LowererMDArch.cpp

//-------------------------------------------------------------------------------------------------------
// Copyright (C) Microsoft. All rights reserved.
// Copyright (c) 2021 ChakraCore Project Contributors. All rights reserved.
// Licensed under the MIT license. See LICENSE.txt file in the project root for full license information.
//-------------------------------------------------------------------------------------------------------

#include "Backend.h"
#include "LowererMDArch.h"
#include "Library/Generators/JavascriptGeneratorFunction.h"

const Js::OpCode LowererMD::MDExtend32Opcode = Js::OpCode::MOVSXD;

extern const IRType RegTypes[RegNumCount];

BYTE
LowererMDArch::GetDefaultIndirScale()
{
    return IndirScale8;
}

RegNum
LowererMDArch::GetRegShiftCount()
{
    return RegRCX;
}

RegNum
LowererMDArch::GetRegReturn(IRType type)
{
    return ( IRType_IsFloat(type) || IRType_IsSimd128(type) ) ? RegXMM0 : RegRAX;
}

RegNum
LowererMDArch::GetRegReturnAsmJs(IRType type)
{
    if (IRType_IsFloat(type))
    {
        return RegXMM0;
    }
    else if (IRType_IsSimd128(type))
    {
        return RegXMM0;
    }
    else
    {
        return RegRAX;
    }
}

RegNum
LowererMDArch::GetRegStackPointer()
{
    return RegRSP;
}

RegNum
LowererMDArch::GetRegBlockPointer()
{
    return RegRBP;
}

RegNum
LowererMDArch::GetRegFramePointer()
{
    return RegRBP;
}

RegNum
LowererMDArch::GetRegChkStkParam()
{
    return RegRAX;
}

RegNum
LowererMDArch::GetRegIMulDestLower()
{
    return RegRAX;
}

RegNum
LowererMDArch::GetRegIMulHighDestLower()
{
    return RegRDX;
}
RegNum
LowererMDArch::GetRegArgI4(int32 argNum)
{
    // TODO: decide on registers to use for int
    return RegNOREG;
}

RegNum
LowererMDArch::GetRegArgR8(int32 argNum)
{
    // TODO: decide on registers to use for double
    return RegNOREG;
}

Js::OpCode
LowererMDArch::GetAssignOp(IRType type)
{
    switch (type)
    {
    case TyFloat64:
        return Js::OpCode::MOVSD;
    case TyFloat32:
        return Js::OpCode::MOVSS;
    case TySimd128F4:
    case TySimd128I4:
    case TySimd128I8:
    case TySimd128I16:
    case TySimd128U4:
    case TySimd128U8:
    case TySimd128U16:
    case TySimd128B4:
    case TySimd128B8:
    case TySimd128B16:
    case TySimd128D2:
    case TySimd128I2:
        return Js::OpCode::MOVUPS;
    default:
        return Js::OpCode::MOV;
    }
}

void
LowererMDArch::Init(LowererMD *lowererMD)
{
    this->lowererMD                 = lowererMD;
    this->helperCallArgsCount       = 0;
}

///----------------------------------------------------------------------------
///
/// LowererMD::LoadInputParamPtr
///
///     Load the address of the start of the passed-in parameters not including
///     the this parameter.
///
///----------------------------------------------------------------------------

IR::Instr *
LowererMDArch::LoadInputParamPtr(IR::Instr *instrInsert, IR::RegOpnd *optionalDstOpnd /* = nullptr */)
{
    if (this->m_func->GetJITFunctionBody()->IsCoroutine())
    {
        IR::RegOpnd * argPtrRegOpnd = Lowerer::LoadGeneratorArgsPtr(instrInsert);
        IR::IndirOpnd * indirOpnd = IR::IndirOpnd::New(argPtrRegOpnd, 1 * MachPtr, TyMachPtr, this->m_func);
        IR::RegOpnd * dstOpnd = optionalDstOpnd != nullptr ? optionalDstOpnd : IR::RegOpnd::New(TyMachPtr, this->m_func);

        return Lowerer::InsertLea(dstOpnd, indirOpnd, instrInsert);
    }
    else
    {
        // Stack looks like (EBP chain)+0, (return addr)+4, (function object)+8, (arg count)+12, (this)+16, actual args
        StackSym *paramSym = StackSym::New(TyMachReg, this->m_func);
        this->m_func->SetArgOffset(paramSym, 5 * MachPtr);
        return this->lowererMD->m_lowerer->InsertLoadStackAddress(paramSym, instrInsert, optionalDstOpnd);
    }
}


IR::Instr *
LowererMDArch::LoadStackArgPtr(IR::Instr * instrArgPtr)
{
    // Get the args pointer relative to the frame pointer.
    // NOTE: This code is sufficient for the apply-args optimization, but not for StackArguments,
    // if and when that is enabled.

    // dst = LEA &[rbp + "this" offset + sizeof(var)]

    IR::Instr * instr = LoadInputParamPtr(instrArgPtr, instrArgPtr->UnlinkDst()->AsRegOpnd());
    instrArgPtr->Remove();

    return instr->m_prev;
}

IR::Instr *
LowererMDArch::LoadHeapArgsCached(IR::Instr *instrArgs)
{
    ASSERT_INLINEE_FUNC(instrArgs);
    Func *func = instrArgs->m_func;
    IR::Instr *instrPrev = instrArgs->m_prev;

    if (instrArgs->m_func->IsStackArgsEnabled())
    {
        instrArgs->m_opcode = Js::OpCode::MOV;
        instrArgs->ReplaceSrc1(IR::AddrOpnd::NewNull(func));

        if (PHASE_TRACE1(Js::StackArgFormalsOptPhase) && func->GetJITFunctionBody()->GetInParamsCount() > 1)
        {
            Output::Print(_u("StackArgFormals : %s (%d) :Removing Heap Arguments object creation in Lowerer. \n"), instrArgs->m_func->GetJITFunctionBody()->GetDisplayName(), instrArgs->m_func->GetFunctionNumber());
            Output::Flush();
        }
    }
    else
    {
        // s7 = formals are let decls
        // s6 = memory context
        // s5 = local frame instance
        // s4 = address of first actual argument (after "this")
        // s3 = formal argument count
        // s2 = actual argument count
        // s1 = current function
        // dst = JavascriptOperators::LoadArguments(s1, s2, s3, s4, s5, s6, s7)

        // s7 = formals are let decls
        IR::Opnd * formalsAreLetDecls = IR::IntConstOpnd::New((IntConstType)(instrArgs->m_opcode == Js::OpCode::LdLetHeapArgsCached), TyUint8, func);
        this->LoadHelperArgument(instrArgs, formalsAreLetDecls);

        // s6 = memory context
        this->lowererMD->m_lowerer->LoadScriptContext(instrArgs);

        // s5 = local frame instance
        IR::Opnd *frameObj = instrArgs->UnlinkSrc1();
        this->LoadHelperArgument(instrArgs, frameObj);

        if (func->IsInlinee())
        {
            // s4 = address of first actual argument (after "this").
            StackSym *firstRealArgSlotSym = func->GetInlineeArgvSlotOpnd()->m_sym->AsStackSym();
            this->m_func->SetArgOffset(firstRealArgSlotSym, firstRealArgSlotSym->m_offset + MachPtr);
            IR::Instr *instr = this->lowererMD->m_lowerer->InsertLoadStackAddress(firstRealArgSlotSym, instrArgs);
            this->LoadHelperArgument(instrArgs, instr->GetDst());

            // s3 = formal argument count (without counting "this").
            uint32 formalsCount = func->GetJITFunctionBody()->GetInParamsCount() - 1;
            this->LoadHelperArgument(instrArgs, IR::IntConstOpnd::New(formalsCount, TyUint32, func));

            // s2 = actual argument count (without counting "this").
            instr = IR::Instr::New(Js::OpCode::MOV,
                IR::RegOpnd::New(TyMachReg, func),
                IR::IntConstOpnd::New(func->actualCount - 1, TyMachReg, func),
                func);
            instrArgs->InsertBefore(instr);
            this->LoadHelperArgument(instrArgs, instr->GetDst());

            // s1 = current function.
            this->LoadHelperArgument(instrArgs, func->GetInlineeFunctionObjectSlotOpnd());

            // Save the newly-created args object to its dedicated stack slot.
            IR::SymOpnd *argObjSlotOpnd = func->GetInlineeArgumentsObjectSlotOpnd();
            instr = IR::Instr::New(Js::OpCode::MOV,
                argObjSlotOpnd,
                instrArgs->GetDst(),
                func);
            instrArgs->InsertAfter(instr);
        }
        else
        {
            // s4 = address of first actual argument (after "this")
            // Stack looks like (EBP chain)+0, (return addr)+4, (function object)+8, (arg count)+12, (this)+16, actual args
            IR::Instr *instr = this->LoadInputParamPtr(instrArgs);
            this->LoadHelperArgument(instrArgs, instr->GetDst());

            // s3 = formal argument count (without counting "this")
            uint32 formalsCount = func->GetInParamsCount() - 1;
            this->LoadHelperArgument(instrArgs, IR::IntConstOpnd::New(formalsCount, TyInt32, func));

            // s2 = actual argument count (without counting "this")
            instr = this->lowererMD->LoadInputParamCount(instrArgs);
            instr = IR::Instr::New(Js::OpCode::DEC, instr->GetDst(), instr->GetDst(), func);
            instrArgs->InsertBefore(instr);
            this->LoadHelperArgument(instrArgs, instr->GetDst());

            // s1 = current function
            StackSym *paramSym = StackSym::New(TyMachReg, func);
            this->m_func->SetArgOffset(paramSym, 2 * MachPtr);
            IR::Opnd * srcOpnd = IR::SymOpnd::New(paramSym, TyMachReg, func);
            this->LoadHelperArgument(instrArgs, srcOpnd);

            // Save the newly-created args object to its dedicated stack slot.
            IR::Opnd *opnd = LowererMD::CreateStackArgumentsSlotOpnd(func);
            instr = IR::Instr::New(Js::OpCode::MOV, opnd, instrArgs->GetDst(), func);
            instrArgs->InsertAfter(instr);
        }

        this->lowererMD->ChangeToHelperCall(instrArgs, IR::HelperOp_LoadHeapArgsCached);
    }
    return instrPrev;
}

///----------------------------------------------------------------------------
///
/// LowererMDArch::LoadHeapArguments
///
///     Load the arguments object
///     NOTE: The same caveat regarding arguments passed on the stack applies here
///           as in LoadInputParamCount above.
///----------------------------------------------------------------------------

IR::Instr *
LowererMDArch::LoadHeapArguments(IR::Instr *instrArgs)
{
    ASSERT_INLINEE_FUNC(instrArgs);
    Func *func = instrArgs->m_func;

    IR::Instr *instrPrev = instrArgs->m_prev;
    if (func->IsStackArgsEnabled())
    {
        instrArgs->m_opcode = Js::OpCode::MOV;
        instrArgs->ReplaceSrc1(IR::AddrOpnd::NewNull(func));
        if (PHASE_TRACE1(Js::StackArgFormalsOptPhase) && func->GetJITFunctionBody()->GetInParamsCount() > 1)
        {
            Output::Print(_u("StackArgFormals : %s (%d) :Removing Heap Arguments object creation in Lowerer. \n"), instrArgs->m_func->GetJITFunctionBody()->GetDisplayName(), instrArgs->m_func->GetFunctionNumber());
            Output::Flush();
        }
    }
    else
    {
        // s7 = formals are let decls
        // s6 = memory context
        // s5 = array of property ID's
        // s4 = local frame instance
        // s3 = address of first actual argument (after "this")
        // s2 = actual argument count
        // s1 = current function
        // dst = JavascriptOperators::LoadHeapArguments(s1, s2, s3, s4, s5, s6, s7)

        // s7 = formals are let decls
        this->LoadHelperArgument(instrArgs, IR::IntConstOpnd::New(instrArgs->m_opcode == Js::OpCode::LdLetHeapArguments ? TRUE : FALSE, TyUint8, func));

        // s6 = memory context
        instrPrev = this->lowererMD->m_lowerer->LoadScriptContext(instrArgs);

        // s5 = array of property ID's
        intptr_t formalsPropIdArray = instrArgs->m_func->GetJITFunctionBody()->GetFormalsPropIdArrayAddr();
        if (!formalsPropIdArray)
        {
            formalsPropIdArray = instrArgs->m_func->GetScriptContextInfo()->GetNullAddr();
        }

        IR::Opnd * argArray = IR::AddrOpnd::New(formalsPropIdArray, IR::AddrOpndKindDynamicMisc, m_func);
        this->LoadHelperArgument(instrArgs, argArray);

        // s4 = local frame instance
        IR::Opnd *frameObj = instrArgs->UnlinkSrc1();
        this->LoadHelperArgument(instrArgs, frameObj);

        if (func->IsInlinee())
        {
            // s3 = address of first actual argument (after "this").
            StackSym *firstRealArgSlotSym = func->GetInlineeArgvSlotOpnd()->m_sym->AsStackSym();
            this->m_func->SetArgOffset(firstRealArgSlotSym, firstRealArgSlotSym->m_offset + MachPtr);
            IR::Instr *instr = this->lowererMD->m_lowerer->InsertLoadStackAddress(firstRealArgSlotSym, instrArgs);
            this->LoadHelperArgument(instrArgs, instr->GetDst());

            // s2 = actual argument count (without counting "this").
            instr = IR::Instr::New(Js::OpCode::MOV,
                                   IR::RegOpnd::New(TyUint32, func),
                                   IR::IntConstOpnd::New(func->actualCount - 1, TyUint32, func),
                                   func);
            instrArgs->InsertBefore(instr);
            this->LoadHelperArgument(instrArgs, instr->GetDst());

            // s1 = current function.
            this->LoadHelperArgument(instrArgs, func->GetInlineeFunctionObjectSlotOpnd());

            // Save the newly-created args object to its dedicated stack slot.
            IR::SymOpnd *argObjSlotOpnd = func->GetInlineeArgumentsObjectSlotOpnd();
            instr = IR::Instr::New(Js::OpCode::MOV,
                                   argObjSlotOpnd,
                                   instrArgs->GetDst(),
                                   func);
            instrArgs->InsertAfter(instr);
        }
        else
        {
            // s3 = address of first actual argument (after "this")
            // Stack looks like (EBP chain)+0, (return addr)+4, (function object)+8, (arg count)+12, (this)+16, actual args
            IR::Instr *instr = this->LoadInputParamPtr(instrArgs);
            this->LoadHelperArgument(instrArgs, instr->GetDst());

            // s2 = actual argument count (without counting "this")
            instr = this->lowererMD->LoadInputParamCount(instrArgs, -1);
            IR::Opnd * opndInputParamCount = instr->GetDst();

            this->LoadHelperArgument(instrArgs, opndInputParamCount);

            // s1 = current function
            StackSym * paramSym = StackSym::New(TyMachReg, func);
            this->m_func->SetArgOffset(paramSym, 2 * MachPtr);
            IR::Opnd * srcOpnd = IR::SymOpnd::New(paramSym, TyMachReg, func);

            if (this->m_func->GetJITFunctionBody()->IsCoroutine())
            {
                // the function object for generator calls is a GeneratorVirtualScriptFunction object
                // and we need to pass the real JavascriptGeneratorFunction object so grab it instead
                IR::RegOpnd *tmpOpnd = IR::RegOpnd::New(TyMachReg, func);
                Lowerer::InsertMove(tmpOpnd, srcOpnd, instrArgs);

                srcOpnd = IR::IndirOpnd::New(tmpOpnd, Js::GeneratorVirtualScriptFunction::GetRealFunctionOffset(), TyMachPtr, func);
            }

            this->LoadHelperArgument(instrArgs, srcOpnd);

            // Save the newly-created args object to its dedicated stack slot.
            IR::Opnd *opnd = LowererMD::CreateStackArgumentsSlotOpnd(func);
            instr = IR::Instr::New(Js::OpCode::MOV, opnd, instrArgs->GetDst(), func);
            instrArgs->InsertAfter(instr);
        }

        this->lowererMD->ChangeToHelperCall(instrArgs, IR::HelperOp_LoadHeapArguments);
    }

    return instrPrev;
}

//
// Load the parameter in the first argument slot
//

IR::Instr *
LowererMDArch::LoadNewScObjFirstArg(IR::Instr * instr, IR::Opnd * dst, ushort extraArgs)
{
    // Spread moves down the argument slot by one.
    IR::Opnd *      argOpnd         = this->GetArgSlotOpnd(3 + extraArgs);
    IR::Instr *     argInstr        = Lowerer::InsertMove(argOpnd, dst, instr);

    return argInstr;
}

inline static RegNum GetRegFromArgPosition(const bool isFloatArg, const uint16 argPosition)
{
    RegNum reg = RegNOREG;

    if (!isFloatArg && argPosition <= IntArgRegsCount)
    {
        switch (argPosition)
        {
#define REG_INT_ARG(Index, Name)    \
case ((Index) + 1):         \
reg = Reg ## Name;      \
break;
#include "RegList.h"
            default:
                Assume(UNREACHED);
        }
    }
    else if (isFloatArg && argPosition <= XmmArgRegsCount)
    {
        switch (argPosition)
        {
#define REG_XMM_ARG(Index, Name)    \
case ((Index) + 1):         \
reg = Reg ## Name;      \
break;
#include "RegList.h"
            default:
                Assume(UNREACHED);
        }
    }

    return reg;
}

int32
LowererMDArch::LowerCallArgs(IR::Instr *callInstr, ushort callFlags, Js::ArgSlot extraParams, IR::IntConstOpnd **callInfoOpndRef /* = nullptr */)
{
    AssertMsg(this->helperCallArgsCount == 0, "We don't support nested helper calls yet");

    const Js::ArgSlot       argOffset       = 1;
    uint32                  argCount        = 0;

    // Lower args and look for StartCall

    IR::Instr * argInstr = callInstr;
    IR::Instr * cfgInsertLoc = callInstr->GetPrevRealInstr();
    IR::Opnd *src2 = argInstr->UnlinkSrc2();
    while (src2->IsSymOpnd())
    {
        IR::SymOpnd *   argLinkOpnd = src2->AsSymOpnd();
        StackSym *      argLinkSym  = argLinkOpnd->m_sym->AsStackSym();
        AssertMsg(argLinkSym->IsArgSlotSym() && argLinkSym->m_isSingleDef, "Arg tree not single def...");
        argLinkOpnd->Free(this->m_func);

        argInstr = argLinkSym->m_instrDef;

        src2 = argInstr->UnlinkSrc2();
        this->lowererMD->ChangeToAssign(argInstr);

        // Mov each arg to its argSlot

        Js::ArgSlot     argPosition = argInstr->GetDst()->AsSymOpnd()->m_sym->AsStackSym()->GetArgSlotNum();
        Js::ArgSlot     index       = argOffset + argPosition;
        if(index < argPosition)
        {
            Js::Throw::OutOfMemory();
        }
        index += extraParams;
        if(index < extraParams)
        {
            Js::Throw::OutOfMemory();
        }

        IR::Opnd *      dstOpnd     = this->GetArgSlotOpnd(index, argLinkSym);

        argInstr->ReplaceDst(dstOpnd);

        cfgInsertLoc = argInstr->GetPrevRealInstr();

        // The arg sym isn't assigned a constant directly anymore
        // TODO: We can just move the instruction down next to the call if it is just a constant assignment
        // but AMD64 doesn't have the MOV mem,imm64 encoding, and we have no code to detect if the value can fit
        // into imm32 and hoist the src if it is not.
        argLinkSym->m_isConst = false;
        argLinkSym->m_isIntConst = false;
        argLinkSym->m_isTaggableIntConst = false;

        argInstr->Unlink();
        callInstr->InsertBefore(argInstr);
        argCount++;
    }


    IR::RegOpnd *       argLinkOpnd = src2->AsRegOpnd();
    StackSym *          argLinkSym  = argLinkOpnd->m_sym->AsStackSym();
    AssertMsg(!argLinkSym->IsArgSlotSym() && argLinkSym->m_isSingleDef, "Arg tree not single def...");

    IR::Instr *startCallInstr = argLinkSym->m_instrDef;

    if (callInstr->m_opcode == Js::OpCode::NewScObject ||
        callInstr->m_opcode == Js::OpCode::NewScObjectSpread ||
        callInstr->m_opcode == Js::OpCode::NewScObjectLiteral ||
        callInstr->m_opcode == Js::OpCode::NewScObjArray ||
        callInstr->m_opcode == Js::OpCode::NewScObjArraySpread)
    {
        // These push an extra arg.
        argCount++;
    }

    AssertMsg(startCallInstr->m_opcode == Js::OpCode::StartCall ||
              startCallInstr->m_opcode == Js::OpCode::LoweredStartCall,
              "Problem with arg chain.");
    AssertMsg(startCallInstr->GetArgOutCount(/*getInterpreterArgOutCount*/ false) == argCount ||
              m_func->GetJITFunctionBody()->IsAsmJsMode(),
        "ArgCount doesn't match StartCall count");
    //
    // Machine dependent lowering
    //

    if (callInstr->m_opcode != Js::OpCode::AsmJsCallI)
    {
        // Push argCount
        IR::IntConstOpnd *argCountOpnd = Lowerer::MakeCallInfoConst(callFlags, argCount, m_func);
        if (callInfoOpndRef)
        {
            argCountOpnd->Use(m_func);
            *callInfoOpndRef = argCountOpnd;
        }
        Lowerer::InsertMove(this->GetArgSlotOpnd(1 + extraParams), argCountOpnd, callInstr);
    }
    startCallInstr = this->LowerStartCall(startCallInstr);

    const uint32 argSlots = argCount + 1 + extraParams; // + 1 for call flags
    this->m_func->m_argSlotsForFunctionsCalled = max(this->m_func->m_argSlotsForFunctionsCalled, argSlots);

    if (m_func->GetJITFunctionBody()->IsAsmJsMode())
    {
        IR::Opnd * functionObjOpnd = callInstr->UnlinkSrc1();
        GeneratePreCall(callInstr, functionObjOpnd, cfgInsertLoc->GetNextRealInstr());
    }

    return argSlots;
}

void
LowererMDArch::SetMaxArgSlots(Js::ArgSlot actualCount /*including this*/)
{
    Js::ArgSlot offset = 3;//For function object & callInfo & this
    if (this->m_func->m_argSlotsForFunctionsCalled < (uint32) (actualCount + offset))
    {
        this->m_func->m_argSlotsForFunctionsCalled = (uint32)(actualCount + offset);
    }
    return;
}

void
LowererMDArch::GenerateMemInit(IR::RegOpnd * opnd, int32 offset, size_t value, IR::Instr * insertBeforeInstr, bool isZeroed)
{
    IRType type = TyVar;
    if (isZeroed)
    {
        if (value == 0)
        {
            // Recycler memory are zero initialized
            return;
        }

        type = value <= UINT_MAX ?
            (value <= USHORT_MAX ?
            (value <= UCHAR_MAX ? TyUint8 : TyUint16) :
                TyUint32) :
            type;
    }
    Func * func = this->m_func;
    lowererMD->GetLowerer()->InsertMove(IR::IndirOpnd::New(opnd, offset, type, func), IR::IntConstOpnd::New(value, type, func), insertBeforeInstr);
}

IR::Instr *
LowererMDArch::LowerCallIDynamic(IR::Instr *callInstr, IR::Instr*saveThisArgOutInstr, IR::Opnd *argsLength, ushort callFlags, IR::Instr * insertBeforeInstrForCFG)
{
    callInstr->InsertBefore(saveThisArgOutInstr); //Move this Argout next to call;
    this->LoadDynamicArgument(saveThisArgOutInstr, 3); //this pointer is the 3rd argument

    /*callInfo*/
    if (callInstr->m_func->IsInlinee())
    {
        Assert(argsLength->AsIntConstOpnd()->GetValue() == callInstr->m_func->actualCount);
        this->SetMaxArgSlots((Js::ArgSlot)callInstr->m_func->actualCount);
    }
    else
    {
        callInstr->InsertBefore(IR::Instr::New(Js::OpCode::ADD, argsLength, argsLength, IR::IntConstOpnd::New(1, TyMachReg, this->m_func), this->m_func));
        this->SetMaxArgSlots(Js::InlineeCallInfo::MaxInlineeArgoutCount);
    }
    callInstr->InsertBefore(IR::Instr::New(Js::OpCode::MOV, this->GetArgSlotOpnd(2), argsLength, this->m_func));

    IR::Opnd    *funcObjOpnd = callInstr->UnlinkSrc1();
    GeneratePreCall(callInstr, funcObjOpnd, insertBeforeInstrForCFG);

    // Normally for dynamic calls we move 4 args to registers and push remaining
    // args onto stack (Windows convention, and unchanged on xplat). We need to
    // manully home 4 args. inlinees lower differently and follow platform ABI.
    // So we need to manually home actualArgsCount + 2 args (function, callInfo).
    const uint32 homeArgs = callInstr->m_func->IsInlinee() ?
                                callInstr->m_func->actualCount + 2 : 4;
    LowerCall(callInstr, homeArgs);

    return callInstr;
}

void
LowererMDArch::GenerateFunctionObjectTest(IR::Instr * callInstr, IR::RegOpnd  *functionObjOpnd, bool isHelper, IR::LabelInstr* continueAfterExLabel /* = nullptr */)
{
    AssertMsg(!m_func->IsJitInDebugMode() || continueAfterExLabel, "When jit is in debug mode, continueAfterExLabel must be provided otherwise continue after exception may cause AV.");

    IR::RegOpnd *functionObjRegOpnd = functionObjOpnd->AsRegOpnd();
    IR::Instr * insertBeforeInstr = callInstr;

    // Need check and error if we are calling a tagged int.
    if (!functionObjRegOpnd->IsNotTaggedValue())
    {
        IR::LabelInstr * helperLabel = IR::LabelInstr::New(Js::OpCode::Label, this->m_func, true);
        if (this->lowererMD->GenerateObjectTest(functionObjRegOpnd, callInstr, helperLabel))
        {

            IR::LabelInstr * callLabel = IR::LabelInstr::New(Js::OpCode::Label, this->m_func, isHelper);
            IR::Instr* instr = IR::BranchInstr::New(Js::OpCode::JMP, callLabel, this->m_func);
            callInstr->InsertBefore(instr);

            callInstr->InsertBefore(helperLabel);
            callInstr->InsertBefore(callLabel);

            insertBeforeInstr = callLabel;
            lowererMD->m_lowerer->GenerateRuntimeError(insertBeforeInstr, JSERR_NeedFunction);

            if (continueAfterExLabel)
            {
                // Under debugger the RuntimeError (exception) can be ignored, generate branch to jmp to safe place
                // (which would normally be debugger bailout check).
                IR::BranchInstr* continueAfterEx = IR::BranchInstr::New(LowererMD::MDUncondBranchOpcode, continueAfterExLabel, this->m_func);
                insertBeforeInstr->InsertBefore(continueAfterEx);
            }
        }
    }
}

void
LowererMDArch::GeneratePreCall(IR::Instr * callInstr, IR::Opnd  *functionObjOpnd, IR::Instr * insertBeforeInstrForCFGCheck)
{
    if (insertBeforeInstrForCFGCheck == nullptr)
    {
        insertBeforeInstrForCFGCheck = callInstr;
    }

    IR::RegOpnd * functionTypeRegOpnd = nullptr;
    IR::IndirOpnd * entryPointIndirOpnd = nullptr;
    if (callInstr->m_opcode == Js::OpCode::AsmJsCallI)
    {
        functionTypeRegOpnd = IR::RegOpnd::New(TyMachReg, m_func);

        IR::IndirOpnd* functionInfoIndirOpnd = IR::IndirOpnd::New(functionObjOpnd->AsRegOpnd(), Js::RecyclableObject::GetOffsetOfType(), TyMachReg, m_func);

        IR::Instr* instr = IR::Instr::New(Js::OpCode::MOV, functionTypeRegOpnd, functionInfoIndirOpnd, m_func);

        insertBeforeInstrForCFGCheck->InsertBefore(instr);

        functionInfoIndirOpnd = IR::IndirOpnd::New(functionTypeRegOpnd, Js::ScriptFunctionType::GetEntryPointInfoOffset(), TyMachReg, m_func);
        instr = IR::Instr::New(Js::OpCode::MOV, functionTypeRegOpnd, functionInfoIndirOpnd, m_func);
        insertBeforeInstrForCFGCheck->InsertBefore(instr);

        uint32 entryPointOffset = Js::ProxyEntryPointInfo::GetAddressOffset();

        entryPointIndirOpnd = IR::IndirOpnd::New(functionTypeRegOpnd, entryPointOffset, TyMachReg, m_func);
    }
    else
    {
        // For calls to fixed functions we load the function's type directly from the known (hard-coded) function object address.
        // For other calls, we need to load it from the function object stored in a register operand.
        if (functionObjOpnd->IsAddrOpnd() && functionObjOpnd->AsAddrOpnd()->m_isFunction)
        {
            functionTypeRegOpnd = this->lowererMD->m_lowerer->GenerateFunctionTypeFromFixedFunctionObject(insertBeforeInstrForCFGCheck, functionObjOpnd);
        }
        else if (functionObjOpnd->IsRegOpnd())
        {
            AssertMsg(functionObjOpnd->AsRegOpnd()->m_sym->IsStackSym(), "Expected call target to be a stack symbol.");

            functionTypeRegOpnd = IR::RegOpnd::New(TyMachReg, m_func);
            // functionTypeRegOpnd(RAX) = MOV function->type
            {
                IR::IndirOpnd * functionTypeIndirOpnd = IR::IndirOpnd::New(functionObjOpnd->AsRegOpnd(),
                    Js::DynamicObject::GetOffsetOfType(), TyMachReg, m_func);
                IR::Instr * mov = IR::Instr::New(Js::OpCode::MOV, functionTypeRegOpnd, functionTypeIndirOpnd, m_func);
                insertBeforeInstrForCFGCheck->InsertBefore(mov);
            }
        }
        else
        {
            AnalysisAssertMsg(false, "Unexpected call target operand type.");
        }
        // entryPointRegOpnd(RAX) = MOV type->entryPoint
        entryPointIndirOpnd = IR::IndirOpnd::New(functionTypeRegOpnd, Js::Type::GetOffsetOfEntryPoint(), TyMachPtr, m_func);
    }

    IR::RegOpnd *entryPointRegOpnd = functionTypeRegOpnd;
    entryPointRegOpnd->m_isCallArg = true;

    IR::Instr *mov = IR::Instr::New(Js::OpCode::MOV, entryPointRegOpnd, entryPointIndirOpnd, m_func);
    insertBeforeInstrForCFGCheck->InsertBefore(mov);

    // entryPointRegOpnd(RAX) = CALL entryPointRegOpnd(RAX)
    callInstr->SetSrc1(entryPointRegOpnd);

#if defined(_CONTROL_FLOW_GUARD)
    // verify that the call target is valid (CFG Check)
    if (!PHASE_OFF(Js::CFGInJitPhase, this->m_func))
    {
        this->lowererMD->GenerateCFGCheck(entryPointRegOpnd, insertBeforeInstrForCFGCheck);
    }
#endif

    // Setup the first call argument - pointer to the function being called.
    IR::Instr * instrMovArg1 = IR::Instr::New(Js::OpCode::MOV, GetArgSlotOpnd(1), functionObjOpnd, m_func);
    callInstr->InsertBefore(instrMovArg1);
}

IR::Instr *
LowererMDArch::LowerCallI(IR::Instr * callInstr, ushort callFlags, bool isHelper, IR::Instr * insertBeforeInstrForCFG)
{
    AssertMsg(this->helperCallArgsCount == 0, "We don't support nested helper calls yet");

    IR::Opnd * functionObjOpnd = callInstr->UnlinkSrc1();
    IR::Instr * insertBeforeInstrForCFGCheck = callInstr;

    // If this is a call for new, we already pass the function operand through NewScObject,
    // which checks if the function operand is a real function or not, don't need to add a check again
    // If this is a call to a fixed function, we've already verified that the target is, indeed, a function.
    if (callInstr->m_opcode != Js::OpCode::CallIFixed && !(callFlags & Js::CallFlags_New))
    {
        Assert(functionObjOpnd->IsRegOpnd());
        IR::LabelInstr* continueAfterExLabel = Lowerer::InsertContinueAfterExceptionLabelForDebugger(m_func, callInstr, isHelper);
        GenerateFunctionObjectTest(callInstr, functionObjOpnd->AsRegOpnd(), isHelper, continueAfterExLabel);
    }
    else if (insertBeforeInstrForCFG != nullptr)
    {
        RegNum dstReg = insertBeforeInstrForCFG->GetDst()->AsRegOpnd()->GetReg();
        AssertMsg(dstReg == RegArg2 || dstReg == RegArg3, "NewScObject should insert the first Argument in RegArg2/RegArg3 only based on Spread call or not.");
        insertBeforeInstrForCFGCheck = insertBeforeInstrForCFG;
    }

    GeneratePreCall(callInstr, functionObjOpnd, insertBeforeInstrForCFGCheck);

    // We need to get the calculated CallInfo in SimpleJit because that doesn't include any changes for stack alignment
    IR::IntConstOpnd *callInfo = nullptr;
    int32 argCount = LowerCallArgs(callInstr, callFlags, 1, &callInfo);


    IR::Opnd *const finalDst = callInstr->GetDst();

    // x64 keeps track of argCount for us, so pass just an arbitrary value there
    IR::Instr* ret = this->LowerCall(callInstr, argCount);

    IR::AutoReuseOpnd autoReuseSavedFunctionObjOpnd;
    if (callInstr->IsJitProfilingInstr())
    {
        Assert(callInstr->m_func->IsSimpleJit());
        Assert(!CONFIG_FLAG(NewSimpleJit));

        if(finalDst &&
            finalDst->IsRegOpnd() &&
            functionObjOpnd->IsRegOpnd() &&
            finalDst->AsRegOpnd()->m_sym == functionObjOpnd->AsRegOpnd()->m_sym)
        {
            // The function object sym is going to be overwritten, so save it in a temp for profiling
            IR::RegOpnd *const savedFunctionObjOpnd = IR::RegOpnd::New(functionObjOpnd->GetType(), callInstr->m_func);
            autoReuseSavedFunctionObjOpnd.Initialize(savedFunctionObjOpnd, callInstr->m_func);
            Lowerer::InsertMove(savedFunctionObjOpnd, functionObjOpnd, callInstr->m_next);
            functionObjOpnd = savedFunctionObjOpnd;
        }

        auto instr = callInstr->AsJitProfilingInstr();
        ret = this->lowererMD->m_lowerer->GenerateCallProfiling(
            instr->profileId,
            instr->inlineCacheIndex,
            instr->GetDst(),
            functionObjOpnd,
            callInfo,
            instr->isProfiledReturnCall,
            callInstr,
            ret);
    }
    return ret;
}

static inline IRType ExtendHelperArg(IRType type)
{
#ifdef __clang__
    // clang expects caller to extend arg size to int
    switch (type)
    {
        case TyInt8:
        case TyInt16:
            return TyInt32;
        case TyUint8:
        case TyUint16:
            return TyUint32;
    }
#endif
    return type;
}

IR::Instr *
LowererMDArch::LowerCall(IR::Instr * callInstr, uint32 argCount)
{
    UNREFERENCED_PARAMETER(argCount);
    IR::Instr *retInstr = callInstr;
    callInstr->m_opcode = Js::OpCode::CALL;

    // This is required here due to calls create during lowering
    callInstr->m_func->SetHasCallsOnSelfAndParents();

    if (callInstr->GetDst())
    {
        IR::Opnd *       dstOpnd;

        this->lowererMD->ForceDstToReg(callInstr);
        dstOpnd = callInstr->GetDst();
        IRType dstType = dstOpnd->GetType();
        Js::OpCode       assignOp = GetAssignOp(dstType);

        if (callInstr->GetSrc1()->IsHelperCallOpnd())
        {
            // Truncate the result of a conversion to 32-bit int, because the C++ code doesn't.
            IR::HelperCallOpnd *helperOpnd = callInstr->GetSrc1()->AsHelperCallOpnd();
            if (helperOpnd->m_fnHelper == IR::HelperConv_ToInt32 ||
                helperOpnd->m_fnHelper == IR::HelperConv_ToInt32_Full ||
                helperOpnd->m_fnHelper == IR::HelperConv_ToInt32Core ||
                helperOpnd->m_fnHelper == IR::HelperConv_ToUInt32 ||
                helperOpnd->m_fnHelper == IR::HelperConv_ToUInt32_Full ||
                helperOpnd->m_fnHelper == IR::HelperConv_ToUInt32Core)
            {
                assignOp = Js::OpCode::MOV_TRUNC;
            }
        }

        IR::Instr * movInstr = callInstr->SinkDst(assignOp);
        RegNum      reg = GetRegReturn(dstType);

        callInstr->GetDst()->AsRegOpnd()->SetReg(reg);
        movInstr->GetSrc1()->AsRegOpnd()->SetReg(reg);
        retInstr = movInstr;
    }

    //
    // assign the arguments to appropriate positions
    //

    AssertMsg(this->helperCallArgsCount >= 0, "Fatal. helper call arguments ought to be positive");
    AssertMsg(this->helperCallArgsCount < MaxArgumentsToHelper && MaxArgumentsToHelper < 255, "Too many helper call arguments");

    uint16 argsLeft = static_cast<uint16>(this->helperCallArgsCount);

    // Sys V x64 ABI assigns int and xmm arg registers separately.
    // e.g. args:   int, double, int, double, int, double
    //  Windows:    int0, xmm1, int2, xmm3, stack, stack
    //  Sys V:      int0, xmm0, int1, xmm1, int2, xmm2
#ifdef _WIN32
#define _V_ARG_INDEX(index) index
#else
    uint16 _vindex[MaxArgumentsToHelper];
    {
        uint16 intIndex = 1, doubleIndex = 1, stackIndex = IntArgRegsCount + 1;
        for (int i = 0; i < this->helperCallArgsCount; i++)
        {
            IR::Opnd * helperSrc = this->helperCallArgs[this->helperCallArgsCount - 1 - i];
            IRType type = helperSrc->GetType();
            if (IRType_IsFloat(type) || IRType_IsSimd128(type))
            {
                if (doubleIndex <= XmmArgRegsCount)
                {
                    _vindex[i] = doubleIndex++;
                }
                else
                {
                    _vindex[i] = stackIndex++;
                }
            }
            else
            {
                if (intIndex <= IntArgRegsCount)
                {
                    _vindex[i] = intIndex++;
                }
                else
                {
                    _vindex[i] = stackIndex++;
                }
            }
        }
    }
#define _V_ARG_INDEX(index) _vindex[(index) - 1]
#endif

    // xplat NOTE: Lower often loads "known args" with LoadHelperArgument() and
    // variadic JS runtime args with LowerCallArgs(). So the full args length is
    //      this->helperCallArgsCount + argCount
    // "argCount > 0" indicates we have variadic JS runtime args and needs to
    // manually home registers on xplat.
    const bool shouldHomeParams = argCount > 0;

    while (argsLeft > 0)
    {
        IR::Opnd * helperSrc = this->helperCallArgs[this->helperCallArgsCount - argsLeft];
        uint16 index = _V_ARG_INDEX(argsLeft);
        StackSym * helperSym = m_func->m_symTable->GetArgSlotSym(index);
        helperSym->m_type = ExtendHelperArg(helperSrc->GetType());
        Lowerer::InsertMove(
            this->GetArgSlotOpnd(index, helperSym, /*isHelper*/!shouldHomeParams),
            helperSrc,
            callInstr, false);
        --argsLeft;
    }

#ifndef _WIN32
    // Manually home args
    if (shouldHomeParams)
    {
        const int callArgCount = this->helperCallArgsCount + static_cast<int>(argCount);

        int argRegs = min(callArgCount, static_cast<int>(XmmArgRegsCount));

        for (int i = argRegs; i > 0; i--)
        {
            IRType type     = this->xplatCallArgs.args[i];
            bool isFloatArg = this->xplatCallArgs.IsFloat(i);

            if ( i > IntArgRegsCount && !isFloatArg ) continue;

            StackSym * sym = this->m_func->m_symTable->GetArgSlotSym(static_cast<uint16>(i));
            RegNum reg = GetRegFromArgPosition(isFloatArg, i);

            IR::RegOpnd *regOpnd = IR::RegOpnd::New(nullptr, reg, type, this->m_func);
            regOpnd->m_isCallArg = true;

            Lowerer::InsertMove(
                IR::SymOpnd::New(sym, type, this->m_func),
                regOpnd,
                callInstr, false);
        }
    }
    this->xplatCallArgs.Reset();
#endif // !_WIN32

    //
    // load the address into a register because we cannot directly access 64 bit constants
    // in CALL instruction. Non helper call methods will already be accessed indirectly.
    //
    // Skip this for bailout calls. The register allocator will lower that as appropriate, without affecting spill choices.
    //
    // Also skip this for relocatable helper calls. These will be turned into indirect
    // calls in lower.

    if (callInstr->GetSrc1()->IsHelperCallOpnd())
    {
        // Helper calls previously do not have bailouts except for bailout call. However with LazyBailOut, they can now have
        // bailouts as well. Handle them the same way as before.
        if (!callInstr->HasBailOutInfo() || callInstr->OnlyHasLazyBailOut())
        {
            IR::RegOpnd *targetOpnd = IR::RegOpnd::New(StackSym::New(TyMachPtr, m_func), RegRAX, TyMachPtr, this->m_func);
            IR::Instr   *movInstr = IR::Instr::New(Js::OpCode::MOV, targetOpnd, callInstr->GetSrc1(), this->m_func);
            targetOpnd->m_isCallArg = true;

            callInstr->UnlinkSrc1();
            callInstr->SetSrc1(targetOpnd);
            callInstr->InsertBefore(movInstr);
        }
    }

    if (callInstr->HasLazyBailOut())
    {
        BailOutInfo *bailOutInfo = callInstr->GetBailOutInfo();
        if (bailOutInfo->bailOutRecord == nullptr)
        {
            bailOutInfo->bailOutRecord = NativeCodeDataNewZ(
                this->m_func->GetNativeCodeDataAllocator(),
                BailOutRecord,
                bailOutInfo->bailOutOffset,
                bailOutInfo->polymorphicCacheIndex,
                callInstr->GetBailOutKind(),
                bailOutInfo->bailOutFunc
            );
        }
    }

    //
    // Reset the call
    //

    this->m_func->m_argSlotsForFunctionsCalled  = max(this->m_func->m_argSlotsForFunctionsCalled , (uint32)this->helperCallArgsCount);
    this->helperCallArgsCount = 0;

    return retInstr;
}

//
// Returns the opnd where the corresponding argument would have been stored. On amd64,
// the first 4 arguments go in registers and the rest are on stack.
//
IR::Opnd *
LowererMDArch::GetArgSlotOpnd(uint16 index, StackSym * argSym, bool isHelper /*= false*/)
{
    Assert(index != 0);

    uint16 argPosition = index;
    IR::Opnd *argSlotOpnd = nullptr;

    if (argSym != nullptr)
    {
        argSym->m_offset = (index - 1) * MachPtr;
        argSym->m_allocated = true;
    }

    IRType type = argSym ? argSym->GetType() : TyMachReg;
    const bool isFloatArg = IRType_IsFloat(type) || IRType_IsSimd128(type);
    RegNum reg = GetRegFromArgPosition(isFloatArg, argPosition);
#ifndef _WIN32
    if (isFloatArg && argPosition <= XmmArgRegsCount)
    {
        this->xplatCallArgs.SetFloat(argPosition);
    }
#endif

    if (reg != RegNOREG)
    {
        IR::RegOpnd *regOpnd = IR::RegOpnd::New(argSym, reg, type, m_func);
        regOpnd->m_isCallArg = true;

        argSlotOpnd = regOpnd;
    }
    else
    {
        if (argSym == nullptr)
        {
            argSym = this->m_func->m_symTable->GetArgSlotSym(index);
        }

#ifndef _WIN32
        // helper does not home args, adjust stack offset
        if (isHelper)
        {
            const uint16 argIndex = index - IntArgRegsCount;
            argSym->m_offset = (argIndex - 1) * MachPtr;
        }
#endif

        argSlotOpnd = IR::SymOpnd::New(argSym, type, this->m_func);
    }

    return argSlotOpnd;
}

IR::Instr *
LowererMDArch::LowerAsmJsCallE(IR::Instr *callInstr)
{
    IR::IntConstOpnd *callInfo = nullptr;
    int32 argCount = this->LowerCallArgs(callInstr, Js::CallFlags_Value, 1, &callInfo);

    IR::Instr* ret = this->LowerCall(callInstr, argCount);

    return ret;
}

IR::Instr *
LowererMDArch::LowerAsmJsCallI(IR::Instr * callInstr)
{
    int32 argCount = this->LowerCallArgs(callInstr, Js::CallFlags_Value, 0);

    IR::Instr* ret = this->LowerCall(callInstr, argCount);

    return ret;
}

IR::Instr *
LowererMDArch::LowerWasmArrayBoundsCheck(IR::Instr * instr, IR::Opnd *addrOpnd)
{
    IR::IndirOpnd * indirOpnd = addrOpnd->AsIndirOpnd();
    IR::RegOpnd * indexOpnd = indirOpnd->GetIndexOpnd();
    if (indexOpnd && !indexOpnd->IsIntegral32())
    {
        // We don't expect the index to be anything other than an int32 or uint32
        // Having an int32 index guaranties that int64 index add doesn't overflow
        // If we're wrong, just throw index out of range
        Assert(UNREACHED);
        lowererMD->m_lowerer->GenerateThrow(IR::IntConstOpnd::New(WASMERR_ArrayIndexOutOfRange, TyInt32, m_func), instr);
        return instr;
    }
    if (m_func->GetJITFunctionBody()->UsesWAsmJsFastVirtualBuffer())
    {
        return instr;
    }

    Assert(instr->GetSrc2());
    IR::LabelInstr * helperLabel = Lowerer::InsertLabel(true, instr);
    IR::LabelInstr * loadLabel = Lowerer::InsertLabel(false, instr);
    IR::LabelInstr * doneLabel = Lowerer::InsertLabel(false, instr);

    // Find array buffer length
    uint32 offset = indirOpnd->GetOffset();
    IR::Opnd *arrayLenOpnd = instr->GetSrc2();
    IR::Int64ConstOpnd * constOffsetOpnd = IR::Int64ConstOpnd::New((int64)addrOpnd->GetSize() + (int64)offset, TyInt64, m_func);
    IR::Opnd *cmpOpnd;
    if (indexOpnd != nullptr)
    {
        // Compare index + memop access length and array buffer length, and generate RuntimeError if greater
        cmpOpnd = IR::RegOpnd::New(TyInt64, m_func);
        Lowerer::InsertAdd(true, cmpOpnd, indexOpnd, constOffsetOpnd, helperLabel);
    }
    else
    {
        cmpOpnd = constOffsetOpnd;
    }
    lowererMD->m_lowerer->InsertCompareBranch(cmpOpnd, arrayLenOpnd, Js::OpCode::BrGt_A, true, helperLabel, helperLabel);
    lowererMD->m_lowerer->GenerateThrow(IR::IntConstOpnd::New(WASMERR_ArrayIndexOutOfRange, TyInt32, m_func), loadLabel);
    Lowerer::InsertBranch(Js::OpCode::Br, loadLabel, helperLabel);
    return doneLabel;
}

void
LowererMDArch::LowerAtomicStore(IR::Opnd * dst, IR::Opnd * src1, IR::Instr * insertBeforeInstr)
{
    Assert(IRType_IsNativeInt(dst->GetType()));
    Assert(IRType_IsNativeInt(src1->GetType()));
    IR::RegOpnd* tmpSrc = IR::RegOpnd::New(dst->GetType(), m_func);
    Lowerer::InsertMove(tmpSrc, src1, insertBeforeInstr);

    // Put src1 as dst to make sure we know that register is modified
    IR::Instr* xchgInstr = IR::Instr::New(Js::OpCode::XCHG, tmpSrc, tmpSrc, dst, insertBeforeInstr->m_func);
    insertBeforeInstr->InsertBefore(xchgInstr);
}

void
LowererMDArch::LowerAtomicLoad(IR::Opnd * dst, IR::Opnd * src1, IR::Instr * insertBeforeInstr)
{
    Assert(IRType_IsNativeInt(dst->GetType()));
    Assert(IRType_IsNativeInt(src1->GetType()));
    IR::Instr* newMove = Lowerer::InsertMove(dst, src1, insertBeforeInstr);

    if (m_func->GetJITFunctionBody()->UsesWAsmJsFastVirtualBuffer())
    {
        // We need to have an AV when accessing out of bounds memory even if the dst is not used
        // Make sure LinearScan doesn't dead store this instruction
        newMove->hasSideEffects = true;
    }

    // Need to add Memory Barrier before the load
    // MemoryBarrier is implemented with `lock or [rsp], 0` on x64
    IR::IndirOpnd* stackTop = IR::IndirOpnd::New(
        IR::RegOpnd::New(nullptr, RegRSP, TyMachReg, m_func),
        0,
        TyMachReg,
        m_func
    );
    IR::IntConstOpnd* zero = IR::IntConstOpnd::New(0, TyMachReg, m_func);
    IR::Instr* memoryBarrier = IR::Instr::New(Js::OpCode::LOCKOR, stackTop, stackTop, zero, m_func);
    newMove->InsertBefore(memoryBarrier);
}

IR::Instr*
LowererMDArch::LowerAsmJsLdElemHelper(IR::Instr * instr, bool isSimdLoad /*= false*/, bool checkEndOffset /*= false*/)
{
    IR::Instr* done;
    IR::Opnd * src1 = instr->UnlinkSrc1();
    IRType type = src1->GetType();
    IR::RegOpnd * indexOpnd = src1->AsIndirOpnd()->GetIndexOpnd();
    const uint8 dataWidth = instr->dataWidth;
    Assert(isSimdLoad == false || dataWidth == 4 || dataWidth == 8 || dataWidth == 12 || dataWidth == 16);

#if ENABLE_FAST_ARRAYBUFFER
    if (CONFIG_FLAG(WasmFastArray) && m_func->GetJITFunctionBody()->IsWasmFunction())
    {
        return instr;
    }
#endif

#ifdef _WIN32
    // For x64, bound checks are required only for SIMD loads.
    if (isSimdLoad)
#else
    // xplat: Always do bound check. We don't support out-of-bound access violation recovery.
    if (true)
#endif
    {
        IR::LabelInstr * helperLabel = Lowerer::InsertLabel(true, instr);
        IR::LabelInstr * loadLabel = Lowerer::InsertLabel(false, instr);
        IR::LabelInstr * doneLabel = Lowerer::InsertLabel(false, instr);
        IR::Opnd *cmpOpnd;
        if (indexOpnd)
        {
            cmpOpnd = indexOpnd;
        }
        else
        {
            cmpOpnd = IR::IntConstOpnd::New(src1->AsIndirOpnd()->GetOffset(), TyUint32, m_func);
        }

        // if dataWidth != byte per element, we need to check end offset
        if (checkEndOffset)
        {
            IR::RegOpnd *tmp = IR::RegOpnd::New(cmpOpnd->GetType(), m_func);
            // MOV tmp, cmpOnd
            Lowerer::InsertMove(tmp, cmpOpnd, helperLabel);
            // ADD tmp, dataWidth
            Lowerer::InsertAdd(true, tmp, tmp, IR::IntConstOpnd::New((uint32)dataWidth, tmp->GetType(), m_func, true), helperLabel);
            // JB helper
            Lowerer::InsertBranch(Js::OpCode::JB, helperLabel, helperLabel);
            // CMP tmp, size
            // JG  $helper
            lowererMD->m_lowerer->InsertCompareBranch(tmp, instr->UnlinkSrc2(), Js::OpCode::BrGt_A, true, helperLabel, helperLabel);
        }
        else
        {
#ifdef ENABLE_WASM_SIMD
            if (m_func->GetJITFunctionBody()->IsWasmFunction() && src1->AsIndirOpnd()->GetOffset()) //WASM
            {
                IR::RegOpnd *tmp = IR::RegOpnd::New(cmpOpnd->GetType(), m_func);
                // MOV tmp, cmpOnd
                Lowerer::InsertMove(tmp, cmpOpnd, helperLabel);
                // ADD tmp, offset
                Lowerer::InsertAdd(true, tmp, tmp, IR::IntConstOpnd::New((uint32)src1->AsIndirOpnd()->GetOffset(), tmp->GetType(), m_func), helperLabel);
                // JB helper
                Lowerer::InsertBranch(Js::OpCode::JB, helperLabel, helperLabel);
                lowererMD->m_lowerer->InsertCompareBranch(tmp, instr->UnlinkSrc2(), Js::OpCode::BrGe_A, true, helperLabel, helperLabel);
            }
            else
#endif
            {
                lowererMD->m_lowerer->InsertCompareBranch(cmpOpnd, instr->UnlinkSrc2(), Js::OpCode::BrGe_A, true, helperLabel, helperLabel);
            }
        }
        Lowerer::InsertBranch(Js::OpCode::Br, loadLabel, helperLabel);

        if (isSimdLoad)
        {
            lowererMD->m_lowerer->GenerateRuntimeError(loadLabel, JSERR_ArgumentOutOfRange, IR::HelperOp_RuntimeRangeError);
        }
        else
        {
            if (IRType_IsFloat(type))
            {
                Lowerer::InsertMove(instr->UnlinkDst(), IR::FloatConstOpnd::New(Js::NumberConstants::NaN, type, m_func), loadLabel);
            }
            else
            {
                Lowerer::InsertMove(instr->UnlinkDst(), IR::IntConstOpnd::New(0, TyInt8, m_func), loadLabel);
            }
        }

        Lowerer::InsertBranch(Js::OpCode::Br, doneLabel, loadLabel);
        done = doneLabel;
    }
    else
    {
        Assert(!instr->GetSrc2());
        done = instr;
    }
    return done;
}

IR::Instr*
LowererMDArch::LowerAsmJsStElemHelper(IR::Instr * instr, bool isSimdStore /*= false*/, bool checkEndOffset /*= false*/)
{
    IR::Instr* done;
    IR::Opnd * dst = instr->UnlinkDst();
    IR::RegOpnd * indexOpnd = dst->AsIndirOpnd()->GetIndexOpnd();
    const uint8 dataWidth = instr->dataWidth;

    Assert(isSimdStore == false || dataWidth == 4 || dataWidth == 8 || dataWidth == 12 || dataWidth == 16);

#ifdef _WIN32
    // For x64, bound checks are required only for SIMD loads.
    if (isSimdStore)
#else
    // xplat: Always do bound check. We don't support out-of-bound access violation recovery.
    if (true)
#endif
    {
        IR::LabelInstr * helperLabel = Lowerer::InsertLabel(true, instr);
        IR::LabelInstr * storeLabel = Lowerer::InsertLabel(false, instr);
        IR::LabelInstr * doneLabel = Lowerer::InsertLabel(false, instr);
        IR::Opnd * cmpOpnd;
        if (indexOpnd)
        {
            cmpOpnd = dst->AsIndirOpnd()->GetIndexOpnd();
        }
        else
        {
            cmpOpnd = IR::IntConstOpnd::New(dst->AsIndirOpnd()->GetOffset(), TyUint32, m_func);
        }

        // if dataWidth != byte per element, we need to check end offset
        if (checkEndOffset)
        {
            IR::RegOpnd *tmp = IR::RegOpnd::New(cmpOpnd->GetType(), m_func);
            // MOV tmp, cmpOnd
            Lowerer::InsertMove(tmp, cmpOpnd, helperLabel);
            // ADD tmp, dataWidth
            Lowerer::InsertAdd(true, tmp, tmp, IR::IntConstOpnd::New((uint32)dataWidth, tmp->GetType(), m_func, true), helperLabel);
            // JB helper
            Lowerer::InsertBranch(Js::OpCode::JB, helperLabel, helperLabel);
            // CMP tmp, size
            // JG  $helper
            lowererMD->m_lowerer->InsertCompareBranch(tmp, instr->UnlinkSrc2(), Js::OpCode::BrGt_A, true, helperLabel, helperLabel);
        }
        else
        {
#ifdef ENABLE_WASM_SIMD
            if (m_func->GetJITFunctionBody()->IsWasmFunction() && dst->AsIndirOpnd()->GetOffset()) //WASM
            {
                IR::RegOpnd *tmp = IR::RegOpnd::New(cmpOpnd->GetType(), m_func);
                // MOV tmp, cmpOnd
                Lowerer::InsertMove(tmp, cmpOpnd, helperLabel);
                // ADD tmp, offset
                Lowerer::InsertAdd(true, tmp, tmp, IR::IntConstOpnd::New((uint32)dst->AsIndirOpnd()->GetOffset(), tmp->GetType(), m_func), helperLabel);
                // JB helper
                Lowerer::InsertBranch(Js::OpCode::JB, helperLabel, helperLabel);
                lowererMD->m_lowerer->InsertCompareBranch(tmp, instr->UnlinkSrc2(), Js::OpCode::BrGe_A, true, helperLabel, helperLabel);
            }
            else
#endif
            {
                lowererMD->m_lowerer->InsertCompareBranch(cmpOpnd, instr->UnlinkSrc2(), Js::OpCode::BrGe_A, true, helperLabel, helperLabel);
            }
        }
        Lowerer::InsertBranch(Js::OpCode::Br, storeLabel, helperLabel);

        if (isSimdStore)
        {
            lowererMD->m_lowerer->GenerateRuntimeError(storeLabel, JSERR_ArgumentOutOfRange, IR::HelperOp_RuntimeRangeError);
        }

        Lowerer::InsertBranch(Js::OpCode::Br, doneLabel, storeLabel);
        done = doneLabel;
    }
    else
    {
        Assert(!instr->GetSrc2());
        done = instr;
    }

    return done;
}

///----------------------------------------------------------------------------
///
/// LowererMDArch::LowerStartCall
///
///
///----------------------------------------------------------------------------

IR::Instr *
LowererMDArch::LowerStartCall(IR::Instr * startCallInstr)
{
    startCallInstr->m_opcode = Js::OpCode::LoweredStartCall;
    return startCallInstr;
}

IR::Instr *
LowererMDArch::LoadInt64HelperArgument(IR::Instr * instrInsert, IR::Opnd * opndArg)
{
    return LoadHelperArgument(instrInsert, opndArg);
}

///----------------------------------------------------------------------------
///
/// LowererMDArch::LoadHelperArgument
///
///     Assign register or push on stack as per AMD64 calling convention
///
///----------------------------------------------------------------------------

IR::Instr *
LowererMDArch::LoadHelperArgument(IR::Instr *instr, IR::Opnd *opndArg)
{
    IR::Opnd *destOpnd;
    IR::Instr *instrToReturn;
    if(opndArg->IsImmediateOpnd())
    {
        destOpnd = opndArg;
        instrToReturn = instr;
    }
    else
    {
        destOpnd = IR::RegOpnd::New(opndArg->GetType(), this->m_func);
        instrToReturn = instr->m_prev;
        Lowerer::InsertMove(destOpnd, opndArg, instr, false);
        instrToReturn = instrToReturn->m_next;
    }

    helperCallArgs[helperCallArgsCount++] = destOpnd;
    AssertMsg(helperCallArgsCount < LowererMDArch::MaxArgumentsToHelper,
              "We do not yet support any no. of arguments to the helper");

    return instrToReturn;
}

IR::Instr *
LowererMDArch::LoadDynamicArgument(IR::Instr *instr, uint argNumber)
{
    Assert(instr->m_opcode == Js::OpCode::ArgOut_A_Dynamic);
    Assert(instr->GetSrc2() == nullptr);

    instr->m_opcode = Js::OpCode::MOV;
    IR::Opnd* dst = GetArgSlotOpnd((Js::ArgSlot) argNumber);
    instr->SetDst(dst);

    if (!dst->IsRegOpnd())
    {
        //TODO: Move it to legalizer.
        IR::RegOpnd *tempOpnd = IR::RegOpnd::New(TyMachReg, instr->m_func);
        instr->InsertBefore(IR::Instr::New(Js::OpCode::MOV, tempOpnd, instr->GetSrc1(), instr->m_func));
        instr->ReplaceSrc1(tempOpnd);
    }
    return instr;
}

IR::Instr *
LowererMDArch::LoadDynamicArgumentUsingLength(IR::Instr *instr)
{
    Assert(instr->m_opcode == Js::OpCode::ArgOut_A_Dynamic);
    IR::RegOpnd* src2 = instr->UnlinkSrc2()->AsRegOpnd();

    IR::Instr*mov = IR::Instr::New(Js::OpCode::MOV, IR::RegOpnd::New(TyMachReg, this->m_func), src2, this->m_func);
    instr->InsertBefore(mov);
    //We need store nth actuals, so stack location is after function object, callinfo & this pointer
    instr->InsertBefore(IR::Instr::New(Js::OpCode::ADD, mov->GetDst(), mov->GetDst(), IR::IntConstOpnd::New(3, TyMachReg, this->m_func), this->m_func));
    IR::RegOpnd *stackPointer   = IR::RegOpnd::New(nullptr, GetRegStackPointer(), TyMachReg, this->m_func);
    IR::IndirOpnd *actualsLocation = IR::IndirOpnd::New(stackPointer, mov->GetDst()->AsRegOpnd(), GetDefaultIndirScale(), TyMachReg, this->m_func);
    instr->SetDst(actualsLocation);

    instr->m_opcode = Js::OpCode::MOV;
    return instr;
}


IR::Instr *
LowererMDArch::LoadDoubleHelperArgument(IR::Instr * instrInsert, IR::Opnd * opndArg)
{
    IR::Opnd * float64Opnd;
    if (opndArg->GetType() == TyFloat32)
    {
        float64Opnd = IR::RegOpnd::New(TyFloat64, m_func);
        IR::Instr * instr = IR::Instr::New(Js::OpCode::CVTSS2SD, float64Opnd, opndArg, this->m_func);
        instrInsert->InsertBefore(instr);
    }
    else
    {
        float64Opnd = opndArg;
    }

    Assert(opndArg->IsFloat());

    return LoadHelperArgument(instrInsert, opndArg);
}

IR::Instr *
LowererMDArch::LoadFloatHelperArgument(IR::Instr * instrInsert, IR::Opnd * opndArg)
{
    Assert(opndArg->IsFloat32());
    return LoadHelperArgument(instrInsert, opndArg);
}

//
// Emits the code to allocate 'size' amount of space on stack. for values smaller than PAGE_SIZE
// this will just emit sub rsp,size otherwise calls _chkstk.
//
void
LowererMDArch::GenerateStackAllocation(IR::Instr *instr, uint32 size)
{
    Assert(size > 0);

    IR::RegOpnd *       rspOpnd         = IR::RegOpnd::New(nullptr, RegRSP, TyMachReg, this->m_func);

    //review: size should fit in 32bits

    IR::IntConstOpnd *  stackSizeOpnd   = IR::IntConstOpnd::New(size, TyMachReg, this->m_func);

    if (size <= PAGESIZE)
    {
        // Generate SUB RSP, stackSize

        IR::Instr * subInstr = IR::Instr::New(Js::OpCode::SUB,
            rspOpnd, rspOpnd, stackSizeOpnd, this->m_func);

        instr->InsertAfter(subInstr);
    }
    else
    {
        // Generate _chkstk call

        //
        // REVIEW: Call to helper functions assume the address of the variable to be present in
        // RAX. But _chkstk method accepts argument in RAX. Hence handling this one manually.
        // fix this later when CALLHELPER leaved dependency on RAX.
        //

        IR::RegOpnd *raxOpnd = IR::RegOpnd::New(nullptr, RegRAX, TyMachReg, this->m_func);
        IR::RegOpnd *rcxOpnd = IR::RegOpnd::New(nullptr, RegRCX, TyMachReg, this->m_func);

        IR::Instr * subInstr = IR::Instr::New(Js::OpCode::SUB, rspOpnd, rspOpnd, stackSizeOpnd, this->m_func);
        instr->InsertAfter(subInstr);

        // Leave off the src until we've calculated it below.
        IR::Instr * callInstr = IR::Instr::New(Js::OpCode::Call, raxOpnd, rcxOpnd, this->m_func);
        instr->InsertAfter(callInstr);

        this->LowerCall(callInstr, 0);

        {
            IR::Instr   *movHelperAddrInstr = IR::Instr::New(
                Js::OpCode::MOV,
                rcxOpnd,
                IR::HelperCallOpnd::New(IR::HelperCRT_chkstk, this->m_func),
                this->m_func);

            instr->InsertAfter(movHelperAddrInstr);
        }

        Lowerer::InsertMove(raxOpnd, stackSizeOpnd, instr->m_next);
    }
}

void
LowererMDArch::MovArgFromReg2Stack(IR::Instr * instr, RegNum reg, uint16 slotNumber, IRType type)
{
    StackSym *          slotSym   = this->m_func->m_symTable->GetArgSlotSym(slotNumber + 1);
    slotSym->m_type = type;
    IR::SymOpnd *       dst       = IR::SymOpnd::New(slotSym, type,  this->m_func);
    IR::RegOpnd *       src       = IR::RegOpnd::New(nullptr, reg, type,  this->m_func);
    IR::Instr *         movInstr =  IR::Instr::New(GetAssignOp(type), dst, src, this->m_func);
    instr->InsertAfter(movInstr);
}

///----------------------------------------------------------------------------
///
/// LowererMDArch::LowerEntryInstr
///
///     Emit prolog.
///
///----------------------------------------------------------------------------
IR::Instr *
LowererMDArch::LowerEntryInstr(IR::EntryInstr * entryInstr)
{
    /*
     * push   rbp
     * mov    rbp,                       rsp
     * sub    rsp,                       localVariablesHeight + floatCalleeSavedRegsSize
     * movsdx qword ptr [rsp + 0],       xmm6  ------\
     * movsdx qword ptr [rsp + 8],       xmm7        |
     * ...                                           |
     * movsdx qword ptr [rsp + (N * 8)], xmmN        |- Callee saved registers.
     * push   rsi                                    |
     * ...                                           |
     * push   rbx                              ------/
     * sub    rsp,                       ArgumentsBacking
     */

    uint savedRegSize           = 0;
    IR::Instr *firstPrologInstr = nullptr;
    IR::Instr *lastPrologInstr  = nullptr;

    // PUSH used callee-saved registers.
    IR::Instr *secondInstr      = entryInstr->m_next;
    AssertMsg(secondInstr, "Instruction chain broken.");

    IR::RegOpnd *stackPointer   = IR::RegOpnd::New(nullptr, GetRegStackPointer(), TyMachReg, this->m_func);
    unsigned     xmmOffset      = 0;

    // PDATA doesn't seem to like two consecutive "SUB RSP, size" instructions. Temporarily save and
    // restore RBX always so that the pattern doesn't occur in the prolog.
    for (RegNum reg = (RegNum)(RegNOREG + 1); reg < RegNumCount; reg = (RegNum)(reg + 1))
    {
        if (LinearScan::IsCalleeSaved(reg) && (this->m_func->HasTry() || this->m_func->m_regsUsed.Test(reg)))
        {
            IRType       type      = RegTypes[reg];
            IR::RegOpnd *regOpnd   = IR::RegOpnd::New(nullptr, reg, type, this->m_func);

            if (type == TyFloat64)
            {
                IR::Instr *saveInstr = IR::Instr::New(Js::OpCode::MOVAPS,
                                                      IR::IndirOpnd::New(stackPointer,
                                                                         xmmOffset,
                                                                         type,
                                                                         this->m_func),
                                                      regOpnd,
                                                      this->m_func);

                xmmOffset += (MachDouble * 2);
                entryInstr->InsertAfter(saveInstr);

                m_func->m_prologEncoder.RecordXmmRegSave();
            }
            else
            {
                Assert(type == TyInt64);

                IR::Instr *pushInstr = IR::Instr::New(Js::OpCode::PUSH, this->m_func);
                pushInstr->SetSrc1(regOpnd);
                entryInstr->InsertAfter(pushInstr);

                m_func->m_prologEncoder.RecordNonVolRegSave();
                savedRegSize += MachPtr;
            }
        }
    }
    //
    // Now that we know the exact stack size, lets fix it for alignment
    // The stack on entry would be aligned. VC++ recommends that the stack
    // should always be 16 byte aligned.
    //
    uint32 argSlotsForFunctionsCalled = this->m_func->m_argSlotsForFunctionsCalled;

    if (Lowerer::IsArgSaveRequired(this->m_func))
    {
        if (argSlotsForFunctionsCalled < IntArgRegsCount)
            argSlotsForFunctionsCalled = IntArgRegsCount;
    }
    else
    {
        argSlotsForFunctionsCalled = 0;
    }

    uint32 stackArgsSize    = MachPtr * (argSlotsForFunctionsCalled + 1);

    this->m_func->m_localStackHeight = Math::Align<int32>(this->m_func->m_localStackHeight, 8);

    // Allocate the inlined arg out stack in the locals. Allocate an additional slot so that
    // we can unconditionally clear the first slot past the current frame.
    this->m_func->m_localStackHeight += m_func->GetMaxInlineeArgOutSize() + MachPtr;

    uint32 stackLocalsSize  = this->m_func->m_localStackHeight;
    if(xmmOffset != 0)
    {
        // Xmm registers need to be saved to 16-byte-aligned addresses. The stack locals size is aligned here and the total
        // size will be aligned below, which guarantees that the offset from rsp will be 16-byte-aligned.
        stackLocalsSize = ::Math::Align(stackLocalsSize + xmmOffset, static_cast<uint32>(MachDouble * 2));
    }

    uint32 totalStackSize   = stackLocalsSize +
                              stackArgsSize   +
                              savedRegSize;

    AssertMsg(0 == (totalStackSize % 8), "Stack should always be 8 byte aligned");
    uint32 alignmentPadding = (totalStackSize % 16) ? MachPtr : 0;

    stackArgsSize += alignmentPadding;
    Assert(
        xmmOffset == 0 ||
        ::Math::Align(stackArgsSize + savedRegSize, static_cast<uint32>(MachDouble * 2)) == stackArgsSize + savedRegSize);

    totalStackSize += alignmentPadding;
    if(totalStackSize > (1u << 20)) // 1 MB
    {
        // Total stack size is > 1 MB, let's just bail. There are things that need to be changed to allow using large stack
        // sizes, for instance in the unwind info, the offset to saved xmm registers can be (1 MB - 16) at most for the op-code
        // we're currently using (UWOP_SAVE_XMM128). To support larger offsets, we need to use a FAR version of the op-code.
        throw Js::OperationAbortedException();
    }

    if (m_func->HasInlinee())
    {
        this->m_func->GetJITOutput()->SetFrameHeight(this->m_func->m_localStackHeight);
    }

    //
    // This is the last instruction so should have been emitted before, register saves.
    // But we did not have 'savedRegSize' by then. So we saved secondInstr. We now insert w.r.t that
    // instruction.
    //
    this->m_func->SetArgsSize(stackArgsSize);
    this->m_func->SetSavedRegSize(savedRegSize);
    this->m_func->SetSpillSize(stackLocalsSize);

    if (secondInstr == entryInstr->m_next)
    {
        // There is no register save at all, just combine the stack allocation
        uint combineStackAllocationSize = stackArgsSize + stackLocalsSize;
        this->GenerateStackAllocation(secondInstr->m_prev, combineStackAllocationSize);
        m_func->m_prologEncoder.RecordAlloca(combineStackAllocationSize);
    }
    else
    {
        this->GenerateStackAllocation(secondInstr->m_prev, stackArgsSize);
        m_func->m_prologEncoder.RecordAlloca(stackArgsSize);

        // Allocate frame.
        if (stackLocalsSize)
        {
            this->GenerateStackAllocation(entryInstr, stackLocalsSize);
            m_func->m_prologEncoder.RecordAlloca(stackLocalsSize);
        }
    }

    lastPrologInstr = secondInstr->m_prev;
    Assert(lastPrologInstr != entryInstr);

    // Zero-initialize dedicated arguments slot.
    IR::Instr *movRax0 = nullptr;
    IR::Opnd *raxOpnd = nullptr;

    if ((this->m_func->HasArgumentSlot() &&
            (this->m_func->IsStackArgsEnabled() ||
            this->m_func->IsJitInDebugMode() ||
            // disabling apply inlining leads to explicit load from the zero-inited slot
            this->m_func->GetJITFunctionBody()->IsInlineApplyDisabled()))
#ifdef BAILOUT_INJECTION
        || Js::Configuration::Global.flags.IsEnabled(Js::BailOutFlag)
        || Js::Configuration::Global.flags.IsEnabled(Js::BailOutAtEveryLineFlag)
        || Js::Configuration::Global.flags.IsEnabled(Js::BailOutAtEveryByteCodeFlag)
        || Js::Configuration::Global.flags.IsEnabled(Js::BailOutByteCodeFlag)
#endif
        )
    {
        // TODO: Support mov [rbp - n], IMM64
        raxOpnd = IR::RegOpnd::New(nullptr, RegRAX, TyUint32, this->m_func);
        movRax0 = IR::Instr::New(Js::OpCode::XOR, raxOpnd, raxOpnd, raxOpnd, this->m_func);
        secondInstr->m_prev->InsertAfter(movRax0);

        IR::Opnd *opnd = LowererMD::CreateStackArgumentsSlotOpnd(this->m_func);
        IR::Instr *movNullInstr = IR::Instr::New(Js::OpCode::MOV, opnd, raxOpnd->UseWithNewType(TyMachReg, this->m_func), this->m_func);
        secondInstr->m_prev->InsertAfter(movNullInstr);
    }

    // Zero initialize the first inlinee frames argc.
    if (m_func->HasInlinee())
    {
        if(!movRax0)
        {
            raxOpnd = IR::RegOpnd::New(nullptr, RegRAX, TyUint32, this->m_func);
            movRax0 = IR::Instr::New(Js::OpCode::XOR, raxOpnd, raxOpnd, raxOpnd, this->m_func);
            secondInstr->m_prev->InsertAfter(movRax0);
        }

        StackSym *sym           = this->m_func->m_symTable->GetArgSlotSym((Js::ArgSlot)-1);
        sym->m_isInlinedArgSlot = true;
        sym->m_offset           = 0;
        IR::Opnd *dst           = IR::SymOpnd::New(sym, 0, TyMachReg, this->m_func);
        secondInstr->m_prev->InsertAfter(IR::Instr::New(Js::OpCode::MOV,
                                                        dst,
                                                        raxOpnd->UseWithNewType(TyMachReg, this->m_func),
                                                        this->m_func));
    }

    // Generate MOV RBP, RSP
    IR::RegOpnd * rbpOpnd = IR::RegOpnd::New(nullptr, RegRBP, TyMachReg, this->m_func);
    IR::RegOpnd * rspOpnd = IR::RegOpnd::New(nullptr, RegRSP, TyMachReg, this->m_func);

    IR::Instr * movInstr = IR::Instr::New(Js::OpCode::MOV, rbpOpnd, rspOpnd, this->m_func);
    entryInstr->InsertAfter(movInstr);

    // Generate PUSH RBP
    IR::Instr * pushInstr = IR::Instr::New(Js::OpCode::PUSH, this->m_func);
    pushInstr->SetSrc1(rbpOpnd);
    entryInstr->InsertAfter(pushInstr);
    m_func->m_prologEncoder.RecordNonVolRegSave();
    firstPrologInstr = pushInstr;


    //
    // Insert pragmas that tell the prolog encoder the extent of the prolog.
    //
    firstPrologInstr->InsertBefore(IR::PragmaInstr::New(Js::OpCode::PrologStart, 0, m_func));
    lastPrologInstr->InsertAfter(IR::PragmaInstr::New(Js::OpCode::PrologEnd, 0, m_func));

#ifdef _WIN32 // home registers
    //
    // Now store all the arguments in the register in the stack slots
    //
    if (m_func->GetJITFunctionBody()->IsAsmJsMode() && !m_func->IsLoopBody())
    {
        uint16 offset = 2;
        this->MovArgFromReg2Stack(entryInstr, RegRCX, 1);
        for (uint16 i = 0; i < m_func->GetJITFunctionBody()->GetAsmJsInfo()->GetArgCount() && i < 3; i++)
        {
            switch (m_func->GetJITFunctionBody()->GetAsmJsInfo()->GetArgType(i))
            {
            case Js::AsmJsVarType::Int:
                this->MovArgFromReg2Stack(entryInstr, i == 0 ? RegRDX : i == 1 ? RegR8 : RegR9, offset, TyInt32);
                offset++;
                break;
            case Js::AsmJsVarType::Int64:
                this->MovArgFromReg2Stack(entryInstr, i == 0 ? RegRDX : i == 1 ? RegR8 : RegR9, offset, TyInt64);
                offset++;
                break;
            case Js::AsmJsVarType::Float:
                // registers we need are contiguous, so calculate it from XMM1
                this->MovArgFromReg2Stack(entryInstr, (RegNum)(RegXMM1 + i), offset, TyFloat32);
                offset++;
                break;
            case Js::AsmJsVarType::Double:
                this->MovArgFromReg2Stack(entryInstr, (RegNum)(RegXMM1 + i), offset, TyFloat64);
                offset++;
                break;
            case Js::AsmJsVarType::Float32x4:
                this->MovArgFromReg2Stack(entryInstr, (RegNum)(RegXMM1 + i), offset, TySimd128F4);
                offset += 2;
                break;
            case Js::AsmJsVarType::Int32x4:
                this->MovArgFromReg2Stack(entryInstr, (RegNum)(RegXMM1 + i), offset, TySimd128I4);
                offset += 2;
                break;
            case Js::AsmJsVarType::Int16x8:
                this->MovArgFromReg2Stack(entryInstr, (RegNum)(RegXMM1 + i), offset, TySimd128I8);
                offset += 2;
                break;
            case Js::AsmJsVarType::Int8x16:
                this->MovArgFromReg2Stack(entryInstr, (RegNum)(RegXMM1 + i), offset, TySimd128I16);
                offset += 2;
                break;
            case Js::AsmJsVarType::Uint32x4:
                this->MovArgFromReg2Stack(entryInstr, (RegNum)(RegXMM1 + i), offset, TySimd128U4);
                offset += 2;
                break;
            case Js::AsmJsVarType::Uint16x8:
                this->MovArgFromReg2Stack(entryInstr, (RegNum)(RegXMM1 + i), offset, TySimd128U8);
                offset += 2;
                break;
            case Js::AsmJsVarType::Uint8x16:
                this->MovArgFromReg2Stack(entryInstr, (RegNum)(RegXMM1 + i), offset, TySimd128U16);
                offset += 2;
                break;
            case Js::AsmJsVarType::Bool32x4:
                this->MovArgFromReg2Stack(entryInstr, (RegNum)(RegXMM1 + i), offset, TySimd128B4);
                offset += 2;
                break;
            case Js::AsmJsVarType::Bool16x8:
                this->MovArgFromReg2Stack(entryInstr, (RegNum)(RegXMM1 + i), offset, TySimd128B8);
                offset += 2;
                break;
            case Js::AsmJsVarType::Bool8x16:
                this->MovArgFromReg2Stack(entryInstr, (RegNum)(RegXMM1 + i), offset, TySimd128B16);
                offset += 2;
                break;
            case Js::AsmJsVarType::Float64x2:
                this->MovArgFromReg2Stack(entryInstr, (RegNum)(RegXMM1 + i), offset, TySimd128D2);
                offset += 2;
                break;
            case Js::AsmJsVarType::Int64x2:
                this->MovArgFromReg2Stack(entryInstr, (RegNum)(RegXMM1 + i), offset, TySimd128I2);
                offset += 2;
                break;
            default:
                Assume(UNREACHED);
            }
        }
    }
    else if (argSlotsForFunctionsCalled)
    {
        this->MovArgFromReg2Stack(entryInstr, RegRCX, 1);
        this->MovArgFromReg2Stack(entryInstr, RegRDX, 2);
        this->MovArgFromReg2Stack(entryInstr, RegR8, 3);
        this->MovArgFromReg2Stack(entryInstr, RegR9, 4);
    }
#endif  // _WIN32

    IntConstType frameSize = Js::Constants::MinStackJIT + stackArgsSize + stackLocalsSize + savedRegSize;
    this->GeneratePrologueStackProbe(entryInstr, frameSize);

    return entryInstr;
}

void
LowererMDArch::GeneratePrologueStackProbe(IR::Instr *entryInstr, IntConstType frameSize)
{
    //
    // Generate a stack overflow check. Since ProbeCurrentStack throws an exception it needs
    // an unwindable stack. Should we need to call ProbeCurrentStack, instead of creating a new frame here,
    // we make it appear like our caller directly called ProbeCurrentStack.
    //
    // For thread-bound thread context
    //    MOV  rax, ThreadContext::scriptStackLimit + frameSize
    //    CMP  rsp, rax
    //    JG   $done
    //    MOV  rax, ThreadContext::ProbeCurrentStack
    //    MOV  rcx, frameSize
    //    MOV  rdx, scriptContext
    //    JMP  rax
    // $done:
    //

    // For thread-agile thread context
    //    MOV  rax, [ThreadContext::scriptStackLimit]
    //    ADD  rax, frameSize
    //    CMP  rsp, rax
    //    JG   $done
    //    MOV  rax, ThreadContext::ProbeCurrentStack
    //    MOV  rcx, frameSize
    //    MOV  rdx, scriptContext
    //    JMP  rax
    // $done:
    //

    // For thread context with script interrupt enabled
    //    MOV  rax, [ThreadContext::scriptStackLimit]
    //    ADD  rax, frameSize
    //    JO   $helper
    //    CMP  rsp, rax
    //    JG   $done
    // $helper:
    //    MOV  rax, ThreadContext::ProbeCurrentStack
    //    MOV  rcx, frameSize
    //    MOV  rdx, scriptContext
    //    JMP  rax
    // $done:
    //

    // Do not insert stack probe for leaf functions which have low stack footprint
    if (this->m_func->IsTrueLeaf() &&
        frameSize - Js::Constants::MinStackJIT < Js::Constants::MaxStackSizeForNoProbe)
    {
        return;
    }

    IR::LabelInstr *helperLabel = IR::LabelInstr::New(Js::OpCode::Label, this->m_func, true);
    IR::Instr *insertInstr = entryInstr->m_next;
    IR::Instr *instr;
    IR::Opnd *stackLimitOpnd;
    bool doInterruptProbe = m_func->GetJITFunctionBody()->DoInterruptProbe();

    // MOV rax, ThreadContext::scriptStackLimit + frameSize
    stackLimitOpnd = IR::RegOpnd::New(nullptr, RegRAX, TyMachReg, this->m_func);
    if (doInterruptProbe || !m_func->GetThreadContextInfo()->IsThreadBound())
    {
        // Load the current stack limit from the ThreadContext and add the current frame size.
        {
            intptr_t pLimit = m_func->GetThreadContextInfo()->GetThreadStackLimitAddr();
            IR::RegOpnd *baseOpnd = IR::RegOpnd::New(nullptr, RegRAX, TyMachReg, this->m_func);
            Lowerer::InsertMove(baseOpnd, IR::AddrOpnd::New(pLimit, IR::AddrOpndKindDynamicMisc, this->m_func), insertInstr);
            IR::IndirOpnd *indirOpnd = IR::IndirOpnd::New(baseOpnd, 0, TyMachReg, this->m_func);

            Lowerer::InsertMove(stackLimitOpnd, indirOpnd, insertInstr);
        }

        instr = IR::Instr::New(Js::OpCode::ADD, stackLimitOpnd, stackLimitOpnd,
                               IR::IntConstOpnd::New(frameSize, TyMachReg, this->m_func), this->m_func);
        insertInstr->InsertBefore(instr);

        if (doInterruptProbe)
        {
            // If the add overflows, call the probe helper.
            instr = IR::BranchInstr::New(Js::OpCode::JO, helperLabel, this->m_func);
            insertInstr->InsertBefore(instr);
        }
    }
    else
    {
        // TODO: michhol, check this math
        size_t scriptStackLimit = m_func->GetThreadContextInfo()->GetScriptStackLimit();
        Lowerer::InsertMove(stackLimitOpnd, IR::IntConstOpnd::New((frameSize + scriptStackLimit), TyMachReg, this->m_func), insertInstr);
    }

    // CMP rsp, rax
    instr = IR::Instr::New(Js::OpCode::CMP, this->m_func);
    instr->SetSrc1(IR::RegOpnd::New(nullptr, GetRegStackPointer(), TyMachReg, m_func));
    instr->SetSrc2(stackLimitOpnd);
    insertInstr->InsertBefore(instr);


    IR::LabelInstr * doneLabel = nullptr;
    if (!PHASE_OFF(Js::LayoutPhase, this->m_func))
    {
        // JLE $helper
        instr = IR::BranchInstr::New(Js::OpCode::JLE, helperLabel, m_func);
        insertInstr->InsertBefore(instr);
        Security::InsertRandomFunctionPad(insertInstr);

        // This is generated after layout. Generate the block at the end of the function manually
        insertInstr = IR::PragmaInstr::New(Js::OpCode::StatementBoundary, Js::Constants::NoStatementIndex, m_func);

        this->m_func->m_tailInstr->InsertAfter(insertInstr);
        this->m_func->m_tailInstr = insertInstr;
    }
    else
    {
        doneLabel = IR::LabelInstr::New(Js::OpCode::Label, this->m_func);
        // JGT $done
        instr = IR::BranchInstr::New(Js::OpCode::JGT, doneLabel, m_func);
        insertInstr->InsertBefore(instr);
    }

    insertInstr->InsertBefore(helperLabel);

    IR::RegOpnd *target;
    {
        // MOV RegArg1, scriptContext
        Lowerer::InsertMove(
            IR::RegOpnd::New(nullptr, RegArg1, TyMachReg, m_func),
            this->lowererMD->m_lowerer->LoadScriptContextOpnd(insertInstr), insertInstr);

        // MOV RegArg0, frameSize
        Lowerer::InsertMove(
            IR::RegOpnd::New(nullptr, RegArg0, TyMachReg, this->m_func),
            IR::IntConstOpnd::New(frameSize, TyMachReg, this->m_func), insertInstr);

        // MOV rax, ThreadContext::ProbeCurrentStack
        target = IR::RegOpnd::New(nullptr, RegRAX, TyMachReg, m_func);
        Lowerer::InsertMove(target, IR::HelperCallOpnd::New(IR::HelperProbeCurrentStack, m_func), insertInstr);
    }

    // JMP rax
    instr = IR::MultiBranchInstr::New(Js::OpCode::JMP, target, m_func);
    insertInstr->InsertBefore(instr);

    if (doneLabel)
    {
        // $done:
        insertInstr->InsertBefore(doneLabel);
        Security::InsertRandomFunctionPad(doneLabel);
    }
}


///----------------------------------------------------------------------------
///
/// LowererMDArch::LowerExitInstr
///
///     Emit epilog.
///
///----------------------------------------------------------------------------

IR::Instr *
LowererMDArch::LowerExitInstr(IR::ExitInstr * exitInstr)
{
    uint32 savedRegSize = 0;

    // POP used callee-saved registers

    IR::Instr * exitPrevInstr = exitInstr->m_prev;
    AssertMsg(exitPrevInstr, "Can a function have only 1 instr ? Or is the instr chain broken");

    IR::RegOpnd *stackPointer = IR::RegOpnd::New(nullptr, GetRegStackPointer(), TyMachReg, this->m_func);

    unsigned xmmOffset = 0;

    for (RegNum reg = (RegNum)(RegNOREG + 1); reg < RegNumCount; reg = (RegNum)(reg+1))
    {
        if (LinearScan::IsCalleeSaved(reg) && (this->m_func->HasTry() || this->m_func->m_regsUsed.Test(reg)))
        {
            IRType       type    = RegTypes[reg];
            IR::RegOpnd *regOpnd = IR::RegOpnd::New(nullptr, reg, type, this->m_func);

            if (type == TyFloat64)
            {
                IR::Instr *restoreInstr = IR::Instr::New(Js::OpCode::MOVAPS,
                                                         regOpnd,
                                                         IR::IndirOpnd::New(stackPointer,
                                                                            xmmOffset,
                                                                            type,
                                                                            this->m_func),
                                                         this->m_func);
                xmmOffset += (MachDouble * 2);
                exitInstr->InsertBefore(restoreInstr);
            }
            else
            {
                Assert(type == TyInt64);

                IR::Instr *popInstr = IR::Instr::New(Js::OpCode::POP, regOpnd, this->m_func);
                exitInstr->InsertBefore(popInstr);

                savedRegSize += MachPtr;
            }
        }
    }

    Assert(savedRegSize == (uint)this->m_func->GetSavedRegSize());


    // Generate ADD RSP, argsStackSize before the register restore (if there are any)
    uint32 stackArgsSize = this->m_func->GetArgsSize();
    Assert(stackArgsSize);
    if (savedRegSize || xmmOffset)
    {
        IR::IntConstOpnd *stackSizeOpnd = IR::IntConstOpnd::New(stackArgsSize, TyMachReg, this->m_func);
        IR::Instr *addInstr = IR::Instr::New(Js::OpCode::ADD, stackPointer, stackPointer, stackSizeOpnd, this->m_func);
        exitPrevInstr->InsertAfter(addInstr);
    }

    //
    // useful register operands
    //
    IR::RegOpnd * rspOpnd = IR::RegOpnd::New(nullptr, RegRSP, TyMachReg, this->m_func);
    IR::RegOpnd * rbpOpnd = IR::RegOpnd::New(nullptr, RegRBP, TyMachReg, this->m_func);

    // Restore frame

    // Generate MOV RSP, RBP

    IR::Instr * movInstr = IR::Instr::New(Js::OpCode::MOV, rspOpnd, rbpOpnd, this->m_func);
    exitInstr->InsertBefore(movInstr);

    // Generate POP RBP

    IR::Instr * pushInstr = IR::Instr::New(Js::OpCode::POP, rbpOpnd, this->m_func);
    exitInstr->InsertBefore(pushInstr);

    // Insert RET
    IR::IntConstOpnd * intSrc = IR::IntConstOpnd::New(0, TyInt32, this->m_func);
    IR::RegOpnd *retReg = nullptr;
    if (m_func->GetJITFunctionBody()->IsAsmJsMode() && !m_func->IsLoopBody())
    {
        switch (m_func->GetJITFunctionBody()->GetAsmJsInfo()->GetRetType())
        {
        case Js::AsmJsRetType::Double:
        case Js::AsmJsRetType::Float:
            retReg = IR::RegOpnd::New(nullptr, this->GetRegReturnAsmJs(TyMachDouble), TyMachDouble, this->m_func);
            break;
#ifdef ENABLE_WASM_SIMD
        case Js::AsmJsRetType::Int32x4:
            retReg = IR::RegOpnd::New(nullptr, this->GetRegReturnAsmJs(TySimd128I4), TySimd128I4, this->m_func);
            break;
        case Js::AsmJsRetType::Int16x8:
            retReg = IR::RegOpnd::New(nullptr, this->GetRegReturnAsmJs(TySimd128I8), TySimd128I8, this->m_func);
            break;
        case Js::AsmJsRetType::Int8x16:
            retReg = IR::RegOpnd::New(nullptr, this->GetRegReturnAsmJs(TySimd128I16), TySimd128U16, this->m_func);
            break;
        case Js::AsmJsRetType::Uint32x4:
            retReg = IR::RegOpnd::New(nullptr, this->GetRegReturnAsmJs(TySimd128U4), TySimd128U4, this->m_func);
            break;
        case Js::AsmJsRetType::Uint16x8:
            retReg = IR::RegOpnd::New(nullptr, this->GetRegReturnAsmJs(TySimd128U8), TySimd128U8, this->m_func);
            break;
        case Js::AsmJsRetType::Uint8x16:
            retReg = IR::RegOpnd::New(nullptr, this->GetRegReturnAsmJs(TySimd128U16), TySimd128U16, this->m_func);
            break;
        case Js::AsmJsRetType::Bool32x4:
            retReg = IR::RegOpnd::New(nullptr, this->GetRegReturnAsmJs(TySimd128B4), TySimd128B4, this->m_func);
            break;
        case Js::AsmJsRetType::Bool16x8:
            retReg = IR::RegOpnd::New(nullptr, this->GetRegReturnAsmJs(TySimd128B8), TySimd128B8, this->m_func);
            break;
        case Js::AsmJsRetType::Bool8x16:
            retReg = IR::RegOpnd::New(nullptr, this->GetRegReturnAsmJs(TySimd128B16), TySimd128B16, this->m_func);
            break;
        case Js::AsmJsRetType::Float32x4:
            retReg = IR::RegOpnd::New(nullptr, this->GetRegReturnAsmJs(TySimd128F4), TySimd128F4, this->m_func);
            break;
        case Js::AsmJsRetType::Float64x2:
            retReg = IR::RegOpnd::New(nullptr, this->GetRegReturnAsmJs(TySimd128D2), TySimd128D2, this->m_func);
            break;
        case Js::AsmJsRetType::Int64x2:
            retReg = IR::RegOpnd::New(nullptr, this->GetRegReturnAsmJs(TySimd128I2), TySimd128I2, this->m_func);
            break;
#endif
        case Js::AsmJsRetType::Int64:
        case Js::AsmJsRetType::Signed:
            retReg = IR::RegOpnd::New(nullptr, this->GetRegReturn(TyMachReg), TyMachReg, this->m_func);
            break;
        case Js::AsmJsRetType::Void:
            break;
        default:
            Assume(UNREACHED);
        }
    }
    else
    {
        retReg = IR::RegOpnd::New(nullptr, this->GetRegReturn(TyMachReg), TyMachReg, this->m_func);
    }


    // Generate RET
    IR::Instr * retInstr = IR::Instr::New(Js::OpCode::RET, this->m_func);
    retInstr->SetSrc1(intSrc);
    if (retReg)
    {
        retInstr->SetSrc2(retReg);
    }
    exitInstr->InsertBefore(retInstr);

    retInstr->m_opcode = Js::OpCode::RET;


    return exitInstr;
}

IR::Instr *
LowererMDArch::LowerExitInstrAsmJs(IR::ExitInstr * exitInstr)
{
    // epilogue is almost identical on x64, except for return register
    return LowerExitInstr(exitInstr);
}

void
LowererMDArch::EmitInt4Instr(IR::Instr *instr, bool signExtend /* = false */)
{
    IR::Opnd *dst       = instr->GetDst();
    IR::Opnd *src1      = instr->GetSrc1();
    IR::Opnd *src2      = instr->GetSrc2();
    IR::Instr *newInstr = nullptr;
    IR::RegOpnd *regEDX;

    bool isInt64Instr = instr->AreAllOpndInt64();
    if (!isInt64Instr)
    {
        if (dst && !dst->IsUInt32())
        {
            dst->SetType(TyInt32);
        }
        if (!src1->IsUInt32())
        {
            src1->SetType(TyInt32);
        }
        if (src2 && !src2->IsUInt32())
        {
            src2->SetType(TyInt32);
        }
    }

    switch (instr->m_opcode)
    {
    case Js::OpCode::Neg_I4:
        instr->m_opcode = Js::OpCode::NEG;
        break;

    case Js::OpCode::Not_I4:
        instr->m_opcode = Js::OpCode::NOT;
        break;

    case Js::OpCode::Add_I4:
        LowererMD::ChangeToAdd(instr, false /* needFlags */);
        break;

    case Js::OpCode::Sub_I4:
        LowererMD::ChangeToSub(instr, false /* needFlags */);
        break;

    case Js::OpCode::Mul_I4:
        instr->m_opcode = Js::OpCode::IMUL2;
        break;

    case Js::OpCode::DivU_I4:
    case Js::OpCode::Div_I4:
        instr->SinkDst(Js::OpCode::MOV, RegRAX);
        goto idiv_common;
    case Js::OpCode::RemU_I4:
    case Js::OpCode::Rem_I4:
        instr->SinkDst(Js::OpCode::MOV, RegRDX);
idiv_common:
        {
            bool isUnsigned = instr->GetSrc1()->IsUnsigned();
            if (isUnsigned)
            {
                Assert(instr->GetSrc2()->IsUnsigned());
                Assert(instr->m_opcode == Js::OpCode::RemU_I4 || instr->m_opcode == Js::OpCode::DivU_I4);
                instr->m_opcode = Js::OpCode::DIV;
            }
            else
            {
                instr->m_opcode = Js::OpCode::IDIV;
            }
            instr->HoistSrc1(Js::OpCode::MOV, RegRAX);

            regEDX = IR::RegOpnd::New(src1->GetType(), instr->m_func);
            regEDX->SetReg(RegRDX);
            if (isUnsigned)
            {
                // we need to ensure that register allocator doesn't muck about with rdx
                instr->HoistSrc2(Js::OpCode::MOV, RegRCX);

                Lowerer::InsertMove(regEDX, IR::IntConstOpnd::New(0, src1->GetType(), instr->m_func), instr);
                // NOP ensures that the EDX = Ld_I4 0 doesn't get deadstored, will be removed in peeps
                instr->InsertBefore(IR::Instr::New(Js::OpCode::NOP, regEDX, regEDX, instr->m_func));
            }
            else
            {
                if (instr->GetSrc2()->IsImmediateOpnd())
                {
                    instr->HoistSrc2(Js::OpCode::MOV);
                }
                instr->InsertBefore(IR::Instr::New(isInt64Instr ? Js::OpCode::CQO : Js::OpCode::CDQ, regEDX, instr->m_func));
            }
            return;
        }
    case Js::OpCode::Or_I4:
        instr->m_opcode = Js::OpCode::OR;
        break;

    case Js::OpCode::Xor_I4:
        instr->m_opcode = Js::OpCode::XOR;
        break;

    case Js::OpCode::And_I4:
        instr->m_opcode = Js::OpCode::AND;
        break;

    case Js::OpCode::Shl_I4:
    case Js::OpCode::ShrU_I4:
    case Js::OpCode::Shr_I4:
    case Js::OpCode::Rol_I4:
    case Js::OpCode::Ror_I4:
        LowererMD::ChangeToShift(instr, false /* needFlags */);
        break;

    case Js::OpCode::BrTrue_I4:
        instr->m_opcode = Js::OpCode::JNE;
        goto br1_Common;

    case Js::OpCode::BrFalse_I4:
        instr->m_opcode = Js::OpCode::JEQ;
br1_Common:
        src1 = instr->UnlinkSrc1();
        newInstr = IR::Instr::New(Js::OpCode::TEST, instr->m_func);
        instr->InsertBefore(newInstr);
        newInstr->SetSrc1(src1);
        newInstr->SetSrc2(src1);
        return;

    case Js::OpCode::BrEq_I4:
        instr->m_opcode = Js::OpCode::JEQ;
        goto br2_Common;

    case Js::OpCode::BrNeq_I4:
        instr->m_opcode = Js::OpCode::JNE;
        goto br2_Common;

    case Js::OpCode::BrUnGt_I4:
        instr->m_opcode = Js::OpCode::JA;
        goto br2_Common;

    case Js::OpCode::BrUnGe_I4:
        instr->m_opcode = Js::OpCode::JAE;
        goto br2_Common;

    case Js::OpCode::BrUnLe_I4:
        instr->m_opcode = Js::OpCode::JBE;
        goto br2_Common;

    case Js::OpCode::BrUnLt_I4:
        instr->m_opcode = Js::OpCode::JB;
        goto br2_Common;

    case Js::OpCode::BrGt_I4:
        instr->m_opcode = Js::OpCode::JGT;
        goto br2_Common;

    case Js::OpCode::BrGe_I4:
        instr->m_opcode = Js::OpCode::JGE;
        goto br2_Common;

    case Js::OpCode::BrLe_I4:
        instr->m_opcode = Js::OpCode::JLE;
        goto br2_Common;

    case Js::OpCode::BrLt_I4:
        instr->m_opcode = Js::OpCode::JLT;
br2_Common:
        src1 = instr->UnlinkSrc1();
        src2 = instr->UnlinkSrc2();
        newInstr = IR::Instr::New(Js::OpCode::CMP, instr->m_func);
        instr->InsertBefore(newInstr);
        newInstr->SetSrc1(src1);
        newInstr->SetSrc2(src2);
        return;

    default:
        AssertMsg(UNREACHED, "Un-implemented int4 opcode");
    }

    if (signExtend)
    {
        Assert(instr->GetDst());
        IR::Opnd *dst64 = instr->GetDst()->Copy(instr->m_func);
        dst64->SetType(TyMachReg);
        instr->InsertAfter(IR::Instr::New(Js::OpCode::MOVSXD, dst64, instr->GetDst(), instr->m_func));
    }

    LowererMD::Legalize(instr);
}

#if !FLOATVAR
void
LowererMDArch::EmitLoadVar(IR::Instr *instrLoad, bool isFromUint32, bool isHelper)
{
    //    e1  = MOV e_src1
    //    e1  = SHL e1, Js::VarTag_Shift
    //          JO $ToVar
    //          JB  $ToVar     [isFromUint32]
    //    e1  = INC e1
    //  r_dst = MOVSXD e1
    //          JMP $done
    //  $ToVar:
    //          EmitLoadVarNoCheck
    //  $Done:

    Assert(instrLoad->GetSrc1()->IsRegOpnd());
    Assert(instrLoad->GetDst()->GetType() == TyVar);

    bool isInt            = false;
    bool isNotInt         = false;
    IR::Opnd *dst         = instrLoad->GetDst();
    IR::RegOpnd *src1     = instrLoad->GetSrc1()->AsRegOpnd();
    IR::LabelInstr *toVar = nullptr;
    IR::LabelInstr *done  = nullptr;

    // TODO: Fix bad lowering. We shouldn't get TyVars here.
    // Assert(instrLoad->GetSrc1()->GetType() == TyInt32);
    src1->SetType(TyInt32);

    if (src1->IsTaggedInt())
    {
        isInt = true;
    }
    else if (src1->IsNotInt())
    {
        isNotInt = true;
    }

    if (!isNotInt)
    {
        // e1 = MOV e_src1
        IR::RegOpnd *e1 = IR::RegOpnd::New(TyInt32, m_func);
        instrLoad->InsertBefore(IR::Instr::New(Js::OpCode::MOV, e1, instrLoad->GetSrc1(), m_func));

        // e1 = SHL e1, Js::VarTag_Shift
        instrLoad->InsertBefore(IR::Instr::New(Js::OpCode::SHL,
            e1,
            e1,
            IR::IntConstOpnd::New(Js::VarTag_Shift, TyInt8, m_func), m_func));


        if (!isInt)
        {
            // JO $ToVar
            toVar = IR::LabelInstr::New(Js::OpCode::Label, m_func, true);
            instrLoad->InsertBefore(IR::BranchInstr::New(Js::OpCode::JO, toVar, m_func));

            if (isFromUint32)
            {
                //       JB  $ToVar     [isFromUint32]
                instrLoad->InsertBefore(IR::BranchInstr::New(Js::OpCode::JB, toVar, this->m_func));
            }
        }

        // e1 = INC e1
        instrLoad->InsertBefore(IR::Instr::New(Js::OpCode::INC, e1, e1, m_func));

        // dst = MOVSXD e1
        instrLoad->InsertBefore(IR::Instr::New(Js::OpCode::MOVSXD, dst, e1, m_func));

        if (!isInt)
        {
            // JMP $done
            done = IR::LabelInstr::New(Js::OpCode::Label, m_func, isHelper);
            instrLoad->InsertBefore(IR::BranchInstr::New(Js::OpCode::JMP, done, m_func));
        }
    }

    IR::Instr *insertInstr = instrLoad;
    if (!isInt)
    {
        // $toVar:
        if (toVar)
        {
            instrLoad->InsertBefore(toVar);
        }

        // ToVar()
        this->lowererMD->EmitLoadVarNoCheck(dst->AsRegOpnd(), src1, instrLoad, isFromUint32, isHelper || toVar != nullptr);
    }

    if (done)
    {
        instrLoad->InsertAfter(done);
    }

    instrLoad->Remove();
}
#else

void
LowererMDArch::EmitLoadVar(IR::Instr *instrLoad, bool isFromUint32, bool isHelper)
{
    //    MOV_TRUNC e1, e_src1
    //    CMP e1, 0             [uint32]
    //    JLT $Helper           [uint32]  -- overflows?
    //    BTS r1, VarTag_Shift
    //    MOV r_dst, r1
    //    JMP $done             [uint32]
    // $helper                  [uint32]
    //    EmitLoadVarNoCheck
    // $done                    [uint32]


    Assert(instrLoad->GetSrc1()->IsRegOpnd());
    Assert(instrLoad->GetDst()->GetType() == TyVar);

    bool isInt            = false;
    IR::Opnd *dst         = instrLoad->GetDst();
    IR::RegOpnd *src1     = instrLoad->GetSrc1()->AsRegOpnd();
    IR::LabelInstr *labelHelper = nullptr;

    // TODO: Fix bad lowering. We shouldn't get TyVars here.
    // Assert(instrLoad->GetSrc1()->GetType() == TyInt32);
    src1->SetType(TyInt32);

    if (src1->IsTaggedInt())
    {
        isInt = true;
    }
    else if (src1->IsNotInt())
    {
        // ToVar()
        this->lowererMD->EmitLoadVarNoCheck(dst->AsRegOpnd(), src1, instrLoad, isFromUint32, isHelper);
        return;
    }

    IR::RegOpnd *r1 = IR::RegOpnd::New(TyVar, m_func);

    // e1 = MOV_TRUNC e_src1
    // (Use MOV_TRUNC here as we rely on the register copy to clear the upper 32 bits.)
    IR::RegOpnd *e1 = r1->Copy(m_func)->AsRegOpnd();
    e1->SetType(TyInt32);
    instrLoad->InsertBefore(IR::Instr::New(Js::OpCode::MOV_TRUNC,
        e1,
        src1,
        m_func));

    if (!isInt && isFromUint32)
    {
        // CMP e1, 0
        IR::Instr *instr = IR::Instr::New(Js::OpCode::CMP, m_func);
        instr->SetSrc1(e1);
        instr->SetSrc2(IR::IntConstOpnd::New(0, TyInt32, m_func));
        instrLoad->InsertBefore(instr);

        Assert(!labelHelper);
        labelHelper = IR::LabelInstr::New(Js::OpCode::Label, m_func, true);

        // JLT $helper
        instr = IR::BranchInstr::New(Js::OpCode::JLT, labelHelper, m_func);
        instrLoad->InsertBefore(instr);
    }

    // The previous operation clears the top 32 bits.
    // BTS r1, VarTag_Shift
    this->lowererMD->GenerateInt32ToVarConversion(r1, instrLoad);

    // REVIEW: We need r1 only if we could generate sn = Ld_A_I4 sn. i.e. the destination and
    // source are the same.
    // r_dst = MOV r1
    instrLoad->InsertBefore(IR::Instr::New(Js::OpCode::MOV,
        dst,
        r1,
        m_func));

    if (labelHelper)
    {
        Assert(isFromUint32);

        // JMP $done
        IR::LabelInstr * labelDone = IR::LabelInstr::New(Js::OpCode::Label, m_func, isHelper);
        instrLoad->InsertBefore(IR::BranchInstr::New(Js::OpCode::JMP, labelDone, m_func));

        // $helper
        instrLoad->InsertBefore(labelHelper);

        // ToVar()
        this->lowererMD->EmitLoadVarNoCheck(dst->AsRegOpnd(), src1, instrLoad, isFromUint32, true);

        // $done
        instrLoad->InsertBefore(labelDone);
    }
    instrLoad->Remove();
}
#endif

void
LowererMDArch::EmitIntToFloat(IR::Opnd *dst, IR::Opnd *src, IR::Instr *instrInsert)
{
    Assert(dst->IsRegOpnd() && dst->IsFloat());
    Assert(src->IsRegOpnd() && src->IsInt32());
    if (dst->IsFloat64())
    {
        // Use MOVD to make sure we sign extended the 32-bit src
        instrInsert->InsertBefore(IR::Instr::New(Js::OpCode::MOVD, dst, src, this->m_func));

        // Convert to float
        instrInsert->InsertBefore(IR::Instr::New(Js::OpCode::CVTDQ2PD, dst, dst, this->m_func));
    }
    else
    {
        Assert(dst->IsFloat32());
        instrInsert->InsertBefore(IR::Instr::New(Js::OpCode::CVTSI2SS, dst, src, this->m_func));
    }
}

void
LowererMDArch::EmitIntToLong(IR::Opnd *dst, IR::Opnd *src, IR::Instr *instrInsert)
{
    Assert(dst->IsRegOpnd() && dst->IsInt64());
    Assert(src->IsInt32());

    Lowerer::InsertMove(dst, src, instrInsert);
}

void
LowererMDArch::EmitUIntToLong(IR::Opnd *dst, IR::Opnd *src, IR::Instr *instrInsert)
{
    Assert(dst->IsRegOpnd() && dst->IsInt64());
    Assert(src->IsUInt32());

    Lowerer::InsertMove(dst, src, instrInsert);
}

void
LowererMDArch::EmitLongToInt(IR::Opnd *dst, IR::Opnd *src, IR::Instr *instrInsert)
{
    Assert(dst->IsRegOpnd() && dst->IsInt32());
    Assert(src->IsInt64());

    instrInsert->InsertBefore(IR::Instr::New(Js::OpCode::MOV_TRUNC, dst, src, instrInsert->m_func));
}

void
LowererMDArch::EmitUIntToFloat(IR::Opnd *dst, IR::Opnd *src, IR::Instr *instrInsert)
{
    Assert(dst->IsRegOpnd() && dst->IsFloat());
    Assert(src->IsRegOpnd() && (src->IsInt32() || src->IsUInt32()));

    // MOV tempReg.i32, src  - make sure the top bits are 0
    IR::RegOpnd * tempReg = IR::RegOpnd::New(TyInt32, this->m_func);
    instrInsert->InsertBefore(IR::Instr::New(Js::OpCode::MOV_TRUNC, tempReg, src, this->m_func));

    // CVTSI2SD dst, tempReg.i64  (Use the tempreg as if it is 64 bit without sign extension)
    instrInsert->InsertBefore(IR::Instr::New(dst->IsFloat64() ? Js::OpCode::CVTSI2SD : Js::OpCode::CVTSI2SS, dst,
        tempReg->UseWithNewType(TyInt64, this->m_func), this->m_func));
}

bool
LowererMDArch::EmitLoadInt32(IR::Instr *instrLoad, bool conversionFromObjectAllowed, bool bailOutOnHelper, IR::LabelInstr * labelBailOut)
{
    //
    //    r1 = MOV src1
    // rtest = MOV src1
    //         SHR rtest, AtomTag_Shift
    //         CMP rtest, 1
    //         JNE $helper or $float
    // r_dst = MOV_TRUNC e_src1
    //         JMP $done
    //  $float:
    //     dst = ConvertToFloat(r1, $helper)
    // $helper:
    // r_dst = ToInt32()
    //

    Assert(instrLoad->GetSrc1()->IsRegOpnd());
    Assert(instrLoad->GetSrc1()->GetType() == TyVar);

    // TODO: Fix bad lowering. We shouldn't see TyVars here.
    // Assert(instrLoad->GetDst()->GetType() == TyInt32);

    bool isInt             = false;
    bool isNotInt          = false;
    IR::Opnd *dst          = instrLoad->GetDst();
    IR::RegOpnd *src1      = instrLoad->GetSrc1()->AsRegOpnd();
    IR::LabelInstr *helper = nullptr;
    IR::LabelInstr *labelFloat = nullptr;
    IR::LabelInstr *done   = nullptr;

    if (src1->IsTaggedInt())
    {
        isInt = true;
    }
    else if (src1->IsNotInt())
    {
        isNotInt = true;
    }

    if (src1->IsEqual(instrLoad->GetDst()) == false)
    {
        // r1 = MOV src1
        IR::RegOpnd *r1 = IR::RegOpnd::New(TyVar, instrLoad->m_func);
        r1->SetValueType(src1->GetValueType());
        instrLoad->InsertBefore(IR::Instr::New(Js::OpCode::MOV, r1, src1, instrLoad->m_func));
        src1 = r1;
    }

    const ValueType src1ValueType(src1->GetValueType());
    const bool doFloatToIntFastPath =
        (src1ValueType.IsLikelyFloat() || src1ValueType.IsLikelyUntaggedInt()) &&
                !(instrLoad->HasBailOutInfo() && (instrLoad->GetBailOutKind() == IR::BailOutIntOnly || instrLoad->GetBailOutKind() == IR::BailOutExpectingInteger));

    if (isNotInt)
    {
        // Known to be non-integer. If we are required to bail out on helper call, just re-jit.
        if (!doFloatToIntFastPath && bailOutOnHelper)
        {
            if(!GlobOpt::DoEliminateArrayAccessHelperCall(this->m_func))
            {
                // Array access helper call removal is already off for some reason. Prevent trying to rejit again
                // because it won't help and the same thing will happen again. Just abort jitting this function.
                if(PHASE_TRACE(Js::BailOutPhase, this->m_func))
                {
                    Output::Print(_u("    Aborting JIT because EliminateArrayAccessHelperCall is already off\n"));
                    Output::Flush();
                }
                throw Js::OperationAbortedException();
            }

            throw Js::RejitException(RejitReason::ArrayAccessHelperCallEliminationDisabled);
        }
    }
    else
    {
        // It could be an integer in this case.
        if (!isInt)
        {
            if(doFloatToIntFastPath)
            {
                labelFloat = IR::LabelInstr::New(Js::OpCode::Label, instrLoad->m_func, false);
            }
            else
            {
                helper = IR::LabelInstr::New(Js::OpCode::Label, instrLoad->m_func, true);
            }

            this->lowererMD->GenerateSmIntTest(src1, instrLoad, labelFloat ? labelFloat : helper);
        }
        IR::RegOpnd *src132 = src1->UseWithNewType(TyInt32, instrLoad->m_func)->AsRegOpnd();
#if !INT32VAR
        // src1 = SAR src1, VarTag_Shift
        instrLoad->InsertBefore(IR::Instr::New(Js::OpCode::SAR,
                                               src132,
                                               src132,
                                               IR::IntConstOpnd::New(Js::VarTag_Shift, TyInt8, instrLoad->m_func),
                                               instrLoad->m_func));

        // r_dst = MOV src1
        // This is only a MOV (and not a MOVSXD) because we do a signed shift right, but we'll copy
        // all 64 bits.


        instrLoad->InsertBefore(IR::Instr::New(Js::OpCode::MOV,
                                               dst->UseWithNewType(TyMachReg, instrLoad->m_func),
                                               src1,
                                               instrLoad->m_func));

#else

        instrLoad->InsertBefore(IR::Instr::New(Js::OpCode::MOV_TRUNC,
                                               dst->UseWithNewType(TyInt32, instrLoad->m_func),
                                               src132,
                                               instrLoad->m_func));
#endif
        if (!isInt)
        {
            // JMP $done
            done = instrLoad->GetOrCreateContinueLabel();
            instrLoad->InsertBefore(IR::BranchInstr::New(Js::OpCode::JMP, done, m_func));
        }
    }

    if (!isInt)
    {
        if(doFloatToIntFastPath)
        {
            if(labelFloat)
            {
                instrLoad->InsertBefore(labelFloat);
            }
            if(!helper)
            {
                helper = IR::LabelInstr::New(Js::OpCode::Label, instrLoad->m_func, true);
            }

            if(!done)
            {
                done = instrLoad->GetOrCreateContinueLabel();
            }
#if FLOATVAR
            IR::RegOpnd* floatOpnd = this->lowererMD->CheckFloatAndUntag(src1, instrLoad, helper);
#else
            this->lowererMD->GenerateFloatTest(src1, instrLoad, helper, instrLoad->HasBailOutInfo());
            IR::IndirOpnd* floatOpnd = IR::IndirOpnd::New(src1, Js::JavascriptNumber::GetValueOffset(), TyMachDouble, this->m_func);
#endif
            this->lowererMD->ConvertFloatToInt32(instrLoad->GetDst(), floatOpnd, helper, done, instrLoad);
        }

        // $helper:
        if (helper)
        {
            instrLoad->InsertBefore(helper);
        }
        if(instrLoad->HasBailOutInfo() && (instrLoad->GetBailOutKind() == IR::BailOutIntOnly || instrLoad->GetBailOutKind() == IR::BailOutExpectingInteger))
        {
            // Avoid bailout if we have a JavascriptNumber whose value is a signed 32-bit integer
            lowererMD->m_lowerer->LoadInt32FromUntaggedVar(instrLoad);

            // Need to bail out instead of calling a helper
            return true;
        }

        if (bailOutOnHelper)
        {
            Assert(labelBailOut);
            lowererMD->m_lowerer->InsertBranch(Js::OpCode::Br, labelBailOut, instrLoad);
            instrLoad->Remove();
        }
        else if (conversionFromObjectAllowed)
        {
            lowererMD->m_lowerer->LowerUnaryHelperMem(instrLoad, IR::HelperConv_ToInt32);
        }
        else
        {
            lowererMD->m_lowerer->LowerUnaryHelperMemWithBoolReference(instrLoad, IR::HelperConv_ToInt32_NoObjects, true /*useBoolForBailout*/);
        }
    }
    else
    {
        instrLoad->Remove();
    }

    return false;
}

IR::Instr *
LowererMDArch::LoadCheckedFloat(IR::RegOpnd *opndOrig, IR::RegOpnd *opndFloat, IR::LabelInstr *labelInline, IR::LabelInstr *labelHelper, IR::Instr *instrInsert, const bool checkForNullInLoopBody)
{
    //
    //   if (TaggedInt::Is(opndOrig))
    //       opndFloat = CVTSI2SD opndOrig_32
    //                   JMP $labelInline
    //   else
    //                   JMP $labelOpndIsNotInt
    //
    // $labelOpndIsNotInt:
    //   if (TaggedFloat::Is(opndOrig))
    //       s2        = MOV opndOrig
    //       s2        = XOR FloatTag_Value
    //       opndFloat = MOVD s2
    //   else
    //                   JMP $labelHelper
    //
    // $labelInline:
    //

    IR::Instr *instrFirst = nullptr;

    IR::LabelInstr *labelOpndIsNotInt = IR::LabelInstr::New(Js::OpCode::Label, this->m_func);
    lowererMD->GenerateSmIntTest(opndOrig, instrInsert, labelOpndIsNotInt, &instrFirst);

    if (opndOrig->GetValueType().IsLikelyFloat())
    {
        // Make this path helper if value is likely a float
        instrInsert->InsertBefore(IR::LabelInstr::New(Js::OpCode::Label, this->m_func, true));
    }

    IR::Opnd    *opndOrig_32 = opndOrig->UseWithNewType(TyInt32, this->m_func);

    IR::Instr   *cvtsi2sd    = IR::Instr::New(Js::OpCode::CVTSI2SD, opndFloat, opndOrig_32, this->m_func);
    instrInsert->InsertBefore(cvtsi2sd);

    IR::Instr   *jmpInline   = IR::BranchInstr::New(Js::OpCode::JMP, labelInline, this->m_func);
    instrInsert->InsertBefore(jmpInline);

    instrInsert->InsertBefore(labelOpndIsNotInt);

    lowererMD->GenerateFloatTest(opndOrig, instrInsert, labelHelper, checkForNullInLoopBody);

    IR::RegOpnd *s2          = IR::RegOpnd::New(TyMachReg, this->m_func);
    IR::Instr   *mov         = IR::Instr::New(Js::OpCode::MOV, s2, opndOrig, this->m_func);
    instrInsert->InsertBefore(mov);

    IR::Instr   *xorTag      = IR::Instr::New(Js::OpCode::XOR,
                                              s2,
                                              s2,
                                              IR::IntConstOpnd::New(Js::FloatTag_Value,
                                                                TyMachReg,
                                                                this->m_func,
                                                                /* dontEncode = */ true),
                                              this->m_func);
    instrInsert->InsertBefore(xorTag);
    LowererMD::Legalize(xorTag);

    IR::Instr   *movFloat    = IR::Instr::New(Js::OpCode::MOVD, opndFloat, s2, this->m_func);
    instrInsert->InsertBefore(movFloat);

    return instrFirst;
}

IR::LabelInstr *
LowererMDArch::GetBailOutStackRestoreLabel(BailOutInfo * bailOutInfo, IR::LabelInstr * exitTargetInstr)
{
    return exitTargetInstr;
}

bool LowererMDArch::GenerateFastAnd(IR::Instr * instrAnd)
{
    return true;
}

bool LowererMDArch::GenerateFastDivAndRem_Signed(IR::Instr* instrDiv)
{
    Assert(instrDiv->m_opcode == Js::OpCode::Div_I4 || instrDiv->m_opcode == Js::OpCode::Rem_I4);

    IR::Opnd* divident  = instrDiv->GetSrc1(); // nominator
    IR::Opnd* divisor   = instrDiv->GetSrc2(); // denominator
    IR::Opnd* dst       = instrDiv->GetDst();
    int constDivisor    = divisor->AsIntConstOpnd()->AsInt32();
    IR::Opnd* result = IR::RegOpnd::New(TyInt32, this->m_func);

    bool isNegDevisor   = false;

    Assert(divisor->GetType() == TyInt32); // constopnd->AsInt32() currently does silent casts between int32 and uint32

    if (constDivisor < 0)
    {
        isNegDevisor = true;
        constDivisor *= -1;
    }

    if (constDivisor <= 0 || constDivisor > INT32_MAX - 1)
    {
        return false;
    }
    if (constDivisor == 1)
    {
        // Wasm expects x/-1 not to be folded to -x.
        if (m_func->GetJITFunctionBody()->IsWasmFunction() && isNegDevisor == true)
        {
            return false;
        }
        Lowerer::InsertMove(dst, divident, instrDiv);
    }
    else if (Math::IsPow2(constDivisor)) // Power of two
    {
        // Negative dividents needs the result incremented by 1
        // Following sequence avoids branch
        // For q = n/d and d = 2^k

        //      sar q n k-1   //2^(k-1) if n < 0 else 0
        //      shr q q 32-k
        //      add q q n
        //      sar q q k

        int k = Math::Log2(constDivisor);
        Lowerer::InsertShift(Js::OpCode::Shr_A, false, result, divident, IR::IntConstOpnd::New(k - 1, TyInt8, this->m_func), instrDiv);
        Lowerer::InsertShift(Js::OpCode::ShrU_A, false, result, result, IR::IntConstOpnd::New(32 - k, TyInt8, this->m_func), instrDiv);
        Lowerer::InsertAdd(false, result, result, divident, instrDiv);
        Lowerer::InsertShift(Js::OpCode::Shr_A, false, dst, result, IR::IntConstOpnd::New(k, TyInt8, this->m_func), instrDiv);
    }
    else
    {
        // Ref: Warren's Hacker's Delight, Chapter 10
        //
        // For q = n/d where d is a signed constant
        // Calculate magic_number (multiplier) and shift amounts (shiftAmt) and replace  div with mul and shift

        Js::NumberUtilities::DivMagicNumber magic_number(Js::NumberUtilities::GenerateDivMagicNumber(constDivisor));
        int32 multiplier = magic_number.multiplier;

        // Compute mulhs divident, multiplier
        IR::Opnd* quotient64reg = IR::RegOpnd::New(TyInt64, this->m_func);
        IR::Opnd* divident64Reg = IR::RegOpnd::New(TyInt64, this->m_func);

        Lowerer::InsertMove(divident64Reg, divident, instrDiv);
        IR::Instr* imul = IR::Instr::New(LowererMD::MDImulOpcode, quotient64reg, IR::IntConstOpnd::New(multiplier, TyInt32, this->m_func), divident64Reg, this->m_func);
        instrDiv->InsertBefore(imul);
        LowererMD::Legalize(imul);

        Lowerer::InsertShift(Js::OpCode::Shr_A, false, quotient64reg, quotient64reg, IR::IntConstOpnd::New(32, TyInt8, this->m_func), instrDiv);
        Lowerer::InsertMove(result, quotient64reg, instrDiv);

        // Special handling when divisor is of type 5 and 7.
        if (multiplier < 0)
        {
            Lowerer::InsertAdd(false, result, result, divident, instrDiv);
        }
        if (magic_number.shiftAmt > 0)
        {
            Lowerer::InsertShift(Js::OpCode::Shr_A, false, result, result, IR::IntConstOpnd::New(magic_number.shiftAmt, TyInt8, this->m_func), instrDiv);
        }
        IR::Opnd* tmpReg2 = IR::RegOpnd::New(TyInt32, this->m_func);
        Lowerer::InsertMove(tmpReg2, divident, instrDiv);

        // Add 1 if divisor is less than 0
        Lowerer::InsertShift(Js::OpCode::ShrU_A, false, tmpReg2, tmpReg2, IR::IntConstOpnd::New(31, TyInt8, this->m_func), instrDiv); // 1 if divident < 0, 0 otherwise
        Lowerer::InsertAdd(false, dst, result, tmpReg2, instrDiv);
    }

    // Negate results if divident is less than zero
    if (isNegDevisor)
    {
        Lowerer::InsertSub(false, dst, IR::IntConstOpnd::New(0, TyInt8, this->m_func), dst, instrDiv);
    }
    return true;
}

bool LowererMDArch::GenerateFastDivAndRem_Unsigned(IR::Instr* instrDiv)
{
    Assert(instrDiv->m_opcode == Js::OpCode::DivU_I4 || instrDiv->m_opcode == Js::OpCode::RemU_I4);

    IR::Opnd* divident  = instrDiv->GetSrc1(); // nominator
    IR::Opnd* divisor   = instrDiv->GetSrc2(); // denominator
    IR::Opnd* dst       = instrDiv->GetDst();
    uint constDivisor   = divisor->AsIntConstOpnd()->AsUint32();

    Assert(divisor->GetType() == TyUint32); // IR::Opnd->AsInt32() and IR::Opnd->AsUint32() allows silent casts.

    if (constDivisor <= 0 || constDivisor > UINT32_MAX - 1)
    {
        return false;
    }

    if (constDivisor == 1)
    {
        Lowerer::InsertMove(dst, divident, instrDiv);
    }
    else if (Math::IsPow2(constDivisor)) // Power of two
    {
        int k = Math::Log2(constDivisor);
        Lowerer::InsertShift(Js::OpCode::ShrU_A, false, dst, divident, IR::IntConstOpnd::New(k, TyInt8, this->m_func), instrDiv);
    }
    else
    {
        // Ref: Warren's Hacker's Delight, Chapter 10
        //
        // For q = n/d where d is an unsigned constant
        // Calculate magic_number (multiplier) and shift amounts (shiftAmt) and replace  div with mul and shift

        Js::NumberUtilities::DivMagicNumber magic_number(Js::NumberUtilities::GenerateDivMagicNumber(constDivisor));
        uint multiplier   = magic_number.multiplier;
        int addIndicator  = magic_number.addIndicator;

        IR::Opnd* quotient64Reg = IR::RegOpnd::New(TyUint64, this->m_func);
        IR::Opnd* multiplierReg = IR::RegOpnd::New(TyUint64, this->m_func);

        Lowerer::InsertMove(multiplierReg, IR::IntConstOpnd::New(multiplier, TyInt64, this->m_func), instrDiv);

        IR::Instr* imul = IR::Instr::New(LowererMD::MDImulOpcode, quotient64Reg, divident, multiplierReg, this->m_func);
        instrDiv->InsertBefore(imul);
        LowererMD::Legalize(imul);
        if (!addIndicator) // Simple case type 3, 5..
        {
            Lowerer::InsertShift(Js::OpCode::ShrU_A, false, quotient64Reg, quotient64Reg, IR::IntConstOpnd::New(32 + magic_number.shiftAmt, TyInt8, this->m_func), instrDiv);
            Lowerer::InsertMove(dst, quotient64Reg, instrDiv);
        }
        else // Special case type 7..
        {
            IR::Opnd* tmpReg = IR::RegOpnd::New(TyUint32, this->m_func);
            Lowerer::InsertMove(dst, divident, instrDiv);

            Lowerer::InsertShift(Js::OpCode::ShrU_A, false, quotient64Reg, quotient64Reg, IR::IntConstOpnd::New(32, TyInt8, this->m_func), instrDiv);
            Lowerer::InsertMove(tmpReg, quotient64Reg, instrDiv);

            Lowerer::InsertSub(false, dst, dst, tmpReg, instrDiv);
            Lowerer::InsertShift(Js::OpCode::ShrU_A, false, dst, dst, IR::IntConstOpnd::New(1, TyInt8, this->m_func), instrDiv);
            Lowerer::InsertAdd(false, dst, dst, tmpReg, instrDiv);
            Lowerer::InsertShift(Js::OpCode::ShrU_A, false, dst, dst, IR::IntConstOpnd::New(magic_number.shiftAmt-1, TyInt8, this->m_func), instrDiv);
        }
    }
    return true;
}

bool LowererMDArch::GenerateFastDivAndRem(IR::Instr* instrDiv, IR::LabelInstr* bailOutLabel)
{
    Assert(instrDiv);
    IR::Opnd* divident  = instrDiv->GetSrc1(); // nominator
    IR::Opnd* divisor   = instrDiv->GetSrc2(); // denominator
    IR::Opnd* dst       = instrDiv->GetDst();

    if (divident->GetType() != TyInt32 && divident->GetType() != TyUint32)
    {
        return false;
    }

    if (divident->IsRegOpnd() && divident->AsRegOpnd()->IsSameRegUntyped(dst))
    {
        if (instrDiv->m_opcode == Js::OpCode::Rem_I4 || instrDiv->m_opcode == Js::OpCode::RemU_I4 || bailOutLabel)
        {
            divident = IR::RegOpnd::New(TyInt32, instrDiv->m_func);
            Lowerer::InsertMove(divident, instrDiv->GetSrc1(), instrDiv);
        }
    }

    if (PHASE_OFF(Js::BitopsFastPathPhase, this->m_func) || !divisor->IsIntConstOpnd())
    {
        return false;
    }

    bool success = false;
    if (instrDiv->m_opcode == Js::OpCode::DivU_I4 || instrDiv->m_opcode == Js::OpCode::RemU_I4)
    {
        success = GenerateFastDivAndRem_Unsigned(instrDiv);
    }
    else if (instrDiv->m_opcode == Js::OpCode::Div_I4 || instrDiv->m_opcode == Js::OpCode::Rem_I4)
    {
        success = GenerateFastDivAndRem_Signed(instrDiv);
    }

    if (!success)
    {
        return false;
    }
    // For reminder/mod ops
    if (instrDiv->m_opcode == Js::OpCode::Rem_I4 || instrDiv->m_opcode == Js::OpCode::RemU_I4 || bailOutLabel)
    {
        // For q = n/d
        // mul dst, dst, divident
        // sub dst, divident, dst
        IR::Opnd* reminderOpnd = dst;;
        if (bailOutLabel)
        {
            reminderOpnd = IR::RegOpnd::New(TyInt32, instrDiv->m_func);
        }

        IR::Instr* imul = IR::Instr::New(LowererMD::MDImulOpcode, reminderOpnd, dst, divisor, instrDiv->m_func);
        instrDiv->InsertBefore(imul);
        LowererMD::Legalize(imul);
        Lowerer::InsertSub(false, reminderOpnd, divident, reminderOpnd, instrDiv);
        if (bailOutLabel)
        {
            Lowerer::InsertTestBranch(reminderOpnd, reminderOpnd, Js::OpCode::BrNeq_A, bailOutLabel, instrDiv);
        }
    }

    // DIV/REM has been optimized and can be removed now.
    instrDiv->Remove();
    return true;
}

bool LowererMDArch::GenerateFastXor(IR::Instr * instrXor)
{
    return true;
}

bool LowererMDArch::GenerateFastOr(IR::Instr * instrOr)
{
    return true;
}

bool LowererMDArch::GenerateFastNot(IR::Instr * instrNot)
{
    return true;
}

bool LowererMDArch::GenerateFastShiftLeft(IR::Instr * instrShift)
{
    return true;
}

bool LowererMDArch::GenerateFastShiftRight(IR::Instr * instrShift)
{
    // Given:
    //
    // dst = Shr/ShrU src1, src2
    //
    // Generate:
    //
    // (If not 2 Int31's, jump to $helper.)
    // s1 = MOV src1
    //RCX = MOV src2
    //      TEST RCX, 0x1F             [unsigned only]  // Bail if unsigned and not shifting,
    //      JEQ $helper                [unsigned only]  // as we may not end up with a taggable int
    // s1 = SAR/SHR s1, RCX
    //      BTS s1, VarTag_Shift
    //dst = MOV s1
    //      JMP $fallthru
    // $helper:
    //      (caller generates helper call)
    // $fallthru:

    IR::Instr *      instr;
    IR::LabelInstr * labelHelper;
    IR::LabelInstr * labelFallThru;
    IR::Opnd *       opndReg;
    IR::Opnd *       opndSrc1;
    IR::Opnd *       opndSrc2;
    Assert(instrShift->m_opcode == Js::OpCode::ShrU_A || instrShift->m_opcode == Js::OpCode::Shr_A);
    bool             isUnsigned = (instrShift->m_opcode == Js::OpCode::ShrU_A);

    opndSrc1 = instrShift->GetSrc1();
    opndSrc2 = instrShift->GetSrc2();
    AssertMsg(opndSrc1 && opndSrc2, "Expected 2 src opnd's on Add instruction");

    // Not int?
    if (opndSrc1->IsRegOpnd() && opndSrc1->AsRegOpnd()->IsNotInt())
    {
        return true;
    }
    if (opndSrc2->IsRegOpnd() && opndSrc2->AsRegOpnd()->IsNotInt())
    {
        return true;
    }

    // Tagged ints?
    bool isTaggedInts = false;
    if (opndSrc1->IsTaggedInt())
    {
        if (opndSrc2->IsTaggedInt())
        {
            isTaggedInts = true;
        }
    }

    IntConstType s2Value = 0;
    bool src2IsIntConst = false;

    if (isUnsigned)
    {
        if (opndSrc2->IsRegOpnd())
        {
            src2IsIntConst = opndSrc2->AsRegOpnd()->m_sym->IsTaggableIntConst();
            if (src2IsIntConst)
            {
                s2Value = opndSrc2->AsRegOpnd()->m_sym->GetIntConstValue();
            }
        }
        else
        {
            AssertMsg(opndSrc2->IsAddrOpnd() && Js::TaggedInt::Is(opndSrc2->AsAddrOpnd()->m_address),
                "Expect src2 of shift right to be reg or Var.");
            src2IsIntConst = true;
            s2Value = Js::TaggedInt::ToInt32(opndSrc2->AsAddrOpnd()->m_address);
        }

        // 32-bit Shifts only uses the bottom 5 bits.
        s2Value &=  0x1F;

        // Unsigned shift by 0 could yield a value not encodable as a tagged int.
        if (isUnsigned && src2IsIntConst && s2Value == 0)
        {
            return true;
        }
    }


    labelHelper = IR::LabelInstr::New(Js::OpCode::Label, this->m_func, true);

    if (!isTaggedInts)
    {
        // (If not 2 Int31's, jump to $helper.)

        this->lowererMD->GenerateSmIntPairTest(instrShift, opndSrc1, opndSrc2, labelHelper);
    }

    opndSrc1    = opndSrc1->UseWithNewType(TyInt32, this->m_func);
    if (src2IsIntConst)
    {
        opndSrc2 = IR::IntConstOpnd::New(s2Value, TyInt32, this->m_func);
    }
    else
    {
        // RCX = MOV src2
        opndSrc2 = opndSrc2->UseWithNewType(TyInt32, this->m_func);
        opndReg = IR::RegOpnd::New(TyInt32, this->m_func);
        opndReg->AsRegOpnd()->SetReg(this->GetRegShiftCount());
        instr = IR::Instr::New(Js::OpCode::MOV, opndReg, opndSrc2, this->m_func);
        instrShift->InsertBefore(instr);
        opndSrc2 = opndReg;
    }

    if (!src2IsIntConst && isUnsigned)
    {
        //      TEST RCX, 0x1F             [unsigned only]  // Bail if unsigned and not shifting,
        instr = IR::Instr::New(Js::OpCode::TEST, this->m_func);
        instr->SetSrc1(opndSrc2);
        instr->SetSrc2(IR::IntConstOpnd::New(0x1F, TyInt32, this->m_func));
        instrShift->InsertBefore(instr);

        //      JEQ $helper                  [unsigned only]  // as we may not end up with a taggable int
        instr = IR::BranchInstr::New(Js::OpCode::JEQ, labelHelper, this->m_func);
        instrShift->InsertBefore(instr);
    }

    // s1 = MOV src1
    opndReg = IR::RegOpnd::New(TyInt32, this->m_func);
    instr = IR::Instr::New(Js::OpCode::MOV, opndReg, opndSrc1, this->m_func);
    instrShift->InsertBefore(instr);

    // s1 = SAR/SHR s1, RCX
    instr = IR::Instr::New(isUnsigned ? Js::OpCode::SHR : Js::OpCode::SAR, opndReg, opndReg, opndSrc2, this->m_func);
    instrShift->InsertBefore(instr);

    //
    // Convert TyInt32 operand, back to TyMachPtr type.
    //

    if(TyMachReg != opndReg->GetType())
    {
        opndReg = opndReg->UseWithNewType(TyMachPtr, this->m_func);
    }

    //    BTS s1, VarTag_Shift
    this->lowererMD->GenerateInt32ToVarConversion(opndReg, instrShift);

    // dst = MOV s1

    instr = IR::Instr::New(Js::OpCode::MOV, instrShift->GetDst(), opndReg, this->m_func);
    instrShift->InsertBefore(instr);

    //      JMP $fallthru

    labelFallThru = IR::LabelInstr::New(Js::OpCode::Label, this->m_func);
    instr = IR::BranchInstr::New(Js::OpCode::JMP, labelFallThru, this->m_func);
    instrShift->InsertBefore(instr);

    // $helper:
    //      (caller generates helper call)
    // $fallthru:

    instrShift->InsertBefore(labelHelper);
    instrShift->InsertAfter(labelFallThru);

    return true;
}

void
LowererMDArch::FinalLower()
{
    IR::IntConstOpnd *intOpnd;

    FOREACH_INSTR_BACKWARD_EDITING_IN_RANGE(instr, instrPrev, this->m_func->m_tailInstr, this->m_func->m_headInstr)
    {
        switch (instr->m_opcode)
        {
        case Js::OpCode::Ret:
            instr->Remove();
            break;
        case Js::OpCode::LdArgSize:
            Assert(this->m_func->HasTry());
            instr->m_opcode = Js::OpCode::MOV;
            intOpnd = IR::IntConstOpnd::New(this->m_func->GetArgsSize(), TyUint32, this->m_func);
            instr->SetSrc1(intOpnd);
            LowererMD::Legalize(instr);
            break;

        case Js::OpCode::LdSpillSize:
            Assert(this->m_func->HasTry());
            instr->m_opcode = Js::OpCode::MOV;
            intOpnd = IR::IntConstOpnd::New(this->m_func->GetSpillSize(), TyUint32, this->m_func);
            instr->SetSrc1(intOpnd);
            LowererMD::Legalize(instr);
            break;

        case Js::OpCode::Leave:
            Assert(this->m_func->DoOptimizeTry() && !this->m_func->IsLoopBodyInTry());
            instrPrev = this->lowererMD->m_lowerer->LowerLeave(instr, instr->AsBranchInstr()->GetTarget(), true /*fromFinalLower*/);
            break;

        case Js::OpCode::CMOVA:
        case Js::OpCode::CMOVAE:
        case Js::OpCode::CMOVB:
        case Js::OpCode::CMOVBE:
        case Js::OpCode::CMOVE:
        case Js::OpCode::CMOVG:
        case Js::OpCode::CMOVGE:
        case Js::OpCode::CMOVL:
        case Js::OpCode::CMOVLE:
        case Js::OpCode::CMOVNE:
        case Js::OpCode::CMOVNO:
        case Js::OpCode::CMOVNP:
        case Js::OpCode::CMOVNS:
        case Js::OpCode::CMOVO:
        case Js::OpCode::CMOVP:
        case Js::OpCode::CMOVS:
            // Get rid of fake src1.
            if (instr->GetSrc2())
            {
                // CMOV inserted before regalloc have a dummy src1 to simulate the fact that
                // CMOV is not a definite def of the dst.
                instr->SwapOpnds();
                instr->FreeSrc2();
            }
            break;
        case Js::OpCode::LOCKCMPXCHG8B:
        case Js::OpCode::CMPXCHG8B:
            // Get rid of the deps and srcs
            instr->FreeDst();
            instr->FreeSrc2();
            break;
        default:
            if (instr->HasLazyBailOut())
            {
                // Since Lowerer and Peeps might have removed instructions with lazy bailout
                // if we attach them to helper calls, FinalLower is the first phase that
                // we can know if the function has any lazy bailouts at all.
                this->m_func->SetHasLazyBailOut();
            }

            break;
        }
    } NEXT_INSTR_BACKWARD_EDITING_IN_RANGE;
}

IR::Opnd*
LowererMDArch::GenerateArgOutForStackArgs(IR::Instr* callInstr, IR::Instr* stackArgsInstr)
{
    return this->lowererMD->m_lowerer->GenerateArgOutForStackArgs(callInstr, stackArgsInstr);
}

void
LowererMDArch::LowerInlineSpreadArgOutLoop(IR::Instr *callInstr, IR::RegOpnd *indexOpnd, IR::RegOpnd *arrayElementsStartOpnd)
{
    this->lowererMD->m_lowerer->LowerInlineSpreadArgOutLoopUsingRegisters(callInstr, indexOpnd, arrayElementsStartOpnd);
}

IR::Instr *
LowererMDArch::LowerEHRegionReturn(IR::Instr * insertBeforeInstr, IR::Opnd * targetOpnd)
{
    IR::RegOpnd *retReg = IR::RegOpnd::New(StackSym::New(TyMachReg, this->m_func), GetRegReturn(TyMachReg), TyMachReg, this->m_func);

    // Load the continuation address into the return register.
    insertBeforeInstr->InsertBefore(IR::Instr::New(Js::OpCode::MOV, retReg, targetOpnd, this->m_func));

    // MOV REG_EH_SPILL_SIZE, spillSize
    IR::Instr *movSpillSize = IR::Instr::New(Js::OpCode::LdSpillSize,
        IR::RegOpnd::New(nullptr, REG_EH_SPILL_SIZE, TyMachReg, m_func),
        m_func);
    insertBeforeInstr->InsertBefore(movSpillSize);


    // MOV REG_EH_ARGS_SIZE, argsSize
    IR::Instr *movArgsSize = IR::Instr::New(Js::OpCode::LdArgSize,
        IR::RegOpnd::New(nullptr, REG_EH_ARGS_SIZE, TyMachReg, m_func),
        m_func);
    insertBeforeInstr->InsertBefore(movArgsSize);

    // MOV REG_EH_TARGET, amd64_ReturnFromCallWithFakeFrame
    // PUSH REG_EH_TARGET
    // RET
    IR::Opnd *endCallWithFakeFrame = endCallWithFakeFrame =
        IR::RegOpnd::New(nullptr, REG_EH_TARGET, TyMachReg, m_func);
    IR::Instr *movTarget = IR::Instr::New(Js::OpCode::MOV,
        endCallWithFakeFrame,
        IR::HelperCallOpnd::New(IR::HelperOp_ReturnFromCallWithFakeFrame, m_func),
        m_func);
    insertBeforeInstr->InsertBefore(movTarget);

    IR::Instr *push = IR::Instr::New(Js::OpCode::PUSH, m_func);
    push->SetSrc1(endCallWithFakeFrame);
    insertBeforeInstr->InsertBefore(push);

#if 0
        // TODO: This block gets deleted if we emit a JMP instead of a RET.
    IR::BranchInstr *jmp = IR::BranchInstr::New(Js::OpCode::JMP,
        nullptr,
        targetOpnd,
        m_func);
    leaveInstr->InsertBefore(jmp);
#endif

    IR::IntConstOpnd *intSrc = IR::IntConstOpnd::New(0, TyInt32, this->m_func);
    IR::Instr * retInstr = IR::Instr::New(Js::OpCode::RET, this->m_func);
    retInstr->SetSrc1(intSrc);
    retInstr->SetSrc2(retReg);
    insertBeforeInstr->InsertBefore(retInstr);

    // return the last instruction inserted
    return retInstr;
}

IR::BranchInstr*
LowererMDArch::InsertMissingItemCompareBranch(IR::Opnd* compareSrc, IR::Opnd* missingItemOpnd, Js::OpCode opcode, IR::LabelInstr* target, IR::Instr* insertBeforeInstr)
{
    Assert(compareSrc->IsFloat64() && missingItemOpnd->IsUint64());

    IR::Opnd * compareSrcUint64Opnd = IR::RegOpnd::New(TyUint64, m_func);

    if (compareSrc->IsRegOpnd())
    {
        this->lowererMD->EmitReinterpretPrimitive(compareSrcUint64Opnd, compareSrc, insertBeforeInstr);
    }
    else if (compareSrc->IsIndirOpnd())
    {
        compareSrcUint64Opnd = compareSrc->UseWithNewType(TyUint64, m_func);
    }

    return this->lowererMD->m_lowerer->InsertCompareBranch(missingItemOpnd, compareSrcUint64Opnd, opcode, target, insertBeforeInstr);
}