511    --I;512    Units.accumulate(*I);513  } while (I != ChainBegin);514 515  // Make sure we allocate in-order, to get the cheapest registers first.516  unsigned RegClassID = ChainBegin->getDesc().operands()[0].RegClass;517  auto Ord = RCI.getOrder(TRI->getRegClass(RegClassID));518  for (auto Reg : Ord) {519    if (!Units.available(Reg))520      continue;521    if (C == getColor(Reg))522      return Reg;523  }524 525  return -1;526}527 528bool AArch64A57FPLoadBalancing::colorChain(Chain *G, Color C,529                                           MachineBasicBlock &MBB) {530  bool Changed = false;531  LLVM_DEBUG(dbgs() << " - colorChain(" << G->str() << ", "532                    << ColorNames[(int)C] << ")\n");533 534  // Try and obtain a free register of the right class. Without a register535  // to play with we cannot continue.536  int Reg = scavengeRegister(G, C, MBB);537  if (Reg == -1) {538    LLVM_DEBUG(dbgs() << "Scavenging (thus coloring) failed!\n");539    return false;540  }541  LLVM_DEBUG(dbgs() << " - Scavenged register: " << printReg(Reg, TRI) << "\n");542 543  std::map<unsigned, unsigned> Substs;544  for (MachineInstr &I : *G) {545    if (!G->contains(I) && (&I != G->getKill() || G->isKillImmutable()))546      continue;547 548    // I is a member of G, or I is a mutable instruction that kills G.549 550    std::vector<unsigned> ToErase;551    for (auto &U : I.operands()) {552      if (U.isReg() && U.isUse() && Substs.find(U.getReg()) != Substs.end()) {553        Register OrigReg = U.getReg();554        U.setReg(Substs[OrigReg]);555        if (U.isKill())556          // Don't erase straight away, because there may be other operands557          // that also reference this substitution!558          ToErase.push_back(OrigReg);559      } else if (U.isRegMask()) {560        for (auto J : Substs) {561          if (U.clobbersPhysReg(J.first))562            ToErase.push_back(J.first);563        }564      }565    }566    // Now it's safe to remove the substs identified earlier.567    for (auto J : ToErase)568      Substs.erase(J);569 570    // Only change the def if this isn't the last instruction.571    if (&I != G->getKill()) {572      MachineOperand &MO = I.getOperand(0);573 574      bool Change = TransformAll || getColor(MO.getReg()) != C;575      if (G->requiresFixup() && &I == G->getLast())576        Change = false;577 578      if (Change) {579        Substs[MO.getReg()] = Reg;580        MO.setReg(Reg);581 582        Changed = true;583      }584    }585  }586  assert(Substs.size() == 0 && "No substitutions should be left active!");587 588  if (G->getKill()) {589    LLVM_DEBUG(dbgs() << " - Kill instruction seen.\n");590  } else {591    // We didn't have a kill instruction, but we didn't seem to need to change592    // the destination register anyway.593    LLVM_DEBUG(dbgs() << " - Destination register not changed.\n");594  }595  return Changed;596}597 598void AArch64A57FPLoadBalancing::scanInstruction(599    MachineInstr *MI, unsigned Idx, std::map<unsigned, Chain *> &ActiveChains,600    std::vector<std::unique_ptr<Chain>> &AllChains) {601  // Inspect "MI", updating ActiveChains and AllChains.602 603  if (isMul(MI)) {604 605    for (auto &I : MI->uses())606      maybeKillChain(I, Idx, ActiveChains);607    for (auto &I : MI->defs())608      maybeKillChain(I, Idx, ActiveChains);609 610    // Create a new chain. Multiplies don't require forwarding so can go on any611    // unit.612    Register DestReg = MI->getOperand(0).getReg();613 614    LLVM_DEBUG(dbgs() << "New chain started for register "615                      << printReg(DestReg, TRI) << " at " << *MI);616 617    auto G = std::make_unique<Chain>(MI, Idx, getColor(DestReg));618    ActiveChains[DestReg] = G.get();619    AllChains.push_back(std::move(G));620 621  } else if (isMla(MI)) {622 623    // It is beneficial to keep MLAs on the same functional unit as their624    // accumulator operand.625    Register DestReg = MI->getOperand(0).getReg();626    Register AccumReg = MI->getOperand(3).getReg();627 628    maybeKillChain(MI->getOperand(1), Idx, ActiveChains);629    maybeKillChain(MI->getOperand(2), Idx, ActiveChains);630    if (DestReg != AccumReg)631      maybeKillChain(MI->getOperand(0), Idx, ActiveChains);632 633    if (ActiveChains.find(AccumReg) != ActiveChains.end()) {634      LLVM_DEBUG(dbgs() << "Chain found for accumulator register "635                        << printReg(AccumReg, TRI) << " in MI " << *MI);636 637      // For simplicity we only chain together sequences of MULs/MLAs where the638      // accumulator register is killed on each instruction. This means we don't639      // need to track other uses of the registers we want to rewrite.640      //641      // FIXME: We could extend to handle the non-kill cases for more coverage.642      if (MI->getOperand(3).isKill()) {643        // Add to chain.644        LLVM_DEBUG(dbgs() << "Instruction was successfully added to chain.\n");645        ActiveChains[AccumReg]->add(MI, Idx, getColor(DestReg));646        // Handle cases where the destination is not the same as the accumulator.647        if (DestReg != AccumReg) {648          ActiveChains[DestReg] = ActiveChains[AccumReg];649          ActiveChains.erase(AccumReg);650        }651        return;652      }653 654      LLVM_DEBUG(655          dbgs() << "Cannot add to chain because accumulator operand wasn't "656                 << "marked <kill>!\n");657      maybeKillChain(MI->getOperand(3), Idx, ActiveChains);658    }659 660    LLVM_DEBUG(dbgs() << "Creating new chain for dest register "661                      << printReg(DestReg, TRI) << "\n");662    auto G = std::make_unique<Chain>(MI, Idx, getColor(DestReg));663    ActiveChains[DestReg] = G.get();664    AllChains.push_back(std::move(G));665 666  } else {667 668    // Non-MUL or MLA instruction. Invalidate any chain in the uses or defs669    // lists.670    for (auto &I : MI->uses())671      maybeKillChain(I, Idx, ActiveChains);672    for (auto &I : MI->defs())673      maybeKillChain(I, Idx, ActiveChains);674 675  }676}677 678void AArch64A57FPLoadBalancing::679maybeKillChain(MachineOperand &MO, unsigned Idx,680               std::map<unsigned, Chain*> &ActiveChains) {681  // Given an operand and the set of active chains (keyed by register),682  // determine if a chain should be ended and remove from ActiveChains.683  MachineInstr *MI = MO.getParent();684 685  if (MO.isReg()) {686 687    // If this is a KILL of a current chain, record it.688    if (MO.isKill() && ActiveChains.find(MO.getReg()) != ActiveChains.end()) {689      LLVM_DEBUG(dbgs() << "Kill seen for chain " << printReg(MO.getReg(), TRI)690                        << "\n");691      ActiveChains[MO.getReg()]->setKill(MI, Idx, /*Immutable=*/MO.isTied());692    }693    ActiveChains.erase(MO.getReg());694 695  } else if (MO.isRegMask()) {696 697    for (auto I = ActiveChains.begin(), E = ActiveChains.end();698         I != E;) {699      if (MO.clobbersPhysReg(I->first)) {700        LLVM_DEBUG(dbgs() << "Kill (regmask) seen for chain "701                          << printReg(I->first, TRI) << "\n");702        I->second->setKill(MI, Idx, /*Immutable=*/true);703        ActiveChains.erase(I++);704      } else705        ++I;706    }707 708  }709}710 711Color AArch64A57FPLoadBalancing::getColor(unsigned Reg) {712  if ((TRI->getEncodingValue(Reg) % 2) == 0)713    return Color::Even;714  else715    return Color::Odd;716}717 718// Factory function used by AArch64TargetMachine to add the pass to the passmanager.719FunctionPass *llvm::createAArch64A57FPLoadBalancing() {720  return new AArch64A57FPLoadBalancing();721}722