Emitter: Changed a lot of 'i's into 'x's, because...

<JakeStine>sudonim says he likes xMOV better than iMOV as an emitter prefix. I'm wondering if I should go ahead and change it.
<JakeStine>I tend to favor logic, but everyone else just thinks it looks like iMac and iPod
<JakeStine>I just don't want to have to change it more than once.
<cotton>well 'x' is like the algebraic variable, which can be anything
<cotton>so it does kindoff make sense
<cotton>cuz like
<cotton>you have xSOMETHING, where SOMETHING is all the different emitter functions

... I'm sold. :p

git-svn-id: http://pcsx2.googlecode.com/svn/trunk@1030 96395faa-99c1-11dd-bbfe-3dabce05a288
This commit is contained in:
Jake.Stine 2009-04-20 03:10:05 +00:00
parent 3ee59f3f4e
commit 18c4765d31
20 changed files with 1105 additions and 1110 deletions

View File

@ -802,12 +802,12 @@ void _saveEAX(VURegs *VU, int x86reg, uptr offset, int info)
// (this is one of my test cases for the new emitter --air)
using namespace x86Emitter;
iAddressReg thisreg( x86reg );
xAddressReg thisreg( x86reg );
if ( _X ) iMOV(ptr32[thisreg+offset], 0x00000000);
if ( _Y ) iMOV(ptr32[thisreg+offset+4], 0x00000000);
if ( _Z ) iMOV(ptr32[thisreg+offset+8], 0x00000000);
if ( _W ) iMOV(ptr32[thisreg+offset+12], 0x3f800000);
if ( _X ) xMOV(ptr32[thisreg+offset], 0x00000000);
if ( _Y ) xMOV(ptr32[thisreg+offset+4], 0x00000000);
if ( _Z ) xMOV(ptr32[thisreg+offset+8], 0x00000000);
if ( _W ) xMOV(ptr32[thisreg+offset+12], 0x3f800000);
}
return;
}

View File

@ -34,20 +34,20 @@ void MOV128_MtoM( x86IntRegType destRm, x86IntRegType srcRm )
{
// (this is one of my test cases for the new emitter --air)
iAddressReg src( srcRm );
iAddressReg dest( destRm );
xAddressReg src( srcRm );
xAddressReg dest( destRm );
iMOV( eax, ptr[src] );
iMOV( ptr[dest], eax );
xMOV( eax, ptr[src] );
xMOV( ptr[dest], eax );
iMOV( eax, ptr[src+4] );
iMOV( ptr[dest+4], eax );
xMOV( eax, ptr[src+4] );
xMOV( ptr[dest+4], eax );
iMOV( eax, ptr[src+8] );
iMOV( ptr[dest+8], eax );
xMOV( eax, ptr[src+8] );
xMOV( ptr[dest+8], eax );
iMOV( eax, ptr[src+12] );
iMOV( ptr[dest+12], eax );
xMOV( eax, ptr[src+12] );
xMOV( ptr[dest+12], eax );
}
/*
@ -200,10 +200,10 @@ void vtlb_DynGenRead64(u32 bits)
SHR32ItoR(EAX,VTLB_PAGE_BITS);
MOV32RmSOffsettoR(EAX,EAX,(int)vtlbdata.vmap,2);
ADD32RtoR(ECX,EAX);
iForwardJS8 _fullread;
xForwardJS8 _fullread;
_vtlb_DynGen_DirectRead( bits, false );
iForwardJump8 cont;
xForwardJump8 cont;
_fullread.SetTarget();
@ -223,10 +223,10 @@ void vtlb_DynGenRead32(u32 bits, bool sign)
SHR32ItoR(EAX,VTLB_PAGE_BITS);
MOV32RmSOffsettoR(EAX,EAX,(int)vtlbdata.vmap,2);
ADD32RtoR(ECX,EAX);
iForwardJS8 _fullread;
xForwardJS8 _fullread;
_vtlb_DynGen_DirectRead( bits, sign );
iForwardJump8 cont;
xForwardJump8 cont;
_fullread.SetTarget();
_vtlb_DynGen_IndirectRead( bits );
@ -478,10 +478,10 @@ void vtlb_DynGenWrite(u32 sz)
SHR32ItoR(EAX,VTLB_PAGE_BITS);
MOV32RmSOffsettoR(EAX,EAX,(int)vtlbdata.vmap,2);
ADD32RtoR(ECX,EAX);
iForwardJS8 _full;
xForwardJS8 _full;
_vtlb_DynGen_DirectWrite( sz );
iForwardJump8 cont;
xForwardJump8 cont;
_full.SetTarget();
_vtlb_DynGen_IndirectWrite( sz );

View File

@ -41,54 +41,54 @@ class Group8Impl
protected:
static const uint OperandSize = sizeof(ImmType);
static void prefix16() { if( OperandSize == 2 ) iWrite<u8>( 0x66 ); }
static void prefix16() { if( OperandSize == 2 ) xWrite<u8>( 0x66 ); }
public:
Group8Impl() {} // For the love of GCC.
// ------------------------------------------------------------------------
static __emitinline void Emit( const iRegister<ImmType>& bitbase, const iRegister<ImmType>& bitoffset )
static __emitinline void Emit( const xRegister<ImmType>& bitbase, const xRegister<ImmType>& bitoffset )
{
prefix16();
iWrite<u8>( 0x0f );
iWrite<u8>( 0xa3 | (InstType << 2) );
xWrite<u8>( 0x0f );
xWrite<u8>( 0xa3 | (InstType << 2) );
ModRM_Direct( bitoffset.Id, bitbase.Id );
}
// ------------------------------------------------------------------------
static __emitinline void Emit( void* bitbase, const iRegister<ImmType>& bitoffset )
static __emitinline void Emit( void* bitbase, const xRegister<ImmType>& bitoffset )
{
prefix16();
iWrite<u8>( 0x0f );
iWrite<u8>( 0xa3 | (InstType << 2) );
iWriteDisp( bitoffset.Id, bitbase );
xWrite<u8>( 0x0f );
xWrite<u8>( 0xa3 | (InstType << 2) );
xWriteDisp( bitoffset.Id, bitbase );
}
// ------------------------------------------------------------------------
static __emitinline void Emit( const ModSibBase& bitbase, const iRegister<ImmType>& bitoffset )
static __emitinline void Emit( const ModSibBase& bitbase, const xRegister<ImmType>& bitoffset )
{
prefix16();
iWrite<u8>( 0x0f );
iWrite<u8>( 0xa3 | (InstType << 2) );
xWrite<u8>( 0x0f );
xWrite<u8>( 0xa3 | (InstType << 2) );
EmitSibMagic( bitoffset.Id, bitbase );
}
// ------------------------------------------------------------------------
static __emitinline void Emit( const iRegister<ImmType>& bitbase, u8 immoffset )
static __emitinline void Emit( const xRegister<ImmType>& bitbase, u8 immoffset )
{
prefix16();
iWrite<u16>( 0xba0f );
xWrite<u16>( 0xba0f );
ModRM_Direct( InstType, bitbase.Id );
iWrite<u8>( immoffset );
xWrite<u8>( immoffset );
}
// ------------------------------------------------------------------------
static __emitinline void Emit( const ModSibStrict<ImmType>& bitbase, u8 immoffset )
{
prefix16();
iWrite<u16>( 0xba0f );
xWrite<u16>( 0xba0f );
EmitSibMagic( InstType, bitbase );
iWrite<u8>( immoffset );
xWrite<u8>( immoffset );
}
};
@ -102,12 +102,12 @@ protected:
typedef Group8Impl<InstType,u32> m_16;
public:
__forceinline void operator()( const iRegister32& bitbase, const iRegister32& bitoffset ) const { m_32::Emit( bitbase, bitoffset ); }
__forceinline void operator()( const iRegister16& bitbase, const iRegister16& bitoffset ) const { m_16::Emit( bitbase, bitoffset ); }
__forceinline void operator()( void* bitbase, const iRegister32& bitoffset ) const { m_32::Emit( bitbase, bitoffset ); }
__forceinline void operator()( void* bitbase, const iRegister16& bitoffset ) const { m_16::Emit( bitbase, bitoffset ); }
__noinline void operator()( const ModSibBase& bitbase, const iRegister32& bitoffset ) const { m_32::Emit( bitbase, bitoffset ); }
__noinline void operator()( const ModSibBase& bitbase, const iRegister16& bitoffset ) const { m_16::Emit( bitbase, bitoffset ); }
__forceinline void operator()( const xRegister32& bitbase, const xRegister32& bitoffset ) const { m_32::Emit( bitbase, bitoffset ); }
__forceinline void operator()( const xRegister16& bitbase, const xRegister16& bitoffset ) const { m_16::Emit( bitbase, bitoffset ); }
__forceinline void operator()( void* bitbase, const xRegister32& bitoffset ) const { m_32::Emit( bitbase, bitoffset ); }
__forceinline void operator()( void* bitbase, const xRegister16& bitoffset ) const { m_16::Emit( bitbase, bitoffset ); }
__noinline void operator()( const ModSibBase& bitbase, const xRegister32& bitoffset ) const { m_32::Emit( bitbase, bitoffset ); }
__noinline void operator()( const ModSibBase& bitbase, const xRegister16& bitoffset ) const { m_16::Emit( bitbase, bitoffset ); }
// Note on Imm forms : use int as the source operand since it's "reasonably inert" from a compiler
// perspective. (using uint tends to make the compiler try and fail to match signed immediates with
@ -115,8 +115,8 @@ public:
__noinline void operator()( const ModSibStrict<u32>& bitbase, u8 immoffset ) const { m_32::Emit( bitbase, immoffset ); }
__noinline void operator()( const ModSibStrict<u16>& bitbase, u8 immoffset ) const { m_16::Emit( bitbase, immoffset ); }
void operator()( const iRegister<u32>& bitbase, u8 immoffset ) const { m_32::Emit( bitbase, immoffset ); }
void operator()( const iRegister<u16>& bitbase, u8 immoffset ) const { m_16::Emit( bitbase, immoffset ); }
void operator()( const xRegister<u32>& bitbase, u8 immoffset ) const { m_32::Emit( bitbase, immoffset ); }
void operator()( const xRegister<u16>& bitbase, u8 immoffset ) const { m_16::Emit( bitbase, immoffset ); }
Group8ImplAll() {}
};
@ -130,33 +130,33 @@ class BitScanImpl
{
protected:
static const uint OperandSize = sizeof(ImmType);
static void prefix16() { if( OperandSize == 2 ) iWrite<u8>( 0x66 ); }
static void prefix16() { if( OperandSize == 2 ) xWrite<u8>( 0x66 ); }
static void emitbase()
{
prefix16();
iWrite<u8>( 0x0f );
iWrite<u8>( isReverse ? 0xbd : 0xbc );
xWrite<u8>( 0x0f );
xWrite<u8>( isReverse ? 0xbd : 0xbc );
}
public:
BitScanImpl() {} // For the love of GCC.
// ------------------------------------------------------------------------
static __emitinline void Emit( const iRegister<ImmType>& to, const iRegister<ImmType>& from )
static __emitinline void Emit( const xRegister<ImmType>& to, const xRegister<ImmType>& from )
{
emitbase();
ModRM_Direct( to.Id, from.Id );
}
// ------------------------------------------------------------------------
static __emitinline void Emit( const iRegister<ImmType>& to, const void* src )
static __emitinline void Emit( const xRegister<ImmType>& to, const void* src )
{
emitbase();
iWriteDisp( to.Id, src );
xWriteDisp( to.Id, src );
}
// ------------------------------------------------------------------------
static __emitinline void Emit( const iRegister<ImmType>& to, const ModSibBase& sibsrc )
static __emitinline void Emit( const xRegister<ImmType>& to, const ModSibBase& sibsrc )
{
emitbase();
EmitSibMagic( to.Id, sibsrc );
@ -175,12 +175,12 @@ protected:
typedef BitScanImpl<isReverse,u32> m_16;
public:
__forceinline void operator()( const iRegister32& to, const iRegister32& from ) const { m_32::Emit( to, from ); }
__forceinline void operator()( const iRegister16& to, const iRegister16& from ) const { m_16::Emit( to, from ); }
__forceinline void operator()( const iRegister32& to, const void* src ) const { m_32::Emit( to, src ); }
__forceinline void operator()( const iRegister16& to, const void* src ) const { m_16::Emit( to, src ); }
__noinline void operator()( const iRegister32& to, const ModSibBase& sibsrc ) const { m_32::Emit( to, sibsrc ); }
__noinline void operator()( const iRegister16& to, const ModSibBase& sibsrc ) const { m_16::Emit( to, sibsrc ); }
__forceinline void operator()( const xRegister32& to, const xRegister32& from ) const { m_32::Emit( to, from ); }
__forceinline void operator()( const xRegister16& to, const xRegister16& from ) const { m_16::Emit( to, from ); }
__forceinline void operator()( const xRegister32& to, const void* src ) const { m_32::Emit( to, src ); }
__forceinline void operator()( const xRegister16& to, const void* src ) const { m_16::Emit( to, src ); }
__noinline void operator()( const xRegister32& to, const ModSibBase& sibsrc ) const { m_32::Emit( to, sibsrc ); }
__noinline void operator()( const xRegister16& to, const ModSibBase& sibsrc ) const { m_16::Emit( to, sibsrc ); }
BitScanImplAll() {}
};

View File

@ -33,7 +33,7 @@ protected:
static const uint OperandSize = sizeof(ImmType);
static bool Is8BitOperand() { return OperandSize == 1; }
static void prefix16() { if( OperandSize == 2 ) iWrite<u8>( 0x66 ); }
static void prefix16() { if( OperandSize == 2 ) xWrite<u8>( 0x66 ); }
static void basesibform( bool isCL )
{
@ -46,7 +46,7 @@ public:
DwordShiftImpl() {} // because GCC doesn't like static classes
// ------------------------------------------------------------------------
static __emitinline void Emit( const iRegister<ImmType>& to, const iRegister<ImmType>& from )
static __emitinline void Emit( const xRegister<ImmType>& to, const xRegister<ImmType>& from )
{
prefix16();
write16( 0xa50f | (isShiftRight ? 0x800 : 0) );
@ -54,7 +54,7 @@ public:
}
// ------------------------------------------------------------------------
static __emitinline void Emit( const iRegister<ImmType>& to, const iRegister<ImmType>& from, u8 imm )
static __emitinline void Emit( const xRegister<ImmType>& to, const xRegister<ImmType>& from, u8 imm )
{
if( imm == 0 ) return;
prefix16();
@ -64,14 +64,14 @@ public:
}
// ------------------------------------------------------------------------
static __emitinline void Emit( const ModSibBase& sibdest, const iRegister<ImmType>& from, __unused const iRegisterCL& clreg )
static __emitinline void Emit( const ModSibBase& sibdest, const xRegister<ImmType>& from, __unused const xRegisterCL& clreg )
{
basesibform();
EmitSibMagic( from.Id, sibdest );
}
// ------------------------------------------------------------------------
static __emitinline void Emit( const ModSibBase& sibdest, const iRegister<ImmType>& from, u8 imm )
static __emitinline void Emit( const ModSibBase& sibdest, const xRegister<ImmType>& from, u8 imm )
{
basesibform();
EmitSibMagic( from.Id, sibdest );
@ -80,18 +80,18 @@ public:
// ------------------------------------------------------------------------
// dest data type is inferred from the 'from' register, so we can do void* resolution :)
static __emitinline void Emit( void* dest, const iRegister<ImmType>& from, __unused const iRegisterCL& clreg )
static __emitinline void Emit( void* dest, const xRegister<ImmType>& from, __unused const xRegisterCL& clreg )
{
basesibform();
iWriteDisp( from.Id, dest );
xWriteDisp( from.Id, dest );
}
// ------------------------------------------------------------------------
// dest data type is inferred from the 'from' register, so we can do void* resolution :)
static __emitinline void Emit( void* dest, const iRegister<ImmType>& from, u8 imm )
static __emitinline void Emit( void* dest, const xRegister<ImmType>& from, u8 imm )
{
basesibform();
iWriteDisp( from.Id, dest );
xWriteDisp( from.Id, dest );
write8( imm );
}
};
@ -110,20 +110,20 @@ protected:
public:
// ---------- 32 Bit Interface -----------
__forceinline void operator()( const iRegister32& to, const iRegister32& from, __unused const iRegisterCL& clreg ) const { m_32::Emit( to, from ); }
__forceinline void operator()( void* dest, const iRegister32& from, __unused const iRegisterCL& clreg ) const { m_32::Emit( dest, from ); }
__noinline void operator()( const ModSibBase& sibdest, const iRegister32& from, __unused const iRegisterCL& clreg ) const { m_32::Emit( sibdest, from ); }
__forceinline void operator()( const iRegister32& to, const iRegister32& from, u8 imm ) const { m_32::Emit( to, from, imm ); }
__forceinline void operator()( void* dest, const iRegister32& from, u8 imm ) const { m_32::Emit( dest, from, imm ); }
__noinline void operator()( const ModSibBase& sibdest, const iRegister32& from, u8 imm ) const { m_32::Emit( sibdest, from ); }
__forceinline void operator()( const xRegister32& to, const xRegister32& from, __unused const xRegisterCL& clreg ) const { m_32::Emit( to, from ); }
__forceinline void operator()( void* dest, const xRegister32& from, __unused const xRegisterCL& clreg ) const { m_32::Emit( dest, from ); }
__noinline void operator()( const ModSibBase& sibdest, const xRegister32& from, __unused const xRegisterCL& clreg ) const { m_32::Emit( sibdest, from ); }
__forceinline void operator()( const xRegister32& to, const xRegister32& from, u8 imm ) const { m_32::Emit( to, from, imm ); }
__forceinline void operator()( void* dest, const xRegister32& from, u8 imm ) const { m_32::Emit( dest, from, imm ); }
__noinline void operator()( const ModSibBase& sibdest, const xRegister32& from, u8 imm ) const { m_32::Emit( sibdest, from ); }
// ---------- 16 Bit Interface -----------
__forceinline void operator()( const iRegister16& to, const iRegister16& from, __unused const iRegisterCL& clreg ) const { m_16::Emit( to, from ); }
__forceinline void operator()( void* dest, const iRegister16& from, __unused const iRegisterCL& clreg ) const { m_16::Emit( dest, from ); }
__noinline void operator()( const ModSibBase& sibdest, const iRegister16& from, __unused const iRegisterCL& clreg ) const { m_16::Emit( sibdest, from ); }
__forceinline void operator()( const iRegister16& to, const iRegister16& from, u8 imm ) const { m_16::Emit( to, from, imm ); }
__forceinline void operator()( void* dest, const iRegister16& from, u8 imm ) const { m_16::Emit( dest, from, imm ); }
__noinline void operator()( const ModSibBase& sibdest, const iRegister16& from, u8 imm ) const { m_16::Emit( sibdest, from ); }
__forceinline void operator()( const xRegister16& to, const xRegister16& from, __unused const xRegisterCL& clreg ) const { m_16::Emit( to, from ); }
__forceinline void operator()( void* dest, const xRegister16& from, __unused const xRegisterCL& clreg ) const { m_16::Emit( dest, from ); }
__noinline void operator()( const ModSibBase& sibdest, const xRegister16& from, __unused const xRegisterCL& clreg ) const { m_16::Emit( sibdest, from ); }
__forceinline void operator()( const xRegister16& to, const xRegister16& from, u8 imm ) const { m_16::Emit( to, from, imm ); }
__forceinline void operator()( void* dest, const xRegister16& from, u8 imm ) const { m_16::Emit( dest, from, imm ); }
__noinline void operator()( const ModSibBase& sibdest, const xRegister16& from, u8 imm ) const { m_16::Emit( sibdest, from ); }
DwordShiftImplAll() {} // Why does GCC need these?
};

View File

@ -41,65 +41,65 @@ protected:
static const uint OperandSize = sizeof(ImmType);
static bool Is8BitOperand() { return OperandSize == 1; }
static void prefix16() { if( OperandSize == 2 ) iWrite<u8>( 0x66 ); }
static void prefix16() { if( OperandSize == 2 ) xWrite<u8>( 0x66 ); }
public:
Group1Impl() {} // because GCC doesn't like static classes
static __emitinline void Emit( G1Type InstType, const iRegister<ImmType>& to, const iRegister<ImmType>& from )
static __emitinline void Emit( G1Type InstType, const xRegister<ImmType>& to, const xRegister<ImmType>& from )
{
prefix16();
iWrite<u8>( (Is8BitOperand() ? 0 : 1) | (InstType<<3) );
xWrite<u8>( (Is8BitOperand() ? 0 : 1) | (InstType<<3) );
ModRM_Direct( from.Id, to.Id );
}
static __emitinline void Emit( G1Type InstType, const ModSibBase& sibdest, const iRegister<ImmType>& from )
static __emitinline void Emit( G1Type InstType, const ModSibBase& sibdest, const xRegister<ImmType>& from )
{
prefix16();
iWrite<u8>( (Is8BitOperand() ? 0 : 1) | (InstType<<3) );
xWrite<u8>( (Is8BitOperand() ? 0 : 1) | (InstType<<3) );
EmitSibMagic( from.Id, sibdest );
}
static __emitinline void Emit( G1Type InstType, const iRegister<ImmType>& to, const ModSibBase& sibsrc )
static __emitinline void Emit( G1Type InstType, const xRegister<ImmType>& to, const ModSibBase& sibsrc )
{
prefix16();
iWrite<u8>( (Is8BitOperand() ? 2 : 3) | (InstType<<3) );
xWrite<u8>( (Is8BitOperand() ? 2 : 3) | (InstType<<3) );
EmitSibMagic( to.Id, sibsrc );
}
static __emitinline void Emit( G1Type InstType, void* dest, const iRegister<ImmType>& from )
static __emitinline void Emit( G1Type InstType, void* dest, const xRegister<ImmType>& from )
{
prefix16();
iWrite<u8>( (Is8BitOperand() ? 0 : 1) | (InstType<<3) );
iWriteDisp( from.Id, dest );
xWrite<u8>( (Is8BitOperand() ? 0 : 1) | (InstType<<3) );
xWriteDisp( from.Id, dest );
}
static __emitinline void Emit( G1Type InstType, const iRegister<ImmType>& to, const void* src )
static __emitinline void Emit( G1Type InstType, const xRegister<ImmType>& to, const void* src )
{
prefix16();
iWrite<u8>( (Is8BitOperand() ? 2 : 3) | (InstType<<3) );
iWriteDisp( to.Id, src );
xWrite<u8>( (Is8BitOperand() ? 2 : 3) | (InstType<<3) );
xWriteDisp( to.Id, src );
}
static __emitinline void Emit( G1Type InstType, const iRegister<ImmType>& to, int imm )
static __emitinline void Emit( G1Type InstType, const xRegister<ImmType>& to, int imm )
{
prefix16();
if( !Is8BitOperand() && is_s8( imm ) )
{
iWrite<u8>( 0x83 );
xWrite<u8>( 0x83 );
ModRM_Direct( InstType, to.Id );
iWrite<s8>( imm );
xWrite<s8>( imm );
}
else
{
if( to.IsAccumulator() )
iWrite<u8>( (Is8BitOperand() ? 4 : 5) | (InstType<<3) );
xWrite<u8>( (Is8BitOperand() ? 4 : 5) | (InstType<<3) );
else
{
iWrite<u8>( Is8BitOperand() ? 0x80 : 0x81 );
xWrite<u8>( Is8BitOperand() ? 0x80 : 0x81 );
ModRM_Direct( InstType, to.Id );
}
iWrite<ImmType>( imm );
xWrite<ImmType>( imm );
}
}
@ -107,19 +107,19 @@ public:
{
if( Is8BitOperand() )
{
iWrite<u8>( 0x80 );
xWrite<u8>( 0x80 );
EmitSibMagic( InstType, sibdest );
iWrite<ImmType>( imm );
xWrite<ImmType>( imm );
}
else
{
prefix16();
iWrite<u8>( is_s8( imm ) ? 0x83 : 0x81 );
xWrite<u8>( is_s8( imm ) ? 0x83 : 0x81 );
EmitSibMagic( InstType, sibdest );
if( is_s8( imm ) )
iWrite<s8>( imm );
xWrite<s8>( imm );
else
iWrite<ImmType>( imm );
xWrite<ImmType>( imm );
}
}
};
@ -132,15 +132,15 @@ class Group1ImplAll
{
public:
template< typename T >
__forceinline void operator()( const iRegister<T>& to, const iRegister<T>& from ) const { Group1Impl<T>::Emit( InstType, to, from ); }
__forceinline void operator()( const xRegister<T>& to, const xRegister<T>& from ) const { Group1Impl<T>::Emit( InstType, to, from ); }
template< typename T >
__forceinline void operator()( const iRegister<T>& to, const void* src ) const { Group1Impl<T>::Emit( InstType, to, src ); }
__forceinline void operator()( const xRegister<T>& to, const void* src ) const { Group1Impl<T>::Emit( InstType, to, src ); }
template< typename T >
__forceinline void operator()( void* dest, const iRegister<T>& from ) const { Group1Impl<T>::Emit( InstType, dest, from ); }
__forceinline void operator()( void* dest, const xRegister<T>& from ) const { Group1Impl<T>::Emit( InstType, dest, from ); }
template< typename T >
__noinline void operator()( const ModSibBase& sibdest, const iRegister<T>& from ) const { Group1Impl<T>::Emit( InstType, sibdest, from ); }
__noinline void operator()( const ModSibBase& sibdest, const xRegister<T>& from ) const { Group1Impl<T>::Emit( InstType, sibdest, from ); }
template< typename T >
__noinline void operator()( const iRegister<T>& to, const ModSibBase& sibsrc ) const { Group1Impl<T>::Emit( InstType, to, sibsrc ); }
__noinline void operator()( const xRegister<T>& to, const ModSibBase& sibsrc ) const { Group1Impl<T>::Emit( InstType, to, sibsrc ); }
// Note on Imm forms : use int as the source operand since it's "reasonably inert" from a compiler
// perspective. (using uint tends to make the compiler try and fail to match signed immediates with
@ -149,38 +149,60 @@ public:
template< typename T >
__noinline void operator()( const ModSibStrict<T>& sibdest, int imm ) const { Group1Impl<T>::Emit( InstType, sibdest, imm ); }
template< typename T >
__forceinline void operator()( const iRegister<T>& to, int imm ) const { Group1Impl<T>::Emit( InstType, to, imm ); }
__forceinline void operator()( const xRegister<T>& to, int imm ) const { Group1Impl<T>::Emit( InstType, to, imm ); }
Group1ImplAll() {} // Why does GCC need these?
};
// ------------------------------------------------------------------------
// This class combines x86 with SSE/SSE2 logic operations (ADD, OR, and NOT).
// Note: ANDN [AndNot] is handled below separately.
//
template< G1Type InstType, u8 OpcodeSSE >
class G1LogicImpl : public Group1ImplAll<InstType>
class G1LogicImpl_PlusSSE : public Group1ImplAll<InstType>
{
public:
using Group1ImplAll<InstType>::operator();
const SSELogicImpl<0x00,OpcodeSSE> PS;
const SSELogicImpl<0x66,OpcodeSSE> PD;
G1LogicImpl() {}
G1LogicImpl_PlusSSE() {}
};
// ------------------------------------------------------------------------
// This calss combines x86 with SSE/SSE2 arithmetic operations (ADD/SUB).
//
template< G1Type InstType, u8 OpcodeSSE >
class G1ArithmeticImpl : public G1LogicImpl<InstType, OpcodeSSE >
class G1ArithmeticImpl_PlusSSE : public G1LogicImpl_PlusSSE<InstType, OpcodeSSE >
{
public:
using Group1ImplAll<InstType>::operator();
const SSELogicImpl<0xf3,OpcodeSSE> SS;
const SSELogicImpl<0xf2,OpcodeSSE> SD;
G1ArithmeticImpl() {}
G1ArithmeticImpl_PlusSSE() {}
};
template< u8 OpcodeSSE >
class SSEAndNotImpl
// ------------------------------------------------------------------------
class G1CompareImpl_PlusSSE : Group1ImplAll< G1Type_CMP >
{
public:
const SSELogicImpl<0x00,OpcodeSSE> PS;
const SSELogicImpl<0x66,OpcodeSSE> PD;
protected:
template< u8 Prefix > struct Woot
{
__forceinline void operator()( const xRegisterSSE& to, const xRegisterSSE& from, SSE2_ComparisonType cmptype ) const { writeXMMop( Prefix, 0xc2, to, from ); xWrite<u8>( cmptype ); }
__forceinline void operator()( const xRegisterSSE& to, const void* from, SSE2_ComparisonType cmptype ) const { writeXMMop( Prefix, 0xc2, to, from ); xWrite<u8>( cmptype ); }
__noinline void operator()( const xRegisterSSE& to, const ModSibBase& from, SSE2_ComparisonType cmptype ) const { writeXMMop( Prefix, 0xc2, to, from ); xWrite<u8>( cmptype ); }
};
SSEAndNotImpl() {}
};
public:
using Group1ImplAll< G1Type_CMP >::operator();
Woot<0x00> PS;
Woot<0x66> PD;
Woot<0xf3> SS;
Woot<0xf2> SD;
G1CompareImpl_PlusSSE() {} //GCWhat?
};

View File

@ -45,21 +45,21 @@ protected:
static const uint OperandSize = sizeof(ImmType);
static bool Is8BitOperand() { return OperandSize == 1; }
static void prefix16() { if( OperandSize == 2 ) iWrite<u8>( 0x66 ); }
static void prefix16() { if( OperandSize == 2 ) xWrite<u8>( 0x66 ); }
public:
Group2Impl() {} // For the love of GCC.
// ------------------------------------------------------------------------
static __emitinline void Emit( const iRegister<ImmType>& to )
static __emitinline void Emit( const xRegister<ImmType>& to )
{
prefix16();
iWrite<u8>( Is8BitOperand() ? 0xd2 : 0xd3 );
xWrite<u8>( Is8BitOperand() ? 0xd2 : 0xd3 );
ModRM_Direct( InstType, to.Id );
}
// ------------------------------------------------------------------------
static __emitinline void Emit( const iRegister<ImmType>& to, u8 imm )
static __emitinline void Emit( const xRegister<ImmType>& to, u8 imm )
{
if( imm == 0 ) return;
@ -67,14 +67,14 @@ public:
if( imm == 1 )
{
// special encoding of 1's
iWrite<u8>( Is8BitOperand() ? 0xd0 : 0xd1 );
xWrite<u8>( Is8BitOperand() ? 0xd0 : 0xd1 );
ModRM_Direct( InstType, to.Id );
}
else
{
iWrite<u8>( Is8BitOperand() ? 0xc0 : 0xc1 );
xWrite<u8>( Is8BitOperand() ? 0xc0 : 0xc1 );
ModRM_Direct( InstType, to.Id );
iWrite<u8>( imm );
xWrite<u8>( imm );
}
}
@ -82,7 +82,7 @@ public:
static __emitinline void Emit( const ModSibStrict<ImmType>& sibdest )
{
prefix16();
iWrite<u8>( Is8BitOperand() ? 0xd2 : 0xd3 );
xWrite<u8>( Is8BitOperand() ? 0xd2 : 0xd3 );
EmitSibMagic( InstType, sibdest );
}
@ -95,14 +95,14 @@ public:
if( imm == 1 )
{
// special encoding of 1's
iWrite<u8>( Is8BitOperand() ? 0xd0 : 0xd1 );
xWrite<u8>( Is8BitOperand() ? 0xd0 : 0xd1 );
EmitSibMagic( InstType, sibdest );
}
else
{
iWrite<u8>( Is8BitOperand() ? 0xc0 : 0xc1 );
xWrite<u8>( Is8BitOperand() ? 0xc0 : 0xc1 );
EmitSibMagic( InstType, sibdest );
iWrite<u8>( imm );
xWrite<u8>( imm );
}
}
};
@ -113,16 +113,16 @@ template< G2Type InstType >
class Group2ImplAll
{
public:
template< typename T > __forceinline void operator()( const iRegister<T>& to, __unused const iRegisterCL& from ) const
template< typename T > __forceinline void operator()( const xRegister<T>& to, __unused const xRegisterCL& from ) const
{ Group2Impl<InstType,T>::Emit( to ); }
template< typename T > __noinline void operator()( const ModSibStrict<T>& sibdest, __unused const iRegisterCL& from ) const
template< typename T > __noinline void operator()( const ModSibStrict<T>& sibdest, __unused const xRegisterCL& from ) const
{ Group2Impl<InstType,T>::Emit( sibdest ); }
template< typename T > __noinline void operator()( const ModSibStrict<T>& sibdest, u8 imm ) const
{ Group2Impl<InstType,T>::Emit( sibdest, imm ); }
template< typename T > __forceinline void operator()( const iRegister<T>& to, u8 imm ) const
template< typename T > __forceinline void operator()( const xRegister<T>& to, u8 imm ) const
{ Group2Impl<InstType,T>::Emit( to, imm ); }
Group2ImplAll() {} // I am a class with no members, so I need an explicit constructor! Sense abounds.

View File

@ -31,6 +31,7 @@ enum G3Type
G3Type_iDIV = 7
};
// ------------------------------------------------------------------------
template< typename ImmType >
class Group3Impl
{
@ -38,22 +39,22 @@ protected:
static const uint OperandSize = sizeof(ImmType);
static bool Is8BitOperand() { return OperandSize == 1; }
static void prefix16() { if( OperandSize == 2 ) iWrite<u8>( 0x66 ); }
static void prefix16() { if( OperandSize == 2 ) xWrite<u8>( 0x66 ); }
public:
Group3Impl() {} // For the love of GCC.
static __emitinline void Emit( G3Type InstType, const iRegister<ImmType>& from )
static __emitinline void Emit( G3Type InstType, const xRegister<ImmType>& from )
{
prefix16();
iWrite<u8>(Is8BitOperand() ? 0xf6 : 0xf7 );
xWrite<u8>(Is8BitOperand() ? 0xf6 : 0xf7 );
ModRM_Direct( InstType, from.Id );
}
static __emitinline void Emit( G3Type InstType, const ModSibStrict<ImmType>& sibsrc )
{
prefix16();
iWrite<u8>( Is8BitOperand() ? 0xf6 : 0xf7 );
xWrite<u8>( Is8BitOperand() ? 0xf6 : 0xf7 );
EmitSibMagic( InstType, sibsrc );
}
};
@ -65,7 +66,7 @@ class Group3ImplAll
{
public:
template< typename T >
__forceinline void operator()( const iRegister<T>& from ) const { Group3Impl<T>::Emit( InstType, from ); }
__forceinline void operator()( const xRegister<T>& from ) const { Group3Impl<T>::Emit( InstType, from ); }
template< typename T >
__noinline void operator()( const ModSibStrict<T>& from ) const { Group3Impl<T>::Emit( InstType, from ); }
@ -73,7 +74,9 @@ public:
Group3ImplAll() {}
};
// ------------------------------------------------------------------------
// This class combines x86 and SSE/SSE2 instructions for iMUL and iDIV.
//
template< G3Type InstType, u8 OpcodeSSE >
class G3Impl_PlusSSE : public Group3ImplAll<InstType>
{
@ -94,11 +97,11 @@ class iMulImpl
{
protected:
static const uint OperandSize = sizeof(ImmType);
static void prefix16() { if( OperandSize == 2 ) iWrite<u8>( 0x66 ); }
static void prefix16() { if( OperandSize == 2 ) xWrite<u8>( 0x66 ); }
public:
// ------------------------------------------------------------------------
static __emitinline void Emit( const iRegister<ImmType>& to, const iRegister<ImmType>& from )
static __emitinline void Emit( const xRegister<ImmType>& to, const xRegister<ImmType>& from )
{
prefix16();
write16( 0xaf0f );
@ -106,15 +109,15 @@ public:
}
// ------------------------------------------------------------------------
static __emitinline void Emit( const iRegister<ImmType>& to, const void* src )
static __emitinline void Emit( const xRegister<ImmType>& to, const void* src )
{
prefix16();
write16( 0xaf0f );
iWriteDisp( to.Id, src );
xWriteDisp( to.Id, src );
}
// ------------------------------------------------------------------------
static __emitinline void Emit( const iRegister<ImmType>& to, const ModSibBase& src )
static __emitinline void Emit( const xRegister<ImmType>& to, const ModSibBase& src )
{
prefix16();
write16( 0xaf0f );
@ -122,7 +125,7 @@ public:
}
// ------------------------------------------------------------------------
static __emitinline void Emit( const iRegister<ImmType>& to, const iRegister<ImmType>& from, ImmType imm )
static __emitinline void Emit( const xRegister<ImmType>& to, const xRegister<ImmType>& from, ImmType imm )
{
prefix16();
write16( is_s8( imm ) ? 0x6b : 0x69 );
@ -130,23 +133,23 @@ public:
if( is_s8( imm ) )
write8( imm );
else
iWrite<ImmType>( imm );
xWrite<ImmType>( imm );
}
// ------------------------------------------------------------------------
static __emitinline void Emit( const iRegister<ImmType>& to, const void* src, ImmType imm )
static __emitinline void Emit( const xRegister<ImmType>& to, const void* src, ImmType imm )
{
prefix16();
write16( is_s8( imm ) ? 0x6b : 0x69 );
iWriteDisp( to.Id, src );
xWriteDisp( to.Id, src );
if( is_s8( imm ) )
write8( imm );
else
iWrite<ImmType>( imm );
xWrite<ImmType>( imm );
}
// ------------------------------------------------------------------------
static __emitinline void Emit( const iRegister<ImmType>& to, const ModSibBase& src, ImmType imm )
static __emitinline void Emit( const xRegister<ImmType>& to, const ModSibBase& src, ImmType imm )
{
prefix16();
write16( is_s8( imm ) ? 0x6b : 0x69 );
@ -154,11 +157,11 @@ public:
if( is_s8( imm ) )
write8( imm );
else
iWrite<ImmType>( imm );
xWrite<ImmType>( imm );
}
};
// ------------------------------------------------------------------------
class iMul_PlusSSE : public G3Impl_PlusSSE<G3Type_iMUL,0x59>
{
protected:
@ -166,24 +169,19 @@ protected:
typedef iMulImpl<u16> iMUL16;
public:
using G3Impl_PlusSSE<G3Type_iMUL,0x59>::operator();
__forceinline void operator()( const xRegister32& to, const xRegister32& from ) const { iMUL32::Emit( to, from ); }
__forceinline void operator()( const xRegister32& to, const void* src ) const { iMUL32::Emit( to, src ); }
__forceinline void operator()( const xRegister32& to, const xRegister32& from, s32 imm ) const{ iMUL32::Emit( to, from, imm ); }
__noinline void operator()( const xRegister32& to, const ModSibBase& src ) const { iMUL32::Emit( to, src ); }
__noinline void operator()( const xRegister32& to, const ModSibBase& from, s32 imm ) const { iMUL32::Emit( to, from, imm ); }
template< typename T >
__forceinline void operator()( const iRegister<T>& from ) const { Group3Impl<T>::Emit( G3Type_iMUL, from ); }
template< typename T >
__noinline void operator()( const ModSibStrict<T>& from ) const { Group3Impl<T>::Emit( G3Type_iMUL, from ); }
__forceinline void operator()( const iRegister32& to, const iRegister32& from ) const { iMUL32::Emit( to, from ); }
__forceinline void operator()( const iRegister32& to, const void* src ) const { iMUL32::Emit( to, src ); }
__forceinline void operator()( const iRegister32& to, const iRegister32& from, s32 imm ) const{ iMUL32::Emit( to, from, imm ); }
__noinline void operator()( const iRegister32& to, const ModSibBase& src ) const { iMUL32::Emit( to, src ); }
__noinline void operator()( const iRegister32& to, const ModSibBase& from, s32 imm ) const { iMUL32::Emit( to, from, imm ); }
__forceinline void operator()( const iRegister16& to, const iRegister16& from ) const { iMUL16::Emit( to, from ); }
__forceinline void operator()( const iRegister16& to, const void* src ) const { iMUL16::Emit( to, src ); }
__forceinline void operator()( const iRegister16& to, const iRegister16& from, s16 imm ) const{ iMUL16::Emit( to, from, imm ); }
__noinline void operator()( const iRegister16& to, const ModSibBase& src ) const { iMUL16::Emit( to, src ); }
__noinline void operator()( const iRegister16& to, const ModSibBase& from, s16 imm ) const { iMUL16::Emit( to, from, imm ); }
__forceinline void operator()( const xRegister16& to, const xRegister16& from ) const { iMUL16::Emit( to, from ); }
__forceinline void operator()( const xRegister16& to, const void* src ) const { iMUL16::Emit( to, src ); }
__forceinline void operator()( const xRegister16& to, const xRegister16& from, s16 imm ) const{ iMUL16::Emit( to, from, imm ); }
__noinline void operator()( const xRegister16& to, const ModSibBase& src ) const { iMUL16::Emit( to, src ); }
__noinline void operator()( const xRegister16& to, const ModSibBase& from, s16 imm ) const { iMUL16::Emit( to, from, imm ); }
iMul_PlusSSE() {}
};

View File

@ -21,6 +21,8 @@
// Implementations found here: Increment and Decrement Instructions!
// Note: This header is meant to be included from within the x86Emitter::Internal namespace.
// ------------------------------------------------------------------------
//
template< typename ImmType >
class IncDecImpl
{
@ -28,12 +30,12 @@ protected:
static const uint OperandSize = sizeof(ImmType);
static bool Is8BitOperand() { return OperandSize == 1; }
static void prefix16() { if( OperandSize == 2 ) iWrite<u8>( 0x66 ); }
static void prefix16() { if( OperandSize == 2 ) xWrite<u8>( 0x66 ); }
public:
IncDecImpl() {} // For the love of GCC.
static __emitinline void Emit( bool isDec, const iRegister<ImmType>& to )
static __emitinline void Emit( bool isDec, const xRegister<ImmType>& to )
{
// There is no valid 8-bit form of direct register inc/dec, so fall
// back on Mod/RM format instead:
@ -67,13 +69,13 @@ protected:
typedef IncDecImpl<u8> m_8;
public:
__forceinline void operator()( const iRegister32& to ) const { m_32::Emit( isDec, to ); }
__forceinline void operator()( const xRegister32& to ) const { m_32::Emit( isDec, to ); }
__noinline void operator()( const ModSibStrict<u32>& sibdest ) const{ m_32::Emit( isDec, sibdest ); }
__forceinline void operator()( const iRegister16& to ) const { m_16::Emit( isDec, to ); }
__forceinline void operator()( const xRegister16& to ) const { m_16::Emit( isDec, to ); }
__noinline void operator()( const ModSibStrict<u16>& sibdest ) const{ m_16::Emit( isDec, sibdest ); }
__forceinline void operator()( const iRegister8& to ) const { m_8::Emit( isDec, to ); }
__forceinline void operator()( const xRegister8& to ) const { m_8::Emit( isDec, to ); }
__noinline void operator()( const ModSibStrict<u8>& sibdest ) const { m_8::Emit( isDec, sibdest ); }
IncDecImplAll() {} // don't ask.

View File

@ -27,22 +27,22 @@ class JmpCallImpl
protected:
static const uint OperandSize = sizeof(ImmType);
static void prefix16() { if( OperandSize == 2 ) iWrite<u8>( 0x66 ); }
static void prefix16() { if( OperandSize == 2 ) xWrite<u8>( 0x66 ); }
public:
JmpCallImpl() {} // For the love of GCC.
static __emitinline void Emit( bool isJmp, const iRegister<ImmType>& absreg )
static __emitinline void Emit( bool isJmp, const xRegister<ImmType>& absreg )
{
prefix16();
iWrite<u8>( 0xff );
xWrite<u8>( 0xff );
ModRM_Direct( isJmp ? 4 : 2, absreg.Id );
}
static __emitinline void Emit( bool isJmp, const ModSibStrict<ImmType>& src )
{
prefix16();
iWrite<u8>( 0xff );
xWrite<u8>( 0xff );
EmitSibMagic( isJmp ? 4 : 2, src );
}
};
@ -58,10 +58,10 @@ protected:
public:
JmpCallImplAll() {}
__forceinline void operator()( const iRegister32& absreg ) const { m_32::Emit( isJmp, absreg ); }
__forceinline void operator()( const xRegister32& absreg ) const { m_32::Emit( isJmp, absreg ); }
__forceinline void operator()( const ModSibStrict<u32>& src ) const { m_32::Emit( isJmp, src ); }
__forceinline void operator()( const iRegister16& absreg ) const { m_16::Emit( isJmp, absreg ); }
__forceinline void operator()( const xRegister16& absreg ) const { m_16::Emit( isJmp, absreg ); }
__forceinline void operator()( const ModSibStrict<u16>& src ) const { m_16::Emit( isJmp, src ); }
// Special form for calling functions. This form automatically resolves the
@ -77,8 +77,8 @@ public:
// always 5 bytes (16 bit calls are bad mojo, so no bother to do special logic).
sptr dest = (sptr)func - ((sptr)iGetPtr() + 5);
iWrite<u8>( 0xe8 );
iWrite<u32>( dest );
xWrite<u8>( 0xe8 );
xWrite<u32>( dest );
}
}

View File

@ -32,23 +32,23 @@ class MovImpl
protected:
static const uint OperandSize = sizeof(ImmType);
static bool Is8BitOperand() { return OperandSize == 1; }
static void prefix16() { if( OperandSize == 2 ) iWrite<u8>( 0x66 ); }
static void prefix16() { if( OperandSize == 2 ) xWrite<u8>( 0x66 ); }
public:
MovImpl() {}
// ------------------------------------------------------------------------
static __emitinline void Emit( const iRegister<ImmType>& to, const iRegister<ImmType>& from )
static __emitinline void Emit( const xRegister<ImmType>& to, const xRegister<ImmType>& from )
{
if( to == from ) return; // ignore redundant MOVs.
prefix16();
iWrite<u8>( Is8BitOperand() ? 0x88 : 0x89 );
xWrite<u8>( Is8BitOperand() ? 0x88 : 0x89 );
ModRM_Direct( from.Id, to.Id );
}
// ------------------------------------------------------------------------
static __emitinline void Emit( const ModSibBase& dest, const iRegister<ImmType>& from )
static __emitinline void Emit( const ModSibBase& dest, const xRegister<ImmType>& from )
{
prefix16();
@ -57,18 +57,18 @@ public:
if( from.IsAccumulator() && dest.Index.IsEmpty() && dest.Base.IsEmpty() )
{
iWrite<u8>( Is8BitOperand() ? 0xa2 : 0xa3 );
iWrite<u32>( dest.Displacement );
xWrite<u8>( Is8BitOperand() ? 0xa2 : 0xa3 );
xWrite<u32>( dest.Displacement );
}
else
{
iWrite<u8>( Is8BitOperand() ? 0x88 : 0x89 );
xWrite<u8>( Is8BitOperand() ? 0x88 : 0x89 );
EmitSibMagic( from.Id, dest );
}
}
// ------------------------------------------------------------------------
static __emitinline void Emit( const iRegister<ImmType>& to, const ModSibBase& src )
static __emitinline void Emit( const xRegister<ImmType>& to, const ModSibBase& src )
{
prefix16();
@ -77,18 +77,18 @@ public:
if( to.IsAccumulator() && src.Index.IsEmpty() && src.Base.IsEmpty() )
{
iWrite<u8>( Is8BitOperand() ? 0xa0 : 0xa1 );
iWrite<u32>( src.Displacement );
xWrite<u8>( Is8BitOperand() ? 0xa0 : 0xa1 );
xWrite<u32>( src.Displacement );
}
else
{
iWrite<u8>( Is8BitOperand() ? 0x8a : 0x8b );
xWrite<u8>( Is8BitOperand() ? 0x8a : 0x8b );
EmitSibMagic( to.Id, src );
}
}
// ------------------------------------------------------------------------
static __emitinline void Emit( void* dest, const iRegister<ImmType>& from )
static __emitinline void Emit( void* dest, const xRegister<ImmType>& from )
{
prefix16();
@ -96,18 +96,18 @@ public:
if( from.IsAccumulator() )
{
iWrite<u8>( Is8BitOperand() ? 0xa2 : 0xa3 );
iWrite<s32>( (s32)dest );
xWrite<u8>( Is8BitOperand() ? 0xa2 : 0xa3 );
xWrite<s32>( (s32)dest );
}
else
{
iWrite<u8>( Is8BitOperand() ? 0x88 : 0x89 );
iWriteDisp( from.Id, dest );
xWrite<u8>( Is8BitOperand() ? 0x88 : 0x89 );
xWriteDisp( from.Id, dest );
}
}
// ------------------------------------------------------------------------
static __emitinline void Emit( const iRegister<ImmType>& to, const void* src )
static __emitinline void Emit( const xRegister<ImmType>& to, const void* src )
{
prefix16();
@ -115,33 +115,33 @@ public:
if( to.IsAccumulator() )
{
iWrite<u8>( Is8BitOperand() ? 0xa0 : 0xa1 );
iWrite<s32>( (s32)src );
xWrite<u8>( Is8BitOperand() ? 0xa0 : 0xa1 );
xWrite<s32>( (s32)src );
}
else
{
iWrite<u8>( Is8BitOperand() ? 0x8a : 0x8b );
iWriteDisp( to.Id, src );
xWrite<u8>( Is8BitOperand() ? 0x8a : 0x8b );
xWriteDisp( to.Id, src );
}
}
// ------------------------------------------------------------------------
static __emitinline void Emit( const iRegister<ImmType>& to, ImmType imm )
static __emitinline void Emit( const xRegister<ImmType>& to, ImmType imm )
{
// Note: MOV does not have (reg16/32,imm8) forms.
prefix16();
iWrite<u8>( (Is8BitOperand() ? 0xb0 : 0xb8) | to.Id );
iWrite<ImmType>( imm );
xWrite<u8>( (Is8BitOperand() ? 0xb0 : 0xb8) | to.Id );
xWrite<ImmType>( imm );
}
// ------------------------------------------------------------------------
static __emitinline void Emit( ModSibStrict<ImmType> dest, ImmType imm )
{
prefix16();
iWrite<u8>( Is8BitOperand() ? 0xc6 : 0xc7 );
xWrite<u8>( Is8BitOperand() ? 0xc6 : 0xc7 );
EmitSibMagic( 0, dest );
iWrite<ImmType>( imm );
xWrite<ImmType>( imm );
}
};
@ -150,15 +150,15 @@ class MovImplAll
{
public:
template< typename T >
__forceinline void operator()( const iRegister<T>& to, const iRegister<T>& from ) const { MovImpl<T>::Emit( to, from ); }
__forceinline void operator()( const xRegister<T>& to, const xRegister<T>& from ) const { MovImpl<T>::Emit( to, from ); }
template< typename T >
__forceinline void operator()( const iRegister<T>& to, const void* src ) const { MovImpl<T>::Emit( to, src ); }
__forceinline void operator()( const xRegister<T>& to, const void* src ) const { MovImpl<T>::Emit( to, src ); }
template< typename T >
__forceinline void operator()( void* dest, const iRegister<T>& from ) const { MovImpl<T>::Emit( dest, from ); }
__forceinline void operator()( void* dest, const xRegister<T>& from ) const { MovImpl<T>::Emit( dest, from ); }
template< typename T >
__noinline void operator()( const ModSibBase& sibdest, const iRegister<T>& from ) const { MovImpl<T>::Emit( sibdest, from ); }
__noinline void operator()( const ModSibBase& sibdest, const xRegister<T>& from ) const { MovImpl<T>::Emit( sibdest, from ); }
template< typename T >
__noinline void operator()( const iRegister<T>& to, const ModSibBase& sibsrc ) const { MovImpl<T>::Emit( to, sibsrc ); }
__noinline void operator()( const xRegister<T>& to, const ModSibBase& sibsrc ) const { MovImpl<T>::Emit( to, sibsrc ); }
template< typename T >
__noinline void operator()( const ModSibStrict<T>& sibdest, int imm ) const { MovImpl<T>::Emit( sibdest, imm ); }
@ -167,10 +167,10 @@ public:
// the flags (namely replacing mov reg,0 with xor).
template< typename T >
__emitinline void operator()( const iRegister<T>& to, int imm, bool preserve_flags=false ) const
__emitinline void operator()( const xRegister<T>& to, int imm, bool preserve_flags=false ) const
{
if( !preserve_flags && (imm == 0) )
iXOR( to, to );
xXOR( to, to );
else
MovImpl<T>::Emit( to, imm );
}
@ -193,7 +193,7 @@ protected:
static const uint OperandSize = sizeof(ImmType);
static bool Is8BitOperand() { return OperandSize == 1; }
static void prefix16() { if( OperandSize == 2 ) iWrite<u8>( 0x66 ); }
static void prefix16() { if( OperandSize == 2 ) xWrite<u8>( 0x66 ); }
static __forceinline void emit_base( JccComparisonType cc )
{
@ -206,27 +206,27 @@ protected:
public:
CMovSetImpl() {}
static __emitinline void Emit( JccComparisonType cc, const iRegister<ImmType>& to, const iRegister<ImmType>& from )
static __emitinline void Emit( JccComparisonType cc, const xRegister<ImmType>& to, const xRegister<ImmType>& from )
{
if( to == from ) return;
emit_base( cc );
ModRM_Direct( to.Id, from.Id );
}
static __emitinline void Emit( JccComparisonType cc, const iRegister<ImmType>& to, const void* src )
static __emitinline void Emit( JccComparisonType cc, const xRegister<ImmType>& to, const void* src )
{
emit_base( cc );
iWriteDisp( to.Id, src );
xWriteDisp( to.Id, src );
}
static __emitinline void Emit( JccComparisonType cc, const iRegister<ImmType>& to, const ModSibBase& sibsrc )
static __emitinline void Emit( JccComparisonType cc, const xRegister<ImmType>& to, const ModSibBase& sibsrc )
{
emit_base( cc );
EmitSibMagic( to.Id, sibsrc );
}
// This form is provided for SETcc only (not available in CMOV)
static __emitinline void EmitSet( JccComparisonType cc, const iRegister<ImmType>& to )
static __emitinline void EmitSet( JccComparisonType cc, const xRegister<ImmType>& to )
{
emit_base( cc );
ModRM_Direct( 0, to.Id );
@ -236,7 +236,7 @@ public:
static __emitinline void EmitSet( JccComparisonType cc, const void* src )
{
emit_base( cc );
iWriteDisp( 0, src );
xWriteDisp( 0, src );
}
// This form is provided for SETcc only (not available in CMOV)
@ -258,13 +258,13 @@ protected:
typedef CMovSetImpl<u16, 0x40> m_16; // 0x40 is the cmov base instruction id
public:
__forceinline void operator()( JccComparisonType ccType, const iRegister32& to, const iRegister32& from ) const { m_32::Emit( ccType, to, from ); }
__forceinline void operator()( JccComparisonType ccType, const iRegister32& to, const void* src ) const { m_32::Emit( ccType, to, src ); }
__noinline void operator()( JccComparisonType ccType, const iRegister32& to, const ModSibBase& sibsrc ) const { m_32::Emit( ccType, to, sibsrc ); }
__forceinline void operator()( JccComparisonType ccType, const xRegister32& to, const xRegister32& from ) const { m_32::Emit( ccType, to, from ); }
__forceinline void operator()( JccComparisonType ccType, const xRegister32& to, const void* src ) const { m_32::Emit( ccType, to, src ); }
__noinline void operator()( JccComparisonType ccType, const xRegister32& to, const ModSibBase& sibsrc ) const { m_32::Emit( ccType, to, sibsrc ); }
__forceinline void operator()( JccComparisonType ccType, const iRegister16& to, const iRegister16& from ) const { m_16::Emit( ccType, to, from ); }
__forceinline void operator()( JccComparisonType ccType, const iRegister16& to, const void* src ) const { m_16::Emit( ccType, to, src ); }
__noinline void operator()( JccComparisonType ccType, const iRegister16& to, const ModSibBase& sibsrc ) const { m_16::Emit( ccType, to, sibsrc ); }
__forceinline void operator()( JccComparisonType ccType, const xRegister16& to, const xRegister16& from ) const { m_16::Emit( ccType, to, from ); }
__forceinline void operator()( JccComparisonType ccType, const xRegister16& to, const void* src ) const { m_16::Emit( ccType, to, src ); }
__noinline void operator()( JccComparisonType ccType, const xRegister16& to, const ModSibBase& sibsrc ) const { m_16::Emit( ccType, to, sibsrc ); }
CMovImplGeneric() {} // don't ask.
};
@ -278,13 +278,13 @@ protected:
typedef CMovSetImpl<u16, 0x40> m_16;
public:
__forceinline void operator()( const iRegister32& to, const iRegister32& from ) const { m_32::Emit( ccType, to, from ); }
__forceinline void operator()( const iRegister32& to, const void* src ) const { m_32::Emit( ccType, to, src ); }
__noinline void operator()( const iRegister32& to, const ModSibBase& sibsrc ) const { m_32::Emit( ccType, to, sibsrc ); }
__forceinline void operator()( const xRegister32& to, const xRegister32& from ) const { m_32::Emit( ccType, to, from ); }
__forceinline void operator()( const xRegister32& to, const void* src ) const { m_32::Emit( ccType, to, src ); }
__noinline void operator()( const xRegister32& to, const ModSibBase& sibsrc ) const { m_32::Emit( ccType, to, sibsrc ); }
__forceinline void operator()( const iRegister16& to, const iRegister16& from ) const { m_16::Emit( ccType, to, from ); }
__forceinline void operator()( const iRegister16& to, const void* src ) const { m_16::Emit( ccType, to, src ); }
__noinline void operator()( const iRegister16& to, const ModSibBase& sibsrc ) const { m_16::Emit( ccType, to, sibsrc ); }
__forceinline void operator()( const xRegister16& to, const xRegister16& from ) const { m_16::Emit( ccType, to, from ); }
__forceinline void operator()( const xRegister16& to, const void* src ) const { m_16::Emit( ccType, to, src ); }
__noinline void operator()( const xRegister16& to, const ModSibBase& sibsrc ) const { m_16::Emit( ccType, to, sibsrc ); }
CMovImplAll() {} // don't ask.
};
@ -296,7 +296,7 @@ protected:
typedef CMovSetImpl<u8, 0x90> Impl; // 0x90 is the SETcc base instruction id
public:
__forceinline void operator()( JccComparisonType cc, const iRegister8& to ) const { Impl::EmitSet( cc, to ); }
__forceinline void operator()( JccComparisonType cc, const xRegister8& to ) const { Impl::EmitSet( cc, to ); }
__forceinline void operator()( JccComparisonType cc, void* dest ) const { Impl::EmitSet( cc, dest ); }
__noinline void operator()( JccComparisonType cc, const ModSibStrict<u8>& dest ) const { Impl::EmitSet( cc, dest ); }
@ -311,7 +311,7 @@ protected:
typedef CMovSetImpl<u8, 0x90> Impl; // 0x90 is the SETcc base instruction id
public:
__forceinline void operator()( const iRegister8& to ) const { Impl::EmitSet( ccType, to ); }
__forceinline void operator()( const xRegister8& to ) const { Impl::EmitSet( ccType, to ); }
__forceinline void operator()( void* dest ) const { Impl::EmitSet( ccType, dest ); }
__noinline void operator()( const ModSibStrict<u8>& dest ) const { Impl::EmitSet( ccType, dest ); }
@ -330,24 +330,24 @@ protected:
static const uint SrcOperandSize = sizeof( SrcImmType );
static bool Is8BitOperand() { return SrcOperandSize == 1; }
static void prefix16() { if( DestOperandSize == 2 ) iWrite<u8>( 0x66 ); }
static void prefix16() { if( DestOperandSize == 2 ) xWrite<u8>( 0x66 ); }
static __forceinline void emit_base( bool SignExtend )
{
prefix16();
iWrite<u8>( 0x0f );
iWrite<u8>( 0xb6 | (Is8BitOperand() ? 0 : 1) | (SignExtend ? 8 : 0 ) );
xWrite<u8>( 0x0f );
xWrite<u8>( 0xb6 | (Is8BitOperand() ? 0 : 1) | (SignExtend ? 8 : 0 ) );
}
public:
MovExtendImpl() {} // For the love of GCC.
static __emitinline void Emit( const iRegister<DestImmType>& to, const iRegister<SrcImmType>& from, bool SignExtend )
static __emitinline void Emit( const xRegister<DestImmType>& to, const xRegister<SrcImmType>& from, bool SignExtend )
{
emit_base( SignExtend );
ModRM_Direct( to.Id, from.Id );
}
static __emitinline void Emit( const iRegister<DestImmType>& to, const ModSibStrict<SrcImmType>& sibsrc, bool SignExtend )
static __emitinline void Emit( const xRegister<DestImmType>& to, const ModSibStrict<SrcImmType>& sibsrc, bool SignExtend )
{
emit_base( SignExtend );
EmitSibMagic( to.Id, sibsrc );
@ -364,14 +364,14 @@ protected:
typedef MovExtendImpl<u16, u8> m_8to16;
public:
__forceinline void operator()( const iRegister32& to, const iRegister16& from ) const { m_16to32::Emit( to, from, SignExtend ); }
__noinline void operator()( const iRegister32& to, const ModSibStrict<u16>& sibsrc ) const { m_16to32::Emit( to, sibsrc, SignExtend ); }
__forceinline void operator()( const xRegister32& to, const xRegister16& from ) const { m_16to32::Emit( to, from, SignExtend ); }
__noinline void operator()( const xRegister32& to, const ModSibStrict<u16>& sibsrc ) const { m_16to32::Emit( to, sibsrc, SignExtend ); }
__forceinline void operator()( const iRegister32& to, const iRegister8& from ) const { m_8to32::Emit( to, from, SignExtend ); }
__noinline void operator()( const iRegister32& to, const ModSibStrict<u8>& sibsrc ) const { m_8to32::Emit( to, sibsrc, SignExtend ); }
__forceinline void operator()( const xRegister32& to, const xRegister8& from ) const { m_8to32::Emit( to, from, SignExtend ); }
__noinline void operator()( const xRegister32& to, const ModSibStrict<u8>& sibsrc ) const { m_8to32::Emit( to, sibsrc, SignExtend ); }
__forceinline void operator()( const iRegister16& to, const iRegister8& from ) const { m_8to16::Emit( to, from, SignExtend ); }
__noinline void operator()( const iRegister16& to, const ModSibStrict<u8>& sibsrc ) const { m_8to16::Emit( to, sibsrc, SignExtend ); }
__forceinline void operator()( const xRegister16& to, const xRegister8& from ) const { m_8to16::Emit( to, from, SignExtend ); }
__noinline void operator()( const xRegister16& to, const ModSibStrict<u8>& sibsrc ) const { m_8to16::Emit( to, sibsrc, SignExtend ); }
MovExtendImplAll() {} // don't ask.
};

View File

@ -27,41 +27,41 @@ class TestImpl
protected:
static const uint OperandSize = sizeof(ImmType);
static bool Is8BitOperand() { return OperandSize == 1; }
static void prefix16() { if( OperandSize == 2 ) iWrite<u8>( 0x66 ); }
static void prefix16() { if( OperandSize == 2 ) xWrite<u8>( 0x66 ); }
public:
TestImpl() {}
// ------------------------------------------------------------------------
static __emitinline void Emit( const iRegister<ImmType>& to, const iRegister<ImmType>& from )
static __emitinline void Emit( const xRegister<ImmType>& to, const xRegister<ImmType>& from )
{
prefix16();
iWrite<u8>( Is8BitOperand() ? 0x84 : 0x85 );
xWrite<u8>( Is8BitOperand() ? 0x84 : 0x85 );
ModRM_Direct( from.Id, to.Id );
}
// ------------------------------------------------------------------------
static __emitinline void Emit( const iRegister<ImmType>& to, ImmType imm )
static __emitinline void Emit( const xRegister<ImmType>& to, ImmType imm )
{
prefix16();
if( to.IsAccumulator() )
iWrite<u8>( Is8BitOperand() ? 0xa8 : 0xa9 );
xWrite<u8>( Is8BitOperand() ? 0xa8 : 0xa9 );
else
{
iWrite<u8>( Is8BitOperand() ? 0xf6 : 0xf7 );
xWrite<u8>( Is8BitOperand() ? 0xf6 : 0xf7 );
ModRM_Direct( 0, to.Id );
}
iWrite<ImmType>( imm );
xWrite<ImmType>( imm );
}
// ------------------------------------------------------------------------
static __emitinline void Emit( ModSibStrict<ImmType> dest, ImmType imm )
{
prefix16();
iWrite<u8>( Is8BitOperand() ? 0xf6 : 0xf7 );
xWrite<u8>( Is8BitOperand() ? 0xf6 : 0xf7 );
EmitSibMagic( 0, dest );
iWrite<ImmType>( imm );
xWrite<ImmType>( imm );
}
};
@ -71,12 +71,12 @@ class TestImplAll
{
public:
template< typename T >
__forceinline void operator()( const iRegister<T>& to, const iRegister<T>& from ) const { TestImpl<T>::Emit( to, from ); }
__forceinline void operator()( const xRegister<T>& to, const xRegister<T>& from ) const { TestImpl<T>::Emit( to, from ); }
template< typename T >
__noinline void operator()( const ModSibStrict<T>& sibdest, T imm ) const { TestImpl<T>::Emit( sibdest, imm ); }
template< typename T >
void operator()( const iRegister<T>& to, T imm ) const { TestImpl<T>::Emit( to, imm ); }
void operator()( const xRegister<T>& to, T imm ) const { TestImpl<T>::Emit( to, imm ); }
TestImplAll() {} // Why does GCC need these?
};

View File

@ -26,11 +26,11 @@ __emitinline void SimdPrefix( u8 opcode, u8 prefix=0 )
{
if( sizeof( T ) == 16 && prefix != 0 )
{
iWrite<u16>( 0x0f00 | prefix );
iWrite<u8>( opcode );
xWrite<u16>( 0x0f00 | prefix );
xWrite<u8>( opcode );
}
else
iWrite<u16>( (opcode<<8) | 0x0f );
xWrite<u16>( (opcode<<8) | 0x0f );
}
// ------------------------------------------------------------------------
@ -40,24 +40,24 @@ __emitinline void SimdPrefix( u8 opcode, u8 prefix=0 )
// instructions violate this "guideline.")
//
template< typename T, typename T2 >
__emitinline void writeXMMop( u8 prefix, u8 opcode, const iRegister<T>& to, const iRegister<T2>& from )
__emitinline void writeXMMop( u8 prefix, u8 opcode, const xRegister<T>& to, const xRegister<T2>& from )
{
SimdPrefix<T>( opcode, prefix );
ModRM_Direct( to.Id, from.Id );
}
template< typename T >
void writeXMMop( u8 prefix, u8 opcode, const iRegister<T>& reg, const ModSibBase& sib )
void writeXMMop( u8 prefix, u8 opcode, const xRegister<T>& reg, const ModSibBase& sib )
{
SimdPrefix<T>( opcode, prefix );
EmitSibMagic( reg.Id, sib );
}
template< typename T >
__emitinline void writeXMMop( u8 prefix, u8 opcode, const iRegister<T>& reg, const void* data )
__emitinline void writeXMMop( u8 prefix, u8 opcode, const xRegister<T>& reg, const void* data )
{
SimdPrefix<T>( opcode, prefix );
iWriteDisp( reg.Id, data );
xWriteDisp( reg.Id, data );
}
// ------------------------------------------------------------------------
@ -66,51 +66,74 @@ __emitinline void writeXMMop( u8 prefix, u8 opcode, const iRegister<T>& reg, con
// some special forms of sse/xmm mov instructions also use them due to prefixing inconsistencies.
//
template< typename T, typename T2 >
__emitinline void writeXMMop( u8 opcode, const iRegister<T>& to, const iRegister<T2>& from )
__emitinline void writeXMMop( u8 opcode, const xRegister<T>& to, const xRegister<T2>& from )
{
SimdPrefix<T>( opcode );
ModRM_Direct( to.Id, from.Id );
}
template< typename T >
void writeXMMop( u8 opcode, const iRegister<T>& reg, const ModSibBase& sib )
void writeXMMop( u8 opcode, const xRegister<T>& reg, const ModSibBase& sib )
{
SimdPrefix<T>( opcode );
EmitSibMagic( reg.Id, sib );
}
template< typename T >
__emitinline void writeXMMop( u8 opcode, const iRegister<T>& reg, const void* data )
__emitinline void writeXMMop( u8 opcode, const xRegister<T>& reg, const void* data )
{
SimdPrefix<T>( opcode );
iWriteDisp( reg.Id, data );
xWriteDisp( reg.Id, data );
}
//////////////////////////////////////////////////////////////////////////////////////////
// Moves to/from high/low portions of an xmm register.
// These instructions cannot be used in reg/reg form.
//
template< u8 Prefix, u8 Opcode >
template< u8 Opcode >
class MovhlImplAll
{
protected:
template< u8 Prefix >
struct Woot
{
__forceinline void operator()( const xRegisterSSE& to, const void* from ) const { writeXMMop( Prefix, Opcode, to, from ); }
__forceinline void operator()( const void* to, const xRegisterSSE& from ) const { writeXMMop( Prefix, Opcode+1, from, to ); }
__noinline void operator()( const xRegisterSSE& to, const ModSibBase& from ) const { writeXMMop( Prefix, Opcode, to, from ); }
__noinline void operator()( const ModSibBase& to, const xRegisterSSE& from ) const { writeXMMop( Prefix, Opcode+1, from, to ); }
};
public:
__forceinline void operator()( const iRegisterSSE& to, const void* from ) const { writeXMMop( Prefix, Opcode, to, from ); }
__forceinline void operator()( const void* to, const iRegisterSSE& from ) const { writeXMMop( Prefix, Opcode+1, from, to ); }
__noinline void operator()( const iRegisterSSE& to, const ModSibBase& from ) const { writeXMMop( Prefix, Opcode, to, from ); }
__noinline void operator()( const ModSibBase& to, const iRegisterSSE& from ) const { writeXMMop( Prefix, Opcode+1, from, to ); }
Woot<0x00> PS;
Woot<0x66> PD;
MovhlImplAll() {} //GCC.
};
// ------------------------------------------------------------------------
// RegtoReg forms of MOVHL/MOVLH -- these are the same opcodes as MOVH/MOVL but
// do something kinda different! Fun!
//
template< u8 Opcode >
class MovhlImpl_RtoR
{
public:
__forceinline void PS( const xRegisterSSE& to, const xRegisterSSE& from ) const { writeXMMop( Opcode, to, from ); }
__forceinline void PD( const xRegisterSSE& to, const xRegisterSSE& from ) const { writeXMMop( 0x66, Opcode, to, from ); }
MovhlImpl_RtoR() {} //GCC.
};
// ------------------------------------------------------------------------
template< u8 Prefix, u8 Opcode, u8 OpcodeAlt >
class MovapsImplAll
{
public:
__forceinline void operator()( const iRegisterSSE& to, const iRegisterSSE& from ) const { if( to != from ) writeXMMop( Prefix, Opcode, to, from ); }
__forceinline void operator()( const iRegisterSSE& to, const void* from ) const { writeXMMop( Prefix, Opcode, to, from ); }
__forceinline void operator()( const void* to, const iRegisterSSE& from ) const { writeXMMop( Prefix, OpcodeAlt, from, to ); }
__noinline void operator()( const iRegisterSSE& to, const ModSibBase& from ) const { writeXMMop( Prefix, Opcode, to, from ); }
__noinline void operator()( const ModSibBase& to, const iRegisterSSE& from ) const { writeXMMop( Prefix, OpcodeAlt, from, to ); }
__forceinline void operator()( const xRegisterSSE& to, const xRegisterSSE& from ) const { if( to != from ) writeXMMop( Prefix, Opcode, to, from ); }
__forceinline void operator()( const xRegisterSSE& to, const void* from ) const { writeXMMop( Prefix, Opcode, to, from ); }
__forceinline void operator()( const void* to, const xRegisterSSE& from ) const { writeXMMop( Prefix, OpcodeAlt, from, to ); }
__noinline void operator()( const xRegisterSSE& to, const ModSibBase& from ) const { writeXMMop( Prefix, Opcode, to, from ); }
__noinline void operator()( const ModSibBase& to, const xRegisterSSE& from ) const { writeXMMop( Prefix, OpcodeAlt, from, to ); }
MovapsImplAll() {} //GCC.
};
@ -124,11 +147,11 @@ class PLogicImplAll
{
public:
template< typename T >
__forceinline void operator()( const iRegisterSIMD<T>& to, const iRegisterSIMD<T>& from ) const { writeXMMop( 0x66, Opcode, to, from ); }
__forceinline void operator()( const xRegisterSIMD<T>& to, const xRegisterSIMD<T>& from ) const { writeXMMop( 0x66, Opcode, to, from ); }
template< typename T >
__forceinline void operator()( const iRegisterSIMD<T>& to, const void* from ) const { writeXMMop( 0x66, Opcode, to, from ); }
__forceinline void operator()( const xRegisterSIMD<T>& to, const void* from ) const { writeXMMop( 0x66, Opcode, to, from ); }
template< typename T >
__noinline void operator()( const iRegisterSIMD<T>& to, const ModSibBase& from ) const { writeXMMop( 0x66, Opcode, to, from ); }
__noinline void operator()( const xRegisterSIMD<T>& to, const ModSibBase& from ) const { writeXMMop( 0x66, Opcode, to, from ); }
PLogicImplAll() {} //GCWho?
};
@ -140,47 +163,42 @@ template< u8 Prefix, u8 Opcode >
class SSELogicImpl
{
public:
__forceinline void operator()( const iRegisterSSE& to, const iRegisterSSE& from ) const { writeXMMop( Prefix, Opcode, to, from ); }
__forceinline void operator()( const iRegisterSSE& to, const void* from ) const { writeXMMop( Prefix, Opcode, to, from ); }
__noinline void operator()( const iRegisterSSE& to, const ModSibBase& from ) const { writeXMMop( Prefix, Opcode, to, from ); }
__forceinline void operator()( const xRegisterSSE& to, const xRegisterSSE& from ) const { writeXMMop( Prefix, Opcode, to, from ); }
__forceinline void operator()( const xRegisterSSE& to, const void* from ) const { writeXMMop( Prefix, Opcode, to, from ); }
__noinline void operator()( const xRegisterSSE& to, const ModSibBase& from ) const { writeXMMop( Prefix, Opcode, to, from ); }
SSELogicImpl() {} //GCWho?
};
// ------------------------------------------------------------------------
//
template< u8 OpcodeSSE >
class SSEAndNotImpl
{
public:
const SSELogicImpl<0x00,OpcodeSSE> PS;
const SSELogicImpl<0x66,OpcodeSSE> PD;
SSEAndNotImpl() {}
};
// ------------------------------------------------------------------------
// For implementing SSE-only comparison operations, like CMPEQPS.
//
enum SSE2_ComparisonType
{
SSE2_Equal = 0,
SSE2_Less,
SSE2_LessOrEqual,
SSE2_Unordered,
SSE2_NotEqual,
SSE2_NotLess,
SSE2_NotLessOrEqual,
SSE2_Ordered
};
template< u8 Prefix >
class SSECompareImplGeneric
{
public:
__forceinline void operator()( const iRegisterSSE& to, const iRegisterSSE& from, u8 cmptype ) const { writeXMMop( Prefix, 0xc2, to, from ); iWrite( cmptype ); }
__forceinline void operator()( const iRegisterSSE& to, const void* from, u8 cmptype ) const { writeXMMop( Prefix, 0xc2, to, from ); iWrite( cmptype ); }
__noinline void operator()( const iRegisterSSE& to, const ModSibBase& from, u8 cmptype ) const { writeXMMop( Prefix, 0xc2, to, from ); iWrite( cmptype ); }
SSECompareImplGeneric() {} //GCWhat?
};
template< u8 Prefix, u8 Opcode, SSE2_ComparisonType CType >
template< SSE2_ComparisonType CType >
class SSECompareImpl
{
protected:
template< u8 Prefix > struct Woot
{
__forceinline void operator()( const xRegisterSSE& to, const xRegisterSSE& from ) const { writeXMMop( Prefix, 0xc2, to, from ); xWrite<u8>( CType ); }
__forceinline void operator()( const xRegisterSSE& to, const void* from ) const { writeXMMop( Prefix, 0xc2, to, from ); xWrite<u8>( CType ); }
__noinline void operator()( const xRegisterSSE& to, const ModSibBase& from ) const { writeXMMop( Prefix, 0xc2, to, from ); xWrite<u8>( CType ); }
};
public:
__forceinline void operator()( const iRegisterSSE& to, const iRegisterSSE& from ) const { writeXMMop( Prefix, 0xc2, to, from ); iWrite( CType ); }
__forceinline void operator()( const iRegisterSSE& to, const void* from ) const { writeXMMop( Prefix, 0xc2, to, from ); iWrite( CType ); }
__noinline void operator()( const iRegisterSSE& to, const ModSibBase& from ) const { writeXMMop( Prefix, 0xc2, to, from ); iWrite( CType ); }
Woot<0x00> PS;
Woot<0x66> PD;
Woot<0xf3> SS;
Woot<0xf2> SD;
SSECompareImpl() {} //GCWhat?
};

View File

@ -66,54 +66,54 @@ __threadlocal XMMSSEType g_xmmtypes[iREGCNT_XMM] = { XMMT_INT };
namespace x86Emitter {
const iAddressIndexerBase ptr;
const iAddressIndexer<u128> ptr128;
const iAddressIndexer<u64> ptr64;
const iAddressIndexer<u32> ptr32;
const iAddressIndexer<u16> ptr16;
const iAddressIndexer<u8> ptr8;
const xAddressIndexerBase ptr;
const xAddressIndexer<u128> ptr128;
const xAddressIndexer<u64> ptr64;
const xAddressIndexer<u32> ptr32;
const xAddressIndexer<u16> ptr16;
const xAddressIndexer<u8> ptr8;
// ------------------------------------------------------------------------
template< typename OperandType > const iRegister<OperandType> iRegister<OperandType>::Empty;
const iAddressReg iAddressReg::Empty;
template< typename OperandType > const xRegister<OperandType> xRegister<OperandType>::Empty;
const xAddressReg xAddressReg::Empty;
const iRegisterSSE
const xRegisterSSE
xmm0( 0 ), xmm1( 1 ),
xmm2( 2 ), xmm3( 3 ),
xmm4( 4 ), xmm5( 5 ),
xmm6( 6 ), xmm7( 7 );
const iRegisterMMX
const xRegisterMMX
mm0( 0 ), mm1( 1 ),
mm2( 2 ), mm3( 3 ),
mm4( 4 ), mm5( 5 ),
mm6( 6 ), mm7( 7 );
const iRegister32
const xRegister32
eax( 0 ), ebx( 3 ),
ecx( 1 ), edx( 2 ),
esi( 6 ), edi( 7 ),
ebp( 5 ), esp( 4 );
const iRegister16
const xRegister16
ax( 0 ), bx( 3 ),
cx( 1 ), dx( 2 ),
si( 6 ), di( 7 ),
bp( 5 ), sp( 4 );
const iRegister8
const xRegister8
al( 0 ),
dl( 2 ), bl( 3 ),
ah( 4 ), ch( 5 ),
dh( 6 ), bh( 7 );
const iRegisterCL cl;
const xRegisterCL cl;
namespace Internal
{
// Performance note: VC++ wants to use byte/word register form for the following
// ModRM/SibSB constructors when we use iWrite<u8>, and furthermore unrolls the
// ModRM/SibSB constructors when we use xWrite<u8>, and furthermore unrolls the
// the shift using a series of ADDs for the following results:
// add cl,cl
// add cl,cl
@ -137,7 +137,7 @@ namespace Internal
__forceinline void ModRM( uint mod, uint reg, uint rm )
{
iWrite<u8>( (mod << 6) | (reg << 3) | rm );
xWrite<u8>( (mod << 6) | (reg << 3) | rm );
//*(u32*)x86Ptr = (mod << 6) | (reg << 3) | rm;
//x86Ptr++;
}
@ -149,20 +149,20 @@ namespace Internal
__forceinline void SibSB( u32 ss, u32 index, u32 base )
{
iWrite<u8>( (ss << 6) | (index << 3) | base );
xWrite<u8>( (ss << 6) | (index << 3) | base );
//*(u32*)x86Ptr = (ss << 6) | (index << 3) | base;
//x86Ptr++;
}
__forceinline void iWriteDisp( int regfield, s32 displacement )
__forceinline void xWriteDisp( int regfield, s32 displacement )
{
ModRM( 0, regfield, ModRm_UseDisp32 );
iWrite<s32>( displacement );
xWrite<s32>( displacement );
}
__forceinline void iWriteDisp( int regfield, const void* address )
__forceinline void xWriteDisp( int regfield, const void* address )
{
iWriteDisp( regfield, (s32)address );
xWriteDisp( regfield, (s32)address );
}
// ------------------------------------------------------------------------
@ -206,7 +206,7 @@ namespace Internal
if( info.Index.IsEmpty() )
{
iWriteDisp( regfield, info.Displacement );
xWriteDisp( regfield, info.Displacement );
return;
}
else
@ -229,7 +229,7 @@ namespace Internal
{
ModRM( 0, regfield, ModRm_UseSib );
SibSB( info.Scale, info.Index.Id, ModRm_UseDisp32 );
iWrite<s32>( info.Displacement );
xWrite<s32>( info.Displacement );
return;
}
else
@ -245,116 +245,116 @@ namespace Internal
if( displacement_size != 0 )
{
if( displacement_size == 1 )
iWrite<s8>( info.Displacement );
xWrite<s8>( info.Displacement );
else
iWrite<s32>( info.Displacement );
xWrite<s32>( info.Displacement );
}
}
}
using namespace Internal;
const MovImplAll iMOV;
const TestImplAll iTEST;
const MovImplAll xMOV;
const TestImplAll xTEST;
const G1LogicImpl<G1Type_AND,0x54> iAND;
const G1LogicImpl<G1Type_OR,0x56> iOR;
const G1LogicImpl<G1Type_XOR,0x57> iXOR;
const G1LogicImpl_PlusSSE<G1Type_AND,0x54> xAND;
const G1LogicImpl_PlusSSE<G1Type_OR,0x56> xOR;
const G1LogicImpl_PlusSSE<G1Type_XOR,0x57> xXOR;
const G1ArithmeticImpl<G1Type_ADD,0x58> iADD;
const G1ArithmeticImpl<G1Type_SUB,0x5c> iSUB;
const G1ArithmeticImpl_PlusSSE<G1Type_ADD,0x58> xADD;
const G1ArithmeticImpl_PlusSSE<G1Type_SUB,0x5c> xSUB;
const Group1ImplAll<G1Type_ADC> iADC;
const Group1ImplAll<G1Type_SBB> iSBB;
const Group1ImplAll<G1Type_CMP> iCMP;
const Group1ImplAll<G1Type_ADC> xADC;
const Group1ImplAll<G1Type_SBB> xSBB;
const G1CompareImpl_PlusSSE xCMP;
const Group2ImplAll<G2Type_ROL> iROL;
const Group2ImplAll<G2Type_ROR> iROR;
const Group2ImplAll<G2Type_RCL> iRCL;
const Group2ImplAll<G2Type_RCR> iRCR;
const Group2ImplAll<G2Type_SHL> iSHL;
const Group2ImplAll<G2Type_SHR> iSHR;
const Group2ImplAll<G2Type_SAR> iSAR;
const Group2ImplAll<G2Type_ROL> xROL;
const Group2ImplAll<G2Type_ROR> xROR;
const Group2ImplAll<G2Type_RCL> xRCL;
const Group2ImplAll<G2Type_RCR> xRCR;
const Group2ImplAll<G2Type_SHL> xSHL;
const Group2ImplAll<G2Type_SHR> xSHR;
const Group2ImplAll<G2Type_SAR> xSAR;
const Group3ImplAll<G3Type_NOT> iNOT;
const Group3ImplAll<G3Type_NEG> iNEG;
const Group3ImplAll<G3Type_MUL> iUMUL;
const Group3ImplAll<G3Type_DIV> iUDIV;
const G3Impl_PlusSSE<G3Type_iDIV,0x5e> iDIV;
const iMul_PlusSSE iMUL;
const Group3ImplAll<G3Type_NOT> xNOT;
const Group3ImplAll<G3Type_NEG> xNEG;
const Group3ImplAll<G3Type_MUL> xUMUL;
const Group3ImplAll<G3Type_DIV> xUDIV;
const G3Impl_PlusSSE<G3Type_iDIV,0x5e> xDIV;
const iMul_PlusSSE xMUL;
const IncDecImplAll<false> iINC;
const IncDecImplAll<true> iDEC;
const IncDecImplAll<false> xINC;
const IncDecImplAll<true> xDEC;
const MovExtendImplAll<false> iMOVZX;
const MovExtendImplAll<true> iMOVSX;
const MovExtendImplAll<false> xMOVZX;
const MovExtendImplAll<true> xMOVSX;
const DwordShiftImplAll<false> iSHLD;
const DwordShiftImplAll<true> iSHRD;
const DwordShiftImplAll<false> xSHLD;
const DwordShiftImplAll<true> xSHRD;
const Group8ImplAll<G8Type_BT> iBT;
const Group8ImplAll<G8Type_BTR> iBTR;
const Group8ImplAll<G8Type_BTS> iBTS;
const Group8ImplAll<G8Type_BTC> iBTC;
const Group8ImplAll<G8Type_BT> xBT;
const Group8ImplAll<G8Type_BTR> xBTR;
const Group8ImplAll<G8Type_BTS> xBTS;
const Group8ImplAll<G8Type_BTC> xBTC;
const BitScanImplAll<false> iBSF;
const BitScanImplAll<true> iBSR;
const BitScanImplAll<false> xBSF;
const BitScanImplAll<true> xBSR;
// ------------------------------------------------------------------------
const CMovImplGeneric iCMOV;
const CMovImplGeneric xCMOV;
const CMovImplAll<Jcc_Above> iCMOVA;
const CMovImplAll<Jcc_AboveOrEqual> iCMOVAE;
const CMovImplAll<Jcc_Below> iCMOVB;
const CMovImplAll<Jcc_BelowOrEqual> iCMOVBE;
const CMovImplAll<Jcc_Above> xCMOVA;
const CMovImplAll<Jcc_AboveOrEqual> xCMOVAE;
const CMovImplAll<Jcc_Below> xCMOVB;
const CMovImplAll<Jcc_BelowOrEqual> xCMOVBE;
const CMovImplAll<Jcc_Greater> iCMOVG;
const CMovImplAll<Jcc_GreaterOrEqual> iCMOVGE;
const CMovImplAll<Jcc_Less> iCMOVL;
const CMovImplAll<Jcc_LessOrEqual> iCMOVLE;
const CMovImplAll<Jcc_Greater> xCMOVG;
const CMovImplAll<Jcc_GreaterOrEqual> xCMOVGE;
const CMovImplAll<Jcc_Less> xCMOVL;
const CMovImplAll<Jcc_LessOrEqual> xCMOVLE;
const CMovImplAll<Jcc_Zero> iCMOVZ;
const CMovImplAll<Jcc_Equal> iCMOVE;
const CMovImplAll<Jcc_NotZero> iCMOVNZ;
const CMovImplAll<Jcc_NotEqual> iCMOVNE;
const CMovImplAll<Jcc_Zero> xCMOVZ;
const CMovImplAll<Jcc_Equal> xCMOVE;
const CMovImplAll<Jcc_NotZero> xCMOVNZ;
const CMovImplAll<Jcc_NotEqual> xCMOVNE;
const CMovImplAll<Jcc_Overflow> iCMOVO;
const CMovImplAll<Jcc_NotOverflow> iCMOVNO;
const CMovImplAll<Jcc_Carry> iCMOVC;
const CMovImplAll<Jcc_NotCarry> iCMOVNC;
const CMovImplAll<Jcc_Overflow> xCMOVO;
const CMovImplAll<Jcc_NotOverflow> xCMOVNO;
const CMovImplAll<Jcc_Carry> xCMOVC;
const CMovImplAll<Jcc_NotCarry> xCMOVNC;
const CMovImplAll<Jcc_Signed> iCMOVS;
const CMovImplAll<Jcc_Unsigned> iCMOVNS;
const CMovImplAll<Jcc_ParityEven> iCMOVPE;
const CMovImplAll<Jcc_ParityOdd> iCMOVPO;
const CMovImplAll<Jcc_Signed> xCMOVS;
const CMovImplAll<Jcc_Unsigned> xCMOVNS;
const CMovImplAll<Jcc_ParityEven> xCMOVPE;
const CMovImplAll<Jcc_ParityOdd> xCMOVPO;
// ------------------------------------------------------------------------
const SetImplGeneric iSET;
const SetImplGeneric xSET;
const SetImplAll<Jcc_Above> iSETA;
const SetImplAll<Jcc_AboveOrEqual> iSETAE;
const SetImplAll<Jcc_Below> iSETB;
const SetImplAll<Jcc_BelowOrEqual> iSETBE;
const SetImplAll<Jcc_Above> xSETA;
const SetImplAll<Jcc_AboveOrEqual> xSETAE;
const SetImplAll<Jcc_Below> xSETB;
const SetImplAll<Jcc_BelowOrEqual> xSETBE;
const SetImplAll<Jcc_Greater> iSETG;
const SetImplAll<Jcc_GreaterOrEqual> iSETGE;
const SetImplAll<Jcc_Less> iSETL;
const SetImplAll<Jcc_LessOrEqual> iSETLE;
const SetImplAll<Jcc_Greater> xSETG;
const SetImplAll<Jcc_GreaterOrEqual> xSETGE;
const SetImplAll<Jcc_Less> xSETL;
const SetImplAll<Jcc_LessOrEqual> xSETLE;
const SetImplAll<Jcc_Zero> iSETZ;
const SetImplAll<Jcc_Equal> iSETE;
const SetImplAll<Jcc_NotZero> iSETNZ;
const SetImplAll<Jcc_NotEqual> iSETNE;
const SetImplAll<Jcc_Zero> xSETZ;
const SetImplAll<Jcc_Equal> xSETE;
const SetImplAll<Jcc_NotZero> xSETNZ;
const SetImplAll<Jcc_NotEqual> xSETNE;
const SetImplAll<Jcc_Overflow> iSETO;
const SetImplAll<Jcc_NotOverflow> iSETNO;
const SetImplAll<Jcc_Carry> iSETC;
const SetImplAll<Jcc_NotCarry> iSETNC;
const SetImplAll<Jcc_Overflow> xSETO;
const SetImplAll<Jcc_NotOverflow> xSETNO;
const SetImplAll<Jcc_Carry> xSETC;
const SetImplAll<Jcc_NotCarry> xSETNC;
const SetImplAll<Jcc_Signed> iSETS;
const SetImplAll<Jcc_Unsigned> iSETNS;
const SetImplAll<Jcc_ParityEven> iSETPE;
const SetImplAll<Jcc_ParityOdd> iSETPO;
const SetImplAll<Jcc_Signed> xSETS;
const SetImplAll<Jcc_Unsigned> xSETNS;
const SetImplAll<Jcc_ParityEven> xSETPE;
const SetImplAll<Jcc_ParityOdd> xSETPO;
// ------------------------------------------------------------------------
@ -389,7 +389,7 @@ __emitinline void iAdvancePtr( uint bytes )
{
// common debugger courtesy: advance with INT3 as filler.
for( uint i=0; i<bytes; i++ )
iWrite<u8>( 0xcc );
xWrite<u8>( 0xcc );
}
else
x86Ptr += bytes;
@ -430,7 +430,7 @@ void ModSibBase::Reduce()
Index = Base;
Scale = 0;
if( !Base.IsStackPointer() ) // prevent ESP from being encoded 'alone'
Base = iAddressReg::Empty;
Base = xAddressReg::Empty;
return;
}
@ -484,9 +484,9 @@ void ModSibBase::Reduce()
// of LEA, which alters flags states.
//
template< typename OperandType >
static void EmitLeaMagic( iRegister<OperandType> to, const ModSibBase& src, bool preserve_flags )
static void EmitLeaMagic( xRegister<OperandType> to, const ModSibBase& src, bool preserve_flags )
{
typedef iRegister<OperandType> ToReg;
typedef xRegister<OperandType> ToReg;
int displacement_size = (src.Displacement == 0) ? 0 :
( ( src.IsByteSizeDisp() ) ? 1 : 2 );
@ -501,12 +501,12 @@ static void EmitLeaMagic( iRegister<OperandType> to, const ModSibBase& src, bool
if( src.Index.IsEmpty() )
{
iMOV( to, src.Displacement );
xMOV( to, src.Displacement );
return;
}
else if( displacement_size == 0 )
{
iMOV( to, ToReg( src.Index.Id ) );
xMOV( to, ToReg( src.Index.Id ) );
return;
}
else
@ -516,8 +516,8 @@ static void EmitLeaMagic( iRegister<OperandType> to, const ModSibBase& src, bool
// encode as MOV and ADD combo. Make sure to use the immediate on the
// ADD since it can encode as an 8-bit sign-extended value.
iMOV( to, ToReg( src.Index.Id ) );
iADD( to, src.Displacement );
xMOV( to, ToReg( src.Index.Id ) );
xADD( to, src.Displacement );
return;
}
else
@ -525,7 +525,7 @@ static void EmitLeaMagic( iRegister<OperandType> to, const ModSibBase& src, bool
// note: no need to do ebp+0 check since we encode all 0 displacements as
// register assignments above (via MOV)
iWrite<u8>( 0x8d );
xWrite<u8>( 0x8d );
ModRM( displacement_size, to.Id, src.Index.Id );
}
}
@ -543,14 +543,14 @@ static void EmitLeaMagic( iRegister<OperandType> to, const ModSibBase& src, bool
// (this does not apply to older model P4s with the broken barrel shifter,
// but we currently aren't optimizing for that target anyway).
iMOV( to, ToReg( src.Index.Id ) );
iSHL( to, src.Scale );
xMOV( to, ToReg( src.Index.Id ) );
xSHL( to, src.Scale );
return;
}
iWrite<u8>( 0x8d );
xWrite<u8>( 0x8d );
ModRM( 0, to.Id, ModRm_UseSib );
SibSB( src.Scale, src.Index.Id, ModRm_UseDisp32 );
iWrite<u32>( src.Displacement );
xWrite<u32>( src.Displacement );
return;
}
else
@ -562,14 +562,14 @@ static void EmitLeaMagic( iRegister<OperandType> to, const ModSibBase& src, bool
if( src.Index == esp )
{
// ESP is not encodable as an index (ix86 ignores it), thus:
iMOV( to, ToReg( src.Base.Id ) ); // will do the trick!
if( src.Displacement ) iADD( to, src.Displacement );
xMOV( to, ToReg( src.Base.Id ) ); // will do the trick!
if( src.Displacement ) xADD( to, src.Displacement );
return;
}
else if( src.Displacement == 0 )
{
iMOV( to, ToReg( src.Base.Id ) );
iADD( to, ToReg( src.Index.Id ) );
xMOV( to, ToReg( src.Base.Id ) );
xADD( to, ToReg( src.Index.Id ) );
return;
}
}
@ -578,7 +578,7 @@ static void EmitLeaMagic( iRegister<OperandType> to, const ModSibBase& src, bool
// special case handling of ESP as Index, which is replaceable with
// a single MOV even when preserve_flags is set! :D
iMOV( to, ToReg( src.Base.Id ) );
xMOV( to, ToReg( src.Base.Id ) );
return;
}
}
@ -586,7 +586,7 @@ static void EmitLeaMagic( iRegister<OperandType> to, const ModSibBase& src, bool
if( src.Base == ebp && displacement_size == 0 )
displacement_size = 1; // forces [ebp] to be encoded as [ebp+0]!
iWrite<u8>( 0x8d );
xWrite<u8>( 0x8d );
ModRM( displacement_size, to.Id, ModRm_UseSib );
SibSB( src.Scale, src.Index.Id, src.Base.Id );
}
@ -595,19 +595,19 @@ static void EmitLeaMagic( iRegister<OperandType> to, const ModSibBase& src, bool
if( displacement_size != 0 )
{
if( displacement_size == 1 )
iWrite<s8>( src.Displacement );
xWrite<s8>( src.Displacement );
else
iWrite<s32>( src.Displacement );
xWrite<s32>( src.Displacement );
}
}
__emitinline void iLEA( iRegister32 to, const ModSibBase& src, bool preserve_flags )
__emitinline void xLEA( xRegister32 to, const ModSibBase& src, bool preserve_flags )
{
EmitLeaMagic( to, src, preserve_flags );
}
__emitinline void iLEA( iRegister16 to, const ModSibBase& src, bool preserve_flags )
__emitinline void xLEA( xRegister16 to, const ModSibBase& src, bool preserve_flags )
{
write8( 0x66 );
EmitLeaMagic( to, src, preserve_flags );
@ -620,21 +620,21 @@ __emitinline void iLEA( iRegister16 to, const ModSibBase& src, bool preserve_fla
// Note: pushad/popad implementations are intentionally left out. The instructions are
// invalid in x64, and are super slow on x32. Use multiple Push/Pop instructions instead.
__emitinline void iPOP( const ModSibBase& from )
__emitinline void xPOP( const ModSibBase& from )
{
iWrite<u8>( 0x8f );
xWrite<u8>( 0x8f );
EmitSibMagic( 0, from );
}
__emitinline void iPUSH( const ModSibBase& from )
__emitinline void xPUSH( const ModSibBase& from )
{
iWrite<u8>( 0xff );
xWrite<u8>( 0xff );
EmitSibMagic( 6, from );
}
//////////////////////////////////////////////////////////////////////////////////////////
//
__emitinline void iBSWAP( const iRegister32& to )
__emitinline void xBSWAP( const xRegister32& to )
{
write8( 0x0F );
write8( 0xC8 | to.Id );
@ -645,66 +645,81 @@ __emitinline void iBSWAP( const iRegister32& to )
// MMX / XMM Instructions
// (these will get put in their own file later)
const MovapsImplAll< 0, 0x28, 0x29 > iMOVAPS;
const MovapsImplAll< 0, 0x10, 0x11 > iMOVUPS;
const MovapsImplAll< 0x66, 0x28, 0x29 > iMOVAPD;
const MovapsImplAll< 0x66, 0x10, 0x11 > iMOVUPD;
const MovapsImplAll< 0, 0x28, 0x29 > xMOVAPS;
const MovapsImplAll< 0, 0x10, 0x11 > xMOVUPS;
const MovapsImplAll< 0x66, 0x28, 0x29 > xMOVAPD;
const MovapsImplAll< 0x66, 0x10, 0x11 > xMOVUPD;
#ifdef ALWAYS_USE_MOVAPS
const MovapsImplAll< 0x66, 0x6f, 0x7f > iMOVDQA;
const MovapsImplAll< 0xf3, 0x6f, 0x7f > iMOVDQU;
const MovapsImplAll< 0x66, 0x6f, 0x7f > xMOVDQA;
const MovapsImplAll< 0xf3, 0x6f, 0x7f > xMOVDQU;
#else
const MovapsImplAll< 0, 0x28, 0x29 > iMOVDQA;
const MovapsImplAll< 0, 0x10, 0x11 > iMOVDQU;
const MovapsImplAll< 0, 0x28, 0x29 > xMOVDQA;
const MovapsImplAll< 0, 0x10, 0x11 > xMOVDQU;
#endif
const MovhlImplAll< 0, 0x16 > iMOVHPS;
const MovhlImplAll< 0, 0x12 > iMOVLPS;
const MovhlImplAll< 0x66, 0x16 > iMOVHPD;
const MovhlImplAll< 0x66, 0x12 > iMOVLPD;
const MovhlImplAll<0x16> xMOVH;
const MovhlImplAll<0x12> xMOVL;
const MovhlImpl_RtoR<0x16> xMOVLH;
const MovhlImpl_RtoR<0x12> xMOVHL;
const PLogicImplAll<0xdb> iPAND;
const PLogicImplAll<0xdf> iPANDN;
const PLogicImplAll<0xeb> iPOR;
const PLogicImplAll<0xef> iPXOR;
const PLogicImplAll<0xdb> xPAND;
const PLogicImplAll<0xdf> xPANDN;
const PLogicImplAll<0xeb> xPOR;
const PLogicImplAll<0xef> xPXOR;
const SSEAndNotImpl<0x55> iANDN;
const SSEAndNotImpl<0x55> xANDN;
// Compute Reciprocal Packed Single-Precision Floating-Point Values
const SSELogicImpl<0,0x53> iRCPPS;
const SSELogicImpl<0,0x53> xRCPPS;
// Compute Reciprocal of Scalar Single-Precision Floating-Point Value
const SSELogicImpl<0xf3,0x53> iRCPSS;
const SSELogicImpl<0xf3,0x53> xRCPSS;
// ------------------------------------------------------------------------
const SSECompareImpl<SSE2_Equal> xCMPEQ;
const SSECompareImpl<SSE2_Less> xCMPLT;
const SSECompareImpl<SSE2_LessOrEqual> xCMPLE;
const SSECompareImpl<SSE2_Unordered> xCMPUNORD;
const SSECompareImpl<SSE2_NotEqual> xCMPNE;
const SSECompareImpl<SSE2_NotLess> xCMPNLT;
const SSECompareImpl<SSE2_NotLessOrEqual> xCMPNLE;
const SSECompareImpl<SSE2_Ordered> xCMPORD;
//////////////////////////////////////////////////////////////////////////////////////////
//
// Moves from XMM to XMM, with the *upper 64 bits* of the destination register
// being cleared to zero.
__forceinline void iMOVQZX( const iRegisterSSE& to, const iRegisterSSE& from ) { writeXMMop( 0xf3, 0x7e, to, from ); }
__forceinline void xMOVQZX( const xRegisterSSE& to, const xRegisterSSE& from ) { writeXMMop( 0xf3, 0x7e, to, from ); }
// Moves from XMM to XMM, with the *upper 64 bits* of the destination register
// being cleared to zero.
__forceinline void iMOVQZX( const iRegisterSSE& to, const ModSibBase& src ) { writeXMMop( 0xf3, 0x7e, to, src ); }
__forceinline void xMOVQZX( const xRegisterSSE& to, const ModSibBase& src ) { writeXMMop( 0xf3, 0x7e, to, src ); }
// Moves from XMM to XMM, with the *upper 64 bits* of the destination register
// being cleared to zero.
__forceinline void iMOVQZX( const iRegisterSSE& to, const void* src ) { writeXMMop( 0xf3, 0x7e, to, src ); }
__forceinline void xMOVQZX( const xRegisterSSE& to, const void* src ) { writeXMMop( 0xf3, 0x7e, to, src ); }
// Moves lower quad of XMM to ptr64 (no bits are cleared)
__forceinline void iMOVQ( const ModSibBase& dest, const iRegisterSSE& from ) { writeXMMop( 0x66, 0xd6, from, dest ); }
__forceinline void xMOVQ( const ModSibBase& dest, const xRegisterSSE& from ) { writeXMMop( 0x66, 0xd6, from, dest ); }
// Moves lower quad of XMM to ptr64 (no bits are cleared)
__forceinline void iMOVQ( void* dest, const iRegisterSSE& from ) { writeXMMop( 0x66, 0xd6, from, dest ); }
__forceinline void xMOVQ( void* dest, const xRegisterSSE& from ) { writeXMMop( 0x66, 0xd6, from, dest ); }
__forceinline void iMOVQ( const iRegisterMMX& to, const iRegisterMMX& from ) { if( to != from ) writeXMMop( 0x6f, to, from ); }
__forceinline void iMOVQ( const iRegisterMMX& to, const ModSibBase& src ) { writeXMMop( 0x6f, to, src ); }
__forceinline void iMOVQ( const iRegisterMMX& to, const void* src ) { writeXMMop( 0x6f, to, src ); }
__forceinline void iMOVQ( const ModSibBase& dest, const iRegisterMMX& from ) { writeXMMop( 0x7f, from, dest ); }
__forceinline void iMOVQ( void* dest, const iRegisterMMX& from ) { writeXMMop( 0x7f, from, dest ); }
__forceinline void xMOVQ( const xRegisterMMX& to, const xRegisterMMX& from ) { if( to != from ) writeXMMop( 0x6f, to, from ); }
__forceinline void xMOVQ( const xRegisterMMX& to, const ModSibBase& src ) { writeXMMop( 0x6f, to, src ); }
__forceinline void xMOVQ( const xRegisterMMX& to, const void* src ) { writeXMMop( 0x6f, to, src ); }
__forceinline void xMOVQ( const ModSibBase& dest, const xRegisterMMX& from ) { writeXMMop( 0x7f, from, dest ); }
__forceinline void xMOVQ( void* dest, const xRegisterMMX& from ) { writeXMMop( 0x7f, from, dest ); }
// This form of iMOVQ is Intel's adeptly named 'MOVQ2DQ'
__forceinline void iMOVQ( const iRegisterSSE& to, const iRegisterMMX& from ) { writeXMMop( 0xf3, 0xd6, to, from ); }
// This form of xMOVQ is Intel's adeptly named 'MOVQ2DQ'
__forceinline void xMOVQ( const xRegisterSSE& to, const xRegisterMMX& from ) { writeXMMop( 0xf3, 0xd6, to, from ); }
// This form of iMOVQ is Intel's adeptly named 'MOVDQ2Q'
__forceinline void iMOVQ( const iRegisterMMX& to, const iRegisterSSE& from )
// This form of xMOVQ is Intel's adeptly named 'MOVDQ2Q'
__forceinline void xMOVQ( const xRegisterMMX& to, const xRegisterSSE& from )
{
// Manual implementation of this form of MOVQ, since its parameters are unique in a way
// that breaks the template inference of writeXMMop();
@ -716,53 +731,42 @@ __forceinline void iMOVQ( const iRegisterMMX& to, const iRegisterSSE& from )
//////////////////////////////////////////////////////////////////////////////////////////
//
#define IMPLEMENT_iMOVS( ssd, prefix ) \
__forceinline void iMOV##ssd( const iRegisterSSE& to, const iRegisterSSE& from ) { if( to != from ) writeXMMop( prefix, 0x10, to, from ); } \
__forceinline void iMOV##ssd##ZX( const iRegisterSSE& to, const void* from ) { writeXMMop( prefix, 0x10, to, from ); } \
__forceinline void iMOV##ssd##ZX( const iRegisterSSE& to, const ModSibBase& from ) { writeXMMop( prefix, 0x10, to, from ); } \
__forceinline void iMOV##ssd( const void* to, const iRegisterSSE& from ) { writeXMMop( prefix, 0x11, from, to ); } \
__forceinline void iMOV##ssd( const ModSibBase& to, const iRegisterSSE& from ) { writeXMMop( prefix, 0x11, from, to ); }
#define IMPLEMENT_xMOVS( ssd, prefix ) \
__forceinline void xMOV##ssd( const xRegisterSSE& to, const xRegisterSSE& from ) { if( to != from ) writeXMMop( prefix, 0x10, to, from ); } \
__forceinline void xMOV##ssd##ZX( const xRegisterSSE& to, const void* from ) { writeXMMop( prefix, 0x10, to, from ); } \
__forceinline void xMOV##ssd##ZX( const xRegisterSSE& to, const ModSibBase& from ) { writeXMMop( prefix, 0x10, to, from ); } \
__forceinline void xMOV##ssd( const void* to, const xRegisterSSE& from ) { writeXMMop( prefix, 0x11, from, to ); } \
__forceinline void xMOV##ssd( const ModSibBase& to, const xRegisterSSE& from ) { writeXMMop( prefix, 0x11, from, to ); }
IMPLEMENT_iMOVS( SS, 0xf3 )
IMPLEMENT_iMOVS( SD, 0xf2 )
IMPLEMENT_xMOVS( SS, 0xf3 )
IMPLEMENT_xMOVS( SD, 0xf2 )
//////////////////////////////////////////////////////////////////////////////////////////
// Non-temporal movs only support a register as a target (ie, load form only, no stores)
//
__forceinline void iMOVNTDQA( const iRegisterSSE& to, const void* from )
__forceinline void xMOVNTDQA( const xRegisterSSE& to, const void* from )
{
iWrite<u32>( 0x2A380f66 );
iWriteDisp( to.Id, from );
xWrite<u32>( 0x2A380f66 );
xWriteDisp( to.Id, from );
}
__noinline void iMOVNTDQA( const iRegisterSSE& to, const ModSibBase& from )
__noinline void xMOVNTDQA( const xRegisterSSE& to, const ModSibBase& from )
{
iWrite<u32>( 0x2A380f66 );
xWrite<u32>( 0x2A380f66 );
EmitSibMagic( to.Id, from );
}
__forceinline void iMOVNTDQ( void* to, const iRegisterSSE& from ) { writeXMMop( 0x66, 0xe7, from, to ); }
__noinline void iMOVNTDQA( const ModSibBase& to, const iRegisterSSE& from ) { writeXMMop( 0x66, 0xe7, from, to ); }
__forceinline void xMOVNTDQ( void* to, const xRegisterSSE& from ) { writeXMMop( 0x66, 0xe7, from, to ); }
__noinline void xMOVNTDQA( const ModSibBase& to, const xRegisterSSE& from ) { writeXMMop( 0x66, 0xe7, from, to ); }
__forceinline void iMOVNTPD( void* to, const iRegisterSSE& from ) { writeXMMop( 0x66, 0x2b, from, to ); }
__noinline void iMOVNTPD( const ModSibBase& to, const iRegisterSSE& from ) { writeXMMop( 0x66, 0x2b, from, to ); }
__forceinline void iMOVNTPS( void* to, const iRegisterSSE& from ) { writeXMMop( 0x2b, from, to ); }
__noinline void iMOVNTPS( const ModSibBase& to, const iRegisterSSE& from ) { writeXMMop( 0x2b, from, to ); }
__forceinline void xMOVNTPD( void* to, const xRegisterSSE& from ) { writeXMMop( 0x66, 0x2b, from, to ); }
__noinline void xMOVNTPD( const ModSibBase& to, const xRegisterSSE& from ) { writeXMMop( 0x66, 0x2b, from, to ); }
__forceinline void xMOVNTPS( void* to, const xRegisterSSE& from ) { writeXMMop( 0x2b, from, to ); }
__noinline void xMOVNTPS( const ModSibBase& to, const xRegisterSSE& from ) { writeXMMop( 0x2b, from, to ); }
__forceinline void iMOVNTQ( void* to, const iRegisterMMX& from ) { writeXMMop( 0xe7, from, to ); }
__noinline void iMOVNTQ( const ModSibBase& to, const iRegisterMMX& from ) { writeXMMop( 0xe7, from, to ); }
//////////////////////////////////////////////////////////////////////////////////////////
// Mov Low to High / High to Low
//
// These instructions come in xmmreg,xmmreg forms only!
//
__forceinline void iMOVLHPS( const iRegisterSSE& to, const iRegisterSSE& from ) { writeXMMop( 0x16, to, from ); }
__forceinline void iMOVHLPS( const iRegisterSSE& to, const iRegisterSSE& from ) { writeXMMop( 0x12, to, from ); }
__forceinline void iMOVLHPD( const iRegisterSSE& to, const iRegisterSSE& from ) { writeXMMop( 0x66, 0x16, to, from ); }
__forceinline void iMOVHLPD( const iRegisterSSE& to, const iRegisterSSE& from ) { writeXMMop( 0x66, 0x12, to, from ); }
__forceinline void xMOVNTQ( void* to, const xRegisterMMX& from ) { writeXMMop( 0xe7, from, to ); }
__noinline void xMOVNTQ( const ModSibBase& to, const xRegisterMMX& from ) { writeXMMop( 0xe7, from, to ); }
}

View File

@ -53,29 +53,29 @@ namespace x86Emitter
//////////////////////////////////////////////////////////////////////////////////////////
// x86Register Method Implementations
//
__forceinline iAddressInfo iAddressReg::operator+( const iAddressReg& right ) const
__forceinline xAddressInfo xAddressReg::operator+( const xAddressReg& right ) const
{
return iAddressInfo( *this, right );
return xAddressInfo( *this, right );
}
__forceinline iAddressInfo iAddressReg::operator+( const iAddressInfo& right ) const
__forceinline xAddressInfo xAddressReg::operator+( const xAddressInfo& right ) const
{
return right + *this;
}
__forceinline iAddressInfo iAddressReg::operator+( s32 right ) const
__forceinline xAddressInfo xAddressReg::operator+( s32 right ) const
{
return iAddressInfo( *this, right );
return xAddressInfo( *this, right );
}
__forceinline iAddressInfo iAddressReg::operator*( u32 right ) const
__forceinline xAddressInfo xAddressReg::operator*( u32 right ) const
{
return iAddressInfo( Empty, *this, right );
return xAddressInfo( Empty, *this, right );
}
__forceinline iAddressInfo iAddressReg::operator<<( u32 shift ) const
__forceinline xAddressInfo xAddressReg::operator<<( u32 shift ) const
{
return iAddressInfo( Empty, *this, 1<<shift );
return xAddressInfo( Empty, *this, 1<<shift );
}
//////////////////////////////////////////////////////////////////////////////////////////
@ -83,7 +83,7 @@ namespace x86Emitter
//
// ------------------------------------------------------------------------
__forceinline ModSibBase::ModSibBase( const iAddressInfo& src ) :
__forceinline ModSibBase::ModSibBase( const xAddressInfo& src ) :
Base( src.Base ),
Index( src.Index ),
Scale( src.Factor ),
@ -93,7 +93,7 @@ namespace x86Emitter
}
// ------------------------------------------------------------------------
__forceinline ModSibBase::ModSibBase( iAddressReg base, iAddressReg index, int scale, s32 displacement ) :
__forceinline ModSibBase::ModSibBase( xAddressReg base, xAddressReg index, int scale, s32 displacement ) :
Base( base ),
Index( index ),
Scale( scale ),
@ -113,9 +113,9 @@ namespace x86Emitter
}
//////////////////////////////////////////////////////////////////////////////////////////
// iAddressInfo Method Implementations
// xAddressInfo Method Implementations
//
__forceinline iAddressInfo& iAddressInfo::Add( const iAddressReg& src )
__forceinline xAddressInfo& xAddressInfo::Add( const xAddressReg& src )
{
if( src == Index )
{
@ -124,7 +124,7 @@ namespace x86Emitter
else if( src == Base )
{
// Compound the existing register reference into the Index/Scale pair.
Base = iAddressReg::Empty;
Base = xAddressReg::Empty;
if( src == Index )
Factor++;
@ -146,7 +146,7 @@ namespace x86Emitter
}
// ------------------------------------------------------------------------
__forceinline iAddressInfo& iAddressInfo::Add( const iAddressInfo& src )
__forceinline xAddressInfo& xAddressInfo::Add( const xAddressInfo& src )
{
Add( src.Base );
Add( src.Displacement );
@ -174,7 +174,7 @@ namespace x86Emitter
// ------------------------------------------------------------------------
template< typename OperandType >
iForwardJump<OperandType>::iForwardJump( JccComparisonType cctype ) :
xForwardJump<OperandType>::xForwardJump( JccComparisonType cctype ) :
BasePtr( (s8*)iGetPtr() +
((OperandSize == 1) ? 2 : // j8's are always 2 bytes.
((cctype==Jcc_Unconditional) ? 5 : 6 )) // j32's are either 5 or 6 bytes
@ -184,15 +184,15 @@ namespace x86Emitter
jASSUME( OperandSize == 1 || OperandSize == 4 );
if( OperandSize == 1 )
iWrite<u8>( (cctype == Jcc_Unconditional) ? 0xeb : (0x70 | cctype) );
xWrite<u8>( (cctype == Jcc_Unconditional) ? 0xeb : (0x70 | cctype) );
else
{
if( cctype == Jcc_Unconditional )
iWrite<u8>( 0xe9 );
xWrite<u8>( 0xe9 );
else
{
iWrite<u8>( 0x0f );
iWrite<u8>( 0x80 | cctype );
xWrite<u8>( 0x0f );
xWrite<u8>( 0x80 | cctype );
}
}
@ -201,7 +201,7 @@ namespace x86Emitter
// ------------------------------------------------------------------------
template< typename OperandType >
void iForwardJump<OperandType>::SetTarget() const
void xForwardJump<OperandType>::SetTarget() const
{
jASSUME( BasePtr != NULL );

View File

@ -38,16 +38,16 @@ namespace x86Emitter
// ------------------------------------------------------------------------
// Group 1 Instruction Class
extern const Internal::G1LogicImpl<Internal::G1Type_AND,0x54> iAND;
extern const Internal::G1LogicImpl<Internal::G1Type_OR,0x56> iOR;
extern const Internal::G1LogicImpl<Internal::G1Type_XOR,0x57> iXOR;
extern const Internal::SSEAndNotImpl<0x55> iANDN;
extern const Internal::G1LogicImpl_PlusSSE<Internal::G1Type_AND,0x54> xAND;
extern const Internal::G1LogicImpl_PlusSSE<Internal::G1Type_OR,0x56> xOR;
extern const Internal::G1LogicImpl_PlusSSE<Internal::G1Type_XOR,0x57> xXOR;
extern const Internal::G1ArithmeticImpl<Internal::G1Type_ADD,0x58> iADD;
extern const Internal::G1ArithmeticImpl<Internal::G1Type_SUB,0x5c> iSUB;
extern const Internal::Group1ImplAll<Internal::G1Type_ADC> iADC;
extern const Internal::Group1ImplAll<Internal::G1Type_SBB> iSBB;
extern const Internal::Group1ImplAll<Internal::G1Type_CMP> iCMP;
extern const Internal::G1ArithmeticImpl_PlusSSE<Internal::G1Type_ADD,0x58> xADD;
extern const Internal::G1ArithmeticImpl_PlusSSE<Internal::G1Type_SUB,0x5c> xSUB;
extern const Internal::G1CompareImpl_PlusSSE xCMP;
extern const Internal::Group1ImplAll<Internal::G1Type_ADC> xADC;
extern const Internal::Group1ImplAll<Internal::G1Type_SBB> xSBB;
// ------------------------------------------------------------------------
// Group 2 Instruction Class
@ -56,174 +56,154 @@ namespace x86Emitter
// zero. This is a safe optimization since any zero-value shift does not affect any
// flags.
extern const Internal::MovImplAll iMOV;
extern const Internal::TestImplAll iTEST;
extern const Internal::MovImplAll xMOV;
extern const Internal::TestImplAll xTEST;
extern const Internal::Group2ImplAll<Internal::G2Type_ROL> iROL;
extern const Internal::Group2ImplAll<Internal::G2Type_ROR> iROR;
extern const Internal::Group2ImplAll<Internal::G2Type_RCL> iRCL;
extern const Internal::Group2ImplAll<Internal::G2Type_RCR> iRCR;
extern const Internal::Group2ImplAll<Internal::G2Type_SHL> iSHL;
extern const Internal::Group2ImplAll<Internal::G2Type_SHR> iSHR;
extern const Internal::Group2ImplAll<Internal::G2Type_SAR> iSAR;
extern const Internal::Group2ImplAll<Internal::G2Type_ROL> xROL;
extern const Internal::Group2ImplAll<Internal::G2Type_ROR> xROR;
extern const Internal::Group2ImplAll<Internal::G2Type_RCL> xRCL;
extern const Internal::Group2ImplAll<Internal::G2Type_RCR> xRCR;
extern const Internal::Group2ImplAll<Internal::G2Type_SHL> xSHL;
extern const Internal::Group2ImplAll<Internal::G2Type_SHR> xSHR;
extern const Internal::Group2ImplAll<Internal::G2Type_SAR> xSAR;
// ------------------------------------------------------------------------
// Group 3 Instruction Class
extern const Internal::Group3ImplAll<Internal::G3Type_NOT> iNOT;
extern const Internal::Group3ImplAll<Internal::G3Type_NEG> iNEG;
extern const Internal::Group3ImplAll<Internal::G3Type_MUL> iUMUL;
extern const Internal::Group3ImplAll<Internal::G3Type_DIV> iUDIV;
extern const Internal::G3Impl_PlusSSE<Internal::G3Type_iDIV,0x5e> iDIV;
extern const Internal::iMul_PlusSSE iMUL;
extern const Internal::Group3ImplAll<Internal::G3Type_NOT> xNOT;
extern const Internal::Group3ImplAll<Internal::G3Type_NEG> xNEG;
extern const Internal::Group3ImplAll<Internal::G3Type_MUL> xUMUL;
extern const Internal::Group3ImplAll<Internal::G3Type_DIV> xUDIV;
extern const Internal::G3Impl_PlusSSE<Internal::G3Type_iDIV,0x5e> xDIV;
extern const Internal::iMul_PlusSSE xMUL;
extern const Internal::IncDecImplAll<false> iINC;
extern const Internal::IncDecImplAll<true> iDEC;
extern const Internal::IncDecImplAll<false> xINC;
extern const Internal::IncDecImplAll<true> xDEC;
extern const Internal::MovExtendImplAll<false> iMOVZX;
extern const Internal::MovExtendImplAll<true> iMOVSX;
extern const Internal::MovExtendImplAll<false> xMOVZX;
extern const Internal::MovExtendImplAll<true> xMOVSX;
extern const Internal::DwordShiftImplAll<false> iSHLD;
extern const Internal::DwordShiftImplAll<true> iSHRD;
extern const Internal::DwordShiftImplAll<false> xSHLD;
extern const Internal::DwordShiftImplAll<true> xSHRD;
extern const Internal::Group8ImplAll<Internal::G8Type_BT> iBT;
extern const Internal::Group8ImplAll<Internal::G8Type_BTR> iBTR;
extern const Internal::Group8ImplAll<Internal::G8Type_BTS> iBTS;
extern const Internal::Group8ImplAll<Internal::G8Type_BTC> iBTC;
extern const Internal::Group8ImplAll<Internal::G8Type_BT> xBT;
extern const Internal::Group8ImplAll<Internal::G8Type_BTR> xBTR;
extern const Internal::Group8ImplAll<Internal::G8Type_BTS> xBTS;
extern const Internal::Group8ImplAll<Internal::G8Type_BTC> xBTC;
extern const Internal::JmpCallImplAll<true> iJMP;
extern const Internal::JmpCallImplAll<false> iCALL;
extern const Internal::JmpCallImplAll<true> xJMP;
extern const Internal::JmpCallImplAll<false> xCALL;
extern const Internal::BitScanImplAll<false> iBSF;
extern const Internal::BitScanImplAll<true> iBSR;
extern const Internal::BitScanImplAll<false> xBSF;
extern const Internal::BitScanImplAll<true> xBSR;
// ------------------------------------------------------------------------
extern const Internal::CMovImplGeneric iCMOV;
extern const Internal::CMovImplGeneric xCMOV;
extern const Internal::CMovImplAll<Jcc_Above> iCMOVA;
extern const Internal::CMovImplAll<Jcc_AboveOrEqual> iCMOVAE;
extern const Internal::CMovImplAll<Jcc_Below> iCMOVB;
extern const Internal::CMovImplAll<Jcc_BelowOrEqual> iCMOVBE;
extern const Internal::CMovImplAll<Jcc_Above> xCMOVA;
extern const Internal::CMovImplAll<Jcc_AboveOrEqual> xCMOVAE;
extern const Internal::CMovImplAll<Jcc_Below> xCMOVB;
extern const Internal::CMovImplAll<Jcc_BelowOrEqual> xCMOVBE;
extern const Internal::CMovImplAll<Jcc_Greater> iCMOVG;
extern const Internal::CMovImplAll<Jcc_GreaterOrEqual> iCMOVGE;
extern const Internal::CMovImplAll<Jcc_Less> iCMOVL;
extern const Internal::CMovImplAll<Jcc_LessOrEqual> iCMOVLE;
extern const Internal::CMovImplAll<Jcc_Greater> xCMOVG;
extern const Internal::CMovImplAll<Jcc_GreaterOrEqual> xCMOVGE;
extern const Internal::CMovImplAll<Jcc_Less> xCMOVL;
extern const Internal::CMovImplAll<Jcc_LessOrEqual> xCMOVLE;
extern const Internal::CMovImplAll<Jcc_Zero> iCMOVZ;
extern const Internal::CMovImplAll<Jcc_Equal> iCMOVE;
extern const Internal::CMovImplAll<Jcc_NotZero> iCMOVNZ;
extern const Internal::CMovImplAll<Jcc_NotEqual> iCMOVNE;
extern const Internal::CMovImplAll<Jcc_Zero> xCMOVZ;
extern const Internal::CMovImplAll<Jcc_Equal> xCMOVE;
extern const Internal::CMovImplAll<Jcc_NotZero> xCMOVNZ;
extern const Internal::CMovImplAll<Jcc_NotEqual> xCMOVNE;
extern const Internal::CMovImplAll<Jcc_Overflow> iCMOVO;
extern const Internal::CMovImplAll<Jcc_NotOverflow> iCMOVNO;
extern const Internal::CMovImplAll<Jcc_Carry> iCMOVC;
extern const Internal::CMovImplAll<Jcc_NotCarry> iCMOVNC;
extern const Internal::CMovImplAll<Jcc_Overflow> xCMOVO;
extern const Internal::CMovImplAll<Jcc_NotOverflow> xCMOVNO;
extern const Internal::CMovImplAll<Jcc_Carry> xCMOVC;
extern const Internal::CMovImplAll<Jcc_NotCarry> xCMOVNC;
extern const Internal::CMovImplAll<Jcc_Signed> iCMOVS;
extern const Internal::CMovImplAll<Jcc_Unsigned> iCMOVNS;
extern const Internal::CMovImplAll<Jcc_ParityEven> iCMOVPE;
extern const Internal::CMovImplAll<Jcc_ParityOdd> iCMOVPO;
extern const Internal::CMovImplAll<Jcc_Signed> xCMOVS;
extern const Internal::CMovImplAll<Jcc_Unsigned> xCMOVNS;
extern const Internal::CMovImplAll<Jcc_ParityEven> xCMOVPE;
extern const Internal::CMovImplAll<Jcc_ParityOdd> xCMOVPO;
// ------------------------------------------------------------------------
extern const Internal::SetImplGeneric iSET;
extern const Internal::SetImplGeneric xSET;
extern const Internal::SetImplAll<Jcc_Above> iSETA;
extern const Internal::SetImplAll<Jcc_AboveOrEqual> iSETAE;
extern const Internal::SetImplAll<Jcc_Below> iSETB;
extern const Internal::SetImplAll<Jcc_BelowOrEqual> iSETBE;
extern const Internal::SetImplAll<Jcc_Above> xSETA;
extern const Internal::SetImplAll<Jcc_AboveOrEqual> xSETAE;
extern const Internal::SetImplAll<Jcc_Below> xSETB;
extern const Internal::SetImplAll<Jcc_BelowOrEqual> xSETBE;
extern const Internal::SetImplAll<Jcc_Greater> iSETG;
extern const Internal::SetImplAll<Jcc_GreaterOrEqual> iSETGE;
extern const Internal::SetImplAll<Jcc_Less> iSETL;
extern const Internal::SetImplAll<Jcc_LessOrEqual> iSETLE;
extern const Internal::SetImplAll<Jcc_Greater> xSETG;
extern const Internal::SetImplAll<Jcc_GreaterOrEqual> xSETGE;
extern const Internal::SetImplAll<Jcc_Less> xSETL;
extern const Internal::SetImplAll<Jcc_LessOrEqual> xSETLE;
extern const Internal::SetImplAll<Jcc_Zero> iSETZ;
extern const Internal::SetImplAll<Jcc_Equal> iSETE;
extern const Internal::SetImplAll<Jcc_NotZero> iSETNZ;
extern const Internal::SetImplAll<Jcc_NotEqual> iSETNE;
extern const Internal::SetImplAll<Jcc_Zero> xSETZ;
extern const Internal::SetImplAll<Jcc_Equal> xSETE;
extern const Internal::SetImplAll<Jcc_NotZero> xSETNZ;
extern const Internal::SetImplAll<Jcc_NotEqual> xSETNE;
extern const Internal::SetImplAll<Jcc_Overflow> iSETO;
extern const Internal::SetImplAll<Jcc_NotOverflow> iSETNO;
extern const Internal::SetImplAll<Jcc_Carry> iSETC;
extern const Internal::SetImplAll<Jcc_NotCarry> iSETNC;
extern const Internal::SetImplAll<Jcc_Overflow> xSETO;
extern const Internal::SetImplAll<Jcc_NotOverflow> xSETNO;
extern const Internal::SetImplAll<Jcc_Carry> xSETC;
extern const Internal::SetImplAll<Jcc_NotCarry> xSETNC;
extern const Internal::SetImplAll<Jcc_Signed> iSETS;
extern const Internal::SetImplAll<Jcc_Unsigned> iSETNS;
extern const Internal::SetImplAll<Jcc_ParityEven> iSETPE;
extern const Internal::SetImplAll<Jcc_ParityOdd> iSETPO;
extern const Internal::SetImplAll<Jcc_Signed> xSETS;
extern const Internal::SetImplAll<Jcc_Unsigned> xSETNS;
extern const Internal::SetImplAll<Jcc_ParityEven> xSETPE;
extern const Internal::SetImplAll<Jcc_ParityOdd> xSETPO;
//////////////////////////////////////////////////////////////////////////////////////////
// Miscellaneous Instructions
// These are all defined inline or in ix86.cpp.
//
extern void iBSWAP( const iRegister32& to );
extern void xBSWAP( const xRegister32& to );
// ----- Lea Instructions (Load Effective Address) -----
// Note: alternate (void*) forms of these instructions are not provided since those
// forms are functionally equivalent to Mov reg,imm, and thus better written as MOVs
// instead.
extern void iLEA( iRegister32 to, const ModSibBase& src, bool preserve_flags=false );
extern void iLEA( iRegister16 to, const ModSibBase& src, bool preserve_flags=false );
extern void xLEA( xRegister32 to, const ModSibBase& src, bool preserve_flags=false );
extern void xLEA( xRegister16 to, const ModSibBase& src, bool preserve_flags=false );
// ----- Push / Pop Instructions -----
// Note: pushad/popad implementations are intentionally left out. The instructions are
// invalid in x64, and are super slow on x32. Use multiple Push/Pop instructions instead.
extern void iPOP( const ModSibBase& from );
extern void iPUSH( const ModSibBase& from );
extern void xPOP( const ModSibBase& from );
extern void xPUSH( const ModSibBase& from );
static __forceinline void iPOP( iRegister32 from ) { write8( 0x58 | from.Id ); }
static __forceinline void iPOP( void* from ) { iPOP( ptr[from] ); }
static __forceinline void xPOP( xRegister32 from ) { write8( 0x58 | from.Id ); }
static __forceinline void xPOP( void* from ) { xPOP( ptr[from] ); }
static __forceinline void iPUSH( u32 imm ) { write8( 0x68 ); write32( imm ); }
static __forceinline void iPUSH( iRegister32 from ) { write8( 0x50 | from.Id ); }
static __forceinline void iPUSH( void* from ) { iPUSH( ptr[from] ); }
static __forceinline void xPUSH( u32 imm ) { write8( 0x68 ); write32( imm ); }
static __forceinline void xPUSH( xRegister32 from ) { write8( 0x50 | from.Id ); }
static __forceinline void xPUSH( void* from ) { xPUSH( ptr[from] ); }
// pushes the EFLAGS register onto the stack
static __forceinline void iPUSHFD() { write8( 0x9C ); }
static __forceinline void xPUSHFD() { write8( 0x9C ); }
// pops the EFLAGS register from the stack
static __forceinline void iPOPFD() { write8( 0x9D ); }
static __forceinline void xPOPFD() { write8( 0x9D ); }
// ----- Miscellaneous Instructions -----
// Various Instructions with no parameter and no special encoding logic.
__forceinline void iRET() { write8( 0xC3 ); }
__forceinline void iCBW() { write16( 0x9866 ); }
__forceinline void iCWD() { write8( 0x98 ); }
__forceinline void iCDQ() { write8( 0x99 ); }
__forceinline void iCWDE() { write8( 0x98 ); }
__forceinline void xRET() { write8( 0xC3 ); }
__forceinline void xCBW() { write16( 0x9866 ); }
__forceinline void xCWD() { write8( 0x98 ); }
__forceinline void xCDQ() { write8( 0x99 ); }
__forceinline void xCWDE() { write8( 0x98 ); }
__forceinline void iLAHF() { write8( 0x9f ); }
__forceinline void iSAHF() { write8( 0x9e ); }
__forceinline void xLAHF() { write8( 0x9f ); }
__forceinline void xSAHF() { write8( 0x9e ); }
__forceinline void iSTC() { write8( 0xF9 ); }
__forceinline void iCLC() { write8( 0xF8 ); }
__forceinline void xSTC() { write8( 0xF9 ); }
__forceinline void xCLC() { write8( 0xF8 ); }
// NOP 1-byte
__forceinline void iNOP() { write8(0x90); }
//////////////////////////////////////////////////////////////////////////////////////////
// MUL / DIV instructions
/*extern void iMUL( const iRegister32& to, const iRegister32& from );
extern void iMUL( const iRegister32& to, const void* src );
extern void iMUL( const iRegister32& to, const iRegister32& from, s32 imm );
extern void iMUL( const iRegister32& to, const ModSibBase& src );
extern void iMUL( const iRegister32& to, const ModSibBase& src, s32 imm );
extern void iMUL( const iRegister16& to, const iRegister16& from );
extern void iMUL( const iRegister16& to, const void* src );
extern void iMUL( const iRegister16& to, const iRegister16& from, s16 imm );
extern void iMUL( const iRegister16& to, const ModSibBase& src );
extern void iMUL( const iRegister16& to, const ModSibBase& src, s16 imm );
template< typename T >
__forceinline void iMUL( const iRegister<T>& from ) { Internal::Group3Impl<T>::Emit( Internal::G3Type_iMUL, from ); }
template< typename T >
__noinline void iMUL( const ModSibStrict<T>& from ) { Internal::Group3Impl<T>::Emit( Internal::G3Type_iMUL, from ); }*/
__forceinline void xNOP() { write8(0x90); }
//////////////////////////////////////////////////////////////////////////////////////////
// JMP / Jcc Instructions!
@ -232,92 +212,92 @@ namespace x86Emitter
#define DEFINE_FORWARD_JUMP( label, cond ) \
template< typename OperandType > \
class iForward##label : public iForwardJump<OperandType> \
class xForward##label : public xForwardJump<OperandType> \
{ \
public: \
iForward##label() : iForwardJump<OperandType>( cond ) {} \
xForward##label() : xForwardJump<OperandType>( cond ) {} \
};
// ------------------------------------------------------------------------
// Note: typedefs below are defined individually in order to appease Intellisense
// resolution. Including them into the class definition macro above breaks it.
typedef iForwardJump<s8> iForwardJump8;
typedef iForwardJump<s32> iForwardJump32;
typedef xForwardJump<s8> xForwardJump8;
typedef xForwardJump<s32> xForwardJump32;
DEFINE_FORWARD_JUMP( JA, Jcc_Above );
DEFINE_FORWARD_JUMP( JB, Jcc_Below );
DEFINE_FORWARD_JUMP( JAE, Jcc_AboveOrEqual );
DEFINE_FORWARD_JUMP( JBE, Jcc_BelowOrEqual );
typedef iForwardJA<s8> iForwardJA8;
typedef iForwardJA<s32> iForwardJA32;
typedef iForwardJB<s8> iForwardJB8;
typedef iForwardJB<s32> iForwardJB32;
typedef iForwardJAE<s8> iForwardJAE8;
typedef iForwardJAE<s32> iForwardJAE32;
typedef iForwardJBE<s8> iForwardJBE8;
typedef iForwardJBE<s32> iForwardJBE32;
typedef xForwardJA<s8> xForwardJA8;
typedef xForwardJA<s32> xForwardJA32;
typedef xForwardJB<s8> xForwardJB8;
typedef xForwardJB<s32> xForwardJB32;
typedef xForwardJAE<s8> xForwardJAE8;
typedef xForwardJAE<s32> xForwardJAE32;
typedef xForwardJBE<s8> xForwardJBE8;
typedef xForwardJBE<s32> xForwardJBE32;
DEFINE_FORWARD_JUMP( JG, Jcc_Greater );
DEFINE_FORWARD_JUMP( JL, Jcc_Less );
DEFINE_FORWARD_JUMP( JGE, Jcc_GreaterOrEqual );
DEFINE_FORWARD_JUMP( JLE, Jcc_LessOrEqual );
typedef iForwardJG<s8> iForwardJG8;
typedef iForwardJG<s32> iForwardJG32;
typedef iForwardJL<s8> iForwardJL8;
typedef iForwardJL<s32> iForwardJL32;
typedef iForwardJGE<s8> iForwardJGE8;
typedef iForwardJGE<s32> iForwardJGE32;
typedef iForwardJLE<s8> iForwardJLE8;
typedef iForwardJLE<s32> iForwardJLE32;
typedef xForwardJG<s8> xForwardJG8;
typedef xForwardJG<s32> xForwardJG32;
typedef xForwardJL<s8> xForwardJL8;
typedef xForwardJL<s32> xForwardJL32;
typedef xForwardJGE<s8> xForwardJGE8;
typedef xForwardJGE<s32> xForwardJGE32;
typedef xForwardJLE<s8> xForwardJLE8;
typedef xForwardJLE<s32> xForwardJLE32;
DEFINE_FORWARD_JUMP( JZ, Jcc_Zero );
DEFINE_FORWARD_JUMP( JE, Jcc_Equal );
DEFINE_FORWARD_JUMP( JNZ, Jcc_NotZero );
DEFINE_FORWARD_JUMP( JNE, Jcc_NotEqual );
typedef iForwardJZ<s8> iForwardJZ8;
typedef iForwardJZ<s32> iForwardJZ32;
typedef iForwardJE<s8> iForwardJE8;
typedef iForwardJE<s32> iForwardJE32;
typedef iForwardJNZ<s8> iForwardJNZ8;
typedef iForwardJNZ<s32> iForwardJNZ32;
typedef iForwardJNE<s8> iForwardJNE8;
typedef iForwardJNE<s32> iForwardJNE32;
typedef xForwardJZ<s8> xForwardJZ8;
typedef xForwardJZ<s32> xForwardJZ32;
typedef xForwardJE<s8> xForwardJE8;
typedef xForwardJE<s32> xForwardJE32;
typedef xForwardJNZ<s8> xForwardJNZ8;
typedef xForwardJNZ<s32> xForwardJNZ32;
typedef xForwardJNE<s8> xForwardJNE8;
typedef xForwardJNE<s32> xForwardJNE32;
DEFINE_FORWARD_JUMP( JS, Jcc_Signed );
DEFINE_FORWARD_JUMP( JNS, Jcc_Unsigned );
typedef iForwardJS<s8> iForwardJS8;
typedef iForwardJS<s32> iForwardJS32;
typedef iForwardJNS<s8> iForwardJNS8;
typedef iForwardJNS<s32> iForwardJNS32;
typedef xForwardJS<s8> xForwardJS8;
typedef xForwardJS<s32> xForwardJS32;
typedef xForwardJNS<s8> xForwardJNS8;
typedef xForwardJNS<s32> xForwardJNS32;
DEFINE_FORWARD_JUMP( JO, Jcc_Overflow );
DEFINE_FORWARD_JUMP( JNO, Jcc_NotOverflow );
typedef iForwardJO<s8> iForwardJO8;
typedef iForwardJO<s32> iForwardJO32;
typedef iForwardJNO<s8> iForwardJNO8;
typedef iForwardJNO<s32> iForwardJNO32;
typedef xForwardJO<s8> xForwardJO8;
typedef xForwardJO<s32> xForwardJO32;
typedef xForwardJNO<s8> xForwardJNO8;
typedef xForwardJNO<s32> xForwardJNO32;
DEFINE_FORWARD_JUMP( JC, Jcc_Carry );
DEFINE_FORWARD_JUMP( JNC, Jcc_NotCarry );
typedef iForwardJC<s8> iForwardJC8;
typedef iForwardJC<s32> iForwardJC32;
typedef iForwardJNC<s8> iForwardJNC8;
typedef iForwardJNC<s32> iForwardJNC32;
typedef xForwardJC<s8> xForwardJC8;
typedef xForwardJC<s32> xForwardJC32;
typedef xForwardJNC<s8> xForwardJNC8;
typedef xForwardJNC<s32> xForwardJNC32;
DEFINE_FORWARD_JUMP( JPE, Jcc_ParityEven );
DEFINE_FORWARD_JUMP( JPO, Jcc_ParityOdd );
typedef iForwardJPE<s8> iForwardJPE8;
typedef iForwardJPE<s32> iForwardJPE32;
typedef iForwardJPO<s8> iForwardJPO8;
typedef iForwardJPO<s32> iForwardJPO32;
typedef xForwardJPE<s8> xForwardJPE8;
typedef xForwardJPE<s32> xForwardJPE32;
typedef xForwardJPO<s8> xForwardJPO8;
typedef xForwardJPO<s32> xForwardJPO32;
//////////////////////////////////////////////////////////////////////////////////////////
// MMX Mov Instructions (MOVD, MOVQ, MOVSS).
@ -332,53 +312,53 @@ namespace x86Emitter
// MOVD has valid forms for MMX and XMM registers.
//
template< typename T >
__emitinline void iMOVDZX( const iRegisterSIMD<T>& to, const iRegister32& from )
__emitinline void xMOVDZX( const xRegisterSIMD<T>& to, const xRegister32& from )
{
Internal::writeXMMop( 0x66, 0x6e, to, from );
}
template< typename T >
__emitinline void iMOVDZX( const iRegisterSIMD<T>& to, const void* src )
__emitinline void xMOVDZX( const xRegisterSIMD<T>& to, const void* src )
{
Internal::writeXMMop( 0x66, 0x6e, to, src );
}
template< typename T >
void iMOVDZX( const iRegisterSIMD<T>& to, const ModSibBase& src )
void xMOVDZX( const xRegisterSIMD<T>& to, const ModSibBase& src )
{
Internal::writeXMMop( 0x66, 0x6e, to, src );
}
template< typename T >
__emitinline void iMOVD( const iRegister32& to, const iRegisterSIMD<T>& from )
__emitinline void xMOVD( const xRegister32& to, const xRegisterSIMD<T>& from )
{
Internal::writeXMMop( 0x66, 0x7e, from, to );
}
template< typename T >
__emitinline void iMOVD( void* dest, const iRegisterSIMD<T>& from )
__emitinline void xMOVD( void* dest, const xRegisterSIMD<T>& from )
{
Internal::writeXMMop( 0x66, 0x7e, from, dest );
}
template< typename T >
void iMOVD( const ModSibBase& dest, const iRegisterSIMD<T>& from )
void xMOVD( const ModSibBase& dest, const xRegisterSIMD<T>& from )
{
Internal::writeXMMop( 0x66, 0x7e, from, dest );
}
// ------------------------------------------------------------------------
// iMASKMOV:
// xMASKMOV:
// Selectively write bytes from mm1/xmm1 to memory location using the byte mask in mm2/xmm2.
// The default memory location is specified by DS:EDI. The most significant bit in each byte
// of the mask operand determines whether the corresponding byte in the source operand is
// written to the corresponding byte location in memory.
template< typename T >
static __forceinline void iMASKMOV( const iRegisterSIMD<T>& to, const iRegisterSIMD<T>& from ) { Internal::writeXMMop( 0x66, 0xf7, to, from ); }
static __forceinline void xMASKMOV( const xRegisterSIMD<T>& to, const xRegisterSIMD<T>& from ) { Internal::writeXMMop( 0x66, 0xf7, to, from ); }
// iPMOVMSKB:
// xPMOVMSKB:
// Creates a mask made up of the most significant bit of each byte of the source
// operand and stores the result in the low byte or word of the destination operand.
// Upper bits of the destination are cleared to zero.
@ -387,93 +367,91 @@ namespace x86Emitter
// 128-bit (SSE) source, the byte mask is 16-bits.
//
template< typename T >
static __forceinline void iPMOVMSKB( const iRegister32& to, const iRegisterSIMD<T>& from ) { Internal::writeXMMop( 0x66, 0xd7, to, from ); }
static __forceinline void xPMOVMSKB( const xRegister32& to, const xRegisterSIMD<T>& from ) { Internal::writeXMMop( 0x66, 0xd7, to, from ); }
// ------------------------------------------------------------------------
extern void iMOVQ( const iRegisterMMX& to, const iRegisterMMX& from );
extern void iMOVQ( const iRegisterMMX& to, const iRegisterSSE& from );
extern void iMOVQ( const iRegisterSSE& to, const iRegisterMMX& from );
extern void xMOVQ( const xRegisterMMX& to, const xRegisterMMX& from );
extern void xMOVQ( const xRegisterMMX& to, const xRegisterSSE& from );
extern void xMOVQ( const xRegisterSSE& to, const xRegisterMMX& from );
extern void iMOVQ( void* dest, const iRegisterSSE& from );
extern void iMOVQ( const ModSibBase& dest, const iRegisterSSE& from );
extern void iMOVQ( void* dest, const iRegisterMMX& from );
extern void iMOVQ( const ModSibBase& dest, const iRegisterMMX& from );
extern void iMOVQ( const iRegisterMMX& to, const void* src );
extern void iMOVQ( const iRegisterMMX& to, const ModSibBase& src );
extern void xMOVQ( void* dest, const xRegisterSSE& from );
extern void xMOVQ( const ModSibBase& dest, const xRegisterSSE& from );
extern void xMOVQ( void* dest, const xRegisterMMX& from );
extern void xMOVQ( const ModSibBase& dest, const xRegisterMMX& from );
extern void xMOVQ( const xRegisterMMX& to, const void* src );
extern void xMOVQ( const xRegisterMMX& to, const ModSibBase& src );
extern void iMOVQZX( const iRegisterSSE& to, const void* src );
extern void iMOVQZX( const iRegisterSSE& to, const ModSibBase& src );
extern void iMOVQZX( const iRegisterSSE& to, const iRegisterSSE& from );
extern void xMOVQZX( const xRegisterSSE& to, const void* src );
extern void xMOVQZX( const xRegisterSSE& to, const ModSibBase& src );
extern void xMOVQZX( const xRegisterSSE& to, const xRegisterSSE& from );
extern void iMOVSS( const iRegisterSSE& to, const iRegisterSSE& from );
extern void iMOVSS( const void* to, const iRegisterSSE& from );
extern void iMOVSS( const ModSibBase& to, const iRegisterSSE& from );
extern void iMOVSD( const iRegisterSSE& to, const iRegisterSSE& from );
extern void iMOVSD( const void* to, const iRegisterSSE& from );
extern void iMOVSD( const ModSibBase& to, const iRegisterSSE& from );
extern void xMOVSS( const xRegisterSSE& to, const xRegisterSSE& from );
extern void xMOVSS( const void* to, const xRegisterSSE& from );
extern void xMOVSS( const ModSibBase& to, const xRegisterSSE& from );
extern void xMOVSD( const xRegisterSSE& to, const xRegisterSSE& from );
extern void xMOVSD( const void* to, const xRegisterSSE& from );
extern void xMOVSD( const ModSibBase& to, const xRegisterSSE& from );
extern void iMOVSSZX( const iRegisterSSE& to, const void* from );
extern void iMOVSSZX( const iRegisterSSE& to, const ModSibBase& from );
extern void iMOVSDZX( const iRegisterSSE& to, const void* from );
extern void iMOVSDZX( const iRegisterSSE& to, const ModSibBase& from );
extern void xMOVSSZX( const xRegisterSSE& to, const void* from );
extern void xMOVSSZX( const xRegisterSSE& to, const ModSibBase& from );
extern void xMOVSDZX( const xRegisterSSE& to, const void* from );
extern void xMOVSDZX( const xRegisterSSE& to, const ModSibBase& from );
extern void iMOVNTDQA( const iRegisterSSE& to, const void* from );
extern void iMOVNTDQA( const iRegisterSSE& to, const ModSibBase& from );
extern void iMOVNTDQ( void* to, const iRegisterSSE& from );
extern void iMOVNTDQA( const ModSibBase& to, const iRegisterSSE& from );
extern void xMOVNTDQA( const xRegisterSSE& to, const void* from );
extern void xMOVNTDQA( const xRegisterSSE& to, const ModSibBase& from );
extern void xMOVNTDQ( void* to, const xRegisterSSE& from );
extern void xMOVNTDQA( const ModSibBase& to, const xRegisterSSE& from );
extern void iMOVNTPD( void* to, const iRegisterSSE& from );
extern void iMOVNTPD( const ModSibBase& to, const iRegisterSSE& from );
extern void iMOVNTPS( void* to, const iRegisterSSE& from );
extern void iMOVNTPS( const ModSibBase& to, const iRegisterSSE& from );
extern void iMOVNTQ( void* to, const iRegisterMMX& from );
extern void iMOVNTQ( const ModSibBase& to, const iRegisterMMX& from );
extern void iMOVLHPS( const iRegisterSSE& to, const iRegisterSSE& from );
extern void iMOVHLPS( const iRegisterSSE& to, const iRegisterSSE& from );
extern void iMOVLHPD( const iRegisterSSE& to, const iRegisterSSE& from );
extern void iMOVHLPD( const iRegisterSSE& to, const iRegisterSSE& from );
extern void xMOVNTPD( void* to, const xRegisterSSE& from );
extern void xMOVNTPD( const ModSibBase& to, const xRegisterSSE& from );
extern void xMOVNTPS( void* to, const xRegisterSSE& from );
extern void xMOVNTPS( const ModSibBase& to, const xRegisterSSE& from );
extern void xMOVNTQ( void* to, const xRegisterMMX& from );
extern void xMOVNTQ( const ModSibBase& to, const xRegisterMMX& from );
//////////////////////////////////////////////////////////////////////////////////////////
//
extern const Internal::MovapsImplAll<0, 0x28, 0x29> iMOVAPS;
extern const Internal::MovapsImplAll<0, 0x10, 0x11> iMOVUPS;
extern const Internal::MovapsImplAll<0, 0x28, 0x29> xMOVAPS;
extern const Internal::MovapsImplAll<0, 0x10, 0x11> xMOVUPS;
extern const Internal::MovapsImplAll<0x66, 0x28, 0x29> iMOVAPD;
extern const Internal::MovapsImplAll<0x66, 0x10, 0x11> iMOVUPD;
extern const Internal::MovapsImplAll<0x66, 0x28, 0x29> xMOVAPD;
extern const Internal::MovapsImplAll<0x66, 0x10, 0x11> xMOVUPD;
#ifdef ALWAYS_USE_MOVAPS
extern const Internal::MovapsImplAll<0x66, 0x6f, 0x7f> iMOVDQA;
extern const Internal::MovapsImplAll<0xf3, 0x6f, 0x7f> iMOVDQU;
extern const Internal::MovapsImplAll<0x66, 0x6f, 0x7f> xMOVDQA;
extern const Internal::MovapsImplAll<0xf3, 0x6f, 0x7f> xMOVDQU;
#else
extern const Internal::MovapsImplAll<0, 0x28, 0x29> iMOVDQA;
extern const Internal::MovapsImplAll<0, 0x10, 0x11> iMOVDQU;
extern const Internal::MovapsImplAll<0, 0x28, 0x29> xMOVDQA;
extern const Internal::MovapsImplAll<0, 0x10, 0x11> xMOVDQU;
#endif
extern const Internal::MovhlImplAll<0, 0x16> iMOVHPS;
extern const Internal::MovhlImplAll<0, 0x12> iMOVLPS;
extern const Internal::MovhlImplAll<0x66, 0x16> iMOVHPD;
extern const Internal::MovhlImplAll<0x66, 0x12> iMOVLPD;
extern const Internal::MovhlImpl_RtoR<0x16> xMOVLH;
extern const Internal::MovhlImpl_RtoR<0x12> xMOVHL;
extern const Internal::PLogicImplAll<0xdb> iPAND;
extern const Internal::PLogicImplAll<0xdf> iPANDN;
extern const Internal::PLogicImplAll<0xeb> iPOR;
extern const Internal::PLogicImplAll<0xef> iPXOR;
extern const Internal::MovhlImplAll<0x16> xMOVH;
extern const Internal::MovhlImplAll<0x12> xMOVL;
extern const Internal::SSELogicImpl<0,0x53> iRCPPS;
extern const Internal::SSELogicImpl<0xf3,0x53> iRCPSS;
extern const Internal::PLogicImplAll<0xdb> xPAND;
extern const Internal::PLogicImplAll<0xdf> xPANDN;
extern const Internal::PLogicImplAll<0xeb> xPOR;
extern const Internal::PLogicImplAll<0xef> xPXOR;
extern const Internal::SSECompareImplGeneric<0x00> iCMPPS;
extern const Internal::SSECompareImplGeneric<0x66> iCMPPD;
extern const Internal::SSECompareImplGeneric<0xf3> iCMPSS;
extern const Internal::SSECompareImplGeneric<0xf2> iCMPSD;
extern const Internal::SSECompareImplGeneric<0x00> iCMPPS;
extern const Internal::SSECompareImplGeneric<0x66> iCMPPD;
extern const Internal::SSECompareImplGeneric<0xf3> iCMPSS;
extern const Internal::SSECompareImplGeneric<0xf2> iCMPSD;
extern const Internal::SSEAndNotImpl<0x55> xANDN;
extern const Internal::SSELogicImpl<0,0x53> xRCPPS;
extern const Internal::SSELogicImpl<0xf3,0x53> xRCPSS;
// ------------------------------------------------------------------------
extern const Internal::SSECompareImpl<SSE2_Equal> xCMPEQ;
extern const Internal::SSECompareImpl<SSE2_Less> xCMPLT;
extern const Internal::SSECompareImpl<SSE2_LessOrEqual> xCMPLE;
extern const Internal::SSECompareImpl<SSE2_Unordered> xCMPUNORD;
extern const Internal::SSECompareImpl<SSE2_NotEqual> xCMPNE;
extern const Internal::SSECompareImpl<SSE2_NotLess> xCMPNLT;
extern const Internal::SSECompareImpl<SSE2_NotLessOrEqual> xCMPNLE;
extern const Internal::SSECompareImpl<SSE2_Ordered> xCMPORD;
}

View File

@ -40,11 +40,11 @@ namespace x86Emitter {
using namespace Internal;
const JmpCallImplAll<true> iJMP;
const JmpCallImplAll<false> iCALL;
const JmpCallImplAll<true> xJMP;
const JmpCallImplAll<false> xCALL;
// ------------------------------------------------------------------------
void iSmartJump::SetTarget()
void xSmartJump::SetTarget()
{
u8* target = iGetPtr();
if( m_baseptr == NULL ) return;
@ -67,7 +67,7 @@ void iSmartJump::SetTarget()
}
}
iSmartJump::~iSmartJump()
xSmartJump::~xSmartJump()
{
SetTarget();
m_baseptr = NULL; // just in case (sometimes helps in debugging too)
@ -78,7 +78,7 @@ iSmartJump::~iSmartJump()
// Writes a jump at the current x86Ptr, which targets a pre-established target address.
// (usually a backwards jump)
//
// slideForward - used internally by iSmartJump to indicate that the jump target is going
// slideForward - used internally by xSmartJump to indicate that the jump target is going
// to slide forward in the event of an 8 bit displacement.
//
// Using this
@ -96,21 +96,21 @@ __emitinline void iJccKnownTarget( JccComparisonType comparison, void* target, b
if( is_s8( displacement8 ) )
{
iWrite<u8>( (comparison == Jcc_Unconditional) ? 0xeb : (0x70 | comparison) );
iWrite<s8>( displacement8 );
xWrite<u8>( (comparison == Jcc_Unconditional) ? 0xeb : (0x70 | comparison) );
xWrite<s8>( displacement8 );
}
else
{
// Perform a 32 bit jump instead. :(
if( comparison == Jcc_Unconditional )
iWrite<u8>( 0xe9 );
xWrite<u8>( 0xe9 );
else
{
iWrite<u8>( 0x0f );
iWrite<u8>( 0x80 | comparison );
xWrite<u8>( 0x0f );
xWrite<u8>( 0x80 | comparison );
}
iWrite<s32>( (sptr)target - ((sptr)iGetPtr() + 4) );
xWrite<s32>( (sptr)target - ((sptr)iGetPtr() + 4) );
}
}

View File

@ -35,9 +35,9 @@
using namespace x86Emitter;
template< typename ImmType >
static __forceinline iRegister<ImmType> _reghlp( x86IntRegType src )
static __forceinline xRegister<ImmType> _reghlp( x86IntRegType src )
{
return iRegister<ImmType>( src );
return xRegister<ImmType>( src );
}
static __forceinline ModSibBase _mrmhlp( x86IntRegType src )
@ -48,43 +48,43 @@ static __forceinline ModSibBase _mrmhlp( x86IntRegType src )
template< typename ImmType >
static __forceinline ModSibStrict<ImmType> _mhlp( x86IntRegType src )
{
return ModSibStrict<ImmType>( iAddressReg::Empty, iAddressReg(src) );
return ModSibStrict<ImmType>( xAddressReg::Empty, xAddressReg(src) );
}
template< typename ImmType >
static __forceinline ModSibStrict<ImmType> _mhlp2( x86IntRegType src1, x86IntRegType src2 )
{
return ModSibStrict<ImmType>( iAddressReg(src2), iAddressReg(src1) );
return ModSibStrict<ImmType>( xAddressReg(src2), xAddressReg(src1) );
}
//////////////////////////////////////////////////////////////////////////////////////////
//
#define DEFINE_LEGACY_HELPER( cod, bits ) \
emitterT void cod##bits##RtoR( x86IntRegType to, x86IntRegType from ) { i##cod( _reghlp<u##bits>(to), _reghlp<u##bits>(from) ); } \
emitterT void cod##bits##ItoR( x86IntRegType to, u##bits imm ) { i##cod( _reghlp<u##bits>(to), imm ); } \
emitterT void cod##bits##MtoR( x86IntRegType to, uptr from ) { i##cod( _reghlp<u##bits>(to), (void*)from ); } \
emitterT void cod##bits##RtoM( uptr to, x86IntRegType from ) { i##cod( (void*)to, _reghlp<u##bits>(from) ); } \
emitterT void cod##bits##ItoM( uptr to, u##bits imm ) { i##cod( ptr##bits[to], imm ); } \
emitterT void cod##bits##ItoRm( x86IntRegType to, u##bits imm, int offset ) { i##cod( _mhlp<u##bits>(to) + offset, imm ); } \
emitterT void cod##bits##RmtoR( x86IntRegType to, x86IntRegType from, int offset ) { i##cod( _reghlp<u##bits>(to), _mhlp<u##bits>(from) + offset ); } \
emitterT void cod##bits##RtoRm( x86IntRegType to, x86IntRegType from, int offset ) { i##cod( _mhlp<u##bits>(to) + offset, _reghlp<u##bits>(from) ); } \
emitterT void cod##bits##RtoR( x86IntRegType to, x86IntRegType from ) { x##cod( _reghlp<u##bits>(to), _reghlp<u##bits>(from) ); } \
emitterT void cod##bits##ItoR( x86IntRegType to, u##bits imm ) { x##cod( _reghlp<u##bits>(to), imm ); } \
emitterT void cod##bits##MtoR( x86IntRegType to, uptr from ) { x##cod( _reghlp<u##bits>(to), (void*)from ); } \
emitterT void cod##bits##RtoM( uptr to, x86IntRegType from ) { x##cod( (void*)to, _reghlp<u##bits>(from) ); } \
emitterT void cod##bits##ItoM( uptr to, u##bits imm ) { x##cod( ptr##bits[to], imm ); } \
emitterT void cod##bits##ItoRm( x86IntRegType to, u##bits imm, int offset ) { x##cod( _mhlp<u##bits>(to) + offset, imm ); } \
emitterT void cod##bits##RmtoR( x86IntRegType to, x86IntRegType from, int offset ) { x##cod( _reghlp<u##bits>(to), _mhlp<u##bits>(from) + offset ); } \
emitterT void cod##bits##RtoRm( x86IntRegType to, x86IntRegType from, int offset ) { x##cod( _mhlp<u##bits>(to) + offset, _reghlp<u##bits>(from) ); } \
emitterT void cod##bits##RtoRmS( x86IntRegType to1, x86IntRegType to2, x86IntRegType from, int offset ) \
{ i##cod( _mhlp2<u##bits>(to1,to2) + offset, _reghlp<u##bits>(from) ); } \
{ x##cod( _mhlp2<u##bits>(to1,to2) + offset, _reghlp<u##bits>(from) ); } \
emitterT void cod##bits##RmStoR( x86IntRegType to, x86IntRegType from1, x86IntRegType from2, int offset ) \
{ i##cod( _reghlp<u##bits>(to), _mhlp2<u##bits>(from1,from2) + offset ); }
{ x##cod( _reghlp<u##bits>(to), _mhlp2<u##bits>(from1,from2) + offset ); }
#define DEFINE_LEGACY_SHIFT_HELPER( cod, bits ) \
emitterT void cod##bits##CLtoR( x86IntRegType to ) { i##cod( _reghlp<u##bits>(to), cl ); } \
emitterT void cod##bits##ItoR( x86IntRegType to, u8 imm ) { i##cod( _reghlp<u##bits>(to), imm ); } \
emitterT void cod##bits##CLtoM( uptr to ) { i##cod( ptr##bits[to], cl ); } \
emitterT void cod##bits##ItoM( uptr to, u8 imm ) { i##cod( ptr##bits[to], imm ); } \
emitterT void cod##bits##ItoRm( x86IntRegType to, u8 imm, int offset ) { i##cod( _mhlp<u##bits>(to) + offset, imm ); } \
emitterT void cod##bits##CLtoRm( x86IntRegType to, int offset ) { i##cod( _mhlp<u##bits>(to) + offset, cl ); }
emitterT void cod##bits##CLtoR( x86IntRegType to ) { x##cod( _reghlp<u##bits>(to), cl ); } \
emitterT void cod##bits##ItoR( x86IntRegType to, u8 imm ) { x##cod( _reghlp<u##bits>(to), imm ); } \
emitterT void cod##bits##CLtoM( uptr to ) { x##cod( ptr##bits[to], cl ); } \
emitterT void cod##bits##ItoM( uptr to, u8 imm ) { x##cod( ptr##bits[to], imm ); } \
emitterT void cod##bits##ItoRm( x86IntRegType to, u8 imm, int offset ) { x##cod( _mhlp<u##bits>(to) + offset, imm ); } \
emitterT void cod##bits##CLtoRm( x86IntRegType to, int offset ) { x##cod( _mhlp<u##bits>(to) + offset, cl ); }
#define DEFINE_LEGACY_ONEREG_HELPER( cod, bits ) \
emitterT void cod##bits##R( x86IntRegType to ) { i##cod( _reghlp<u##bits>(to) ); } \
emitterT void cod##bits##M( uptr to ) { i##cod( ptr##bits[to] ); } \
emitterT void cod##bits##Rm( x86IntRegType to, uptr offset ) { i##cod( _mhlp<u##bits>(to) + offset ); }
emitterT void cod##bits##R( x86IntRegType to ) { x##cod( _reghlp<u##bits>(to) ); } \
emitterT void cod##bits##M( uptr to ) { x##cod( ptr##bits[to] ); } \
emitterT void cod##bits##Rm( x86IntRegType to, uptr offset ) { x##cod( _mhlp<u##bits>(to) + offset ); }
//emitterT void cod##bits##RtoRmS( x86IntRegType to1, x86IntRegType to2, x86IntRegType from, int offset ) \
// { cod( _mhlp2<u##bits>(to1,to2) + offset, _reghlp<u##bits>(from) ); } \
@ -133,9 +133,9 @@ DEFINE_OPCODE_ONEREG_LEGACY( NEG )
// ------------------------------------------------------------------------
#define DEFINE_LEGACY_MOVEXTEND( form, destbits, srcbits ) \
emitterT void MOV##form##destbits##R##srcbits##toR( x86IntRegType to, x86IntRegType from ) { iMOV##form( iRegister##destbits( to ), iRegister##srcbits( from ) ); } \
emitterT void MOV##form##destbits##Rm##srcbits##toR( x86IntRegType to, x86IntRegType from, int offset ) { iMOV##form( iRegister##destbits( to ), ptr##srcbits[iAddressReg( from ) + offset] ); } \
emitterT void MOV##form##destbits##M##srcbits##toR( x86IntRegType to, u32 from ) { iMOV##form( iRegister##destbits( to ), ptr##srcbits[from] ); }
emitterT void MOV##form##destbits##R##srcbits##toR( x86IntRegType to, x86IntRegType from ) { xMOV##form( xRegister##destbits( to ), xRegister##srcbits( from ) ); } \
emitterT void MOV##form##destbits##Rm##srcbits##toR( x86IntRegType to, x86IntRegType from, int offset ) { xMOV##form( xRegister##destbits( to ), ptr##srcbits[xAddressReg( from ) + offset] ); } \
emitterT void MOV##form##destbits##M##srcbits##toR( x86IntRegType to, u32 from ) { xMOV##form( xRegister##destbits( to ), ptr##srcbits[from] ); }
DEFINE_LEGACY_MOVEXTEND( SX, 32, 16 )
DEFINE_LEGACY_MOVEXTEND( ZX, 32, 16 )
@ -145,164 +145,164 @@ DEFINE_LEGACY_MOVEXTEND( ZX, 32, 8 )
DEFINE_LEGACY_MOVEXTEND( SX, 16, 8 )
DEFINE_LEGACY_MOVEXTEND( ZX, 16, 8 )
emitterT void TEST32ItoR( x86IntRegType to, u32 from ) { iTEST( iRegister32(to), from ); }
emitterT void TEST32ItoM( uptr to, u32 from ) { iTEST( ptr32[to], from ); }
emitterT void TEST32RtoR( x86IntRegType to, x86IntRegType from ) { iTEST( iRegister32(to), iRegister32(from) ); }
emitterT void TEST32ItoRm( x86IntRegType to, u32 from ) { iTEST( ptr32[iAddressReg(to)], from ); }
emitterT void TEST32ItoR( x86IntRegType to, u32 from ) { xTEST( xRegister32(to), from ); }
emitterT void TEST32ItoM( uptr to, u32 from ) { xTEST( ptr32[to], from ); }
emitterT void TEST32RtoR( x86IntRegType to, x86IntRegType from ) { xTEST( xRegister32(to), xRegister32(from) ); }
emitterT void TEST32ItoRm( x86IntRegType to, u32 from ) { xTEST( ptr32[xAddressReg(to)], from ); }
emitterT void TEST16ItoR( x86IntRegType to, u16 from ) { iTEST( iRegister16(to), from ); }
emitterT void TEST16ItoM( uptr to, u16 from ) { iTEST( ptr16[to], from ); }
emitterT void TEST16RtoR( x86IntRegType to, x86IntRegType from ) { iTEST( iRegister16(to), iRegister16(from) ); }
emitterT void TEST16ItoRm( x86IntRegType to, u16 from ) { iTEST( ptr16[iAddressReg(to)], from ); }
emitterT void TEST16ItoR( x86IntRegType to, u16 from ) { xTEST( xRegister16(to), from ); }
emitterT void TEST16ItoM( uptr to, u16 from ) { xTEST( ptr16[to], from ); }
emitterT void TEST16RtoR( x86IntRegType to, x86IntRegType from ) { xTEST( xRegister16(to), xRegister16(from) ); }
emitterT void TEST16ItoRm( x86IntRegType to, u16 from ) { xTEST( ptr16[xAddressReg(to)], from ); }
emitterT void TEST8ItoR( x86IntRegType to, u8 from ) { iTEST( iRegister8(to), from ); }
emitterT void TEST8ItoM( uptr to, u8 from ) { iTEST( ptr8[to], from ); }
emitterT void TEST8RtoR( x86IntRegType to, x86IntRegType from ) { iTEST( iRegister8(to), iRegister8(from) ); }
emitterT void TEST8ItoRm( x86IntRegType to, u8 from ) { iTEST( ptr8[iAddressReg(to)], from ); }
emitterT void TEST8ItoR( x86IntRegType to, u8 from ) { xTEST( xRegister8(to), from ); }
emitterT void TEST8ItoM( uptr to, u8 from ) { xTEST( ptr8[to], from ); }
emitterT void TEST8RtoR( x86IntRegType to, x86IntRegType from ) { xTEST( xRegister8(to), xRegister8(from) ); }
emitterT void TEST8ItoRm( x86IntRegType to, u8 from ) { xTEST( ptr8[xAddressReg(to)], from ); }
// mov r32 to [r32<<scale+from2]
emitterT void MOV32RmSOffsettoR( x86IntRegType to, x86IntRegType from1, s32 from2, int scale )
{
iMOV( iRegister32(to), ptr[(iAddressReg(from1)<<scale) + from2] );
xMOV( xRegister32(to), ptr[(xAddressReg(from1)<<scale) + from2] );
}
emitterT void MOV16RmSOffsettoR( x86IntRegType to, x86IntRegType from1, s32 from2, int scale )
{
iMOV( iRegister16(to), ptr[(iAddressReg(from1)<<scale) + from2] );
xMOV( xRegister16(to), ptr[(xAddressReg(from1)<<scale) + from2] );
}
emitterT void MOV8RmSOffsettoR( x86IntRegType to, x86IntRegType from1, s32 from2, int scale )
{
iMOV( iRegister8(to), ptr[(iAddressReg(from1)<<scale) + from2] );
xMOV( xRegister8(to), ptr[(xAddressReg(from1)<<scale) + from2] );
}
emitterT void AND32I8toR( x86IntRegType to, s8 from )
{
iAND( _reghlp<u32>(to), from );
xAND( _reghlp<u32>(to), from );
}
emitterT void AND32I8toM( uptr to, s8 from )
{
iAND( ptr8[to], from );
xAND( ptr8[to], from );
}
/* cmove r32 to r32*/
emitterT void CMOVE32RtoR( x86IntRegType to, x86IntRegType from )
{
iCMOVE( iRegister32(to), iRegister32(from) );
xCMOVE( xRegister32(to), xRegister32(from) );
}
// shld imm8 to r32
emitterT void SHLD32ItoR( x86IntRegType to, x86IntRegType from, u8 shift )
{
iSHLD( iRegister32(to), iRegister32(from), shift );
xSHLD( xRegister32(to), xRegister32(from), shift );
}
// shrd imm8 to r32
emitterT void SHRD32ItoR( x86IntRegType to, x86IntRegType from, u8 shift )
{
iSHRD( iRegister32(to), iRegister32(from), shift );
xSHRD( xRegister32(to), xRegister32(from), shift );
}
/* mul eax by r32 to edx:eax */
emitterT void MUL32R( x86IntRegType from ) { iUMUL( iRegister32(from) ); }
emitterT void MUL32R( x86IntRegType from ) { xUMUL( xRegister32(from) ); }
/* imul eax by r32 to edx:eax */
emitterT void IMUL32R( x86IntRegType from ) { iMUL( iRegister32(from) ); }
emitterT void IMUL32R( x86IntRegType from ) { xMUL( xRegister32(from) ); }
/* mul eax by m32 to edx:eax */
emitterT void MUL32M( u32 from ) { iUMUL( ptr32[from] ); }
emitterT void MUL32M( u32 from ) { xUMUL( ptr32[from] ); }
/* imul eax by m32 to edx:eax */
emitterT void IMUL32M( u32 from ) { iMUL( ptr32[from] ); }
emitterT void IMUL32M( u32 from ) { xMUL( ptr32[from] ); }
/* imul r32 by r32 to r32 */
emitterT void IMUL32RtoR( x86IntRegType to, x86IntRegType from )
{
iMUL( iRegister32(to), iRegister32(from) );
xMUL( xRegister32(to), xRegister32(from) );
}
/* div eax by r32 to edx:eax */
emitterT void DIV32R( x86IntRegType from ) { iUDIV( iRegister32(from) ); }
emitterT void DIV32R( x86IntRegType from ) { xUDIV( xRegister32(from) ); }
/* idiv eax by r32 to edx:eax */
emitterT void IDIV32R( x86IntRegType from ) { iDIV( iRegister32(from) ); }
emitterT void IDIV32R( x86IntRegType from ) { xDIV( xRegister32(from) ); }
/* div eax by m32 to edx:eax */
emitterT void DIV32M( u32 from ) { iUDIV( ptr32[from] ); }
emitterT void DIV32M( u32 from ) { xUDIV( ptr32[from] ); }
/* idiv eax by m32 to edx:eax */
emitterT void IDIV32M( u32 from ) { iDIV( ptr32[from] ); }
emitterT void IDIV32M( u32 from ) { xDIV( ptr32[from] ); }
emitterT void LEA32RtoR(x86IntRegType to, x86IntRegType from, s32 offset)
{
iLEA( iRegister32( to ), ptr[iAddressReg(from)+offset] );
xLEA( xRegister32( to ), ptr[xAddressReg(from)+offset] );
}
emitterT void LEA32RRtoR(x86IntRegType to, x86IntRegType from0, x86IntRegType from1)
{
iLEA( iRegister32( to ), ptr[iAddressReg(from0)+iAddressReg(from1)] );
xLEA( xRegister32( to ), ptr[xAddressReg(from0)+xAddressReg(from1)] );
}
// Don't inline recursive functions
emitterT void LEA32RStoR(x86IntRegType to, x86IntRegType from, u32 scale)
{
iLEA( iRegister32( to ), ptr[iAddressReg(from)*(1<<scale)] );
xLEA( xRegister32( to ), ptr[xAddressReg(from)*(1<<scale)] );
}
// to = from + offset
emitterT void LEA16RtoR(x86IntRegType to, x86IntRegType from, s16 offset)
{
iLEA( iRegister16( to ), ptr[iAddressReg(from)+offset] );
xLEA( xRegister16( to ), ptr[xAddressReg(from)+offset] );
}
// to = from0 + from1
emitterT void LEA16RRtoR(x86IntRegType to, x86IntRegType from0, x86IntRegType from1)
{
iLEA( iRegister16( to ), ptr[iAddressReg(from0)+iAddressReg(from1)] );
xLEA( xRegister16( to ), ptr[xAddressReg(from0)+xAddressReg(from1)] );
}
// to = from << scale (max is 3)
emitterT void LEA16RStoR(x86IntRegType to, x86IntRegType from, u32 scale)
{
iLEA( iRegister16( to ), ptr[iAddressReg(from)*(1<<scale)] );
xLEA( xRegister16( to ), ptr[xAddressReg(from)*(1<<scale)] );
}
emitterT void BT32ItoR( x86IntRegType to, u8 from ) { iBT( iRegister32(to), from ); }
emitterT void BTR32ItoR( x86IntRegType to, u8 from ) { iBTR( iRegister32(to), from ); }
emitterT void BT32ItoR( x86IntRegType to, u8 from ) { xBT( xRegister32(to), from ); }
emitterT void BTR32ItoR( x86IntRegType to, u8 from ) { xBTR( xRegister32(to), from ); }
emitterT void SETS8R( x86IntRegType to ) { iSETS( iRegister8(to) ); }
emitterT void SETL8R( x86IntRegType to ) { iSETL( iRegister8(to) ); }
emitterT void SETGE8R( x86IntRegType to ) { iSETGE( iRegister8(to) ); }
emitterT void SETG8R( x86IntRegType to ) { iSETG( iRegister8(to) ); }
emitterT void SETA8R( x86IntRegType to ) { iSETA( iRegister8(to) ); }
emitterT void SETAE8R( x86IntRegType to ) { iSETAE( iRegister8(to) ); }
emitterT void SETB8R( x86IntRegType to ) { iSETB( iRegister8(to) ); }
emitterT void SETNZ8R( x86IntRegType to ) { iSETNZ( iRegister8(to) ); }
emitterT void SETZ8R( x86IntRegType to ) { iSETZ( iRegister8(to) ); }
emitterT void SETE8R( x86IntRegType to ) { iSETE( iRegister8(to) ); }
emitterT void SETS8R( x86IntRegType to ) { xSETS( xRegister8(to) ); }
emitterT void SETL8R( x86IntRegType to ) { xSETL( xRegister8(to) ); }
emitterT void SETGE8R( x86IntRegType to ) { xSETGE( xRegister8(to) ); }
emitterT void SETG8R( x86IntRegType to ) { xSETG( xRegister8(to) ); }
emitterT void SETA8R( x86IntRegType to ) { xSETA( xRegister8(to) ); }
emitterT void SETAE8R( x86IntRegType to ) { xSETAE( xRegister8(to) ); }
emitterT void SETB8R( x86IntRegType to ) { xSETB( xRegister8(to) ); }
emitterT void SETNZ8R( x86IntRegType to ) { xSETNZ( xRegister8(to) ); }
emitterT void SETZ8R( x86IntRegType to ) { xSETZ( xRegister8(to) ); }
emitterT void SETE8R( x86IntRegType to ) { xSETE( xRegister8(to) ); }
/* push imm32 */
emitterT void PUSH32I( u32 from ) { iPUSH( from ); }
emitterT void PUSH32I( u32 from ) { xPUSH( from ); }
/* push r32 */
emitterT void PUSH32R( x86IntRegType from ) { iPUSH( iRegister32( from ) ); }
emitterT void PUSH32R( x86IntRegType from ) { xPUSH( xRegister32( from ) ); }
/* push m32 */
emitterT void PUSH32M( u32 from )
{
iPUSH( ptr[from] );
xPUSH( ptr[from] );
}
/* pop r32 */
emitterT void POP32R( x86IntRegType from ) { iPOP( iRegister32( from ) ); }
emitterT void PUSHFD( void ) { iPUSHFD(); }
emitterT void POPFD( void ) { iPOPFD(); }
emitterT void POP32R( x86IntRegType from ) { xPOP( xRegister32( from ) ); }
emitterT void PUSHFD( void ) { xPUSHFD(); }
emitterT void POPFD( void ) { xPOPFD(); }
emitterT void RET( void ) { iRET(); }
emitterT void RET( void ) { xRET(); }
emitterT void CBW( void ) { iCBW(); }
emitterT void CWD( void ) { iCWD(); }
emitterT void CDQ( void ) { iCDQ(); }
emitterT void CWDE() { iCWDE(); }
emitterT void CBW( void ) { xCBW(); }
emitterT void CWD( void ) { xCWD(); }
emitterT void CDQ( void ) { xCDQ(); }
emitterT void CWDE() { xCWDE(); }
emitterT void LAHF() { iLAHF(); }
emitterT void SAHF() { iSAHF(); }
emitterT void LAHF() { xLAHF(); }
emitterT void SAHF() { xSAHF(); }
//////////////////////////////////////////////////////////////////////////////////////////
//////////////////////////////////////////////////////////////////////////////////////////
@ -437,9 +437,9 @@ emitterT void x86Align( int bytes )
/* IX86 instructions */
/********************/
emitterT void STC( void ) { iSTC(); }
emitterT void CLC( void ) { iCLC(); }
emitterT void NOP( void ) { iNOP(); }
emitterT void STC( void ) { xSTC(); }
emitterT void CLC( void ) { xCLC(); }
emitterT void NOP( void ) { xNOP(); }
////////////////////////////////////
// jump instructions /
@ -465,13 +465,13 @@ emitterT u32* JMP32( uptr to )
/* jmp r32/r64 */
emitterT void JMPR( x86IntRegType to )
{
iJMP( iRegister32(to) );
xJMP( xRegister32(to) );
}
// jmp m32
emitterT void JMP32M( uptr to )
{
iJMP( ptr32[to] );
xJMP( ptr32[to] );
}
/* jp rel8 */
@ -720,27 +720,27 @@ emitterT u32* JNO32( u32 to )
/* call func */
emitterT void CALLFunc( uptr func )
{
iCALL( (void*)func );
xCALL( (void*)func );
}
/* call r32 */
emitterT void CALL32R( x86IntRegType to )
{
iCALL( iRegister32( to ) );
xCALL( xRegister32( to ) );
}
/* call m32 */
emitterT void CALL32M( u32 to )
{
iCALL( ptr32[to] );
xCALL( ptr32[to] );
}
emitterT void BSRRtoR(x86IntRegType to, x86IntRegType from)
{
iBSR( iRegister32(to), iRegister32(from) );
xBSR( xRegister32(to), xRegister32(from) );
}
emitterT void BSWAP32R( x86IntRegType to )
{
iBSWAP( iRegister32(to) );
xBSWAP( xRegister32(to) );
}

View File

@ -27,26 +27,26 @@
using namespace x86Emitter;
emitterT void MOVQMtoR( x86MMXRegType to, uptr from ) { iMOVQ( iRegisterMMX(to), (void*)from ); }
emitterT void MOVQRtoM( uptr to, x86MMXRegType from ) { iMOVQ( (void*)to, iRegisterMMX(from) ); }
emitterT void MOVQRtoR( x86MMXRegType to, x86MMXRegType from ) { iMOVQ( iRegisterMMX(to), iRegisterMMX(from) ); }
emitterT void MOVQRmtoR( x86MMXRegType to, x86IntRegType from, int offset ) { iMOVQ( iRegisterMMX(to), ptr[iAddressReg(from)+offset] ); }
emitterT void MOVQRtoRm( x86IntRegType to, x86MMXRegType from, int offset ) { iMOVQ( ptr[iAddressReg(to)+offset], iRegisterMMX(from) ); }
emitterT void MOVQMtoR( x86MMXRegType to, uptr from ) { xMOVQ( xRegisterMMX(to), (void*)from ); }
emitterT void MOVQRtoM( uptr to, x86MMXRegType from ) { xMOVQ( (void*)to, xRegisterMMX(from) ); }
emitterT void MOVQRtoR( x86MMXRegType to, x86MMXRegType from ) { xMOVQ( xRegisterMMX(to), xRegisterMMX(from) ); }
emitterT void MOVQRmtoR( x86MMXRegType to, x86IntRegType from, int offset ) { xMOVQ( xRegisterMMX(to), ptr[xAddressReg(from)+offset] ); }
emitterT void MOVQRtoRm( x86IntRegType to, x86MMXRegType from, int offset ) { xMOVQ( ptr[xAddressReg(to)+offset], xRegisterMMX(from) ); }
emitterT void MOVDMtoMMX( x86MMXRegType to, uptr from ) { iMOVDZX( iRegisterMMX(to), (void*)from ); }
emitterT void MOVDMMXtoM( uptr to, x86MMXRegType from ) { iMOVD( (void*)to, iRegisterMMX(from) ); }
emitterT void MOVD32RtoMMX( x86MMXRegType to, x86IntRegType from ) { iMOVDZX( iRegisterMMX(to), iRegister32(from) ); }
emitterT void MOVD32RmtoMMX( x86MMXRegType to, x86IntRegType from, int offset ) { iMOVDZX( iRegisterMMX(to), ptr[iAddressReg(from)+offset] ); }
emitterT void MOVD32MMXtoR( x86IntRegType to, x86MMXRegType from ) { iMOVD( iRegister32(to), iRegisterMMX(from) ); }
emitterT void MOVD32MMXtoRm( x86IntRegType to, x86MMXRegType from, int offset ) { iMOVD( ptr[iAddressReg(to)+offset], iRegisterMMX(from) ); }
emitterT void MOVDMtoMMX( x86MMXRegType to, uptr from ) { xMOVDZX( xRegisterMMX(to), (void*)from ); }
emitterT void MOVDMMXtoM( uptr to, x86MMXRegType from ) { xMOVD( (void*)to, xRegisterMMX(from) ); }
emitterT void MOVD32RtoMMX( x86MMXRegType to, x86IntRegType from ) { xMOVDZX( xRegisterMMX(to), xRegister32(from) ); }
emitterT void MOVD32RmtoMMX( x86MMXRegType to, x86IntRegType from, int offset ) { xMOVDZX( xRegisterMMX(to), ptr[xAddressReg(from)+offset] ); }
emitterT void MOVD32MMXtoR( x86IntRegType to, x86MMXRegType from ) { xMOVD( xRegister32(to), xRegisterMMX(from) ); }
emitterT void MOVD32MMXtoRm( x86IntRegType to, x86MMXRegType from, int offset ) { xMOVD( ptr[xAddressReg(to)+offset], xRegisterMMX(from) ); }
emitterT void PMOVMSKBMMXtoR(x86IntRegType to, x86MMXRegType from) { iPMOVMSKB( iRegister32(to), iRegisterMMX(from) ); }
emitterT void PMOVMSKBMMXtoR(x86IntRegType to, x86MMXRegType from) { xPMOVMSKB( xRegister32(to), xRegisterMMX(from) ); }
#define DEFINE_LEGACY_LOGIC_OPCODE( mod ) \
emitterT void P##mod##RtoR( x86MMXRegType to, x86MMXRegType from ) { iP##mod( iRegisterMMX(to), iRegisterMMX(from) ); } \
emitterT void P##mod##MtoR( x86MMXRegType to, uptr from ) { iP##mod( iRegisterMMX(to), (void*)from ); } \
emitterT void SSE2_P##mod##_XMM_to_XMM( x86SSERegType to, x86SSERegType from ) { iP##mod( iRegisterSSE(to), iRegisterSSE(from) ); } \
emitterT void SSE2_P##mod##_M128_to_XMM( x86SSERegType to, uptr from ) { iP##mod( iRegisterSSE(to), (void*)from ); }
emitterT void P##mod##RtoR( x86MMXRegType to, x86MMXRegType from ) { xP##mod( xRegisterMMX(to), xRegisterMMX(from) ); } \
emitterT void P##mod##MtoR( x86MMXRegType to, uptr from ) { xP##mod( xRegisterMMX(to), (void*)from ); } \
emitterT void SSE2_P##mod##_XMM_to_XMM( x86SSERegType to, x86SSERegType from ) { xP##mod( xRegisterSSE(to), xRegisterSSE(from) ); } \
emitterT void SSE2_P##mod##_M128_to_XMM( x86SSERegType to, uptr from ) { xP##mod( xRegisterSSE(to), (void*)from ); }
DEFINE_LEGACY_LOGIC_OPCODE( AND )
DEFINE_LEGACY_LOGIC_OPCODE( ANDN )
@ -460,4 +460,4 @@ emitterT void PSHUFWMtoR(x86MMXRegType to, uptr from, u8 imm8)
write8(imm8);
}
emitterT void MASKMOVQRtoR(x86MMXRegType to, x86MMXRegType from) { iMASKMOV( iRegisterMMX(to), iRegisterMMX(from) ); }
emitterT void MASKMOVQRtoR(x86MMXRegType to, x86MMXRegType from) { xMASKMOV( xRegisterMMX(to), xRegisterMMX(from) ); }

View File

@ -137,96 +137,95 @@ using namespace x86Emitter;
write8( op )
#define DEFINE_LEGACY_MOV_OPCODE( mod, sse ) \
emitterT void sse##_MOV##mod##_M128_to_XMM( x86SSERegType to, uptr from ) { iMOV##mod( iRegisterSSE(to), (void*)from ); } \
emitterT void sse##_MOV##mod##_XMM_to_M128( uptr to, x86SSERegType from ) { iMOV##mod( (void*)to, iRegisterSSE(from) ); } \
emitterT void sse##_MOV##mod##RmtoR( x86SSERegType to, x86IntRegType from, int offset ) { iMOV##mod( iRegisterSSE(to), ptr[iAddressReg(from)+offset] ); } \
emitterT void sse##_MOV##mod##RtoRm( x86IntRegType to, x86SSERegType from, int offset ) { iMOV##mod( ptr[iAddressReg(to)+offset], iRegisterSSE(from) ); } \
emitterT void sse##_MOV##mod##_M128_to_XMM( x86SSERegType to, uptr from ) { xMOV##mod( xRegisterSSE(to), (void*)from ); } \
emitterT void sse##_MOV##mod##_XMM_to_M128( uptr to, x86SSERegType from ) { xMOV##mod( (void*)to, xRegisterSSE(from) ); } \
emitterT void sse##_MOV##mod##RmtoR( x86SSERegType to, x86IntRegType from, int offset ) { xMOV##mod( xRegisterSSE(to), ptr[xAddressReg(from)+offset] ); } \
emitterT void sse##_MOV##mod##RtoRm( x86IntRegType to, x86SSERegType from, int offset ) { xMOV##mod( ptr[xAddressReg(to)+offset], xRegisterSSE(from) ); } \
emitterT void sse##_MOV##mod##RmStoR( x86SSERegType to, x86IntRegType from, x86IntRegType from2, int scale ) \
{ iMOV##mod( iRegisterSSE(to), ptr[iAddressReg(from)+iAddressReg(from2)] ); } \
{ xMOV##mod( xRegisterSSE(to), ptr[xAddressReg(from)+xAddressReg(from2)] ); } \
emitterT void sse##_MOV##mod##RtoRmS( x86IntRegType to, x86SSERegType from, x86IntRegType from2, int scale ) \
{ iMOV##mod( ptr[iAddressReg(to)+iAddressReg(from2)], iRegisterSSE(from) ); }
{ xMOV##mod( ptr[xAddressReg(to)+xAddressReg(from2)], xRegisterSSE(from) ); }
DEFINE_LEGACY_MOV_OPCODE( UPS, SSE )
DEFINE_LEGACY_MOV_OPCODE( APS, SSE )
DEFINE_LEGACY_MOV_OPCODE( LPS, SSE )
DEFINE_LEGACY_MOV_OPCODE( HPS, SSE )
DEFINE_LEGACY_MOV_OPCODE( DQA, SSE2 )
DEFINE_LEGACY_MOV_OPCODE( DQU, SSE2 )
//**********************************************************************************/
//MOVAPS: Move aligned Packed Single Precision FP values *
//**********************************************************************************
emitterT void SSE_MOVAPS_XMM_to_XMM( x86SSERegType to, x86SSERegType from ) { iMOVAPS( iRegisterSSE(to), iRegisterSSE(from) ); }
emitterT void SSE_MOVAPS_XMM_to_XMM( x86SSERegType to, x86SSERegType from ) { xMOVAPS( xRegisterSSE(to), xRegisterSSE(from) ); }
emitterT void SSE2_MOVQ_M64_to_XMM( x86SSERegType to, uptr from ) { iMOVQZX( iRegisterSSE(to), (void*)from ); }
emitterT void SSE2_MOVQ_XMM_to_XMM( x86SSERegType to, x86SSERegType from ) { iMOVQZX( iRegisterSSE(to), iRegisterSSE(from) ); }
emitterT void SSE2_MOVQ_XMM_to_M64( u32 to, x86SSERegType from ) { iMOVQ( (void*)to, iRegisterSSE(from) ); }
emitterT void SSE2_MOVDQ2Q_XMM_to_MM( x86MMXRegType to, x86SSERegType from) { iMOVQ( iRegisterMMX(to), iRegisterSSE(from) ); }
emitterT void SSE2_MOVQ2DQ_MM_to_XMM( x86SSERegType to, x86MMXRegType from) { iMOVQ( iRegisterSSE(to), iRegisterMMX(from) ); }
emitterT void SSE2_MOVQ_M64_to_XMM( x86SSERegType to, uptr from ) { xMOVQZX( xRegisterSSE(to), (void*)from ); }
emitterT void SSE2_MOVQ_XMM_to_XMM( x86SSERegType to, x86SSERegType from ) { xMOVQZX( xRegisterSSE(to), xRegisterSSE(from) ); }
emitterT void SSE2_MOVQ_XMM_to_M64( u32 to, x86SSERegType from ) { xMOVQ( (void*)to, xRegisterSSE(from) ); }
emitterT void SSE2_MOVDQ2Q_XMM_to_MM( x86MMXRegType to, x86SSERegType from) { xMOVQ( xRegisterMMX(to), xRegisterSSE(from) ); }
emitterT void SSE2_MOVQ2DQ_MM_to_XMM( x86SSERegType to, x86MMXRegType from) { xMOVQ( xRegisterSSE(to), xRegisterMMX(from) ); }
//**********************************************************************************/
//MOVSS: Move Scalar Single-Precision FP value *
//**********************************************************************************
emitterT void SSE_MOVSS_M32_to_XMM( x86SSERegType to, uptr from ) { iMOVSSZX( iRegisterSSE(to), (void*)from ); }
emitterT void SSE_MOVSS_XMM_to_M32( u32 to, x86SSERegType from ) { iMOVSS( (void*)to, iRegisterSSE(from) ); }
emitterT void SSE_MOVSS_XMM_to_XMM( x86SSERegType to, x86SSERegType from ) { iMOVSS( iRegisterSSE(to), iRegisterSSE(from) ); }
emitterT void SSE_MOVSS_Rm_to_XMM( x86SSERegType to, x86IntRegType from, int offset ) { iMOVSSZX( iRegisterSSE(to), ptr[iAddressReg(from)+offset] ); }
emitterT void SSE_MOVSS_XMM_to_Rm( x86IntRegType to, x86SSERegType from, int offset ) { iMOVSS( ptr[iAddressReg(to)+offset], iRegisterSSE(from) ); }
emitterT void SSE_MOVSS_M32_to_XMM( x86SSERegType to, uptr from ) { xMOVSSZX( xRegisterSSE(to), (void*)from ); }
emitterT void SSE_MOVSS_XMM_to_M32( u32 to, x86SSERegType from ) { xMOVSS( (void*)to, xRegisterSSE(from) ); }
emitterT void SSE_MOVSS_XMM_to_XMM( x86SSERegType to, x86SSERegType from ) { xMOVSS( xRegisterSSE(to), xRegisterSSE(from) ); }
emitterT void SSE_MOVSS_Rm_to_XMM( x86SSERegType to, x86IntRegType from, int offset ) { xMOVSSZX( xRegisterSSE(to), ptr[xAddressReg(from)+offset] ); }
emitterT void SSE_MOVSS_XMM_to_Rm( x86IntRegType to, x86SSERegType from, int offset ) { xMOVSS( ptr[xAddressReg(to)+offset], xRegisterSSE(from) ); }
emitterT void SSE2_MOVSD_M32_to_XMM( x86SSERegType to, uptr from ) { iMOVSDZX( iRegisterSSE(to), (void*)from ); }
emitterT void SSE2_MOVSD_XMM_to_M32( u32 to, x86SSERegType from ) { iMOVSD( (void*)to, iRegisterSSE(from) ); }
emitterT void SSE2_MOVSD_XMM_to_XMM( x86SSERegType to, x86SSERegType from ) { iMOVSD( iRegisterSSE(to), iRegisterSSE(from) ); }
emitterT void SSE2_MOVSD_Rm_to_XMM( x86SSERegType to, x86IntRegType from, int offset ) { iMOVSDZX( iRegisterSSE(to), ptr[iAddressReg(from)+offset] ); }
emitterT void SSE2_MOVSD_XMM_to_Rm( x86IntRegType to, x86SSERegType from, int offset ) { iMOVSD( ptr[iAddressReg(to)+offset], iRegisterSSE(from) ); }
emitterT void SSE2_MOVSD_M32_to_XMM( x86SSERegType to, uptr from ) { xMOVSDZX( xRegisterSSE(to), (void*)from ); }
emitterT void SSE2_MOVSD_XMM_to_M32( u32 to, x86SSERegType from ) { xMOVSD( (void*)to, xRegisterSSE(from) ); }
emitterT void SSE2_MOVSD_XMM_to_XMM( x86SSERegType to, x86SSERegType from ) { xMOVSD( xRegisterSSE(to), xRegisterSSE(from) ); }
emitterT void SSE2_MOVSD_Rm_to_XMM( x86SSERegType to, x86IntRegType from, int offset ) { xMOVSDZX( xRegisterSSE(to), ptr[xAddressReg(from)+offset] ); }
emitterT void SSE2_MOVSD_XMM_to_Rm( x86IntRegType to, x86SSERegType from, int offset ) { xMOVSD( ptr[xAddressReg(to)+offset], xRegisterSSE(from) ); }
emitterT void SSE_MASKMOVDQU_XMM_to_XMM( x86SSERegType to, x86SSERegType from ) { iMASKMOV( iRegisterSSE(to), iRegisterSSE(from) ); }
emitterT void SSE_MASKMOVDQU_XMM_to_XMM( x86SSERegType to, x86SSERegType from ) { xMASKMOV( xRegisterSSE(to), xRegisterSSE(from) ); }
//**********************************************************************************/
//MOVLPS: Move low Packed Single-Precision FP *
//**********************************************************************************
emitterT void SSE_MOVLPS_M64_to_XMM( x86SSERegType to, uptr from ) { iMOVLPS( iRegisterSSE(to), (void*)from ); }
emitterT void SSE_MOVLPS_XMM_to_M64( u32 to, x86SSERegType from ) { iMOVLPS( (void*)to, iRegisterSSE(from) ); }
emitterT void SSE_MOVLPS_Rm_to_XMM( x86SSERegType to, x86IntRegType from, int offset ) { iMOVLPS( iRegisterSSE(to), ptr[iAddressReg(from)+offset] ); }
emitterT void SSE_MOVLPS_XMM_to_Rm( x86IntRegType to, x86SSERegType from, int offset ) { iMOVLPS( ptr[iAddressReg(to)+offset], iRegisterSSE(from) ); }
emitterT void SSE_MOVLPS_M64_to_XMM( x86SSERegType to, uptr from ) { xMOVL.PS( xRegisterSSE(to), (void*)from ); }
emitterT void SSE_MOVLPS_XMM_to_M64( u32 to, x86SSERegType from ) { xMOVL.PS( (void*)to, xRegisterSSE(from) ); }
emitterT void SSE_MOVLPS_Rm_to_XMM( x86SSERegType to, x86IntRegType from, int offset ) { xMOVL.PS( xRegisterSSE(to), ptr[xAddressReg(from)+offset] ); }
emitterT void SSE_MOVLPS_XMM_to_Rm( x86IntRegType to, x86SSERegType from, int offset ) { xMOVL.PS( ptr[xAddressReg(to)+offset], xRegisterSSE(from) ); }
/////////////////////////////////////////////////////////////////////////////////////
//**********************************************************************************/
//MOVHPS: Move High Packed Single-Precision FP *
//**********************************************************************************
emitterT void SSE_MOVHPS_M64_to_XMM( x86SSERegType to, uptr from ) { iMOVHPS( iRegisterSSE(to), (void*)from ); }
emitterT void SSE_MOVHPS_XMM_to_M64( u32 to, x86SSERegType from ) { iMOVHPS( (void*)to, iRegisterSSE(from) ); }
emitterT void SSE_MOVHPS_Rm_to_XMM( x86SSERegType to, x86IntRegType from, int offset ) { iMOVHPS( iRegisterSSE(to), ptr[iAddressReg(from)+offset] ); }
emitterT void SSE_MOVHPS_XMM_to_Rm( x86IntRegType to, x86SSERegType from, int offset ) { iMOVHPS( ptr[iAddressReg(to)+offset], iRegisterSSE(from) ); }
emitterT void SSE_MOVHPS_M64_to_XMM( x86SSERegType to, uptr from ) { xMOVH.PS( xRegisterSSE(to), (void*)from ); }
emitterT void SSE_MOVHPS_XMM_to_M64( u32 to, x86SSERegType from ) { xMOVH.PS( (void*)to, xRegisterSSE(from) ); }
emitterT void SSE_MOVHPS_Rm_to_XMM( x86SSERegType to, x86IntRegType from, int offset ) { xMOVH.PS( xRegisterSSE(to), ptr[xAddressReg(from)+offset] ); }
emitterT void SSE_MOVHPS_XMM_to_Rm( x86IntRegType to, x86SSERegType from, int offset ) { xMOVH.PS( ptr[xAddressReg(to)+offset], xRegisterSSE(from) ); }
/////////////////////////////////////////////////////////////////////////////////////
//**********************************************************************************/
//MOVLHPS: Moved packed Single-Precision FP low to high *
//**********************************************************************************
emitterT void SSE_MOVLHPS_XMM_to_XMM( x86SSERegType to, x86SSERegType from ) { iMOVLHPS( iRegisterSSE(to), iRegisterSSE(from) ); }
emitterT void SSE_MOVLHPS_XMM_to_XMM( x86SSERegType to, x86SSERegType from ) { xMOVLH.PS( xRegisterSSE(to), xRegisterSSE(from) ); }
//////////////////////////////////////////////////////////////////////////////////////
//**********************************************************************************/
//MOVHLPS: Moved packed Single-Precision FP High to Low *
//**********************************************************************************
emitterT void SSE_MOVHLPS_XMM_to_XMM( x86SSERegType to, x86SSERegType from ) { iMOVHLPS( iRegisterSSE(to), iRegisterSSE(from) ); }
emitterT void SSE_MOVHLPS_XMM_to_XMM( x86SSERegType to, x86SSERegType from ) { xMOVHL.PS( xRegisterSSE(to), xRegisterSSE(from) ); }
emitterT void SSE2_PMOVMSKB_XMM_to_R32(x86IntRegType to, x86SSERegType from) { iPMOVMSKB( iRegister32(to), iRegisterSSE(from) ); }
emitterT void SSE2_PMOVMSKB_XMM_to_R32(x86IntRegType to, x86SSERegType from) { xPMOVMSKB( xRegister32(to), xRegisterSSE(from) ); }
#define DEFINE_LEGACY_PSD_OPCODE( mod ) \
emitterT void SSE_##mod##PS_M128_to_XMM( x86SSERegType to, uptr from ) { i##mod.PS( iRegisterSSE(to), (void*)from ); } \
emitterT void SSE_##mod##PS_XMM_to_XMM( x86SSERegType to, x86SSERegType from ) { i##mod.PS( iRegisterSSE(to), iRegisterSSE(from) ); } \
emitterT void SSE2_##mod##PD_M128_to_XMM( x86SSERegType to, uptr from ) { i##mod.PD( iRegisterSSE(to), (void*)from ); } \
emitterT void SSE2_##mod##PD_XMM_to_XMM( x86SSERegType to, x86SSERegType from ) { i##mod.PD( iRegisterSSE(to), iRegisterSSE(from) ); }
emitterT void SSE_##mod##PS_M128_to_XMM( x86SSERegType to, uptr from ) { x##mod.PS( xRegisterSSE(to), (void*)from ); } \
emitterT void SSE_##mod##PS_XMM_to_XMM( x86SSERegType to, x86SSERegType from ) { x##mod.PS( xRegisterSSE(to), xRegisterSSE(from) ); } \
emitterT void SSE2_##mod##PD_M128_to_XMM( x86SSERegType to, uptr from ) { x##mod.PD( xRegisterSSE(to), (void*)from ); } \
emitterT void SSE2_##mod##PD_XMM_to_XMM( x86SSERegType to, x86SSERegType from ) { x##mod.PD( xRegisterSSE(to), xRegisterSSE(from) ); }
#define DEFINE_LEGACY_PSSD_OPCODE( mod ) \
DEFINE_LEGACY_PSD_OPCODE( mod ) \
emitterT void SSE_##mod##SS_M32_to_XMM( x86SSERegType to, uptr from ) { i##mod.SS( iRegisterSSE(to), (void*)from ); } \
emitterT void SSE_##mod##SS_XMM_to_XMM( x86SSERegType to, x86SSERegType from ) { i##mod.SS( iRegisterSSE(to), iRegisterSSE(from) ); } \
emitterT void SSE2_##mod##SD_M32_to_XMM( x86SSERegType to, uptr from ) { i##mod.SD( iRegisterSSE(to), (void*)from ); } \
emitterT void SSE2_##mod##SD_XMM_to_XMM( x86SSERegType to, x86SSERegType from ) { i##mod.SD( iRegisterSSE(to), iRegisterSSE(from) ); }
emitterT void SSE_##mod##SS_M32_to_XMM( x86SSERegType to, uptr from ) { x##mod.SS( xRegisterSSE(to), (void*)from ); } \
emitterT void SSE_##mod##SS_XMM_to_XMM( x86SSERegType to, x86SSERegType from ) { x##mod.SS( xRegisterSSE(to), xRegisterSSE(from) ); } \
emitterT void SSE2_##mod##SD_M32_to_XMM( x86SSERegType to, uptr from ) { x##mod.SD( xRegisterSSE(to), (void*)from ); } \
emitterT void SSE2_##mod##SD_XMM_to_XMM( x86SSERegType to, x86SSERegType from ) { x##mod.SD( xRegisterSSE(to), xRegisterSSE(from) ); }
DEFINE_LEGACY_PSD_OPCODE( AND )
DEFINE_LEGACY_PSD_OPCODE( ANDN )
@ -238,76 +237,35 @@ DEFINE_LEGACY_PSSD_OPCODE( ADD )
DEFINE_LEGACY_PSSD_OPCODE( MUL )
DEFINE_LEGACY_PSSD_OPCODE( DIV )
emitterT void SSE_RCPPS_XMM_to_XMM( x86SSERegType to, x86SSERegType from ) { iRCPPS( iRegisterSSE(to), iRegisterSSE(from) ); }
emitterT void SSE_RCPPS_M128_to_XMM( x86SSERegType to, uptr from ) { iRCPPS( iRegisterSSE(to), (void*)from ); }
emitterT void SSE_RCPPS_XMM_to_XMM( x86SSERegType to, x86SSERegType from ) { xRCPPS( xRegisterSSE(to), xRegisterSSE(from) ); }
emitterT void SSE_RCPPS_M128_to_XMM( x86SSERegType to, uptr from ) { xRCPPS( xRegisterSSE(to), (void*)from ); }
emitterT void SSE_RCPSS_XMM_to_XMM( x86SSERegType to, x86SSERegType from ) { iRCPSS( iRegisterSSE(to), iRegisterSSE(from) ); }
emitterT void SSE_RCPSS_M32_to_XMM( x86SSERegType to, uptr from ) { iRCPSS( iRegisterSSE(to), (void*)from ); }
emitterT void SSE_RCPSS_XMM_to_XMM( x86SSERegType to, x86SSERegType from ) { xRCPSS( xRegisterSSE(to), xRegisterSSE(from) ); }
emitterT void SSE_RCPSS_M32_to_XMM( x86SSERegType to, uptr from ) { xRCPSS( xRegisterSSE(to), (void*)from ); }
////////////////////////////////////////////////////////////////////////////////////////////
//**********************************************************************************/
//Packed Single-Precision FP compare (CMPccPS) *
//**********************************************************************************
//missing SSE_CMPPS_I8_to_XMM
// SSE_CMPPS_M32_to_XMM
// SSE_CMPPS_XMM_to_XMM
emitterT void SSE_CMPEQPS_M128_to_XMM( x86SSERegType to, uptr from ) { CMPPSMtoR( 0 ); }
emitterT void SSE_CMPEQPS_XMM_to_XMM( x86SSERegType to, x86SSERegType from ) { CMPPSRtoR( 0 ); }
emitterT void SSE_CMPLTPS_M128_to_XMM( x86SSERegType to, uptr from ) { CMPPSMtoR( 1 ); }
emitterT void SSE_CMPLTPS_XMM_to_XMM( x86SSERegType to, x86SSERegType from ) { CMPPSRtoR( 1 ); }
emitterT void SSE_CMPLEPS_M128_to_XMM( x86SSERegType to, uptr from ) { CMPPSMtoR( 2 ); }
emitterT void SSE_CMPLEPS_XMM_to_XMM( x86SSERegType to, x86SSERegType from ) { CMPPSRtoR( 2 ); }
emitterT void SSE_CMPUNORDPS_M128_to_XMM( x86SSERegType to, uptr from ) { CMPPSMtoR( 3 ); }
emitterT void SSE_CMPUNORDPS_XMM_to_XMM( x86SSERegType to, x86SSERegType from ) { CMPPSRtoR( 3 ); }
emitterT void SSE_CMPNEPS_M128_to_XMM( x86SSERegType to, uptr from ) { CMPPSMtoR( 4 ); }
emitterT void SSE_CMPNEPS_XMM_to_XMM( x86SSERegType to, x86SSERegType from ) { CMPPSRtoR( 4 ); }
emitterT void SSE_CMPNLTPS_M128_to_XMM( x86SSERegType to, uptr from ) { CMPPSMtoR( 5 ); }
emitterT void SSE_CMPNLTPS_XMM_to_XMM( x86SSERegType to, x86SSERegType from ) { CMPPSRtoR( 5 ); }
emitterT void SSE_CMPNLEPS_M128_to_XMM( x86SSERegType to, uptr from ) { CMPPSMtoR( 6 ); }
emitterT void SSE_CMPNLEPS_XMM_to_XMM( x86SSERegType to, x86SSERegType from ) { CMPPSRtoR( 6 ); }
emitterT void SSE_CMPORDPS_M128_to_XMM( x86SSERegType to, uptr from ) { CMPPSMtoR( 7 ); }
emitterT void SSE_CMPORDPS_XMM_to_XMM( x86SSERegType to, x86SSERegType from ) { CMPPSRtoR( 7 ); }
///////////////////////////////////////////////////////////////////////////////////////////
//**********************************************************************************/
//Scalar Single-Precision FP compare (CMPccSS) *
//**********************************************************************************
//missing SSE_CMPSS_I8_to_XMM
// SSE_CMPSS_M32_to_XMM
// SSE_CMPSS_XMM_to_XMM
emitterT void SSE_CMPEQSS_M32_to_XMM( x86SSERegType to, uptr from ) { CMPSSMtoR( 0 ); }
emitterT void SSE_CMPEQSS_XMM_to_XMM( x86SSERegType to, x86SSERegType from ) { CMPSSRtoR( 0 ); }
emitterT void SSE_CMPLTSS_M32_to_XMM( x86SSERegType to, uptr from ) { CMPSSMtoR( 1 ); }
emitterT void SSE_CMPLTSS_XMM_to_XMM( x86SSERegType to, x86SSERegType from ) { CMPSSRtoR( 1 ); }
emitterT void SSE_CMPLESS_M32_to_XMM( x86SSERegType to, uptr from ) { CMPSSMtoR( 2 ); }
emitterT void SSE_CMPLESS_XMM_to_XMM( x86SSERegType to, x86SSERegType from ) { CMPSSRtoR( 2 ); }
emitterT void SSE_CMPUNORDSS_M32_to_XMM( x86SSERegType to, uptr from ) { CMPSSMtoR( 3 ); }
emitterT void SSE_CMPUNORDSS_XMM_to_XMM( x86SSERegType to, x86SSERegType from ) { CMPSSRtoR( 3 ); }
emitterT void SSE_CMPNESS_M32_to_XMM( x86SSERegType to, uptr from ) { CMPSSMtoR( 4 ); }
emitterT void SSE_CMPNESS_XMM_to_XMM( x86SSERegType to, x86SSERegType from ) { CMPSSRtoR( 4 ); }
emitterT void SSE_CMPNLTSS_M32_to_XMM( x86SSERegType to, uptr from ) { CMPSSMtoR( 5 ); }
emitterT void SSE_CMPNLTSS_XMM_to_XMM( x86SSERegType to, x86SSERegType from ) { CMPSSRtoR( 5 ); }
emitterT void SSE_CMPNLESS_M32_to_XMM( x86SSERegType to, uptr from ) { CMPSSMtoR( 6 ); }
emitterT void SSE_CMPNLESS_XMM_to_XMM( x86SSERegType to, x86SSERegType from ) { CMPSSRtoR( 6 ); }
emitterT void SSE_CMPORDSS_M32_to_XMM( x86SSERegType to, uptr from ) { CMPSSMtoR( 7 ); }
emitterT void SSE_CMPORDSS_XMM_to_XMM( x86SSERegType to, x86SSERegType from ) { CMPSSRtoR( 7 ); }
#define DEFINE_LEGACY_CMP_OPCODE( comp ) \
emitterT void SSE_CMP##comp##PS_M128_to_XMM( x86SSERegType to, uptr from ) { xCMP##comp.PS( xRegisterSSE(to), (void*)from ); } \
emitterT void SSE_CMP##comp##PS_XMM_to_XMM( x86SSERegType to, x86SSERegType from ) { xCMP##comp.PS( xRegisterSSE(to), xRegisterSSE(from) ); } \
emitterT void SSE2_CMP##comp##PD_M128_to_XMM( x86SSERegType to, uptr from ) { xCMP##comp.PD( xRegisterSSE(to), (void*)from ); } \
emitterT void SSE2_CMP##comp##PD_XMM_to_XMM( x86SSERegType to, x86SSERegType from ) { xCMP##comp.PD( xRegisterSSE(to), xRegisterSSE(from) ); } \
emitterT void SSE_CMP##comp##SS_M128_to_XMM( x86SSERegType to, uptr from ) { xCMP##comp.SS( xRegisterSSE(to), (void*)from ); } \
emitterT void SSE_CMP##comp##SS_XMM_to_XMM( x86SSERegType to, x86SSERegType from ) { xCMP##comp.SS( xRegisterSSE(to), xRegisterSSE(from) ); } \
emitterT void SSE2_CMP##comp##SD_M128_to_XMM( x86SSERegType to, uptr from ) { xCMP##comp.SD( xRegisterSSE(to), (void*)from ); } \
emitterT void SSE2_CMP##comp##SD_XMM_to_XMM( x86SSERegType to, x86SSERegType from ) { xCMP##comp.SD( xRegisterSSE(to), xRegisterSSE(from) ); }
emitterT void SSE2_CMPEQSD_M64_to_XMM( x86SSERegType to, uptr from ) { CMPSDMtoR( 0 ); }
emitterT void SSE2_CMPEQSD_XMM_to_XMM( x86SSERegType to, x86SSERegType from ) { CMPSDRtoR( 0 ); }
emitterT void SSE2_CMPLTSD_M64_to_XMM( x86SSERegType to, uptr from ) { CMPSDMtoR( 1 ); }
emitterT void SSE2_CMPLTSD_XMM_to_XMM( x86SSERegType to, x86SSERegType from ) { CMPSDRtoR( 1 ); }
emitterT void SSE2_CMPLESD_M64_to_XMM( x86SSERegType to, uptr from ) { CMPSDMtoR( 2 ); }
emitterT void SSE2_CMPLESD_XMM_to_XMM( x86SSERegType to, x86SSERegType from ) { CMPSDRtoR( 2 ); }
emitterT void SSE2_CMPUNORDSD_M64_to_XMM( x86SSERegType to, uptr from ) { CMPSDMtoR( 3 ); }
emitterT void SSE2_CMPUNORDSD_XMM_to_XMM( x86SSERegType to, x86SSERegType from ) { CMPSDRtoR( 3 ); }
emitterT void SSE2_CMPNESD_M64_to_XMM( x86SSERegType to, uptr from ) { CMPSDMtoR( 4 ); }
emitterT void SSE2_CMPNESD_XMM_to_XMM( x86SSERegType to, x86SSERegType from ) { CMPSDRtoR( 4 ); }
emitterT void SSE2_CMPNLTSD_M64_to_XMM( x86SSERegType to, uptr from ) { CMPSDMtoR( 5 ); }
emitterT void SSE2_CMPNLTSD_XMM_to_XMM( x86SSERegType to, x86SSERegType from ) { CMPSDRtoR( 5 ); }
emitterT void SSE2_CMPNLESD_M64_to_XMM( x86SSERegType to, uptr from ) { CMPSDMtoR( 6 ); }
emitterT void SSE2_CMPNLESD_XMM_to_XMM( x86SSERegType to, x86SSERegType from ) { CMPSDRtoR( 6 ); }
emitterT void SSE2_CMPORDSD_M64_to_XMM( x86SSERegType to, uptr from ) { CMPSDMtoR( 7 ); }
emitterT void SSE2_CMPORDSD_XMM_to_XMM( x86SSERegType to, x86SSERegType from ) { CMPSDRtoR( 7 ); }
DEFINE_LEGACY_CMP_OPCODE( EQ )
DEFINE_LEGACY_CMP_OPCODE( LT )
DEFINE_LEGACY_CMP_OPCODE( LE )
DEFINE_LEGACY_CMP_OPCODE( UNORD )
DEFINE_LEGACY_CMP_OPCODE( NE )
DEFINE_LEGACY_CMP_OPCODE( NLT )
DEFINE_LEGACY_CMP_OPCODE( NLE )
DEFINE_LEGACY_CMP_OPCODE( ORD )
emitterT void SSE_UCOMISS_M32_to_XMM( x86SSERegType to, uptr from )
{
@ -642,7 +600,7 @@ emitterT void SSE2_MOVD_XMM_to_Rm( x86IntRegType to, x86SSERegType from, int off
///////////////////////////////////////////////////////////////////////////////////////
emitterT void SSE2_MOVDQA_XMM_to_XMM( x86SSERegType to, x86SSERegType from) { iMOVDQA( iRegisterSSE(to), iRegisterSSE(from) ); }
emitterT void SSE2_MOVDQA_XMM_to_XMM( x86SSERegType to, x86SSERegType from) { xMOVDQA( xRegisterSSE(to), xRegisterSSE(from) ); }
// shift right logical

View File

@ -115,7 +115,7 @@ template< typename T >
static __forceinline bool is_s8( T imm ) { return (s8)imm == (s32)imm; }
template< typename T >
static __forceinline void iWrite( T val )
static __forceinline void xWrite( T val )
{
*(T*)x86Ptr = val;
x86Ptr += sizeof(T);
@ -159,7 +159,7 @@ namespace x86Emitter
static const int ModRm_UseSib = 4; // same index value as ESP (used in RM field)
static const int ModRm_UseDisp32 = 5; // same index value as EBP (used in Mod field)
class iAddressInfo;
class xAddressInfo;
class ModSibBase;
extern void iSetPtr( void* ptr );
@ -170,12 +170,12 @@ namespace x86Emitter
static __forceinline void write8( u8 val )
{
iWrite( val );
xWrite( val );
}
static __forceinline void write16( u16 val )
{
iWrite( val );
xWrite( val );
}
static __forceinline void write24( u32 val )
@ -186,30 +186,30 @@ namespace x86Emitter
static __forceinline void write32( u32 val )
{
iWrite( val );
xWrite( val );
}
static __forceinline void write64( u64 val )
{
iWrite( val );
xWrite( val );
}
//////////////////////////////////////////////////////////////////////////////////////////
// iRegister
// Unless templating some fancy stuff, use the friendly iRegister32/16/8 typedefs instead.
// xRegister
// Unless templating some fancy stuff, use the friendly xRegister32/16/8 typedefs instead.
//
template< typename OperandType >
class iRegister
class xRegister
{
public:
static const uint OperandSize = sizeof( OperandType );
static const iRegister Empty; // defined as an empty/unused value (-1)
static const xRegister Empty; // defined as an empty/unused value (-1)
int Id;
iRegister( const iRegister<OperandType>& src ) : Id( src.Id ) {}
iRegister(): Id( -1 ) {}
explicit iRegister( int regId ) : Id( regId ) { jASSUME( Id >= -1 && Id < 8 ); }
xRegister( const xRegister<OperandType>& src ) : Id( src.Id ) {}
xRegister(): Id( -1 ) {}
explicit xRegister( int regId ) : Id( regId ) { jASSUME( Id >= -1 && Id < 8 ); }
bool IsEmpty() const { return Id < 0; }
@ -219,17 +219,17 @@ namespace x86Emitter
// returns true if the register is a valid MMX or XMM register.
bool IsSIMD() const { return OperandSize == 8 || OperandSize == 16; }
bool operator==( const iRegister<OperandType>& src ) const
bool operator==( const xRegister<OperandType>& src ) const
{
return (Id == src.Id);
}
bool operator!=( const iRegister<OperandType>& src ) const
bool operator!=( const xRegister<OperandType>& src ) const
{
return (Id != src.Id);
}
iRegister<OperandType>& operator=( const iRegister<OperandType>& src )
xRegister<OperandType>& operator=( const xRegister<OperandType>& src )
{
Id = src.Id;
return *this;
@ -239,20 +239,20 @@ namespace x86Emitter
//////////////////////////////////////////////////////////////////////////////////////////
//
template< typename OperandType >
class iRegisterSIMD : public iRegister<OperandType>
class xRegisterSIMD : public xRegister<OperandType>
{
public:
static const iRegisterSIMD Empty; // defined as an empty/unused value (-1)
static const xRegisterSIMD Empty; // defined as an empty/unused value (-1)
public:
iRegisterSIMD(): iRegister<OperandType>() {}
iRegisterSIMD( const iRegisterSIMD& src ) : iRegister<OperandType>( src.Id ) {}
iRegisterSIMD( const iRegister<OperandType>& src ) : iRegister<OperandType>( src ) {}
explicit iRegisterSIMD( int regId ) : iRegister<OperandType>( regId ) {}
xRegisterSIMD(): xRegister<OperandType>() {}
xRegisterSIMD( const xRegisterSIMD& src ) : xRegister<OperandType>( src.Id ) {}
xRegisterSIMD( const xRegister<OperandType>& src ) : xRegister<OperandType>( src ) {}
explicit xRegisterSIMD( int regId ) : xRegister<OperandType>( regId ) {}
iRegisterSIMD<OperandType>& operator=( const iRegisterSIMD<OperandType>& src )
xRegisterSIMD<OperandType>& operator=( const xRegisterSIMD<OperandType>& src )
{
iRegister<OperandType>::Id = src.Id;
xRegister<OperandType>::Id = src.Id;
return *this;
}
};
@ -266,66 +266,66 @@ namespace x86Emitter
// all about the the templated code in haphazard fashion. Yay.. >_<
//
typedef iRegisterSIMD<u128> iRegisterSSE;
typedef iRegisterSIMD<u64> iRegisterMMX;
typedef iRegister<u32> iRegister32;
typedef iRegister<u16> iRegister16;
typedef iRegister<u8> iRegister8;
typedef xRegisterSIMD<u128> xRegisterSSE;
typedef xRegisterSIMD<u64> xRegisterMMX;
typedef xRegister<u32> xRegister32;
typedef xRegister<u16> xRegister16;
typedef xRegister<u8> xRegister8;
class iRegisterCL : public iRegister8
class xRegisterCL : public xRegister8
{
public:
iRegisterCL(): iRegister8( 1 ) {}
xRegisterCL(): xRegister8( 1 ) {}
};
extern const iRegisterSSE
extern const xRegisterSSE
xmm0, xmm1, xmm2, xmm3,
xmm4, xmm5, xmm6, xmm7;
extern const iRegisterMMX
extern const xRegisterMMX
mm0, mm1, mm2, mm3,
mm4, mm5, mm6, mm7;
extern const iRegister32
extern const xRegister32
eax, ebx, ecx, edx,
esi, edi, ebp, esp;
extern const iRegister16
extern const xRegister16
ax, bx, cx, dx,
si, di, bp, sp;
extern const iRegister8
extern const xRegister8
al, dl, bl,
ah, ch, dh, bh;
extern const iRegisterCL cl; // I'm special!
extern const xRegisterCL cl; // I'm special!
//////////////////////////////////////////////////////////////////////////////////////////
// Use 32 bit registers as out index register (for ModSib memory address calculations)
// Only iAddressReg provides operators for constructing iAddressInfo types.
// Only xAddressReg provides operators for constructing xAddressInfo types.
//
class iAddressReg : public iRegister32
class xAddressReg : public xRegister32
{
public:
static const iAddressReg Empty; // defined as an empty/unused value (-1)
static const xAddressReg Empty; // defined as an empty/unused value (-1)
public:
iAddressReg(): iRegister32() {}
iAddressReg( const iAddressReg& src ) : iRegister32( src.Id ) {}
iAddressReg( const iRegister32& src ) : iRegister32( src ) {}
explicit iAddressReg( int regId ) : iRegister32( regId ) {}
xAddressReg(): xRegister32() {}
xAddressReg( const xAddressReg& src ) : xRegister32( src.Id ) {}
xAddressReg( const xRegister32& src ) : xRegister32( src ) {}
explicit xAddressReg( int regId ) : xRegister32( regId ) {}
// Returns true if the register is the stack pointer: ESP.
bool IsStackPointer() const { return Id == 4; }
iAddressInfo operator+( const iAddressReg& right ) const;
iAddressInfo operator+( const iAddressInfo& right ) const;
iAddressInfo operator+( s32 right ) const;
xAddressInfo operator+( const xAddressReg& right ) const;
xAddressInfo operator+( const xAddressInfo& right ) const;
xAddressInfo operator+( s32 right ) const;
iAddressInfo operator*( u32 factor ) const;
iAddressInfo operator<<( u32 shift ) const;
xAddressInfo operator*( u32 factor ) const;
xAddressInfo operator<<( u32 shift ) const;
iAddressReg& operator=( const iRegister32& src )
xAddressReg& operator=( const xRegister32& src )
{
Id = src.Id;
return *this;
@ -334,16 +334,16 @@ namespace x86Emitter
//////////////////////////////////////////////////////////////////////////////////////////
//
class iAddressInfo
class xAddressInfo
{
public:
iAddressReg Base; // base register (no scale)
iAddressReg Index; // index reg gets multiplied by the scale
xAddressReg Base; // base register (no scale)
xAddressReg Index; // index reg gets multiplied by the scale
int Factor; // scale applied to the index register, in factor form (not a shift!)
s32 Displacement; // address displacement
public:
__forceinline iAddressInfo( const iAddressReg& base, const iAddressReg& index, int factor=1, s32 displacement=0 ) :
__forceinline xAddressInfo( const xAddressReg& base, const xAddressReg& index, int factor=1, s32 displacement=0 ) :
Base( base ),
Index( index ),
Factor( factor ),
@ -351,7 +351,7 @@ namespace x86Emitter
{
}
__forceinline explicit iAddressInfo( const iAddressReg& index, int displacement=0 ) :
__forceinline explicit xAddressInfo( const xAddressReg& index, int displacement=0 ) :
Base(),
Index( index ),
Factor(0),
@ -359,7 +359,7 @@ namespace x86Emitter
{
}
__forceinline explicit iAddressInfo( s32 displacement ) :
__forceinline explicit xAddressInfo( s32 displacement ) :
Base(),
Index(),
Factor(0),
@ -367,24 +367,24 @@ namespace x86Emitter
{
}
static iAddressInfo FromIndexReg( const iAddressReg& index, int scale=0, s32 displacement=0 );
static xAddressInfo FromIndexReg( const xAddressReg& index, int scale=0, s32 displacement=0 );
public:
bool IsByteSizeDisp() const { return is_s8( Displacement ); }
__forceinline iAddressInfo& Add( s32 imm )
__forceinline xAddressInfo& Add( s32 imm )
{
Displacement += imm;
return *this;
}
__forceinline iAddressInfo& Add( const iAddressReg& src );
__forceinline iAddressInfo& Add( const iAddressInfo& src );
__forceinline xAddressInfo& Add( const xAddressReg& src );
__forceinline xAddressInfo& Add( const xAddressInfo& src );
__forceinline iAddressInfo operator+( const iAddressReg& right ) const { return iAddressInfo( *this ).Add( right ); }
__forceinline iAddressInfo operator+( const iAddressInfo& right ) const { return iAddressInfo( *this ).Add( right ); }
__forceinline iAddressInfo operator+( s32 imm ) const { return iAddressInfo( *this ).Add( imm ); }
__forceinline iAddressInfo operator-( s32 imm ) const { return iAddressInfo( *this ).Add( -imm ); }
__forceinline xAddressInfo operator+( const xAddressReg& right ) const { return xAddressInfo( *this ).Add( right ); }
__forceinline xAddressInfo operator+( const xAddressInfo& right ) const { return xAddressInfo( *this ).Add( right ); }
__forceinline xAddressInfo operator+( s32 imm ) const { return xAddressInfo( *this ).Add( imm ); }
__forceinline xAddressInfo operator-( s32 imm ) const { return xAddressInfo( *this ).Add( -imm ); }
};
//////////////////////////////////////////////////////////////////////////////////////////
@ -392,25 +392,25 @@ namespace x86Emitter
//
// This class serves two purposes: It houses 'reduced' ModRM/SIB info only, which means
// that the Base, Index, Scale, and Displacement values are all in the correct arrange-
// ments, and it serves as a type-safe layer between the iRegister's operators (which
// generate iAddressInfo types) and the emitter's ModSib instruction forms. Without this,
// the iRegister would pass as a ModSib type implicitly, and that would cause ambiguity
// ments, and it serves as a type-safe layer between the xRegister's operators (which
// generate xAddressInfo types) and the emitter's ModSib instruction forms. Without this,
// the xRegister would pass as a ModSib type implicitly, and that would cause ambiguity
// on a number of instructions.
//
// End users should always use iAddressInfo instead.
// End users should always use xAddressInfo instead.
//
class ModSibBase
{
public:
iAddressReg Base; // base register (no scale)
iAddressReg Index; // index reg gets multiplied by the scale
xAddressReg Base; // base register (no scale)
xAddressReg Index; // index reg gets multiplied by the scale
uint Scale; // scale applied to the index register, in scale/shift form
s32 Displacement; // offset applied to the Base/Index registers.
public:
explicit ModSibBase( const iAddressInfo& src );
explicit ModSibBase( const xAddressInfo& src );
explicit ModSibBase( s32 disp );
ModSibBase( iAddressReg base, iAddressReg index, int scale=0, s32 displacement=0 );
ModSibBase( xAddressReg base, xAddressReg index, int scale=0, s32 displacement=0 );
bool IsByteSizeDisp() const { return is_s8( Displacement ); }
@ -437,9 +437,9 @@ namespace x86Emitter
public:
static const uint OperandSize = sizeof( OperandType );
__forceinline explicit ModSibStrict( const iAddressInfo& src ) : ModSibBase( src ) {}
__forceinline explicit ModSibStrict( const xAddressInfo& src ) : ModSibBase( src ) {}
__forceinline explicit ModSibStrict( s32 disp ) : ModSibBase( disp ) {}
__forceinline ModSibStrict( iAddressReg base, iAddressReg index, int scale=0, s32 displacement=0 ) :
__forceinline ModSibStrict( xAddressReg base, xAddressReg index, int scale=0, s32 displacement=0 ) :
ModSibBase( base, index, scale, displacement ) {}
__forceinline ModSibStrict<OperandType>& Add( s32 imm )
@ -453,20 +453,20 @@ namespace x86Emitter
};
//////////////////////////////////////////////////////////////////////////////////////////
// iAddressIndexerBase - This is a static class which provisions our ptr[] syntax.
// xAddressIndexerBase - This is a static class which provisions our ptr[] syntax.
//
struct iAddressIndexerBase
struct xAddressIndexerBase
{
// passthrough instruction, allows ModSib to pass silently through ptr translation
// without doing anything and without compiler error.
const ModSibBase& operator[]( const ModSibBase& src ) const { return src; }
__forceinline ModSibBase operator[]( iAddressReg src ) const
__forceinline ModSibBase operator[]( xAddressReg src ) const
{
return ModSibBase( src, iAddressReg::Empty );
return ModSibBase( src, xAddressReg::Empty );
}
__forceinline ModSibBase operator[]( const iAddressInfo& src ) const
__forceinline ModSibBase operator[]( const xAddressInfo& src ) const
{
return ModSibBase( src );
}
@ -481,7 +481,7 @@ namespace x86Emitter
return ModSibBase( (uptr)src );
}
iAddressIndexerBase() {} // appease the GCC gods
xAddressIndexerBase() {} // appease the GCC gods
};
//////////////////////////////////////////////////////////////////////////////////////////
@ -489,7 +489,7 @@ namespace x86Emitter
// specification of the operand size for ImmToMem operations.
//
template< typename OperandType >
struct iAddressIndexer
struct xAddressIndexer
{
static const uint OperandSize = sizeof( OperandType );
@ -497,12 +497,12 @@ namespace x86Emitter
// without doing anything and without compiler error.
const ModSibStrict<OperandType>& operator[]( const ModSibStrict<OperandType>& src ) const { return src; }
__forceinline ModSibStrict<OperandType> operator[]( iAddressReg src ) const
__forceinline ModSibStrict<OperandType> operator[]( xAddressReg src ) const
{
return ModSibStrict<OperandType>( src, iAddressReg::Empty );
return ModSibStrict<OperandType>( src, xAddressReg::Empty );
}
__forceinline ModSibStrict<OperandType> operator[]( const iAddressInfo& src ) const
__forceinline ModSibStrict<OperandType> operator[]( const xAddressInfo& src ) const
{
return ModSibStrict<OperandType>( src );
}
@ -517,17 +517,17 @@ namespace x86Emitter
return ModSibStrict<OperandType>( (uptr)src );
}
iAddressIndexer() {} // GCC initialization dummy
xAddressIndexer() {} // GCC initialization dummy
};
// ptr[] - use this form for instructions which can resolve the address operand size from
// the other register operand sizes.
extern const iAddressIndexerBase ptr;
extern const iAddressIndexer<u128> ptr128;
extern const iAddressIndexer<u64> ptr64;
extern const iAddressIndexer<u32> ptr32; // explicitly typed addressing, usually needed for '[dest],imm' instruction forms
extern const iAddressIndexer<u16> ptr16; // explicitly typed addressing, usually needed for '[dest],imm' instruction forms
extern const iAddressIndexer<u8> ptr8; // explicitly typed addressing, usually needed for '[dest],imm' instruction forms
extern const xAddressIndexerBase ptr;
extern const xAddressIndexer<u128> ptr128;
extern const xAddressIndexer<u64> ptr64;
extern const xAddressIndexer<u32> ptr32; // explicitly typed addressing, usually needed for '[dest],imm' instruction forms
extern const xAddressIndexer<u16> ptr16; // explicitly typed addressing, usually needed for '[dest],imm' instruction forms
extern const xAddressIndexer<u8> ptr8; // explicitly typed addressing, usually needed for '[dest],imm' instruction forms
//////////////////////////////////////////////////////////////////////////////////////////
// JccComparisonType - enumerated possibilities for inspired code branching!
@ -561,26 +561,41 @@ namespace x86Emitter
// Not supported yet:
//E3 cb JECXZ rel8 Jump short if ECX register is 0.
//////////////////////////////////////////////////////////////////////////////////////////
// SSE2_ComparisonType - enumerated possibilities for SIMD data comparison!
//
enum SSE2_ComparisonType
{
SSE2_Equal = 0,
SSE2_Less,
SSE2_LessOrEqual,
SSE2_Unordered,
SSE2_NotEqual,
SSE2_NotLess,
SSE2_NotLessOrEqual,
SSE2_Ordered
};
//////////////////////////////////////////////////////////////////////////////////////////
// iSmartJump
// xSmartJump
// This class provides an interface for generating forward-based j8's or j32's "smartly"
// as per the measured displacement distance. If the displacement is a valid s8, then
// a j8 is inserted, else a j32.
//
// Note: This class is inherently unsafe, and so it's recommended to use iForwardJump8/32
// Note: This class is inherently unsafe, and so it's recommended to use xForwardJump8/32
// whenever it is known that the jump destination is (or is not) short. Only use
// iSmartJump in cases where it's unknown what jump encoding will be ideal.
// xSmartJump in cases where it's unknown what jump encoding will be ideal.
//
// Important: Use this tool with caution! iSmartJump cannot be used in cases where jump
// Important: Use this tool with caution! xSmartJump cannot be used in cases where jump
// targets overlap, since the writeback of the second target will alter the position of
// the first target (which breaks the relative addressing). To assist in avoiding such
// errors, iSmartJump works based on C++ block scope, where the destruction of the
// iSmartJump object (invoked by a '}') signals the target of the jump. Example:
// errors, xSmartJump works based on C++ block scope, where the destruction of the
// xSmartJump object (invoked by a '}') signals the target of the jump. Example:
//
// {
// iCMP( EAX, ECX );
// iSmartJump jumpTo( Jcc_Above );
// xSmartJump jumpTo( Jcc_Above );
// [... conditional code ...]
// } // smartjump targets this spot.
//
@ -593,7 +608,7 @@ namespace x86Emitter
// speed benefits in the form of L1/L2 cache clutter, on any CPU. They're also notably
// faster on P4's, and mildly faster on AMDs. (Core2's and i7's don't care)
//
class iSmartJump : public NoncopyableObject
class xSmartJump : public NoncopyableObject
{
protected:
u8* m_baseptr; // base address of the instruction (passed to the instruction emitter)
@ -607,12 +622,12 @@ namespace x86Emitter
}
JccComparisonType GetCondition() const { return m_cc; }
virtual ~iSmartJump();
virtual ~xSmartJump();
// ------------------------------------------------------------------------
// ccType - Comparison type to be written back to the jump instruction position.
//
iSmartJump( JccComparisonType ccType )
xSmartJump( JccComparisonType ccType )
{
jASSUME( ccType != Jcc_Unknown );
m_baseptr = iGetPtr();
@ -625,12 +640,12 @@ namespace x86Emitter
};
//////////////////////////////////////////////////////////////////////////////////////////
// iForwardJump
// Primary use of this class is through the various iForwardJA8/iForwardJLE32/etc. helpers
// xForwardJump
// Primary use of this class is through the various xForwardJA8/xForwardJLE32/etc. helpers
// defined later in this header. :)
//
template< typename OperandType >
class iForwardJump
class xForwardJump
{
public:
static const uint OperandSize = sizeof( OperandType );
@ -641,7 +656,7 @@ namespace x86Emitter
// The jump instruction is emitted at the point of object construction. The conditional
// type must be valid (Jcc_Unknown generates an assertion).
iForwardJump( JccComparisonType cctype = Jcc_Unconditional );
xForwardJump( JccComparisonType cctype = Jcc_Unconditional );
// Sets the jump target by writing back the current x86Ptr to the jump instruction.
// This method can be called multiple times, re-writing the jump instruction's target
@ -656,8 +671,8 @@ namespace x86Emitter
extern void ModRM( uint mod, uint reg, uint rm );
extern void ModRM_Direct( uint reg, uint rm );
extern void SibSB( u32 ss, u32 index, u32 base );
extern void iWriteDisp( int regfield, s32 displacement );
extern void iWriteDisp( int regfield, const void* address );
extern void xWriteDisp( int regfield, s32 displacement );
extern void xWriteDisp( int regfield, const void* address );
extern void EmitSibMagic( uint regfield, const ModSibBase& info );