Port x64 fixes from Insurgency branch

This commit is contained in:
Nicholas Hastings 2025-11-09 12:13:45 -05:00
parent 4ca98539d2
commit ed67695b02
26 changed files with 61 additions and 49 deletions

View File

@ -329,7 +329,7 @@ enum XUSER_SIGNIN_STATE
#if (defined(POSIX))
typedef size_t ULONG_PTR;
#else
#elif !(defined(WIN64))
typedef _W64 unsigned long ULONG_PTR;
#endif
typedef ULONG_PTR DWORD_PTR, *PDWORD_PTR;

Binary file not shown.

Binary file not shown.

Binary file not shown.

Binary file not shown.

BIN
lib/public/win64/tier0.lib Normal file

Binary file not shown.

BIN
lib/public/win64/tier1.lib Normal file

Binary file not shown.

Binary file not shown.

View File

@ -57,7 +57,7 @@ void ColorQuantize(uint8 const *Image,
{
struct QuantizedValue *v=FindQNode(q,p);
if (v)
for(int c=0;c<3;c++)
for(c=0;c<3;c++)
out_palette[p*3+c]=v->Mean[c];
}
memset(Error,0,sizeof(Error));

View File

@ -3236,6 +3236,7 @@ void MathLib_Init( float gamma, float texGamma, float brightness, int overbright
{
s_bSSEEnabled = true;
#ifndef _WIN64
// Select the SSE specific routines if available
pfVectorNormalizeFast = _SSE_VectorNormalizeFast;
pfInvRSquared = _SSE_InvRSquared;
@ -3245,6 +3246,7 @@ void MathLib_Init( float gamma, float texGamma, float brightness, int overbright
#ifdef _WIN32
pfFastSinCos = _SSE_SinCos;
pfFastCos = _SSE_cos;
#endif
#endif
}
else
@ -3255,7 +3257,7 @@ void MathLib_Init( float gamma, float texGamma, float brightness, int overbright
if ( bAllowSSE2 && pi.m_bSSE2 )
{
s_bSSE2Enabled = true;
#ifdef _WIN32
#if defined(_WIN32) && !defined(_WIN64)
pfFastSinCos = _SSE2_SinCos;
pfFastCos = _SSE2_cos;
#endif

View File

@ -198,8 +198,8 @@ static void UpdateStats(struct QuantizedValue *v)
N+=s->Count;
for(j=0;j<current_ndims;j++)
{
uint8 v=s->Value[j];
Means[j]+=v*s->Count;
uint8 v2=s->Value[j];
Means[j]+=v2*s->Count;
}
}
for(j=0;j<current_ndims;j++)
@ -316,7 +316,7 @@ static void SubdivideNode(struct QuantizedValue *n, int whichdim)
// extrema instead. LocalMean[i][0] will be the point with the lowest
// value on the dimension and LocalMean[i][1] the one with the lowest
// value.
for(int i=0;i<current_ndims;i++)
for(i=0;i<current_ndims;i++)
{
LocalMean[i][0]=minS->Value[i];
LocalMean[i][1]=maxS->Value[i];

View File

@ -16,6 +16,9 @@
// memdbgon must be the last include file in a .cpp file!!!
#include "tier0/memdbgon.h"
#ifndef COMPILER_MSVC64
// Implement for 64-bit Windows if needed.
static const uint32 _sincos_masks[] = { (uint32)0x0, (uint32)~0x0 };
static const uint32 _sincos_inv_masks[] = { (uint32)~0x0, (uint32)0x0 };
@ -843,3 +846,4 @@ vec_t DotProduct (const vec_t *a, const vec_t *c)
}
*/
#endif // COMPILER_MSVC64

View File

@ -25,7 +25,6 @@ enum NormalDecodeMode_t
// Forward declaration
#ifdef _WIN32
typedef enum _D3DFORMAT D3DFORMAT;
typedef enum DXGI_FORMAT;
#endif
//-----------------------------------------------------------------------------

View File

@ -43,7 +43,7 @@ typedef unsigned short MDLHandle_t;
inline MDLHandle_t VoidPtrToMDLHandle( void *ptr )
{
return (MDLHandle_t)(int)ptr;
return (MDLHandle_t)(uintp)ptr;
}
enum

View File

@ -478,23 +478,23 @@ FORCEINLINE fltx4 ArcTan2SIMD( const fltx4 &a, const fltx4 &b )
return result;
}
FORCEINLINE fltx4 MaxSIMD( const fltx4 & a, const fltx4 & b ) // max(a,b)
FORCEINLINE fltx4 MaxSIMD( const fltx4 & a, const fltx4 & b ) // V_max(a,b)
{
fltx4 retVal;
SubFloat( retVal, 0 ) = max( SubFloat( a, 0 ), SubFloat( b, 0 ) );
SubFloat( retVal, 1 ) = max( SubFloat( a, 1 ), SubFloat( b, 1 ) );
SubFloat( retVal, 2 ) = max( SubFloat( a, 2 ), SubFloat( b, 2 ) );
SubFloat( retVal, 3 ) = max( SubFloat( a, 3 ), SubFloat( b, 3 ) );
SubFloat( retVal, 0 ) = V_max( SubFloat( a, 0 ), SubFloat( b, 0 ) );
SubFloat( retVal, 1 ) = V_max( SubFloat( a, 1 ), SubFloat( b, 1 ) );
SubFloat( retVal, 2 ) = V_max( SubFloat( a, 2 ), SubFloat( b, 2 ) );
SubFloat( retVal, 3 ) = V_max( SubFloat( a, 3 ), SubFloat( b, 3 ) );
return retVal;
}
FORCEINLINE fltx4 MinSIMD( const fltx4 & a, const fltx4 & b ) // min(a,b)
FORCEINLINE fltx4 MinSIMD( const fltx4 & a, const fltx4 & b ) // V_min(a,b)
{
fltx4 retVal;
SubFloat( retVal, 0 ) = min( SubFloat( a, 0 ), SubFloat( b, 0 ) );
SubFloat( retVal, 1 ) = min( SubFloat( a, 1 ), SubFloat( b, 1 ) );
SubFloat( retVal, 2 ) = min( SubFloat( a, 2 ), SubFloat( b, 2 ) );
SubFloat( retVal, 3 ) = min( SubFloat( a, 3 ), SubFloat( b, 3 ) );
SubFloat( retVal, 0 ) = V_min( SubFloat( a, 0 ), SubFloat( b, 0 ) );
SubFloat( retVal, 1 ) = V_min( SubFloat( a, 1 ), SubFloat( b, 1 ) );
SubFloat( retVal, 2 ) = V_min( SubFloat( a, 2 ), SubFloat( b, 2 ) );
SubFloat( retVal, 3 ) = V_min( SubFloat( a, 3 ), SubFloat( b, 3 ) );
return retVal;
}
@ -945,7 +945,7 @@ FORCEINLINE void TransposeSIMD( fltx4 & x, fltx4 & y, fltx4 & z, fltx4 & w )
// and replicate it to the whole return value.
FORCEINLINE fltx4 FindLowestSIMD3( const fltx4 & a )
{
float lowest = min( min( SubFloat(a, 0), SubFloat(a, 1) ), SubFloat(a, 2));
float lowest = V_min( V_min( SubFloat(a, 0), SubFloat(a, 1) ), SubFloat(a, 2));
return ReplicateX4(lowest);
}
@ -953,7 +953,7 @@ FORCEINLINE fltx4 FindLowestSIMD3( const fltx4 & a )
// and replicate it to the whole return value.
FORCEINLINE fltx4 FindHighestSIMD3( const fltx4 & a )
{
float highest = max( max( SubFloat(a, 0), SubFloat(a, 1) ), SubFloat(a, 2));
float highest = V_max( V_max( SubFloat(a, 0), SubFloat(a, 1) ), SubFloat(a, 2));
return ReplicateX4(highest);
}
@ -1167,12 +1167,12 @@ FORCEINLINE fltx4 ArcTan2SIMD( const fltx4 &a, const fltx4 &b )
// DivSIMD defined further down, since it uses ReciprocalSIMD
FORCEINLINE fltx4 MaxSIMD( const fltx4 & a, const fltx4 & b ) // max(a,b)
FORCEINLINE fltx4 MaxSIMD( const fltx4 & a, const fltx4 & b ) // V_max(a,b)
{
return __vmaxfp( a, b );
}
FORCEINLINE fltx4 MinSIMD( const fltx4 & a, const fltx4 & b ) // min(a,b)
FORCEINLINE fltx4 MinSIMD( const fltx4 & a, const fltx4 & b ) // V_min(a,b)
{
return __vminfp( a, b );
}
@ -1938,11 +1938,11 @@ FORCEINLINE fltx4 FindLowestSIMD3( const fltx4 & a )
compareOne = __vrlimi( compareOne, a, 8 | 4 , 1 );
// compareOne is [y,z,G,G]
fltx4 retval = MinSIMD( a, compareOne );
// retVal is [min(x,y), min(y,z), G, G]
// retVal is [min(x,y), V_min(y,z), G, G]
compareOne = __vrlimi( compareOne, a, 8 , 2);
// compareOne is [z, G, G, G]
retval = MinSIMD( retval, compareOne );
// retVal = [ min(min(x,y),z), G, G, G ]
// retVal = [ V_min(min(x,y),z), G, G, G ]
// splat the x component out to the whole vector and return
return SplatXSIMD( retval );
@ -1962,11 +1962,11 @@ FORCEINLINE fltx4 FindHighestSIMD3( const fltx4 & a )
compareOne = __vrlimi( compareOne, a, 8 | 4 , 1 );
// compareOne is [y,z,G,G]
fltx4 retval = MaxSIMD( a, compareOne );
// retVal is [max(x,y), max(y,z), G, G]
// retVal is [max(x,y), V_max(y,z), G, G]
compareOne = __vrlimi( compareOne, a, 8 , 2);
// compareOne is [z, G, G, G]
retval = MaxSIMD( retval, compareOne );
// retVal = [ max(max(x,y),z), G, G, G ]
// retVal = [ V_max(max(x,y),z), G, G, G ]
// splat the x component out to the whole vector and return
return SplatXSIMD( retval );
@ -2596,12 +2596,12 @@ FORCEINLINE fltx4 CmpInBoundsSIMD( const fltx4 & a, const fltx4 & b ) // (a <=
return AndSIMD( CmpLeSIMD(a,b), CmpGeSIMD(a, NegSIMD(b)) );
}
FORCEINLINE fltx4 MinSIMD( const fltx4 & a, const fltx4 & b ) // min(a,b)
FORCEINLINE fltx4 MinSIMD( const fltx4 & a, const fltx4 & b ) // V_min(a,b)
{
return _mm_min_ps( a, b );
}
FORCEINLINE fltx4 MaxSIMD( const fltx4 & a, const fltx4 & b ) // max(a,b)
FORCEINLINE fltx4 MaxSIMD( const fltx4 & a, const fltx4 & b ) // V_max(a,b)
{
return _mm_max_ps( a, b );
}
@ -2756,7 +2756,7 @@ FORCEINLINE fltx4 FindLowestSIMD3( const fltx4 &a )
compareOne = RotateLeft2( a );
// compareOne is [z, G, x, y]
retval = MinSIMD( retval, compareOne );
// retVal = [ min(min(x,y),z)..]
// retVal = [ V_min(min(x,y),z)..]
// splat the x component out to the whole vector and return
return SplatXSIMD( retval );
@ -2773,7 +2773,7 @@ FORCEINLINE fltx4 FindHighestSIMD3( const fltx4 &a )
compareOne = RotateLeft2( a );
// compareOne is [z, G, x, y]
retval = MaxSIMD( retval, compareOne );
// retVal = [ max(max(x,y),z)..]
// retVal = [ V_max(max(x,y),z)..]
// splat the x component out to the whole vector and return
return SplatXSIMD( retval );
@ -2931,6 +2931,12 @@ FORCEINLINE i32x4 IntShiftLeftWordSIMD(const i32x4 &vSrcA, const i32x4 &vSrcB)
// like this.
FORCEINLINE void ConvertStoreAsIntsSIMD(intx4 * RESTRICT pDest, const fltx4 &vSrc)
{
#ifdef COMPILER_MSVC64
(*pDest)[0] = SubFloat(vSrc, 0);
(*pDest)[1] = SubFloat(vSrc, 1);
(*pDest)[2] = SubFloat(vSrc, 2);
(*pDest)[3] = SubFloat(vSrc, 3);
#else
__m64 bottom = _mm_cvttps_pi32( vSrc );
__m64 top = _mm_cvttps_pi32( _mm_movehl_ps(vSrc,vSrc) );
@ -2938,6 +2944,7 @@ FORCEINLINE void ConvertStoreAsIntsSIMD(intx4 * RESTRICT pDest, const fltx4 &vSr
*reinterpret_cast<__m64 *>(&(*pDest)[2]) = top;
_mm_empty();
#endif
}

View File

@ -9,11 +9,11 @@
#ifndef MINMAX_H
#define MINMAX_H
#ifndef min
#define min(a,b) (((a) < (b)) ? (a) : (b))
#ifndef V_min
#define V_min(a,b) (((a) < (b)) ? (a) : (b))
#endif
#ifndef max
#define max(a,b) (((a) > (b)) ? (a) : (b))
#ifndef V_max
#define V_max(a,b) (((a) > (b)) ? (a) : (b))
#endif
#endif // MINMAX_H

View File

@ -123,12 +123,12 @@ inline CTimingInfo< T > CalcSmoothAverage_Struct( const T &value, int nTimes, co
{
if ( i != newValueIndex )
{
info.m_HighAverage = max( pInfo->m_Values[i].m_Average, info.m_HighAverage );
info.m_LowAverage = min( pInfo->m_Values[i].m_Average, info.m_LowAverage );
info.m_HighAverage = V_max( pInfo->m_Values[i].m_Average, info.m_HighAverage );
info.m_LowAverage = V_min( pInfo->m_Values[i].m_Average, info.m_LowAverage );
}
info.m_HighValue = max( pInfo->m_Values[i].m_Value, info.m_HighValue );
info.m_LowValue = min( pInfo->m_Values[i].m_Value, info.m_LowValue );
info.m_HighValue = V_max( pInfo->m_Values[i].m_Value, info.m_HighValue );
info.m_LowValue = V_min( pInfo->m_Values[i].m_Value, info.m_LowValue );
info.m_AverageValue += pInfo->m_Values[i].m_Value;
}

View File

@ -22,7 +22,7 @@ typedef unsigned char uint8;
#define POSIX 1
#endif
#if defined(__x86_64__) || defined(_WIN64)
#if (defined(__x86_64__) || defined(_WIN64)) && !defined(X64BITS)
#define X64BITS
#endif

View File

@ -1123,7 +1123,7 @@ FORCEINLINE void CBitRead::GrabNextDWord( bool bOverFlowImmediately )
}
else
{
Assert( reinterpret_cast<int>(m_pDataIn) + 3 < reinterpret_cast<int>(m_pBufferEnd));
Assert( reinterpret_cast<uintp>(m_pDataIn) + 3 < reinterpret_cast<uintp>(m_pBufferEnd));
m_nInBufWord = LittleDWord( *( m_pDataIn++ ) );
}
}

View File

@ -137,7 +137,7 @@ protected:
const int MAX_GROW = 128;
if ( m_nGrowSize * sizeof(T) > MAX_GROW )
{
m_nGrowSize = max( 1, MAX_GROW / sizeof(T) );
m_nGrowSize = V_max( 1, MAX_GROW / sizeof(T) );
}
}
#endif

View File

@ -2228,7 +2228,7 @@ bool KeyValues::WriteAsBinary( CUtlBuffer &buffer )
}
case TYPE_PTR:
{
buffer.PutUnsignedInt( (int)dat->m_pValue );
buffer.PutPtr( dat->m_pValue );
}
default:
@ -2323,7 +2323,7 @@ bool KeyValues::ReadAsBinary( CUtlBuffer &buffer )
}
case TYPE_PTR:
{
dat->m_pValue = (void*)buffer.GetUnsignedInt();
dat->m_pValue = buffer.GetPtr();
}
default:

View File

@ -331,7 +331,7 @@ bool bf_write::WriteBits(const void *pInData, int nBits)
}
// Align output to dword boundary
while (((unsigned long)pOut & 3) != 0 && nBitsLeft >= 8)
while (((uintp)pOut & 3) != 0 && nBitsLeft >= 8)
{
WriteUBitLong( *pOut, 8, false );
@ -808,7 +808,7 @@ void old_bf_read::ReadBits(void *pOutData, int nBits)
// align output to dword boundary
while( ((unsigned long)pOut & 3) != 0 && nBitsLeft >= 8 )
while( ((uintp)pOut & 3) != 0 && nBitsLeft >= 8 )
{
*pOut = (unsigned char)ReadUBitLong(8);
++pOut;

View File

@ -147,7 +147,7 @@ JustAfew:
// The low-order two bits of pb and nBuffer in total control the
// upfront work.
//
nFront = ((unsigned int)pb) & 3;
nFront = ((uintp)pb) & 3;
nBuffer -= nFront;
switch (nFront)
{

View File

@ -154,7 +154,7 @@ static HMODULE InternalLoadLibrary( const char *pName )
return LoadLibraryEx( pName, NULL, LOAD_WITH_ALTERED_SEARCH_PATH );
#endif
}
unsigned ThreadedLoadLibraryFunc( void *pParam )
uintp ThreadedLoadLibraryFunc( void *pParam )
{
ThreadedLoadLibaryContext_t *pContext = (ThreadedLoadLibaryContext_t*)pParam;
pContext->m_hLibrary = InternalLoadLibrary(pContext->m_pLibraryName);

View File

@ -483,7 +483,7 @@ void CBitRead::ReadBits(void *pOutData, int nBits)
// align output to dword boundary
while( ((unsigned long)pOut & 3) != 0 && nBitsLeft >= 8 )
while( ((uintp)pOut & 3) != 0 && nBitsLeft >= 8 )
{
*pOut = (unsigned char)ReadUBitLong(8);
++pOut;

View File

@ -10,7 +10,7 @@
#include "processor_detect_linux.cpp"
#elif defined( _X360 )
#elif defined( _X360 ) || defined( WIN64 )
bool CheckMMXTechnology(void) { return false; }
bool CheckSSETechnology(void) { return false; }