/// AddOperand - Add the specified operand to the specified machine instr.  II
/// specifies the instruction information for the node, and IIOpNum is the
/// operand number (in the II) that we are adding. IIOpNum and II are used for
/// assertions only.
void InstrEmitter::AddOperand(MachineInstr *MI, SDValue Op,
                              unsigned IIOpNum,
                              const MCInstrDesc *II,
                              DenseMap<SDValue, unsigned> &VRBaseMap,
                              bool IsDebug, bool IsClone, bool IsCloned) {
    if (Op.isMachineOpcode()) {
        AddRegisterOperand(MI, Op, IIOpNum, II, VRBaseMap,
                           IsDebug, IsClone, IsCloned);
    } else if (ConstantSDNode *C = dyn_cast<ConstantSDNode>(Op)) {
        MI->addOperand(MachineOperand::CreateImm(C->getSExtValue()));
    } else if (ConstantFPSDNode *F = dyn_cast<ConstantFPSDNode>(Op)) {
        const ConstantFP *CFP = F->getConstantFPValue();
        MI->addOperand(MachineOperand::CreateFPImm(CFP));
    } else if (RegisterSDNode *R = dyn_cast<RegisterSDNode>(Op)) {
        MI->addOperand(MachineOperand::CreateReg(R->getReg(), false));
    } else if (RegisterMaskSDNode *RM = dyn_cast<RegisterMaskSDNode>(Op)) {
        MI->addOperand(MachineOperand::CreateRegMask(RM->getRegMask()));
    } else if (GlobalAddressSDNode *TGA = dyn_cast<GlobalAddressSDNode>(Op)) {
        MI->addOperand(MachineOperand::CreateGA(TGA->getGlobal(), TGA->getOffset(),
                                                TGA->getTargetFlags()));
    } else if (BasicBlockSDNode *BBNode = dyn_cast<BasicBlockSDNode>(Op)) {
        MI->addOperand(MachineOperand::CreateMBB(BBNode->getBasicBlock()));
    } else if (FrameIndexSDNode *FI = dyn_cast<FrameIndexSDNode>(Op)) {
        MI->addOperand(MachineOperand::CreateFI(FI->getIndex()));
    } else if (JumpTableSDNode *JT = dyn_cast<JumpTableSDNode>(Op)) {
        MI->addOperand(MachineOperand::CreateJTI(JT->getIndex(),
                       JT->getTargetFlags()));
    } else if (ConstantPoolSDNode *CP = dyn_cast<ConstantPoolSDNode>(Op)) {
        int Offset = CP->getOffset();
        unsigned Align = CP->getAlignment();
        Type *Type = CP->getType();
        // MachineConstantPool wants an explicit alignment.
        if (Align == 0) {
            Align = TM->getTargetData()->getPrefTypeAlignment(Type);
            if (Align == 0) {
                // Alignment of vector types.  FIXME!
                Align = TM->getTargetData()->getTypeAllocSize(Type);
            }
        }

        unsigned Idx;
        MachineConstantPool *MCP = MF->getConstantPool();
        if (CP->isMachineConstantPoolEntry())
            Idx = MCP->getConstantPoolIndex(CP->getMachineCPVal(), Align);
        else
            Idx = MCP->getConstantPoolIndex(CP->getConstVal(), Align);
        MI->addOperand(MachineOperand::CreateCPI(Idx, Offset,
                       CP->getTargetFlags()));
    } else if (ExternalSymbolSDNode *ES = dyn_cast<ExternalSymbolSDNode>(Op)) {
        MI->addOperand(MachineOperand::CreateES(ES->getSymbol(),
                                                ES->getTargetFlags()));
    } else if (BlockAddressSDNode *BA = dyn_cast<BlockAddressSDNode>(Op)) {
        MI->addOperand(MachineOperand::CreateBA(BA->getBlockAddress(),
                                                BA->getTargetFlags()));
    } else {
        assert(Op.getValueType() != MVT::Other &&
               Op.getValueType() != MVT::Glue &&
               "Chain and glue operands should occur at end of operand list!");
        AddRegisterOperand(MI, Op, IIOpNum, II, VRBaseMap,
                           IsDebug, IsClone, IsCloned);
    }
}

SDValue
AlphaTargetLowering::LowerReturn(SDValue Chain,
                                 CallingConv::ID CallConv, bool isVarArg,
                                 const SmallVectorImpl<ISD::OutputArg> &Outs,
                                 DebugLoc dl, SelectionDAG &DAG) {

  SDValue Copy = DAG.getCopyToReg(Chain, dl, Alpha::R26,
                                  DAG.getNode(AlphaISD::GlobalRetAddr,
                                              DebugLoc::getUnknownLoc(),
                                              MVT::i64),
                                  SDValue());
  switch (Outs.size()) {
  default:
    llvm_unreachable("Do not know how to return this many arguments!");
  case 0:
    break;
    //return SDValue(); // ret void is legal
  case 1: {
    EVT ArgVT = Outs[0].Val.getValueType();
    unsigned ArgReg;
    if (ArgVT.isInteger())
      ArgReg = Alpha::R0;
    else {
      assert(ArgVT.isFloatingPoint());
      ArgReg = Alpha::F0;
    }
    Copy = DAG.getCopyToReg(Copy, dl, ArgReg,
                            Outs[0].Val, Copy.getValue(1));
    if (DAG.getMachineFunction().getRegInfo().liveout_empty())
      DAG.getMachineFunction().getRegInfo().addLiveOut(ArgReg);
    break;
  }
  case 2: {
    EVT ArgVT = Outs[0].Val.getValueType();
    unsigned ArgReg1, ArgReg2;
    if (ArgVT.isInteger()) {
      ArgReg1 = Alpha::R0;
      ArgReg2 = Alpha::R1;
    } else {
      assert(ArgVT.isFloatingPoint());
      ArgReg1 = Alpha::F0;
      ArgReg2 = Alpha::F1;
    }
    Copy = DAG.getCopyToReg(Copy, dl, ArgReg1,
                            Outs[0].Val, Copy.getValue(1));
    if (std::find(DAG.getMachineFunction().getRegInfo().liveout_begin(),
                  DAG.getMachineFunction().getRegInfo().liveout_end(), ArgReg1)
        == DAG.getMachineFunction().getRegInfo().liveout_end())
      DAG.getMachineFunction().getRegInfo().addLiveOut(ArgReg1);
    Copy = DAG.getCopyToReg(Copy, dl, ArgReg2,
                            Outs[1].Val, Copy.getValue(1));
    if (std::find(DAG.getMachineFunction().getRegInfo().liveout_begin(),
                   DAG.getMachineFunction().getRegInfo().liveout_end(), ArgReg2)
        == DAG.getMachineFunction().getRegInfo().liveout_end())
      DAG.getMachineFunction().getRegInfo().addLiveOut(ArgReg2);
    break;
  }
  }
  return DAG.getNode(AlphaISD::RET_FLAG, dl,
                     MVT::Other, Copy, Copy.getValue(1));
}

/// ComplexPattern used on MipsInstrInfo
/// Used on Mips Load/Store instructions
bool MipsDAGToDAGISel::
SelectAddr(SDNode *Parent, SDValue Addr, SDValue &Base, SDValue &Offset) {
  EVT ValTy = Addr.getValueType();

  // if Address is FI, get the TargetFrameIndex.
  if (FrameIndexSDNode *FIN = dyn_cast<FrameIndexSDNode>(Addr)) {
    Base   = CurDAG->getTargetFrameIndex(FIN->getIndex(), ValTy);
    Offset = CurDAG->getTargetConstant(0, ValTy);
    return true;
  }

  // on PIC code Load GA
  if (Addr.getOpcode() == MipsISD::Wrapper) {
    Base   = Addr.getOperand(0);
    Offset = Addr.getOperand(1);
    return true;
  }

  if (TM.getRelocationModel() != Reloc::PIC_) {
    if ((Addr.getOpcode() == ISD::TargetExternalSymbol ||
        Addr.getOpcode() == ISD::TargetGlobalAddress))
      return false;
  }

  // Addresses of the form FI+const or FI|const
  if (CurDAG->isBaseWithConstantOffset(Addr)) {
    ConstantSDNode *CN = dyn_cast<ConstantSDNode>(Addr.getOperand(1));
    if (isInt<16>(CN->getSExtValue())) {

      // If the first operand is a FI, get the TargetFI Node
      if (FrameIndexSDNode *FIN = dyn_cast<FrameIndexSDNode>
                                  (Addr.getOperand(0)))
        Base = CurDAG->getTargetFrameIndex(FIN->getIndex(), ValTy);
      else
        Base = Addr.getOperand(0);

      Offset = CurDAG->getTargetConstant(CN->getZExtValue(), ValTy);
      return true;
    }
  }

  // Operand is a result from an ADD.
  if (Addr.getOpcode() == ISD::ADD) {
    // When loading from constant pools, load the lower address part in
    // the instruction itself. Example, instead of:
    //  lui $2, %hi($CPI1_0)
    //  addiu $2, $2, %lo($CPI1_0)
    //  lwc1 $f0, 0($2)
    // Generate:
    //  lui $2, %hi($CPI1_0)
    //  lwc1 $f0, %lo($CPI1_0)($2)
    if (Addr.getOperand(1).getOpcode() == MipsISD::Lo) {
      SDValue LoVal = Addr.getOperand(1), Opnd0 = LoVal.getOperand(0);
      if (isa<ConstantPoolSDNode>(Opnd0) || isa<GlobalAddressSDNode>(Opnd0) ||
          isa<JumpTableSDNode>(Opnd0)) {
        Base = Addr.getOperand(0);
        Offset = Opnd0;
        return true;
      }
    }

    // If an indexed floating point load/store can be emitted, return false.
    const LSBaseSDNode *LS = dyn_cast<LSBaseSDNode>(Parent);

    if (LS &&
        (LS->getMemoryVT() == MVT::f32 || LS->getMemoryVT() == MVT::f64) &&
        Subtarget.hasMips32r2Or64())
      return false;
  }

  Base   = Addr;
  Offset = CurDAG->getTargetConstant(0, ValTy);
  return true;
}

/// LowerCCCCallTo - functions arguments are copied from virtual regs to
/// (physical regs)/(stack frame), CALLSEQ_START and CALLSEQ_END are emitted.
/// TODO: sret.
SDValue
NvfcTargetLowering::LowerCCCCallTo(SDValue Chain, SDValue Callee,
                                     CallingConv::ID CallConv, bool isVarArg,
                                     bool isTailCall,
                                     const SmallVectorImpl<ISD::OutputArg>
                                       &Outs,
                                     const SmallVectorImpl<SDValue> &OutVals,
                                     const SmallVectorImpl<ISD::InputArg> &Ins,
                                     DebugLoc dl, SelectionDAG &DAG,
                                     SmallVectorImpl<SDValue> &InVals) const {
  // Analyze operands of the call, assigning locations to each operand.
  SmallVector<CCValAssign, 16> ArgLocs;
  CCState CCInfo(CallConv, isVarArg, getTargetMachine(),
                 ArgLocs, *DAG.getContext());

  CCInfo.AnalyzeCallOperands(Outs, CC_NVFUC);

  // Get a count of how many bytes are to be pushed on the stack.
  unsigned NumBytes = CCInfo.getNextStackOffset();

  Chain = DAG.getCALLSEQ_START(Chain ,DAG.getConstant(NumBytes,
                                                      getPointerTy(), true));

  SmallVector<std::pair<unsigned, SDValue>, 4> RegsToPass;
  SmallVector<SDValue, 12> MemOpChains;
  SDValue StackPtr;

  // Walk the register/memloc assignments, inserting copies/loads.
  for (unsigned i = 0, e = ArgLocs.size(); i != e; ++i) {
    CCValAssign &VA = ArgLocs[i];

    SDValue Arg = OutVals[i];

    // Promote the value if needed.
    switch (VA.getLocInfo()) {
      default: llvm_unreachable("Unknown loc info!");
      case CCValAssign::Full: break;
    }

    // Arguments that can be passed on register must be kept at RegsToPass
    // vector
    if (VA.isRegLoc()) {
      RegsToPass.push_back(std::make_pair(VA.getLocReg(), Arg));
    } else {
      assert(VA.isMemLoc());

      // BUG: Special registers cannot be accessed...
      if (StackPtr.getNode() == 0)
        StackPtr = DAG.getCopyFromReg(Chain, dl, Nvfc::SP, getPointerTy());

      SDValue PtrOff = DAG.getNode(ISD::ADD, dl, getPointerTy(),
                                   StackPtr,
                                   DAG.getIntPtrConstant(VA.getLocMemOffset()));


      MemOpChains.push_back(DAG.getStore(Chain, dl, Arg, PtrOff,
                                         PseudoSourceValue::getStack(),
                                         VA.getLocMemOffset(), false, false, 0));
    }
  }

  // Transform all store nodes into one single node because all store nodes are
  // independent of each other.
  if (!MemOpChains.empty())
    Chain = DAG.getNode(ISD::TokenFactor, dl, MVT::Other,
                        &MemOpChains[0], MemOpChains.size());

  // Build a sequence of copy-to-reg nodes chained together with token chain and
  // flag operands which copy the outgoing args into registers.  The InFlag in
  // necessary since all emited instructions must be stuck together.
  SDValue InFlag;
  for (unsigned i = 0, e = RegsToPass.size(); i != e; ++i) {
    Chain = DAG.getCopyToReg(Chain, dl, RegsToPass[i].first,
                             RegsToPass[i].second, InFlag);
    InFlag = Chain.getValue(1);
  }

  // If the callee is a GlobalAddress node (quite common, every direct call is)
  // turn it into a TargetGlobalAddress node so that legalize doesn't hack it.
  // Likewise ExternalSymbol -> TargetExternalSymbol.
  if (GlobalAddressSDNode *G = dyn_cast<GlobalAddressSDNode>(Callee))
    Callee = DAG.getTargetGlobalAddress(G->getGlobal(), dl, MVT::i32);
  else if (ExternalSymbolSDNode *E = dyn_cast<ExternalSymbolSDNode>(Callee))
    Callee = DAG.getTargetExternalSymbol(E->getSymbol(), MVT::i32);

  // Returns a chain & a flag for retval copy to use.
  SDVTList NodeTys = DAG.getVTList(MVT::Other, MVT::Flag);
  SmallVector<SDValue, 8> Ops;  
  Ops.push_back(Chain);
  Ops.push_back(Callee);

  // Add argument registers to the end of the list so that they are
  // known live into the call.
  for (unsigned i = 0, e = RegsToPass.size(); i != e; ++i)
    Ops.push_back(DAG.getRegister(RegsToPass[i].first,
                                  RegsToPass[i].second.getValueType()));

//.........这里部分代码省略.........

SDValue NvfcTargetLowering::LowerSETCC(SDValue Op, SelectionDAG &DAG) const {
  SDValue LHS   = Op.getOperand(0);
  SDValue RHS   = Op.getOperand(1);
  DebugLoc dl   = Op.getDebugLoc();

  // If we are doing an AND and testing against zero, then the CMP
  // will not be generated.  The AND (or OR) will generate the condition codes,
  // but they are different from CMP.
  // FIXME: since we're doing a post-processing, use a pseudoinstr here, so
  // lowering & isel wouldn't diverge.
  bool andCC = false;
  if (ConstantSDNode *RHSC = dyn_cast<ConstantSDNode>(RHS)) {
    if (RHSC->isNullValue() && LHS.hasOneUse() &&
        (LHS.getOpcode() == ISD::AND ||
         (LHS.getOpcode() == ISD::TRUNCATE &&
          LHS.getOperand(0).getOpcode() == ISD::AND))) {
      andCC = true;
    }
  }
  ISD::CondCode CC = cast<CondCodeSDNode>(Op.getOperand(2))->get();
  SDValue TargetCC;
  SDValue Flag = EmitCMP(LHS, RHS, TargetCC, CC, dl, DAG);

  // Get the condition codes directly from the status register, if its easy.
  // Otherwise a branch will be generated.  Note that the AND and OR
  // instructions generate different flags than CMP, the carry bit can be used
  // for NE/EQ.
  bool Invert = false;
  bool Convert = true;
  int Shamt = 0; 
  switch (cast<ConstantSDNode>(TargetCC)->getZExtValue()) {
   default:
    Convert = false;
    break;
   case NvfcCC::COND_HS:
     // Res = (FLG >> 8) & 1
     Shamt = 8;
     break;
   case NvfcCC::COND_LO:
     // Res = ~((FLG >> 8) & 1)
     Shamt = 8;    
     Invert = true;
     break;
   case NvfcCC::COND_NE:
     if (andCC) {
       // C = ~Z, thus Res = (FLG >> 8) & 1
       Shamt = 8;
     } else {
       // Res = ~((FLG >> 0xb) & 1)
       Shamt = 0xb;
       Invert = true;
     }
     break;
   case NvfcCC::COND_E:
     // Res = (FLG >> 0xb) & 1
     Shamt = 0xb;
     break;
  }
  EVT VT = Op.getValueType();
  SDValue One  = DAG.getConstant(1, VT);
  if (Convert) {
    SDValue Shift = DAG.getConstant(Shamt, VT);
    SDValue FLG = DAG.getCopyFromReg(DAG.getEntryNode(), dl, Nvfc::FLG,
    				     MVT::i32, Flag);
    FLG = DAG.getNode(NvfcISD::XBIT, dl, MVT::i32, FLG, Shift, Flag);
    if (Invert) {
	FLG = DAG.getNode(ISD::XOR, dl, MVT::i32, FLG, One);
	//FLG = DAG.getNode(ISD::NOT, dl, MVT::i32, FLG);
    }
    
    return FLG;
  } else {
    SDValue Zero = DAG.getConstant(0, VT);
    SDVTList VTs = DAG.getVTList(Op.getValueType(), MVT::Flag);
    SmallVector<SDValue, 4> Ops;
    Ops.push_back(One);
    Ops.push_back(Zero);
    Ops.push_back(TargetCC);
    Ops.push_back(Flag);
    return DAG.getNode(NvfcISD::SELECT_CC, dl, VTs, &Ops[0], Ops.size());
  }
}

bool SparcDAGToDAGISel::SelectADDRri(SDValue Addr,
                                     SDValue &Base, SDValue &Offset) {
  if (FrameIndexSDNode *FIN = dyn_cast<FrameIndexSDNode>(Addr)) {
    Base = CurDAG->getTargetFrameIndex(FIN->getIndex(),
                                       getTargetLowering()->getPointerTy());
    Offset = CurDAG->getTargetConstant(0, MVT::i32);
    return true;
  }
  if (Addr.getOpcode() == ISD::TargetExternalSymbol ||
      Addr.getOpcode() == ISD::TargetGlobalAddress ||
      Addr.getOpcode() == ISD::TargetGlobalTLSAddress)
    return false;  // direct calls.

  if (Addr.getOpcode() == ISD::ADD) {
    if (ConstantSDNode *CN = dyn_cast<ConstantSDNode>(Addr.getOperand(1))) {
      if (isInt<13>(CN->getSExtValue())) {
        if (FrameIndexSDNode *FIN =
                dyn_cast<FrameIndexSDNode>(Addr.getOperand(0))) {
          // Constant offset from frame ref.
          Base = CurDAG->getTargetFrameIndex(FIN->getIndex(),
                                           getTargetLowering()->getPointerTy());
        } else {
          Base = Addr.getOperand(0);
        }
        Offset = CurDAG->getTargetConstant(CN->getZExtValue(), MVT::i32);
        return true;
      }
    }
    if (Addr.getOperand(0).getOpcode() == SPISD::Lo) {
      Base = Addr.getOperand(1);
      Offset = Addr.getOperand(0).getOperand(0);
      return true;
    }
    if (Addr.getOperand(1).getOpcode() == SPISD::Lo) {
      Base = Addr.getOperand(0);
      Offset = Addr.getOperand(1).getOperand(0);
      return true;
    }
  }
  Base = Addr;
  Offset = CurDAG->getTargetConstant(0, MVT::i32);
  return true;
}

void Mips16TargetLowering::
getOpndList(SmallVectorImpl<SDValue> &Ops,
            std::deque< std::pair<unsigned, SDValue> > &RegsToPass,
            bool IsPICCall, bool GlobalOrExternal, bool InternalLinkage,
            CallLoweringInfo &CLI, SDValue Callee, SDValue Chain) const {
  SelectionDAG &DAG = CLI.DAG;
  MachineFunction &MF = DAG.getMachineFunction();
  MipsFunctionInfo *FuncInfo = MF.getInfo<MipsFunctionInfo>();
  const char* Mips16HelperFunction = nullptr;
  bool NeedMips16Helper = false;

  if (Subtarget.inMips16HardFloat()) {
    //
    // currently we don't have symbols tagged with the mips16 or mips32
    // qualifier so we will assume that we don't know what kind it is.
    // and generate the helper
    //
    bool LookupHelper = true;
    if (ExternalSymbolSDNode *S = dyn_cast<ExternalSymbolSDNode>(CLI.Callee)) {
      Mips16Libcall Find = { RTLIB::UNKNOWN_LIBCALL, S->getSymbol() };

      if (std::binary_search(std::begin(HardFloatLibCalls),
                             std::end(HardFloatLibCalls), Find))
        LookupHelper = false;
      else {
        const char *Symbol = S->getSymbol();
        Mips16IntrinsicHelperType IntrinsicFind = { Symbol, "" };
        const Mips16HardFloatInfo::FuncSignature *Signature =
            Mips16HardFloatInfo::findFuncSignature(Symbol);
        if (!IsPICCall && (Signature && (FuncInfo->StubsNeeded.find(Symbol) ==
                                         FuncInfo->StubsNeeded.end()))) {
          FuncInfo->StubsNeeded[Symbol] = Signature;
          //
          // S2 is normally saved if the stub is for a function which
          // returns a float or double value and is not otherwise. This is
          // because more work is required after the function the stub
          // is calling completes, and so the stub cannot directly return
          // and the stub has no stack space to store the return address so
          // S2 is used for that purpose.
          // In order to take advantage of not saving S2, we need to also
          // optimize the call in the stub and this requires some further
          // functionality in MipsAsmPrinter which we don't have yet.
          // So for now we always save S2. The optimization will be done
          // in a follow-on patch.
          //
          if (1 || (Signature->RetSig != Mips16HardFloatInfo::NoFPRet))
            FuncInfo->setSaveS2();
        }
        // one more look at list of intrinsics
        const Mips16IntrinsicHelperType *Helper =
            std::lower_bound(std::begin(Mips16IntrinsicHelper),
                             std::end(Mips16IntrinsicHelper), IntrinsicFind);
        if (Helper != std::end(Mips16IntrinsicHelper) &&
            *Helper == IntrinsicFind) {
          Mips16HelperFunction = Helper->Helper;
          NeedMips16Helper = true;
          LookupHelper = false;
        }

      }
    } else if (GlobalAddressSDNode *G =
                   dyn_cast<GlobalAddressSDNode>(CLI.Callee)) {
      Mips16Libcall Find = { RTLIB::UNKNOWN_LIBCALL,
                             G->getGlobal()->getName().data() };

      if (std::binary_search(std::begin(HardFloatLibCalls),
                             std::end(HardFloatLibCalls), Find))
        LookupHelper = false;
    }
    if (LookupHelper)
      Mips16HelperFunction =
        getMips16HelperFunction(CLI.RetTy, CLI.getArgs(), NeedMips16Helper);
  }

  SDValue JumpTarget = Callee;

  // T9 should contain the address of the callee function if
  // -reloction-model=pic or it is an indirect call.
  if (IsPICCall || !GlobalOrExternal) {
    unsigned V0Reg = Mips::V0;
    if (NeedMips16Helper) {
      RegsToPass.push_front(std::make_pair(V0Reg, Callee));
      JumpTarget = DAG.getExternalSymbol(Mips16HelperFunction, getPointerTy());
      ExternalSymbolSDNode *S = cast<ExternalSymbolSDNode>(JumpTarget);
      JumpTarget = getAddrGlobal(S, JumpTarget.getValueType(), DAG,
                                 MipsII::MO_GOT, Chain,
                                 FuncInfo->callPtrInfo(S->getSymbol()));
    } else
      RegsToPass.push_front(std::make_pair((unsigned)Mips::T9, Callee));
  }

  Ops.push_back(JumpTarget);

  MipsTargetLowering::getOpndList(Ops, RegsToPass, IsPICCall, GlobalOrExternal,
                                  InternalLinkage, CLI, Callee, Chain);
}

SDValue X86SelectionDAGInfo::EmitTargetCodeForMemcpy(
    SelectionDAG &DAG, SDLoc dl, SDValue Chain, SDValue Dst, SDValue Src,
    SDValue Size, unsigned Align, bool isVolatile, bool AlwaysInline,
    MachinePointerInfo DstPtrInfo, MachinePointerInfo SrcPtrInfo) const {
  // This requires the copy size to be a constant, preferably
  // within a subtarget-specific limit.
  ConstantSDNode *ConstantSize = dyn_cast<ConstantSDNode>(Size);
  const X86Subtarget &Subtarget =
      DAG.getMachineFunction().getSubtarget<X86Subtarget>();
  if (!ConstantSize)
    return SDValue();
  uint64_t SizeVal = ConstantSize->getZExtValue();
  if (!AlwaysInline && SizeVal > Subtarget.getMaxInlineSizeThreshold())
    return SDValue();

  /// If not DWORD aligned, it is more efficient to call the library.  However
  /// if calling the library is not allowed (AlwaysInline), then soldier on as
  /// the code generated here is better than the long load-store sequence we
  /// would otherwise get.
  if (!AlwaysInline && (Align & 3) != 0)
    return SDValue();

  // If to a segment-relative address space, use the default lowering.
  if (DstPtrInfo.getAddrSpace() >= 256 ||
      SrcPtrInfo.getAddrSpace() >= 256)
    return SDValue();

  // If the base register might conflict with our physical registers, bail out.
  const unsigned ClobberSet[] = {X86::RCX, X86::RSI, X86::RDI,
                                 X86::ECX, X86::ESI, X86::EDI};
  if (isBaseRegConflictPossible(DAG, ClobberSet))
    return SDValue();

  MVT AVT;
  if (Align & 1)
    AVT = MVT::i8;
  else if (Align & 2)
    AVT = MVT::i16;
  else if (Align & 4)
    // DWORD aligned
    AVT = MVT::i32;
  else
    // QWORD aligned
    AVT = Subtarget.is64Bit() ? MVT::i64 : MVT::i32;

  unsigned UBytes = AVT.getSizeInBits() / 8;
  unsigned CountVal = SizeVal / UBytes;
  SDValue Count = DAG.getIntPtrConstant(CountVal, dl);
  unsigned BytesLeft = SizeVal % UBytes;

  SDValue InFlag;
  Chain = DAG.getCopyToReg(Chain, dl, Subtarget.is64Bit() ? X86::RCX : X86::ECX,
                           Count, InFlag);
  InFlag = Chain.getValue(1);
  Chain = DAG.getCopyToReg(Chain, dl, Subtarget.is64Bit() ? X86::RDI : X86::EDI,
                           Dst, InFlag);
  InFlag = Chain.getValue(1);
  Chain = DAG.getCopyToReg(Chain, dl, Subtarget.is64Bit() ? X86::RSI : X86::ESI,
                           Src, InFlag);
  InFlag = Chain.getValue(1);

  SDVTList Tys = DAG.getVTList(MVT::Other, MVT::Glue);
  SDValue Ops[] = { Chain, DAG.getValueType(AVT), InFlag };
  SDValue RepMovs = DAG.getNode(X86ISD::REP_MOVS, dl, Tys, Ops);

  SmallVector<SDValue, 4> Results;
  Results.push_back(RepMovs);
  if (BytesLeft) {
    // Handle the last 1 - 7 bytes.
    unsigned Offset = SizeVal - BytesLeft;
    EVT DstVT = Dst.getValueType();
    EVT SrcVT = Src.getValueType();
    EVT SizeVT = Size.getValueType();
    Results.push_back(DAG.getMemcpy(Chain, dl,
                                    DAG.getNode(ISD::ADD, dl, DstVT, Dst,
                                                DAG.getConstant(Offset, dl,
                                                                DstVT)),
                                    DAG.getNode(ISD::ADD, dl, SrcVT, Src,
                                                DAG.getConstant(Offset, dl,
                                                                SrcVT)),
                                    DAG.getConstant(BytesLeft, dl, SizeVT),
                                    Align, isVolatile, AlwaysInline, false,
                                    DstPtrInfo.getWithOffset(Offset),
                                    SrcPtrInfo.getWithOffset(Offset)));
  }

  return DAG.getNode(ISD::TokenFactor, dl, MVT::Other, Results);
}

SDValue X86SelectionDAGInfo::EmitTargetCodeForMemset(
    SelectionDAG &DAG, SDLoc dl, SDValue Chain, SDValue Dst, SDValue Src,
    SDValue Size, unsigned Align, bool isVolatile,
    MachinePointerInfo DstPtrInfo) const {
  ConstantSDNode *ConstantSize = dyn_cast<ConstantSDNode>(Size);
  const X86Subtarget &Subtarget =
      DAG.getMachineFunction().getSubtarget<X86Subtarget>();

#ifndef NDEBUG
  // If the base register might conflict with our physical registers, bail out.
  const unsigned ClobberSet[] = {X86::RCX, X86::RAX, X86::RDI,
                                 X86::ECX, X86::EAX, X86::EDI};
  assert(!isBaseRegConflictPossible(DAG, ClobberSet));
#endif

  // If to a segment-relative address space, use the default lowering.
  if (DstPtrInfo.getAddrSpace() >= 256)
    return SDValue();

  // If not DWORD aligned or size is more than the threshold, call the library.
  // The libc version is likely to be faster for these cases. It can use the
  // address value and run time information about the CPU.
  if ((Align & 3) != 0 || !ConstantSize ||
      ConstantSize->getZExtValue() > Subtarget.getMaxInlineSizeThreshold()) {
    // Check to see if there is a specialized entry-point for memory zeroing.
    ConstantSDNode *V = dyn_cast<ConstantSDNode>(Src);

    if (const char *bzeroEntry = V &&
        V->isNullValue() ? Subtarget.getBZeroEntry() : nullptr) {
      const TargetLowering &TLI = DAG.getTargetLoweringInfo();
      EVT IntPtr = TLI.getPointerTy(DAG.getDataLayout());
      Type *IntPtrTy = DAG.getDataLayout().getIntPtrType(*DAG.getContext());
      TargetLowering::ArgListTy Args;
      TargetLowering::ArgListEntry Entry;
      Entry.Node = Dst;
      Entry.Ty = IntPtrTy;
      Args.push_back(Entry);
      Entry.Node = Size;
      Args.push_back(Entry);

      TargetLowering::CallLoweringInfo CLI(DAG);
      CLI.setDebugLoc(dl).setChain(Chain)
        .setCallee(CallingConv::C, Type::getVoidTy(*DAG.getContext()),
                   DAG.getExternalSymbol(bzeroEntry, IntPtr), std::move(Args),
                   0)
        .setDiscardResult();

      std::pair<SDValue,SDValue> CallResult = TLI.LowerCallTo(CLI);
      return CallResult.second;
    }

    // Otherwise have the target-independent code call memset.
    return SDValue();
  }

  uint64_t SizeVal = ConstantSize->getZExtValue();
  SDValue InFlag;
  EVT AVT;
  SDValue Count;
  ConstantSDNode *ValC = dyn_cast<ConstantSDNode>(Src);
  unsigned BytesLeft = 0;
  bool TwoRepStos = false;
  if (ValC) {
    unsigned ValReg;
    uint64_t Val = ValC->getZExtValue() & 255;

    // If the value is a constant, then we can potentially use larger sets.
    switch (Align & 3) {
    case 2:   // WORD aligned
      AVT = MVT::i16;
      ValReg = X86::AX;
      Val = (Val << 8) | Val;
      break;
    case 0:  // DWORD aligned
      AVT = MVT::i32;
      ValReg = X86::EAX;
      Val = (Val << 8)  | Val;
      Val = (Val << 16) | Val;
      if (Subtarget.is64Bit() && ((Align & 0x7) == 0)) {  // QWORD aligned
        AVT = MVT::i64;
        ValReg = X86::RAX;
        Val = (Val << 32) | Val;
      }
      break;
    default:  // Byte aligned
      AVT = MVT::i8;
      ValReg = X86::AL;
      Count = DAG.getIntPtrConstant(SizeVal, dl);
      break;
    }

    if (AVT.bitsGT(MVT::i8)) {
      unsigned UBytes = AVT.getSizeInBits() / 8;
      Count = DAG.getIntPtrConstant(SizeVal / UBytes, dl);
      BytesLeft = SizeVal % UBytes;
    }

    Chain = DAG.getCopyToReg(Chain, dl, ValReg, DAG.getConstant(Val, dl, AVT),
                             InFlag);
    InFlag = Chain.getValue(1);
//.........这里部分代码省略.........

// Emit, if possible, a specialized version of the given Libcall. Typically this
// means selecting the appropriately aligned version, but we also convert memset
// of 0 into memclr.
SDValue ARMSelectionDAGInfo::EmitSpecializedLibcall(
    SelectionDAG &DAG, const SDLoc &dl, SDValue Chain, SDValue Dst, SDValue Src,
    SDValue Size, unsigned Align, RTLIB::Libcall LC) const {
  const ARMSubtarget &Subtarget =
      DAG.getMachineFunction().getSubtarget<ARMSubtarget>();
  const ARMTargetLowering *TLI = Subtarget.getTargetLowering();

  // Only use a specialized AEABI function if the default version of this
  // Libcall is an AEABI function.
  if (std::strncmp(TLI->getLibcallName(LC), "__aeabi", 7) != 0)
    return SDValue();

  // Translate RTLIB::Libcall to AEABILibcall. We only do this in order to be
  // able to translate memset to memclr and use the value to index the function
  // name array.
  enum {
    AEABI_MEMCPY = 0,
    AEABI_MEMMOVE,
    AEABI_MEMSET,
    AEABI_MEMCLR
  } AEABILibcall;
  switch (LC) {
  case RTLIB::MEMCPY:
    AEABILibcall = AEABI_MEMCPY;
    break;
  case RTLIB::MEMMOVE:
    AEABILibcall = AEABI_MEMMOVE;
    break;
  case RTLIB::MEMSET: 
    AEABILibcall = AEABI_MEMSET;
    if (ConstantSDNode *ConstantSrc = dyn_cast<ConstantSDNode>(Src))
      if (ConstantSrc->getZExtValue() == 0)
        AEABILibcall = AEABI_MEMCLR;
    break;
  default:
    return SDValue();
  }

  // Choose the most-aligned libcall variant that we can
  enum {
    ALIGN1 = 0,
    ALIGN4,
    ALIGN8
  } AlignVariant;
  if ((Align & 7) == 0)
    AlignVariant = ALIGN8;
  else if ((Align & 3) == 0)
    AlignVariant = ALIGN4;
  else
    AlignVariant = ALIGN1;

  TargetLowering::ArgListTy Args;
  TargetLowering::ArgListEntry Entry;
  Entry.Ty = DAG.getDataLayout().getIntPtrType(*DAG.getContext());
  Entry.Node = Dst;
  Args.push_back(Entry);
  if (AEABILibcall == AEABI_MEMCLR) {
    Entry.Node = Size;
    Args.push_back(Entry);
  } else if (AEABILibcall == AEABI_MEMSET) {
    // Adjust parameters for memset, EABI uses format (ptr, size, value),
    // GNU library uses (ptr, value, size)
    // See RTABI section 4.3.4
    Entry.Node = Size;
    Args.push_back(Entry);

    // Extend or truncate the argument to be an i32 value for the call.
    if (Src.getValueType().bitsGT(MVT::i32))
      Src = DAG.getNode(ISD::TRUNCATE, dl, MVT::i32, Src);
    else if (Src.getValueType().bitsLT(MVT::i32))
      Src = DAG.getNode(ISD::ZERO_EXTEND, dl, MVT::i32, Src);

    Entry.Node = Src; 
    Entry.Ty = Type::getInt32Ty(*DAG.getContext());
    Entry.isSExt = false;
    Args.push_back(Entry);
  } else {
    Entry.Node = Src;
    Args.push_back(Entry);
    
    Entry.Node = Size;
    Args.push_back(Entry);
  }

  char const *FunctionNames[4][3] = {
    { "__aeabi_memcpy",  "__aeabi_memcpy4",  "__aeabi_memcpy8"  },
    { "__aeabi_memmove", "__aeabi_memmove4", "__aeabi_memmove8" },
    { "__aeabi_memset",  "__aeabi_memset4",  "__aeabi_memset8"  },
    { "__aeabi_memclr",  "__aeabi_memclr4",  "__aeabi_memclr8"  }
  };
  TargetLowering::CallLoweringInfo CLI(DAG);
  CLI.setDebugLoc(dl)
      .setChain(Chain)
      .setCallee(
           TLI->getLibcallCallingConv(LC), Type::getVoidTy(*DAG.getContext()),
           DAG.getExternalSymbol(FunctionNames[AEABILibcall][AlignVariant],
                                 TLI->getPointerTy(DAG.getDataLayout())),
//.........这里部分代码省略.........

/// Select instructions not customized! Used for
/// expanded, promoted and normal instructions
SDNode* MBlazeDAGToDAGISel::Select(SDNode *Node) {
  unsigned Opcode = Node->getOpcode();
  DebugLoc dl = Node->getDebugLoc();

  // If we have a custom node, we already have selected!
  if (Node->isMachineOpcode())
    return NULL;

  ///
  // Instruction Selection not handled by the auto-generated
  // tablegen selection should be handled here.
  ///
  switch (Opcode) {
    default: break;

    // Get target GOT address.
    case ISD::GLOBAL_OFFSET_TABLE:
      return getGlobalBaseReg();

    case ISD::FrameIndex: {
        SDValue imm = CurDAG->getTargetConstant(0, MVT::i32);
        int FI = dyn_cast<FrameIndexSDNode>(Node)->getIndex();
        EVT VT = Node->getValueType(0);
        SDValue TFI = CurDAG->getTargetFrameIndex(FI, VT);
        unsigned Opc = MBlaze::ADDIK;
        if (Node->hasOneUse())
          return CurDAG->SelectNodeTo(Node, Opc, VT, TFI, imm);
        return CurDAG->getMachineNode(Opc, dl, VT, TFI, imm);
    }


    /// Handle direct and indirect calls when using PIC. On PIC, when
    /// GOT is smaller than about 64k (small code) the GA target is
    /// loaded with only one instruction. Otherwise GA's target must
    /// be loaded with 3 instructions.
    case MBlazeISD::JmpLink: {
      if (TM.getRelocationModel() == Reloc::PIC_) {
        SDValue Chain  = Node->getOperand(0);
        SDValue Callee = Node->getOperand(1);
        SDValue R20Reg = CurDAG->getRegister(MBlaze::R20, MVT::i32);
        SDValue InFlag(0, 0);

        if ((isa<GlobalAddressSDNode>(Callee)) ||
            (isa<ExternalSymbolSDNode>(Callee)))
        {
          /// Direct call for global addresses and external symbols
          SDValue GPReg = CurDAG->getRegister(MBlaze::R15, MVT::i32);

          // Use load to get GOT target
          SDValue Ops[] = { Callee, GPReg, Chain };
          SDValue Load = SDValue(CurDAG->getMachineNode(MBlaze::LW, dl,
                                 MVT::i32, MVT::Other, Ops), 0);
          Chain = Load.getValue(1);

          // Call target must be on T9
          Chain = CurDAG->getCopyToReg(Chain, dl, R20Reg, Load, InFlag);
        } else
          /// Indirect call
          Chain = CurDAG->getCopyToReg(Chain, dl, R20Reg, Callee, InFlag);

        // Emit Jump and Link Register
        SDNode *ResNode = CurDAG->getMachineNode(MBlaze::BRLID, dl, MVT::Other,
                                                 MVT::Glue, R20Reg, Chain);
        Chain  = SDValue(ResNode, 0);
        InFlag = SDValue(ResNode, 1);
        ReplaceUses(SDValue(Node, 0), Chain);
        ReplaceUses(SDValue(Node, 1), InFlag);
        return ResNode;
      }
    }
  }

  // Select the default instruction
  SDNode *ResNode = SelectCode(Node);

  DEBUG(errs() << "=> ");
  if (ResNode == NULL || ResNode == Node)
    DEBUG(Node->dump(CurDAG));
  else
    DEBUG(ResNode->dump(CurDAG));
  DEBUG(errs() << "\n");
  return ResNode;
}

/// PerformExpensiveChecks - Do extensive, expensive, sanity checking.
void DAGTypeLegalizer::PerformExpensiveChecks() {
  // If a node is not processed, then none of its values should be mapped by any
  // of PromotedIntegers, ExpandedIntegers, ..., ReplacedValues.

  // If a node is processed, then each value with an illegal type must be mapped
  // by exactly one of PromotedIntegers, ExpandedIntegers, ..., ReplacedValues.
  // Values with a legal type may be mapped by ReplacedValues, but not by any of
  // the other maps.

  // Note that these invariants may not hold momentarily when processing a node:
  // the node being processed may be put in a map before being marked Processed.

  // Note that it is possible to have nodes marked NewNode in the DAG.  This can
  // occur in two ways.  Firstly, a node may be created during legalization but
  // never passed to the legalization core.  This is usually due to the implicit
  // folding that occurs when using the DAG.getNode operators.  Secondly, a new
  // node may be passed to the legalization core, but when analyzed may morph
  // into a different node, leaving the original node as a NewNode in the DAG.
  // A node may morph if one of its operands changes during analysis.  Whether
  // it actually morphs or not depends on whether, after updating its operands,
  // it is equivalent to an existing node: if so, it morphs into that existing
  // node (CSE).  An operand can change during analysis if the operand is a new
  // node that morphs, or it is a processed value that was mapped to some other
  // value (as recorded in ReplacedValues) in which case the operand is turned
  // into that other value.  If a node morphs then the node it morphed into will
  // be used instead of it for legalization, however the original node continues
  // to live on in the DAG.
  // The conclusion is that though there may be nodes marked NewNode in the DAG,
  // all uses of such nodes are also marked NewNode: the result is a fungus of
  // NewNodes growing on top of the useful nodes, and perhaps using them, but
  // not used by them.

  // If a value is mapped by ReplacedValues, then it must have no uses, except
  // by nodes marked NewNode (see above).

  // The final node obtained by mapping by ReplacedValues is not marked NewNode.
  // Note that ReplacedValues should be applied iteratively.

  // Note that the ReplacedValues map may also map deleted nodes (by iterating
  // over the DAG we never dereference deleted nodes).  This means that it may
  // also map nodes marked NewNode if the deallocated memory was reallocated as
  // another node, and that new node was not seen by the LegalizeTypes machinery
  // (for example because it was created but not used).  In general, we cannot
  // distinguish between new nodes and deleted nodes.
  SmallVector<SDNode*, 16> NewNodes;
  for (SelectionDAG::allnodes_iterator I = DAG.allnodes_begin(),
       E = DAG.allnodes_end(); I != E; ++I) {
    // Remember nodes marked NewNode - they are subject to extra checking below.
    if (I->getNodeId() == NewNode)
      NewNodes.push_back(I);

    for (unsigned i = 0, e = I->getNumValues(); i != e; ++i) {
      SDValue Res(I, i);
      bool Failed = false;

      unsigned Mapped = 0;
      if (ReplacedValues.find(Res) != ReplacedValues.end()) {
        Mapped |= 1;
        // Check that remapped values are only used by nodes marked NewNode.
        for (SDNode::use_iterator UI = I->use_begin(), UE = I->use_end();
             UI != UE; ++UI)
          if (UI.getUse().getResNo() == i)
            assert(UI->getNodeId() == NewNode &&
                   "Remapped value has non-trivial use!");

        // Check that the final result of applying ReplacedValues is not
        // marked NewNode.
        SDValue NewVal = ReplacedValues[Res];
        DenseMap<SDValue, SDValue>::iterator I = ReplacedValues.find(NewVal);
        while (I != ReplacedValues.end()) {
          NewVal = I->second;
          I = ReplacedValues.find(NewVal);
        }
        assert(NewVal.getNode()->getNodeId() != NewNode &&
               "ReplacedValues maps to a new node!");
      }
      if (PromotedIntegers.find(Res) != PromotedIntegers.end())
        Mapped |= 2;
      if (SoftenedFloats.find(Res) != SoftenedFloats.end())
        Mapped |= 4;
      if (ScalarizedVectors.find(Res) != ScalarizedVectors.end())
        Mapped |= 8;
      if (ExpandedIntegers.find(Res) != ExpandedIntegers.end())
        Mapped |= 16;
      if (ExpandedFloats.find(Res) != ExpandedFloats.end())
        Mapped |= 32;
      if (SplitVectors.find(Res) != SplitVectors.end())
        Mapped |= 64;
      if (WidenedVectors.find(Res) != WidenedVectors.end())
        Mapped |= 128;

      if (I->getNodeId() != Processed) {
        // Since we allow ReplacedValues to map deleted nodes, it may map nodes
        // marked NewNode too, since a deleted node may have been reallocated as
        // another node that has not been seen by the LegalizeTypes machinery.
        if ((I->getNodeId() == NewNode && Mapped > 1) ||
            (I->getNodeId() != NewNode && Mapped != 0)) {
          dbgs() << "Unprocessed value in a map!";
          Failed = true;
//.........这里部分代码省略.........

/// EmitCopyFromReg - Generate machine code for an CopyFromReg node or an
/// implicit physical register output.
void InstrEmitter::
EmitCopyFromReg(SDNode *Node, unsigned ResNo, bool IsClone, bool IsCloned,
                unsigned SrcReg, DenseMap<SDValue, unsigned> &VRBaseMap) {
    unsigned VRBase = 0;
    if (TargetRegisterInfo::isVirtualRegister(SrcReg)) {
        // Just use the input register directly!
        SDValue Op(Node, ResNo);
        if (IsClone)
            VRBaseMap.erase(Op);
        bool isNew = VRBaseMap.insert(std::make_pair(Op, SrcReg)).second;
        (void)isNew; // Silence compiler warning.
        assert(isNew && "Node emitted out of order - early");
        return;
    }

    // If the node is only used by a CopyToReg and the dest reg is a vreg, use
    // the CopyToReg'd destination register instead of creating a new vreg.
    bool MatchReg = true;
    const TargetRegisterClass *UseRC = NULL;
    EVT VT = Node->getValueType(ResNo);

    // Stick to the preferred register classes for legal types.
    if (TLI->isTypeLegal(VT))
        UseRC = TLI->getRegClassFor(VT);

    if (!IsClone && !IsCloned)
        for (SDNode::use_iterator UI = Node->use_begin(), E = Node->use_end();
                UI != E; ++UI) {
            SDNode *User = *UI;
            bool Match = true;
            if (User->getOpcode() == ISD::CopyToReg &&
                    User->getOperand(2).getNode() == Node &&
                    User->getOperand(2).getResNo() == ResNo) {
                unsigned DestReg = cast<RegisterSDNode>(User->getOperand(1))->getReg();
                if (TargetRegisterInfo::isVirtualRegister(DestReg)) {
                    VRBase = DestReg;
                    Match = false;
                } else if (DestReg != SrcReg)
                    Match = false;
            } else {
                for (unsigned i = 0, e = User->getNumOperands(); i != e; ++i) {
                    SDValue Op = User->getOperand(i);
                    if (Op.getNode() != Node || Op.getResNo() != ResNo)
                        continue;
                    EVT VT = Node->getValueType(Op.getResNo());
                    if (VT == MVT::Other || VT == MVT::Glue)
                        continue;
                    Match = false;
                    if (User->isMachineOpcode()) {
                        const MCInstrDesc &II = TII->get(User->getMachineOpcode());
                        const TargetRegisterClass *RC = 0;
                        if (i+II.getNumDefs() < II.getNumOperands())
                            RC = TII->getRegClass(II, i+II.getNumDefs(), TRI);
                        if (!UseRC)
                            UseRC = RC;
                        else if (RC) {
                            const TargetRegisterClass *ComRC =
                                TRI->getCommonSubClass(UseRC, RC);
                            // If multiple uses expect disjoint register classes, we emit
                            // copies in AddRegisterOperand.
                            if (ComRC)
                                UseRC = ComRC;
                        }
                    }
                }
            }
            MatchReg &= Match;
            if (VRBase)
                break;
        }

    const TargetRegisterClass *SrcRC = 0, *DstRC = 0;
    SrcRC = TRI->getMinimalPhysRegClass(SrcReg, VT);

    // Figure out the register class to create for the destreg.
    if (VRBase) {
        DstRC = MRI->getRegClass(VRBase);
    } else if (UseRC) {
        assert(UseRC->hasType(VT) && "Incompatible phys register def and uses!");
        DstRC = UseRC;
    } else {
        DstRC = TLI->getRegClassFor(VT);
    }

    // If all uses are reading from the src physical register and copying the
    // register is either impossible or very expensive, then don't create a copy.
    if (MatchReg && SrcRC->getCopyCost() < 0) {
        VRBase = SrcReg;
    } else {
        // Create the reg, emit the copy.
        VRBase = MRI->createVirtualRegister(DstRC);
        BuildMI(*MBB, InsertPos, Node->getDebugLoc(), TII->get(TargetO