H3cJP/source-engine
1
0
Fork 0
mirror of https://github.com/nillerusr/source-engine.git synced 2025-04-11 18:57:47 +00:00
Code Issues Actions Packages Projects Releases Wiki Activity

Merge 9cc3539f6d into b8558de63e

Browse Source
This commit is contained in:
Er2 2023-10-23 21:32:05 +02:00 committed by GitHub
commit 072921189c
No known key found for this signature in database
GPG Key ID: 4AEE18F83AFDEB23
43 changed files with 3234 additions and 2792 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

1
.gitignore vendored
View File

@ -1,3 +1,4 @@
*~
*.mak
*.mak.vpc_crc
*.vpc_crc

5657
common/sse2neon.h
View File

File diff suppressed because it is too large Load Diff

4
engine/cmodel.cpp
View File

@ -862,7 +862,7 @@ BOX TRACING
// Custom SIMD implementation for box brushes
const fltx4 Four_DistEpsilons={DIST_EPSILON,DIST_EPSILON,DIST_EPSILON,DIST_EPSILON};
const fltx4 Four_DistEpsilons=FLTX4(DIST_EPSILON,DIST_EPSILON,DIST_EPSILON,DIST_EPSILON);
const int32 ALIGN16 g_CubeFaceIndex0[4] ALIGN16_POST = {0,1,2,-1};
const int32 ALIGN16 g_CubeFaceIndex1[4] ALIGN16_POST = {3,4,5,-1};
bool IntersectRayWithBoxBrush( TraceInfo_t *pTraceInfo, const cbrush_t *pBrush, cboxbrush_t *pBox )
@ -1572,7 +1572,7 @@ void FASTCALL CM_TraceToLeaf( TraceInfo_t * RESTRICT pTraceInfo, int ndxLeaf, fl
fltx4 traceStart = LoadAlignedSIMD(pTraceInfo->m_start.Base());
fltx4 traceDelta = LoadAlignedSIMD(pTraceInfo->m_delta.Base());
fltx4 traceInvDelta = LoadAlignedSIMD(pTraceInfo->m_invDelta.Base());
static const fltx4 vecEpsilon = {DISPCOLL_DIST_EPSILON,DISPCOLL_DIST_EPSILON,DISPCOLL_DIST_EPSILON,DISPCOLL_DIST_EPSILON};
static const fltx4 vecEpsilon = FLTX4(DISPCOLL_DIST_EPSILON,DISPCOLL_DIST_EPSILON,DISPCOLL_DIST_EPSILON,DISPCOLL_DIST_EPSILON);
// only used in !IS_POINT version:
fltx4 extents;
if (!IS_POINT)

2
engine/gl_lightmap.cpp
View File

@ -1333,6 +1333,7 @@ void CacheAndUnloadLightmapData()
host_state.worldbrush->unloadedlightmaps = true;
}
#pragma optimize("", off)
//sorts the surfaces in place
static void SortSurfacesByLightmapID( SurfaceHandle_t *pToSort, int iSurfaceCount )
{
@ -1448,6 +1449,7 @@ void R_RedownloadAllLightmaps()
g_RebuildLightmaps = false;
}
#pragma optimize("", on)
//-----------------------------------------------------------------------------
// Purpose: flag the lightmaps as needing to be rebuilt (gamma change)

2
engine/l_studio.cpp
View File

@ -40,7 +40,7 @@
#include "materialsystem/materialsystem_config.h"
#include "materialsystem/itexture.h"
#include "IHammer.h"
#if defined( _WIN32 ) && !defined( _X360 )
#if defined( _WIN32 ) && defined( _M_IX86 )
#include <xmmintrin.h>
#endif
#include "staticpropmgr.h"

2
engine/sys_engine.cpp
View File

@ -104,7 +104,7 @@ extern ConVar host_timer_spin_ms;
extern float host_nexttick;
extern IVEngineClient *engineClient;
#ifdef WIN32
#if defined(_WIN32)
static void cpu_frequency_monitoring_callback( IConVar *var, const char *pOldValue, float flOldValue )
{
// Set the specified interval for CPU frequency monitoring

2
game/client/client_virtualreality.cpp
View File

@ -152,7 +152,7 @@ CON_COMMAND( vr_toggle, "Toggles VR mode" )
// --------------------------------------------------------------------
// Purpose: Returns true if the matrix is orthonormal
// --------------------------------------------------------------------
bool IsOrthonormal ( VMatrix Mat, float fTolerance )
bool IsOrthonormal ( VMatrix& Mat, float fTolerance )
{
float LenFwd = Mat.GetForward().Length();
float LenUp = Mat.GetUp().Length();

4
game/client/detailobjectsystem.cpp
View File

@ -2122,8 +2122,8 @@ int CDetailObjectSystem::SortSpritesBackToFront( int nLeaf, const Vector &viewOr
#else
#define MANTISSA_LSB_OFFSET 0
#endif
static fltx4 Four_MagicNumbers={ MAGIC_NUMBER, MAGIC_NUMBER, MAGIC_NUMBER, MAGIC_NUMBER };
static fltx4 Four_255s={ 255.0, 255.0, 255.0, 255.0 };
static fltx4 Four_MagicNumbers=FLTX4( MAGIC_NUMBER, MAGIC_NUMBER, MAGIC_NUMBER, MAGIC_NUMBER );
static fltx4 Four_255s=FLTX4( 255.0, 255.0, 255.0, 255.0 );
static ALIGN16 int32 And255Mask[4] ALIGN16_POST = {0xff,0xff,0xff,0xff};
#define PIXMASK ( * ( reinterpret_cast< fltx4 *>( &And255Mask ) ) )

2
game/client/touch.cpp
View File

@ -23,7 +23,7 @@ extern ConVar default_fov;
extern IMatSystemSurface *g_pMatSystemSurface;
#ifdef ANDROID
#ifdef PLATFORM_MOBILE
#define TOUCH_DEFAULT "1"
#else
#define TOUCH_DEFAULT "0"

6
gameui/OptionsDialog.cpp
View File

@ -87,9 +87,9 @@ COptionsDialog::COptionsDialog(vgui::Panel *parent) : PropertyDialog(parent, "Op
AddPage(new COptionsSubKeyboard(this), "#GameUI_Keyboard");
AddPage(new COptionsSubMouse(this), "#GameUI_Mouse");
#ifdef ANDROID
AddPage(new COptionsSubTouch(this), "Touch");
#endif
// Requires additional res file from extras_dir.vpk
if (IsMobile())
AddPage(new COptionsSubTouch(this), "Touch");
m_pOptionsSubAudio = new COptionsSubAudio(this);
AddPage(m_pOptionsSubAudio, "#GameUI_Audio");

2
ivp

@ -1 +1 @@
Subproject commit 4098acbbe3bc48320496f7533851640cc40cbb89
Subproject commit 2b7c913bb0c804eab2203eef30b9ba4559a44ef4

10
launcher/launcher.cpp
View File

@ -653,6 +653,11 @@ void ReportDirtyDiskNoMaterialSystem()
//-----------------------------------------------------------------------------
bool CSourceAppSystemGroup::Create()
{
if ( !CommandLine()->FindParm( "-nolog" ) )
DebugLogger()->Init("engine.log");
else
DebugLogger()->Disable();
IFileSystem *pFileSystem = (IFileSystem*)FindSystem( FILESYSTEM_INTERFACE_VERSION );
pFileSystem->InstallDirtyDiskReportFunc( ReportDirtyDiskNoMaterialSystem );
@ -767,11 +772,6 @@ bool CSourceAppSystemGroup::Create()
bool CSourceAppSystemGroup::PreInit()
{
if ( !CommandLine()->FindParm( "-nolog" ) )
DebugLogger()->Init("engine.log");
else
DebugLogger()->Disable();
CreateInterfaceFn factory = GetFactory();
ConnectTier1Libraries( &factory, 1 );
ConVar_Register( );

2
launcher_main/wscript
View File

@ -15,7 +15,7 @@ def configure(conf):
return
def build(bld):
if bld.env.DEST_OS == 'android':
if bld.env.MOBILE and bld.env.DEST_OS=='android':
return
source = ['main.cpp']

2
lib

@ -1 +1 @@
Subproject commit 86a66ee92d9fda0a09f54a435e850faa7ab5d0fa
Subproject commit 09cc8293779b1f2c879cca276e9a3f715ab5309d

2
materialsystem/colorspace.h
View File

@ -287,7 +287,7 @@ namespace ColorSpace
{
// preload 3.0f onto the returns so that we don't need to multiply the bumpAverage by it
// straight away (eg, reschedule this dependent op)
static const fltx4 vThree = { 3.0f, 3.0f, 3.0f, 0.0f };
static const fltx4 vThree = FLTX4( 3.0f, 3.0f, 3.0f, 0.0f );
fltx4 retValBump1 = MulSIMD( vThree, linearBumpColor1);
fltx4 retValBump2 = MulSIMD( vThree, linearBumpColor2);
fltx4 retValBump3 = MulSIMD( vThree, linearBumpColor3);

2
mathlib/3dnow.cpp
View File

@ -16,7 +16,7 @@
// memdbgon must be the last include file in a .cpp file!!!
#include "tier0/memdbgon.h"
#if !defined(COMPILER_MSVC64) && !defined(LINUX) && !defined(COMPILER_CLANG)
#if defined(_M_IX86) && !defined(LINUX) && !defined(COMPILER_CLANG)
// Implement for 64-bit Windows if needed.
// Clang hits "fatal error: error in backend:" and other errors when trying
// to compile the inline assembly below. 3DNow support is highly unlikely to

6
mathlib/mathlib_base.cpp
View File

@ -3258,7 +3258,7 @@ void MathLib_Init( float gamma, float texGamma, float brightness, int overbright
// SSE Generally performs better than 3DNow when present, so this is placed
// first to allow SSE to override these settings.
#if !defined( OSX ) && !defined( PLATFORM_WINDOWS_PC64 ) && !defined(LINUX) && !defined(PLATFORM_BSD)
#ifdef _M_IX86
if ( bAllow3DNow && pi.m_b3DNow )
{
s_b3DNowEnabled = true;
@ -3291,7 +3291,7 @@ void MathLib_Init( float gamma, float texGamma, float brightness, int overbright
pfRSqrt = _SSE_RSqrtAccurate;
pfRSqrtFast = _SSE_RSqrtFast;
#endif
#ifdef PLATFORM_WINDOWS_PC32
#ifdef _M_IX86
pfFastSinCos = _SSE_SinCos;
pfFastCos = _SSE_cos;
#endif
@ -3304,7 +3304,7 @@ void MathLib_Init( float gamma, float texGamma, float brightness, int overbright
if ( bAllowSSE2 && pi.m_bSSE2 )
{
s_bSSE2Enabled = true;
#ifdef PLATFORM_WINDOWS_PC32
#ifdef _M_IX86
pfFastSinCos = _SSE2_SinCos;
pfFastCos = _SSE2_cos;
#endif

26
mathlib/sse.cpp
View File

@ -91,13 +91,13 @@ float _SSE_Sqrt(float x)
{
Assert( s_bMathlibInitialized );
float root = 0.f;
#ifdef _WIN32
#if defined(_WIN32) && !defined(_M_ARM) && !defined(_M_ARM64)
_asm
{
sqrtss xmm0, x
movss root, xmm0
}
#elif POSIX
#else
_mm_store_ss( &root, _mm_sqrt_ss( _mm_load_ss( &x ) ) );
#endif
return root;
@ -122,7 +122,7 @@ float _SSE_RSqrtAccurate(float x)
}
#else
#ifdef POSIX
#if POSIX || defined(_M_ARM) || defined(_M_ARM64)
const __m128 f3 = _mm_set_ss(3.0f); // 3 as SSE value
const __m128 f05 = _mm_set_ss(0.5f); // 0.5 as SSE value
#endif
@ -131,7 +131,7 @@ const __m128 f05 = _mm_set_ss(0.5f); // 0.5 as SSE value
float _SSE_RSqrtAccurate(float a)
{
#ifdef _WIN32
#if defined(_WIN32) && !defined(_M_ARM) && !defined(_M_ARM64)
float x;
float half = 0.5f;
float three = 3.f;
@ -153,8 +153,8 @@ float _SSE_RSqrtAccurate(float a)
}
return x;
#elif POSIX
__m128 xx = _mm_load_ss( &a );
#else
__m128 xx = _mm_load_ss( &a );
__m128 xr = _mm_rsqrt_ss( xx );
__m128 xt;
@ -166,8 +166,6 @@ float _SSE_RSqrtAccurate(float a)
_mm_store_ss( &a, xr );
return a;
#else
#error "Not Implemented"
#endif
}
@ -764,7 +762,7 @@ float _SSE_cos( float x )
//-----------------------------------------------------------------------------
// SSE2 implementations of optimized routines:
//-----------------------------------------------------------------------------
#ifdef PLATFORM_WINDOWS_PC32
#if defined(_M_IX86)
void _SSE2_SinCos(float x, float* s, float* c) // any x
{
#ifdef _WIN32
@ -850,9 +848,7 @@ void _SSE2_SinCos(float x, float* s, float* c) // any x
#error "Not Implemented"
#endif
}
#endif // PLATFORM_WINDOWS_PC32
#ifdef PLATFORM_WINDOWS_PC32
float _SSE2_cos(float x)
{
#ifdef _WIN32
@ -970,9 +966,7 @@ void VectorTransformSSE(const float *in1, const matrix3x4_t& in2, float *out1)
#error "Not Implemented"
#endif
}
#endif
#if 0
void VectorRotateSSE( const float *in1, const matrix3x4_t& in2, float *out1 )
{
Assert( s_bMathlibInitialized );
@ -1026,9 +1020,7 @@ void VectorRotateSSE( const float *in1, const matrix3x4_t& in2, float *out1 )
#error "Not Implemented"
#endif
}
#endif
#ifdef _WIN32
void _declspec(naked) _SSE_VectorMA( const float *start, float scale, const float *direction, float *dest )
{
// FIXME: This don't work!! It will overwrite memory in the write to dest
@ -1057,7 +1049,6 @@ void _declspec(naked) _SSE_VectorMA( const float *start, float scale, const floa
#endif
}
}
#endif
#ifdef _WIN32
#ifdef PFN_VECTORMA
@ -1101,7 +1092,6 @@ float (__cdecl *pfVectorMA)(Vector& v) = _VectorMA;
// NJS: (Nov 1 2002) -NOT- faster. may time a couple cycles faster in a single function like
// this, but when inlined, and instruction scheduled, the C version is faster.
// Verified this via VTune
/*
vec_t DotProduct (const vec_t *a, const vec_t *c)
{
vec_t temp;
@ -1124,6 +1114,6 @@ vec_t DotProduct (const vec_t *a, const vec_t *c)
ret
}
}
*/
#endif
#endif // COMPILER_MSVC64

42
mathlib/sseconst.cpp
View File

@ -7,35 +7,35 @@
#include "mathlib/ssemath.h"
#include "mathlib/ssequaternion.h"
const fltx4 Four_PointFives={0.5,0.5,0.5,0.5};
const fltx4 Four_PointFives=FLTX4(0.5,0.5,0.5,0.5);
#ifndef _X360
const fltx4 Four_Zeros={0.0,0.0,0.0,0.0};
const fltx4 Four_Ones={1.0,1.0,1.0,1.0};
const fltx4 Four_Zeros=FLTX4(0.0,0.0,0.0,0.0);
const fltx4 Four_Ones=FLTX4(1.0,1.0,1.0,1.0);
#endif
const fltx4 Four_Twos={2.0,2.0,2.0,2.0};
const fltx4 Four_Threes={3.0,3.0,3.0,3.0};
const fltx4 Four_Fours={4.0,4.0,4.0,4.0};
const fltx4 Four_Origin={0,0,0,1};
const fltx4 Four_NegativeOnes={-1,-1,-1,-1};
const fltx4 Four_Twos=FLTX4(2.0,2.0,2.0,2.0);
const fltx4 Four_Threes=FLTX4(3.0,3.0,3.0,3.0);
const fltx4 Four_Fours=FLTX4(4.0,4.0,4.0,4.0);
const fltx4 Four_Origin=FLTX4(0,0,0,1);
const fltx4 Four_NegativeOnes=FLTX4(-1,-1,-1,-1);
const fltx4 Four_2ToThe21s={ (float) (1<<21), (float) (1<<21), (float) (1<<21), (float)(1<<21) };
const fltx4 Four_2ToThe22s={ (float) (1<<22), (float) (1<<22), (float) (1<<22), (float)(1<<22) };
const fltx4 Four_2ToThe23s={ (float) (1<<23), (float) (1<<23), (float) (1<<23), (float)(1<<23) };
const fltx4 Four_2ToThe24s={ (float) (1<<24), (float) (1<<24), (float) (1<<24), (float)(1<<24) };
const fltx4 Four_2ToThe21s=FLTX4( (float) (1<<21), (float) (1<<21), (float) (1<<21), (float)(1<<21) );
const fltx4 Four_2ToThe22s=FLTX4( (float) (1<<22), (float) (1<<22), (float) (1<<22), (float)(1<<22) );
const fltx4 Four_2ToThe23s=FLTX4( (float) (1<<23), (float) (1<<23), (float) (1<<23), (float)(1<<23) );
const fltx4 Four_2ToThe24s=FLTX4( (float) (1<<24), (float) (1<<24), (float) (1<<24), (float)(1<<24) );
const fltx4 Four_Point225s={ .225, .225, .225, .225 };
const fltx4 Four_Epsilons={FLT_EPSILON,FLT_EPSILON,FLT_EPSILON,FLT_EPSILON};
const fltx4 Four_Point225s=FLTX4( .225, .225, .225, .225 );
const fltx4 Four_Epsilons=FLTX4(FLT_EPSILON,FLT_EPSILON,FLT_EPSILON,FLT_EPSILON);
const fltx4 Four_FLT_MAX={FLT_MAX,FLT_MAX,FLT_MAX,FLT_MAX};
const fltx4 Four_Negative_FLT_MAX={-FLT_MAX,-FLT_MAX,-FLT_MAX,-FLT_MAX};
const fltx4 g_SIMD_0123 = { 0., 1., 2., 3. };
const fltx4 Four_FLT_MAX=FLTX4(FLT_MAX,FLT_MAX,FLT_MAX,FLT_MAX);
const fltx4 Four_Negative_FLT_MAX=FLTX4(-FLT_MAX,-FLT_MAX,-FLT_MAX,-FLT_MAX);
const fltx4 g_SIMD_0123 = FLTX4( 0., 1., 2., 3. );
const fltx4 g_QuatMultRowSign[4] =
{
{ 1.0f, 1.0f, -1.0f, 1.0f },
{ -1.0f, 1.0f, 1.0f, 1.0f },
{ 1.0f, -1.0f, 1.0f, 1.0f },
{ -1.0f, -1.0f, -1.0f, 1.0f }
FLTX4( 1.0f, 1.0f, -1.0f, 1.0f ),
FLTX4( -1.0f, 1.0f, 1.0f, 1.0f ),
FLTX4( 1.0f, -1.0f, 1.0f, 1.0f ),
FLTX4( -1.0f, -1.0f, -1.0f, 1.0f )
};
const uint32 ALIGN16 g_SIMD_clear_signmask[4] ALIGN16_POST = {0x7fffffff,0x7fffffff,0x7fffffff,0x7fffffff};

2
mathlib/ssenoise.cpp
View File

@ -20,7 +20,7 @@
#define MAGIC_NUMBER (1<<15) // gives 8 bits of fraction
static fltx4 Four_MagicNumbers = { MAGIC_NUMBER, MAGIC_NUMBER, MAGIC_NUMBER, MAGIC_NUMBER };
static fltx4 Four_MagicNumbers = FLTX4( MAGIC_NUMBER, MAGIC_NUMBER, MAGIC_NUMBER, MAGIC_NUMBER );
static ALIGN16 int32 idx_mask[4]= {0xffff, 0xffff, 0xffff, 0xffff};

29
public/materialsystem/imesh.h
View File

@ -1220,7 +1220,7 @@ inline void CVertexBuilder::FastVertexSSE( const ModelVertexDX7_t &vertex )
Assert( m_CompressionType == VERTEX_COMPRESSION_NONE ); // FIXME: support compressed verts if needed
Assert( m_nCurrentVertex < m_nMaxVertexCount );
#if defined( _WIN32 ) && !defined( _X360 ) && !defined( PLATFORM_64BITS )
#if defined( _WIN32 ) && defined( _M_IX86 )
const void *pRead = &vertex;
void *pCurrPos = m_pCurrPosition;
__asm
@ -1236,7 +1236,7 @@ inline void CVertexBuilder::FastVertexSSE( const ModelVertexDX7_t &vertex )
movntps [edi + 16], xmm1
movntps [edi + 32], xmm2
}
#elif defined(GNUC) || defined(PLATFORM_WINDOWS_PC64)
#else
const char *pRead = (char *)&vertex;
char *pCurrPos = (char *)m_pCurrPosition;
__m128 m1 = _mm_load_ps( (float *)pRead );
@ -1245,8 +1245,6 @@ inline void CVertexBuilder::FastVertexSSE( const ModelVertexDX7_t &vertex )
_mm_stream_ps( (float *)pCurrPos, m1 );
_mm_stream_ps( (float *)((intp)pCurrPos + 16), m2 );
_mm_stream_ps( (float *)((intp)pCurrPos + 32), m3 );
#else
Error( "Implement CMeshBuilder::FastVertexSSE(dx7)" );
#endif
IncrementFloatPointer( m_pCurrPosition, m_VertexSize_Position );
@ -1267,7 +1265,7 @@ inline void CVertexBuilder::Fast4VerticesSSE(
Assert( m_CompressionType == VERTEX_COMPRESSION_NONE ); // FIXME: support compressed verts if needed
Assert( m_nCurrentVertex < m_nMaxVertexCount-3 );
#if defined( _WIN32 ) && !defined( _X360 ) && !defined( PLATFORM_64BITS )
#if defined( _WIN32 ) && defined( _M_IX86 )
void *pCurrPos = m_pCurrPosition;
__asm
{
@ -1309,7 +1307,7 @@ inline void CVertexBuilder::Fast4VerticesSSE(
movntps [edi + 80+96], xmm5
}
#elif defined(__arm__) || defined(PLATFORM_WINDOWS_PC64)
#else
const void *pReadA = &vtx_a;
const void *pReadB = &vtx_b;
const void *pReadC = &vtx_c;
@ -1341,8 +1339,6 @@ inline void CVertexBuilder::Fast4VerticesSSE(
_mm_stream_ps( (float *)((intp)pCurrPos + 48+96), m4 );
_mm_stream_ps( (float *)((intp)pCurrPos + 64+96), m5 );
_mm_stream_ps( (float *)((intp)pCurrPos + 80+96), m6 );
#else
Error( "Implement CMeshBuilder::Fast4VerticesSSE\n");
#endif
IncrementFloatPointer( m_pCurrPosition, 4*m_VertexSize_Position );
@ -1430,7 +1426,7 @@ inline void CVertexBuilder::FastVertexSSE( const ModelVertexDX8_t &vertex )
Assert( m_CompressionType == VERTEX_COMPRESSION_NONE ); // FIXME: support compressed verts if needed
Assert( m_nCurrentVertex < m_nMaxVertexCount );
#if defined( _WIN32 ) && !defined( _X360 ) && !defined( PLATFORM_64BITS )
#if defined( _WIN32 ) && defined( _M_IX86 )
const void *pRead = &vertex;
void *pCurrPos = m_pCurrPosition;
__asm
@ -1448,21 +1444,10 @@ inline void CVertexBuilder::FastVertexSSE( const ModelVertexDX8_t &vertex )
movntps [edi + 32], xmm2
movntps [edi + 48], xmm3
}
#elif defined(GNUC) || defined(PLATFORM_WINDOWS_PC64)
#else
const void *pRead = &vertex;
void *pCurrPos = m_pCurrPosition;
/* __asm__ __volatile__ (
"movaps (%0), %%xmm0\n"
"movaps 16(%0), %%xmm1\n"
"movaps 32(%0), %%xmm2\n"
"movaps 48(%0), %%xmm3\n"
"movntps %%xmm0, (%1)\n"
"movntps %%xmm1, 16(%1)\n"
"movntps %%xmm2, 32(%1)\n"
"movntps %%xmm3, 48(%1)\n"
:: "r" (pRead), "r" (pCurrPos) : "memory"); */
__m128 m1 = _mm_load_ps( (float *)pRead );
__m128 m2 = _mm_load_ps( (float *)((intp)pRead + 16) );
__m128 m3 = _mm_load_ps( (float *)((intp)pRead + 32) );
@ -1471,8 +1456,6 @@ inline void CVertexBuilder::FastVertexSSE( const ModelVertexDX8_t &vertex )
_mm_stream_ps( (float *)((intp)pCurrPos + 16), m2 );
_mm_stream_ps( (float *)((intp)pCurrPos + 32), m3 );
_mm_stream_ps( (float *)((intp)pCurrPos + 48), m4 );
#else
Error( "Implement CMeshBuilder::FastVertexSSE((dx8)" );
#endif
IncrementFloatPointer( m_pCurrPosition, m_VertexSize_Position );

10
public/mathlib/mathlib.h
View File

@ -405,6 +405,9 @@ void inline SinCos( float radians, float *sine, float *cosine )
{
#if defined( _X360 )
XMScalarSinCos( sine, cosine, radians );
#elif defined( PLATFORM_WINDOWS_PC64 ) || defined(_M_ARM) || defined(_M_ARM64)
*sine = sin( radians );
*cosine = cos( radians );
#elif defined( PLATFORM_WINDOWS_PC32 )
_asm
{
@ -417,11 +420,8 @@ void inline SinCos( float radians, float *sine, float *cosine )
fstp DWORD PTR [edx]
fstp DWORD PTR [eax]
}
#elif defined( PLATFORM_WINDOWS_PC64 )
*sine = sin( radians );
*cosine = cos( radians );
#elif defined( OSX )
__sincosf(radians, sine, cosine);
__sincosf(radians, sine, cosine);
#elif defined( POSIX )
sincosf(radians, sine, cosine);
#endif
@ -1188,7 +1188,7 @@ inline float SimpleSplineRemapValClamped( float val, float A, float B, float C,
FORCEINLINE int RoundFloatToInt(float f)
{
#if defined(__i386__) || defined(_M_IX86) || defined( PLATFORM_WINDOWS_PC64 ) || defined(__x86_64__)
#if defined(__i386__) || defined(__x86_64__) || defined(_M_IX86) || defined(PLATFORM_WINDOWS_PC64)
return _mm_cvtss_si32(_mm_load_ss(&f));
#elif defined( _X360 )
#ifdef Assert

5
public/mathlib/simdvectormatrix.h
View File

@ -135,7 +135,12 @@ public:
Assert( m_pData );
static FourVectors value{Four_Zeros, Four_Zeros, Four_Zeros};
#if defined(_M_ARM) // fuck msvc with C2719 error
for (size_t n = m_nHeight*m_nPaddedWidth; n; n--)
*(m_pData+n) = value;
#else
memutils::set( m_pData, value, m_nHeight*m_nPaddedWidth );
#endif
}
void RaiseToPower( float power );

20
public/mathlib/ssemath.h
View File

@ -8,7 +8,7 @@
#if defined( _X360 )
#include <xboxmath.h>
#elif defined(__arm__) || defined(__aarch64__)
#elif defined(__arm__) || defined(__aarch64__) || defined(_M_ARM) || defined(_M_ARM64)
#include "sse2neon.h"
#else
#include <xmmintrin.h>
@ -63,6 +63,12 @@ typedef __m128 fltx4;
typedef __m128 i32x4;
typedef __m128 u32x4;
#if defined(_M_ARM) || defined(_M_ARM64)
#define FLTX4(w, x, y, z) {(w) + (unsigned long long(x) << 32), (y) + (unsigned long long(z) << 32)}
#else
#define FLTX4(w, x, y, z) {w, x, y, z}
#endif
#endif
// The FLTX4 type is a fltx4 used as a parameter to a function.
@ -1828,7 +1834,7 @@ FORCEINLINE fltx4 ReplicateX4( float flValue )
FORCEINLINE float SubFloat( const fltx4 & a, int idx )
{
// NOTE: if the output goes into a register, this causes a Load-Hit-Store stall (don't mix fpu/vpu math!)
#ifndef POSIX
#if defined(_WIN32) && (defined(__i386__) || defined(__x86_64__))
return a.m128_f32[ idx ];
#else
return (reinterpret_cast<float const *>(&a))[idx];
@ -1837,7 +1843,7 @@ FORCEINLINE float SubFloat( const fltx4 & a, int idx )
FORCEINLINE float & SubFloat( fltx4 & a, int idx )
{
#ifndef POSIX
#if defined(_WIN32) && (defined(__i386__) || defined(__x86_64__))
return a.m128_f32[ idx ];
#else
return (reinterpret_cast<float *>(&a))[idx];
@ -1851,8 +1857,8 @@ FORCEINLINE uint32 SubFloatConvertToInt( const fltx4 & a, int idx )
FORCEINLINE uint32 SubInt( const fltx4 & a, int idx )
{
#ifndef POSIX
return a.m128_u32[idx];
#if defined(_WIN32) && (defined(__i386__) || defined(__x86_64__))
return a.m128_u32[ idx ];
#else
return (reinterpret_cast<uint32 const *>(&a))[idx];
#endif
@ -1860,8 +1866,8 @@ FORCEINLINE uint32 SubInt( const fltx4 & a, int idx )
FORCEINLINE uint32 & SubInt( fltx4 & a, int idx )
{
#ifndef POSIX
return a.m128_u32[idx];
#if defined(_WIN32) && (defined(__i386__) || defined(__x86_64__))
return a.m128_u32[ idx ];
#else
return (reinterpret_cast<uint32 *>(&a))[idx];
#endif

2
public/tier0/commonmacros.h
View File

@ -68,7 +68,7 @@ inline bool IsPowerOfTwo( T value )
// From crtdefs.h
#if !defined(UNALIGNED)
#if defined(_M_IA64) || defined(_M_AMD64)
#if defined(_M_AMD64) || defined(_M_ARM) || defined(_M_ARM64)
#define UNALIGNED __unaligned
#else
#define UNALIGNED

54
public/tier0/platform.h
View File

@ -276,6 +276,13 @@ typedef signed char int8;
#else
#define IsAndroid() false
#endif
#ifdef PLATFORM_MOBILE
#define IsMobile() true
#else
#define IsMobile() false
#endif
// From steam/steamtypes.h
// RTime32
// We use this 32 bit time representing real world time.
@ -405,6 +412,18 @@ FIXME: Enable this when we no longer fear change =)
#define __i386__ 1
#endif
#ifdef _M_ARM64EC
#define _M_ARM64 1
#endif
#ifdef _M_ARM
#define __arm__ 1
#endif
#ifdef _M_ARM64
#define __aarch64__ 1
#endif
#elif POSIX
#if defined( OSX ) && defined( CARBON_WORKAROUND )
#define DWORD unsigned int
@ -852,7 +871,9 @@ static FORCEINLINE double fsel(double fComparand, double fValGE, double fLT)
//-----------------------------------------------------------------------------
//#define CHECK_FLOAT_EXCEPTIONS 1
#if !defined( _X360 )
#if defined (__arm__) || defined (__aarch64__)
inline void SetupFPUControlWord() {}
#elif !defined( _X360 )
#if defined( _MSC_VER )
#if defined( PLATFORM_WINDOWS_PC64 )
@ -898,8 +919,6 @@ static FORCEINLINE double fsel(double fComparand, double fValGE, double fLT)
#endif
#endif
#elif defined (__arm__) || defined (__aarch64__)
inline void SetupFPUControlWord() {}
#else
inline void SetupFPUControlWord()
{
@ -1025,7 +1044,7 @@ inline T QWordSwapC( T dw )
return output;
}
#elif defined( _MSC_VER ) && !defined( PLATFORM_WINDOWS_PC64 )
#elif defined( _MSC_VER ) && defined( _M_X86 )
#define WordSwap WordSwapAsm
#define DWordSwap DWordSwapAsm
@ -1229,8 +1248,18 @@ PLATFORM_INTERFACE time_t Plat_timegm( struct tm *timeptr );
PLATFORM_INTERFACE struct tm * Plat_localtime( const time_t *timep, struct tm *result );
#if defined( _WIN32 ) && defined( _MSC_VER ) && ( _MSC_VER >= 1400 )
#ifdef _M_X64
extern "C" unsigned __int64 __rdtsc();
#pragma intrinsic(__rdtsc)
#else
#include <intrin.h>
#define MSVC_ARM_SYSREG(op0, op1, crn, crm, op2) \
( ((op0 & 1) << 14) | \
((op1 & 7) << 11) | \
((crn & 15) << 7) | \
((crm & 15) << 3) | \
((op2 & 7) << 0) )
#endif
#endif
inline uint64 Plat_Rdtsc()
@ -1241,15 +1270,16 @@ inline uint64 Plat_Rdtsc()
return t.tv_sec * 1000000000ULL + t.tv_nsec;
#elif defined( _X360 )
return ( uint64 )__mftb32();
#elif defined( _WIN64 )
#elif defined( _M_IX86 )
_asm rdtsc
#elif defined( _M_ARM )
return __rdpmccntr64();
//uint32 val = _MoveFromCoprocessor(15,0, 9,13,0);
//return ((uint64)val) << 6;
#elif defined( _M_ARM64 ) || defined( _M_ARM64EC )
return _ReadStatusReg(MSVC_ARM_SYSREG(3,3, 9,12,5));
#elif defined( COMPILER_MSVC )
return ( uint64 )__rdtsc();
#elif defined( _WIN32 )
#if defined( _MSC_VER ) && ( _MSC_VER >= 1400 )
return ( uint64 )__rdtsc();
#else
__asm rdtsc;
__asm ret;
#endif
#elif defined( __i386__ )
uint64 val;
__asm__ __volatile__ ( "rdtsc" : "=A" (val) );

8
public/tier0/threadtools.h
View File

@ -241,6 +241,8 @@ inline void ThreadPause()
_mm_pause();
#elif defined( COMPILER_MSVC32 )
__asm pause;
#elif defined(_M_ARM) || defined(_M_ARM64)
__yield();
#elif defined( COMPILER_MSVCX360 )
YieldProcessor();
__asm { or r0,r0,r0 }
@ -445,7 +447,7 @@ PLATFORM_INTERFACE bool ThreadInterlockedAssignIf64( volatile int64 *pDest, int6
PLATFORM_INTERFACE int64 ThreadInterlockedExchange64( int64 volatile *, int64 value ) NOINLINE;
#ifdef COMPILER_MSVC32
#if COMPILER_MSVC32 || _M_ARM
PLATFORM_INTERFACE int64 ThreadInterlockedIncrement64( int64 volatile * ) NOINLINE;
PLATFORM_INTERFACE int64 ThreadInterlockedDecrement64( int64 volatile * ) NOINLINE;
PLATFORM_INTERFACE int64 ThreadInterlockedExchangeAdd64( int64 volatile *, int64 value ) NOINLINE;
@ -465,7 +467,7 @@ inline int64 ThreadInterlockedDecrement64( int64 volatile *p )
#endif
#ifdef COMPILER_MSVC64
#if COMPILER_MSVC64 || _M_ARM64
// 64 bit windows can use intrinsics for these, 32-bit can't
#pragma intrinsic( _InterlockedCompareExchange64 )
#pragma intrinsic( _InterlockedExchange64 )
@ -492,7 +494,7 @@ inline bool ThreadInterlockedAssignIf( uint32 volatile *p, uint32 value, uint32
//inline bool ThreadInterlockedAssignIf( int volatile *p, int value, int comperand ) { return ThreadInterlockedAssignIf( (int32 volatile *)p, value, comperand ); }
#if defined( _WIN64 )
#if defined( _WIN64 ) && !defined(_M_ARM64)
typedef __m128i int128;
inline int128 int128_zero() { return _mm_setzero_si128(); }
PLATFORM_INTERFACE bool ThreadInterlockedAssignIf128( volatile int128 *pDest, const int128 &value, const int128 &comperand ) NOINLINE;

8
public/tier0/tslist.h
View File

@ -34,7 +34,7 @@
//-----------------------------------------------------------------------------
#if defined( PLATFORM_64BITS )
#if defined( PLATFORM_64BITS ) && !defined(_M_ARM64)
#if defined (PLATFORM_WINDOWS)
//typedef __m128i int128;
@ -140,7 +140,7 @@ union TSLIST_HEAD_ALIGN TSLHead_t
uint32 DepthAndSequence;
} value32;
#ifdef PLATFORM_64BITS
#if defined(PLATFORM_64BITS) && !defined(_M_ARM64)
int128 value64x128;
#else
int64 value64x128;
@ -194,7 +194,7 @@ public:
#ifdef USE_NATIVE_SLIST
InitializeSListHead( &m_Head );
#elif defined(PLATFORM_64BITS)
#elif defined(PLATFORM_64BITS) && !defined(_M_ARM64)
m_Head.value64x128 = int128_zero();
#else
m_Head.value64x128 = (int64)0;
@ -720,7 +720,7 @@ public:
intp sequence;
} value;
#ifdef PLATFORM_64BITS
#if defined(PLATFORM_64BITS) && !defined(_M_ARM64)
int128 value64x128;
#else
int64 value64x128;

2
serverbrowser/BaseGamesPage.cpp
View File

@ -582,7 +582,7 @@ void CBaseGamesPage::LoadFilterSettings()
m_bFilterNoEmptyServers = filter->GetInt("NoEmpty");
m_bFilterNoPasswordedServers = filter->GetInt("NoPassword");
m_bFilterReplayServers = filter->GetInt("Replay");
m_pQuickListCheckButton->SetSelected( filter->GetInt( "QuickList", IsAndroid() ) );
m_pQuickListCheckButton->SetSelected( filter->GetInt( "QuickList", IsMobile() ) );
int secureFilter = filter->GetInt("Secure");
m_pSecureFilter->ActivateItem(secureFilter);

6
studiorender/r_studiodraw.cpp
View File

@ -657,7 +657,7 @@ static matrix3x4_t *ComputeSkinMatrix( mstudioboneweight_t &boneweights, matrix3
static matrix3x4_t *ComputeSkinMatrixSSE( mstudioboneweight_t &boneweights, matrix3x4_t *pPoseToWorld, matrix3x4_t &result )
{
// NOTE: pPoseToWorld, being cache aligned, doesn't need explicit initialization
#if defined( _WIN32 ) && !defined( _X360 ) && !defined( PLATFORM_64BITS )
#if defined( _WIN32 ) && defined(_M_IX86)
switch( boneweights.numbones )
{
default:
@ -866,11 +866,9 @@ static matrix3x4_t *ComputeSkinMatrixSSE( mstudioboneweight_t &boneweights, matr
return &result;
#endif
}
#elif POSIX || PLATFORM_WINDOWS_PC64
#elif POSIX || _WIN32
// #warning "ComputeSkinMatrixSSE C implementation only"
return ComputeSkinMatrix( boneweights, pPoseToWorld, result );
#elif defined( _X360 )
return ComputeSkinMatrix( boneweights, pPoseToWorld, result );
#else
#error
#endif

2
studiorender/r_studiolight.cpp
View File

@ -72,7 +72,7 @@ void R_LightAmbient_4D( const FourVectors& normal, Vector4D* pLightBoxColor, Fou
// VPROF( "R_LightAmbient" );
// !!speed!! compute ambient color cube in sse format
static fltx4 FourZeros={0.,0.,0.,.0};
static fltx4 FourZeros=FLTX4(0.,0.,0.,.0);
// find the contributions from each axis
fltx4 NegMask=CmpLtSIMD(normal.x,FourZeros);

3
studiorender/r_studiolight.h
View File

@ -12,8 +12,9 @@
#include "tier0/platform.h"
#include "mathlib/ssemath.h"
#if defined( _WIN32 ) && !defined( _X360 )
#if defined( _WIN32 ) && !defined( _X360 ) && (defined(_M_IX86) || defined(_M_AMD64))
#include <xmmintrin.h>
#endif

3
studiorender/studiorender.h
View File

@ -22,9 +22,6 @@
#include "flexrenderdata.h"
#include "mathlib/compressed_vector.h"
#include "r_studiolight.h"
#if defined( _WIN32 ) && !defined( _X360 )
#include <xmmintrin.h>
#endif
#include "tier0/dbg.h"

2
tier0/PMELib.cpp
View File

@ -6,7 +6,7 @@
//
//===========================================================================//
#ifdef _WIN32
#if defined(_WIN32) && !defined(_M_ARM) && !defined(_M_ARM64)
#include <windows.h>
#pragma warning( disable : 4530 ) // warning: exception handler -GX option

2
tier0/cpu.cpp
View File

@ -22,7 +22,7 @@ const tchar* GetProcessorVendorId();
static bool cpuid(uint32 function, uint32& out_eax, uint32& out_ebx, uint32& out_ecx, uint32& out_edx)
{
#if defined (__arm__) || defined (__aarch64__) || defined( _X360 )
#if !defined(__i386__) && !defined(__x86_64)
return false;
#elif defined(GNUC)

2
tier0/cpumonitoring.cpp
View File

@ -23,7 +23,7 @@
#include "pch_tier0.h"
#include "tier0/cpumonitoring.h"
#ifdef PLATFORM_WINDOWS_PC32
#ifdef _M_IX86
#include "tier0/threadtools.h"
#define NOMINMAX
#undef min

6
tier0/stacktools.cpp
View File

@ -175,7 +175,7 @@ int GetCallStack_Fast( void **pReturnAddressesOut, int iArrayCount, int iSkipCou
{
//Only tested in windows. This function won't work with frame pointer omission enabled. "vpc /nofpo" all projects
#if (defined( TIER0_FPO_DISABLED ) || defined( _DEBUG )) &&\
(defined( WIN32 ) && !defined( _X360 ) && !defined(_M_X64))
(defined( WIN32 ) && defined(_M_X86))
void *pStackCrawlEBP;
__asm
{
@ -1467,7 +1467,7 @@ CStackTop_CopyParentStack::CStackTop_CopyParentStack( void * const *pParentStack
#if defined( ENABLE_RUNTIME_STACK_TRANSLATION )
//miniature version of GetCallStack_Fast()
#if (defined( TIER0_FPO_DISABLED ) || defined( _DEBUG )) &&\
(defined( WIN32 ) && !defined( _X360 ) && !defined(_M_X64))
(defined( WIN32 ) && defined(_M_X86))
void *pStackCrawlEBP;
__asm
{
@ -1525,7 +1525,7 @@ CStackTop_ReferenceParentStack::CStackTop_ReferenceParentStack( void * const *pP
#if defined( ENABLE_RUNTIME_STACK_TRANSLATION )
//miniature version of GetCallStack_Fast()
#if (defined( TIER0_FPO_DISABLED ) || defined( _DEBUG )) &&\
(defined( WIN32 ) && !defined( _X360 ) && !defined(_M_X64))
(defined( WIN32 ) && defined(_M_X86))
void *pStackCrawlEBP;
__asm
{

14
tier0/threadtools.cpp
View File

@ -1740,7 +1740,7 @@ bool ThreadInterlockedAssignIf( int32 volatile *pDest, int32 value, int32 comper
{
Assert( (size_t)pDest % 4 == 0 );
#if !(defined(_WIN64) || defined (_X360))
#ifdef _M_IX86
__asm
{
mov eax,comperand
@ -1773,7 +1773,7 @@ void *ThreadInterlockedCompareExchangePointer( void * volatile *pDest, void *val
bool ThreadInterlockedAssignPointerIf( void * volatile *pDest, void *value, void *comperand )
{
Assert( (size_t)pDest % 4 == 0 );
#if !(defined(_WIN64) || defined (_X360))
#ifdef _M_IX86
__asm
{
mov eax,comperand
@ -1807,13 +1807,19 @@ int64 ThreadInterlockedCompareExchange64( int64 volatile *pDest, int64 value, in
lock CMPXCHG8B [esi];
}
}
#elif defined(_M_ARM)
int64 ThreadInterlockedCompareExchange64( int64 volatile *pDest, int64 value, int64 comperand )
{
Assert( (size_t)pDest % 8 == 0 );
return InterlockedCompareExchange64( pDest, value, comperand );
}
#endif
bool ThreadInterlockedAssignIf64(volatile int64 *pDest, int64 value, int64 comperand )
{
Assert( (size_t)pDest % 8 == 0 );
#if defined(_X360) || defined(_WIN64)
#if defined(_X360) || defined(_WIN64) || defined(_M_ARM) || defined(_M_ARM64)
return ( ThreadInterlockedCompareExchange64( pDest, value, comperand ) == comperand );
#else
__asm
@ -1833,7 +1839,7 @@ bool ThreadInterlockedAssignIf64(volatile int64 *pDest, int64 value, int64 compe
#endif
}
#ifdef _WIN64
#if defined(_WIN64) && !defined(_M_ARM64)
bool ThreadInterlockedAssignIf128( volatile int128 *pDest, const int128 &value, const int128 &comperand )
{
DbgAssert( ( (size_t)pDest % 16 ) == 0 );

5
tier0/wscript
View File

@ -54,10 +54,13 @@ def build(bld):
'assert_dialog.rc',
#'etwprof.cpp', [$WINDOWS]
'platform.cpp',
'pme.cpp',
'vcrmode.cpp',
'win32consoleio.cpp'
]
if bld.env.DEST_CPU in ['arm', 'arm64']:
source += ['pme_posix.cpp']
else:
source += ['pme.cpp']
if bld.env.DEST_CPU == 'amd64':
source += [
'InterlockedCompareExchange128.masm'

2
tier1/processor_detect.cpp
View File

@ -6,7 +6,7 @@
// $NoKeywords: $
//=============================================================================//
#if defined( _X360 ) || defined( WIN64 )
#if defined( _X360 ) || defined( WIN64 ) || defined(_M_ARM) || defined(_M_ARM64)
bool CheckMMXTechnology(void) { return false; }
bool CheckSSETechnology(void) { return false; }

4
vphysics/trace.cpp
View File

@ -453,7 +453,7 @@ private:
#ifdef WIN32
static const
#endif
fltx4 g_IVPToHLDir = { 1.0f, -1.0f, 1.0f, 1.0f };
fltx4 g_IVPToHLDir = FLTX4( 1.0f, -1.0f, 1.0f, 1.0f );
//static const fltx4 g_IVPToHLPosition = { IVP2HL(1.0f), -IVP2HL(1.0f), IVP2HL(1.0f), IVP2HL(1.0f) };
@ -680,7 +680,7 @@ bool CTraceIVP::BuildLeafmapCache( const leafmap_t * RESTRICT pLeafmap )
#endif
}
static const fltx4 g_IndexBase = {0,1,2,3};
static const fltx4 g_IndexBase =FLTX4(0,1,2,3);
int CTraceIVP::SupportMapCached( const Vector &dir, Vector *pOut ) const
{
VPROF("SupportMapCached");

10
vstdlib/coroutine.cpp
View File

@ -218,7 +218,11 @@ extern "C" byte *GetStackPtr64();
#define GetStackPtr( pStackPtr) byte *pStackPtr = GetStackPtr64();
#else
#ifdef WIN32
#define GetStackPtr( pStackPtr ) byte *pStackPtr; __asm mov pStackPtr, esp
# if defined(_M_ARM) || defined(_M_ARM64)
# define GetStackPtr( pStackPtr ) byte x; byte *pStackPtr = &x
# else
# define GetStackPtr( pStackPtr ) byte *pStackPtr; __asm mov pStackPtr, esp
# endif
#elif defined(GNUC)
// Apple's version of gcc/g++ doesn't return the expected value using the intrinsic, so
// do it the old fashioned way - this will also use asm on linux (since we don't compile
@ -649,7 +653,7 @@ bool Internal_Coroutine_Continue( HCoroutine hCoroutine, const char *pchDebugMsg
bool bInCoroutineAlready = GCoroutineMgr().IsAnyCoroutineActive();
#ifdef _WIN32
#ifndef _WIN64
#ifdef _M_IX86
// make sure nobody has a try/catch block and then yielded
// because we hate that and we will crash
uint32 topofexceptionchain;
@ -897,7 +901,7 @@ void Coroutine_YieldToMain()
CoroutineDbgMsg( g_fmtstr.sprintf( "Coroutine_YieldToMain() %s#%x -> %s#%x\n", coroutine.m_pchName, coroutine.m_hCoroutine, coroutinePrev.m_pchName, coroutinePrev.m_hCoroutine ) );
#ifdef _WIN32
#ifndef _WIN64
#ifdef _M_IX86
// make sure nobody has a try/catch block and then yielded
// because we hate that and we will crash
uint32 topofexceptionchain;

49
wscript
View File

@ -1,5 +1,6 @@
#! /usr/bin/env python
# encoding: utf-8
# vim: noexpandtab
# nillerusr
from __future__ import print_function
@ -189,6 +190,11 @@ def define_platform(conf):
if conf.options.ALLOW64:
conf.define('PLATFORM_64BITS', 1)
if (conf.env.DEST_OS == 'android' or
(conf.env.DEST_OS == 'win32' and conf.env.DEST_CPU in ['arm', 'arm64'])):
conf.env.MOBILE = True
conf.env.append_unique('DEFINES', ['PLATFORM_MOBILE=1'])
if conf.env.DEST_OS == 'linux':
conf.define('_GLIBCXX_USE_CXX11_ABI',0)
conf.env.append_unique('DEFINES', [
@ -388,14 +394,18 @@ def check_deps(conf):
conf.check(lib='opus', uselib_store='OPUS')
if conf.env.DEST_OS == 'win32':
conf.check(lib='libz', uselib_store='ZLIB', define_name='USE_ZLIB')
# conf.check(lib='nvtc', uselib_store='NVTC')
# conf.check(lib='ati_compress_mt_vc10', uselib_store='ATI_COMPRESS_MT_VC10')
conf.check(lib='SDL2', uselib_store='SDL2')
conf.check(lib='libjpeg', uselib_store='JPEG', define_name='HAVE_JPEG')
conf.check(lib='libpng', uselib_store='PNG', define_name='HAVE_PNG')
conf.check(lib='d3dx9', uselib_store='D3DX9')
conf.check(lib='d3d9', uselib_store='D3D9')
if conf.env.DEST_CPU in ['arm', 'arm64']:
conf.check(lib='d3d9', uselib_store='D3D9')
conf.check(lib='d3dcompiler', uselib_store='D3DCOMPILER')
conf.check(lib='d3dx9', uselib_store='D3DX9')
conf.check(lib='SDL2', uselib_store='SDL2')
else:
conf.check(lib='libz', uselib_store='ZLIB', define_name='USE_ZLIB')
conf.check(lib='SDL2', uselib_store='SDL2')
conf.check(lib='libjpeg', uselib_store='JPEG', define_name='HAVE_JPEG')
conf.check(lib='libpng', uselib_store='PNG', define_name='HAVE_PNG')
conf.check(lib='d3dx9', uselib_store='D3DX9')
conf.check(lib='d3d9', uselib_store='D3D9')
conf.check(lib='dsound', uselib_store='DSOUND')
conf.check(lib='dxguid', uselib_store='DXGUID')
if conf.options.OPUS:
@ -409,10 +419,12 @@ def configure(conf):
# Force XP compability, all build targets should add
# subsystem=bld.env.MSVC_SUBSYSTEM
# TODO: wrapper around bld.stlib, bld.shlib and so on?
conf.env.MSVC_SUBSYSTEM = 'WINDOWS,5.01'
conf.env.MSVC_TARGETS = ['x86'] # explicitly request x86 target for MSVC
if conf.options.ALLOW64:
conf.env.MSVC_TARGETS = ['x64']
conf.env.MSVC_TARGETS = ['x86' if not conf.options.ALLOW64 else 'x64']
if conf.env.MSVC_TARGETS[0] == 'x86':
conf.env.MSVC_SUBSYSTEM = 'WINDOWS,5.01'
else:
conf.env.MSVC_SUBSYSTEM = 'WINDOWS'
if sys.platform == 'win32':
conf.load('msvc_pdb_ext msdev msvs')
conf.load('subproject xcompile compiler_c compiler_cxx gitversion clang_compilation_database strip_on_install_v2 waf_unit_test enforce_pic')
@ -508,7 +520,6 @@ def configure(conf):
else:
cflags += [
'/I'+os.path.abspath('.')+'/thirdparty/SDL',
'/arch:SSE' if conf.env.DEST_CPU == 'x86' else '/arch:AVX',
'/GF',
'/Gy',
'/fp:fast',
@ -518,6 +529,8 @@ def configure(conf):
'/TP',
'/EHsc'
]
if conf.env.DEST_CPU in ['x86', 'x86_64', 'amd64']:
cflags += ['/arch:SSE' if conf.env.DEST_CPU == 'x86' else '/arch:AVX']
if conf.options.BUILD_TYPE == 'debug':
linkflags += [
@ -597,11 +610,19 @@ def configure(conf):
def build(bld):
os.environ["CCACHE_DIR"] = os.path.abspath('.ccache/'+bld.env.COMPILER_CC+'/'+bld.env.DEST_OS+'/'+bld.env.DEST_CPU)
base_lib_path = os.path.join('lib', bld.env.DEST_OS, bld.env.DEST_CPU)
if bld.env.DEST_OS in ['win32', 'android']:
sdl_name = 'SDL2.dll' if bld.env.DEST_OS == 'win32' else 'libSDL2.so'
sdl_path = os.path.join('lib', bld.env.DEST_OS, bld.env.DEST_CPU, sdl_name)
sdl_path = os.path.join(base_lib_path, sdl_name)
bld.install_files(bld.env.LIBDIR, [sdl_path])
if bld.env.DEST_OS == 'win32' and bld.env.DEST_CPU in ['arm', 'arm64']:
# because Windows ARM doesn't have D3DX and it is deprecated since Windows 8
bld.install_files(bld.env.LIBDIR, [
os.path.join(base_lib_path, 'd3dx9.dll'),
os.path.join(base_lib_path, 'd3dxof.dll'),
])
if bld.env.DEST_OS == 'win32':
projects['game'] += ['utils/bzip2']
projects['dedicated'] += ['utils/bzip2']