//===- AArch64InstructionSelector.cpp ----------------------------*- C++ -*-==//
//
// Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
// See https://llvm.org/LICENSE.txt for license information.
// SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
//
//===----------------------------------------------------------------------===//
/// \file
/// This file implements the targeting of the InstructionSelector class for
/// AArch64.
/// \todo This should be generated by TableGen.
//===----------------------------------------------------------------------===//

#include "AArch64GlobalISelUtils.h"
#include "AArch64InstrInfo.h"
#include "AArch64MachineFunctionInfo.h"
#include "AArch64RegisterBankInfo.h"
#include "AArch64RegisterInfo.h"
#include "AArch64Subtarget.h"
#include "AArch64TargetMachine.h"
#include "MCTargetDesc/AArch64AddressingModes.h"
#include "MCTargetDesc/AArch64MCTargetDesc.h"
#include "llvm/BinaryFormat/Dwarf.h"
#include "llvm/CodeGen/GlobalISel/GenericMachineInstrs.h"
#include "llvm/CodeGen/GlobalISel/InstructionSelector.h"
#include "llvm/CodeGen/GlobalISel/InstructionSelectorImpl.h"
#include "llvm/CodeGen/GlobalISel/MIPatternMatch.h"
#include "llvm/CodeGen/GlobalISel/MachineIRBuilder.h"
#include "llvm/CodeGen/GlobalISel/Utils.h"
#include "llvm/CodeGen/MachineBasicBlock.h"
#include "llvm/CodeGen/MachineConstantPool.h"
#include "llvm/CodeGen/MachineFrameInfo.h"
#include "llvm/CodeGen/MachineFunction.h"
#include "llvm/CodeGen/MachineInstr.h"
#include "llvm/CodeGen/MachineInstrBuilder.h"
#include "llvm/CodeGen/MachineMemOperand.h"
#include "llvm/CodeGen/MachineOperand.h"
#include "llvm/CodeGen/MachineRegisterInfo.h"
#include "llvm/CodeGen/TargetOpcodes.h"
#include "llvm/IR/Constants.h"
#include "llvm/IR/DerivedTypes.h"
#include "llvm/IR/Instructions.h"
#include "llvm/IR/IntrinsicsAArch64.h"
#include "llvm/IR/PatternMatch.h"
#include "llvm/IR/Type.h"
#include "llvm/Pass.h"
#include "llvm/Support/Debug.h"
#include "llvm/Support/raw_ostream.h"
#include <optional>

#define DEBUG_TYPE "aarch64-isel"

using namespace llvm;
using namespace MIPatternMatch;
using namespace AArch64GISelUtils;

namespace llvm {
class BlockFrequencyInfo;
class ProfileSummaryInfo;
}

namespace {

#define GET_GLOBALISEL_PREDICATE_BITSET
#include "AArch64GenGlobalISel.inc"
#undef GET_GLOBALISEL_PREDICATE_BITSET


class AArch64InstructionSelector : public InstructionSelector {
public:
  AArch64InstructionSelector(const AArch64TargetMachine &TM,
                             const AArch64Subtarget &STI,
                             const AArch64RegisterBankInfo &RBI);

  bool select(MachineInstr &I) override;
  static const char *getName() { return DEBUG_TYPE; }

  void setupMF(MachineFunction &MF, GISelKnownBits *KB,
               CodeGenCoverage &CoverageInfo, ProfileSummaryInfo *PSI,
               BlockFrequencyInfo *BFI) override {
    InstructionSelector::setupMF(MF, KB, CoverageInfo, PSI, BFI);
    MIB.setMF(MF);

    // hasFnAttribute() is expensive to call on every BRCOND selection, so
    // cache it here for each run of the selector.
    ProduceNonFlagSettingCondBr =
        !MF.getFunction().hasFnAttribute(Attribute::SpeculativeLoadHardening);
    MFReturnAddr = Register();

    processPHIs(MF);
  }

private:
  /// tblgen-erated 'select' implementation, used as the initial selector for
  /// the patterns that don't require complex C++.
  bool selectImpl(MachineInstr &I, CodeGenCoverage &CoverageInfo) const;

  // A lowering phase that runs before any selection attempts.
  // Returns true if the instruction was modified.
  bool preISelLower(MachineInstr &I);

  // An early selection function that runs before the selectImpl() call.
  bool earlySelect(MachineInstr &I);

  // Do some preprocessing of G_PHIs before we begin selection.
  void processPHIs(MachineFunction &MF);

  bool earlySelectSHL(MachineInstr &I, MachineRegisterInfo &MRI);

  /// Eliminate same-sized cross-bank copies into stores before selectImpl().
  bool contractCrossBankCopyIntoStore(MachineInstr &I,
                                      MachineRegisterInfo &MRI);

  bool convertPtrAddToAdd(MachineInstr &I, MachineRegisterInfo &MRI);

  bool selectVaStartAAPCS(MachineInstr &I, MachineFunction &MF,
                          MachineRegisterInfo &MRI) const;
  bool selectVaStartDarwin(MachineInstr &I, MachineFunction &MF,
                           MachineRegisterInfo &MRI) const;

  ///@{
  /// Helper functions for selectCompareBranch.
  bool selectCompareBranchFedByFCmp(MachineInstr &I, MachineInstr &FCmp,
                                    MachineIRBuilder &MIB) const;
  bool selectCompareBranchFedByICmp(MachineInstr &I, MachineInstr &ICmp,
                                    MachineIRBuilder &MIB) const;
  bool tryOptCompareBranchFedByICmp(MachineInstr &I, MachineInstr &ICmp,
                                    MachineIRBuilder &MIB) const;
  bool tryOptAndIntoCompareBranch(MachineInstr &AndInst, bool Invert,
                                  MachineBasicBlock *DstMBB,
                                  MachineIRBuilder &MIB) const;
  ///@}

  bool selectCompareBranch(MachineInstr &I, MachineFunction &MF,
                           MachineRegisterInfo &MRI);

  bool selectVectorAshrLshr(MachineInstr &I, MachineRegisterInfo &MRI);
  bool selectVectorSHL(MachineInstr &I, MachineRegisterInfo &MRI);

  // Helper to generate an equivalent of scalar_to_vector into a new register,
  // returned via 'Dst'.
  MachineInstr *emitScalarToVector(unsigned EltSize,
                                   const TargetRegisterClass *DstRC,
                                   Register Scalar,
                                   MachineIRBuilder &MIRBuilder) const;

  /// Emit a lane insert into \p DstReg, or a new vector register if
  /// std::nullopt is provided.
  ///
  /// The lane inserted into is defined by \p LaneIdx. The vector source
  /// register is given by \p SrcReg. The register containing the element is
  /// given by \p EltReg.
  MachineInstr *emitLaneInsert(std::optional<Register> DstReg, Register SrcReg,
                               Register EltReg, unsigned LaneIdx,
                               const RegisterBank &RB,
                               MachineIRBuilder &MIRBuilder) const;

  /// Emit a sequence of instructions representing a constant \p CV for a
  /// vector register \p Dst. (E.g. a MOV, or a load from a constant pool.)
  ///
  /// \returns the last instruction in the sequence on success, and nullptr
  /// otherwise.
  MachineInstr *emitConstantVector(Register Dst, Constant *CV,
                                   MachineIRBuilder &MIRBuilder,
                                   MachineRegisterInfo &MRI);

  bool selectInsertElt(MachineInstr &I, MachineRegisterInfo &MRI);
  bool tryOptConstantBuildVec(MachineInstr &MI, LLT DstTy,
                              MachineRegisterInfo &MRI);
  /// \returns true if a G_BUILD_VECTOR instruction \p MI can be selected as a
  /// SUBREG_TO_REG.
  bool tryOptBuildVecToSubregToReg(MachineInstr &MI, MachineRegisterInfo &MRI);
  bool selectBuildVector(MachineInstr &I, MachineRegisterInfo &MRI);
  bool selectMergeValues(MachineInstr &I, MachineRegisterInfo &MRI);
  bool selectUnmergeValues(MachineInstr &I, MachineRegisterInfo &MRI);

  bool selectShuffleVector(MachineInstr &I, MachineRegisterInfo &MRI);
  bool selectExtractElt(MachineInstr &I, MachineRegisterInfo &MRI);
  bool selectConcatVectors(MachineInstr &I, MachineRegisterInfo &MRI);
  bool selectSplitVectorUnmerge(MachineInstr &I, MachineRegisterInfo &MRI);

  /// Helper function to select vector load intrinsics like
  /// @llvm.aarch64.neon.ld2.*, @llvm.aarch64.neon.ld4.*, etc.
  /// \p Opc is the opcode that the selected instruction should use.
  /// \p NumVecs is the number of vector destinations for the instruction.
  /// \p I is the original G_INTRINSIC_W_SIDE_EFFECTS instruction.
  bool selectVectorLoadIntrinsic(unsigned Opc, unsigned NumVecs,
                                 MachineInstr &I);
  bool selectIntrinsicWithSideEffects(MachineInstr &I,
                                      MachineRegisterInfo &MRI);
  bool selectIntrinsic(MachineInstr &I, MachineRegisterInfo &MRI);
  bool selectVectorICmp(MachineInstr &I, MachineRegisterInfo &MRI);
  bool selectIntrinsicTrunc(MachineInstr &I, MachineRegisterInfo &MRI) const;
  bool selectIntrinsicRound(MachineInstr &I, MachineRegisterInfo &MRI) const;
  bool selectJumpTable(MachineInstr &I, MachineRegisterInfo &MRI);
  bool selectBrJT(MachineInstr &I, MachineRegisterInfo &MRI);
  bool selectTLSGlobalValue(MachineInstr &I, MachineRegisterInfo &MRI);
  bool selectReduction(MachineInstr &I, MachineRegisterInfo &MRI);
  bool selectMOPS(MachineInstr &I, MachineRegisterInfo &MRI);
  bool selectUSMovFromExtend(MachineInstr &I, MachineRegisterInfo &MRI);

  unsigned emitConstantPoolEntry(const Constant *CPVal,
                                 MachineFunction &MF) const;
  MachineInstr *emitLoadFromConstantPool(const Constant *CPVal,
                                         MachineIRBuilder &MIRBuilder) const;

  // Emit a vector concat operation.
  MachineInstr *emitVectorConcat(std::optional<Register> Dst, Register Op1,
                                 Register Op2,
                                 MachineIRBuilder &MIRBuilder) const;

  // Emit an integer compare between LHS and RHS, which checks for Predicate.
  MachineInstr *emitIntegerCompare(MachineOperand &LHS, MachineOperand &RHS,
                                   MachineOperand &Predicate,
                                   MachineIRBuilder &MIRBuilder) const;

  /// Emit a floating point comparison between \p LHS and \p RHS.
  /// \p Pred if given is the intended predicate to use.
  MachineInstr *
  emitFPCompare(Register LHS, Register RHS, MachineIRBuilder &MIRBuilder,
                std::optional<CmpInst::Predicate> = std::nullopt) const;

  MachineInstr *
  emitInstr(unsigned Opcode, std::initializer_list<llvm::DstOp> DstOps,
            std::initializer_list<llvm::SrcOp> SrcOps,
            MachineIRBuilder &MIRBuilder,
            const ComplexRendererFns &RenderFns = std::nullopt) const;
  /// Helper function to emit an add or sub instruction.
  ///
  /// \p AddrModeAndSizeToOpcode must contain each of the opcode variants above
  /// in a specific order.
  ///
  /// Below is an example of the expected input to \p AddrModeAndSizeToOpcode.
  ///
  /// \code
  ///   const std::array<std::array<unsigned, 2>, 4> Table {
  ///    {{AArch64::ADDXri, AArch64::ADDWri},
  ///     {AArch64::ADDXrs, AArch64::ADDWrs},
  ///     {AArch64::ADDXrr, AArch64::ADDWrr},
  ///     {AArch64::SUBXri, AArch64::SUBWri},
  ///     {AArch64::ADDXrx, AArch64::ADDWrx}}};
  /// \endcode
  ///
  /// Each row in the table corresponds to a different addressing mode. Each
  /// column corresponds to a different register size.
  ///
  /// \attention Rows must be structured as follows:
  ///   - Row 0: The ri opcode variants
  ///   - Row 1: The rs opcode variants
  ///   - Row 2: The rr opcode variants
  ///   - Row 3: The ri opcode variants for negative immediates
  ///   - Row 4: The rx opcode variants
  ///
  /// \attention Columns must be structured as follows:
  ///   - Column 0: The 64-bit opcode variants
  ///   - Column 1: The 32-bit opcode variants
  ///
  /// \p Dst is the destination register of the binop to emit.
  /// \p LHS is the left-hand operand of the binop to emit.
  /// \p RHS is the right-hand operand of the binop to emit.
  MachineInstr *emitAddSub(
      const std::array<std::array<unsigned, 2>, 5> &AddrModeAndSizeToOpcode,
      Register Dst, MachineOperand &LHS, MachineOperand &RHS,
      MachineIRBuilder &MIRBuilder) const;
  MachineInstr *emitADD(Register DefReg, MachineOperand &LHS,
                        MachineOperand &RHS,
                        MachineIRBuilder &MIRBuilder) const;
  MachineInstr *emitADDS(Register Dst, MachineOperand &LHS, MachineOperand &RHS,
                         MachineIRBuilder &MIRBuilder) const;
  MachineInstr *emitSUBS(Register Dst, MachineOperand &LHS, MachineOperand &RHS,
                         MachineIRBuilder &MIRBuilder) const;
  MachineInstr *emitCMN(MachineOperand &LHS, MachineOperand &RHS,
                        MachineIRBuilder &MIRBuilder) const;
  MachineInstr *emitTST(MachineOperand &LHS, MachineOperand &RHS,
                        MachineIRBuilder &MIRBuilder) const;
  MachineInstr *emitSelect(Register Dst, Register LHS, Register RHS,
                           AArch64CC::CondCode CC,
                           MachineIRBuilder &MIRBuilder) const;
  MachineInstr *emitExtractVectorElt(std::optional<Register> DstReg,
                                     const RegisterBank &DstRB, LLT ScalarTy,
                                     Register VecReg, unsigned LaneIdx,
                                     MachineIRBuilder &MIRBuilder) const;
  MachineInstr *emitCSINC(Register Dst, Register Src1, Register Src2,
                          AArch64CC::CondCode Pred,
                          MachineIRBuilder &MIRBuilder) const;
  /// Emit a CSet for a FP compare.
  ///
  /// \p Dst is expected to be a 32-bit scalar register.
  MachineInstr *emitCSetForFCmp(Register Dst, CmpInst::Predicate Pred,
                                MachineIRBuilder &MIRBuilder) const;

  /// Emit the overflow op for \p Opcode.
  ///
  /// \p Opcode is expected to be an overflow op's opcode, e.g. G_UADDO,
  /// G_USUBO, etc.
  std::pair<MachineInstr *, AArch64CC::CondCode>
  emitOverflowOp(unsigned Opcode, Register Dst, MachineOperand &LHS,
                 MachineOperand &RHS, MachineIRBuilder &MIRBuilder) const;

  /// Emit expression as a conjunction (a series of CCMP/CFCMP ops).
  /// In some cases this is even possible with OR operations in the expression.
  MachineInstr *emitConjunction(Register Val, AArch64CC::CondCode &OutCC,
                                MachineIRBuilder &MIB) const;
  MachineInstr *emitConditionalComparison(Register LHS, Register RHS,
                                          CmpInst::Predicate CC,
                                          AArch64CC::CondCode Predicate,
                                          AArch64CC::CondCode OutCC,
                                          MachineIRBuilder &MIB) const;
  MachineInstr *emitConjunctionRec(Register Val, AArch64CC::CondCode &OutCC,
                                   bool Negate, Register CCOp,
                                   AArch64CC::CondCode Predicate,
                                   MachineIRBuilder &MIB) const;

  /// Emit a TB(N)Z instruction which tests \p Bit in \p TestReg.
  /// \p IsNegative is true if the test should be "not zero".
  /// This will also optimize the test bit instruction when possible.
  MachineInstr *emitTestBit(Register TestReg, uint64_t Bit, bool IsNegative,
                            MachineBasicBlock *DstMBB,
                            MachineIRBuilder &MIB) const;

  /// Emit a CB(N)Z instruction which branches to \p DestMBB.
  MachineInstr *emitCBZ(Register CompareReg, bool IsNegative,
                        MachineBasicBlock *DestMBB,
                        MachineIRBuilder &MIB) const;

  // Equivalent to the i32shift_a and friends from AArch64InstrInfo.td.
  // We use these manually instead of using the importer since it doesn't
  // support SDNodeXForm.
  ComplexRendererFns selectShiftA_32(const MachineOperand &Root) const;
  ComplexRendererFns selectShiftB_32(const MachineOperand &Root) const;
  ComplexRendererFns selectShiftA_64(const MachineOperand &Root) const;
  ComplexRendererFns selectShiftB_64(const MachineOperand &Root) const;

  ComplexRendererFns select12BitValueWithLeftShift(uint64_t Immed) const;
  ComplexRendererFns selectArithImmed(MachineOperand &Root) const;
  ComplexRendererFns selectNegArithImmed(MachineOperand &Root) const;

  ComplexRendererFns selectAddrModeUnscaled(MachineOperand &Root,
                                            unsigned Size) const;

  ComplexRendererFns selectAddrModeUnscaled8(MachineOperand &Root) const {
    return selectAddrModeUnscaled(Root, 1);
  }
  ComplexRendererFns selectAddrModeUnscaled16(MachineOperand &Root) const {
    return selectAddrModeUnscaled(Root, 2);
  }
  ComplexRendererFns selectAddrModeUnscaled32(MachineOperand &Root) const {
    return selectAddrModeUnscaled(Root, 4);
  }
  ComplexRendererFns selectAddrModeUnscaled64(MachineOperand &Root) const {
    return selectAddrModeUnscaled(Root, 8);
  }
  ComplexRendererFns selectAddrModeUnscaled128(MachineOperand &Root) const {
    return selectAddrModeUnscaled(Root, 16);
  }

  /// Helper to try to fold in a GISEL_ADD_LOW into an immediate, to be used
  /// from complex pattern matchers like selectAddrModeIndexed().
  ComplexRendererFns tryFoldAddLowIntoImm(MachineInstr &RootDef, unsigned Size,
                                          MachineRegisterInfo &MRI) const;

  ComplexRendererFns selectAddrModeIndexed(MachineOperand &Root,
                                           unsigned Size) const;
  template <int Width>
  ComplexRendererFns selectAddrModeIndexed(MachineOperand &Root) const {
    return selectAddrModeIndexed(Root, Width / 8);
  }

  bool isWorthFoldingIntoExtendedReg(MachineInstr &MI,
                                     const MachineRegisterInfo &MRI) const;
  ComplexRendererFns
  selectAddrModeShiftedExtendXReg(MachineOperand &Root,
                                  unsigned SizeInBytes) const;

  /// Returns a \p ComplexRendererFns which contains a base, offset, and whether
  /// or not a shift + extend should be folded into an addressing mode. Returns
  /// None when this is not profitable or possible.
  ComplexRendererFns
  selectExtendedSHL(MachineOperand &Root, MachineOperand &Base,
                    MachineOperand &Offset, unsigned SizeInBytes,
                    bool WantsExt) const;
  ComplexRendererFns selectAddrModeRegisterOffset(MachineOperand &Root) const;
  ComplexRendererFns selectAddrModeXRO(MachineOperand &Root,
                                       unsigned SizeInBytes) const;
  template <int Width>
  ComplexRendererFns selectAddrModeXRO(MachineOperand &Root) const {
    return selectAddrModeXRO(Root, Width / 8);
  }

  ComplexRendererFns selectAddrModeWRO(MachineOperand &Root,
                                       unsigned SizeInBytes) const;
  template <int Width>
  ComplexRendererFns selectAddrModeWRO(MachineOperand &Root) const {
    return selectAddrModeWRO(Root, Width / 8);
  }

  ComplexRendererFns selectShiftedRegister(MachineOperand &Root,
                                           bool AllowROR = false) const;

  ComplexRendererFns selectArithShiftedRegister(MachineOperand &Root) const {
    return selectShiftedRegister(Root);
  }

  ComplexRendererFns selectLogicalShiftedRegister(MachineOperand &Root) const {
    return selectShiftedRegister(Root, true);
  }

  /// Given an extend instruction, determine the correct shift-extend type for
  /// that instruction.
  ///
  /// If the instruction is going to be used in a load or store, pass
  /// \p IsLoadStore = true.
  AArch64_AM::ShiftExtendType
  getExtendTypeForInst(MachineInstr &MI, MachineRegisterInfo &MRI,
                       bool IsLoadStore = false) const;

  /// Move \p Reg to \p RC if \p Reg is not already on \p RC.
  ///
  /// \returns Either \p Reg if no change was necessary, or the new register
  /// created by moving \p Reg.
  ///
  /// Note: This uses emitCopy right now.
  Register moveScalarRegClass(Register Reg, const TargetRegisterClass &RC,
                              MachineIRBuilder &MIB) const;

  ComplexRendererFns selectArithExtendedRegister(MachineOperand &Root) const;

  void renderTruncImm(MachineInstrBuilder &MIB, const MachineInstr &MI,
                      int OpIdx = -1) const;
  void renderLogicalImm32(MachineInstrBuilder &MIB, const MachineInstr &I,
                          int OpIdx = -1) const;
  void renderLogicalImm64(MachineInstrBuilder &MIB, const MachineInstr &I,
                          int OpIdx = -1) const;
  void renderFPImm16(MachineInstrBuilder &MIB, const MachineInstr &MI,
                     int OpIdx = -1) const;
  void renderFPImm32(MachineInstrBuilder &MIB, const MachineInstr &MI,
                     int OpIdx = -1) const;
  void renderFPImm64(MachineInstrBuilder &MIB, const MachineInstr &MI,
                     int OpIdx = -1) const;
  void renderFPImm32SIMDModImmType4(MachineInstrBuilder &MIB,
                                    const MachineInstr &MI,
                                    int OpIdx = -1) const;

  // Materialize a GlobalValue or BlockAddress using a movz+movk sequence.
  void materializeLargeCMVal(MachineInstr &I, const Value *V, unsigned OpFlags);

  // Optimization methods.
  bool tryOptSelect(GSelect &Sel);
  bool tryOptSelectConjunction(GSelect &Sel, MachineInstr &CondMI);
  MachineInstr *tryFoldIntegerCompare(MachineOperand &LHS, MachineOperand &RHS,
                                      MachineOperand &Predicate,
                                      MachineIRBuilder &MIRBuilder) const;

  /// Return true if \p MI is a load or store of \p NumBytes bytes.
  bool isLoadStoreOfNumBytes(const MachineInstr &MI, unsigned NumBytes) const;

  /// Returns true if \p MI is guaranteed to have the high-half of a 64-bit
  /// register zeroed out. In other words, the result of MI has been explicitly
  /// zero extended.
  bool isDef32(const MachineInstr &MI) const;

  const AArch64TargetMachine &TM;
  const AArch64Subtarget &STI;
  const AArch64InstrInfo &TII;
  const AArch64RegisterInfo &TRI;
  const AArch64RegisterBankInfo &RBI;

  bool ProduceNonFlagSettingCondBr = false;

  // Some cached values used during selection.
  // We use LR as a live-in register, and we keep track of it here as it can be
  // clobbered by calls.
  Register MFReturnAddr;

  MachineIRBuilder MIB;

#define GET_GLOBALISEL_PREDICATES_DECL
#include "AArch64GenGlobalISel.inc"
#undef GET_GLOBALISEL_PREDICATES_DECL

// We declare the temporaries used by selectImpl() in the class to minimize the
// cost of constructing placeholder values.
#define GET_GLOBALISEL_TEMPORARIES_DECL
#include "AArch64GenGlobalISel.inc"
#undef GET_GLOBALISEL_TEMPORARIES_DECL
};

} // end anonymous namespace

#define GET_GLOBALISEL_IMPL
#include "AArch64GenGlobalISel.inc"
#undef GET_GLOBALISEL_IMPL

AArch64InstructionSelector::AArch64InstructionSelector(
    const AArch64TargetMachine &TM, const AArch64Subtarget &STI,
    const AArch64RegisterBankInfo &RBI)
    : TM(TM), STI(STI), TII(*STI.getInstrInfo()), TRI(*STI.getRegisterInfo()),
      RBI(RBI),
#define GET_GLOBALISEL_PREDICATES_INIT
#include "AArch64GenGlobalISel.inc"
#undef GET_GLOBALISEL_PREDICATES_INIT
#define GET_GLOBALISEL_TEMPORARIES_INIT
#include "AArch64GenGlobalISel.inc"
#undef GET_GLOBALISEL_TEMPORARIES_INIT
{
}

// FIXME: This should be target-independent, inferred from the types declared
// for each class in the bank.
//
/// Given a register bank, and a type, return the smallest register class that
/// can represent that combination.
static const TargetRegisterClass *
getRegClassForTypeOnBank(LLT Ty, const RegisterBank &RB,
                         bool GetAllRegSet = false) {
  if (RB.getID() == AArch64::GPRRegBankID) {
    if (Ty.getSizeInBits() <= 32)
      return GetAllRegSet ? &AArch64::GPR32allRegClass
                          : &AArch64::GPR32RegClass;
    if (Ty.getSizeInBits() == 64)
      return GetAllRegSet ? &AArch64::GPR64allRegClass
                          : &AArch64::GPR64RegClass;
    if (Ty.getSizeInBits() == 128)
      return &AArch64::XSeqPairsClassRegClass;
    return nullptr;
  }

  if (RB.getID() == AArch64::FPRRegBankID) {
    switch (Ty.getSizeInBits()) {
    case 8:
      return &AArch64::FPR8RegClass;
    case 16:
      return &AArch64::FPR16RegClass;
    case 32:
      return &AArch64::FPR32RegClass;
    case 64:
      return &AArch64::FPR64RegClass;
    case 128:
      return &AArch64::FPR128RegClass;
    }
    return nullptr;
  }

  return nullptr;
}

/// Given a register bank, and size in bits, return the smallest register class
/// that can represent that combination.
static const TargetRegisterClass *
getMinClassForRegBank(const RegisterBank &RB, unsigned SizeInBits,
                      bool GetAllRegSet = false) {
  unsigned RegBankID = RB.getID();

  if (RegBankID == AArch64::GPRRegBankID) {
    if (SizeInBits <= 32)
      return GetAllRegSet ? &AArch64::GPR32allRegClass
                          : &AArch64::GPR32RegClass;
    if (SizeInBits == 64)
      return GetAllRegSet ? &AArch64::GPR64allRegClass
                          : &AArch64::GPR64RegClass;
    if (SizeInBits == 128)
      return &AArch64::XSeqPairsClassRegClass;
  }

  if (RegBankID == AArch64::FPRRegBankID) {
    switch (SizeInBits) {
    default:
      return nullptr;
    case 8:
      return &AArch64::FPR8RegClass;
    case 16:
      return &AArch64::FPR16RegClass;
    case 32:
      return &AArch64::FPR32RegClass;
    case 64:
      return &AArch64::FPR64RegClass;
    case 128:
      return &AArch64::FPR128RegClass;
    }
  }

  return nullptr;
}

/// Returns the correct subregister to use for a given register class.
static bool getSubRegForClass(const TargetRegisterClass *RC,
                              const TargetRegisterInfo &TRI, unsigned &SubReg) {
  switch (TRI.getRegSizeInBits(*RC)) {
  case 8:
    SubReg = AArch64::bsub;
    break;
  case 16:
    SubReg = AArch64::hsub;
    break;
  case 32:
    if (RC != &AArch64::FPR32RegClass)
      SubReg = AArch64::sub_32;
    else
      SubReg = AArch64::ssub;
    break;
  case 64:
    SubReg = AArch64::dsub;
    break;
  default:
    LLVM_DEBUG(
        dbgs() << "Couldn't find appropriate subregister for register class.");
    return false;
  }

  return true;
}

/// Returns the minimum size the given register bank can hold.
static unsigned getMinSizeForRegBank(const RegisterBank &RB) {
  switch (RB.getID()) {
  case AArch64::GPRRegBankID:
    return 32;
  case AArch64::FPRRegBankID:
    return 8;
  default:
    llvm_unreachable("Tried to get minimum size for unknown register bank.");
  }
}

/// Create a REG_SEQUENCE instruction using the registers in \p Regs.
/// Helper function for functions like createDTuple and createQTuple.
///
/// \p RegClassIDs - The list of register class IDs available for some tuple of
/// a scalar class. E.g. QQRegClassID, QQQRegClassID, QQQQRegClassID. This is
/// expected to contain between 2 and 4 tuple classes.
///
/// \p SubRegs - The list of subregister classes associated with each register
/// class ID in \p RegClassIDs. E.g., QQRegClassID should use the qsub0
/// subregister class. The index of each subregister class is expected to
/// correspond with the index of each register class.
///
/// \returns Either the destination register of REG_SEQUENCE instruction that
/// was created, or the 0th element of \p Regs if \p Regs contains a single
/// element.
static Register createTuple(ArrayRef<Register> Regs,
                            const unsigned RegClassIDs[],
                            const unsigned SubRegs[], MachineIRBuilder &MIB) {
  unsigned NumRegs = Regs.size();
  if (NumRegs == 1)
    return Regs[0];
  assert(NumRegs >= 2 && NumRegs <= 4 &&
         "Only support between two and 4 registers in a tuple!");
  const TargetRegisterInfo *TRI = MIB.getMF().getSubtarget().getRegisterInfo();
  auto *DesiredClass = TRI->getRegClass(RegClassIDs[NumRegs - 2]);
  auto RegSequence =
      MIB.buildInstr(TargetOpcode::REG_SEQUENCE, {DesiredClass}, {});
  for (unsigned I = 0, E = Regs.size(); I < E; ++I) {
    RegSequence.addUse(Regs[I]);
    RegSequence.addImm(SubRegs[I]);
  }
  return RegSequence.getReg(0);
}

/// Create a tuple of D-registers using the registers in \p Regs.
static Register createDTuple(ArrayRef<Register> Regs, MachineIRBuilder &MIB) {
  static const unsigned RegClassIDs[] = {
      AArch64::DDRegClassID, AArch64::DDDRegClassID, AArch64::DDDDRegClassID};
  static const unsigned SubRegs[] = {AArch64::dsub0, AArch64::dsub1,
                                     AArch64::dsub2, AArch64::dsub3};
  return createTuple(Regs, RegClassIDs, SubRegs, MIB);
}

/// Create a tuple of Q-registers using the registers in \p Regs.
static Register createQTuple(ArrayRef<Register> Regs, MachineIRBuilder &MIB) {
  static const unsigned RegClassIDs[] = {
      AArch64::QQRegClassID, AArch64::QQQRegClassID, AArch64::QQQQRegClassID};
  static const unsigned SubRegs[] = {AArch64::qsub0, AArch64::qsub1,
                                     AArch64::qsub2, AArch64::qsub3};
  return createTuple(Regs, RegClassIDs, SubRegs, MIB);
}

static std::optional<uint64_t> getImmedFromMO(const MachineOperand &Root) {
  auto &MI = *Root.getParent();
  auto &MBB = *MI.getParent();
  auto &MF = *MBB.getParent();
  auto &MRI = MF.getRegInfo();
  uint64_t Immed;
  if (Root.isImm())
    Immed = Root.getImm();
  else if (Root.isCImm())
    Immed = Root.getCImm()->getZExtValue();
  else if (Root.isReg()) {
    auto ValAndVReg =
        getIConstantVRegValWithLookThrough(Root.getReg(), MRI, true);
    if (!ValAndVReg)
      return std::nullopt;
    Immed = ValAndVReg->Value.getSExtValue();
  } else
    return std::nullopt;
  return Immed;
}

/// Check whether \p I is a currently unsupported binary operation:
/// - it has an unsized type
/// - an operand is not a vreg
/// - all operands are not in the same bank
/// These are checks that should someday live in the verifier, but right now,
/// these are mostly limitations of the aarch64 selector.
static bool unsupportedBinOp(const MachineInstr &I,
                             const AArch64RegisterBankInfo &RBI,
                             const MachineRegisterInfo &MRI,
                             const AArch64RegisterInfo &TRI) {
  LLT Ty = MRI.getType(I.getOperand(0).getReg());
  if (!Ty.isValid()) {
    LLVM_DEBUG(dbgs() << "Generic binop register should be typed\n");
    return true;
  }

  const RegisterBank *PrevOpBank = nullptr;
  for (auto &MO : I.operands()) {
    // FIXME: Support non-register operands.
    if (!MO.isReg()) {
      LLVM_DEBUG(dbgs() << "Generic inst non-reg operands are unsupported\n");
      return true;
    }

    // FIXME: Can generic operations have physical registers operands? If
    // so, this will need to be taught about that, and we'll need to get the
    // bank out of the minimal class for the register.
    // Either way, this needs to be documented (and possibly verified).
    if (!MO.getReg().isVirtual()) {
      LLVM_DEBUG(dbgs() << "Generic inst has physical register operand\n");
      return true;
    }

    const RegisterBank *OpBank = RBI.getRegBank(MO.getReg(), MRI, TRI);
    if (!OpBank) {
      LLVM_DEBUG(dbgs() << "Generic register has no bank or class\n");
      return true;
    }

    if (PrevOpBank && OpBank != PrevOpBank) {
      LLVM_DEBUG(dbgs() << "Generic inst operands have different banks\n");
      return true;
    }
    PrevOpBank = OpBank;
  }
  return false;
}

/// Select the AArch64 opcode for the basic binary operation \p GenericOpc
/// (such as G_OR or G_SDIV), appropriate for the register bank \p RegBankID
/// and of size \p OpSize.
/// \returns \p GenericOpc if the combination is unsupported.
static unsigned selectBinaryOp(unsigned GenericOpc, unsigned RegBankID,
                               unsigned OpSize) {
  switch (RegBankID) {
  case AArch64::GPRRegBankID:
    if (OpSize == 32) {
      switch (GenericOpc) {
      case TargetOpcode::G_SHL:
        return AArch64::LSLVWr;
      case TargetOpcode::G_LSHR:
        return AArch64::LSRVWr;
      case TargetOpcode::G_ASHR:
        return AArch64::ASRVWr;
      default:
        return GenericOpc;
      }
    } else if (OpSize == 64) {
      switch (GenericOpc) {
      case TargetOpcode::G_PTR_ADD:
        return AArch64::ADDXrr;
      case TargetOpcode::G_SHL:
        return AArch64::LSLVXr;
      case TargetOpcode::G_LSHR:
        return AArch64::LSRVXr;
      case TargetOpcode::G_ASHR:
        return AArch64::ASRVXr;
      default:
        return GenericOpc;
      }
    }
    break;
  case AArch64::FPRRegBankID:
    switch (OpSize) {
    case 32:
      switch (GenericOpc) {
      case TargetOpcode::G_FADD:
        return AArch64::FADDSrr;
      case TargetOpcode::G_FSUB:
        return AArch64::FSUBSrr;
      case TargetOpcode::G_FMUL:
        return AArch64::FMULSrr;
      case TargetOpcode::G_FDIV:
        return AArch64::FDIVSrr;
      default:
        return GenericOpc;
      }
    case 64:
      switch (GenericOpc) {
      case TargetOpcode::G_FADD:
        return AArch64::FADDDrr;
      case TargetOpcode::G_FSUB:
        return AArch64::FSUBDrr;
      case TargetOpcode::G_FMUL:
        return AArch64::FMULDrr;
      case TargetOpcode::G_FDIV:
        return AArch64::FDIVDrr;
      case TargetOpcode::G_OR:
        return AArch64::ORRv8i8;
      default:
        return GenericOpc;
      }
    }
    break;
  }
  return GenericOpc;
}

/// Select the AArch64 opcode for the G_LOAD or G_STORE operation \p GenericOpc,
/// appropriate for the (value) register bank \p RegBankID and of memory access
/// size \p OpSize.  This returns the variant with the base+unsigned-immediate
/// addressing mode (e.g., LDRXui).
/// \returns \p GenericOpc if the combination is unsupported.
static unsigned selectLoadStoreUIOp(unsigned GenericOpc, unsigned RegBankID,
                                    unsigned OpSize) {
  const bool isStore = GenericOpc == TargetOpcode::G_STORE;
  switch (RegBankID) {
  case AArch64::GPRRegBankID:
    switch (OpSize) {
    case 8:
      return isStore ? AArch64::STRBBui : AArch64::LDRBBui;
    case 16:
      return isStore ? AArch64::STRHHui : AArch64::LDRHHui;
    case 32:
      return isStore ? AArch64::STRWui : AArch64::LDRWui;
    case 64:
      return isStore ? AArch64::STRXui : AArch64::LDRXui;
    }
    break;
  case AArch64::FPRRegBankID:
    switch (OpSize) {
    case 8:
      return isStore ? AArch64::STRBui : AArch64::LDRBui;
    case 16:
      return isStore ? AArch64::STRHui : AArch64::LDRHui;
    case 32:
      return isStore ? AArch64::STRSui : AArch64::LDRSui;
    case 64:
      return isStore ? AArch64::STRDui : AArch64::LDRDui;
    case 128:
      return isStore ? AArch64::STRQui : AArch64::LDRQui;
    }
    break;
  }
  return GenericOpc;
}

/// Helper function for selectCopy. Inserts a subregister copy from \p SrcReg
/// to \p *To.
///
/// E.g "To = COPY SrcReg:SubReg"
static bool copySubReg(MachineInstr &I, MachineRegisterInfo &MRI,
                       const RegisterBankInfo &RBI, Register SrcReg,
                       const TargetRegisterClass *To, unsigned SubReg) {
  assert(SrcReg.isValid() && "Expected a valid source register?");
  assert(To && "Destination register class cannot be null");
  assert(SubReg && "Expected a valid subregister");

  MachineIRBuilder MIB(I);
  auto SubRegCopy =
      MIB.buildInstr(TargetOpcode::COPY, {To}, {}).addReg(SrcReg, 0, SubReg);
  MachineOperand &RegOp = I.getOperand(1);
  RegOp.setReg(SubRegCopy.getReg(0));

  // It's possible that the destination register won't be constrained. Make
  // sure that happens.
  if (!I.getOperand(0).getReg().isPhysical())
    RBI.constrainGenericRegister(I.getOperand(0).getReg(), *To, MRI);

  return true;
}

/// Helper function to get the source and destination register classes for a
/// copy. Returns a std::pair containing the source register class for the
/// copy, and the destination register class for the copy. If a register class
/// cannot be determined, then it will be nullptr.
static std::pair<const TargetRegisterClass *, const TargetRegisterClass *>
getRegClassesForCopy(MachineInstr &I, const TargetInstrInfo &TII,
                     MachineRegisterInfo &MRI, const TargetRegisterInfo &TRI,
                     const RegisterBankInfo &RBI) {
  Register DstReg = I.getOperand(0).getReg();
  Register SrcReg = I.getOperand(1).getReg();
  const RegisterBank &DstRegBank = *RBI.getRegBank(DstReg, MRI, TRI);
  const RegisterBank &SrcRegBank = *RBI.getRegBank(SrcReg, MRI, TRI);
  unsigned DstSize = RBI.getSizeInBits(DstReg, MRI, TRI);
  unsigned SrcSize = RBI.getSizeInBits(SrcReg, MRI, TRI);

  // Special casing for cross-bank copies of s1s. We can technically represent
  // a 1-bit value with any size of register. The minimum size for a GPR is 32
  // bits. So, we need to put the FPR on 32 bits as well.
  //
  // FIXME: I'm not sure if this case holds true outside of copies. If it does,
  // then we can pull it into the helpers that get the appropriate class for a
  // register bank. Or make a new helper that carries along some constraint
  // information.
  if (SrcRegBank != DstRegBank && (DstSize == 1 && SrcSize == 1))
    SrcSize = DstSize = 32;

  return {getMinClassForRegBank(SrcRegBank, SrcSize, true),
          getMinClassForRegBank(DstRegBank, DstSize, true)};
}

// FIXME: We need some sort of API in RBI/TRI to allow generic code to
// constrain operands of simple instructions given a TargetRegisterClass
// and LLT
static bool selectDebugInstr(MachineInstr &I, MachineRegisterInfo &MRI,
                             const RegisterBankInfo &RBI) {
  for (MachineOperand &MO : I.operands()) {
    if (!MO.isReg())
      continue;
    Register Reg = MO.getReg();
    if (!Reg)
      continue;
    if (Reg.isPhysical())
      continue;
    LLT Ty = MRI.getType(Reg);
    const RegClassOrRegBank &RegClassOrBank = MRI.getRegClassOrRegBank(Reg);
    const TargetRegisterClass *RC =
        RegClassOrBank.dyn_cast<const TargetRegisterClass *>();
    if (!RC) {
      const RegisterBank &RB = *RegClassOrBank.get<const RegisterBank *>();
      RC = getRegClassForTypeOnBank(Ty, RB);
      if (!RC) {
        LLVM_DEBUG(
            dbgs() << "Warning: DBG_VALUE operand has unexpected size/bank\n");
        break;
      }
    }
    RBI.constrainGenericRegister(Reg, *RC, MRI);
  }

  return true;
}

static bool selectCopy(MachineInstr &I, const TargetInstrInfo &TII,
                       MachineRegisterInfo &MRI, const TargetRegisterInfo &TRI,
                       const RegisterBankInfo &RBI) {
  Register DstReg = I.getOperand(0).getReg();
  Register SrcReg = I.getOperand(1).getReg();
  const RegisterBank &DstRegBank = *RBI.getRegBank(DstReg, MRI, TRI);
  const RegisterBank &SrcRegBank = *RBI.getRegBank(SrcReg, MRI, TRI);

  // Find the correct register classes for the source and destination registers.
  const TargetRegisterClass *SrcRC;
  const TargetRegisterClass *DstRC;
  std::tie(SrcRC, DstRC) = getRegClassesForCopy(I, TII, MRI, TRI, RBI);

  if (!DstRC) {
    LLVM_DEBUG(dbgs() << "Unexpected dest size "
                      << RBI.getSizeInBits(DstReg, MRI, TRI) << '\n');
    return false;
  }

  // Is this a copy? If so, then we may need to insert a subregister copy.
  if (I.isCopy()) {
    // Yes. Check if there's anything to fix up.
    if (!SrcRC) {
      LLVM_DEBUG(dbgs() << "Couldn't determine source register class\n");
      return false;
    }

    unsigned SrcSize = TRI.getRegSizeInBits(*SrcRC);
    unsigned DstSize = TRI.getRegSizeInBits(*DstRC);
    unsigned SubReg;

    // If the source bank doesn't support a subregister copy small enough,
    // then we first need to copy to the destination bank.
    if (getMinSizeForRegBank(SrcRegBank) > DstSize) {
      const TargetRegisterClass *DstTempRC =
          getMinClassForRegBank(DstRegBank, SrcSize, /* GetAllRegSet */ true);
      getSubRegForClass(DstRC, TRI, SubReg);

      MachineIRBuilder MIB(I);
      auto Copy = MIB.buildCopy({DstTempRC}, {SrcReg});
      copySubReg(I, MRI, RBI, Copy.getReg(0), DstRC, SubReg);
    } else if (SrcSize > DstSize) {
      // If the source register is bigger than the destination we need to
      // perform a subregister copy.
      const TargetRegisterClass *SubRegRC =
          getMinClassForRegBank(SrcRegBank, DstSize, /* GetAllRegSet */ true);
      getSubRegForClass(SubRegRC, TRI, SubReg);
      copySubReg(I, MRI, RBI, SrcReg, DstRC, SubReg);
    } else if (DstSize > SrcSize) {
      // If the destination register is bigger than the source we need to do
      // a promotion using SUBREG_TO_REG.
      const TargetRegisterClass *PromotionRC =
          getMinClassForRegBank(SrcRegBank, DstSize, /* GetAllRegSet */ true);
      getSubRegForClass(SrcRC, TRI, SubReg);

      Register PromoteReg = MRI.createVirtualRegister(PromotionRC);
      BuildMI(*I.getParent(), I, I.getDebugLoc(),
              TII.get(AArch64::SUBREG_TO_REG), PromoteReg)
          .addImm(0)
          .addUse(SrcReg)
          .addImm(SubReg);
      MachineOperand &RegOp = I.getOperand(1);
      RegOp.setReg(PromoteReg);
    }

    // If the destination is a physical register, then there's nothing to
    // change, so we're done.
    if (DstReg.isPhysical())
      return true;
  }

  // No need to constrain SrcReg. It will get constrained when we hit another
  // of its use or its defs. Copies do not have constraints.
  if (!RBI.constrainGenericRegister(DstReg, *DstRC, MRI)) {
    LLVM_DEBUG(dbgs() << "Failed to constrain " << TII.getName(I.getOpcode())
                      << " operand\n");
    return false;
  }

  // If this a GPR ZEXT that we want to just reduce down into a copy.
  // The sizes will be mismatched with the source < 32b but that's ok.
  if (I.getOpcode() == TargetOpcode::G_ZEXT) {
    I.setDesc(TII.get(AArch64::COPY));
    assert(SrcRegBank.getID() == AArch64::GPRRegBankID);
    return selectCopy(I, TII, MRI, TRI, RBI);
  }

  I.setDesc(TII.get(AArch64::COPY));
  return true;
}

static unsigned selectFPConvOpc(unsigned GenericOpc, LLT DstTy, LLT SrcTy) {
  if (!DstTy.isScalar() || !SrcTy.isScalar())
    return GenericOpc;

  const unsigned DstSize = DstTy.getSizeInBits();
  const unsigned SrcSize = SrcTy.getSizeInBits();

  switch (DstSize) {
  case 32:
    switch (SrcSize) {
    case 32:
      switch (GenericOpc) {
      case TargetOpcode::G_SITOFP:
        return AArch64::SCVTFUWSri;
      case TargetOpcode::G_UITOFP:
        return AArch64::UCVTFUWSri;
      case TargetOpcode::G_FPTOSI:
        return AArch64::FCVTZSUWSr;
      case TargetOpcode::G_FPTOUI:
        return AArch64::FCVTZUUWSr;
      default:
        return GenericOpc;
      }
    case 64:
      switch (GenericOpc) {
      case TargetOpcode::G_SITOFP:
        return AArch64::SCVTFUXSri;
      case TargetOpcode::G_UITOFP:
        return AArch64::UCVTFUXSri;
      case TargetOpcode::G_FPTOSI:
        return AArch64::FCVTZSUWDr;
      case TargetOpcode::G_FPTOUI:
        return AArch64::FCVTZUUWDr;
      default:
        return GenericOpc;
      }
    default:
      return GenericOpc;
    }
  case 64:
    switch (SrcSize) {
    case 32:
      switch (GenericOpc) {
      case TargetOpcode::G_SITOFP:
        return AArch64::SCVTFUWDri;
      case TargetOpcode::G_UITOFP:
        return AArch64::UCVTFUWDri;
      case TargetOpcode::G_FPTOSI:
        return AArch64::FCVTZSUXSr;
      case TargetOpcode::G_FPTOUI:
        return AArch64::FCVTZUUXSr;
      default:
        return GenericOpc;
      }
    case 64:
      switch (GenericOpc) {
      case TargetOpcode::G_SITOFP:
        return AArch64::SCVTFUXDri;
      case TargetOpcode::G_UITOFP:
        return AArch64::UCVTFUXDri;
      case TargetOpcode::G_FPTOSI:
        return AArch64::FCVTZSUXDr;
      case TargetOpcode::G_FPTOUI:
        return AArch64::FCVTZUUXDr;
      default:
        return GenericOpc;
      }
    default:
      return GenericOpc;
    }
  default:
    return GenericOpc;
  };
  return GenericOpc;
}

MachineInstr *
AArch64InstructionSelector::emitSelect(Register Dst, Register True,
                                       Register False, AArch64CC::CondCode CC,
                                       MachineIRBuilder &MIB) const {
  MachineRegisterInfo &MRI = *MIB.getMRI();
  assert(RBI.getRegBank(False, MRI, TRI)->getID() ==
             RBI.getRegBank(True, MRI, TRI)->getID() &&
         "Expected both select operands to have the same regbank?");
  LLT Ty = MRI.getType(True);
  if (Ty.isVector())
    return nullptr;
  const unsigned Size = Ty.getSizeInBits();
  assert((Size == 32 || Size == 64) &&
         "Expected 32 bit or 64 bit select only?");
  const bool Is32Bit = Size == 32;
  if (RBI.getRegBank(True, MRI, TRI)->getID() != AArch64::GPRRegBankID) {
    unsigned Opc = Is32Bit ? AArch64::FCSELSrrr : AArch64::FCSELDrrr;
    auto FCSel = MIB.buildInstr(Opc, {Dst}, {True, False}).addImm(CC);
    constrainSelectedInstRegOperands(*FCSel, TII, TRI, RBI);
    return &*FCSel;
  }

  // By default, we'll try and emit a CSEL.
  unsigned Opc = Is32Bit ? AArch64::CSELWr : AArch64::CSELXr;
  bool Optimized = false;
  auto TryFoldBinOpIntoSelect = [&Opc, Is32Bit, &CC, &MRI,
                                 &Optimized](Register &Reg, Register &OtherReg,
                                             bool Invert) {
    if (Optimized)
      return false;

    // Attempt to fold:
    //
    // %sub = G_SUB 0, %x
    // %select = G_SELECT cc, %reg, %sub
    //
    // Into:
    // %select = CSNEG %reg, %x, cc
    Register MatchReg;
    if (mi_match(Reg, MRI, m_Neg(m_Reg(MatchReg)))) {
      Opc = Is32Bit ? AArch64::CSNEGWr : AArch64::CSNEGXr;
      Reg = MatchReg;
      if (Invert) {
        CC = AArch64CC::getInvertedCondCode(CC);
        std::swap(Reg, OtherReg);
      }
      return true;
    }

    // Attempt to fold:
    //
    // %xor = G_XOR %x, -1
    // %select = G_SELECT cc, %reg, %xor
    //
    // Into:
    // %select = CSINV %reg, %x, cc
    if (mi_match(Reg, MRI, m_Not(m_Reg(MatchReg)))) {
      Opc = Is32Bit ? AArch64::CSINVWr : AArch64::CSINVXr;
      Reg = MatchReg;
      if (Invert) {
        CC = AArch64CC::getInvertedCondCode(CC);
        std::swap(Reg, OtherReg);
      }
      return true;
    }

    // Attempt to fold:
    //
    // %add = G_ADD %x, 1
    // %select = G_SELECT cc, %reg, %add
    //
    // Into:
    // %select = CSINC %reg, %x, cc
    if (mi_match(Reg, MRI,
                 m_any_of(m_GAdd(m_Reg(MatchReg), m_SpecificICst(1)),
                          m_GPtrAdd(m_Reg(MatchReg), m_SpecificICst(1))))) {
      Opc = Is32Bit ? AArch64::CSINCWr : AArch64::CSINCXr;
      Reg = MatchReg;
      if (Invert) {
        CC = AArch64CC::getInvertedCondCode(CC);
        std::swap(Reg, OtherReg);
      }
      return true;
    }

    return false;
  };

  // Helper lambda which tries to use CSINC/CSINV for the instruction when its
  // true/false values are constants.
  // FIXME: All of these patterns already exist in tablegen. We should be
  // able to import these.
  auto TryOptSelectCst = [&Opc, &True, &False, &CC, Is32Bit, &MRI,
                          &Optimized]() {
    if (Optimized)
      return false;
    auto TrueCst = getIConstantVRegValWithLookThrough(True, MRI);
    auto FalseCst = getIConstantVRegValWithLookThrough(False, MRI);
    if (!TrueCst && !FalseCst)
      return false;

    Register ZReg = Is32Bit ? AArch64::WZR : AArch64::XZR;
    if (TrueCst && FalseCst) {
      int64_t T = TrueCst->Value.getSExtValue();
      int64_t F = FalseCst->Value.getSExtValue();

      if (T == 0 && F == 1) {
        // G_SELECT cc, 0, 1 -> CSINC zreg, zreg, cc
        Opc = Is32Bit ? AArch64::CSINCWr : AArch64::CSINCXr;
        True = ZReg;
        False = ZReg;
        return true;
      }

      if (T == 0 && F == -1) {
        // G_SELECT cc 0, -1 -> CSINV zreg, zreg cc
        Opc = Is32Bit ? AArch64::CSINVWr : AArch64::CSINVXr;
        True = ZReg;
        False = ZReg;
        return true;
      }
    }

    if (TrueCst) {
      int64_t T = TrueCst->Value.getSExtValue();
      if (T == 1) {
        // G_SELECT cc, 1, f -> CSINC f, zreg, inv_cc
        Opc = Is32Bit ? AArch64::CSINCWr : AArch64::CSINCXr;
        True = False;
        False = ZReg;
        CC = AArch64CC::getInvertedCondCode(CC);
        return true;
      }

      if (T == -1) {
        // G_SELECT cc, -1, f -> CSINV f, zreg, inv_cc
        Opc = Is32Bit ? AArch64::CSINVWr : AArch64::CSINVXr;
        True = False;
        False = ZReg;
        CC = AArch64CC::getInvertedCondCode(CC);
        return true;
      }
    }

    if (FalseCst) {
      int64_t F = FalseCst->Value.getSExtValue();
      if (F == 1) {
        // G_SELECT cc, t, 1 -> CSINC t, zreg, cc
        Opc = Is32Bit ? AArch64::CSINCWr : AArch64::CSINCXr;
        False = ZReg;
        return true;
      }

      if (F == -1) {
        // G_SELECT cc, t, -1 -> CSINC t, zreg, cc
        Opc = Is32Bit ? AArch64::CSINVWr : AArch64::CSINVXr;
        False = ZReg;
        return true;
      }
    }
    return false;
  };

  Optimized |= TryFoldBinOpIntoSelect(False, True, /*Invert = */ false);
  Optimized |= TryFoldBinOpIntoSelect(True, False, /*Invert = */ true);
  Optimized |= TryOptSelectCst();
  auto SelectInst = MIB.buildInstr(Opc, {Dst}, {True, False}).addImm(CC);
  constrainSelectedInstRegOperands(*SelectInst, TII, TRI, RBI);
  return &*SelectInst;
}

static AArch64CC::CondCode changeICMPPredToAArch64CC(CmpInst::Predicate P) {
  switch (P) {
  default:
    llvm_unreachable("Unknown condition code!");
  case CmpInst::ICMP_NE:
    return AArch64CC::NE;
  case CmpInst::ICMP_EQ:
    return AArch64CC::EQ;
  case CmpInst::ICMP_SGT:
    return AArch64CC::GT;
  case CmpInst::ICMP_SGE:
    return AArch64CC::GE;
  case CmpInst::ICMP_SLT:
    return AArch64CC::LT;
  case CmpInst::ICMP_SLE:
    return AArch64CC::LE;
  case CmpInst::ICMP_UGT:
    return AArch64CC::HI;
  case CmpInst::ICMP_UGE:
    return AArch64CC::HS;
  case CmpInst::ICMP_ULT:
    return AArch64CC::LO;
  case CmpInst::ICMP_ULE:
    return AArch64CC::LS;
  }
}

/// changeFPCCToORAArch64CC - Convert an IR fp condition code to an AArch64 CC.
static void changeFPCCToORAArch64CC(CmpInst::Predicate CC,
                                    AArch64CC::CondCode &CondCode,
                                    AArch64CC::CondCode &CondCode2) {
  CondCode2 = AArch64CC::AL;
  switch (CC) {
  default:
    llvm_unreachable("Unknown FP condition!");
  case CmpInst::FCMP_OEQ:
    CondCode = AArch64CC::EQ;
    break;
  case CmpInst::FCMP_OGT:
    CondCode = AArch64CC::GT;
    break;
  case CmpInst::FCMP_OGE:
    CondCode = AArch64CC::GE;
    break;
  case CmpInst::FCMP_OLT:
    CondCode = AArch64CC::MI;
    break;
  case CmpInst::FCMP_OLE:
    CondCode = AArch64CC::LS;
    break;
  case CmpInst::FCMP_ONE:
    CondCode = AArch64CC::MI;
    CondCode2 = AArch64CC::GT;
    break;
  case CmpInst::FCMP_ORD:
    CondCode = AArch64CC::VC;
    break;
  case CmpInst::FCMP_UNO:
    CondCode = AArch64CC::VS;
    break;
  case CmpInst::FCMP_UEQ:
    CondCode = AArch64CC::EQ;
    CondCode2 = AArch64CC::VS;
    break;
  case CmpInst::FCMP_UGT:
    CondCode = AArch64CC::HI;
    break;
  case CmpInst::FCMP_UGE:
    CondCode = AArch64CC::PL;
    break;
  case CmpInst::FCMP_ULT:
    CondCode = AArch64CC::LT;
    break;
  case CmpInst::FCMP_ULE:
    CondCode = AArch64CC::LE;
    break;
  case CmpInst::FCMP_UNE:
    CondCode = AArch64CC::NE;
    break;
  }
}

/// Convert an IR fp condition code to an AArch64 CC.
/// This differs from changeFPCCToAArch64CC in that it returns cond codes that
/// should be AND'ed instead of OR'ed.
static void changeFPCCToANDAArch64CC(CmpInst::Predicate CC,
                                     AArch64CC::CondCode &CondCode,
                                     AArch64CC::CondCode &CondCode2) {
  CondCode2 = AArch64CC::AL;
  switch (CC) {
  default:
    changeFPCCToORAArch64CC(CC, CondCode, CondCode2);
    assert(CondCode2 == AArch64CC::AL);
    break;
  case CmpInst::FCMP_ONE:
    // (a one b)
    // == ((a olt b) || (a ogt b))
    // == ((a ord b) && (a une b))
    CondCode = AArch64CC::VC;
    CondCode2 = AArch64CC::NE;
    break;
  case CmpInst::FCMP_UEQ:
    // (a ueq b)
    // == ((a uno b) || (a oeq b))
    // == ((a ule b) && (a uge b))
    CondCode = AArch64CC::PL;
    CondCode2 = AArch64CC::LE;
    break;
  }
}

/// Return a register which can be used as a bit to test in a TB(N)Z.
static Register getTestBitReg(Register Reg, uint64_t &Bit, bool &Invert,
                              MachineRegisterInfo &MRI) {
  assert(Reg.isValid() && "Expected valid register!");
  bool HasZext = false;
  while (MachineInstr *MI = getDefIgnoringCopies(Reg, MRI)) {
    unsigned Opc = MI->getOpcode();

    if (!MI->getOperand(0).isReg() ||
        !MRI.hasOneNonDBGUse(MI->getOperand(0).getReg()))
      break;

    // (tbz (any_ext x), b) -> (tbz x, b) if we don't use the extended bits.
    //
    // (tbz (trunc x), b) -> (tbz x, b) is always safe, because the bit number
    // on the truncated x is the same as the bit number on x.
    if (Opc == TargetOpcode::G_ANYEXT || Opc == TargetOpcode::G_ZEXT ||
        Opc == TargetOpcode::G_TRUNC) {
      if (Opc == TargetOpcode::G_ZEXT)
        HasZext = true;

      Register NextReg = MI->getOperand(1).getReg();
      // Did we find something worth folding?
      if (!NextReg.isValid() || !MRI.hasOneNonDBGUse(NextReg))
        break;

      // NextReg is worth folding. Keep looking.
      Reg = NextReg;
      continue;
    }

    // Attempt to find a suitable operation with a constant on one side.
    std::optional<uint64_t> C;
    Register TestReg;
    switch (Opc) {
    default:
      break;
    case TargetOpcode::G_AND:
    case TargetOpcode::G_XOR: {
      TestReg = MI->getOperand(1).getReg();
      Register ConstantReg = MI->getOperand(2).getReg();
      auto VRegAndVal = getIConstantVRegValWithLookThrough(ConstantReg, MRI);
      if (!VRegAndVal) {
        // AND commutes, check the other side for a constant.
        // FIXME: Can we canonicalize the constant so that it's always on the
        // same side at some point earlier?
        std::swap(ConstantReg, TestReg);
        VRegAndVal = getIConstantVRegValWithLookThrough(ConstantReg, MRI);
      }
      if (VRegAndVal) {
        if (HasZext)
          C = VRegAndVal->Value.getZExtValue();
        else
          C = VRegAndVal->Value.getSExtValue();
      }
      break;
    }
    case TargetOpcode::G_ASHR:
    case TargetOpcode::G_LSHR:
    case TargetOpcode::G_SHL: {
      TestReg = MI->getOperand(1).getReg();
      auto VRegAndVal =
          getIConstantVRegValWithLookThrough(MI->getOperand(2).getReg(), MRI);
      if (VRegAndVal)
        C = VRegAndVal->Value.getSExtValue();
      break;
    }
    }

    // Didn't find a constant or viable register. Bail out of the loop.
    if (!C || !TestReg.isValid())
      break;

    // We found a suitable instruction with a constant. Check to see if we can
    // walk through the instruction.
    Register NextReg;
    unsigned TestRegSize = MRI.getType(TestReg).getSizeInBits();
    switch (Opc) {
    default:
      break;
    case TargetOpcode::G_AND:
      // (tbz (and x, m), b) -> (tbz x, b) when the b-th bit of m is set.
      if ((*C >> Bit) & 1)
        NextReg = TestReg;
      break;
    case TargetOpcode::G_SHL:
      // (tbz (shl x, c), b) -> (tbz x, b-c) when b-c is positive and fits in
      // the type of the register.
      if (*C <= Bit && (Bit - *C) < TestRegSize) {
        NextReg = TestReg;
        Bit = Bit - *C;
      }
      break;
    case TargetOpcode::G_ASHR:
      // (tbz (ashr x, c), b) -> (tbz x, b+c) or (tbz x, msb) if b+c is > # bits
      // in x
      NextReg = TestReg;
      Bit = Bit + *C;
      if (Bit >= TestRegSize)
        Bit = TestRegSize - 1;
      break;
    case TargetOpcode::G_LSHR:
      // (tbz (lshr x, c), b) -> (tbz x, b+c) when b + c is < # bits in x
      if ((Bit + *C) < TestRegSize) {
        NextReg = TestReg;
        Bit = Bit + *C;
      }
      break;
    case TargetOpcode::G_XOR:
      // We can walk through a G_XOR by inverting whether we use tbz/tbnz when
      // appropriate.
      //
      // e.g. If x' = xor x, c, and the b-th bit is set in c then
      //
      // tbz x', b -> tbnz x, b
      //
      // Because x' only has the b-th bit set if x does not.
      if ((*C >> Bit) & 1)
        Invert = !Invert;
      NextReg = TestReg;
      break;
    }

    // Check if we found anything worth folding.
    if (!NextReg.isValid())
      return Reg;
    Reg = NextReg;
  }

  return Reg;
}

MachineInstr *AArch64InstructionSelector::emitTestBit(
    Register TestReg, uint64_t Bit, bool IsNegative, MachineBasicBlock *DstMBB,
    MachineIRBuilder &MIB) const {
  assert(TestReg.isValid());
  assert(ProduceNonFlagSettingCondBr &&
         "Cannot emit TB(N)Z with speculation tracking!");
  MachineRegisterInfo &MRI = *MIB.getMRI();

  // Attempt to optimize the test bit by walking over instructions.
  TestReg = getTestBitReg(TestReg, Bit, IsNegative, MRI);
  LLT Ty = MRI.getType(TestReg);
  unsigned Size = Ty.getSizeInBits();
  assert(!Ty.isVector() && "Expected a scalar!");
  assert(Bit < 64 && "Bit is too large!");

  // When the test register is a 64-bit register, we have to narrow to make
  // TBNZW work.
  bool UseWReg = Bit < 32;
  unsigned NecessarySize = UseWReg ? 32 : 64;
  if (Size != NecessarySize)
    TestReg = moveScalarRegClass(
        TestReg, UseWReg ? AArch64::GPR32RegClass : AArch64::GPR64RegClass,
        MIB);

  static const unsigned OpcTable[2][2] = {{AArch64::TBZX, AArch64::TBNZX},
                                          {AArch64::TBZW, AArch64::TBNZW}};
  unsigned Opc = OpcTable[UseWReg][IsNegative];
  auto TestBitMI =
      MIB.buildInstr(Opc).addReg(TestReg).addImm(Bit).addMBB(DstMBB);
  constrainSelectedInstRegOperands(*TestBitMI, TII, TRI, RBI);
  return &*TestBitMI;
}

bool AArch64InstructionSelector::tryOptAndIntoCompareBranch(
    MachineInstr &AndInst, bool Invert, MachineBasicBlock *DstMBB,
    MachineIRBuilder &MIB) const {
  assert(AndInst.getOpcode() == TargetOpcode::G_AND && "Expected G_AND only?");
  // Given something like this:
  //
  //  %x = ...Something...
  //  %one = G_CONSTANT i64 1
  //  %zero = G_CONSTANT i64 0
  //  %and = G_AND %x, %one
  //  %cmp = G_ICMP intpred(ne), %and, %zero
  //  %cmp_trunc = G_TRUNC %cmp
  //  G_BRCOND %cmp_trunc, %bb.3
  //
  // We want to try and fold the AND into the G_BRCOND and produce either a
  // TBNZ (when we have intpred(ne)) or a TBZ (when we have intpred(eq)).
  //
  // In this case, we'd get
  //
  // TBNZ %x %bb.3
  //

  // Check if the AND has a constant on its RHS which we can use as a mask.
  // If it's a power of 2, then it's the same as checking a specific bit.
  // (e.g, ANDing with 8 == ANDing with 000...100 == testing if bit 3 is set)
  auto MaybeBit = getIConstantVRegValWithLookThrough(
      AndInst.getOperand(2).getReg(), *MIB.getMRI());
  if (!MaybeBit)
    return false;

  int32_t Bit = MaybeBit->Value.exactLogBase2();
  if (Bit < 0)
    return false;

  Register TestReg = AndInst.getOperand(1).getReg();

  // Emit a TB(N)Z.
  emitTestBit(TestReg, Bit, Invert, DstMBB, MIB);
  return true;
}

MachineInstr *AArch64InstructionSelector::emitCBZ(Register CompareReg,
                                                  bool IsNegative,
                                                  MachineBasicBlock *DestMBB,
                                                  MachineIRBuilder &MIB) const {
  assert(ProduceNonFlagSettingCondBr && "CBZ does not set flags!");
  MachineRegisterInfo &MRI = *MIB.getMRI();
  assert(RBI.getRegBank(CompareReg, MRI, TRI)->getID() ==
             AArch64::GPRRegBankID &&
         "Expected GPRs only?");
  auto Ty = MRI.getType(CompareReg);
  unsigned Width = Ty.getSizeInBits();
  assert(!Ty.isVector() && "Expected scalar only?");
  assert(Width <= 64 && "Expected width to be at most 64?");
  static const unsigned OpcTable[2][2] = {{AArch64::CBZW, AArch64::CBZX},
                                          {AArch64::CBNZW, AArch64::CBNZX}};
  unsigned Opc = OpcTable[IsNegative][Width == 64];
  auto BranchMI = MIB.buildInstr(Opc, {}, {CompareReg}).addMBB(DestMBB);
  constrainSelectedInstRegOperands(*BranchMI, TII, TRI, RBI);
  return &*BranchMI;
}

bool AArch64InstructionSelector::selectCompareBranchFedByFCmp(
    MachineInstr &I, MachineInstr &FCmp, MachineIRBuilder &MIB) const {
  assert(FCmp.getOpcode() == TargetOpcode::G_FCMP);
  assert(I.getOpcode() == TargetOpcode::G_BRCOND);
  // Unfortunately, the mapping of LLVM FP CC's onto AArch64 CC's isn't
  // totally clean.  Some of them require two branches to implement.
  auto Pred = (CmpInst::Predicate)FCmp.getOperand(1).getPredicate();
  emitFPCompare(FCmp.getOperand(2).getReg(), FCmp.getOperand(3).getReg(), MIB,
                Pred);
  AArch64CC::CondCode CC1, CC2;
  changeFCMPPredToAArch64CC(static_cast<CmpInst::Predicate>(Pred), CC1, CC2);
  MachineBasicBlock *DestMBB = I.getOperand(1).getMBB();
  MIB.buildInstr(AArch64::Bcc, {}, {}).addImm(CC1).addMBB(DestMBB);
  if (CC2 != AArch64CC::AL)
    MIB.buildInstr(AArch64::Bcc, {}, {}).addImm(CC2).addMBB(DestMBB);
  I.eraseFromParent();
  return true;
}

bool AArch64InstructionSelector::tryOptCompareBranchFedByICmp(
    MachineInstr &I, MachineInstr &ICmp, MachineIRBuilder &MIB) const {
  assert(ICmp.getOpcode() == TargetOpcode::G_ICMP);
  assert(I.getOpcode() == TargetOpcode::G_BRCOND);
  // Attempt to optimize the G_BRCOND + G_ICMP into a TB(N)Z/CB(N)Z.
  //
  // Speculation tracking/SLH assumes that optimized TB(N)Z/CB(N)Z
  // instructions will not be produced, as they are conditional branch
  // instructions that do not set flags.
  if (!ProduceNonFlagSettingCondBr)
    return false;

  MachineRegisterInfo &MRI = *MIB.getMRI();
  MachineBasicBlock *DestMBB = I.getOperand(1).getMBB();
  auto Pred =
      static_cast<CmpInst::Predicate>(ICmp.getOperand(1).getPredicate());
  Register LHS = ICmp.getOperand(2).getReg();
  Register RHS = ICmp.getOperand(3).getReg();

  // We're allowed to emit a TB(N)Z/CB(N)Z. Try to do that.
  auto VRegAndVal = getIConstantVRegValWithLookThrough(RHS, MRI);
  MachineInstr *AndInst = getOpcodeDef(TargetOpcode::G_AND, LHS, MRI);

  // When we can emit a TB(N)Z, prefer that.
  //
  // Handle non-commutative condition codes first.
  // Note that we don't want to do this when we have a G_AND because it can
  // become a tst. The tst will make the test bit in the TB(N)Z redundant.
  if (VRegAndVal && !AndInst) {
    int64_t C = VRegAndVal->Value.getSExtValue();

    // When we have a greater-than comparison, we can just test if the msb is
    // zero.
    if (C == -1 && Pred == CmpInst::ICMP_SGT) {
      uint64_t Bit = MRI.getType(LHS).getSizeInBits() - 1;
      emitTestBit(LHS, Bit, /*IsNegative = */ false, DestMBB, MIB);
      I.eraseFromParent();
      return true;
    }

    // When we have a less than comparison, we can just test if the msb is not
    // zero.
    if (C == 0 && Pred == CmpInst::ICMP_SLT) {
      uint64_t Bit = MRI.getType(LHS).getSizeInBits() - 1;
      emitTestBit(LHS, Bit, /*IsNegative = */ true, DestMBB, MIB);
      I.eraseFromParent();
      return true;
    }

    // Inversely, if we have a signed greater-than-or-equal comparison to zero,
    // we can test if the msb is zero.
    if (C == 0 && Pred == CmpInst::ICMP_SGE) {
      uint64_t Bit = MRI.getType(LHS).getSizeInBits() - 1;
      emitTestBit(LHS, Bit, /*IsNegative = */ false, DestMBB, MIB);
      I.eraseFromParent();
      return true;
    }
  }

  // Attempt to handle commutative condition codes. Right now, that's only
  // eq/ne.
  if (ICmpInst::isEquality(Pred)) {
    if (!VRegAndVal) {
      std::swap(RHS, LHS);
      VRegAndVal = getIConstantVRegValWithLookThrough(RHS, MRI);
      AndInst = getOpcodeDef(TargetOpcode::G_AND, LHS, MRI);
    }

    if (VRegAndVal && VRegAndVal->Value == 0) {
      // If there's a G_AND feeding into this branch, try to fold it away by
      // emitting a TB(N)Z instead.
      //
      // Note: If we have LT, then it *is* possible to fold, but it wouldn't be
      // beneficial. When we have an AND and LT, we need a TST/ANDS, so folding
      // would be redundant.
      if (AndInst &&
          tryOptAndIntoCompareBranch(
              *AndInst, /*Invert = */ Pred == CmpInst::ICMP_NE, DestMBB, MIB)) {
        I.eraseFromParent();
        return true;
      }

      // Otherwise, try to emit a CB(N)Z instead.
      auto LHSTy = MRI.getType(LHS);
      if (!LHSTy.isVector() && LHSTy.getSizeInBits() <= 64) {
        emitCBZ(LHS, /*IsNegative = */ Pred == CmpInst::ICMP_NE, DestMBB, MIB);
        I.eraseFromParent();
        return true;
      }
    }
  }

  return false;
}

bool AArch64InstructionSelector::selectCompareBranchFedByICmp(
    MachineInstr &I, MachineInstr &ICmp, MachineIRBuilder &MIB) const {
  assert(ICmp.getOpcode() == TargetOpcode::G_ICMP);
  assert(I.getOpcode() == TargetOpcode::G_BRCOND);
  if (tryOptCompareBranchFedByICmp(I, ICmp, MIB))
    return true;

  // Couldn't optimize. Emit a compare + a Bcc.
  MachineBasicBlock *DestMBB = I.getOperand(1).getMBB();
  auto PredOp = ICmp.getOperand(1);
  emitIntegerCompare(ICmp.getOperand(2), ICmp.getOperand(3), PredOp, MIB);
  const AArch64CC::CondCode CC = changeICMPPredToAArch64CC(
      static_cast<CmpInst::Predicate>(PredOp.getPredicate()));
  MIB.buildInstr(AArch64::Bcc, {}, {}).addImm(CC).addMBB(DestMBB);
  I.eraseFromParent();
  return true;
}

bool AArch64InstructionSelector::selectCompareBranch(
    MachineInstr &I, MachineFunction &MF, MachineRegisterInfo &MRI) {
  Register CondReg = I.getOperand(0).getReg();
  MachineInstr *CCMI = MRI.getVRegDef(CondReg);
  // Try to select the G_BRCOND using whatever is feeding the condition if
  // possible.
  unsigned CCMIOpc = CCMI->getOpcode();
  if (CCMIOpc == TargetOpcode::G_FCMP)
    return selectCompareBranchFedByFCmp(I, *CCMI, MIB);
  if (CCMIOpc == TargetOpcode::G_ICMP)
    return selectCompareBranchFedByICmp(I, *CCMI, MIB);

  // Speculation tracking/SLH assumes that optimized TB(N)Z/CB(N)Z
  // instructions will not be produced, as they are conditional branch
  // instructions that do not set flags.
  if (ProduceNonFlagSettingCondBr) {
    emitTestBit(CondReg, /*Bit = */ 0, /*IsNegative = */ true,
                I.getOperand(1).getMBB(), MIB);
    I.eraseFromParent();
    return true;
  }

  // Can't emit TB(N)Z/CB(N)Z. Emit a tst + bcc instead.
  auto TstMI =
      MIB.buildInstr(AArch64::ANDSWri, {LLT::scalar(32)}, {CondReg}).addImm(1);
  constrainSelectedInstRegOperands(*TstMI, TII, TRI, RBI);
  auto Bcc = MIB.buildInstr(AArch64::Bcc)
                 .addImm(AArch64CC::EQ)
                 .addMBB(I.getOperand(1).getMBB());
  I.eraseFromParent();
  return constrainSelectedInstRegOperands(*Bcc, TII, TRI, RBI);
}

/// Returns the element immediate value of a vector shift operand if found.
/// This needs to detect a splat-like operation, e.g. a G_BUILD_VECTOR.
static std::optional<int64_t> getVectorShiftImm(Register Reg,
                                                MachineRegisterInfo &MRI) {
  assert(MRI.getType(Reg).isVector() && "Expected a *vector* shift operand");
  MachineInstr *OpMI = MRI.getVRegDef(Reg);
  return getAArch64VectorSplatScalar(*OpMI, MRI);
}

/// Matches and returns the shift immediate value for a SHL instruction given
/// a shift operand.
static std::optional<int64_t> getVectorSHLImm(LLT SrcTy, Register Reg,
                                              MachineRegisterInfo &MRI) {
  std::optional<int64_t> ShiftImm = getVectorShiftImm(Reg, MRI);
  if (!ShiftImm)
    return std::nullopt;
  // Check the immediate is in range for a SHL.
  int64_t Imm = *ShiftImm;
  if (Imm < 0)
    return std::nullopt;
  switch (SrcTy.getElementType().getSizeInBits()) {
  default:
    LLVM_DEBUG(dbgs() << "Unhandled element type for vector shift");
    return std::nullopt;
  case 8:
    if (Imm > 7)
      return std::nullopt;
    break;
  case 16:
    if (Imm > 15)
      return std::nullopt;
    break;
  case 32:
    if (Imm > 31)
      return std::nullopt;
    break;
  case 64:
    if (Imm > 63)
      return std::nullopt;
    break;
  }
  return Imm;
}

bool AArch64InstructionSelector::selectVectorSHL(MachineInstr &I,
                                                 MachineRegisterInfo &MRI) {
  assert(I.getOpcode() == TargetOpcode::G_SHL);
  Register DstReg = I.getOperand(0).getReg();
  const LLT Ty = MRI.getType(DstReg);
  Register Src1Reg = I.getOperand(1).getReg();
  Register Src2Reg = I.getOperand(2).getReg();

  if (!Ty.isVector())
    return false;

  // Check if we have a vector of constants on RHS that we can select as the
  // immediate form.
  std::optional<int64_t> ImmVal = getVectorSHLImm(Ty, Src2Reg, MRI);

  unsigned Opc = 0;
  if (Ty == LLT::fixed_vector(2, 64)) {
    Opc = ImmVal ? AArch64::SHLv2i64_shift : AArch64::USHLv2i64;
  } else if (Ty == LLT::fixed_vector(4, 32)) {
    Opc = ImmVal ? AArch64::SHLv4i32_shift : AArch64::USHLv4i32;
  } else if (Ty == LLT::fixed_vector(2, 32)) {
    Opc = ImmVal ? AArch64::SHLv2i32_shift : AArch64::USHLv2i32;
  } else if (Ty == LLT::fixed_vector(4, 16)) {
    Opc = ImmVal ? AArch64::SHLv4i16_shift : AArch64::USHLv4i16;
  } else if (Ty == LLT::fixed_vector(8, 16)) {
    Opc = ImmVal ? AArch64::SHLv8i16_shift : AArch64::USHLv8i16;
  } else if (Ty == LLT::fixed_vector(16, 8)) {
    Opc = ImmVal ? AArch64::SHLv16i8_shift : AArch64::USHLv16i8;
  } else if (Ty == LLT::fixed_vector(8, 8)) {
    Opc = ImmVal ? AArch64::SHLv8i8_shift : AArch64::USHLv8i8;
  } else {
    LLVM_DEBUG(dbgs() << "Unhandled G_SHL type");
    return false;
  }

  auto Shl = MIB.buildInstr(Opc, {DstReg}, {Src1Reg});
  if (ImmVal)
    Shl.addImm(*ImmVal);
  else
    Shl.addUse(Src2Reg);
  constrainSelectedInstRegOperands(*Shl, TII, TRI, RBI);
  I.eraseFromParent();
  return true;
}

bool AArch64InstructionSelector::selectVectorAshrLshr(
    MachineInstr &I, MachineRegisterInfo &MRI) {
  assert(I.getOpcode() == TargetOpcode::G_ASHR ||
         I.getOpcode() == TargetOpcode::G_LSHR);
  Register DstReg = I.getOperand(0).getReg();
  const LLT Ty = MRI.getType(DstReg);
  Register Src1Reg = I.getOperand(1).getReg();
  Register Src2Reg = I.getOperand(2).getReg();

  if (!Ty.isVector())
    return false;

  bool IsASHR = I.getOpcode() == TargetOpcode::G_ASHR;

  // We expect the immediate case to be lowered in the PostLegalCombiner to
  // AArch64ISD::VASHR or AArch64ISD::VLSHR equivalents.

  // There is not a shift right register instruction, but the shift left
  // register instruction takes a signed value, where negative numbers specify a
  // right shift.

  unsigned Opc = 0;
  unsigned NegOpc = 0;
  const TargetRegisterClass *RC =
      getRegClassForTypeOnBank(Ty, RBI.getRegBank(AArch64::FPRRegBankID));
  if (Ty == LLT::fixed_vector(2, 64)) {
    Opc = IsASHR ? AArch64::SSHLv2i64 : AArch64::USHLv2i64;
    NegOpc = AArch64::NEGv2i64;
  } else if (Ty == LLT::fixed_vector(4, 32)) {
    Opc = IsASHR ? AArch64::SSHLv4i32 : AArch64::USHLv4i32;
    NegOpc = AArch64::NEGv4i32;
  } else if (Ty == LLT::fixed_vector(2, 32)) {
    Opc = IsASHR ? AArch64::SSHLv2i32 : AArch64::USHLv2i32;
    NegOpc = AArch64::NEGv2i32;
  } else if (Ty == LLT::fixed_vector(4, 16)) {
    Opc = IsASHR ? AArch64::SSHLv4i16 : AArch64::USHLv4i16;
    NegOpc = AArch64::NEGv4i16;
  } else if (Ty == LLT::fixed_vector(8, 16)) {
    Opc = IsASHR ? AArch64::SSHLv8i16 : AArch64::USHLv8i16;
    NegOpc = AArch64::NEGv8i16;
  } else if (Ty == LLT::fixed_vector(16, 8)) {
    Opc = IsASHR ? AArch64::SSHLv16i8 : AArch64::USHLv16i8;
    NegOpc = AArch64::NEGv16i8;
  } else if (Ty == LLT::fixed_vector(8, 8)) {
    Opc = IsASHR ? AArch64::SSHLv8i8 : AArch64::USHLv8i8;
    NegOpc = AArch64::NEGv8i8;
  } else {
    LLVM_DEBUG(dbgs() << "Unhandled G_ASHR type");
    return false;
  }

  auto Neg = MIB.buildInstr(NegOpc, {RC}, {Src2Reg});
  constrainSelectedInstRegOperands(*Neg, TII, TRI, RBI);
  auto SShl = MIB.buildInstr(Opc, {DstReg}, {Src1Reg, Neg});
  constrainSelectedInstRegOperands(*SShl, TII, TRI, RBI);
  I.eraseFromParent();
  return true;
}

bool AArch64InstructionSelector::selectVaStartAAPCS(
    MachineInstr &I, MachineFunction &MF, MachineRegisterInfo &MRI) const {
  return false;
}

bool AArch64InstructionSelector::selectVaStartDarwin(
    MachineInstr &I, MachineFunction &MF, MachineRegisterInfo &MRI) const {
  AArch64FunctionInfo *FuncInfo = MF.getInfo<AArch64FunctionInfo>();
  Register ListReg = I.getOperand(0).getReg();

  Register ArgsAddrReg = MRI.createVirtualRegister(&AArch64::GPR64RegClass);

  auto MIB =
      BuildMI(*I.getParent(), I, I.getDebugLoc(), TII.get(AArch64::ADDXri))
          .addDef(ArgsAddrReg)
          .addFrameIndex(FuncInfo->getVarArgsStackIndex())
          .addImm(0)
          .addImm(0);

  constrainSelectedInstRegOperands(*MIB, TII, TRI, RBI);

  MIB = BuildMI(*I.getParent(), I, I.getDebugLoc(), TII.get(AArch64::STRXui))
            .addUse(ArgsAddrReg)
            .addUse(ListReg)
            .addImm(0)
            .addMemOperand(*I.memoperands_begin());

  constrainSelectedInstRegOperands(*MIB, TII, TRI, RBI);
  I.eraseFromParent();
  return true;
}

void AArch64InstructionSelector::materializeLargeCMVal(
    MachineInstr &I, const Value *V, unsigned OpFlags) {
  MachineBasicBlock &MBB = *I.getParent();
  MachineFunction &MF = *MBB.getParent();
  MachineRegisterInfo &MRI = MF.getRegInfo();

  auto MovZ = MIB.buildInstr(AArch64::MOVZXi, {&AArch64::GPR64RegClass}, {});
  MovZ->addOperand(MF, I.getOperand(1));
  MovZ->getOperand(1).setTargetFlags(OpFlags | AArch64II::MO_G0 |
                                     AArch64II::MO_NC);
  MovZ->addOperand(MF, MachineOperand::CreateImm(0));
  constrainSelectedInstRegOperands(*MovZ, TII, TRI, RBI);

  auto BuildMovK = [&](Register SrcReg, unsigned char Flags, unsigned Offset,
                       Register ForceDstReg) {
    Register DstReg = ForceDstReg
                          ? ForceDstReg
                          : MRI.createVirtualRegister(&AArch64::GPR64RegClass);
    auto MovI = MIB.buildInstr(AArch64::MOVKXi).addDef(DstReg).addUse(SrcReg);
    if (auto *GV = dyn_cast<GlobalValue>(V)) {
      MovI->addOperand(MF, MachineOperand::CreateGA(
                               GV, MovZ->getOperand(1).getOffset(), Flags));
    } else {
      MovI->addOperand(
          MF, MachineOperand::CreateBA(cast<BlockAddress>(V),
                                       MovZ->getOperand(1).getOffset(), Flags));
    }
    MovI->addOperand(MF, MachineOperand::CreateImm(Offset));
    constrainSelectedInstRegOperands(*MovI, TII, TRI, RBI);
    return DstReg;
  };
  Register DstReg = BuildMovK(MovZ.getReg(0),
                              AArch64II::MO_G1 | AArch64II::MO_NC, 16, 0);
  DstReg = BuildMovK(DstReg, AArch64II::MO_G2 | AArch64II::MO_NC, 32, 0);
  BuildMovK(DstReg, AArch64II::MO_G3, 48, I.getOperand(0).getReg());
}

bool AArch64InstructionSelector::preISelLower(MachineInstr &I) {
  MachineBasicBlock &MBB = *I.getParent();
  MachineFunction &MF = *MBB.getParent();
  MachineRegisterInfo &MRI = MF.getRegInfo();

  switch (I.getOpcode()) {
  case TargetOpcode::G_STORE: {
    bool Changed = contractCrossBankCopyIntoStore(I, MRI);
    MachineOperand &SrcOp = I.getOperand(0);
    if (MRI.getType(SrcOp.getReg()).isPointer()) {
      // Allow matching with imported patterns for stores of pointers. Unlike
      // G_LOAD/G_PTR_ADD, we may not have selected all users. So, emit a copy
      // and constrain.
      auto Copy = MIB.buildCopy(LLT::scalar(64), SrcOp);
      Register NewSrc = Copy.getReg(0);
      SrcOp.setReg(NewSrc);
      RBI.constrainGenericRegister(NewSrc, AArch64::GPR64RegClass, MRI);
      Changed = true;
    }
    return Changed;
  }
  case TargetOpcode::G_PTR_ADD:
    return convertPtrAddToAdd(I, MRI);
  case TargetOpcode::G_LOAD: {
    // For scalar loads of pointers, we try to convert the dest type from p0
    // to s64 so that our imported patterns can match. Like with the G_PTR_ADD
    // conversion, this should be ok because all users should have been
    // selected already, so the type doesn't matter for them.
    Register DstReg = I.getOperand(0).getReg();
    const LLT DstTy = MRI.getType(DstReg);
    if (!DstTy.isPointer())
      return false;
    MRI.setType(DstReg, LLT::scalar(64));
    return true;
  }
  case AArch64::G_DUP: {
    // Convert the type from p0 to s64 to help selection.
    LLT DstTy = MRI.getType(I.getOperand(0).getReg());
    if (!DstTy.getElementType().isPointer())
      return false;
    auto NewSrc = MIB.buildCopy(LLT::scalar(64), I.getOperand(1).getReg());
    MRI.setType(I.getOperand(0).getReg(),
                DstTy.changeElementType(LLT::scalar(64)));
    MRI.setRegClass(NewSrc.getReg(0), &AArch64::GPR64RegClass);
    I.getOperand(1).setReg(NewSrc.getReg(0));
    return true;
  }
  case TargetOpcode::G_UITOFP:
  case TargetOpcode::G_SITOFP: {
    // If both source and destination regbanks are FPR, then convert the opcode
    // to G_SITOF so that the importer can select it to an fpr variant.
    // Otherwise, it ends up matching an fpr/gpr variant and adding a cross-bank
    // copy.
    Register SrcReg = I.getOperand(1).getReg();
    LLT SrcTy = MRI.getType(SrcReg);
    LLT DstTy = MRI.getType(I.getOperand(0).getReg());
    if (SrcTy.isVector() || SrcTy.getSizeInBits() != DstTy.getSizeInBits())
      return false;

    if (RBI.getRegBank(SrcReg, MRI, TRI)->getID() == AArch64::FPRRegBankID) {
      if (I.getOpcode() == TargetOpcode::G_SITOFP)
        I.setDesc(TII.get(AArch64::G_SITOF));
      else
        I.setDesc(TII.get(AArch64::G_UITOF));
      return true;
    }
    return false;
  }
  default:
    return false;
  }
}

/// This lowering tries to look for G_PTR_ADD instructions and then converts
/// them to a standard G_ADD with a COPY on the source.
///
/// The motivation behind this is to expose the add semantics to the imported
/// tablegen patterns. We shouldn't need to check for uses being loads/stores,
/// because the selector works bottom up, uses before defs. By the time we
/// end up trying to select a G_PTR_ADD, we should have already attempted to
/// fold this into addressing modes and were therefore unsuccessful.
bool AArch64InstructionSelector::convertPtrAddToAdd(
    MachineInstr &I, MachineRegisterInfo &MRI) {
  assert(I.getOpcode() == TargetOpcode::G_PTR_ADD && "Expected G_PTR_ADD");
  Register DstReg = I.getOperand(0).getReg();
  Register AddOp1Reg = I.getOperand(1).getReg();
  const LLT PtrTy = MRI.getType(DstReg);
  if (PtrTy.getAddressSpace() != 0)
    return false;

  const LLT CastPtrTy =
      PtrTy.isVector() ? LLT::fixed_vector(2, 64) : LLT::scalar(64);
  auto PtrToInt = MIB.buildPtrToInt(CastPtrTy, AddOp1Reg);
  // Set regbanks on the registers.
  if (PtrTy.isVector())
    MRI.setRegBank(PtrToInt.getReg(0), RBI.getRegBank(AArch64::FPRRegBankID));
  else
    MRI.setRegBank(PtrToInt.getReg(0), RBI.getRegBank(AArch64::GPRRegBankID));

  // Now turn the %dst(p0) = G_PTR_ADD %base, off into:
  // %dst(intty) = G_ADD %intbase, off
  I.setDesc(TII.get(TargetOpcode::G_ADD));
  MRI.setType(DstReg, CastPtrTy);
  I.getOperand(1).setReg(PtrToInt.getReg(0));
  if (!select(*PtrToInt)) {
    LLVM_DEBUG(dbgs() << "Failed to select G_PTRTOINT in convertPtrAddToAdd");
    return false;
  }

  // Also take the opportunity here to try to do some optimization.
  // Try to convert this into a G_SUB if the offset is a 0-x negate idiom.
  Register NegatedReg;
  if (!mi_match(I.getOperand(2).getReg(), MRI, m_Neg(m_Reg(NegatedReg))))
    return true;
  I.getOperand(2).setReg(NegatedReg);
  I.setDesc(TII.get(TargetOpcode::G_SUB));
  return true;
}

bool AArch64InstructionSelector::earlySelectSHL(MachineInstr &I,
                                                MachineRegisterInfo &MRI) {
  // We try to match the immediate variant of LSL, which is actually an alias
  // for a special case of UBFM. Otherwise, we fall back to the imported
  // selector which will match the register variant.
  assert(I.getOpcode() == TargetOpcode::G_SHL && "unexpected op");
  const auto &MO = I.getOperand(2);
  auto VRegAndVal = getIConstantVRegVal(MO.getReg(), MRI);
  if (!VRegAndVal)
    return false;

  const LLT DstTy = MRI.getType(I.getOperand(0).getReg());
  if (DstTy.isVector())
    return false;
  bool Is64Bit = DstTy.getSizeInBits() == 64;
  auto Imm1Fn = Is64Bit ? selectShiftA_64(MO) : selectShiftA_32(MO);
  auto Imm2Fn = Is64Bit ? selectShiftB_64(MO) : selectShiftB_32(MO);

  if (!Imm1Fn || !Imm2Fn)
    return false;

  auto NewI =
      MIB.buildInstr(Is64Bit ? AArch64::UBFMXri : AArch64::UBFMWri,
                     {I.getOperand(0).getReg()}, {I.getOperand(1).getReg()});

  for (auto &RenderFn : *Imm1Fn)
    RenderFn(NewI);
  for (auto &RenderFn : *Imm2Fn)
    RenderFn(NewI);

  I.eraseFromParent();
  return constrainSelectedInstRegOperands(*NewI, TII, TRI, RBI);
}

bool AArch64InstructionSelector::contractCrossBankCopyIntoStore(
    MachineInstr &I, MachineRegisterInfo &MRI) {
  assert(I.getOpcode() == TargetOpcode::G_STORE && "Expected G_STORE");
  // If we're storing a scalar, it doesn't matter what register bank that
  // scalar is on. All that matters is the size.
  //
  // So, if we see something like this (with a 32-bit scalar as an example):
  //
  // %x:gpr(s32) = ... something ...
  // %y:fpr(s32) = COPY %x:gpr(s32)
  // G_STORE %y:fpr(s32)
  //
  // We can fix this up into something like this:
  //
  // G_STORE %x:gpr(s32)
  //
  // And then continue the selection process normally.
  Register DefDstReg = getSrcRegIgnoringCopies(I.getOperand(0).getReg(), MRI);
  if (!DefDstReg.isValid())
    return false;
  LLT DefDstTy = MRI.getType(DefDstReg);
  Register StoreSrcReg = I.getOperand(0).getReg();
  LLT StoreSrcTy = MRI.getType(StoreSrcReg);

  // If we get something strange like a physical register, then we shouldn't
  // go any further.
  if (!DefDstTy.isValid())
    return false;

  // Are the source and dst types the same size?
  if (DefDstTy.getSizeInBits() != StoreSrcTy.getSizeInBits())
    return false;

  if (RBI.getRegBank(StoreSrcReg, MRI, TRI) ==
      RBI.getRegBank(DefDstReg, MRI, TRI))
    return false;

  // We have a cross-bank copy, which is entering a store. Let's fold it.
  I.getOperand(0).setReg(DefDstReg);
  return true;
}

bool AArch64InstructionSelector::earlySelect(MachineInstr &I) {
  assert(I.getParent() && "Instruction should be in a basic block!");
  assert(I.getParent()->getParent() && "Instruction should be in a function!");

  MachineBasicBlock &MBB = *I.getParent();
  MachineFunction &MF = *MBB.getParent();
  MachineRegisterInfo &MRI = MF.getRegInfo();

  switch (I.getOpcode()) {
  case AArch64::G_DUP: {
    // Before selecting a DUP instruction, check if it is better selected as a
    // MOV or load from a constant pool.
    Register Src = I.getOperand(1).getReg();
    auto ValAndVReg = getIConstantVRegValWithLookThrough(Src, MRI);
    if (!ValAndVReg)
      return false;
    LLVMContext &Ctx = MF.getFunction().getContext();
    Register Dst = I.getOperand(0).getReg();
    auto *CV = ConstantDataVector::getSplat(
        MRI.getType(Dst).getNumElements(),
        ConstantInt::get(Type::getIntNTy(Ctx, MRI.getType(Src).getSizeInBits()),
                         ValAndVReg->Value));
    if (!emitConstantVector(Dst, CV, MIB, MRI))
      return false;
    I.eraseFromParent();
    return true;
  }
  case TargetOpcode::G_SEXT:
    // Check for i64 sext(i32 vector_extract) prior to tablegen to select SMOV
    // over a normal extend.
    if (selectUSMovFromExtend(I, MRI))
      return true;
    return false;
  case TargetOpcode::G_BR:
    return false;
  case TargetOpcode::G_SHL:
    return earlySelectSHL(I, MRI);
  case TargetOpcode::G_CONSTANT: {
    bool IsZero = false;
    if (I.getOperand(1).isCImm())
      IsZero = I.getOperand(1).getCImm()->getZExtValue() == 0;
    else if (I.getOperand(1).isImm())
      IsZero = I.getOperand(1).getImm() == 0;

    if (!IsZero)
      return false;

    Register DefReg = I.getOperand(0).getReg();
    LLT Ty = MRI.getType(DefReg);
    if (Ty.getSizeInBits() == 64) {
      I.getOperand(1).ChangeToRegister(AArch64::XZR, false);
      RBI.constrainGenericRegister(DefReg, AArch64::GPR64RegClass, MRI);
    } else if (Ty.getSizeInBits() == 32) {
      I.getOperand(1).ChangeToRegister(AArch64::WZR, false);
      RBI.constrainGenericRegister(DefReg, AArch64::GPR32RegClass, MRI);
    } else
      return false;

    I.setDesc(TII.get(TargetOpcode::COPY));
    return true;
  }

  case TargetOpcode::G_ADD: {
    // Check if this is being fed by a G_ICMP on either side.
    //
    // (cmp pred, x, y) + z
    //
    // In the above case, when the cmp is true, we increment z by 1. So, we can
    // fold the add into the cset for the cmp by using cinc.
    //
    // FIXME: This would probably be a lot nicer in PostLegalizerLowering.
    Register AddDst = I.getOperand(0).getReg();
    Register AddLHS = I.getOperand(1).getReg();
    Register AddRHS = I.getOperand(2).getReg();
    // Only handle scalars.
    LLT Ty = MRI.getType(AddLHS);
    if (Ty.isVector())
      return false;
    // Since G_ICMP is modeled as ADDS/SUBS/ANDS, we can handle 32 bits or 64
    // bits.
    unsigned Size = Ty.getSizeInBits();
    if (Size != 32 && Size != 64)
      return false;
    auto MatchCmp = [&](Register Reg) -> MachineInstr * {
      if (!MRI.hasOneNonDBGUse(Reg))
        return nullptr;
      // If the LHS of the add is 32 bits, then we want to fold a 32-bit
      // compare.
      if (Size == 32)
        return getOpcodeDef(TargetOpcode::G_ICMP, Reg, MRI);
      // We model scalar compares using 32-bit destinations right now.
      // If it's a 64-bit compare, it'll have 64-bit sources.
      Register ZExt;
      if (!mi_match(Reg, MRI,
                    m_OneNonDBGUse(m_GZExt(m_OneNonDBGUse(m_Reg(ZExt))))))
        return nullptr;
      auto *Cmp = getOpcodeDef(TargetOpcode::G_ICMP, ZExt, MRI);
      if (!Cmp ||
          MRI.getType(Cmp->getOperand(2).getReg()).getSizeInBits() != 64)
        return nullptr;
      return Cmp;
    };
    // Try to match
    // z + (cmp pred, x, y)
    MachineInstr *Cmp = MatchCmp(AddRHS);
    if (!Cmp) {
      // (cmp pred, x, y) + z
      std::swap(AddLHS, AddRHS);
      Cmp = MatchCmp(AddRHS);
      if (!Cmp)
        return false;
    }
    auto &PredOp = Cmp->getOperand(1);
    auto Pred = static_cast<CmpInst::Predicate>(PredOp.getPredicate());
    const AArch64CC::CondCode InvCC =
        changeICMPPredToAArch64CC(CmpInst::getInversePredicate(Pred));
    MIB.setInstrAndDebugLoc(I);
    emitIntegerCompare(/*LHS=*/Cmp->getOperand(2),
                       /*RHS=*/Cmp->getOperand(3), PredOp, MIB);
    emitCSINC(/*Dst=*/AddDst, /*Src =*/AddLHS, /*Src2=*/AddLHS, InvCC, MIB);
    I.eraseFromParent();
    return true;
  }
  case TargetOpcode::G_OR: {
    // Look for operations that take the lower `Width=Size-ShiftImm` bits of
    // `ShiftSrc` and insert them into the upper `Width` bits of `MaskSrc` via
    // shifting and masking that we can replace with a BFI (encoded as a BFM).
    Register Dst = I.getOperand(0).getReg();
    LLT Ty = MRI.getType(Dst);

    if (!Ty.isScalar())
      return false;

    unsigned Size = Ty.getSizeInBits();
    if (Size != 32 && Size != 64)
      return false;

    Register ShiftSrc;
    int64_t ShiftImm;
    Register MaskSrc;
    int64_t MaskImm;
    if (!mi_match(
            Dst, MRI,
            m_GOr(m_OneNonDBGUse(m_GShl(m_Reg(ShiftSrc), m_ICst(ShiftImm))),
                  m_OneNonDBGUse(m_GAnd(m_Reg(MaskSrc), m_ICst(MaskImm))))))
      return false;

    if (ShiftImm > Size || ((1ULL << ShiftImm) - 1ULL) != uint64_t(MaskImm))
      return false;

    int64_t Immr = Size - ShiftImm;
    int64_t Imms = Size - ShiftImm - 1;
    unsigned Opc = Size == 32 ? AArch64::BFMWri : AArch64::BFMXri;
    emitInstr(Opc, {Dst}, {MaskSrc, ShiftSrc, Immr, Imms}, MIB);
    I.eraseFromParent();
    return true;
  }
  case TargetOpcode::G_FENCE: {
    if (I.getOperand(1).getImm() == 0)
      BuildMI(MBB, I, MIMetadata(I), TII.get(TargetOpcode::MEMBARRIER));
    else
      BuildMI(MBB, I, MIMetadata(I), TII.get(AArch64::DMB))
          .addImm(I.getOperand(0).getImm() == 4 ? 0x9 : 0xb);
    I.eraseFromParent();
    return true;
  }
  default:
    return false;
  }
}

bool AArch64InstructionSelector::select(MachineInstr &I) {
  assert(I.getParent() && "Instruction should be in a basic block!");
  assert(I.getParent()->getParent() && "Instruction should be in a function!");

  MachineBasicBlock &MBB = *I.getParent();
  MachineFunction &MF = *MBB.getParent();
  MachineRegisterInfo &MRI = MF.getRegInfo();

  const AArch64Subtarget *Subtarget = &MF.getSubtarget<AArch64Subtarget>();
  if (Subtarget->requiresStrictAlign()) {
    // We don't support this feature yet.
    LLVM_DEBUG(dbgs() << "AArch64 GISel does not support strict-align yet\n");
    return false;
  }

  MIB.setInstrAndDebugLoc(I);

  unsigned Opcode = I.getOpcode();
  // G_PHI requires same handling as PHI
  if (!I.isPreISelOpcode() || Opcode == TargetOpcode::G_PHI) {
    // Certain non-generic instructions also need some special handling.

    if (Opcode ==  TargetOpcode::LOAD_STACK_GUARD)
      return constrainSelectedInstRegOperands(I, TII, TRI, RBI);

    if (Opcode == TargetOpcode::PHI || Opcode == TargetOpcode::G_PHI) {
      const Register DefReg = I.getOperand(0).getReg();
      const LLT DefTy = MRI.getType(DefReg);

      const RegClassOrRegBank &RegClassOrBank =
        MRI.getRegClassOrRegBank(DefReg);

      const TargetRegisterClass *DefRC
        = RegClassOrBank.dyn_cast<const TargetRegisterClass *>();
      if (!DefRC) {
        if (!DefTy.isValid()) {
          LLVM_DEBUG(dbgs() << "PHI operand has no type, not a gvreg?\n");
          return false;
        }
        const RegisterBank &RB = *RegClassOrBank.get<const RegisterBank *>();
        DefRC = getRegClassForTypeOnBank(DefTy, RB);
        if (!DefRC) {
          LLVM_DEBUG(dbgs() << "PHI operand has unexpected size/bank\n");
          return false;
        }
      }

      I.setDesc(TII.get(TargetOpcode::PHI));

      return RBI.constrainGenericRegister(DefReg, *DefRC, MRI);
    }

    if (I.isCopy())
      return selectCopy(I, TII, MRI, TRI, RBI);

    if (I.isDebugInstr())
      return selectDebugInstr(I, MRI, RBI);

    return true;
  }


  if (I.getNumOperands() != I.getNumExplicitOperands()) {
    LLVM_DEBUG(
        dbgs() << "Generic instruction has unexpected implicit operands\n");
    return false;
  }

  // Try to do some lowering before we start instruction selecting. These
  // lowerings are purely transformations on the input G_MIR and so selection
  // must continue after any modification of the instruction.
  if (preISelLower(I)) {
    Opcode = I.getOpcode(); // The opcode may have been modified, refresh it.
  }

  // There may be patterns where the importer can't deal with them optimally,
  // but does select it to a suboptimal sequence so our custom C++ selection
  // code later never has a chance to work on it. Therefore, we have an early
  // selection attempt here to give priority to certain selection routines
  // over the imported ones.
  if (earlySelect(I))
    return true;

  if (selectImpl(I, *CoverageInfo))
    return true;

  LLT Ty =
      I.getOperand(0).isReg() ? MRI.getType(I.getOperand(0).getReg()) : LLT{};

  switch (Opcode) {
  case TargetOpcode::G_SBFX:
  case TargetOpcode::G_UBFX: {
    static const unsigned OpcTable[2][2] = {
        {AArch64::UBFMWri, AArch64::UBFMXri},
        {AArch64::SBFMWri, AArch64::SBFMXri}};
    bool IsSigned = Opcode == TargetOpcode::G_SBFX;
    unsigned Size = Ty.getSizeInBits();
    unsigned Opc = OpcTable[IsSigned][Size == 64];
    auto Cst1 =
        getIConstantVRegValWithLookThrough(I.getOperand(2).getReg(), MRI);
    assert(Cst1 && "Should have gotten a constant for src 1?");
    auto Cst2 =
        getIConstantVRegValWithLookThrough(I.getOperand(3).getReg(), MRI);
    assert(Cst2 && "Should have gotten a constant for src 2?");
    auto LSB = Cst1->Value.getZExtValue();
    auto Width = Cst2->Value.getZExtValue();
    auto BitfieldInst =
        MIB.buildInstr(Opc, {I.getOperand(0)}, {I.getOperand(1)})
            .addImm(LSB)
            .addImm(LSB + Width - 1);
    I.eraseFromParent();
    return constrainSelectedInstRegOperands(*BitfieldInst, TII, TRI, RBI);
  }
  case TargetOpcode::G_BRCOND:
    return selectCompareBranch(I, MF, MRI);

  case TargetOpcode::G_BRINDIRECT: {
    I.setDesc(TII.get(AArch64::BR));
    return constrainSelectedInstRegOperands(I, TII, TRI, RBI);
  }

  case TargetOpcode::G_BRJT:
    return selectBrJT(I, MRI);

  case AArch64::G_ADD_LOW: {
    // This op may have been separated from it's ADRP companion by the localizer
    // or some other code motion pass. Given that many CPUs will try to
    // macro fuse these operations anyway, select this into a MOVaddr pseudo
    // which will later be expanded into an ADRP+ADD pair after scheduling.
    MachineInstr *BaseMI = MRI.getVRegDef(I.getOperand(1).getReg());
    if (BaseMI->getOpcode() != AArch64::ADRP) {
      I.setDesc(TII.get(AArch64::ADDXri));
      I.addOperand(MachineOperand::CreateImm(0));
      return constrainSelectedInstRegOperands(I, TII, TRI, RBI);
    }
    assert(TM.getCodeModel() == CodeModel::Small &&
           "Expected small code model");
    auto Op1 = BaseMI->getOperand(1);
    auto Op2 = I.getOperand(2);
    auto MovAddr = MIB.buildInstr(AArch64::MOVaddr, {I.getOperand(0)}, {})
                       .addGlobalAddress(Op1.getGlobal(), Op1.getOffset(),
                                         Op1.getTargetFlags())
                       .addGlobalAddress(Op2.getGlobal(), Op2.getOffset(),
                                         Op2.getTargetFlags());
    I.eraseFromParent();
    return constrainSelectedInstRegOperands(*MovAddr, TII, TRI, RBI);
  }

  case TargetOpcode::G_BSWAP: {
    // Handle vector types for G_BSWAP directly.
    Register DstReg = I.getOperand(0).getReg();
    LLT DstTy = MRI.getType(DstReg);

    // We should only get vector types here; everything else is handled by the
    // importer right now.
    if (!DstTy.isVector() || DstTy.getSizeInBits() > 128) {
      LLVM_DEBUG(dbgs() << "Dst type for G_BSWAP currently unsupported.\n");
      return false;
    }

    // Only handle 4 and 2 element vectors for now.
    // TODO: 16-bit elements.
    unsigned NumElts = DstTy.getNumElements();
    if (NumElts != 4 && NumElts != 2) {
      LLVM_DEBUG(dbgs() << "Unsupported number of elements for G_BSWAP.\n");
      return false;
    }

    // Choose the correct opcode for the supported types. Right now, that's
    // v2s32, v4s32, and v2s64.
    unsigned Opc = 0;
    unsigned EltSize = DstTy.getElementType().getSizeInBits();
    if (EltSize == 32)
      Opc = (DstTy.getNumElements() == 2) ? AArch64::REV32v8i8
                                          : AArch64::REV32v16i8;
    else if (EltSize == 64)
      Opc = AArch64::REV64v16i8;

    // We should always get something by the time we get here...
    assert(Opc != 0 && "Didn't get an opcode for G_BSWAP?");

    I.setDesc(TII.get(Opc));
    return constrainSelectedInstRegOperands(I, TII, TRI, RBI);
  }

  case TargetOpcode::G_FCONSTANT:
  case TargetOpcode::G_CONSTANT: {
    const bool isFP = Opcode == TargetOpcode::G_FCONSTANT;

    const LLT s8 = LLT::scalar(8);
    const LLT s16 = LLT::scalar(16);
    const LLT s32 = LLT::scalar(32);
    const LLT s64 = LLT::scalar(64);
    const LLT s128 = LLT::scalar(128);
    const LLT p0 = LLT::pointer(0, 64);

    const Register DefReg = I.getOperand(0).getReg();
    const LLT DefTy = MRI.getType(DefReg);
    const unsigned DefSize = DefTy.getSizeInBits();
    const RegisterBank &RB = *RBI.getRegBank(DefReg, MRI, TRI);

    // FIXME: Redundant check, but even less readable when factored out.
    if (isFP) {
      if (Ty != s16 && Ty != s32 && Ty != s64 && Ty != s128) {
        LLVM_DEBUG(dbgs() << "Unable to materialize FP " << Ty
                          << " constant, expected: " << s16 << " or " << s32
                          << " or " << s64 << " or " << s128 << '\n');
        return false;
      }

      if (RB.getID() != AArch64::FPRRegBankID) {
        LLVM_DEBUG(dbgs() << "Unable to materialize FP " << Ty
                          << " constant on bank: " << RB
                          << ", expected: FPR\n");
        return false;
      }

      // The case when we have 0.0 is covered by tablegen. Reject it here so we
      // can be sure tablegen works correctly and isn't rescued by this code.
      // 0.0 is not covered by tablegen for FP128. So we will handle this
      // scenario in the code here.
      if (DefSize != 128 && I.getOperand(1).getFPImm()->isExactlyValue(0.0))
        return false;
    } else {
      // s32 and s64 are covered by tablegen.
      if (Ty != p0 && Ty != s8 && Ty != s16) {
        LLVM_DEBUG(dbgs() << "Unable to materialize integer " << Ty
                          << " constant, expected: " << s32 << ", " << s64
                          << ", or " << p0 << '\n');
        return false;
      }

      if (RB.getID() != AArch64::GPRRegBankID) {
        LLVM_DEBUG(dbgs() << "Unable to materialize integer " << Ty
                          << " constant on bank: " << RB
                          << ", expected: GPR\n");
        return false;
      }
    }

    if (isFP) {
      const TargetRegisterClass &FPRRC = *getRegClassForTypeOnBank(DefTy, RB);
      // For 16, 64, and 128b values, emit a constant pool load.
      switch (DefSize) {
      default:
        llvm_unreachable("Unexpected destination size for G_FCONSTANT?");
      case 32:
        // For s32, use a cp load if we have optsize/minsize.
        if (!shouldOptForSize(&MF))
          break;
        [[fallthrough]];
      case 16:
      case 64:
      case 128: {
        auto *FPImm = I.getOperand(1).getFPImm();
        auto *LoadMI = emitLoadFromConstantPool(FPImm, MIB);
        if (!LoadMI) {
          LLVM_DEBUG(dbgs() << "Failed to load double constant pool entry\n");
          return false;
        }
        MIB.buildCopy({DefReg}, {LoadMI->getOperand(0).getReg()});
        I.eraseFromParent();
        return RBI.constrainGenericRegister(DefReg, FPRRC, MRI);
      }
      }

      // Either emit a FMOV, or emit a copy to emit a normal mov.
      assert(DefSize == 32 &&
             "Expected constant pool loads for all sizes other than 32!");
      const Register DefGPRReg =
          MRI.createVirtualRegister(&AArch64::GPR32RegClass);
      MachineOperand &RegOp = I.getOperand(0);
      RegOp.setReg(DefGPRReg);
      MIB.setInsertPt(MIB.getMBB(), std::next(I.getIterator()));
      MIB.buildCopy({DefReg}, {DefGPRReg});

      if (!RBI.constrainGenericRegister(DefReg, FPRRC, MRI)) {
        LLVM_DEBUG(dbgs() << "Failed to constrain G_FCONSTANT def operand\n");
        return false;
      }

      MachineOperand &ImmOp = I.getOperand(1);
      // FIXME: Is going through int64_t always correct?
      ImmOp.ChangeToImmediate(
          ImmOp.getFPImm()->getValueAPF().bitcastToAPInt().getZExtValue());
    } else if (I.getOperand(1).isCImm()) {
      uint64_t Val = I.getOperand(1).getCImm()->getZExtValue();
      I.getOperand(1).ChangeToImmediate(Val);
    } else if (I.getOperand(1).isImm()) {
      uint64_t Val = I.getOperand(1).getImm();
      I.getOperand(1).ChangeToImmediate(Val);
    }

    const unsigned MovOpc =
        DefSize == 64 ? AArch64::MOVi64imm : AArch64::MOVi32imm;
    I.setDesc(TII.get(MovOpc));
    constrainSelectedInstRegOperands(I, TII, TRI, RBI);
    return true;
  }
  case TargetOpcode::G_EXTRACT: {
    Register DstReg = I.getOperand(0).getReg();
    Register SrcReg = I.getOperand(1).getReg();
    LLT SrcTy = MRI.getType(SrcReg);
    LLT DstTy = MRI.getType(DstReg);
    (void)DstTy;
    unsigned SrcSize = SrcTy.getSizeInBits();

    if (SrcTy.getSizeInBits() > 64) {
      // This should be an extract of an s128, which is like a vector extract.
      if (SrcTy.getSizeInBits() != 128)
        return false;
      // Only support extracting 64 bits from an s128 at the moment.
      if (DstTy.getSizeInBits() != 64)
        return false;

      unsigned Offset = I.getOperand(2).getImm();
      if (Offset % 64 != 0)
        return false;

      // Check we have the right regbank always.
      const RegisterBank &SrcRB = *RBI.getRegBank(SrcReg, MRI, TRI);
      const RegisterBank &DstRB = *RBI.getRegBank(DstReg, MRI, TRI);
      assert(SrcRB.getID() == DstRB.getID() && "Wrong extract regbank!");

      if (SrcRB.getID() == AArch64::GPRRegBankID) {
        auto NewI =
            MIB.buildInstr(TargetOpcode::COPY, {DstReg}, {})
                .addUse(SrcReg, 0,
                        Offset == 0 ? AArch64::sube64 : AArch64::subo64);
        constrainOperandRegClass(MF, TRI, MRI, TII, RBI, *NewI,
                                 AArch64::GPR64RegClass, NewI->getOperand(0));
        I.eraseFromParent();
        return true;
      }

      // Emit the same code as a vector extract.
      // Offset must be a multiple of 64.
      unsigned LaneIdx = Offset / 64;
      MachineInstr *Extract = emitExtractVectorElt(
          DstReg, DstRB, LLT::scalar(64), SrcReg, LaneIdx, MIB);
      if (!Extract)
        return false;
      I.eraseFromParent();
      return true;
    }

    I.setDesc(TII.get(SrcSize == 64 ? AArch64::UBFMXri : AArch64::UBFMWri));
    MachineInstrBuilder(MF, I).addImm(I.getOperand(2).getImm() +
                                      Ty.getSizeInBits() - 1);

    if (SrcSize < 64) {
      assert(SrcSize == 32 && DstTy.getSizeInBits() == 16 &&
             "unexpected G_EXTRACT types");
      return constrainSelectedInstRegOperands(I, TII, TRI, RBI);
    }

    DstReg = MRI.createGenericVirtualRegister(LLT::scalar(64));
    MIB.setInsertPt(MIB.getMBB(), std::next(I.getIterator()));
    MIB.buildInstr(TargetOpcode::COPY, {I.getOperand(0).getReg()}, {})
        .addReg(DstReg, 0, AArch64::sub_32);
    RBI.constrainGenericRegister(I.getOperand(0).getReg(),
                                 AArch64::GPR32RegClass, MRI);
    I.getOperand(0).setReg(DstReg);

    return constrainSelectedInstRegOperands(I, TII, TRI, RBI);
  }

  case TargetOpcode::G_INSERT: {
    LLT SrcTy = MRI.getType(I.getOperand(2).getReg());
    LLT DstTy = MRI.getType(I.getOperand(0).getReg());
    unsigned DstSize = DstTy.getSizeInBits();
    // Larger inserts are vectors, same-size ones should be something else by
    // now (split up or turned into COPYs).
    if (Ty.getSizeInBits() > 64 || SrcTy.getSizeInBits() > 32)
      return false;

    I.setDesc(TII.get(DstSize == 64 ? AArch64::BFMXri : AArch64::BFMWri));
    unsigned LSB = I.getOperand(3).getImm();
    unsigned Width = MRI.getType(I.getOperand(2).getReg()).getSizeInBits();
    I.getOperand(3).setImm((DstSize - LSB) % DstSize);
    MachineInstrBuilder(MF, I).addImm(Width - 1);

    if (DstSize < 64) {
      assert(DstSize == 32 && SrcTy.getSizeInBits() == 16 &&
             "unexpected G_INSERT types");
      return constrainSelectedInstRegOperands(I, TII, TRI, RBI);
    }

    Register SrcReg = MRI.createGenericVirtualRegister(LLT::scalar(64));
    BuildMI(MBB, I.getIterator(), I.getDebugLoc(),
            TII.get(AArch64::SUBREG_TO_REG))
        .addDef(SrcReg)
        .addImm(0)
        .addUse(I.getOperand(2).getReg())
        .addImm(AArch64::sub_32);
    RBI.constrainGenericRegister(I.getOperand(2).getReg(),
                                 AArch64::GPR32RegClass, MRI);
    I.getOperand(2).setReg(SrcReg);

    return constrainSelectedInstRegOperands(I, TII, TRI, RBI);
  }
  case TargetOpcode::G_FRAME_INDEX: {
    // allocas and G_FRAME_INDEX are only supported in addrspace(0).
    if (Ty != LLT::pointer(0, 64)) {
      LLVM_DEBUG(dbgs() << "G_FRAME_INDEX pointer has type: " << Ty
                        << ", expected: " << LLT::pointer(0, 64) << '\n');
      return false;
    }
    I.setDesc(TII.get(AArch64::ADDXri));

    // MOs for a #0 shifted immediate.
    I.addOperand(MachineOperand::CreateImm(0));
    I.addOperand(MachineOperand::CreateImm(0));

    return constrainSelectedInstRegOperands(I, TII, TRI, RBI);
  }

  case TargetOpcode::G_GLOBAL_VALUE: {
    auto GV = I.getOperand(1).getGlobal();
    if (GV->isThreadLocal())
      return selectTLSGlobalValue(I, MRI);

    unsigned OpFlags = STI.ClassifyGlobalReference(GV, TM);
    if (OpFlags & AArch64II::MO_GOT) {
      I.setDesc(TII.get(AArch64::LOADgot));
      I.getOperand(1).setTargetFlags(OpFlags);
    } else if (TM.getCodeModel() == CodeModel::Large) {
      // Materialize the global using movz/movk instructions.
      materializeLargeCMVal(I, GV, OpFlags);
      I.eraseFromParent();
      return true;
    } else if (TM.getCodeModel() == CodeModel::Tiny) {
      I.setDesc(TII.get(AArch64::ADR));
      I.getOperand(1).setTargetFlags(OpFlags);
    } else {
      I.setDesc(TII.get(AArch64::MOVaddr));
      I.getOperand(1).setTargetFlags(OpFlags | AArch64II::MO_PAGE);
      MachineInstrBuilder MIB(MF, I);
      MIB.addGlobalAddress(GV, I.getOperand(1).getOffset(),
                           OpFlags | AArch64II::MO_PAGEOFF | AArch64II::MO_NC);
    }
    return constrainSelectedInstRegOperands(I, TII, TRI, RBI);
  }

  case TargetOpcode::G_ZEXTLOAD:
  case TargetOpcode::G_LOAD:
  case TargetOpcode::G_STORE: {
    GLoadStore &LdSt = cast<GLoadStore>(I);
    bool IsZExtLoad = I.getOpcode() == TargetOpcode::G_ZEXTLOAD;
    LLT PtrTy = MRI.getType(LdSt.getPointerReg());

    if (PtrTy != LLT::pointer(0, 64)) {
      LLVM_DEBUG(dbgs() << "Load/Store pointer has type: " << PtrTy
                        << ", expected: " << LLT::pointer(0, 64) << '\n');
      return false;
    }

    uint64_t MemSizeInBytes = LdSt.getMemSize();
    unsigned MemSizeInBits = LdSt.getMemSizeInBits();
    AtomicOrdering Order = LdSt.getMMO().getSuccessOrdering();

    // Need special instructions for atomics that affect ordering.
    if (Order != AtomicOrdering::NotAtomic &&
        Order != AtomicOrdering::Unordered &&
        Order != AtomicOrdering::Monotonic) {
      assert(!isa<GZExtLoad>(LdSt));
      if (MemSizeInBytes > 64)
        return false;

      if (isa<GLoad>(LdSt)) {
        static constexpr unsigned LDAPROpcodes[] = {
            AArch64::LDAPRB, AArch64::LDAPRH, AArch64::LDAPRW, AArch64::LDAPRX};
        static constexpr unsigned LDAROpcodes[] = {
            AArch64::LDARB, AArch64::LDARH, AArch64::LDARW, AArch64::LDARX};
        ArrayRef<unsigned> Opcodes =
            STI.hasRCPC() && Order != AtomicOrdering::SequentiallyConsistent
                ? LDAPROpcodes
                : LDAROpcodes;
        I.setDesc(TII.get(Opcodes[Log2_32(MemSizeInBytes)]));
      } else {
        static constexpr unsigned Opcodes[] = {AArch64::STLRB, AArch64::STLRH,
                                               AArch64::STLRW, AArch64::STLRX};
        Register ValReg = LdSt.getReg(0);
        if (MRI.getType(ValReg).getSizeInBits() == 64 && MemSizeInBits != 64) {
          // Emit a subreg copy of 32 bits.
          Register NewVal = MRI.createVirtualRegister(&AArch64::GPR32RegClass);
          MIB.buildInstr(TargetOpcode::COPY, {NewVal}, {})
              .addReg(I.getOperand(0).getReg(), 0, AArch64::sub_32);
          I.getOperand(0).setReg(NewVal);
        }
        I.setDesc(TII.get(Opcodes[Log2_32(MemSizeInBytes)]));
      }
      constrainSelectedInstRegOperands(I, TII, TRI, RBI);
      return true;
    }

#ifndef NDEBUG
    const Register PtrReg = LdSt.getPointerReg();
    const RegisterBank &PtrRB = *RBI.getRegBank(PtrReg, MRI, TRI);
    // Check that the pointer register is valid.
    assert(PtrRB.getID() == AArch64::GPRRegBankID &&
           "Load/Store pointer operand isn't a GPR");
    assert(MRI.getType(PtrReg).isPointer() &&
           "Load/Store pointer operand isn't a pointer");
#endif

    const Register ValReg = LdSt.getReg(0);
    const LLT ValTy = MRI.getType(ValReg);
    const RegisterBank &RB = *RBI.getRegBank(ValReg, MRI, TRI);

    // The code below doesn't support truncating stores, so we need to split it
    // again.
    if (isa<GStore>(LdSt) && ValTy.getSizeInBits() > MemSizeInBits) {
      unsigned SubReg;
      LLT MemTy = LdSt.getMMO().getMemoryType();
      auto *RC = getRegClassForTypeOnBank(MemTy, RB);
      if (!getSubRegForClass(RC, TRI, SubReg))
        return false;

      // Generate a subreg copy.
      auto Copy = MIB.buildInstr(TargetOpcode::COPY, {MemTy}, {})
                      .addReg(ValReg, 0, SubReg)
                      .getReg(0);
      RBI.constrainGenericRegister(Copy, *RC, MRI);
      LdSt.getOperand(0).setReg(Copy);
    } else if (isa<GLoad>(LdSt) && ValTy.getSizeInBits() > MemSizeInBits) {
      // If this is an any-extending load from the FPR bank, split it into a regular
      // load + extend.
      if (RB.getID() == AArch64::FPRRegBankID) {
        unsigned SubReg;
        LLT MemTy = LdSt.getMMO().getMemoryType();
        auto *RC = getRegClassForTypeOnBank(MemTy, RB);
        if (!getSubRegForClass(RC, TRI, SubReg))
          return false;
        Register OldDst = LdSt.getReg(0);
        Register NewDst =
            MRI.createGenericVirtualRegister(LdSt.getMMO().getMemoryType());
        LdSt.getOperand(0).setReg(NewDst);
        MRI.setRegBank(NewDst, RB);
        // Generate a SUBREG_TO_REG to extend it.
        MIB.setInsertPt(MIB.getMBB(), std::next(LdSt.getIterator()));
        MIB.buildInstr(AArch64::SUBREG_TO_REG, {OldDst}, {})
            .addImm(0)
            .addUse(NewDst)
            .addImm(SubReg);
        auto SubRegRC = getRegClassForTypeOnBank(MRI.getType(OldDst), RB);
        RBI.constrainGenericRegister(OldDst, *SubRegRC, MRI);
        MIB.setInstr(LdSt);
      }
    }

    // Helper lambda for partially selecting I. Either returns the original
    // instruction with an updated opcode, or a new instruction.
    auto SelectLoadStoreAddressingMode = [&]() -> MachineInstr * {
      bool IsStore = isa<GStore>(I);
      const unsigned NewOpc =
          selectLoadStoreUIOp(I.getOpcode(), RB.getID(), MemSizeInBits);
      if (NewOpc == I.getOpcode())
        return nullptr;
      // Check if we can fold anything into the addressing mode.
      auto AddrModeFns =
          selectAddrModeIndexed(I.getOperand(1), MemSizeInBytes);
      if (!AddrModeFns) {
        // Can't fold anything. Use the original instruction.
        I.setDesc(TII.get(NewOpc));
        I.addOperand(MachineOperand::CreateImm(0));
        return &I;
      }

      // Folded something. Create a new instruction and return it.
      auto NewInst = MIB.buildInstr(NewOpc, {}, {}, I.getFlags());
      Register CurValReg = I.getOperand(0).getReg();
      IsStore ? NewInst.addUse(CurValReg) : NewInst.addDef(CurValReg);
      NewInst.cloneMemRefs(I);
      for (auto &Fn : *AddrModeFns)
        Fn(NewInst);
      I.eraseFromParent();
      return &*NewInst;
    };

    MachineInstr *LoadStore = SelectLoadStoreAddressingMode();
    if (!LoadStore)
      return false;

    // If we're storing a 0, use WZR/XZR.
    if (Opcode == TargetOpcode::G_STORE) {
      auto CVal = getIConstantVRegValWithLookThrough(
          LoadStore->getOperand(0).getReg(), MRI);
      if (CVal && CVal->Value == 0) {
        switch (LoadStore->getOpcode()) {
        case AArch64::STRWui:
        case AArch64::STRHHui:
        case AArch64::STRBBui:
          LoadStore->getOperand(0).setReg(AArch64::WZR);
          break;
        case AArch64::STRXui:
          LoadStore->getOperand(0).setReg(AArch64::XZR);
          break;
        }
      }
    }

    if (IsZExtLoad) {
      // The zextload from a smaller type to i32 should be handled by the
      // importer.
      if (MRI.getType(LoadStore->getOperand(0).getReg()).getSizeInBits() != 64)
        return false;
      // If we have a ZEXTLOAD then change the load's type to be a narrower reg
      // and zero_extend with SUBREG_TO_REG.
      Register LdReg = MRI.createVirtualRegister(&AArch64::GPR32RegClass);
      Register DstReg = LoadStore->getOperand(0).getReg();
      LoadStore->getOperand(0).setReg(LdReg);

      MIB.setInsertPt(MIB.getMBB(), std::next(LoadStore->getIterator()));
      MIB.buildInstr(AArch64::SUBREG_TO_REG, {DstReg}, {})
          .addImm(0)
          .addUse(LdReg)
          .addImm(AArch64::sub_32);
      constrainSelectedInstRegOperands(*LoadStore, TII, TRI, RBI);
      return RBI.constrainGenericRegister(DstReg, AArch64::GPR64allRegClass,
                                          MRI);
    }
    return constrainSelectedInstRegOperands(*LoadStore, TII, TRI, RBI);
  }

  case TargetOpcode::G_SMULH:
  case TargetOpcode::G_UMULH: {
    // Reject the various things we don't support yet.
    if (unsupportedBinOp(I, RBI, MRI, TRI))
      return false;

    const Register DefReg = I.getOperand(0).getReg();
    const RegisterBank &RB = *RBI.getRegBank(DefReg, MRI, TRI);

    if (RB.getID() != AArch64::GPRRegBankID) {
      LLVM_DEBUG(dbgs() << "G_[SU]MULH on bank: " << RB << ", expected: GPR\n");
      return false;
    }

    if (Ty != LLT::scalar(64)) {
      LLVM_DEBUG(dbgs() << "G_[SU]MULH has type: " << Ty
                        << ", expected: " << LLT::scalar(64) << '\n');
      return false;
    }

    unsigned NewOpc = I.getOpcode() == TargetOpcode::G_SMULH ? AArch64::SMULHrr
                                                             : AArch64::UMULHrr;
    I.setDesc(TII.get(NewOpc));

    // Now that we selected an opcode, we need to constrain the register
    // operands to use appropriate classes.
    return constrainSelectedInstRegOperands(I, TII, TRI, RBI);
  }
  case TargetOpcode::G_LSHR:
  case TargetOpcode::G_ASHR:
    if (MRI.getType(I.getOperand(0).getReg()).isVector())
      return selectVectorAshrLshr(I, MRI);
    [[fallthrough]];
  case TargetOpcode::G_SHL:
    if (Opcode == TargetOpcode::G_SHL &&
        MRI.getType(I.getOperand(0).getReg()).isVector())
      return selectVectorSHL(I, MRI);

    // These shifts were legalized to have 64 bit shift amounts because we
    // want to take advantage of the selection patterns that assume the
    // immediates are s64s, however, selectBinaryOp will assume both operands
    // will have the same bit size.
    {
      Register SrcReg = I.getOperand(1).getReg();
      Register ShiftReg = I.getOperand(2).getReg();
      const LLT ShiftTy = MRI.getType(ShiftReg);
      const LLT SrcTy = MRI.getType(SrcReg);
      if (!SrcTy.isVector() && SrcTy.getSizeInBits() == 32 &&
          ShiftTy.getSizeInBits() == 64) {
        assert(!ShiftTy.isVector() && "unexpected vector shift ty");
        // Insert a subregister copy to implement a 64->32 trunc
        auto Trunc = MIB.buildInstr(TargetOpcode::COPY, {SrcTy}, {})
                         .addReg(ShiftReg, 0, AArch64::sub_32);
        MRI.setRegBank(Trunc.getReg(0), RBI.getRegBank(AArch64::GPRRegBankID));
        I.getOperand(2).setReg(Trunc.getReg(0));
      }
    }
    [[fallthrough]];
  case TargetOpcode::G_OR: {
    // Reject the various things we don't support yet.
    if (unsupportedBinOp(I, RBI, MRI, TRI))
      return false;

    const unsigned OpSize = Ty.getSizeInBits();

    const Register DefReg = I.getOperand(0).getReg();
    const RegisterBank &RB = *RBI.getRegBank(DefReg, MRI, TRI);

    const unsigned NewOpc = selectBinaryOp(I.getOpcode(), RB.getID(), OpSize);
    if (NewOpc == I.getOpcode())
      return false;

    I.setDesc(TII.get(NewOpc));
    // FIXME: Should the type be always reset in setDesc?

    // Now that we selected an opcode, we need to constrain the register
    // operands to use appropriate classes.
    return constrainSelectedInstRegOperands(I, TII, TRI, RBI);
  }

  case TargetOpcode::G_PTR_ADD: {
    emitADD(I.getOperand(0).getReg(), I.getOperand(1), I.getOperand(2), MIB);
    I.eraseFromParent();
    return true;
  }
  case TargetOpcode::G_SADDO:
  case TargetOpcode::G_UADDO:
  case TargetOpcode::G_SSUBO:
  case TargetOpcode::G_USUBO: {
    // Emit the operation and get the correct condition code.
    auto OpAndCC = emitOverflowOp(Opcode, I.getOperand(0).getReg(),
                                  I.getOperand(2), I.getOperand(3), MIB);

    // Now, put the overflow result in the register given by the first operand
    // to the overflow op. CSINC increments the result when the predicate is
    // false, so to get the increment when it's true, we need to use the
    // inverse. In this case, we want to increment when carry is set.
    Register ZReg = AArch64::WZR;
    emitCSINC(/*Dst=*/I.getOperand(1).getReg(), /*Src1=*/ZReg, /*Src2=*/ZReg,
              getInvertedCondCode(OpAndCC.second), MIB);
    I.eraseFromParent();
    return true;
  }

  case TargetOpcode::G_PTRMASK: {
    Register MaskReg = I.getOperand(2).getReg();
    std::optional<int64_t> MaskVal = getIConstantVRegSExtVal(MaskReg, MRI);
    // TODO: Implement arbitrary cases
    if (!MaskVal || !isShiftedMask_64(*MaskVal))
      return false;

    uint64_t Mask = *MaskVal;
    I.setDesc(TII.get(AArch64::ANDXri));
    I.getOperand(2).ChangeToImmediate(
        AArch64_AM::encodeLogicalImmediate(Mask, 64));

    return constrainSelectedInstRegOperands(I, TII, TRI, RBI);
  }
  case TargetOpcode::G_PTRTOINT:
  case TargetOpcode::G_TRUNC: {
    const LLT DstTy = MRI.getType(I.getOperand(0).getReg());
    const LLT SrcTy = MRI.getType(I.getOperand(1).getReg());

    const Register DstReg = I.getOperand(0).getReg();
    const Register SrcReg = I.getOperand(1).getReg();

    const RegisterBank &DstRB = *RBI.getRegBank(DstReg, MRI, TRI);
    const RegisterBank &SrcRB = *RBI.getRegBank(SrcReg, MRI, TRI);

    if (DstRB.getID() != SrcRB.getID()) {
      LLVM_DEBUG(
          dbgs() << "G_TRUNC/G_PTRTOINT input/output on different banks\n");
      return false;
    }

    if (DstRB.getID() == AArch64::GPRRegBankID) {
      const TargetRegisterClass *DstRC = getRegClassForTypeOnBank(DstTy, DstRB);
      if (!DstRC)
        return false;

      const TargetRegisterClass *SrcRC = getRegClassForTypeOnBank(SrcTy, SrcRB);
      if (!SrcRC)
        return false;

      if (!RBI.constrainGenericRegister(SrcReg, *SrcRC, MRI) ||
          !RBI.constrainGenericRegister(DstReg, *DstRC, MRI)) {
        LLVM_DEBUG(dbgs() << "Failed to constrain G_TRUNC/G_PTRTOINT\n");
        return false;
      }

      if (DstRC == SrcRC) {
        // Nothing to be done
      } else if (Opcode == TargetOpcode::G_TRUNC && DstTy == LLT::scalar(32) &&
                 SrcTy == LLT::scalar(64)) {
        llvm_unreachable("TableGen can import this case");
        return false;
      } else if (DstRC == &AArch64::GPR32RegClass &&
                 SrcRC == &AArch64::GPR64RegClass) {
        I.getOperand(1).setSubReg(AArch64::sub_32);
      } else {
        LLVM_DEBUG(
            dbgs() << "Unhandled mismatched classes in G_TRUNC/G_PTRTOINT\n");
        return false;
      }

      I.setDesc(TII.get(TargetOpcode::COPY));
      return true;
    } else if (DstRB.getID() == AArch64::FPRRegBankID) {
      if (DstTy == LLT::fixed_vector(4, 16) &&
          SrcTy == LLT::fixed_vector(4, 32)) {
        I.setDesc(TII.get(AArch64::XTNv4i16));
        constrainSelectedInstRegOperands(I, TII, TRI, RBI);
        return true;
      }

      if (!SrcTy.isVector() && SrcTy.getSizeInBits() == 128) {
        MachineInstr *Extract = emitExtractVectorElt(
            DstReg, DstRB, LLT::scalar(DstTy.getSizeInBits()), SrcReg, 0, MIB);
        if (!Extract)
          return false;
        I.eraseFromParent();
        return true;
      }

      // We might have a vector G_PTRTOINT, in which case just emit a COPY.
      if (Opcode == TargetOpcode::G_PTRTOINT) {
        assert(DstTy.isVector() && "Expected an FPR ptrtoint to be a vector");
        I.setDesc(TII.get(TargetOpcode::COPY));
        return selectCopy(I, TII, MRI, TRI, RBI);
      }
    }

    return false;
  }

  case TargetOpcode::G_ANYEXT: {
    if (selectUSMovFromExtend(I, MRI))
      return true;

    const Register DstReg = I.getOperand(0).getReg();
    const Register SrcReg = I.getOperand(1).getReg();

    const RegisterBank &RBDst = *RBI.getRegBank(DstReg, MRI, TRI);
    if (RBDst.getID() != AArch64::GPRRegBankID) {
      LLVM_DEBUG(dbgs() << "G_ANYEXT on bank: " << RBDst
                        << ", expected: GPR\n");
      return false;
    }

    const RegisterBank &RBSrc = *RBI.getRegBank(SrcReg, MRI, TRI);
    if (RBSrc.getID() != AArch64::GPRRegBankID) {
      LLVM_DEBUG(dbgs() << "G_ANYEXT on bank: " << RBSrc
                        << ", expected: GPR\n");
      return false;
    }

    const unsigned DstSize = MRI.getType(DstReg).getSizeInBits();

    if (DstSize == 0) {
      LLVM_DEBUG(dbgs() << "G_ANYEXT operand has no size, not a gvreg?\n");
      return false;
    }

    if (DstSize != 64 && DstSize > 32) {
      LLVM_DEBUG(dbgs() << "G_ANYEXT to size: " << DstSize
                        << ", expected: 32 or 64\n");
      return false;
    }
    // At this point G_ANYEXT is just like a plain COPY, but we need
    // to explicitly form the 64-bit value if any.
    if (DstSize > 32) {
      Register ExtSrc = MRI.createVirtualRegister(&AArch64::GPR64allRegClass);
      BuildMI(MBB, I, I.getDebugLoc(), TII.get(AArch64::SUBREG_TO_REG))
          .addDef(ExtSrc)
          .addImm(0)
          .addUse(SrcReg)
          .addImm(AArch64::sub_32);
      I.getOperand(1).setReg(ExtSrc);
    }
    return selectCopy(I, TII, MRI, TRI, RBI);
  }

  case TargetOpcode::G_ZEXT:
  case TargetOpcode::G_SEXT_INREG:
  case TargetOpcode::G_SEXT: {
    if (selectUSMovFromExtend(I, MRI))
      return true;

    unsigned Opcode = I.getOpcode();
    const bool IsSigned = Opcode != TargetOpcode::G_ZEXT;
    const Register DefReg = I.getOperand(0).getReg();
    Register SrcReg = I.getOperand(1).getReg();
    const LLT DstTy = MRI.getType(DefReg);
    const LLT SrcTy = MRI.getType(SrcReg);
    unsigned DstSize = DstTy.getSizeInBits();
    unsigned SrcSize = SrcTy.getSizeInBits();

    // SEXT_INREG has the same src reg size as dst, the size of the value to be
    // extended is encoded in the imm.
    if (Opcode == TargetOpcode::G_SEXT_INREG)
      SrcSize = I.getOperand(2).getImm();

    if (DstTy.isVector())
      return false; // Should be handled by imported patterns.

    assert((*RBI.getRegBank(DefReg, MRI, TRI)).getID() ==
               AArch64::GPRRegBankID &&
           "Unexpected ext regbank");

    MachineInstr *ExtI;

    // First check if we're extending the result of a load which has a dest type
    // smaller than 32 bits, then this zext is redundant. GPR32 is the smallest
    // GPR register on AArch64 and all loads which are smaller automatically
    // zero-extend the upper bits. E.g.
    // %v(s8) = G_LOAD %p, :: (load 1)
    // %v2(s32) = G_ZEXT %v(s8)
    if (!IsSigned) {
      auto *LoadMI = getOpcodeDef(TargetOpcode::G_LOAD, SrcReg, MRI);
      bool IsGPR =
          RBI.getRegBank(SrcReg, MRI, TRI)->getID() == AArch64::GPRRegBankID;
      if (LoadMI && IsGPR) {
        const MachineMemOperand *MemOp = *LoadMI->memoperands_begin();
        unsigned BytesLoaded = MemOp->getSize();
        if (BytesLoaded < 4 && SrcTy.getSizeInBytes() == BytesLoaded)
          return selectCopy(I, TII, MRI, TRI, RBI);
      }

      // For the 32-bit -> 64-bit case, we can emit a mov (ORRWrs)
      // + SUBREG_TO_REG.
      if (IsGPR && SrcSize == 32 && DstSize == 64) {
        Register SubregToRegSrc =
            MRI.createVirtualRegister(&AArch64::GPR32RegClass);
        const Register ZReg = AArch64::WZR;
        MIB.buildInstr(AArch64::ORRWrs, {SubregToRegSrc}, {ZReg, SrcReg})
            .addImm(0);

        MIB.buildInstr(AArch64::SUBREG_TO_REG, {DefReg}, {})
            .addImm(0)
            .addUse(SubregToRegSrc)
            .addImm(AArch64::sub_32);

        if (!RBI.constrainGenericRegister(DefReg, AArch64::GPR64RegClass,
                                          MRI)) {
          LLVM_DEBUG(dbgs() << "Failed to constrain G_ZEXT destination\n");
          return false;
        }

        if (!RBI.constrainGenericRegister(SrcReg, AArch64::GPR32RegClass,
                                          MRI)) {
          LLVM_DEBUG(dbgs() << "Failed to constrain G_ZEXT source\n");
          return false;
        }

        I.eraseFromParent();
        return true;
      }
    }

    if (DstSize == 64) {
      if (Opcode != TargetOpcode::G_SEXT_INREG) {
        // FIXME: Can we avoid manually doing this?
        if (!RBI.constrainGenericRegister(SrcReg, AArch64::GPR32RegClass,
                                          MRI)) {
          LLVM_DEBUG(dbgs() << "Failed to constrain " << TII.getName(Opcode)
                            << " operand\n");
          return false;
        }
        SrcReg = MIB.buildInstr(AArch64::SUBREG_TO_REG,
                                {&AArch64::GPR64RegClass}, {})
                     .addImm(0)
                     .addUse(SrcReg)
                     .addImm(AArch64::sub_32)
                     .getReg(0);
      }

      ExtI = MIB.buildInstr(IsSigned ? AArch64::SBFMXri : AArch64::UBFMXri,
                             {DefReg}, {SrcReg})
                  .addImm(0)
                  .addImm(SrcSize - 1);
    } else if (DstSize <= 32) {
      ExtI = MIB.buildInstr(IsSigned ? AArch64::SBFMWri : AArch64::UBFMWri,
                             {DefReg}, {SrcReg})
                  .addImm(0)
                  .addImm(SrcSize - 1);
    } else {
      return false;
    }

    constrainSelectedInstRegOperands(*ExtI, TII, TRI, RBI);
    I.eraseFromParent();
    return true;
  }

  case TargetOpcode::G_SITOFP:
  case TargetOpcode::G_UITOFP:
  case TargetOpcode::G_FPTOSI:
  case TargetOpcode::G_FPTOUI: {
    const LLT DstTy = MRI.getType(I.getOperand(0).getReg()),
              SrcTy = MRI.getType(I.getOperand(1).getReg());
    const unsigned NewOpc = selectFPConvOpc(Opcode, DstTy, SrcTy);
    if (NewOpc == Opcode)
      return false;

    I.setDesc(TII.get(NewOpc));
    constrainSelectedInstRegOperands(I, TII, TRI, RBI);
    I.setFlags(MachineInstr::NoFPExcept);

    return true;
  }

  case TargetOpcode::G_FREEZE:
    return selectCopy(I, TII, MRI, TRI, RBI);

  case TargetOpcode::G_INTTOPTR:
    // The importer is currently unable to import pointer types since they
    // didn't exist in SelectionDAG.
    return selectCopy(I, TII, MRI, TRI, RBI);

  case TargetOpcode::G_BITCAST:
    // Imported SelectionDAG rules can handle every bitcast except those that
    // bitcast from a type to the same type. Ideally, these shouldn't occur
    // but we might not run an optimizer that deletes them. The other exception
    // is bitcasts involving pointer types, as SelectionDAG has no knowledge
    // of them.
    return selectCopy(I, TII, MRI, TRI, RBI);

  case TargetOpcode::G_SELECT: {
    auto &Sel = cast<GSelect>(I);
    const Register CondReg = Sel.getCondReg();
    const Register TReg = Sel.getTrueReg();
    const Register FReg = Sel.getFalseReg();

    if (tryOptSelect(Sel))
      return true;

    // Make sure to use an unused vreg instead of wzr, so that the peephole
    // optimizations will be able to optimize these.
    Register DeadVReg = MRI.createVirtualRegister(&AArch64::GPR32RegClass);
    auto TstMI = MIB.buildInstr(AArch64::ANDSWri, {DeadVReg}, {CondReg})
                     .addImm(AArch64_AM::encodeLogicalImmediate(1, 32));
    constrainSelectedInstRegOperands(*TstMI, TII, TRI, RBI);
    if (!emitSelect(Sel.getReg(0), TReg, FReg, AArch64CC::NE, MIB))
      return false;
    Sel.eraseFromParent();
    return true;
  }
  case TargetOpcode::G_ICMP: {
    if (Ty.isVector())
      return selectVectorICmp(I, MRI);

    if (Ty != LLT::scalar(32)) {
      LLVM_DEBUG(dbgs() << "G_ICMP result has type: " << Ty
                        << ", expected: " << LLT::scalar(32) << '\n');
      return false;
    }

    auto Pred = static_cast<CmpInst::Predicate>(I.getOperand(1).getPredicate());
    const AArch64CC::CondCode InvCC =
        changeICMPPredToAArch64CC(CmpInst::getInversePredicate(Pred));
    emitIntegerCompare(I.getOperand(2), I.getOperand(3), I.getOperand(1), MIB);
    emitCSINC(/*Dst=*/I.getOperand(0).getReg(), /*Src1=*/AArch64::WZR,
              /*Src2=*/AArch64::WZR, InvCC, MIB);
    I.eraseFromParent();
    return true;
  }

  case TargetOpcode::G_FCMP: {
    CmpInst::Predicate Pred =
        static_cast<CmpInst::Predicate>(I.getOperand(1).getPredicate());
    if (!emitFPCompare(I.getOperand(2).getReg(), I.getOperand(3).getReg(), MIB,
                       Pred) ||
        !emitCSetForFCmp(I.getOperand(0).getReg(), Pred, MIB))
      return false;
    I.eraseFromParent();
    return true;
  }
  case TargetOpcode::G_VASTART:
    return STI.isTargetDarwin() ? selectVaStartDarwin(I, MF, MRI)
                                : selectVaStartAAPCS(I, MF, MRI);
  case TargetOpcode::G_INTRINSIC:
    return selectIntrinsic(I, MRI);
  case TargetOpcode::G_INTRINSIC_W_SIDE_EFFECTS:
    return selectIntrinsicWithSideEffects(I, MRI);
  case TargetOpcode::G_IMPLICIT_DEF: {
    I.setDesc(TII.get(TargetOpcode::IMPLICIT_DEF));
    const LLT DstTy = MRI.getType(I.getOperand(0).getReg());
    const Register DstReg = I.getOperand(0).getReg();
    const RegisterBank &DstRB = *RBI.getRegBank(DstReg, MRI, TRI);
    const TargetRegisterClass *DstRC = getRegClassForTypeOnBank(DstTy, DstRB);
    RBI.constrainGenericRegister(DstReg, *DstRC, MRI);
    return true;
  }
  case TargetOpcode::G_BLOCK_ADDR: {
    if (TM.getCodeModel() == CodeModel::Large) {
      materializeLargeCMVal(I, I.getOperand(1).getBlockAddress(), 0);
      I.eraseFromParent();
      return true;
    } else {
      I.setDesc(TII.get(AArch64::MOVaddrBA));
      auto MovMI = BuildMI(MBB, I, I.getDebugLoc(), TII.get(AArch64::MOVaddrBA),
                           I.getOperand(0).getReg())
                       .addBlockAddress(I.getOperand(1).getBlockAddress(),
                                        /* Offset */ 0, AArch64II::MO_PAGE)
                       .addBlockAddress(
                           I.getOperand(1).getBlockAddress(), /* Offset */ 0,
                           AArch64II::MO_NC | AArch64II::MO_PAGEOFF);
      I.eraseFromParent();
      return constrainSelectedInstRegOperands(*MovMI, TII, TRI, RBI);
    }
  }
  case AArch64::G_DUP: {
    // When the scalar of G_DUP is an s8/s16 gpr, they can't be selected by
    // imported patterns. Do it manually here. Avoiding generating s16 gpr is
    // difficult because at RBS we may end up pessimizing the fpr case if we
    // decided to add an anyextend to fix this. Manual selection is the most
    // robust solution for now.
    if (RBI.getRegBank(I.getOperand(1).getReg(), MRI, TRI)->getID() !=
        AArch64::GPRRegBankID)
      return false; // We expect the fpr regbank case to be imported.
    LLT VecTy = MRI.getType(I.getOperand(0).getReg());
    if (VecTy == LLT::fixed_vector(8, 8))
      I.setDesc(TII.get(AArch64::DUPv8i8gpr));
    else if (VecTy == LLT::fixed_vector(16, 8))
      I.setDesc(TII.get(AArch64::DUPv16i8gpr));
    else if (VecTy == LLT::fixed_vector(4, 16))
      I.setDesc(TII.get(AArch64::DUPv4i16gpr));
    else if (VecTy == LLT::fixed_vector(8, 16))
      I.setDesc(TII.get(AArch64::DUPv8i16gpr));
    else
      return false;
    return constrainSelectedInstRegOperands(I, TII, TRI, RBI);
  }
  case TargetOpcode::G_INTRINSIC_TRUNC:
    return selectIntrinsicTrunc(I, MRI);
  case TargetOpcode::G_INTRINSIC_ROUND:
    return selectIntrinsicRound(I, MRI);
  case TargetOpcode::G_BUILD_VECTOR:
    return selectBuildVector(I, MRI);
  case TargetOpcode::G_MERGE_VALUES:
    return selectMergeValues(I, MRI);
  case TargetOpcode::G_UNMERGE_VALUES:
    return selectUnmergeValues(I, MRI);
  case TargetOpcode::G_SHUFFLE_VECTOR:
    return selectShuffleVector(I, MRI);
  case TargetOpcode::G_EXTRACT_VECTOR_ELT:
    return selectExtractElt(I, MRI);
  case TargetOpcode::G_INSERT_VECTOR_ELT:
    return selectInsertElt(I, MRI);
  case TargetOpcode::G_CONCAT_VECTORS:
    return selectConcatVectors(I, MRI);
  case TargetOpcode::G_JUMP_TABLE:
    return selectJumpTable(I, MRI);
  case TargetOpcode::G_VECREDUCE_FADD:
  case TargetOpcode::G_VECREDUCE_ADD:
    return selectReduction(I, MRI);
  case TargetOpcode::G_MEMCPY:
  case TargetOpcode::G_MEMCPY_INLINE:
  case TargetOpcode::G_MEMMOVE:
  case TargetOpcode::G_MEMSET:
    assert(STI.hasMOPS() && "Shouldn't get here without +mops feature");
    return selectMOPS(I, MRI);
  }

  return false;
}

bool AArch64InstructionSelector::selectReduction(MachineInstr &I,
                                                 MachineRegisterInfo &MRI) {
  Register VecReg = I.getOperand(1).getReg();
  LLT VecTy = MRI.getType(VecReg);
  if (I.getOpcode() == TargetOpcode::G_VECREDUCE_ADD) {
    // For <2 x i32> ADDPv2i32 generates an FPR64 value, so we need to emit
    // a subregister copy afterwards.
    if (VecTy == LLT::fixed_vector(2, 32)) {
      Register DstReg = I.getOperand(0).getReg();
      auto AddP = MIB.buildInstr(AArch64::ADDPv2i32, {&AArch64::FPR64RegClass},
                                 {VecReg, VecReg});
      auto Copy = MIB.buildInstr(TargetOpcode::COPY, {DstReg}, {})
                      .addReg(AddP.getReg(0), 0, AArch64::ssub)
                      .getReg(0);
      RBI.constrainGenericRegister(Copy, AArch64::FPR32RegClass, MRI);
      I.eraseFromParent();
      return constrainSelectedInstRegOperands(*AddP, TII, TRI, RBI);
    }

    unsigned Opc = 0;
    if (VecTy == LLT::fixed_vector(16, 8))
      Opc = AArch64::ADDVv16i8v;
    else if (VecTy == LLT::fixed_vector(8, 16))
      Opc = AArch64::ADDVv8i16v;
    else if (VecTy == LLT::fixed_vector(4, 32))
      Opc = AArch64::ADDVv4i32v;
    else if (VecTy == LLT::fixed_vector(2, 64))
      Opc = AArch64::ADDPv2i64p;
    else {
      LLVM_DEBUG(dbgs() << "Unhandled type for add reduction");
      return false;
    }
    I.setDesc(TII.get(Opc));
    return constrainSelectedInstRegOperands(I, TII, TRI, RBI);
  }

  if (I.getOpcode() == TargetOpcode::G_VECREDUCE_FADD) {
    unsigned Opc = 0;
    if (VecTy == LLT::fixed_vector(2, 32))
      Opc = AArch64::FADDPv2i32p;
    else if (VecTy == LLT::fixed_vector(2, 64))
      Opc = AArch64::FADDPv2i64p;
    else {
      LLVM_DEBUG(dbgs() << "Unhandled type for fadd reduction");
      return false;
    }
    I.setDesc(TII.get(Opc));
    return constrainSelectedInstRegOperands(I, TII, TRI, RBI);
  }
  return false;
}

bool AArch64InstructionSelector::selectMOPS(MachineInstr &GI,
                                            MachineRegisterInfo &MRI) {
  unsigned Mopcode;
  switch (GI.getOpcode()) {
  case TargetOpcode::G_MEMCPY:
  case TargetOpcode::G_MEMCPY_INLINE:
    Mopcode = AArch64::MOPSMemoryCopyPseudo;
    break;
  case TargetOpcode::G_MEMMOVE:
    Mopcode = AArch64::MOPSMemoryMovePseudo;
    break;
  case TargetOpcode::G_MEMSET:
    // For tagged memset see llvm.aarch64.mops.memset.tag
    Mopcode = AArch64::MOPSMemorySetPseudo;
    break;
  }

  auto &DstPtr = GI.getOperand(0);
  auto &SrcOrVal = GI.getOperand(1);
  auto &Size = GI.getOperand(2);

  // Create copies of the registers that can be clobbered.
  const Register DstPtrCopy = MRI.cloneVirtualRegister(DstPtr.getReg());
  const Register SrcValCopy = MRI.cloneVirtualRegister(SrcOrVal.getReg());
  const Register SizeCopy = MRI.cloneVirtualRegister(Size.getReg());

  const bool IsSet = Mopcode == AArch64::MOPSMemorySetPseudo;
  const auto &SrcValRegClass =
      IsSet ? AArch64::GPR64RegClass : AArch64::GPR64commonRegClass;

  // Constrain to specific registers
  RBI.constrainGenericRegister(DstPtrCopy, AArch64::GPR64commonRegClass, MRI);
  RBI.constrainGenericRegister(SrcValCopy, SrcValRegClass, MRI);
  RBI.constrainGenericRegister(SizeCopy, AArch64::GPR64RegClass, MRI);

  MIB.buildCopy(DstPtrCopy, DstPtr);
  MIB.buildCopy(SrcValCopy, SrcOrVal);
  MIB.buildCopy(SizeCopy, Size);

  // New instruction uses the copied registers because it must update them.
  // The defs are not used since they don't exist in G_MEM*. They are still
  // tied.
  // Note: order of operands is different from G_MEMSET, G_MEMCPY, G_MEMMOVE
  Register DefDstPtr = MRI.createVirtualRegister(&AArch64::GPR64commonRegClass);
  Register DefSize = MRI.createVirtualRegister(&AArch64::GPR64RegClass);
  if (IsSet) {
    MIB.buildInstr(Mopcode, {DefDstPtr, DefSize},
                   {DstPtrCopy, SizeCopy, SrcValCopy});
  } else {
    Register DefSrcPtr = MRI.createVirtualRegister(&SrcValRegClass);
    MIB.buildInstr(Mopcode, {DefDstPtr, DefSrcPtr, DefSize},
                   {DstPtrCopy, SrcValCopy, SizeCopy});
  }

  GI.eraseFromParent();
  return true;
}

bool AArch64InstructionSelector::selectBrJT(MachineInstr &I,
                                            MachineRegisterInfo &MRI) {
  assert(I.getOpcode() == TargetOpcode::G_BRJT && "Expected G_BRJT");
  Register JTAddr = I.getOperand(0).getReg();
  unsigned JTI = I.getOperand(1).getIndex();
  Register Index = I.getOperand(2).getReg();

  Register TargetReg = MRI.createVirtualRegister(&AArch64::GPR64RegClass);
  Register ScratchReg = MRI.createVirtualRegister(&AArch64::GPR64spRegClass);

  MF->getInfo<AArch64FunctionInfo>()->setJumpTableEntryInfo(JTI, 4, nullptr);
  auto JumpTableInst = MIB.buildInstr(AArch64::JumpTableDest32,
                                      {TargetReg, ScratchReg}, {JTAddr, Index})
                           .addJumpTableIndex(JTI);
  // Build the indirect branch.
  MIB.buildInstr(AArch64::BR, {}, {TargetReg});
  I.eraseFromParent();
  return constrainSelectedInstRegOperands(*JumpTableInst, TII, TRI, RBI);
}

bool AArch64InstructionSelector::selectJumpTable(MachineInstr &I,
                                                 MachineRegisterInfo &MRI) {
  assert(I.getOpcode() == TargetOpcode::G_JUMP_TABLE && "Expected jump table");
  assert(I.getOperand(1).isJTI() && "Jump table op should have a JTI!");

  Register DstReg = I.getOperand(0).getReg();
  unsigned JTI = I.getOperand(1).getIndex();
  // We generate a MOVaddrJT which will get expanded to an ADRP + ADD later.
  auto MovMI =
    MIB.buildInstr(AArch64::MOVaddrJT, {DstReg}, {})
          .addJumpTableIndex(JTI, AArch64II::MO_PAGE)
          .addJumpTableIndex(JTI, AArch64II::MO_NC | AArch64II::MO_PAGEOFF);
  I.eraseFromParent();
  return constrainSelectedInstRegOperands(*MovMI, TII, TRI, RBI);
}

bool AArch64InstructionSelector::selectTLSGlobalValue(
    MachineInstr &I, MachineRegisterInfo &MRI) {
  if (!STI.isTargetMachO())
    return false;
  MachineFunction &MF = *I.getParent()->getParent();
  MF.getFrameInfo().setAdjustsStack(true);

  const auto &GlobalOp = I.getOperand(1);
  assert(GlobalOp.getOffset() == 0 &&
         "Shouldn't have an offset on TLS globals!");
  const GlobalValue &GV = *GlobalOp.getGlobal();

  auto LoadGOT =
      MIB.buildInstr(AArch64::LOADgot, {&AArch64::GPR64commonRegClass}, {})
          .addGlobalAddress(&GV, 0, AArch64II::MO_TLS);

  auto Load = MIB.buildInstr(AArch64::LDRXui, {&AArch64::GPR64commonRegClass},
                             {LoadGOT.getReg(0)})
                  .addImm(0);

  MIB.buildCopy(Register(AArch64::X0), LoadGOT.getReg(0));
  // TLS calls preserve all registers except those that absolutely must be
  // trashed: X0 (it takes an argument), LR (it's a call) and NZCV (let's not be
  // silly).
  MIB.buildInstr(getBLRCallOpcode(MF), {}, {Load})
      .addUse(AArch64::X0, RegState::Implicit)
      .addDef(AArch64::X0, RegState::Implicit)
      .addRegMask(TRI.getTLSCallPreservedMask());

  MIB.buildCopy(I.getOperand(0).getReg(), Register(AArch64::X0));
  RBI.constrainGenericRegister(I.getOperand(0).getReg(), AArch64::GPR64RegClass,
                               MRI);
  I.eraseFromParent();
  return true;
}

bool AArch64InstructionSelector::selectIntrinsicTrunc(
    MachineInstr &I, MachineRegisterInfo &MRI) const {
  const LLT SrcTy = MRI.getType(I.getOperand(0).getReg());

  // Select the correct opcode.
  unsigned Opc = 0;
  if (!SrcTy.isVector()) {
    switch (SrcTy.getSizeInBits()) {
    default:
    case 16:
      Opc = AArch64::FRINTZHr;
      break;
    case 32:
      Opc = AArch64::FRINTZSr;
      break;
    case 64:
      Opc = AArch64::FRINTZDr;
      break;
    }
  } else {
    unsigned NumElts = SrcTy.getNumElements();
    switch (SrcTy.getElementType().getSizeInBits()) {
    default:
      break;
    case 16:
      if (NumElts == 4)
        Opc = AArch64::FRINTZv4f16;
      else if (NumElts == 8)
        Opc = AArch64::FRINTZv8f16;
      break;
    case 32:
      if (NumElts == 2)
        Opc = AArch64::FRINTZv2f32;
      else if (NumElts == 4)
        Opc = AArch64::FRINTZv4f32;
      break;
    case 64:
      if (NumElts == 2)
        Opc = AArch64::FRINTZv2f64;
      break;
    }
  }

  if (!Opc) {
    // Didn't get an opcode above, bail.
    LLVM_DEBUG(dbgs() << "Unsupported type for G_INTRINSIC_TRUNC!\n");
    return false;
  }

  // Legalization would have set us up perfectly for this; we just need to
  // set the opcode and move on.
  I.setDesc(TII.get(Opc));
  return constrainSelectedInstRegOperands(I, TII, TRI, RBI);
}

bool AArch64InstructionSelector::selectIntrinsicRound(
    MachineInstr &I, MachineRegisterInfo &MRI) const {
  const LLT SrcTy = MRI.getType(I.getOperand(0).getReg());

  // Select the correct opcode.
  unsigned Opc = 0;
  if (!SrcTy.isVector()) {
    switch (SrcTy.getSizeInBits()) {
    default:
    case 16:
      Opc = AArch64::FRINTAHr;
      break;
    case 32:
      Opc = AArch64::FRINTASr;
      break;
    case 64:
      Opc = AArch64::FRINTADr;
      break;
    }
  } else {
    unsigned NumElts = SrcTy.getNumElements();
    switch (SrcTy.getElementType().getSizeInBits()) {
    default:
      break;
    case 16:
      if (NumElts == 4)
        Opc = AArch64::FRINTAv4f16;
      else if (NumElts == 8)
        Opc = AArch64::FRINTAv8f16;
      break;
    case 32:
      if (NumElts == 2)
        Opc = AArch64::FRINTAv2f32;
      else if (NumElts == 4)
        Opc = AArch64::FRINTAv4f32;
      break;
    case 64:
      if (NumElts == 2)
        Opc = AArch64::FRINTAv2f64;
      break;
    }
  }

  if (!Opc) {
    // Didn't get an opcode above, bail.
    LLVM_DEBUG(dbgs() << "Unsupported type for G_INTRINSIC_ROUND!\n");
    return false;
  }

  // Legalization would have set us up perfectly for this; we just need to
  // set the opcode and move on.
  I.setDesc(TII.get(Opc));
  return constrainSelectedInstRegOperands(I, TII, TRI, RBI);
}

bool AArch64InstructionSelector::selectVectorICmp(
    MachineInstr &I, MachineRegisterInfo &MRI) {
  Register DstReg = I.getOperand(0).getReg();
  LLT DstTy = MRI.getType(DstReg);
  Register SrcReg = I.getOperand(2).getReg();
  Register Src2Reg = I.getOperand(3).getReg();
  LLT SrcTy = MRI.getType(SrcReg);

  unsigned SrcEltSize = SrcTy.getElementType().getSizeInBits();
  unsigned NumElts = DstTy.getNumElements();

  // First index is element size, 0 == 8b, 1 == 16b, 2 == 32b, 3 == 64b
  // Second index is num elts, 0 == v2, 1 == v4, 2 == v8, 3 == v16
  // Third index is cc opcode:
  // 0 == eq
  // 1 == ugt
  // 2 == uge
  // 3 == ult
  // 4 == ule
  // 5 == sgt
  // 6 == sge
  // 7 == slt
  // 8 == sle
  // ne is done by negating 'eq' result.

  // This table below assumes that for some comparisons the operands will be
  // commuted.
  // ult op == commute + ugt op
  // ule op == commute + uge op
  // slt op == commute + sgt op
  // sle op == commute + sge op
  unsigned PredIdx = 0;
  bool SwapOperands = false;
  CmpInst::Predicate Pred = (CmpInst::Predicate)I.getOperand(1).getPredicate();
  switch (Pred) {
  case CmpInst::ICMP_NE:
  case CmpInst::ICMP_EQ:
    PredIdx = 0;
    break;
  case CmpInst::ICMP_UGT:
    PredIdx = 1;
    break;
  case CmpInst::ICMP_UGE:
    PredIdx = 2;
    break;
  case CmpInst::ICMP_ULT:
    PredIdx = 3;
    SwapOperands = true;
    break;
  case CmpInst::ICMP_ULE:
    PredIdx = 4;
    SwapOperands = true;
    break;
  case CmpInst::ICMP_SGT:
    PredIdx = 5;
    break;
  case CmpInst::ICMP_SGE:
    PredIdx = 6;
    break;
  case CmpInst::ICMP_SLT:
    PredIdx = 7;
    SwapOperands = true;
    break;
  case CmpInst::ICMP_SLE:
    PredIdx = 8;
    SwapOperands = true;
    break;
  default:
    llvm_unreachable("Unhandled icmp predicate");
    return false;
  }

  // This table obviously should be tablegen'd when we have our GISel native
  // tablegen selector.

  static const unsigned OpcTable[4][4][9] = {
      {
          {0 /* invalid */, 0 /* invalid */, 0 /* invalid */, 0 /* invalid */,
           0 /* invalid */, 0 /* invalid */, 0 /* invalid */, 0 /* invalid */,
           0 /* invalid */},
          {0 /* invalid */, 0 /* invalid */, 0 /* invalid */, 0 /* invalid */,
           0 /* invalid */, 0 /* invalid */, 0 /* invalid */, 0 /* invalid */,
           0 /* invalid */},
          {AArch64::CMEQv8i8, AArch64::CMHIv8i8, AArch64::CMHSv8i8,
           AArch64::CMHIv8i8, AArch64::CMHSv8i8, AArch64::CMGTv8i8,
           AArch64::CMGEv8i8, AArch64::CMGTv8i8, AArch64::CMGEv8i8},
          {AArch64::CMEQv16i8, AArch64::CMHIv16i8, AArch64::CMHSv16i8,
           AArch64::CMHIv16i8, AArch64::CMHSv16i8, AArch64::CMGTv16i8,
           AArch64::CMGEv16i8, AArch64::CMGTv16i8, AArch64::CMGEv16i8}
      },
      {
          {0 /* invalid */, 0 /* invalid */, 0 /* invalid */, 0 /* invalid */,
           0 /* invalid */, 0 /* invalid */, 0 /* invalid */, 0 /* invalid */,
           0 /* invalid */},
          {AArch64::CMEQv4i16, AArch64::CMHIv4i16, AArch64::CMHSv4i16,
           AArch64::CMHIv4i16, AArch64::CMHSv4i16, AArch64::CMGTv4i16,
           AArch64::CMGEv4i16, AArch64::CMGTv4i16, AArch64::CMGEv4i16},
          {AArch64::CMEQv8i16, AArch64::CMHIv8i16, AArch64::CMHSv8i16,
           AArch64::CMHIv8i16, AArch64::CMHSv8i16, AArch64::CMGTv8i16,
           AArch64::CMGEv8i16, AArch64::CMGTv8i16, AArch64::CMGEv8i16},
          {0 /* invalid */, 0 /* invalid */, 0 /* invalid */, 0 /* invalid */,
           0 /* invalid */, 0 /* invalid */, 0 /* invalid */, 0 /* invalid */,
           0 /* invalid */}
      },
      {
          {AArch64::CMEQv2i32, AArch64::CMHIv2i32, AArch64::CMHSv2i32,
           AArch64::CMHIv2i32, AArch64::CMHSv2i32, AArch64::CMGTv2i32,
           AArch64::CMGEv2i32, AArch64::CMGTv2i32, AArch64::CMGEv2i32},
          {AArch64::CMEQv4i32, AArch64::CMHIv4i32, AArch64::CMHSv4i32,
           AArch64::CMHIv4i32, AArch64::CMHSv4i32, AArch64::CMGTv4i32,
           AArch64::CMGEv4i32, AArch64::CMGTv4i32, AArch64::CMGEv4i32},
          {0 /* invalid */, 0 /* invalid */, 0 /* invalid */, 0 /* invalid */,
           0 /* invalid */, 0 /* invalid */, 0 /* invalid */, 0 /* invalid */,
           0 /* invalid */},
          {0 /* invalid */, 0 /* invalid */, 0 /* invalid */, 0 /* invalid */,
           0 /* invalid */, 0 /* invalid */, 0 /* invalid */, 0 /* invalid */,
           0 /* invalid */}
      },
      {
          {AArch64::CMEQv2i64, AArch64::CMHIv2i64, AArch64::CMHSv2i64,
           AArch64::CMHIv2i64, AArch64::CMHSv2i64, AArch64::CMGTv2i64,
           AArch64::CMGEv2i64, AArch64::CMGTv2i64, AArch64::CMGEv2i64},
          {0 /* invalid */, 0 /* invalid */, 0 /* invalid */, 0 /* invalid */,
           0 /* invalid */, 0 /* invalid */, 0 /* invalid */, 0 /* invalid */,
           0 /* invalid */},
          {0 /* invalid */, 0 /* invalid */, 0 /* invalid */, 0 /* invalid */,
           0 /* invalid */, 0 /* invalid */, 0 /* invalid */, 0 /* invalid */,
           0 /* invalid */},
          {0 /* invalid */, 0 /* invalid */, 0 /* invalid */, 0 /* invalid */,
           0 /* invalid */, 0 /* invalid */, 0 /* invalid */, 0 /* invalid */,
           0 /* invalid */}
      },
  };
  unsigned EltIdx = Log2_32(SrcEltSize / 8);
  unsigned NumEltsIdx = Log2_32(NumElts / 2);
  unsigned Opc = OpcTable[EltIdx][NumEltsIdx][PredIdx];
  if (!Opc) {
    LLVM_DEBUG(dbgs() << "Could not map G_ICMP to cmp opcode");
    return false;
  }

  const RegisterBank &VecRB = *RBI.getRegBank(SrcReg, MRI, TRI);
  const TargetRegisterClass *SrcRC =
      getRegClassForTypeOnBank(SrcTy, VecRB, true);
  if (!SrcRC) {
    LLVM_DEBUG(dbgs() << "Could not determine source register class.\n");
    return false;
  }

  unsigned NotOpc = Pred == ICmpInst::ICMP_NE ? AArch64::NOTv8i8 : 0;
  if (SrcTy.getSizeInBits() == 128)
    NotOpc = NotOpc ? AArch64::NOTv16i8 : 0;

  if (SwapOperands)
    std::swap(SrcReg, Src2Reg);

  auto Cmp = MIB.buildInstr(Opc, {SrcRC}, {SrcReg, Src2Reg});
  constrainSelectedInstRegOperands(*Cmp, TII, TRI, RBI);

  // Invert if we had a 'ne' cc.
  if (NotOpc) {
    Cmp = MIB.buildInstr(NotOpc, {DstReg}, {Cmp});
    constrainSelectedInstRegOperands(*Cmp, TII, TRI, RBI);
  } else {
    MIB.buildCopy(DstReg, Cmp.getReg(0));
  }
  RBI.constrainGenericRegister(DstReg, *SrcRC, MRI);
  I.eraseFromParent();
  return true;
}

MachineInstr *AArch64InstructionSelector::emitScalarToVector(
    unsigned EltSize, const TargetRegisterClass *DstRC, Register Scalar,
    MachineIRBuilder &MIRBuilder) const {
  auto Undef = MIRBuilder.buildInstr(TargetOpcode::IMPLICIT_DEF, {DstRC}, {});

  auto BuildFn = [&](unsigned SubregIndex) {
    auto Ins =
        MIRBuilder
            .buildInstr(TargetOpcode::INSERT_SUBREG, {DstRC}, {Undef, Scalar})
            .addImm(SubregIndex);
    constrainSelectedInstRegOperands(*Undef, TII, TRI, RBI);
    constrainSelectedInstRegOperands(*Ins, TII, TRI, RBI);
    return &*Ins;
  };

  switch (EltSize) {
  case 16:
    return BuildFn(AArch64::hsub);
  case 32:
    return BuildFn(AArch64::ssub);
  case 64:
    return BuildFn(AArch64::dsub);
  default:
    return nullptr;
  }
}

bool AArch64InstructionSelector::selectMergeValues(
    MachineInstr &I, MachineRegisterInfo &MRI) {
  assert(I.getOpcode() == TargetOpcode::G_MERGE_VALUES && "unexpected opcode");
  const LLT DstTy = MRI.getType(I.getOperand(0).getReg());
  const LLT SrcTy = MRI.getType(I.getOperand(1).getReg());
  assert(!DstTy.isVector() && !SrcTy.isVector() && "invalid merge operation");
  const RegisterBank &RB = *RBI.getRegBank(I.getOperand(1).getReg(), MRI, TRI);

  if (I.getNumOperands() != 3)
    return false;

  // Merging 2 s64s into an s128.
  if (DstTy == LLT::scalar(128)) {
    if (SrcTy.getSizeInBits() != 64)
      return false;
    Register DstReg = I.getOperand(0).getReg();
    Register Src1Reg = I.getOperand(1).getReg();
    Register Src2Reg = I.getOperand(2).getReg();
    auto Tmp = MIB.buildInstr(TargetOpcode::IMPLICIT_DEF, {DstTy}, {});
    MachineInstr *InsMI = emitLaneInsert(std::nullopt, Tmp.getReg(0), Src1Reg,
                                         /* LaneIdx */ 0, RB, MIB);
    if (!InsMI)
      return false;
    MachineInstr *Ins2MI = emitLaneInsert(DstReg, InsMI->getOperand(0).getReg(),
                                          Src2Reg, /* LaneIdx */ 1, RB, MIB);
    if (!Ins2MI)
      return false;
    constrainSelectedInstRegOperands(*InsMI, TII, TRI, RBI);
    constrainSelectedInstRegOperands(*Ins2MI, TII, TRI, RBI);
    I.eraseFromParent();
    return true;
  }

  if (RB.getID() != AArch64::GPRRegBankID)
    return false;

  if (DstTy.getSizeInBits() != 64 || SrcTy.getSizeInBits() != 32)
    return false;

  auto *DstRC = &AArch64::GPR64RegClass;
  Register SubToRegDef = MRI.createVirtualRegister(DstRC);
  MachineInstr &SubRegMI = *BuildMI(*I.getParent(), I, I.getDebugLoc(),
                                    TII.get(TargetOpcode::SUBREG_TO_REG))
                                .addDef(SubToRegDef)
                                .addImm(0)
                                .addUse(I.getOperand(1).getReg())
                                .addImm(AArch64::sub_32);
  Register SubToRegDef2 = MRI.createVirtualRegister(DstRC);
  // Need to anyext the second scalar before we can use bfm
  MachineInstr &SubRegMI2 = *BuildMI(*I.getParent(), I, I.getDebugLoc(),
                                    TII.get(TargetOpcode::SUBREG_TO_REG))
                                .addDef(SubToRegDef2)
                                .addImm(0)
                                .addUse(I.getOperand(2).getReg())
                                .addImm(AArch64::sub_32);
  MachineInstr &BFM =
      *BuildMI(*I.getParent(), I, I.getDebugLoc(), TII.get(AArch64::BFMXri))
           .addDef(I.getOperand(0).getReg())
           .addUse(SubToRegDef)
           .addUse(SubToRegDef2)
           .addImm(32)
           .addImm(31);
  constrainSelectedInstRegOperands(SubRegMI, TII, TRI, RBI);
  constrainSelectedInstRegOperands(SubRegMI2, TII, TRI, RBI);
  constrainSelectedInstRegOperands(BFM, TII, TRI, RBI);
  I.eraseFromParent();
  return true;
}

static bool getLaneCopyOpcode(unsigned &CopyOpc, unsigned &ExtractSubReg,
                              const unsigned EltSize) {
  // Choose a lane copy opcode and subregister based off of the size of the
  // vector's elements.
  switch (EltSize) {
  case 8:
    CopyOpc = AArch64::DUPi8;
    ExtractSubReg = AArch64::bsub;
    break;
  case 16:
    CopyOpc = AArch64::DUPi16;
    ExtractSubReg = AArch64::hsub;
    break;
  case 32:
    CopyOpc = AArch64::DUPi32;
    ExtractSubReg = AArch64::ssub;
    break;
  case 64:
    CopyOpc = AArch64::DUPi64;
    ExtractSubReg = AArch64::dsub;
    break;
  default:
    // Unknown size, bail out.
    LLVM_DEBUG(dbgs() << "Elt size '" << EltSize << "' unsupported.\n");
    return false;
  }
  return true;
}

MachineInstr *AArch64InstructionSelector::emitExtractVectorElt(
    std::optional<Register> DstReg, const RegisterBank &DstRB, LLT ScalarTy,
    Register VecReg, unsigned LaneIdx, MachineIRBuilder &MIRBuilder) const {
  MachineRegisterInfo &MRI = *MIRBuilder.getMRI();
  unsigned CopyOpc = 0;
  unsigned ExtractSubReg = 0;
  if (!getLaneCopyOpcode(CopyOpc, ExtractSubReg, ScalarTy.getSizeInBits())) {
    LLVM_DEBUG(
        dbgs() << "Couldn't determine lane copy opcode for instruction.\n");
    return nullptr;
  }

  const TargetRegisterClass *DstRC =
      getRegClassForTypeOnBank(ScalarTy, DstRB, true);
  if (!DstRC) {
    LLVM_DEBUG(dbgs() << "Could not determine destination register class.\n");
    return nullptr;
  }

  const RegisterBank &VecRB = *RBI.getRegBank(VecReg, MRI, TRI);
  const LLT &VecTy = MRI.getType(VecReg);
  const TargetRegisterClass *VecRC =
      getRegClassForTypeOnBank(VecTy, VecRB, true);
  if (!VecRC) {
    LLVM_DEBUG(dbgs() << "Could not determine source register class.\n");
    return nullptr;
  }

  // The register that we're going to copy into.
  Register InsertReg = VecReg;
  if (!DstReg)
    DstReg = MRI.createVirtualRegister(DstRC);
  // If the lane index is 0, we just use a subregister COPY.
  if (LaneIdx == 0) {
    auto Copy = MIRBuilder.buildInstr(TargetOpcode::COPY, {*DstReg}, {})
                    .addReg(VecReg, 0, ExtractSubReg);
    RBI.constrainGenericRegister(*DstReg, *DstRC, MRI);
    return &*Copy;
  }

  // Lane copies require 128-bit wide registers. If we're dealing with an
  // unpacked vector, then we need to move up to that width. Insert an implicit
  // def and a subregister insert to get us there.
  if (VecTy.getSizeInBits() != 128) {
    MachineInstr *ScalarToVector = emitScalarToVector(
        VecTy.getSizeInBits(), &AArch64::FPR128RegClass, VecReg, MIRBuilder);
    if (!ScalarToVector)
      return nullptr;
    InsertReg = ScalarToVector->getOperand(0).getReg();
  }

  MachineInstr *LaneCopyMI =
      MIRBuilder.buildInstr(CopyOpc, {*DstReg}, {InsertReg}).addImm(LaneIdx);
  constrainSelectedInstRegOperands(*LaneCopyMI, TII, TRI, RBI);

  // Make sure that we actually constrain the initial copy.
  RBI.constrainGenericRegister(*DstReg, *DstRC, MRI);
  return LaneCopyMI;
}

bool AArch64InstructionSelector::selectExtractElt(
    MachineInstr &I, MachineRegisterInfo &MRI) {
  assert(I.getOpcode() == TargetOpcode::G_EXTRACT_VECTOR_ELT &&
         "unexpected opcode!");
  Register DstReg = I.getOperand(0).getReg();
  const LLT NarrowTy = MRI.getType(DstReg);
  const Register SrcReg = I.getOperand(1).getReg();
  const LLT WideTy = MRI.getType(SrcReg);
  (void)WideTy;
  assert(WideTy.getSizeInBits() >= NarrowTy.getSizeInBits() &&
         "source register size too small!");
  assert(!NarrowTy.isVector() && "cannot extract vector into vector!");

  // Need the lane index to determine the correct copy opcode.
  MachineOperand &LaneIdxOp = I.getOperand(2);
  assert(LaneIdxOp.isReg() && "Lane index operand was not a register?");

  if (RBI.getRegBank(DstReg, MRI, TRI)->getID() != AArch64::FPRRegBankID) {
    LLVM_DEBUG(dbgs() << "Cannot extract into GPR.\n");
    return false;
  }

  // Find the index to extract from.
  auto VRegAndVal = getIConstantVRegValWithLookThrough(LaneIdxOp.getReg(), MRI);
  if (!VRegAndVal)
    return false;
  unsigned LaneIdx = VRegAndVal->Value.getSExtValue();


  const RegisterBank &DstRB = *RBI.getRegBank(DstReg, MRI, TRI);
  MachineInstr *Extract = emitExtractVectorElt(DstReg, DstRB, NarrowTy, SrcReg,
                                               LaneIdx, MIB);
  if (!Extract)
    return false;

  I.eraseFromParent();
  return true;
}

bool AArch64InstructionSelector::selectSplitVectorUnmerge(
    MachineInstr &I, MachineRegisterInfo &MRI) {
  unsigned NumElts = I.getNumOperands() - 1;
  Register SrcReg = I.getOperand(NumElts).getReg();
  const LLT NarrowTy = MRI.getType(I.getOperand(0).getReg());
  const LLT SrcTy = MRI.getType(SrcReg);

  assert(NarrowTy.isVector() && "Expected an unmerge into vectors");
  if (SrcTy.getSizeInBits() > 128) {
    LLVM_DEBUG(dbgs() << "Unexpected vector type for vec split unmerge");
    return false;
  }

  // We implement a split vector operation by treating the sub-vectors as
  // scalars and extracting them.
  const RegisterBank &DstRB =
      *RBI.getRegBank(I.getOperand(0).getReg(), MRI, TRI);
  for (unsigned OpIdx = 0; OpIdx < NumElts; ++OpIdx) {
    Register Dst = I.getOperand(OpIdx).getReg();
    MachineInstr *Extract =
        emitExtractVectorElt(Dst, DstRB, NarrowTy, SrcReg, OpIdx, MIB);
    if (!Extract)
      return false;
  }
  I.eraseFromParent();
  return true;
}

bool AArch64InstructionSelector::selectUnmergeValues(MachineInstr &I,
                                                     MachineRegisterInfo &MRI) {
  assert(I.getOpcode() == TargetOpcode::G_UNMERGE_VALUES &&
         "unexpected opcode");

  // TODO: Handle unmerging into GPRs and from scalars to scalars.
  if (RBI.getRegBank(I.getOperand(0).getReg(), MRI, TRI)->getID() !=
          AArch64::FPRRegBankID ||
      RBI.getRegBank(I.getOperand(1).getReg(), MRI, TRI)->getID() !=
          AArch64::FPRRegBankID) {
    LLVM_DEBUG(dbgs() << "Unmerging vector-to-gpr and scalar-to-scalar "
                         "currently unsupported.\n");
    return false;
  }

  // The last operand is the vector source register, and every other operand is
  // a register to unpack into.
  unsigned NumElts = I.getNumOperands() - 1;
  Register SrcReg = I.getOperand(NumElts).getReg();
  const LLT NarrowTy = MRI.getType(I.getOperand(0).getReg());
  const LLT WideTy = MRI.getType(SrcReg);
  (void)WideTy;
  assert((WideTy.isVector() || WideTy.getSizeInBits() == 128) &&
         "can only unmerge from vector or s128 types!");
  assert(WideTy.getSizeInBits() > NarrowTy.getSizeInBits() &&
         "source register size too small!");

  if (!NarrowTy.isScalar())
    return selectSplitVectorUnmerge(I, MRI);

  // Choose a lane copy opcode and subregister based off of the size of the
  // vector's elements.
  unsigned CopyOpc = 0;
  unsigned ExtractSubReg = 0;
  if (!getLaneCopyOpcode(CopyOpc, ExtractSubReg, NarrowTy.getSizeInBits()))
    return false;

  // Set up for the lane copies.
  MachineBasicBlock &MBB = *I.getParent();

  // Stores the registers we'll be copying from.
  SmallVector<Register, 4> InsertRegs;

  // We'll use the first register twice, so we only need NumElts-1 registers.
  unsigned NumInsertRegs = NumElts - 1;

  // If our elements fit into exactly 128 bits, then we can copy from the source
  // directly. Otherwise, we need to do a bit of setup with some subregister
  // inserts.
  if (NarrowTy.getSizeInBits() * NumElts == 128) {
    InsertRegs = SmallVector<Register, 4>(NumInsertRegs, SrcReg);
  } else {
    // No. We have to perform subregister inserts. For each insert, create an
    // implicit def and a subregister insert, and save the register we create.
    const TargetRegisterClass *RC = getRegClassForTypeOnBank(
        LLT::fixed_vector(NumElts, WideTy.getScalarSizeInBits()),
        *RBI.getRegBank(SrcReg, MRI, TRI));
    unsigned SubReg = 0;
    bool Found = getSubRegForClass(RC, TRI, SubReg);
    (void)Found;
    assert(Found && "expected to find last operand's subeg idx");
    for (unsigned Idx = 0; Idx < NumInsertRegs; ++Idx) {
      Register ImpDefReg = MRI.createVirtualRegister(&AArch64::FPR128RegClass);
      MachineInstr &ImpDefMI =
          *BuildMI(MBB, I, I.getDebugLoc(), TII.get(TargetOpcode::IMPLICIT_DEF),
                   ImpDefReg);

      // Now, create the subregister insert from SrcReg.
      Register InsertReg = MRI.createVirtualRegister(&AArch64::FPR128RegClass);
      MachineInstr &InsMI =
          *BuildMI(MBB, I, I.getDebugLoc(),
                   TII.get(TargetOpcode::INSERT_SUBREG), InsertReg)
               .addUse(ImpDefReg)
               .addUse(SrcReg)
               .addImm(SubReg);

      constrainSelectedInstRegOperands(ImpDefMI, TII, TRI, RBI);
      constrainSelectedInstRegOperands(InsMI, TII, TRI, RBI);

      // Save the register so that we can copy from it after.
      InsertRegs.push_back(InsertReg);
    }
  }

  // Now that we've created any necessary subregister inserts, we can
  // create the copies.
  //
  // Perform the first copy separately as a subregister copy.
  Register CopyTo = I.getOperand(0).getReg();
  auto FirstCopy = MIB.buildInstr(TargetOpcode::COPY, {CopyTo}, {})
                       .addReg(InsertRegs[0], 0, ExtractSubReg);
  constrainSelectedInstRegOperands(*FirstCopy, TII, TRI, RBI);

  // Now, perform the remaining copies as vector lane copies.
  unsigned LaneIdx = 1;
  for (Register InsReg : InsertRegs) {
    Register CopyTo = I.getOperand(LaneIdx).getReg();
    MachineInstr &CopyInst =
        *BuildMI(MBB, I, I.getDebugLoc(), TII.get(CopyOpc), CopyTo)
             .addUse(InsReg)
             .addImm(LaneIdx);
    constrainSelectedInstRegOperands(CopyInst, TII, TRI, RBI);
    ++LaneIdx;
  }

  // Separately constrain the first copy's destination. Because of the
  // limitation in constrainOperandRegClass, we can't guarantee that this will
  // actually be constrained. So, do it ourselves using the second operand.
  const TargetRegisterClass *RC =
      MRI.getRegClassOrNull(I.getOperand(1).getReg());
  if (!RC) {
    LLVM_DEBUG(dbgs() << "Couldn't constrain copy destination.\n");
    return false;
  }

  RBI.constrainGenericRegister(CopyTo, *RC, MRI);
  I.eraseFromParent();
  return true;
}

bool AArch64InstructionSelector::selectConcatVectors(
    MachineInstr &I, MachineRegisterInfo &MRI)  {
  assert(I.getOpcode() == TargetOpcode::G_CONCAT_VECTORS &&
         "Unexpected opcode");
  Register Dst = I.getOperand(0).getReg();
  Register Op1 = I.getOperand(1).getReg();
  Register Op2 = I.getOperand(2).getReg();
  MachineInstr *ConcatMI = emitVectorConcat(Dst, Op1, Op2, MIB);
  if (!ConcatMI)
    return false;
  I.eraseFromParent();
  return true;
}

unsigned
AArch64InstructionSelector::emitConstantPoolEntry(const Constant *CPVal,
                                                  MachineFunction &MF) const {
  Type *CPTy = CPVal->getType();
  Align Alignment = MF.getDataLayout().getPrefTypeAlign(CPTy);

  MachineConstantPool *MCP = MF.getConstantPool();
  return MCP->getConstantPoolIndex(CPVal, Alignment);
}

MachineInstr *AArch64InstructionSelector::emitLoadFromConstantPool(
    const Constant *CPVal, MachineIRBuilder &MIRBuilder) const {
  auto &MF = MIRBuilder.getMF();
  unsigned CPIdx = emitConstantPoolEntry(CPVal, MF);

  auto Adrp =
      MIRBuilder.buildInstr(AArch64::ADRP, {&AArch64::GPR64RegClass}, {})
          .addConstantPoolIndex(CPIdx, 0, AArch64II::MO_PAGE);

  MachineInstr *LoadMI = nullptr;
  MachinePointerInfo PtrInfo = MachinePointerInfo::getConstantPool(MF);
  unsigned Size = MIRBuilder.getDataLayout().getTypeStoreSize(CPVal->getType());
  switch (Size) {
  case 16:
    LoadMI =
        &*MIRBuilder
              .buildInstr(AArch64::LDRQui, {&AArch64::FPR128RegClass}, {Adrp})
              .addConstantPoolIndex(CPIdx, 0,
                                    AArch64II::MO_PAGEOFF | AArch64II::MO_NC);
    break;
  case 8:
    LoadMI =
        &*MIRBuilder
              .buildInstr(AArch64::LDRDui, {&AArch64::FPR64RegClass}, {Adrp})
              .addConstantPoolIndex(CPIdx, 0,
                                    AArch64II::MO_PAGEOFF | AArch64II::MO_NC);
    break;
  case 4:
    LoadMI =
        &*MIRBuilder
              .buildInstr(AArch64::LDRSui, {&AArch64::FPR32RegClass}, {Adrp})
              .addConstantPoolIndex(CPIdx, 0,
                                    AArch64II::MO_PAGEOFF | AArch64II::MO_NC);
    break;
  case 2:
    LoadMI =
        &*MIRBuilder
              .buildInstr(AArch64::LDRHui, {&AArch64::FPR16RegClass}, {Adrp})
              .addConstantPoolIndex(CPIdx, 0,
                                    AArch64II::MO_PAGEOFF | AArch64II::MO_NC);
    break;
  default:
    LLVM_DEBUG(dbgs() << "Could not load from constant pool of type "
                      << *CPVal->getType());
    return nullptr;
  }
  LoadMI->addMemOperand(MF, MF.getMachineMemOperand(PtrInfo,
                                                    MachineMemOperand::MOLoad,
                                                    Size, Align(Size)));
  constrainSelectedInstRegOperands(*Adrp, TII, TRI, RBI);
  constrainSelectedInstRegOperands(*LoadMI, TII, TRI, RBI);
  return LoadMI;
}

/// Return an <Opcode, SubregIndex> pair to do an vector elt insert of a given
/// size and RB.
static std::pair<unsigned, unsigned>
getInsertVecEltOpInfo(const RegisterBank &RB, unsigned EltSize) {
  unsigned Opc, SubregIdx;
  if (RB.getID() == AArch64::GPRRegBankID) {
    if (EltSize == 16) {
      Opc = AArch64::INSvi16gpr;
      SubregIdx = AArch64::ssub;
    } else if (EltSize == 32) {
      Opc = AArch64::INSvi32gpr;
      SubregIdx = AArch64::ssub;
    } else if (EltSize == 64) {
      Opc = AArch64::INSvi64gpr;
      SubregIdx = AArch64::dsub;
    } else {
      llvm_unreachable("invalid elt size!");
    }
  } else {
    if (EltSize == 8) {
      Opc = AArch64::INSvi8lane;
      SubregIdx = AArch64::bsub;
    } else if (EltSize == 16) {
      Opc = AArch64::INSvi16lane;
      SubregIdx = AArch64::hsub;
    } else if (EltSize == 32) {
      Opc = AArch64::INSvi32lane;
      SubregIdx = AArch64::ssub;
    } else if (EltSize == 64) {
      Opc = AArch64::INSvi64lane;
      SubregIdx = AArch64::dsub;
    } else {
      llvm_unreachable("invalid elt size!");
    }
  }
  return std::make_pair(Opc, SubregIdx);
}

MachineInstr *AArch64InstructionSelector::emitInstr(
    unsigned Opcode, std::initializer_list<llvm::DstOp> DstOps,
    std::initializer_list<llvm::SrcOp> SrcOps, MachineIRBuilder &MIRBuilder,
    const ComplexRendererFns &RenderFns) const {
  assert(Opcode && "Expected an opcode?");
  assert(!isPreISelGenericOpcode(Opcode) &&
         "Function should only be used to produce selected instructions!");
  auto MI = MIRBuilder.buildInstr(Opcode, DstOps, SrcOps);
  if (RenderFns)
    for (auto &Fn : *RenderFns)
      Fn(MI);
  constrainSelectedInstRegOperands(*MI, TII, TRI, RBI);
  return &*MI;
}

MachineInstr *AArch64InstructionSelector::emitAddSub(
    const std::array<std::array<unsigned, 2>, 5> &AddrModeAndSizeToOpcode,
    Register Dst, MachineOperand &LHS, MachineOperand &RHS,
    MachineIRBuilder &MIRBuilder) const {
  MachineRegisterInfo &MRI = MIRBuilder.getMF().getRegInfo();
  assert(LHS.isReg() && RHS.isReg() && "Expected register operands?");
  auto Ty = MRI.getType(LHS.getReg());
  assert(!Ty.isVector() && "Expected a scalar or pointer?");
  unsigned Size = Ty.getSizeInBits();
  assert((Size == 32 || Size == 64) && "Expected a 32-bit or 64-bit type only");
  bool Is32Bit = Size == 32;

  // INSTRri form with positive arithmetic immediate.
  if (auto Fns = selectArithImmed(RHS))
    return emitInstr(AddrModeAndSizeToOpcode[0][Is32Bit], {Dst}, {LHS},
                     MIRBuilder, Fns);

  // INSTRri form with negative arithmetic immediate.
  if (auto Fns = selectNegArithImmed(RHS))
    return emitInstr(AddrModeAndSizeToOpcode[3][Is32Bit], {Dst}, {LHS},
                     MIRBuilder, Fns);

  // INSTRrx form.
  if (auto Fns = selectArithExtendedRegister(RHS))
    return emitInstr(AddrModeAndSizeToOpcode[4][Is32Bit], {Dst}, {LHS},
                     MIRBuilder, Fns);

  // INSTRrs form.
  if (auto Fns = selectShiftedRegister(RHS))
    return emitInstr(AddrModeAndSizeToOpcode[1][Is32Bit], {Dst}, {LHS},
                     MIRBuilder, Fns);
  return emitInstr(AddrModeAndSizeToOpcode[2][Is32Bit], {Dst}, {LHS, RHS},
                   MIRBuilder);
}

MachineInstr *
AArch64InstructionSelector::emitADD(Register DefReg, MachineOperand &LHS,
                                    MachineOperand &RHS,
                                    MachineIRBuilder &MIRBuilder) const {
  const std::array<std::array<unsigned, 2>, 5> OpcTable{
      {{AArch64::ADDXri, AArch64::ADDWri},
       {AArch64::ADDXrs, AArch64::ADDWrs},
       {AArch64::ADDXrr, AArch64::ADDWrr},
       {AArch64::SUBXri, AArch64::SUBWri},
       {AArch64::ADDXrx, AArch64::ADDWrx}}};
  return emitAddSub(OpcTable, DefReg, LHS, RHS, MIRBuilder);
}

MachineInstr *
AArch64InstructionSelector::emitADDS(Register Dst, MachineOperand &LHS,
                                     MachineOperand &RHS,
                                     MachineIRBuilder &MIRBuilder) const {
  const std::array<std::array<unsigned, 2>, 5> OpcTable{
      {{AArch64::ADDSXri, AArch64::ADDSWri},
       {AArch64::ADDSXrs, AArch64::ADDSWrs},
       {AArch64::ADDSXrr, AArch64::ADDSWrr},
       {AArch64::SUBSXri, AArch64::SUBSWri},
       {AArch64::ADDSXrx, AArch64::ADDSWrx}}};
  return emitAddSub(OpcTable, Dst, LHS, RHS, MIRBuilder);
}

MachineInstr *
AArch64InstructionSelector::emitSUBS(Register Dst, MachineOperand &LHS,
                                     MachineOperand &RHS,
                                     MachineIRBuilder &MIRBuilder) const {
  const std::array<std::array<unsigned, 2>, 5> OpcTable{
      {{AArch64::SUBSXri, AArch64::SUBSWri},
       {AArch64::SUBSXrs, AArch64::SUBSWrs},
       {AArch64::SUBSXrr, AArch64::SUBSWrr},
       {AArch64::ADDSXri, AArch64::ADDSWri},
       {AArch64::SUBSXrx, AArch64::SUBSWrx}}};
  return emitAddSub(OpcTable, Dst, LHS, RHS, MIRBuilder);
}

MachineInstr *
AArch64InstructionSelector::emitCMN(MachineOperand &LHS, MachineOperand &RHS,
                                    MachineIRBuilder &MIRBuilder) const {
  MachineRegisterInfo &MRI = MIRBuilder.getMF().getRegInfo();
  bool Is32Bit = (MRI.getType(LHS.getReg()).getSizeInBits() == 32);
  auto RC = Is32Bit ? &AArch64::GPR32RegClass : &AArch64::GPR64RegClass;
  return emitADDS(MRI.createVirtualRegister(RC), LHS, RHS, MIRBuilder);
}

MachineInstr *
AArch64InstructionSelector::emitTST(MachineOperand &LHS, MachineOperand &RHS,
                                    MachineIRBuilder &MIRBuilder) const {
  assert(LHS.isReg() && RHS.isReg() && "Expected register operands?");
  MachineRegisterInfo &MRI = MIRBuilder.getMF().getRegInfo();
  LLT Ty = MRI.getType(LHS.getReg());
  unsigned RegSize = Ty.getSizeInBits();
  bool Is32Bit = (RegSize == 32);
  const unsigned OpcTable[3][2] = {{AArch64::ANDSXri, AArch64::ANDSWri},
                                   {AArch64::ANDSXrs, AArch64::ANDSWrs},
                                   {AArch64::ANDSXrr, AArch64::ANDSWrr}};
  // ANDS needs a logical immediate for its immediate form. Check if we can
  // fold one in.
  if (auto ValAndVReg = getIConstantVRegValWithLookThrough(RHS.getReg(), MRI)) {
    int64_t Imm = ValAndVReg->Value.getSExtValue();

    if (AArch64_AM::isLogicalImmediate(Imm, RegSize)) {
      auto TstMI = MIRBuilder.buildInstr(OpcTable[0][Is32Bit], {Ty}, {LHS});
      TstMI.addImm(AArch64_AM::encodeLogicalImmediate(Imm, RegSize));
      constrainSelectedInstRegOperands(*TstMI, TII, TRI, RBI);
      return &*TstMI;
    }
  }

  if (auto Fns = selectLogicalShiftedRegister(RHS))
    return emitInstr(OpcTable[1][Is32Bit], {Ty}, {LHS}, MIRBuilder, Fns);
  return emitInstr(OpcTable[2][Is32Bit], {Ty}, {LHS, RHS}, MIRBuilder);
}

MachineInstr *AArch64InstructionSelector::emitIntegerCompare(
    MachineOperand &LHS, MachineOperand &RHS, MachineOperand &Predicate,
    MachineIRBuilder &MIRBuilder) const {
  assert(LHS.isReg() && RHS.isReg() && "Expected LHS and RHS to be registers!");
  assert(Predicate.isPredicate() && "Expected predicate?");
  MachineRegisterInfo &MRI = MIRBuilder.getMF().getRegInfo();
  LLT CmpTy = MRI.getType(LHS.getReg());
  assert(!CmpTy.isVector() && "Expected scalar or pointer");
  unsigned Size = CmpTy.getSizeInBits();
  (void)Size;
  assert((Size == 32 || Size == 64) && "Expected a 32-bit or 64-bit LHS/RHS?");
  // Fold the compare into a cmn or tst if possible.
  if (auto FoldCmp = tryFoldIntegerCompare(LHS, RHS, Predicate, MIRBuilder))
    return FoldCmp;
  auto Dst = MRI.cloneVirtualRegister(LHS.getReg());
  return emitSUBS(Dst, LHS, RHS, MIRBuilder);
}

MachineInstr *AArch64InstructionSelector::emitCSetForFCmp(
    Register Dst, CmpInst::Predicate Pred, MachineIRBuilder &MIRBuilder) const {
  MachineRegisterInfo &MRI = *MIRBuilder.getMRI();
#ifndef NDEBUG
  LLT Ty = MRI.getType(Dst);
  assert(!Ty.isVector() && Ty.getSizeInBits() == 32 &&
         "Expected a 32-bit scalar register?");
#endif
  const Register ZReg = AArch64::WZR;
  AArch64CC::CondCode CC1, CC2;
  changeFCMPPredToAArch64CC(Pred, CC1, CC2);
  auto InvCC1 = AArch64CC::getInvertedCondCode(CC1);
  if (CC2 == AArch64CC::AL)
    return emitCSINC(/*Dst=*/Dst, /*Src1=*/ZReg, /*Src2=*/ZReg, InvCC1,
                     MIRBuilder);
  const TargetRegisterClass *RC = &AArch64::GPR32RegClass;
  Register Def1Reg = MRI.createVirtualRegister(RC);
  Register Def2Reg = MRI.createVirtualRegister(RC);
  auto InvCC2 = AArch64CC::getInvertedCondCode(CC2);
  emitCSINC(/*Dst=*/Def1Reg, /*Src1=*/ZReg, /*Src2=*/ZReg, InvCC1, MIRBuilder);
  emitCSINC(/*Dst=*/Def2Reg, /*Src1=*/ZReg, /*Src2=*/ZReg, InvCC2, MIRBuilder);
  auto OrMI = MIRBuilder.buildInstr(AArch64::ORRWrr, {Dst}, {Def1Reg, Def2Reg});
  constrainSelectedInstRegOperands(*OrMI, TII, TRI, RBI);
  return &*OrMI;
}

MachineInstr *AArch64InstructionSelector::emitFPCompare(
    Register LHS, Register RHS, MachineIRBuilder &MIRBuilder,
    std::optional<CmpInst::Predicate> Pred) const {
  MachineRegisterInfo &MRI = *MIRBuilder.getMRI();
  LLT Ty = MRI.getType(LHS);
  if (Ty.isVector())
    return nullptr;
  unsigned OpSize = Ty.getSizeInBits();
  if (OpSize != 32 && OpSize != 64)
    return nullptr;

  // If this is a compare against +0.0, then we don't have
  // to explicitly materialize a constant.
  const ConstantFP *FPImm = getConstantFPVRegVal(RHS, MRI);
  bool ShouldUseImm = FPImm && (FPImm->isZero() && !FPImm->isNegative());

  auto IsEqualityPred = [](CmpInst::Predicate P) {
    return P == CmpInst::FCMP_OEQ || P == CmpInst::FCMP_ONE ||
           P == CmpInst::FCMP_UEQ || P == CmpInst::FCMP_UNE;
  };
  if (!ShouldUseImm && Pred && IsEqualityPred(*Pred)) {
    // Try commutating the operands.
    const ConstantFP *LHSImm = getConstantFPVRegVal(LHS, MRI);
    if (LHSImm && (LHSImm->isZero() && !LHSImm->isNegative())) {
      ShouldUseImm = true;
      std::swap(LHS, RHS);
    }
  }
  unsigned CmpOpcTbl[2][2] = {{AArch64::FCMPSrr, AArch64::FCMPDrr},
                              {AArch64::FCMPSri, AArch64::FCMPDri}};
  unsigned CmpOpc = CmpOpcTbl[ShouldUseImm][OpSize == 64];

  // Partially build the compare. Decide if we need to add a use for the
  // third operand based off whether or not we're comparing against 0.0.
  auto CmpMI = MIRBuilder.buildInstr(CmpOpc).addUse(LHS);
  CmpMI.setMIFlags(MachineInstr::NoFPExcept);
  if (!ShouldUseImm)
    CmpMI.addUse(RHS);
  constrainSelectedInstRegOperands(*CmpMI, TII, TRI, RBI);
  return &*CmpMI;
}

MachineInstr *AArch64InstructionSelector::emitVectorConcat(
    std::optional<Register> Dst, Register Op1, Register Op2,
    MachineIRBuilder &MIRBuilder) const {
  // We implement a vector concat by:
  // 1. Use scalar_to_vector to insert the lower vector into the larger dest
  // 2. Insert the upper vector into the destination's upper element
  // TODO: some of this code is common with G_BUILD_VECTOR handling.
  MachineRegisterInfo &MRI = MIRBuilder.getMF().getRegInfo();

  const LLT Op1Ty = MRI.getType(Op1);
  const LLT Op2Ty = MRI.getType(Op2);

  if (Op1Ty != Op2Ty) {
    LLVM_DEBUG(dbgs() << "Could not do vector concat of differing vector tys");
    return nullptr;
  }
  assert(Op1Ty.isVector() && "Expected a vector for vector concat");

  if (Op1Ty.getSizeInBits() >= 128) {
    LLVM_DEBUG(dbgs() << "Vector concat not supported for full size vectors");
    return nullptr;
  }

  // At the moment we just support 64 bit vector concats.
  if (Op1Ty.getSizeInBits() != 64) {
    LLVM_DEBUG(dbgs() << "Vector concat supported for 64b vectors");
    return nullptr;
  }

  const LLT ScalarTy = LLT::scalar(Op1Ty.getSizeInBits());
  const RegisterBank &FPRBank = *RBI.getRegBank(Op1, MRI, TRI);
  const TargetRegisterClass *DstRC =
      getRegClassForTypeOnBank(Op1Ty.multiplyElements(2), FPRBank);

  MachineInstr *WidenedOp1 =
      emitScalarToVector(ScalarTy.getSizeInBits(), DstRC, Op1, MIRBuilder);
  MachineInstr *WidenedOp2 =
      emitScalarToVector(ScalarTy.getSizeInBits(), DstRC, Op2, MIRBuilder);
  if (!WidenedOp1 || !WidenedOp2) {
    LLVM_DEBUG(dbgs() << "Could not emit a vector from scalar value");
    return nullptr;
  }

  // Now do the insert of the upper element.
  unsigned InsertOpc, InsSubRegIdx;
  std::tie(InsertOpc, InsSubRegIdx) =
      getInsertVecEltOpInfo(FPRBank, ScalarTy.getSizeInBits());

  if (!Dst)
    Dst = MRI.createVirtualRegister(DstRC);
  auto InsElt =
      MIRBuilder
          .buildInstr(InsertOpc, {*Dst}, {WidenedOp1->getOperand(0).getReg()})
          .addImm(1) /* Lane index */
          .addUse(WidenedOp2->getOperand(0).getReg())
          .addImm(0);
  constrainSelectedInstRegOperands(*InsElt, TII, TRI, RBI);
  return &*InsElt;
}

MachineInstr *
AArch64InstructionSelector::emitCSINC(Register Dst, Register Src1,
                                      Register Src2, AArch64CC::CondCode Pred,
                                      MachineIRBuilder &MIRBuilder) const {
  auto &MRI = *MIRBuilder.getMRI();
  const RegClassOrRegBank &RegClassOrBank = MRI.getRegClassOrRegBank(Dst);
  // If we used a register class, then this won't necessarily have an LLT.
  // Compute the size based off whether or not we have a class or bank.
  unsigned Size;
  if (const auto *RC = RegClassOrBank.dyn_cast<const TargetRegisterClass *>())
    Size = TRI.getRegSizeInBits(*RC);
  else
    Size = MRI.getType(Dst).getSizeInBits();
  // Some opcodes use s1.
  assert(Size <= 64 && "Expected 64 bits or less only!");
  static const unsigned OpcTable[2] = {AArch64::CSINCWr, AArch64::CSINCXr};
  unsigned Opc = OpcTable[Size == 64];
  auto CSINC = MIRBuilder.buildInstr(Opc, {Dst}, {Src1, Src2}).addImm(Pred);
  constrainSelectedInstRegOperands(*CSINC, TII, TRI, RBI);
  return &*CSINC;
}

std::pair<MachineInstr *, AArch64CC::CondCode>
AArch64InstructionSelector::emitOverflowOp(unsigned Opcode, Register Dst,
                                           MachineOperand &LHS,
                                           MachineOperand &RHS,
                                           MachineIRBuilder &MIRBuilder) const {
  switch (Opcode) {
  default:
    llvm_unreachable("Unexpected opcode!");
  case TargetOpcode::G_SADDO:
    return std::make_pair(emitADDS(Dst, LHS, RHS, MIRBuilder), AArch64CC::VS);
  case TargetOpcode::G_UADDO:
    return std::make_pair(emitADDS(Dst, LHS, RHS, MIRBuilder), AArch64CC::HS);
  case TargetOpcode::G_SSUBO:
    return std::make_pair(emitSUBS(Dst, LHS, RHS, MIRBuilder), AArch64CC::VS);
  case TargetOpcode::G_USUBO:
    return std::make_pair(emitSUBS(Dst, LHS, RHS, MIRBuilder), AArch64CC::LO);
  }
}

/// Returns true if @p Val is a tree of AND/OR/CMP operations that can be
/// expressed as a conjunction.
/// \param CanNegate    Set to true if we can negate the whole sub-tree just by
///                     changing the conditions on the CMP tests.
///                     (this means we can call emitConjunctionRec() with
///                      Negate==true on this sub-tree)
/// \param MustBeFirst  Set to true if this subtree needs to be negated and we
///                     cannot do the negation naturally. We are required to
///                     emit the subtree first in this case.
/// \param WillNegate   Is true if are called when the result of this
///                     subexpression must be negated. This happens when the
///                     outer expression is an OR. We can use this fact to know
///                     that we have a double negation (or (or ...) ...) that
///                     can be implemented for free.
static bool canEmitConjunction(Register Val, bool &CanNegate, bool &MustBeFirst,
                               bool WillNegate, MachineRegisterInfo &MRI,
                               unsigned Depth = 0) {
  if (!MRI.hasOneNonDBGUse(Val))
    return false;
  MachineInstr *ValDef = MRI.getVRegDef(Val);
  unsigned Opcode = ValDef->getOpcode();
  if (isa<GAnyCmp>(ValDef)) {
    CanNegate = true;
    MustBeFirst = false;
    return true;
  }
  // Protect against exponential runtime and stack overflow.
  if (Depth > 6)
    return false;
  if (Opcode == TargetOpcode::G_AND || Opcode == TargetOpcode::G_OR) {
    bool IsOR = Opcode == TargetOpcode::G_OR;
    Register O0 = ValDef->getOperand(1).getReg();
    Register O1 = ValDef->getOperand(2).getReg();
    bool CanNegateL;
    bool MustBeFirstL;
    if (!canEmitConjunction(O0, CanNegateL, MustBeFirstL, IsOR, MRI, Depth + 1))
      return false;
    bool CanNegateR;
    bool MustBeFirstR;
    if (!canEmitConjunction(O1, CanNegateR, MustBeFirstR, IsOR, MRI, Depth + 1))
      return false;

    if (MustBeFirstL && MustBeFirstR)
      return false;

    if (IsOR) {
      // For an OR expression we need to be able to naturally negate at least
      // one side or we cannot do the transformation at all.
      if (!CanNegateL && !CanNegateR)
        return false;
      // If we the result of the OR will be negated and we can naturally negate
      // the leaves, then this sub-tree as a whole negates naturally.
      CanNegate = WillNegate && CanNegateL && CanNegateR;
      // If we cannot naturally negate the whole sub-tree, then this must be
      // emitted first.
      MustBeFirst = !CanNegate;
    } else {
      assert(Opcode == TargetOpcode::G_AND && "Must be G_AND");
      // We cannot naturally negate an AND operation.
      CanNegate = false;
      MustBeFirst = MustBeFirstL || MustBeFirstR;
    }
    return true;
  }
  return false;
}

MachineInstr *AArch64InstructionSelector::emitConditionalComparison(
    Register LHS, Register RHS, CmpInst::Predicate CC,
    AArch64CC::CondCode Predicate, AArch64CC::CondCode OutCC,
    MachineIRBuilder &MIB) const {
  // TODO: emit CMN as an optimization.
  auto &MRI = *MIB.getMRI();
  LLT OpTy = MRI.getType(LHS);
  assert(OpTy.getSizeInBits() == 32 || OpTy.getSizeInBits() == 64);
  unsigned CCmpOpc;
  std::optional<ValueAndVReg> C;
  if (CmpInst::isIntPredicate(CC)) {
    C = getIConstantVRegValWithLookThrough(RHS, MRI);
    if (C && C->Value.ult(32))
      CCmpOpc = OpTy.getSizeInBits() == 32 ? AArch64::CCMPWi : AArch64::CCMPXi;
    else
      CCmpOpc = OpTy.getSizeInBits() == 32 ? AArch64::CCMPWr : AArch64::CCMPXr;
  } else {
    switch (OpTy.getSizeInBits()) {
    case 16:
      CCmpOpc = AArch64::FCCMPHrr;
      break;
    case 32:
      CCmpOpc = AArch64::FCCMPSrr;
      break;
    case 64:
      CCmpOpc = AArch64::FCCMPDrr;
      break;
    default:
      return nullptr;
    }
  }
  AArch64CC::CondCode InvOutCC = AArch64CC::getInvertedCondCode(OutCC);
  unsigned NZCV = AArch64CC::getNZCVToSatisfyCondCode(InvOutCC);
  auto CCmp =
      MIB.buildInstr(CCmpOpc, {}, {LHS});
  if (CCmpOpc == AArch64::CCMPWi || CCmpOpc == AArch64::CCMPXi)
    CCmp.addImm(C->Value.getZExtValue());
  else
    CCmp.addReg(RHS);
  CCmp.addImm(NZCV).addImm(Predicate);
  constrainSelectedInstRegOperands(*CCmp, TII, TRI, RBI);
  return &*CCmp;
}

MachineInstr *AArch64InstructionSelector::emitConjunctionRec(
    Register Val, AArch64CC::CondCode &OutCC, bool Negate, Register CCOp,
    AArch64CC::CondCode Predicate, MachineIRBuilder &MIB) const {
  // We're at a tree leaf, produce a conditional comparison operation.
  auto &MRI = *MIB.getMRI();
  MachineInstr *ValDef = MRI.getVRegDef(Val);
  unsigned Opcode = ValDef->getOpcode();
  if (auto *Cmp = dyn_cast<GAnyCmp>(ValDef)) {
    Register LHS = Cmp->getLHSReg();
    Register RHS = Cmp->getRHSReg();
    CmpInst::Predicate CC = Cmp->getCond();
    if (Negate)
      CC = CmpInst::getInversePredicate(CC);
    if (isa<GICmp>(Cmp)) {
      OutCC = changeICMPPredToAArch64CC(CC);
    } else {
      // Handle special FP cases.
      AArch64CC::CondCode ExtraCC;
      changeFPCCToANDAArch64CC(CC, OutCC, ExtraCC);
      // Some floating point conditions can't be tested with a single condition
      // code. Construct an additional comparison in this case.
      if (ExtraCC != AArch64CC::AL) {
        MachineInstr *ExtraCmp;
        if (!CCOp)
          ExtraCmp = emitFPCompare(LHS, RHS, MIB, CC);
        else
          ExtraCmp =
              emitConditionalComparison(LHS, RHS, CC, Predicate, ExtraCC, MIB);
        CCOp = ExtraCmp->getOperand(0).getReg();
        Predicate = ExtraCC;
      }
    }

    // Produce a normal comparison if we are first in the chain
    if (!CCOp) {
      auto Dst = MRI.cloneVirtualRegister(LHS);
      if (isa<GICmp>(Cmp))
        return emitSUBS(Dst, Cmp->getOperand(2), Cmp->getOperand(3), MIB);
      return emitFPCompare(Cmp->getOperand(2).getReg(),
                           Cmp->getOperand(3).getReg(), MIB);
    }
    // Otherwise produce a ccmp.
    return emitConditionalComparison(LHS, RHS, CC, Predicate, OutCC, MIB);
  }
  assert(MRI.hasOneNonDBGUse(Val) && "Valid conjunction/disjunction tree");

  bool IsOR = Opcode == TargetOpcode::G_OR;

  Register LHS = ValDef->getOperand(1).getReg();
  bool CanNegateL;
  bool MustBeFirstL;
  bool ValidL = canEmitConjunction(LHS, CanNegateL, MustBeFirstL, IsOR, MRI);
  assert(ValidL && "Valid conjunction/disjunction tree");
  (void)ValidL;

  Register RHS = ValDef->getOperand(2).getReg();
  bool CanNegateR;
  bool MustBeFirstR;
  bool ValidR = canEmitConjunction(RHS, CanNegateR, MustBeFirstR, IsOR, MRI);
  assert(ValidR && "Valid conjunction/disjunction tree");
  (void)ValidR;

  // Swap sub-tree that must come first to the right side.
  if (MustBeFirstL) {
    assert(!MustBeFirstR && "Valid conjunction/disjunction tree");
    std::swap(LHS, RHS);
    std::swap(CanNegateL, CanNegateR);
    std::swap(MustBeFirstL, MustBeFirstR);
  }

  bool NegateR;
  bool NegateAfterR;
  bool NegateL;
  bool NegateAfterAll;
  if (Opcode == TargetOpcode::G_OR) {
    // Swap the sub-tree that we can negate naturally to the left.
    if (!CanNegateL) {
      assert(CanNegateR && "at least one side must be negatable");
      assert(!MustBeFirstR && "invalid conjunction/disjunction tree");
      assert(!Negate);
      std::swap(LHS, RHS);
      NegateR = false;
      NegateAfterR = true;
    } else {
      // Negate the left sub-tree if possible, otherwise negate the result.
      NegateR = CanNegateR;
      NegateAfterR = !CanNegateR;
    }
    NegateL = true;
    NegateAfterAll = !Negate;
  } else {
    assert(Opcode == TargetOpcode::G_AND &&
           "Valid conjunction/disjunction tree");
    assert(!Negate && "Valid conjunction/disjunction tree");

    NegateL = false;
    NegateR = false;
    NegateAfterR = false;
    NegateAfterAll = false;
  }

  // Emit sub-trees.
  AArch64CC::CondCode RHSCC;
  MachineInstr *CmpR =
      emitConjunctionRec(RHS, RHSCC, NegateR, CCOp, Predicate, MIB);
  if (NegateAfterR)
    RHSCC = AArch64CC::getInvertedCondCode(RHSCC);
  MachineInstr *CmpL = emitConjunctionRec(
      LHS, OutCC, NegateL, CmpR->getOperand(0).getReg(), RHSCC, MIB);
  if (NegateAfterAll)
    OutCC = AArch64CC::getInvertedCondCode(OutCC);
  return CmpL;
}

MachineInstr *AArch64InstructionSelector::emitConjunction(
    Register Val, AArch64CC::CondCode &OutCC, MachineIRBuilder &MIB) const {
  bool DummyCanNegate;
  bool DummyMustBeFirst;
  if (!canEmitConjunction(Val, DummyCanNegate, DummyMustBeFirst, false,
                          *MIB.getMRI()))
    return nullptr;
  return emitConjunctionRec(Val, OutCC, false, Register(), AArch64CC::AL, MIB);
}

bool AArch64InstructionSelector::tryOptSelectConjunction(GSelect &SelI,
                                                         MachineInstr &CondMI) {
  AArch64CC::CondCode AArch64CC;
  MachineInstr *ConjMI = emitConjunction(SelI.getCondReg(), AArch64CC, MIB);
  if (!ConjMI)
    return false;

  emitSelect(SelI.getReg(0), SelI.getTrueReg(), SelI.getFalseReg(), AArch64CC, MIB);
  SelI.eraseFromParent();
  return true;
}

bool AArch64InstructionSelector::tryOptSelect(GSelect &I) {
  MachineRegisterInfo &MRI = *MIB.getMRI();
  // We want to recognize this pattern:
  //
  // $z = G_FCMP pred, $x, $y
  // ...
  // $w = G_SELECT $z, $a, $b
  //
  // Where the value of $z is *only* ever used by the G_SELECT (possibly with
  // some copies/truncs in between.)
  //
  // If we see this, then we can emit something like this:
  //
  // fcmp $x, $y
  // fcsel $w, $a, $b, pred
  //
  // Rather than emitting both of the rather long sequences in the standard
  // G_FCMP/G_SELECT select methods.

  // First, check if the condition is defined by a compare.
  MachineInstr *CondDef = MRI.getVRegDef(I.getOperand(1).getReg());

  // We can only fold if all of the defs have one use.
  Register CondDefReg = CondDef->getOperand(0).getReg();
  if (!MRI.hasOneNonDBGUse(CondDefReg)) {
    // Unless it's another select.
    for (const MachineInstr &UI : MRI.use_nodbg_instructions(CondDefReg)) {
      if (CondDef == &UI)
        continue;
      if (UI.getOpcode() != TargetOpcode::G_SELECT)
        return false;
    }
  }

  // Is the condition defined by a compare?
  unsigned CondOpc = CondDef->getOpcode();
  if (CondOpc != TargetOpcode::G_ICMP && CondOpc != TargetOpcode::G_FCMP) {
    if (tryOptSelectConjunction(I, *CondDef))
      return true;
    return false;
  }

  AArch64CC::CondCode CondCode;
  if (CondOpc == TargetOpcode::G_ICMP) {
    auto Pred =
        static_cast<CmpInst::Predicate>(CondDef->getOperand(1).getPredicate());
    CondCode = changeICMPPredToAArch64CC(Pred);
    emitIntegerCompare(CondDef->getOperand(2), CondDef->getOperand(3),
                       CondDef->getOperand(1), MIB);
  } else {
    // Get the condition code for the select.
    auto Pred =
        static_cast<CmpInst::Predicate>(CondDef->getOperand(1).getPredicate());
    AArch64CC::CondCode CondCode2;
    changeFCMPPredToAArch64CC(Pred, CondCode, CondCode2);

    // changeFCMPPredToAArch64CC sets CondCode2 to AL when we require two
    // instructions to emit the comparison.
    // TODO: Handle FCMP_UEQ and FCMP_ONE. After that, this check will be
    // unnecessary.
    if (CondCode2 != AArch64CC::AL)
      return false;

    if (!emitFPCompare(CondDef->getOperand(2).getReg(),
                       CondDef->getOperand(3).getReg(), MIB)) {
      LLVM_DEBUG(dbgs() << "Couldn't emit compare for select!\n");
      return false;
    }
  }

  // Emit the select.
  emitSelect(I.getOperand(0).getReg(), I.getOperand(2).getReg(),
             I.getOperand(3).getReg(), CondCode, MIB);
  I.eraseFromParent();
  return true;
}

MachineInstr *AArch64InstructionSelector::tryFoldIntegerCompare(
    MachineOperand &LHS, MachineOperand &RHS, MachineOperand &Predicate,
    MachineIRBuilder &MIRBuilder) const {
  assert(LHS.isReg() && RHS.isReg() && Predicate.isPredicate() &&
         "Unexpected MachineOperand");
  MachineRegisterInfo &MRI = *MIRBuilder.getMRI();
  // We want to find this sort of thing:
  // x = G_SUB 0, y
  // G_ICMP z, x
  //
  // In this case, we can fold the G_SUB into the G_ICMP using a CMN instead.
  // e.g:
  //
  // cmn z, y

  // Check if the RHS or LHS of the G_ICMP is defined by a SUB
  MachineInstr *LHSDef = getDefIgnoringCopies(LHS.getReg(), MRI);
  MachineInstr *RHSDef = getDefIgnoringCopies(RHS.getReg(), MRI);
  auto P = static_cast<CmpInst::Predicate>(Predicate.getPredicate());
  // Given this:
  //
  // x = G_SUB 0, y
  // G_ICMP x, z
  //
  // Produce this:
  //
  // cmn y, z
  if (isCMN(LHSDef, P, MRI))
    return emitCMN(LHSDef->getOperand(2), RHS, MIRBuilder);

  // Same idea here, but with the RHS of the compare instead:
  //
  // Given this:
  //
  // x = G_SUB 0, y
  // G_ICMP z, x
  //
  // Produce this:
  //
  // cmn z, y
  if (isCMN(RHSDef, P, MRI))
    return emitCMN(LHS, RHSDef->getOperand(2), MIRBuilder);

  // Given this:
  //
  // z = G_AND x, y
  // G_ICMP z, 0
  //
  // Produce this if the compare is signed:
  //
  // tst x, y
  if (!CmpInst::isUnsigned(P) && LHSDef &&
      LHSDef->getOpcode() == TargetOpcode::G_AND) {
    // Make sure that the RHS is 0.
    auto ValAndVReg = getIConstantVRegValWithLookThrough(RHS.getReg(), MRI);
    if (!ValAndVReg || ValAndVReg->Value != 0)
      return nullptr;

    return emitTST(LHSDef->getOperand(1),
                   LHSDef->getOperand(2), MIRBuilder);
  }

  return nullptr;
}

bool AArch64InstructionSelector::selectShuffleVector(
    MachineInstr &I, MachineRegisterInfo &MRI) {
  const LLT DstTy = MRI.getType(I.getOperand(0).getReg());
  Register Src1Reg = I.getOperand(1).getReg();
  const LLT Src1Ty = MRI.getType(Src1Reg);
  Register Src2Reg = I.getOperand(2).getReg();
  const LLT Src2Ty = MRI.getType(Src2Reg);
  ArrayRef<int> Mask = I.getOperand(3).getShuffleMask();

  MachineBasicBlock &MBB = *I.getParent();
  MachineFunction &MF = *MBB.getParent();
  LLVMContext &Ctx = MF.getFunction().getContext();

  // G_SHUFFLE_VECTOR is weird in that the source operands can be scalars, if
  // it's originated from a <1 x T> type. Those should have been lowered into
  // G_BUILD_VECTOR earlier.
  if (!Src1Ty.isVector() || !Src2Ty.isVector()) {
    LLVM_DEBUG(dbgs() << "Could not select a \"scalar\" G_SHUFFLE_VECTOR\n");
    return false;
  }

  unsigned BytesPerElt = DstTy.getElementType().getSizeInBits() / 8;

  SmallVector<Constant *, 64> CstIdxs;
  for (int Val : Mask) {
    // For now, any undef indexes we'll just assume to be 0. This should be
    // optimized in future, e.g. to select DUP etc.
    Val = Val < 0 ? 0 : Val;
    for (unsigned Byte = 0; Byte < BytesPerElt; ++Byte) {
      unsigned Offset = Byte + Val * BytesPerElt;
      CstIdxs.emplace_back(ConstantInt::get(Type::getInt8Ty(Ctx), Offset));
    }
  }

  // Use a constant pool to load the index vector for TBL.
  Constant *CPVal = ConstantVector::get(CstIdxs);
  MachineInstr *IndexLoad = emitLoadFromConstantPool(CPVal, MIB);
  if (!IndexLoad) {
    LLVM_DEBUG(dbgs() << "Could not load from a constant pool");
    return false;
  }

  if (DstTy.getSizeInBits() != 128) {
    assert(DstTy.getSizeInBits() == 64 && "Unexpected shuffle result ty");
    // This case can be done with TBL1.
    MachineInstr *Concat =
        emitVectorConcat(std::nullopt, Src1Reg, Src2Reg, MIB);
    if (!Concat) {
      LLVM_DEBUG(dbgs() << "Could not do vector concat for tbl1");
      return false;
    }

    // The constant pool load will be 64 bits, so need to convert to FPR128 reg.
    IndexLoad = emitScalarToVector(64, &AArch64::FPR128RegClass,
                                   IndexLoad->getOperand(0).getReg(), MIB);

    auto TBL1 = MIB.buildInstr(
        AArch64::TBLv16i8One, {&AArch64::FPR128RegClass},
        {Concat->getOperand(0).getReg(), IndexLoad->getOperand(0).getReg()});
    constrainSelectedInstRegOperands(*TBL1, TII, TRI, RBI);

    auto Copy =
        MIB.buildInstr(TargetOpcode::COPY, {I.getOperand(0).getReg()}, {})
            .addReg(TBL1.getReg(0), 0, AArch64::dsub);
    RBI.constrainGenericRegister(Copy.getReg(0), AArch64::FPR64RegClass, MRI);
    I.eraseFromParent();
    return true;
  }

  // For TBL2 we need to emit a REG_SEQUENCE to tie together two consecutive
  // Q registers for regalloc.
  SmallVector<Register, 2> Regs = {Src1Reg, Src2Reg};
  auto RegSeq = createQTuple(Regs, MIB);
  auto TBL2 = MIB.buildInstr(AArch64::TBLv16i8Two, {I.getOperand(0)},
                             {RegSeq, IndexLoad->getOperand(0)});
  constrainSelectedInstRegOperands(*TBL2, TII, TRI, RBI);
  I.eraseFromParent();
  return true;
}

MachineInstr *AArch64InstructionSelector::emitLaneInsert(
    std::optional<Register> DstReg, Register SrcReg, Register EltReg,
    unsigned LaneIdx, const RegisterBank &RB,
    MachineIRBuilder &MIRBuilder) const {
  MachineInstr *InsElt = nullptr;
  const TargetRegisterClass *DstRC = &AArch64::FPR128RegClass;
  MachineRegisterInfo &MRI = *MIRBuilder.getMRI();

  // Create a register to define with the insert if one wasn't passed in.
  if (!DstReg)
    DstReg = MRI.createVirtualRegister(DstRC);

  unsigned EltSize = MRI.getType(EltReg).getSizeInBits();
  unsigned Opc = getInsertVecEltOpInfo(RB, EltSize).first;

  if (RB.getID() == AArch64::FPRRegBankID) {
    auto InsSub = emitScalarToVector(EltSize, DstRC, EltReg, MIRBuilder);
    InsElt = MIRBuilder.buildInstr(Opc, {*DstReg}, {SrcReg})
                 .addImm(LaneIdx)
                 .addUse(InsSub->getOperand(0).getReg())
                 .addImm(0);
  } else {
    InsElt = MIRBuilder.buildInstr(Opc, {*DstReg}, {SrcReg})
                 .addImm(LaneIdx)
                 .addUse(EltReg);
  }

  constrainSelectedInstRegOperands(*InsElt, TII, TRI, RBI);
  return InsElt;
}

bool AArch64InstructionSelector::selectUSMovFromExtend(
    MachineInstr &MI, MachineRegisterInfo &MRI) {
  if (MI.getOpcode() != TargetOpcode::G_SEXT &&
      MI.getOpcode() != TargetOpcode::G_ZEXT &&
      MI.getOpcode() != TargetOpcode::G_ANYEXT)
    return false;
  bool IsSigned = MI.getOpcode() == TargetOpcode::G_SEXT;
  const Register DefReg = MI.getOperand(0).getReg();
  const LLT DstTy = MRI.getType(DefReg);
  unsigned DstSize = DstTy.getSizeInBits();

  if (DstSize != 32 && DstSize != 64)
    return false;

  MachineInstr *Extract = getOpcodeDef(TargetOpcode::G_EXTRACT_VECTOR_ELT,
                                       MI.getOperand(1).getReg(), MRI);
  int64_t Lane;
  if (!Extract || !mi_match(Extract->getOperand(2).getReg(), MRI, m_ICst(Lane)))
    return false;
  Register Src0 = Extract->getOperand(1).getReg();

  const LLT &VecTy = MRI.getType(Src0);

  if (VecTy.getSizeInBits() != 128) {
    const MachineInstr *ScalarToVector = emitScalarToVector(
        VecTy.getSizeInBits(), &AArch64::FPR128RegClass, Src0, MIB);
    assert(ScalarToVector && "Didn't expect emitScalarToVector to fail!");
    Src0 = ScalarToVector->getOperand(0).getReg();
  }

  unsigned Opcode;
  if (DstSize == 64 && VecTy.getScalarSizeInBits() == 32)
    Opcode = IsSigned ? AArch64::SMOVvi32to64 : AArch64::UMOVvi32;
  else if (DstSize == 64 && VecTy.getScalarSizeInBits() == 16)
    Opcode = IsSigned ? AArch64::SMOVvi16to64 : AArch64::UMOVvi16;
  else if (DstSize == 64 && VecTy.getScalarSizeInBits() == 8)
    Opcode = IsSigned ? AArch64::SMOVvi8to64 : AArch64::UMOVvi8;
  else if (DstSize == 32 && VecTy.getScalarSizeInBits() == 16)
    Opcode = IsSigned ? AArch64::SMOVvi16to32 : AArch64::UMOVvi16;
  else if (DstSize == 32 && VecTy.getScalarSizeInBits() == 8)
    Opcode = IsSigned ? AArch64::SMOVvi8to32 : AArch64::UMOVvi8;
  else
    llvm_unreachable("Unexpected type combo for S/UMov!");

  // We may need to generate one of these, depending on the type and sign of the
  // input:
  //  DstReg = SMOV Src0, Lane;
  //  NewReg = UMOV Src0, Lane; DstReg = SUBREG_TO_REG NewReg, sub_32;
  MachineInstr *ExtI = nullptr;
  if (DstSize == 64 && !IsSigned) {
    Register NewReg = MRI.createVirtualRegister(&AArch64::GPR32RegClass);
    MIB.buildInstr(Opcode, {NewReg}, {Src0}).addImm(Lane);
    ExtI = MIB.buildInstr(AArch64::SUBREG_TO_REG, {DefReg}, {})
               .addImm(0)
               .addUse(NewReg)
               .addImm(AArch64::sub_32);
    RBI.constrainGenericRegister(DefReg, AArch64::GPR64RegClass, MRI);
  } else
    ExtI = MIB.buildInstr(Opcode, {DefReg}, {Src0}).addImm(Lane);

  constrainSelectedInstRegOperands(*ExtI, TII, TRI, RBI);
  MI.eraseFromParent();
  return true;
}

bool AArch64InstructionSelector::selectInsertElt(MachineInstr &I,
                                                 MachineRegisterInfo &MRI) {
  assert(I.getOpcode() == TargetOpcode::G_INSERT_VECTOR_ELT);

  // Get information on the destination.
  Register DstReg = I.getOperand(0).getReg();
  const LLT DstTy = MRI.getType(DstReg);
  unsigned VecSize = DstTy.getSizeInBits();

  // Get information on the element we want to insert into the destination.
  Register EltReg = I.getOperand(2).getReg();
  const LLT EltTy = MRI.getType(EltReg);
  unsigned EltSize = EltTy.getSizeInBits();
  if (EltSize < 16 || EltSize > 64)
    return false; // Don't support all element types yet.

  // Find the definition of the index. Bail out if it's not defined by a
  // G_CONSTANT.
  Register IdxReg = I.getOperand(3).getReg();
  auto VRegAndVal = getIConstantVRegValWithLookThrough(IdxReg, MRI);
  if (!VRegAndVal)
    return false;
  unsigned LaneIdx = VRegAndVal->Value.getSExtValue();

  // Perform the lane insert.
  Register SrcReg = I.getOperand(1).getReg();
  const RegisterBank &EltRB = *RBI.getRegBank(EltReg, MRI, TRI);

  if (VecSize < 128) {
    // If the vector we're inserting into is smaller than 128 bits, widen it
    // to 128 to do the insert.
    MachineInstr *ScalarToVec =
        emitScalarToVector(VecSize, &AArch64::FPR128RegClass, SrcReg, MIB);
    if (!ScalarToVec)
      return false;
    SrcReg = ScalarToVec->getOperand(0).getReg();
  }

  // Create an insert into a new FPR128 register.
  // Note that if our vector is already 128 bits, we end up emitting an extra
  // register.
  MachineInstr *InsMI =
      emitLaneInsert(std::nullopt, SrcReg, EltReg, LaneIdx, EltRB, MIB);

  if (VecSize < 128) {
    // If we had to widen to perform the insert, then we have to demote back to
    // the original size to get the result we want.
    Register DemoteVec = InsMI->getOperand(0).getReg();
    const TargetRegisterClass *RC =
        getRegClassForTypeOnBank(DstTy, *RBI.getRegBank(DemoteVec, MRI, TRI));
    if (RC != &AArch64::FPR32RegClass && RC != &AArch64::FPR64RegClass) {
      LLVM_DEBUG(dbgs() << "Unsupported register class!\n");
      return false;
    }
    unsigned SubReg = 0;
    if (!getSubRegForClass(RC, TRI, SubReg))
      return false;
    if (SubReg != AArch64::ssub && SubReg != AArch64::dsub) {
      LLVM_DEBUG(dbgs() << "Unsupported destination size! (" << VecSize
                        << "\n");
      return false;
    }
    MIB.buildInstr(TargetOpcode::COPY, {DstReg}, {})
        .addReg(DemoteVec, 0, SubReg);
    RBI.constrainGenericRegister(DstReg, *RC, MRI);
  } else {
    // No widening needed.
    InsMI->getOperand(0).setReg(DstReg);
    constrainSelectedInstRegOperands(*InsMI, TII, TRI, RBI);
  }

  I.eraseFromParent();
  return true;
}

MachineInstr *
AArch64InstructionSelector::emitConstantVector(Register Dst, Constant *CV,
                                               MachineIRBuilder &MIRBuilder,
                                               MachineRegisterInfo &MRI) {
  LLT DstTy = MRI.getType(Dst);
  unsigned DstSize = DstTy.getSizeInBits();
  if (CV->isNullValue()) {
    if (DstSize == 128) {
      auto Mov =
          MIRBuilder.buildInstr(AArch64::MOVIv2d_ns, {Dst}, {}).addImm(0);
      constrainSelectedInstRegOperands(*Mov, TII, TRI, RBI);
      return &*Mov;
    }

    if (DstSize == 64) {
      auto Mov =
          MIRBuilder
              .buildInstr(AArch64::MOVIv2d_ns, {&AArch64::FPR128RegClass}, {})
              .addImm(0);
      auto Copy = MIRBuilder.buildInstr(TargetOpcode::COPY, {Dst}, {})
                      .addReg(Mov.getReg(0), 0, AArch64::dsub);
      RBI.constrainGenericRegister(Dst, AArch64::FPR64RegClass, MRI);
      return &*Copy;
    }
  }

  auto *CPLoad = emitLoadFromConstantPool(CV, MIRBuilder);
  if (!CPLoad) {
    LLVM_DEBUG(dbgs() << "Could not generate cp load for constant vector!");
    return nullptr;
  }

  auto Copy = MIRBuilder.buildCopy(Dst, CPLoad->getOperand(0));
  RBI.constrainGenericRegister(
      Dst, *MRI.getRegClass(CPLoad->getOperand(0).getReg()), MRI);
  return &*Copy;
}

bool AArch64InstructionSelector::tryOptConstantBuildVec(
    MachineInstr &I, LLT DstTy, MachineRegisterInfo &MRI) {
  assert(I.getOpcode() == TargetOpcode::G_BUILD_VECTOR);
  unsigned DstSize = DstTy.getSizeInBits();
  assert(DstSize <= 128 && "Unexpected build_vec type!");
  if (DstSize < 32)
    return false;
  // Check if we're building a constant vector, in which case we want to
  // generate a constant pool load instead of a vector insert sequence.
  SmallVector<Constant *, 16> Csts;
  for (unsigned Idx = 1; Idx < I.getNumOperands(); ++Idx) {
    // Try to find G_CONSTANT or G_FCONSTANT
    auto *OpMI =
        getOpcodeDef(TargetOpcode::G_CONSTANT, I.getOperand(Idx).getReg(), MRI);
    if (OpMI)
      Csts.emplace_back(
          const_cast<ConstantInt *>(OpMI->getOperand(1).getCImm()));
    else if ((OpMI = getOpcodeDef(TargetOpcode::G_FCONSTANT,
                                  I.getOperand(Idx).getReg(), MRI)))
      Csts.emplace_back(
          const_cast<ConstantFP *>(OpMI->getOperand(1).getFPImm()));
    else
      return false;
  }
  Constant *CV = ConstantVector::get(Csts);
  if (!emitConstantVector(I.getOperand(0).getReg(), CV, MIB, MRI))
    return false;
  I.eraseFromParent();
  return true;
}

bool AArch64InstructionSelector::tryOptBuildVecToSubregToReg(
    MachineInstr &I, MachineRegisterInfo &MRI) {
  // Given:
  //  %vec = G_BUILD_VECTOR %elt, %undef, %undef, ... %undef
  //
  // Select the G_BUILD_VECTOR as a SUBREG_TO_REG from %elt.
  Register Dst = I.getOperand(0).getReg();
  Register EltReg = I.getOperand(1).getReg();
  LLT EltTy = MRI.getType(EltReg);
  // If the index isn't on the same bank as its elements, then this can't be a
  // SUBREG_TO_REG.
  const RegisterBank &EltRB = *RBI.getRegBank(EltReg, MRI, TRI);
  const RegisterBank &DstRB = *RBI.getRegBank(Dst, MRI, TRI);
  if (EltRB != DstRB)
    return false;
  if (any_of(make_range(I.operands_begin() + 2, I.operands_end()),
             [&MRI](const MachineOperand &Op) {
               return !getOpcodeDef(TargetOpcode::G_IMPLICIT_DEF, Op.getReg(),
                                    MRI);
             }))
    return false;
  unsigned SubReg;
  const TargetRegisterClass *EltRC = getRegClassForTypeOnBank(EltTy, EltRB);
  if (!EltRC)
    return false;
  const TargetRegisterClass *DstRC =
      getRegClassForTypeOnBank(MRI.getType(Dst), DstRB);
  if (!DstRC)
    return false;
  if (!getSubRegForClass(EltRC, TRI, SubReg))
    return false;
  auto SubregToReg = MIB.buildInstr(AArch64::SUBREG_TO_REG, {Dst}, {})
                         .addImm(0)
                         .addUse(EltReg)
                         .addImm(SubReg);
  I.eraseFromParent();
  constrainSelectedInstRegOperands(*SubregToReg, TII, TRI, RBI);
  return RBI.constrainGenericRegister(Dst, *DstRC, MRI);
}

bool AArch64InstructionSelector::selectBuildVector(MachineInstr &I,
                                                   MachineRegisterInfo &MRI) {
  assert(I.getOpcode() == TargetOpcode::G_BUILD_VECTOR);
  // Until we port more of the optimized selections, for now just use a vector
  // insert sequence.
  const LLT DstTy = MRI.getType(I.getOperand(0).getReg());
  const LLT EltTy = MRI.getType(I.getOperand(1).getReg());
  unsigned EltSize = EltTy.getSizeInBits();

  if (tryOptConstantBuildVec(I, DstTy, MRI))
    return true;
  if (tryOptBuildVecToSubregToReg(I, MRI))
    return true;

  if (EltSize < 16 || EltSize > 64)
    return false; // Don't support all element types yet.
  const RegisterBank &RB = *RBI.getRegBank(I.getOperand(1).getReg(), MRI, TRI);

  const TargetRegisterClass *DstRC = &AArch64::FPR128RegClass;
  MachineInstr *ScalarToVec =
      emitScalarToVector(DstTy.getElementType().getSizeInBits(), DstRC,
                         I.getOperand(1).getReg(), MIB);
  if (!ScalarToVec)
    return false;

  Register DstVec = ScalarToVec->getOperand(0).getReg();
  unsigned DstSize = DstTy.getSizeInBits();

  // Keep track of the last MI we inserted. Later on, we might be able to save
  // a copy using it.
  MachineInstr *PrevMI = nullptr;
  for (unsigned i = 2, e = DstSize / EltSize + 1; i < e; ++i) {
    // Note that if we don't do a subregister copy, we can end up making an
    // extra register.
    PrevMI = &*emitLaneInsert(std::nullopt, DstVec, I.getOperand(i).getReg(),
                              i - 1, RB, MIB);
    DstVec = PrevMI->getOperand(0).getReg();
  }

  // If DstTy's size in bits is less than 128, then emit a subregister copy
  // from DstVec to the last register we've defined.
  if (DstSize < 128) {
    // Force this to be FPR using the destination vector.
    const TargetRegisterClass *RC =
        getRegClassForTypeOnBank(DstTy, *RBI.getRegBank(DstVec, MRI, TRI));
    if (!RC)
      return false;
    if (RC != &AArch64::FPR32RegClass && RC != &AArch64::FPR64RegClass) {
      LLVM_DEBUG(dbgs() << "Unsupported register class!\n");
      return false;
    }

    unsigned SubReg = 0;
    if (!getSubRegForClass(RC, TRI, SubReg))
      return false;
    if (SubReg != AArch64::ssub && SubReg != AArch64::dsub) {
      LLVM_DEBUG(dbgs() << "Unsupported destination size! (" << DstSize
                        << "\n");
      return false;
    }

    Register Reg = MRI.createVirtualRegister(RC);
    Register DstReg = I.getOperand(0).getReg();

    MIB.buildInstr(TargetOpcode::COPY, {DstReg}, {}).addReg(DstVec, 0, SubReg);
    MachineOperand &RegOp = I.getOperand(1);
    RegOp.setReg(Reg);
    RBI.constrainGenericRegister(DstReg, *RC, MRI);
  } else {
    // We don't need a subregister copy. Save a copy by re-using the
    // destination register on the final insert.
    assert(PrevMI && "PrevMI was null?");
    PrevMI->getOperand(0).setReg(I.getOperand(0).getReg());
    constrainSelectedInstRegOperands(*PrevMI, TII, TRI, RBI);
  }

  I.eraseFromParent();
  return true;
}

bool AArch64InstructionSelector::selectVectorLoadIntrinsic(unsigned Opc,
                                                           unsigned NumVecs,
                                                           MachineInstr &I) {
  assert(I.getOpcode() == TargetOpcode::G_INTRINSIC_W_SIDE_EFFECTS);
  assert(Opc && "Expected an opcode?");
  assert(NumVecs > 1 && NumVecs < 5 && "Only support 2, 3, or 4 vectors");
  auto &MRI = *MIB.getMRI();
  LLT Ty = MRI.getType(I.getOperand(0).getReg());
  unsigned Size = Ty.getSizeInBits();
  assert((Size == 64 || Size == 128) &&
         "Destination must be 64 bits or 128 bits?");
  unsigned SubReg = Size == 64 ? AArch64::dsub0 : AArch64::qsub0;
  auto Ptr = I.getOperand(I.getNumOperands() - 1).getReg();
  assert(MRI.getType(Ptr).isPointer() && "Expected a pointer type?");
  auto Load = MIB.buildInstr(Opc, {Ty}, {Ptr});
  Load.cloneMemRefs(I);
  constrainSelectedInstRegOperands(*Load, TII, TRI, RBI);
  Register SelectedLoadDst = Load->getOperand(0).getReg();
  for (unsigned Idx = 0; Idx < NumVecs; ++Idx) {
    auto Vec = MIB.buildInstr(TargetOpcode::COPY, {I.getOperand(Idx)}, {})
                   .addReg(SelectedLoadDst, 0, SubReg + Idx);
    // Emit the subreg copies and immediately select them.
    // FIXME: We should refactor our copy code into an emitCopy helper and
    // clean up uses of this pattern elsewhere in the selector.
    selectCopy(*Vec, TII, MRI, TRI, RBI);
  }
  return true;
}

bool AArch64InstructionSelector::selectIntrinsicWithSideEffects(
    MachineInstr &I, MachineRegisterInfo &MRI) {
  // Find the intrinsic ID.
  unsigned IntrinID = I.getIntrinsicID();

  const LLT S8 = LLT::scalar(8);
  const LLT S16 = LLT::scalar(16);
  const LLT S32 = LLT::scalar(32);
  const LLT S64 = LLT::scalar(64);
  const LLT P0 = LLT::pointer(0, 64);
  // Select the instruction.
  switch (IntrinID) {
  default:
    return false;
  case Intrinsic::aarch64_ldxp:
  case Intrinsic::aarch64_ldaxp: {
    auto NewI = MIB.buildInstr(
        IntrinID == Intrinsic::aarch64_ldxp ? AArch64::LDXPX : AArch64::LDAXPX,
        {I.getOperand(0).getReg(), I.getOperand(1).getReg()},
        {I.getOperand(3)});
    NewI.cloneMemRefs(I);
    constrainSelectedInstRegOperands(*NewI, TII, TRI, RBI);
    break;
  }
  case Intrinsic::trap:
    MIB.buildInstr(AArch64::BRK, {}, {}).addImm(1);
    break;
  case Intrinsic::debugtrap:
    MIB.buildInstr(AArch64::BRK, {}, {}).addImm(0xF000);
    break;
  case Intrinsic::ubsantrap:
    MIB.buildInstr(AArch64::BRK, {}, {})
        .addImm(I.getOperand(1).getImm() | ('U' << 8));
    break;
  case Intrinsic::aarch64_neon_ld2: {
    LLT Ty = MRI.getType(I.getOperand(0).getReg());
    unsigned Opc = 0;
    if (Ty == LLT::fixed_vector(8, S8))
      Opc = AArch64::LD2Twov8b;
    else if (Ty == LLT::fixed_vector(16, S8))
      Opc = AArch64::LD2Twov16b;
    else if (Ty == LLT::fixed_vector(4, S16))
      Opc = AArch64::LD2Twov4h;
    else if (Ty == LLT::fixed_vector(8, S16))
      Opc = AArch64::LD2Twov8h;
    else if (Ty == LLT::fixed_vector(2, S32))
      Opc = AArch64::LD2Twov2s;
    else if (Ty == LLT::fixed_vector(4, S32))
      Opc = AArch64::LD2Twov4s;
    else if (Ty == LLT::fixed_vector(2, S64) || Ty == LLT::fixed_vector(2, P0))
      Opc = AArch64::LD2Twov2d;
    else if (Ty == S64 || Ty == P0)
      Opc = AArch64::LD1Twov1d;
    else
      llvm_unreachable("Unexpected type for ld2!");
    selectVectorLoadIntrinsic(Opc, 2, I);
    break;
  }
  case Intrinsic::aarch64_neon_ld4: {
    LLT Ty = MRI.getType(I.getOperand(0).getReg());
    unsigned Opc = 0;
    if (Ty == LLT::fixed_vector(8, S8))
      Opc = AArch64::LD4Fourv8b;
    else if (Ty == LLT::fixed_vector(16, S8))
      Opc = AArch64::LD4Fourv16b;
    else if (Ty == LLT::fixed_vector(4, S16))
      Opc = AArch64::LD4Fourv4h;
    else if (Ty == LLT::fixed_vector(8, S16))
      Opc = AArch64::LD4Fourv8h;
    else if (Ty == LLT::fixed_vector(2, S32))
      Opc = AArch64::LD4Fourv2s;
    else if (Ty == LLT::fixed_vector(4, S32))
      Opc = AArch64::LD4Fourv4s;
    else if (Ty == LLT::fixed_vector(2, S64) || Ty == LLT::fixed_vector(2, P0))
      Opc = AArch64::LD4Fourv2d;
    else if (Ty == S64 || Ty == P0)
      Opc = AArch64::LD1Fourv1d;
    else
      llvm_unreachable("Unexpected type for ld4!");
    selectVectorLoadIntrinsic(Opc, 4, I);
    break;
  }
  case Intrinsic::aarch64_neon_st2: {
    Register Src1 = I.getOperand(1).getReg();
    Register Src2 = I.getOperand(2).getReg();
    Register Ptr = I.getOperand(3).getReg();
    LLT Ty = MRI.getType(Src1);
    unsigned Opc;
    if (Ty == LLT::fixed_vector(8, S8))
      Opc = AArch64::ST2Twov8b;
    else if (Ty == LLT::fixed_vector(16, S8))
      Opc = AArch64::ST2Twov16b;
    else if (Ty == LLT::fixed_vector(4, S16))
      Opc = AArch64::ST2Twov4h;
    else if (Ty == LLT::fixed_vector(8, S16))
      Opc = AArch64::ST2Twov8h;
    else if (Ty == LLT::fixed_vector(2, S32))
      Opc = AArch64::ST2Twov2s;
    else if (Ty == LLT::fixed_vector(4, S32))
      Opc = AArch64::ST2Twov4s;
    else if (Ty == LLT::fixed_vector(2, S64) || Ty == LLT::fixed_vector(2, P0))
      Opc = AArch64::ST2Twov2d;
    else if (Ty == S64 || Ty == P0)
      Opc = AArch64::ST1Twov1d;
    else
      llvm_unreachable("Unexpected type for st2!");
    SmallVector<Register, 2> Regs = {Src1, Src2};
    Register Tuple = Ty.getSizeInBits() == 128 ? createQTuple(Regs, MIB)
                                               : createDTuple(Regs, MIB);
    auto Store = MIB.buildInstr(Opc, {}, {Tuple, Ptr});
    Store.cloneMemRefs(I);
    constrainSelectedInstRegOperands(*Store, TII, TRI, RBI);
    break;
  }
  case Intrinsic::aarch64_mops_memset_tag: {
    // Transform
    //    %dst:gpr(p0) = \
    //      G_INTRINSIC_W_SIDE_EFFECTS intrinsic(@llvm.aarch64.mops.memset.tag),
    //      \ %dst:gpr(p0), %val:gpr(s64), %n:gpr(s64)
    // where %dst is updated, into
    //    %Rd:GPR64common, %Rn:GPR64) = \
    //      MOPSMemorySetTaggingPseudo \
    //      %Rd:GPR64common, %Rn:GPR64, %Rm:GPR64
    // where Rd and Rn are tied.
    // It is expected that %val has been extended to s64 in legalization.
    // Note that the order of the size/value operands are swapped.

    Register DstDef = I.getOperand(0).getReg();
    // I.getOperand(1) is the intrinsic function
    Register DstUse = I.getOperand(2).getReg();
    Register ValUse = I.getOperand(3).getReg();
    Register SizeUse = I.getOperand(4).getReg();

    // MOPSMemorySetTaggingPseudo has two defs; the intrinsic call has only one.
    // Therefore an additional virtual register is requried for the updated size
    // operand. This value is not accessible via the semantics of the intrinsic.
    Register SizeDef = MRI.createGenericVirtualRegister(LLT::scalar(64));

    auto Memset = MIB.buildInstr(AArch64::MOPSMemorySetTaggingPseudo,
                                 {DstDef, SizeDef}, {DstUse, SizeUse, ValUse});
    Memset.cloneMemRefs(I);
    constrainSelectedInstRegOperands(*Memset, TII, TRI, RBI);
    break;
  }
  }

  I.eraseFromParent();
  return true;
}

bool AArch64InstructionSelector::selectIntrinsic(MachineInstr &I,
                                                 MachineRegisterInfo &MRI) {
  unsigned IntrinID = I.getIntrinsicID();

  switch (IntrinID) {
  default:
    break;
  case Intrinsic::aarch64_crypto_sha1h: {
    Register DstReg = I.getOperand(0).getReg();
    Register SrcReg = I.getOperand(2).getReg();

    // FIXME: Should this be an assert?
    if (MRI.getType(DstReg).getSizeInBits() != 32 ||
        MRI.getType(SrcReg).getSizeInBits() != 32)
      return false;

    // The operation has to happen on FPRs. Set up some new FPR registers for
    // the source and destination if they are on GPRs.
    if (RBI.getRegBank(SrcReg, MRI, TRI)->getID() != AArch64::FPRRegBankID) {
      SrcReg = MRI.createVirtualRegister(&AArch64::FPR32RegClass);
      MIB.buildCopy({SrcReg}, {I.getOperand(2)});

      // Make sure the copy ends up getting constrained properly.
      RBI.constrainGenericRegister(I.getOperand(2).getReg(),
                                   AArch64::GPR32RegClass, MRI);
    }

    if (RBI.getRegBank(DstReg, MRI, TRI)->getID() != AArch64::FPRRegBankID)
      DstReg = MRI.createVirtualRegister(&AArch64::FPR32RegClass);

    // Actually insert the instruction.
    auto SHA1Inst = MIB.buildInstr(AArch64::SHA1Hrr, {DstReg}, {SrcReg});
    constrainSelectedInstRegOperands(*SHA1Inst, TII, TRI, RBI);

    // Did we create a new register for the destination?
    if (DstReg != I.getOperand(0).getReg()) {
      // Yep. Copy the result of the instruction back into the original
      // destination.
      MIB.buildCopy({I.getOperand(0)}, {DstReg});
      RBI.constrainGenericRegister(I.getOperand(0).getReg(),
                                   AArch64::GPR32RegClass, MRI);
    }

    I.eraseFromParent();
    return true;
  }
  case Intrinsic::ptrauth_sign: {
    Register DstReg = I.getOperand(0).getReg();
    Register ValReg = I.getOperand(2).getReg();
    uint64_t Key = I.getOperand(3).getImm();
    Register DiscReg = I.getOperand(4).getReg();
    auto DiscVal = getIConstantVRegVal(DiscReg, MRI);
    bool IsDiscZero = DiscVal && DiscVal->isNullValue();

    if (Key > AArch64PACKey::LAST)
      return false;

    unsigned Opcodes[][4] = {
        {AArch64::PACIA, AArch64::PACIB, AArch64::PACDA, AArch64::PACDB},
        {AArch64::PACIZA, AArch64::PACIZB, AArch64::PACDZA, AArch64::PACDZB}};
    unsigned Opcode = Opcodes[IsDiscZero][Key];

    auto PAC = MIB.buildInstr(Opcode, {DstReg}, {ValReg});

    if (!IsDiscZero) {
      PAC.addUse(DiscReg);
      RBI.constrainGenericRegister(DiscReg, AArch64::GPR64spRegClass, MRI);
    }

    RBI.constrainGenericRegister(DstReg, AArch64::GPR64RegClass, MRI);
    I.eraseFromParent();
    return true;
  }
  case Intrinsic::ptrauth_strip: {
    Register DstReg = I.getOperand(0).getReg();
    Register ValReg = I.getOperand(2).getReg();
    uint64_t Key = I.getOperand(3).getImm();

    if (Key > AArch64PACKey::LAST)
      return false;
    unsigned Opcode = getXPACOpcodeForKey((AArch64PACKey::ID)Key);

    MIB.buildInstr(Opcode, {DstReg}, {ValReg});

    RBI.constrainGenericRegister(DstReg, AArch64::GPR64RegClass, MRI);
    RBI.constrainGenericRegister(ValReg, AArch64::GPR64RegClass, MRI);
    I.eraseFromParent();
    return true;
  }
  case Intrinsic::frameaddress:
  case Intrinsic::returnaddress: {
    MachineFunction &MF = *I.getParent()->getParent();
    MachineFrameInfo &MFI = MF.getFrameInfo();

    unsigned Depth = I.getOperand(2).getImm();
    Register DstReg = I.getOperand(0).getReg();
    RBI.constrainGenericRegister(DstReg, AArch64::GPR64RegClass, MRI);

    if (Depth == 0 && IntrinID == Intrinsic::returnaddress) {
      if (!MFReturnAddr) {
        // Insert the copy from LR/X30 into the entry block, before it can be
        // clobbered by anything.
        MFI.setReturnAddressIsTaken(true);
        MFReturnAddr = getFunctionLiveInPhysReg(
            MF, TII, AArch64::LR, AArch64::GPR64RegClass, I.getDebugLoc());
      }

      if (STI.hasPAuth()) {
        MIB.buildInstr(AArch64::XPACI, {DstReg}, {MFReturnAddr});
      } else {
        MIB.buildCopy({Register(AArch64::LR)}, {MFReturnAddr});
        MIB.buildInstr(AArch64::XPACLRI);
        MIB.buildCopy({DstReg}, {Register(AArch64::LR)});
      }

      I.eraseFromParent();
      return true;
    }

    MFI.setFrameAddressIsTaken(true);
    Register FrameAddr(AArch64::FP);
    while (Depth--) {
      Register NextFrame = MRI.createVirtualRegister(&AArch64::GPR64spRegClass);
      auto Ldr =
          MIB.buildInstr(AArch64::LDRXui, {NextFrame}, {FrameAddr}).addImm(0);
      constrainSelectedInstRegOperands(*Ldr, TII, TRI, RBI);
      FrameAddr = NextFrame;
    }

    if (IntrinID == Intrinsic::frameaddress)
      MIB.buildCopy({DstReg}, {FrameAddr});
    else {
      MFI.setReturnAddressIsTaken(true);

      if (STI.hasPAuth()) {
        Register TmpReg = MRI.createVirtualRegister(&AArch64::GPR64RegClass);
        MIB.buildInstr(AArch64::LDRXui, {TmpReg}, {FrameAddr}).addImm(1);
        MIB.buildInstr(AArch64::XPACI, {DstReg}, {TmpReg});
      } else {
        MIB.buildInstr(AArch64::LDRXui, {Register(AArch64::LR)}, {FrameAddr})
            .addImm(1);
        MIB.buildInstr(AArch64::XPACLRI);
        MIB.buildCopy({DstReg}, {Register(AArch64::LR)});
      }
    }

    I.eraseFromParent();
    return true;
  }
  case Intrinsic::swift_async_context_addr:
    auto Sub = MIB.buildInstr(AArch64::SUBXri, {I.getOperand(0).getReg()},
                              {Register(AArch64::FP)})
                   .addImm(8)
                   .addImm(0);
    constrainSelectedInstRegOperands(*Sub, TII, TRI, RBI);

    MF->getFrameInfo().setFrameAddressIsTaken(true);
    MF->getInfo<AArch64FunctionInfo>()->setHasSwiftAsyncContext(true);
    I.eraseFromParent();
    return true;
  }
  return false;
}

InstructionSelector::ComplexRendererFns
AArch64InstructionSelector::selectShiftA_32(const MachineOperand &Root) const {
  auto MaybeImmed = getImmedFromMO(Root);
  if (MaybeImmed == std::nullopt || *MaybeImmed > 31)
    return std::nullopt;
  uint64_t Enc = (32 - *MaybeImmed) & 0x1f;
  return {{[=](MachineInstrBuilder &MIB) { MIB.addImm(Enc); }}};
}

InstructionSelector::ComplexRendererFns
AArch64InstructionSelector::selectShiftB_32(const MachineOperand &Root) const {
  auto MaybeImmed = getImmedFromMO(Root);
  if (MaybeImmed == std::nullopt || *MaybeImmed > 31)
    return std::nullopt;
  uint64_t Enc = 31 - *MaybeImmed;
  return {{[=](MachineInstrBuilder &MIB) { MIB.addImm(Enc); }}};
}

InstructionSelector::ComplexRendererFns
AArch64InstructionSelector::selectShiftA_64(const MachineOperand &Root) const {
  auto MaybeImmed = getImmedFromMO(Root);
  if (MaybeImmed == std::nullopt || *MaybeImmed > 63)
    return std::nullopt;
  uint64_t Enc = (64 - *MaybeImmed) & 0x3f;
  return {{[=](MachineInstrBuilder &MIB) { MIB.addImm(Enc); }}};
}

InstructionSelector::ComplexRendererFns
AArch64InstructionSelector::selectShiftB_64(const MachineOperand &Root) const {
  auto MaybeImmed = getImmedFromMO(Root);
  if (MaybeImmed == std::nullopt || *MaybeImmed > 63)
    return std::nullopt;
  uint64_t Enc = 63 - *MaybeImmed;
  return {{[=](MachineInstrBuilder &MIB) { MIB.addImm(Enc); }}};
}

/// Helper to select an immediate value that can be represented as a 12-bit
/// value shifted left by either 0 or 12. If it is possible to do so, return
/// the immediate and shift value. If not, return std::nullopt.
///
/// Used by selectArithImmed and selectNegArithImmed.
InstructionSelector::ComplexRendererFns
AArch64InstructionSelector::select12BitValueWithLeftShift(
    uint64_t Immed) const {
  unsigned ShiftAmt;
  if (Immed >> 12 == 0) {
    ShiftAmt = 0;
  } else if ((Immed & 0xfff) == 0 && Immed >> 24 == 0) {
    ShiftAmt = 12;
    Immed = Immed >> 12;
  } else
    return std::nullopt;

  unsigned ShVal = AArch64_AM::getShifterImm(AArch64_AM::LSL, ShiftAmt);
  return {{
      [=](MachineInstrBuilder &MIB) { MIB.addImm(Immed); },
      [=](MachineInstrBuilder &MIB) { MIB.addImm(ShVal); },
  }};
}

/// SelectArithImmed - Select an immediate value that can be represented as
/// a 12-bit value shifted left by either 0 or 12.  If so, return true with
/// Val set to the 12-bit value and Shift set to the shifter operand.
InstructionSelector::ComplexRendererFns
AArch64InstructionSelector::selectArithImmed(MachineOperand &Root) const {
  // This function is called from the addsub_shifted_imm ComplexPattern,
  // which lists [imm] as the list of opcode it's interested in, however
  // we still need to check whether the operand is actually an immediate
  // here because the ComplexPattern opcode list is only used in
  // root-level opcode matching.
  auto MaybeImmed = getImmedFromMO(Root);
  if (MaybeImmed == std::nullopt)
    return std::nullopt;
  return select12BitValueWithLeftShift(*MaybeImmed);
}

/// SelectNegArithImmed - As above, but negates the value before trying to
/// select it.
InstructionSelector::ComplexRendererFns
AArch64InstructionSelector::selectNegArithImmed(MachineOperand &Root) const {
  // We need a register here, because we need to know if we have a 64 or 32
  // bit immediate.
  if (!Root.isReg())
    return std::nullopt;
  auto MaybeImmed = getImmedFromMO(Root);
  if (MaybeImmed == std::nullopt)
    return std::nullopt;
  uint64_t Immed = *MaybeImmed;

  // This negation is almost always valid, but "cmp wN, #0" and "cmn wN, #0"
  // have the opposite effect on the C flag, so this pattern mustn't match under
  // those circumstances.
  if (Immed == 0)
    return std::nullopt;

  // Check if we're dealing with a 32-bit type on the root or a 64-bit type on
  // the root.
  MachineRegisterInfo &MRI = Root.getParent()->getMF()->getRegInfo();
  if (MRI.getType(Root.getReg()).getSizeInBits() == 32)
    Immed = ~((uint32_t)Immed) + 1;
  else
    Immed = ~Immed + 1ULL;

  if (Immed & 0xFFFFFFFFFF000000ULL)
    return std::nullopt;

  Immed &= 0xFFFFFFULL;
  return select12BitValueWithLeftShift(Immed);
}

/// Return true if it is worth folding MI into an extended register. That is,
/// if it's safe to pull it into the addressing mode of a load or store as a
/// shift.
bool AArch64InstructionSelector::isWorthFoldingIntoExtendedReg(
    MachineInstr &MI, const MachineRegisterInfo &MRI) const {
  // Always fold if there is one use, or if we're optimizing for size.
  Register DefReg = MI.getOperand(0).getReg();
  if (MRI.hasOneNonDBGUse(DefReg) ||
      MI.getParent()->getParent()->getFunction().hasOptSize())
    return true;

  // It's better to avoid folding and recomputing shifts when we don't have a
  // fastpath.
  if (!STI.hasLSLFast())
    return false;

  // We have a fastpath, so folding a shift in and potentially computing it
  // many times may be beneficial. Check if this is only used in memory ops.
  // If it is, then we should fold.
  return all_of(MRI.use_nodbg_instructions(DefReg),
                [](MachineInstr &Use) { return Use.mayLoadOrStore(); });
}

static bool isSignExtendShiftType(AArch64_AM::ShiftExtendType Type) {
  switch (Type) {
  case AArch64_AM::SXTB:
  case AArch64_AM::SXTH:
  case AArch64_AM::SXTW:
    return true;
  default:
    return false;
  }
}

InstructionSelector::ComplexRendererFns
AArch64InstructionSelector::selectExtendedSHL(
    MachineOperand &Root, MachineOperand &Base, MachineOperand &Offset,
    unsigned SizeInBytes, bool WantsExt) const {
  assert(Base.isReg() && "Expected base to be a register operand");
  assert(Offset.isReg() && "Expected offset to be a register operand");

  MachineRegisterInfo &MRI = Root.getParent()->getMF()->getRegInfo();
  MachineInstr *OffsetInst = MRI.getVRegDef(Offset.getReg());

  unsigned OffsetOpc = OffsetInst->getOpcode();
  bool LookedThroughZExt = false;
  if (OffsetOpc != TargetOpcode::G_SHL && OffsetOpc != TargetOpcode::G_MUL) {
    // Try to look through a ZEXT.
    if (OffsetOpc != TargetOpcode::G_ZEXT || !WantsExt)
      return std::nullopt;

    OffsetInst = MRI.getVRegDef(OffsetInst->getOperand(1).getReg());
    OffsetOpc = OffsetInst->getOpcode();
    LookedThroughZExt = true;

    if (OffsetOpc != TargetOpcode::G_SHL && OffsetOpc != TargetOpcode::G_MUL)
      return std::nullopt;
  }
  // Make sure that the memory op is a valid size.
  int64_t LegalShiftVal = Log2_32(SizeInBytes);
  if (LegalShiftVal == 0)
    return std::nullopt;
  if (!isWorthFoldingIntoExtendedReg(*OffsetInst, MRI))
    return std::nullopt;

  // Now, try to find the specific G_CONSTANT. Start by assuming that the
  // register we will offset is the LHS, and the register containing the
  // constant is the RHS.
  Register OffsetReg = OffsetInst->getOperand(1).getReg();
  Register ConstantReg = OffsetInst->getOperand(2).getReg();
  auto ValAndVReg = getIConstantVRegValWithLookThrough(ConstantReg, MRI);
  if (!ValAndVReg) {
    // We didn't get a constant on the RHS. If the opcode is a shift, then
    // we're done.
    if (OffsetOpc == TargetOpcode::G_SHL)
      return std::nullopt;

    // If we have a G_MUL, we can use either register. Try looking at the RHS.
    std::swap(OffsetReg, ConstantReg);
    ValAndVReg = getIConstantVRegValWithLookThrough(ConstantReg, MRI);
    if (!ValAndVReg)
      return std::nullopt;
  }

  // The value must fit into 3 bits, and must be positive. Make sure that is
  // true.
  int64_t ImmVal = ValAndVReg->Value.getSExtValue();

  // Since we're going to pull this into a shift, the constant value must be
  // a power of 2. If we got a multiply, then we need to check this.
  if (OffsetOpc == TargetOpcode::G_MUL) {
    if (!isPowerOf2_32(ImmVal))
      return std::nullopt;

    // Got a power of 2. So, the amount we'll shift is the log base-2 of that.
    ImmVal = Log2_32(ImmVal);
  }

  if ((ImmVal & 0x7) != ImmVal)
    return std::nullopt;

  // We are only allowed to shift by LegalShiftVal. This shift value is built
  // into the instruction, so we can't just use whatever we want.
  if (ImmVal != LegalShiftVal)
    return std::nullopt;

  unsigned SignExtend = 0;
  if (WantsExt) {
    // Check if the offset is defined by an extend, unless we looked through a
    // G_ZEXT earlier.
    if (!LookedThroughZExt) {
      MachineInstr *ExtInst = getDefIgnoringCopies(OffsetReg, MRI);
      auto Ext = getExtendTypeForInst(*ExtInst, MRI, true);
      if (Ext == AArch64_AM::InvalidShiftExtend)
        return std::nullopt;

      SignExtend = isSignExtendShiftType(Ext) ? 1 : 0;
      // We only support SXTW for signed extension here.
      if (SignExtend && Ext != AArch64_AM::SXTW)
        return std::nullopt;
      OffsetReg = ExtInst->getOperand(1).getReg();
    }

    // Need a 32-bit wide register here.
    MachineIRBuilder MIB(*MRI.getVRegDef(Root.getReg()));
    OffsetReg = moveScalarRegClass(OffsetReg, AArch64::GPR32RegClass, MIB);
  }

  // We can use the LHS of the GEP as the base, and the LHS of the shift as an
  // offset. Signify that we are shifting by setting the shift flag to 1.
  return {{[=](MachineInstrBuilder &MIB) { MIB.addUse(Base.getReg()); },
           [=](MachineInstrBuilder &MIB) { MIB.addUse(OffsetReg); },
           [=](MachineInstrBuilder &MIB) {
             // Need to add both immediates here to make sure that they are both
             // added to the instruction.
             MIB.addImm(SignExtend);
             MIB.addImm(1);
           }}};
}

/// This is used for computing addresses like this:
///
/// ldr x1, [x2, x3, lsl #3]
///
/// Where x2 is the base register, and x3 is an offset register. The shift-left
/// is a constant value specific to this load instruction. That is, we'll never
/// see anything other than a 3 here (which corresponds to the size of the
/// element being loaded.)
InstructionSelector::ComplexRendererFns
AArch64InstructionSelector::selectAddrModeShiftedExtendXReg(
    MachineOperand &Root, unsigned SizeInBytes) const {
  if (!Root.isReg())
    return std::nullopt;
  MachineRegisterInfo &MRI = Root.getParent()->getMF()->getRegInfo();

  // We want to find something like this:
  //
  // val = G_CONSTANT LegalShiftVal
  // shift = G_SHL off_reg val
  // ptr = G_PTR_ADD base_reg shift
  // x = G_LOAD ptr
  //
  // And fold it into this addressing mode:
  //
  // ldr x, [base_reg, off_reg, lsl #LegalShiftVal]

  // Check if we can find the G_PTR_ADD.
  MachineInstr *PtrAdd =
      getOpcodeDef(TargetOpcode::G_PTR_ADD, Root.getReg(), MRI);
  if (!PtrAdd || !isWorthFoldingIntoExtendedReg(*PtrAdd, MRI))
    return std::nullopt;

  // Now, try to match an opcode which will match our specific offset.
  // We want a G_SHL or a G_MUL.
  MachineInstr *OffsetInst =
      getDefIgnoringCopies(PtrAdd->getOperand(2).getReg(), MRI);
  return selectExtendedSHL(Root, PtrAdd->getOperand(1),
                           OffsetInst->getOperand(0), SizeInBytes,
                           /*WantsExt=*/false);
}

/// This is used for computing addresses like this:
///
/// ldr x1, [x2, x3]
///
/// Where x2 is the base register, and x3 is an offset register.
///
/// When possible (or profitable) to fold a G_PTR_ADD into the address
/// calculation, this will do so. Otherwise, it will return std::nullopt.
InstructionSelector::ComplexRendererFns
AArch64InstructionSelector::selectAddrModeRegisterOffset(
    MachineOperand &Root) const {
  MachineRegisterInfo &MRI = Root.getParent()->getMF()->getRegInfo();

  // We need a GEP.
  MachineInstr *Gep = MRI.getVRegDef(Root.getReg());
  if (Gep->getOpcode() != TargetOpcode::G_PTR_ADD)
    return std::nullopt;

  // If this is used more than once, let's not bother folding.
  // TODO: Check if they are memory ops. If they are, then we can still fold
  // without having to recompute anything.
  if (!MRI.hasOneNonDBGUse(Gep->getOperand(0).getReg()))
    return std::nullopt;

  // Base is the GEP's LHS, offset is its RHS.
  return {{[=](MachineInstrBuilder &MIB) {
             MIB.addUse(Gep->getOperand(1).getReg());
           },
           [=](MachineInstrBuilder &MIB) {
             MIB.addUse(Gep->getOperand(2).getReg());
           },
           [=](MachineInstrBuilder &MIB) {
             // Need to add both immediates here to make sure that they are both
             // added to the instruction.
             MIB.addImm(0);
             MIB.addImm(0);
           }}};
}

/// This is intended to be equivalent to selectAddrModeXRO in
/// AArch64ISelDAGtoDAG. It's used for selecting X register offset loads.
InstructionSelector::ComplexRendererFns
AArch64InstructionSelector::selectAddrModeXRO(MachineOperand &Root,
                                              unsigned SizeInBytes) const {
  MachineRegisterInfo &MRI = Root.getParent()->getMF()->getRegInfo();
  if (!Root.isReg())
    return std::nullopt;
  MachineInstr *PtrAdd =
      getOpcodeDef(TargetOpcode::G_PTR_ADD, Root.getReg(), MRI);
  if (!PtrAdd)
    return std::nullopt;

  // Check for an immediates which cannot be encoded in the [base + imm]
  // addressing mode, and can't be encoded in an add/sub. If this happens, we'll
  // end up with code like:
  //
  // mov x0, wide
  // add x1 base, x0
  // ldr x2, [x1, x0]
  //
  // In this situation, we can use the [base, xreg] addressing mode to save an
  // add/sub:
  //
  // mov x0, wide
  // ldr x2, [base, x0]
  auto ValAndVReg =
      getIConstantVRegValWithLookThrough(PtrAdd->getOperand(2).getReg(), MRI);
  if (ValAndVReg) {
    unsigned Scale = Log2_32(SizeInBytes);
    int64_t ImmOff = ValAndVReg->Value.getSExtValue();

    // Skip immediates that can be selected in the load/store addresing
    // mode.
    if (ImmOff % SizeInBytes == 0 && ImmOff >= 0 &&
        ImmOff < (0x1000 << Scale))
      return std::nullopt;

    // Helper lambda to decide whether or not it is preferable to emit an add.
    auto isPreferredADD = [](int64_t ImmOff) {
      // Constants in [0x0, 0xfff] can be encoded in an add.
      if ((ImmOff & 0xfffffffffffff000LL) == 0x0LL)
        return true;

      // Can it be encoded in an add lsl #12?
      if ((ImmOff & 0xffffffffff000fffLL) != 0x0LL)
        return false;

      // It can be encoded in an add lsl #12, but we may not want to. If it is
      // possible to select this as a single movz, then prefer that. A single
      // movz is faster than an add with a shift.
      return (ImmOff & 0xffffffffff00ffffLL) != 0x0LL &&
             (ImmOff & 0xffffffffffff0fffLL) != 0x0LL;
    };

    // If the immediate can be encoded in a single add/sub, then bail out.
    if (isPreferredADD(ImmOff) || isPreferredADD(-ImmOff))
      return std::nullopt;
  }

  // Try to fold shifts into the addressing mode.
  auto AddrModeFns = selectAddrModeShiftedExtendXReg(Root, SizeInBytes);
  if (AddrModeFns)
    return AddrModeFns;

  // If that doesn't work, see if it's possible to fold in registers from
  // a GEP.
  return selectAddrModeRegisterOffset(Root);
}

/// This is used for computing addresses like this:
///
/// ldr x0, [xBase, wOffset, sxtw #LegalShiftVal]
///
/// Where we have a 64-bit base register, a 32-bit offset register, and an
/// extend (which may or may not be signed).
InstructionSelector::ComplexRendererFns
AArch64InstructionSelector::selectAddrModeWRO(MachineOperand &Root,
                                              unsigned SizeInBytes) const {
  MachineRegisterInfo &MRI = Root.getParent()->getMF()->getRegInfo();

  MachineInstr *PtrAdd =
      getOpcodeDef(TargetOpcode::G_PTR_ADD, Root.getReg(), MRI);
  if (!PtrAdd || !isWorthFoldingIntoExtendedReg(*PtrAdd, MRI))
    return std::nullopt;

  MachineOperand &LHS = PtrAdd->getOperand(1);
  MachineOperand &RHS = PtrAdd->getOperand(2);
  MachineInstr *OffsetInst = getDefIgnoringCopies(RHS.getReg(), MRI);

  // The first case is the same as selectAddrModeXRO, except we need an extend.
  // In this case, we try to find a shift and extend, and fold them into the
  // addressing mode.
  //
  // E.g.
  //
  // off_reg = G_Z/S/ANYEXT ext_reg
  // val = G_CONSTANT LegalShiftVal
  // shift = G_SHL off_reg val
  // ptr = G_PTR_ADD base_reg shift
  // x = G_LOAD ptr
  //
  // In this case we can get a load like this:
  //
  // ldr x0, [base_reg, ext_reg, sxtw #LegalShiftVal]
  auto ExtendedShl = selectExtendedSHL(Root, LHS, OffsetInst->getOperand(0),
                                       SizeInBytes, /*WantsExt=*/true);
  if (ExtendedShl)
    return ExtendedShl;

  // There was no shift. We can try and fold a G_Z/S/ANYEXT in alone though.
  //
  // e.g.
  // ldr something, [base_reg, ext_reg, sxtw]
  if (!isWorthFoldingIntoExtendedReg(*OffsetInst, MRI))
    return std::nullopt;

  // Check if this is an extend. We'll get an extend type if it is.
  AArch64_AM::ShiftExtendType Ext =
      getExtendTypeForInst(*OffsetInst, MRI, /*IsLoadStore=*/true);
  if (Ext == AArch64_AM::InvalidShiftExtend)
    return std::nullopt;

  // Need a 32-bit wide register.
  MachineIRBuilder MIB(*PtrAdd);
  Register ExtReg = moveScalarRegClass(OffsetInst->getOperand(1).getReg(),
                                       AArch64::GPR32RegClass, MIB);
  unsigned SignExtend = Ext == AArch64_AM::SXTW;

  // Base is LHS, offset is ExtReg.
  return {{[=](MachineInstrBuilder &MIB) { MIB.addUse(LHS.getReg()); },
           [=](MachineInstrBuilder &MIB) { MIB.addUse(ExtReg); },
           [=](MachineInstrBuilder &MIB) {
             MIB.addImm(SignExtend);
             MIB.addImm(0);
           }}};
}

/// Select a "register plus unscaled signed 9-bit immediate" address.  This
/// should only match when there is an offset that is not valid for a scaled
/// immediate addressing mode.  The "Size" argument is the size in bytes of the
/// memory reference, which is needed here to know what is valid for a scaled
/// immediate.
InstructionSelector::ComplexRendererFns
AArch64InstructionSelector::selectAddrModeUnscaled(MachineOperand &Root,
                                                   unsigned Size) const {
  MachineRegisterInfo &MRI =
      Root.getParent()->getParent()->getParent()->getRegInfo();

  if (!Root.isReg())
    return std::nullopt;

  if (!isBaseWithConstantOffset(Root, MRI))
    return std::nullopt;

  MachineInstr *RootDef = MRI.getVRegDef(Root.getReg());

  MachineOperand &OffImm = RootDef->getOperand(2);
  if (!OffImm.isReg())
    return std::nullopt;
  MachineInstr *RHS = MRI.getVRegDef(OffImm.getReg());
  if (RHS->getOpcode() != TargetOpcode::G_CONSTANT)
    return std::nullopt;
  int64_t RHSC;
  MachineOperand &RHSOp1 = RHS->getOperand(1);
  if (!RHSOp1.isCImm() || RHSOp1.getCImm()->getBitWidth() > 64)
    return std::nullopt;
  RHSC = RHSOp1.getCImm()->getSExtValue();

  // If the offset is valid as a scaled immediate, don't match here.
  if ((RHSC & (Size - 1)) == 0 && RHSC >= 0 && RHSC < (0x1000 << Log2_32(Size)))
    return std::nullopt;
  if (RHSC >= -256 && RHSC < 256) {
    MachineOperand &Base = RootDef->getOperand(1);
    return {{
        [=](MachineInstrBuilder &MIB) { MIB.add(Base); },
        [=](MachineInstrBuilder &MIB) { MIB.addImm(RHSC); },
    }};
  }
  return std::nullopt;
}

InstructionSelector::ComplexRendererFns
AArch64InstructionSelector::tryFoldAddLowIntoImm(MachineInstr &RootDef,
                                                 unsigned Size,
                                                 MachineRegisterInfo &MRI) const {
  if (RootDef.getOpcode() != AArch64::G_ADD_LOW)
    return std::nullopt;
  MachineInstr &Adrp = *MRI.getVRegDef(RootDef.getOperand(1).getReg());
  if (Adrp.getOpcode() != AArch64::ADRP)
    return std::nullopt;

  // TODO: add heuristics like isWorthFoldingADDlow() from SelectionDAG.
  auto Offset = Adrp.getOperand(1).getOffset();
  if (Offset % Size != 0)
    return std::nullopt;

  auto GV = Adrp.getOperand(1).getGlobal();
  if (GV->isThreadLocal())
    return std::nullopt;

  auto &MF = *RootDef.getParent()->getParent();
  if (GV->getPointerAlignment(MF.getDataLayout()) < Size)
    return std::nullopt;

  unsigned OpFlags = STI.ClassifyGlobalReference(GV, MF.getTarget());
  MachineIRBuilder MIRBuilder(RootDef);
  Register AdrpReg = Adrp.getOperand(0).getReg();
  return {{[=](MachineInstrBuilder &MIB) { MIB.addUse(AdrpReg); },
           [=](MachineInstrBuilder &MIB) {
             MIB.addGlobalAddress(GV, Offset,
                                  OpFlags | AArch64II::MO_PAGEOFF |
                                      AArch64II::MO_NC);
           }}};
}

/// Select a "register plus scaled unsigned 12-bit immediate" address.  The
/// "Size" argument is the size in bytes of the memory reference, which
/// determines the scale.
InstructionSelector::ComplexRendererFns
AArch64InstructionSelector::selectAddrModeIndexed(MachineOperand &Root,
                                                  unsigned Size) const {
  MachineFunction &MF = *Root.getParent()->getParent()->getParent();
  MachineRegisterInfo &MRI = MF.getRegInfo();

  if (!Root.isReg())
    return std::nullopt;

  MachineInstr *RootDef = MRI.getVRegDef(Root.getReg());
  if (RootDef->getOpcode() == TargetOpcode::G_FRAME_INDEX) {
    return {{
        [=](MachineInstrBuilder &MIB) { MIB.add(RootDef->getOperand(1)); },
        [=](MachineInstrBuilder &MIB) { MIB.addImm(0); },
    }};
  }

  CodeModel::Model CM = MF.getTarget().getCodeModel();
  // Check if we can fold in the ADD of small code model ADRP + ADD address.
  if (CM == CodeModel::Small) {
    auto OpFns = tryFoldAddLowIntoImm(*RootDef, Size, MRI);
    if (OpFns)
      return OpFns;
  }

  if (isBaseWithConstantOffset(Root, MRI)) {
    MachineOperand &LHS = RootDef->getOperand(1);
    MachineOperand &RHS = RootDef->getOperand(2);
    MachineInstr *LHSDef = MRI.getVRegDef(LHS.getReg());
    MachineInstr *RHSDef = MRI.getVRegDef(RHS.getReg());

    int64_t RHSC = (int64_t)RHSDef->getOperand(1).getCImm()->getZExtValue();
    unsigned Scale = Log2_32(Size);
    if ((RHSC & (Size - 1)) == 0 && RHSC >= 0 && RHSC < (0x1000 << Scale)) {
      if (LHSDef->getOpcode() == TargetOpcode::G_FRAME_INDEX)
        return {{
            [=](MachineInstrBuilder &MIB) { MIB.add(LHSDef->getOperand(1)); },
            [=](MachineInstrBuilder &MIB) { MIB.addImm(RHSC >> Scale); },
        }};

      return {{
          [=](MachineInstrBuilder &MIB) { MIB.add(LHS); },
          [=](MachineInstrBuilder &MIB) { MIB.addImm(RHSC >> Scale); },
      }};
    }
  }

  // Before falling back to our general case, check if the unscaled
  // instructions can handle this. If so, that's preferable.
  if (selectAddrModeUnscaled(Root, Size))
    return std::nullopt;

  return {{
      [=](MachineInstrBuilder &MIB) { MIB.add(Root); },
      [=](MachineInstrBuilder &MIB) { MIB.addImm(0); },
  }};
}

/// Given a shift instruction, return the correct shift type for that
/// instruction.
static AArch64_AM::ShiftExtendType getShiftTypeForInst(MachineInstr &MI) {
  switch (MI.getOpcode()) {
  default:
    return AArch64_AM::InvalidShiftExtend;
  case TargetOpcode::G_SHL:
    return AArch64_AM::LSL;
  case TargetOpcode::G_LSHR:
    return AArch64_AM::LSR;
  case TargetOpcode::G_ASHR:
    return AArch64_AM::ASR;
  case TargetOpcode::G_ROTR:
    return AArch64_AM::ROR;
  }
}

/// Select a "shifted register" operand. If the value is not shifted, set the
/// shift operand to a default value of "lsl 0".
InstructionSelector::ComplexRendererFns
AArch64InstructionSelector::selectShiftedRegister(MachineOperand &Root,
                                                  bool AllowROR) const {
  if (!Root.isReg())
    return std::nullopt;
  MachineRegisterInfo &MRI =
      Root.getParent()->getParent()->getParent()->getRegInfo();

  // Check if the operand is defined by an instruction which corresponds to
  // a ShiftExtendType. E.g. a G_SHL, G_LSHR, etc.
  MachineInstr *ShiftInst = MRI.getVRegDef(Root.getReg());
  AArch64_AM::ShiftExtendType ShType = getShiftTypeForInst(*ShiftInst);
  if (ShType == AArch64_AM::InvalidShiftExtend)
    return std::nullopt;
  if (ShType == AArch64_AM::ROR && !AllowROR)
    return std::nullopt;
  if (!isWorthFoldingIntoExtendedReg(*ShiftInst, MRI))
    return std::nullopt;

  // Need an immediate on the RHS.
  MachineOperand &ShiftRHS = ShiftInst->getOperand(2);
  auto Immed = getImmedFromMO(ShiftRHS);
  if (!Immed)
    return std::nullopt;

  // We have something that we can fold. Fold in the shift's LHS and RHS into
  // the instruction.
  MachineOperand &ShiftLHS = ShiftInst->getOperand(1);
  Register ShiftReg = ShiftLHS.getReg();

  unsigned NumBits = MRI.getType(ShiftReg).getSizeInBits();
  unsigned Val = *Immed & (NumBits - 1);
  unsigned ShiftVal = AArch64_AM::getShifterImm(ShType, Val);

  return {{[=](MachineInstrBuilder &MIB) { MIB.addUse(ShiftReg); },
           [=](MachineInstrBuilder &MIB) { MIB.addImm(ShiftVal); }}};
}

AArch64_AM::ShiftExtendType AArch64InstructionSelector::getExtendTypeForInst(
    MachineInstr &MI, MachineRegisterInfo &MRI, bool IsLoadStore) const {
  unsigned Opc = MI.getOpcode();

  // Handle explicit extend instructions first.
  if (Opc == TargetOpcode::G_SEXT || Opc == TargetOpcode::G_SEXT_INREG) {
    unsigned Size;
    if (Opc == TargetOpcode::G_SEXT)
      Size = MRI.getType(MI.getOperand(1).getReg()).getSizeInBits();
    else
      Size = MI.getOperand(2).getImm();
    assert(Size != 64 && "Extend from 64 bits?");
    switch (Size) {
    case 8:
      return IsLoadStore ? AArch64_AM::InvalidShiftExtend : AArch64_AM::SXTB;
    case 16:
      return IsLoadStore ? AArch64_AM::InvalidShiftExtend : AArch64_AM::SXTH;
    case 32:
      return AArch64_AM::SXTW;
    default:
      return AArch64_AM::InvalidShiftExtend;
    }
  }

  if (Opc == TargetOpcode::G_ZEXT || Opc == TargetOpcode::G_ANYEXT) {
    unsigned Size = MRI.getType(MI.getOperand(1).getReg()).getSizeInBits();
    assert(Size != 64 && "Extend from 64 bits?");
    switch (Size) {
    case 8:
      return IsLoadStore ? AArch64_AM::InvalidShiftExtend : AArch64_AM::UXTB;
    case 16:
      return IsLoadStore ? AArch64_AM::InvalidShiftExtend : AArch64_AM::UXTH;
    case 32:
      return AArch64_AM::UXTW;
    default:
      return AArch64_AM::InvalidShiftExtend;
    }
  }

  // Don't have an explicit extend. Try to handle a G_AND with a constant mask
  // on the RHS.
  if (Opc != TargetOpcode::G_AND)
    return AArch64_AM::InvalidShiftExtend;

  std::optional<uint64_t> MaybeAndMask = getImmedFromMO(MI.getOperand(2));
  if (!MaybeAndMask)
    return AArch64_AM::InvalidShiftExtend;
  uint64_t AndMask = *MaybeAndMask;
  switch (AndMask) {
  default:
    return AArch64_AM::InvalidShiftExtend;
  case 0xFF:
    return !IsLoadStore ? AArch64_AM::UXTB : AArch64_AM::InvalidShiftExtend;
  case 0xFFFF:
    return !IsLoadStore ? AArch64_AM::UXTH : AArch64_AM::InvalidShiftExtend;
  case 0xFFFFFFFF:
    return AArch64_AM::UXTW;
  }
}

Register AArch64InstructionSelector::moveScalarRegClass(
    Register Reg, const TargetRegisterClass &RC, MachineIRBuilder &MIB) const {
  MachineRegisterInfo &MRI = *MIB.getMRI();
  auto Ty = MRI.getType(Reg);
  assert(!Ty.isVector() && "Expected scalars only!");
  if (Ty.getSizeInBits() == TRI.getRegSizeInBits(RC))
    return Reg;

  // Create a copy and immediately select it.
  // FIXME: We should have an emitCopy function?
  auto Copy = MIB.buildCopy({&RC}, {Reg});
  selectCopy(*Copy, TII, MRI, TRI, RBI);
  return Copy.getReg(0);
}

/// Select an "extended register" operand. This operand folds in an extend
/// followed by an optional left shift.
InstructionSelector::ComplexRendererFns
AArch64InstructionSelector::selectArithExtendedRegister(
    MachineOperand &Root) const {
  if (!Root.isReg())
    return std::nullopt;
  MachineRegisterInfo &MRI =
      Root.getParent()->getParent()->getParent()->getRegInfo();

  uint64_t ShiftVal = 0;
  Register ExtReg;
  AArch64_AM::ShiftExtendType Ext;
  MachineInstr *RootDef = getDefIgnoringCopies(Root.getReg(), MRI);
  if (!RootDef)
    return std::nullopt;

  if (!isWorthFoldingIntoExtendedReg(*RootDef, MRI))
    return std::nullopt;

  // Check if we can fold a shift and an extend.
  if (RootDef->getOpcode() == TargetOpcode::G_SHL) {
    // Look for a constant on the RHS of the shift.
    MachineOperand &RHS = RootDef->getOperand(2);
    std::optional<uint64_t> MaybeShiftVal = getImmedFromMO(RHS);
    if (!MaybeShiftVal)
      return std::nullopt;
    ShiftVal = *MaybeShiftVal;
    if (ShiftVal > 4)
      return std::nullopt;
    // Look for a valid extend instruction on the LHS of the shift.
    MachineOperand &LHS = RootDef->getOperand(1);
    MachineInstr *ExtDef = getDefIgnoringCopies(LHS.getReg(), MRI);
    if (!ExtDef)
      return std::nullopt;
    Ext = getExtendTypeForInst(*ExtDef, MRI);
    if (Ext == AArch64_AM::InvalidShiftExtend)
      return std::nullopt;
    ExtReg = ExtDef->getOperand(1).getReg();
  } else {
    // Didn't get a shift. Try just folding an extend.
    Ext = getExtendTypeForInst(*RootDef, MRI);
    if (Ext == AArch64_AM::InvalidShiftExtend)
      return std::nullopt;
    ExtReg = RootDef->getOperand(1).getReg();

    // If we have a 32 bit instruction which zeroes out the high half of a
    // register, we get an implicit zero extend for free. Check if we have one.
    // FIXME: We actually emit the extend right now even though we don't have
    // to.
    if (Ext == AArch64_AM::UXTW && MRI.getType(ExtReg).getSizeInBits() == 32) {
      MachineInstr *ExtInst = MRI.getVRegDef(ExtReg);
      if (isDef32(*ExtInst))
        return std::nullopt;
    }
  }

  // We require a GPR32 here. Narrow the ExtReg if needed using a subregister
  // copy.
  MachineIRBuilder MIB(*RootDef);
  ExtReg = moveScalarRegClass(ExtReg, AArch64::GPR32RegClass, MIB);

  return {{[=](MachineInstrBuilder &MIB) { MIB.addUse(ExtReg); },
           [=](MachineInstrBuilder &MIB) {
             MIB.addImm(getArithExtendImm(Ext, ShiftVal));
           }}};
}

void AArch64InstructionSelector::renderTruncImm(MachineInstrBuilder &MIB,
                                                const MachineInstr &MI,
                                                int OpIdx) const {
  const MachineRegisterInfo &MRI = MI.getParent()->getParent()->getRegInfo();
  assert(MI.getOpcode() == TargetOpcode::G_CONSTANT && OpIdx == -1 &&
         "Expected G_CONSTANT");
  std::optional<int64_t> CstVal =
      getIConstantVRegSExtVal(MI.getOperand(0).getReg(), MRI);
  assert(CstVal && "Expected constant value");
  MIB.addImm(*CstVal);
}

void AArch64InstructionSelector::renderLogicalImm32(
  MachineInstrBuilder &MIB, const MachineInstr &I, int OpIdx) const {
  assert(I.getOpcode() == TargetOpcode::G_CONSTANT && OpIdx == -1 &&
         "Expected G_CONSTANT");
  uint64_t CstVal = I.getOperand(1).getCImm()->getZExtValue();
  uint64_t Enc = AArch64_AM::encodeLogicalImmediate(CstVal, 32);
  MIB.addImm(Enc);
}

void AArch64InstructionSelector::renderLogicalImm64(
  MachineInstrBuilder &MIB, const MachineInstr &I, int OpIdx) const {
  assert(I.getOpcode() == TargetOpcode::G_CONSTANT && OpIdx == -1 &&
         "Expected G_CONSTANT");
  uint64_t CstVal = I.getOperand(1).getCImm()->getZExtValue();
  uint64_t Enc = AArch64_AM::encodeLogicalImmediate(CstVal, 64);
  MIB.addImm(Enc);
}

void AArch64InstructionSelector::renderFPImm16(MachineInstrBuilder &MIB,
                                               const MachineInstr &MI,
                                               int OpIdx) const {
  assert(MI.getOpcode() == TargetOpcode::G_FCONSTANT && OpIdx == -1 &&
         "Expected G_FCONSTANT");
  MIB.addImm(
      AArch64_AM::getFP16Imm(MI.getOperand(1).getFPImm()->getValueAPF()));
}

void AArch64InstructionSelector::renderFPImm32(MachineInstrBuilder &MIB,
                                               const MachineInstr &MI,
                                               int OpIdx) const {
  assert(MI.getOpcode() == TargetOpcode::G_FCONSTANT && OpIdx == -1 &&
         "Expected G_FCONSTANT");
  MIB.addImm(
      AArch64_AM::getFP32Imm(MI.getOperand(1).getFPImm()->getValueAPF()));
}

void AArch64InstructionSelector::renderFPImm64(MachineInstrBuilder &MIB,
                                               const MachineInstr &MI,
                                               int OpIdx) const {
  assert(MI.getOpcode() == TargetOpcode::G_FCONSTANT && OpIdx == -1 &&
         "Expected G_FCONSTANT");
  MIB.addImm(
      AArch64_AM::getFP64Imm(MI.getOperand(1).getFPImm()->getValueAPF()));
}

void AArch64InstructionSelector::renderFPImm32SIMDModImmType4(
    MachineInstrBuilder &MIB, const MachineInstr &MI, int OpIdx) const {
  assert(MI.getOpcode() == TargetOpcode::G_FCONSTANT && OpIdx == -1 &&
         "Expected G_FCONSTANT");
  MIB.addImm(AArch64_AM::encodeAdvSIMDModImmType4(MI.getOperand(1)
                                                      .getFPImm()
                                                      ->getValueAPF()
                                                      .bitcastToAPInt()
                                                      .getZExtValue()));
}

bool AArch64InstructionSelector::isLoadStoreOfNumBytes(
    const MachineInstr &MI, unsigned NumBytes) const {
  if (!MI.mayLoadOrStore())
    return false;
  assert(MI.hasOneMemOperand() &&
         "Expected load/store to have only one mem op!");
  return (*MI.memoperands_begin())->getSize() == NumBytes;
}

bool AArch64InstructionSelector::isDef32(const MachineInstr &MI) const {
  const MachineRegisterInfo &MRI = MI.getParent()->getParent()->getRegInfo();
  if (MRI.getType(MI.getOperand(0).getReg()).getSizeInBits() != 32)
    return false;

  // Only return true if we know the operation will zero-out the high half of
  // the 64-bit register. Truncates can be subregister copies, which don't
  // zero out the high bits. Copies and other copy-like instructions can be
  // fed by truncates, or could be lowered as subregister copies.
  switch (MI.getOpcode()) {
  default:
    return true;
  case TargetOpcode::COPY:
  case TargetOpcode::G_BITCAST:
  case TargetOpcode::G_TRUNC:
  case TargetOpcode::G_PHI:
    return false;
  }
}


// Perform fixups on the given PHI instruction's operands to force them all
// to be the same as the destination regbank.
static void fixupPHIOpBanks(MachineInstr &MI, MachineRegisterInfo &MRI,
                            const AArch64RegisterBankInfo &RBI) {
  assert(MI.getOpcode() == TargetOpcode::G_PHI && "Expected a G_PHI");
  Register DstReg = MI.getOperand(0).getReg();
  const RegisterBank *DstRB = MRI.getRegBankOrNull(DstReg);
  assert(DstRB && "Expected PHI dst to have regbank assigned");
  MachineIRBuilder MIB(MI);

  // Go through each operand and ensure it has the same regbank.
  for (MachineOperand &MO : llvm::drop_begin(MI.operands())) {
    if (!MO.isReg())
      continue;
    Register OpReg = MO.getReg();
    const RegisterBank *RB = MRI.getRegBankOrNull(OpReg);
    if (RB != DstRB) {
      // Insert a cross-bank copy.
      auto *OpDef = MRI.getVRegDef(OpReg);
      const LLT &Ty = MRI.getType(OpReg);
      MachineBasicBlock &OpDefBB = *OpDef->getParent();

      // Any instruction we insert must appear after all PHIs in the block
      // for the block to be valid MIR.
      MachineBasicBlock::iterator InsertPt = std::next(OpDef->getIterator());
      if (InsertPt != OpDefBB.end() && InsertPt->isPHI())
        InsertPt = OpDefBB.getFirstNonPHI();
      MIB.setInsertPt(*OpDef->getParent(), InsertPt);
      auto Copy = MIB.buildCopy(Ty, OpReg);
      MRI.setRegBank(Copy.getReg(0), *DstRB);
      MO.setReg(Copy.getReg(0));
    }
  }
}

void AArch64InstructionSelector::processPHIs(MachineFunction &MF) {
  // We're looking for PHIs, build a list so we don't invalidate iterators.
  MachineRegisterInfo &MRI = MF.getRegInfo();
  SmallVector<MachineInstr *, 32> Phis;
  for (auto &BB : MF) {
    for (auto &MI : BB) {
      if (MI.getOpcode() == TargetOpcode::G_PHI)
        Phis.emplace_back(&MI);
    }
  }

  for (auto *MI : Phis) {
    // We need to do some work here if the operand types are < 16 bit and they
    // are split across fpr/gpr banks. Since all types <32b on gpr
    // end up being assigned gpr32 regclasses, we can end up with PHIs here
    // which try to select between a gpr32 and an fpr16. Ideally RBS shouldn't
    // be selecting heterogenous regbanks for operands if possible, but we
    // still need to be able to deal with it here.
    //
    // To fix this, if we have a gpr-bank operand < 32b in size and at least
    // one other operand is on the fpr bank, then we add cross-bank copies
    // to homogenize the operand banks. For simplicity the bank that we choose
    // to settle on is whatever bank the def operand has. For example:
    //
    // %endbb:
    //   %dst:gpr(s16) = G_PHI %in1:gpr(s16), %bb1, %in2:fpr(s16), %bb2
    //  =>
    // %bb2:
    //   ...
    //   %in2_copy:gpr(s16) = COPY %in2:fpr(s16)
    //   ...
    // %endbb:
    //   %dst:gpr(s16) = G_PHI %in1:gpr(s16), %bb1, %in2_copy:gpr(s16), %bb2
    bool HasGPROp = false, HasFPROp = false;
    for (const MachineOperand &MO : llvm::drop_begin(MI->operands())) {
      if (!MO.isReg())
        continue;
      const LLT &Ty = MRI.getType(MO.getReg());
      if (!Ty.isValid() || !Ty.isScalar())
        break;
      if (Ty.getSizeInBits() >= 32)
        break;
      const RegisterBank *RB = MRI.getRegBankOrNull(MO.getReg());
      // If for some reason we don't have a regbank yet. Don't try anything.
      if (!RB)
        break;

      if (RB->getID() == AArch64::GPRRegBankID)
        HasGPROp = true;
      else
        HasFPROp = true;
    }
    // We have heterogenous regbanks, need to fixup.
    if (HasGPROp && HasFPROp)
      fixupPHIOpBanks(*MI, MRI, RBI);
  }
}

namespace llvm {
InstructionSelector *
createAArch64InstructionSelector(const AArch64TargetMachine &TM,
                                 AArch64Subtarget &Subtarget,
                                 AArch64RegisterBankInfo &RBI) {
  return new AArch64InstructionSelector(TM, Subtarget, RBI);
}
}