zzogl-pg: Here I go, splitting code into separate files again...

git-svn-id: http://pcsx2.googlecode.com/svn/trunk@3446 96395faa-99c1-11dd-bbfe-3dabce05a288
This commit is contained in:
arcum42 2010-07-10 08:08:36 +00:00
parent 088aa28cde
commit 4c6dc19aa0
9 changed files with 728 additions and 844 deletions

View File

@ -51,6 +51,7 @@ set(zzoglSources
GLWin32.cpp GLWin32.cpp
GLWinX11.cpp GLWinX11.cpp
GSmain.cpp GSmain.cpp
HostMemory.cpp
Mem.cpp Mem.cpp
memcpy_amd.cpp memcpy_amd.cpp
Mem_Swizzle.cpp Mem_Swizzle.cpp

View File

@ -177,8 +177,7 @@ template<int index> void _GSgifTransfer(u32 *pMem, u32 size)
break; break;
case 2: case 2:
//Move(); //ZeroGS::TransferLocalLocal();
//ZZLog::Error_Log("GIF_FLG_IMAGE MOVE");
break; break;
case 3: case 3:

View File

@ -0,0 +1,711 @@
/* ZeroGS KOSMOS
* Copyright (C) 2005-2006 zerofrog@gmail.com
*
* This program is free software; you can redistribute it and/or modify
* it under the terms of the GNU General Public License as published by
* the Free Software Foundation; either version 2 of the License, or
* (at your option) any later version.
*
* This program is distributed in the hope that it will be useful,
* but WITHOUT ANY WARRANTY; without even the implied warranty of
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
* GNU General Public License for more details.
*
* You should have received a copy of the GNU General Public License
* along with this program; if not, write to the Free Software
* Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA
*/
#include "GS.h"
#include <Cg/cg.h>
#include <Cg/cgGL.h>
#include <stdlib.h>
#include "Mem.h"
#include "x86.h"
#include "zerogs.h"
#include "targets.h"
namespace ZeroGS
{
extern CRangeManager s_RangeMngr; // manages overwritten memory
extern void ResolveInRange(int start, int end);
static vector<u8> s_vTempBuffer, s_vTransferCache;
static int gs_imageEnd = 0;
void GetRectMemAddress(int& start, int& end, int psm, int x, int y, int w, int h, int bp, int bw)
{
FUNCLOG
if (m_Blocks[psm].bpp == 0)
{
ZZLog::Error_Log("ZeroGS: Bad psm 0x%x.", psm);
start = 0;
end = 0x00400000;
return;
}
if (PSMT_ISZTEX(psm) || psm == PSMCT16S)
{
const BLOCK& b = m_Blocks[psm];
bw = (bw + b.width - 1) / b.width;
start = bp * 256 + ((y / b.height) * bw + (x / b.width)) * 0x2000;
end = bp * 256 + (((y + h - 1) / b.height) * bw + (x + w + b.width - 1) / b.width) * 0x2000;
}
else
{
// just take the addresses
switch (psm)
{
case PSMCT32:
case PSMCT24:
case PSMT8H:
case PSMT4HL:
case PSMT4HH:
start = 4 * getPixelAddress32(x, y, bp, bw);
end = 4 * getPixelAddress32(x + w - 1, y + h - 1, bp, bw) + 4;
break;
case PSMCT16:
start = 2 * getPixelAddress16(x, y, bp, bw);
end = 2 * getPixelAddress16(x + w - 1, y + h - 1, bp, bw) + 2;
break;
case PSMT8:
start = getPixelAddress8(x, y, bp, bw);
end = getPixelAddress8(x + w - 1, y + h - 1, bp, bw) + 1;
break;
case PSMT4:
{
start = getPixelAddress4(x, y, bp, bw) / 2;
int newx = ((x + w - 1 + 31) & ~31) - 1;
int newy = ((y + h - 1 + 15) & ~15) - 1;
end = (getPixelAddress4(max(newx, x), max(newy, y), bp, bw) + 2) / 2;
break;
}
}
}
}
void InitTransferHostLocal()
{
FUNCLOG
//if (g_bIsLost) return;
#if defined(ZEROGS_DEVBUILD)
if (gs.trxpos.dx + gs.imageWnew > gs.dstbuf.bw)
ZZLog::Warn_Log("Transfer error, width exceeded.");
#endif
//bool bHasFlushed = false;
gs.imageX = gs.trxpos.dx;
gs.imageY = gs.trxpos.dy;
gs.imageEndX = gs.imageX + gs.imageWnew;
gs.imageEndY = gs.imageY + gs.imageHnew;
assert(gs.imageEndX < 2048 && gs.imageEndY < 2048);
// hack! viewful joe
if (gs.dstbuf.psm == 63) gs.dstbuf.psm = 0;
int start, end;
GetRectMemAddress(start, end, gs.dstbuf.psm, gs.trxpos.dx, gs.trxpos.dy, gs.imageWnew, gs.imageHnew, gs.dstbuf.bp, gs.dstbuf.bw);
if (end > 0x00400000)
{
ZZLog::Warn_Log("Host local out of bounds!");
//gs.imageTransfer = -1;
end = 0x00400000;
}
gs_imageEnd = end;
if (vb[0].nCount > 0) Flush(0);
if (vb[1].nCount > 0) Flush(1);
//ZZLog::Prim_Log("trans: bp:%x x:%x y:%x w:%x h:%x\n", gs.dstbuf.bp, gs.trxpos.dx, gs.trxpos.dy, gs.imageWnew, gs.imageHnew);
}
void TransferHostLocal(const void* pbyMem, u32 nQWordSize)
{
FUNCLOG
// if (g_bIsLost) return;
int start, end;
GetRectMemAddress(start, end, gs.dstbuf.psm, gs.imageX, gs.imageY, gs.imageWnew, gs.imageHnew, gs.dstbuf.bp, gs.dstbuf.bw);
assert(start < gs_imageEnd);
end = gs_imageEnd;
// sometimes games can decompress to alpha channel of render target only, in this case
// do a resolve right away. wolverine x2
if (((gs.dstbuf.psm == PSMT8H) || (gs.dstbuf.psm == PSMT4HL) || (gs.dstbuf.psm == PSMT4HH)) && !(conf.settings().gust))
{
list<CRenderTarget*> listTransmissionUpdateTargs;
s_RTs.GetTargs(start, end, listTransmissionUpdateTargs);
for (list<CRenderTarget*>::iterator it = listTransmissionUpdateTargs.begin(); it != listTransmissionUpdateTargs.end(); ++it)
{
CRenderTarget* ptarg = *it;
if ((ptarg->status & CRenderTarget::TS_Virtual)) continue;
//ZZLog::Error_Log("Resolving to alpha channel.");
ptarg->Resolve();
}
}
s_RangeMngr.Insert(start, min(end, start + (int)nQWordSize*16));
const u8* porgend = (const u8*)pbyMem + 4 * nQWordSize;
if (s_vTransferCache.size() > 0)
{
int imagecache = s_vTransferCache.size();
s_vTempBuffer.resize(imagecache + nQWordSize*4);
memcpy(&s_vTempBuffer[0], &s_vTransferCache[0], imagecache);
memcpy(&s_vTempBuffer[imagecache], pbyMem, nQWordSize*4);
pbyMem = (const void*) & s_vTempBuffer[0];
porgend = &s_vTempBuffer[0] + s_vTempBuffer.size();
int wordinc = imagecache / 4;
if ((nQWordSize * 4 + imagecache) / 3 == ((nQWordSize + wordinc) * 4) / 3)
{
// can use the data
nQWordSize += wordinc;
}
}
int leftover = m_Blocks[gs.dstbuf.psm].TransferHostLocal(pbyMem, nQWordSize);
if (leftover > 0)
{
// copy the last gs.image24bitOffset to the cache
s_vTransferCache.resize(leftover);
memcpy(&s_vTransferCache[0], porgend - leftover, leftover);
}
else
{
s_vTransferCache.resize(0);
}
#if defined(_DEBUG)
if (g_bSaveTrans)
{
tex0Info t;
t.tbp0 = gs.dstbuf.bp;
t.tw = gs.imageWnew;
t.th = gs.imageHnew;
t.tbw = gs.dstbuf.bw;
t.psm = gs.dstbuf.psm;
SaveTex(&t, 0);
}
#endif
}
void InitTransferLocalHost()
{
FUNCLOG
assert(gs.trxpos.sx + gs.imageWnew <= 2048 && gs.trxpos.sy + gs.imageHnew <= 2048);
#if defined(ZEROGS_DEVBUILD)
if (gs.trxpos.sx + gs.imageWnew > gs.srcbuf.bw)
ZZLog::Warn_Log("Transfer error, width exceeded.");
#endif
gs.imageX = gs.trxpos.sx;
gs.imageY = gs.trxpos.sy;
gs.imageEndX = gs.imageX + gs.imageWnew;
gs.imageEndY = gs.imageY + gs.imageHnew;
s_vTransferCache.resize(0);
int start, end;
GetRectMemAddress(start, end, gs.srcbuf.psm, gs.trxpos.sx, gs.trxpos.sy, gs.imageWnew, gs.imageHnew, gs.srcbuf.bp, gs.srcbuf.bw);
ResolveInRange(start, end);
}
template <class T>
void TransferLocalHost(void* pbyMem, u32 nQWordSize, int& x, int& y, u8 *pstart, _readPixel_0 rp)
{
int i = x, j = y;
T* pbuf = (T*)pbyMem;
u32 nSize = nQWordSize * 16 / sizeof(T);
for (; i < gs.imageEndY; ++i)
{
for (; j < gs.imageEndX && nSize > 0; ++j, --nSize)
{
*pbuf++ = rp(pstart, j % 2048, i % 2048, gs.srcbuf.bw);
}
if (j >= gs.imageEndX)
{
assert(j == gs.imageEndX);
j = gs.trxpos.sx;
}
else
{
assert(nSize == 0);
break;
}
}
}
void TransferLocalHost_24(void* pbyMem, u32 nQWordSize, int& x, int& y, u8 *pstart, _readPixel_0 rp)
{
int i = x, j = y;
u8* pbuf = (u8*)pbyMem;
u32 nSize = nQWordSize * 16 / 3;
for (; i < gs.imageEndY; ++i)
{
for (; j < gs.imageEndX && nSize > 0; ++j, --nSize)
{
u32 p = rp(pstart, j % 2048, i % 2048, gs.srcbuf.bw);
pbuf[0] = (u8)p;
pbuf[1] = (u8)(p >> 8);
pbuf[2] = (u8)(p >> 16);
pbuf += 3;
}
if (j >= gs.imageEndX)
{
assert(j == gs.imageEndX);
j = gs.trxpos.sx;
}
else
{
assert(nSize == 0);
break;
}
}
}
// left/right, top/down
void TransferLocalHost(void* pbyMem, u32 nQWordSize)
{
FUNCLOG
assert(gs.imageTransfer == 1);
u8* pstart = g_pbyGSMemory + 256 * gs.srcbuf.bp;
int i = gs.imageY, j = gs.imageX;
switch (gs.srcbuf.psm)
{
case PSMCT32:
TransferLocalHost<u32>(pbyMem, nQWordSize, i, j, pstart, readPixel32_0);
break;
case PSMCT24:
TransferLocalHost_24(pbyMem, nQWordSize, i, j, pstart, readPixel24_0);
break;
case PSMCT16:
TransferLocalHost<u16>(pbyMem, nQWordSize, i, j, pstart, readPixel16_0);
break;
case PSMCT16S:
TransferLocalHost<u16>(pbyMem, nQWordSize, i, j, pstart, readPixel16S_0);
break;
case PSMT8:
TransferLocalHost<u8>(pbyMem, nQWordSize, i, j, pstart, readPixel8_0);
break;
case PSMT8H:
TransferLocalHost<u8>(pbyMem, nQWordSize, i, j, pstart, readPixel8H_0);
break;
case PSMT32Z:
TransferLocalHost<u32>(pbyMem, nQWordSize, i, j, pstart, readPixel32Z_0);
break;
case PSMT24Z:
TransferLocalHost_24(pbyMem, nQWordSize, i, j, pstart, readPixel24Z_0);
break;
case PSMT16Z:
TransferLocalHost<u16>(pbyMem, nQWordSize, i, j, pstart, readPixel16Z_0);
break;
case PSMT16SZ:
TransferLocalHost<u16>(pbyMem, nQWordSize, i, j, pstart, readPixel16SZ_0);
break;
default:
assert(0);
}
gs.imageY = i;
gs.imageX = j;
if (gs.imageY >= gs.imageEndY)
{
assert(gs.imageY == gs.imageEndY);
gs.imageTransfer = -1;
}
}
// dir depends on trxpos.dirx & trxpos.diry
void TransferLocalLocal()
{
FUNCLOG
//ZZLog::Error_Log("I'z in your code, transferring your memory...");
assert(gs.imageTransfer == 2);
assert(gs.trxpos.sx + gs.imageWnew < 2048 && gs.trxpos.sy + gs.imageHnew < 2048);
assert(gs.trxpos.dx + gs.imageWnew < 2048 && gs.trxpos.dy + gs.imageHnew < 2048);
assert((gs.srcbuf.psm&0x7) == (gs.dstbuf.psm&0x7));
if (gs.trxpos.sx + gs.imageWnew > gs.srcbuf.bw)
ZZLog::Warn_Log("Transfer error, src width exceeded.");
if (gs.trxpos.dx + gs.imageWnew > gs.dstbuf.bw)
ZZLog::Warn_Log("Transfer error, dst width exceeded.");
int srcstart, srcend, dststart, dstend;
GetRectMemAddress(srcstart, srcend, gs.srcbuf.psm, gs.trxpos.sx, gs.trxpos.sy, gs.imageWnew, gs.imageHnew, gs.srcbuf.bp, gs.srcbuf.bw);
GetRectMemAddress(dststart, dstend, gs.dstbuf.psm, gs.trxpos.dx, gs.trxpos.dy, gs.imageWnew, gs.imageHnew, gs.dstbuf.bp, gs.dstbuf.bw);
// resolve the targs
ResolveInRange(srcstart, srcend);
list<CRenderTarget*> listTargs;
s_RTs.GetTargs(dststart, dstend, listTargs);
for (list<CRenderTarget*>::iterator it = listTargs.begin(); it != listTargs.end(); ++it)
{
if (!((*it)->status & CRenderTarget::TS_Virtual))
{
(*it)->Resolve();
//(*it)->status |= CRenderTarget::TS_NeedUpdate;
}
}
u8* pSrcBuf = g_pbyGSMemory + gs.srcbuf.bp * 256;
u8* pDstBuf = g_pbyGSMemory + gs.dstbuf.bp * 256;
#define TRANSFERLOCALLOCAL(srcpsm, dstpsm, widthlimit) { \
if( (gs.imageWnew&widthlimit)!=0 ) break; \
assert( (gs.imageWnew&widthlimit)==0 && widthlimit <= 4); \
for(int i = gs.trxpos.sy, i2 = gs.trxpos.dy; i < gs.trxpos.sy+gs.imageHnew; i++, i2++) { \
for(int j = gs.trxpos.sx, j2 = gs.trxpos.dx; j < gs.trxpos.sx+gs.imageWnew; j+=widthlimit, j2+=widthlimit) { \
\
writePixel##dstpsm##_0(pDstBuf, j2%2048, i2%2048, \
readPixel##srcpsm##_0(pSrcBuf, j%2048, i%2048, gs.srcbuf.bw), gs.dstbuf.bw); \
\
if( widthlimit > 1 ) { \
writePixel##dstpsm##_0(pDstBuf, (j2+1)%2048, i2%2048, \
readPixel##srcpsm##_0(pSrcBuf, (j+1)%2048, i%2048, gs.srcbuf.bw), gs.dstbuf.bw); \
\
if( widthlimit > 2 ) { \
writePixel##dstpsm##_0(pDstBuf, (j2+2)%2048, i2%2048, \
readPixel##srcpsm##_0(pSrcBuf, (j+2)%2048, i%2048, gs.srcbuf.bw), gs.dstbuf.bw); \
\
if( widthlimit > 3 ) { \
writePixel##dstpsm##_0(pDstBuf, (j2+3)%2048, i2%2048, \
readPixel##srcpsm##_0(pSrcBuf, (j+3)%2048, i%2048, gs.srcbuf.bw), gs.dstbuf.bw); \
} \
} \
} \
} \
} \
} \
#define TRANSFERLOCALLOCAL_4(srcpsm, dstpsm) { \
assert( (gs.imageWnew%8) == 0 ); \
for(int i = gs.trxpos.sy, i2 = gs.trxpos.dy; i < gs.trxpos.sy+gs.imageHnew; ++i, ++i2) { \
for(int j = gs.trxpos.sx, j2 = gs.trxpos.dx; j < gs.trxpos.sx+gs.imageWnew; j+=8, j2+=8) { \
/* NOTE: the 2 conseq 4bit values are in NOT in the same byte */ \
u32 read = getPixelAddress##srcpsm##_0(j%2048, i%2048, gs.srcbuf.bw); \
u32 write = getPixelAddress##dstpsm##_0(j2%2048, i2%2048, gs.dstbuf.bw); \
pDstBuf[write] = (pDstBuf[write]&0xf0)|(pSrcBuf[read]&0x0f); \
\
read = getPixelAddress##srcpsm##_0((j+1)%2048, i%2048, gs.srcbuf.bw); \
write = getPixelAddress##dstpsm##_0((j2+1)%2048, i2%2048, gs.dstbuf.bw); \
pDstBuf[write] = (pDstBuf[write]&0x0f)|(pSrcBuf[read]&0xf0); \
\
read = getPixelAddress##srcpsm##_0((j+2)%2048, i%2048, gs.srcbuf.bw); \
write = getPixelAddress##dstpsm##_0((j2+2)%2048, i2%2048, gs.dstbuf.bw); \
pDstBuf[write] = (pDstBuf[write]&0xf0)|(pSrcBuf[read]&0x0f); \
\
read = getPixelAddress##srcpsm##_0((j+3)%2048, i%2048, gs.srcbuf.bw); \
write = getPixelAddress##dstpsm##_0((j2+3)%2048, i2%2048, gs.dstbuf.bw); \
pDstBuf[write] = (pDstBuf[write]&0x0f)|(pSrcBuf[read]&0xf0); \
\
read = getPixelAddress##srcpsm##_0((j+2)%2048, i%2048, gs.srcbuf.bw); \
write = getPixelAddress##dstpsm##_0((j2+2)%2048, i2%2048, gs.dstbuf.bw); \
pDstBuf[write] = (pDstBuf[write]&0xf0)|(pSrcBuf[read]&0x0f); \
\
read = getPixelAddress##srcpsm##_0((j+3)%2048, i%2048, gs.srcbuf.bw); \
write = getPixelAddress##dstpsm##_0((j2+3)%2048, i2%2048, gs.dstbuf.bw); \
pDstBuf[write] = (pDstBuf[write]&0x0f)|(pSrcBuf[read]&0xf0); \
\
read = getPixelAddress##srcpsm##_0((j+2)%2048, i%2048, gs.srcbuf.bw); \
write = getPixelAddress##dstpsm##_0((j2+2)%2048, i2%2048, gs.dstbuf.bw); \
pDstBuf[write] = (pDstBuf[write]&0xf0)|(pSrcBuf[read]&0x0f); \
\
read = getPixelAddress##srcpsm##_0((j+3)%2048, i%2048, gs.srcbuf.bw); \
write = getPixelAddress##dstpsm##_0((j2+3)%2048, i2%2048, gs.dstbuf.bw); \
pDstBuf[write] = (pDstBuf[write]&0x0f)|(pSrcBuf[read]&0xf0); \
} \
} \
} \
switch (gs.srcbuf.psm)
{
case PSMCT32:
if (gs.dstbuf.psm == PSMCT32)
{
TRANSFERLOCALLOCAL(32, 32, 2);
}
else
{
TRANSFERLOCALLOCAL(32, 32Z, 2);
}
break;
case PSMCT24:
if (gs.dstbuf.psm == PSMCT24)
{
TRANSFERLOCALLOCAL(24, 24, 4);
}
else
{
TRANSFERLOCALLOCAL(24, 24Z, 4);
}
break;
case PSMCT16:
switch (gs.dstbuf.psm)
{
case PSMCT16:
TRANSFERLOCALLOCAL(16, 16, 4);
break;
case PSMCT16S:
TRANSFERLOCALLOCAL(16, 16S, 4);
break;
case PSMT16Z:
TRANSFERLOCALLOCAL(16, 16Z, 4);
break;
case PSMT16SZ:
TRANSFERLOCALLOCAL(16, 16SZ, 4);
break;
}
break;
case PSMCT16S:
switch (gs.dstbuf.psm)
{
case PSMCT16:
TRANSFERLOCALLOCAL(16S, 16, 4);
break;
case PSMCT16S:
TRANSFERLOCALLOCAL(16S, 16S, 4);
break;
case PSMT16Z:
TRANSFERLOCALLOCAL(16S, 16Z, 4);
break;
case PSMT16SZ:
TRANSFERLOCALLOCAL(16S, 16SZ, 4);
break;
}
break;
case PSMT8:
if (gs.dstbuf.psm == PSMT8)
{
TRANSFERLOCALLOCAL(8, 8, 4);
}
else
{
TRANSFERLOCALLOCAL(8, 8H, 4);
}
break;
case PSMT4:
switch (gs.dstbuf.psm)
{
case PSMT4:
TRANSFERLOCALLOCAL_4(4, 4);
break;
case PSMT4HL:
TRANSFERLOCALLOCAL_4(4, 4HL);
break;
case PSMT4HH:
TRANSFERLOCALLOCAL_4(4, 4HH);
break;
}
break;
case PSMT8H:
if (gs.dstbuf.psm == PSMT8)
{
TRANSFERLOCALLOCAL(8H, 8, 4);
}
else
{
TRANSFERLOCALLOCAL(8H, 8H, 4);
}
break;
case PSMT4HL:
switch (gs.dstbuf.psm)
{
case PSMT4:
TRANSFERLOCALLOCAL_4(4HL, 4);
break;
case PSMT4HL:
TRANSFERLOCALLOCAL_4(4HL, 4HL);
break;
case PSMT4HH:
TRANSFERLOCALLOCAL_4(4HL, 4HH);
break;
}
break;
case PSMT4HH:
switch (gs.dstbuf.psm)
{
case PSMT4:
TRANSFERLOCALLOCAL_4(4HH, 4);
break;
case PSMT4HL:
TRANSFERLOCALLOCAL_4(4HH, 4HL);
break;
case PSMT4HH:
TRANSFERLOCALLOCAL_4(4HH, 4HH);
break;
}
break;
case PSMT32Z:
if (gs.dstbuf.psm == PSMCT32)
{
TRANSFERLOCALLOCAL(32Z, 32, 2);
}
else
{
TRANSFERLOCALLOCAL(32Z, 32Z, 2);
}
break;
case PSMT24Z:
if (gs.dstbuf.psm == PSMCT24)
{
TRANSFERLOCALLOCAL(24Z, 24, 4);
}
else
{
TRANSFERLOCALLOCAL(24Z, 24Z, 4);
}
break;
case PSMT16Z:
switch (gs.dstbuf.psm)
{
case PSMCT16:
TRANSFERLOCALLOCAL(16Z, 16, 4);
break;
case PSMCT16S:
TRANSFERLOCALLOCAL(16Z, 16S, 4);
break;
case PSMT16Z:
TRANSFERLOCALLOCAL(16Z, 16Z, 4);
break;
case PSMT16SZ:
TRANSFERLOCALLOCAL(16Z, 16SZ, 4);
break;
}
break;
case PSMT16SZ:
switch (gs.dstbuf.psm)
{
case PSMCT16:
TRANSFERLOCALLOCAL(16SZ, 16, 4);
break;
case PSMCT16S:
TRANSFERLOCALLOCAL(16SZ, 16S, 4);
break;
case PSMT16Z:
TRANSFERLOCALLOCAL(16SZ, 16Z, 4);
break;
case PSMT16SZ:
TRANSFERLOCALLOCAL(16SZ, 16SZ, 4);
break;
}
break;
}
g_MemTargs.ClearRange(dststart, dstend);
#ifdef DEVBUILD
if (g_bSaveTrans)
{
tex0Info t;
t.tbp0 = gs.dstbuf.bp;
t.tw = gs.imageWnew;
t.th = gs.imageHnew;
t.tbw = gs.dstbuf.bw;
t.psm = gs.dstbuf.psm;
SaveTex(&t, 0);
t.tbp0 = gs.srcbuf.bp;
t.tw = gs.imageWnew;
t.th = gs.imageHnew;
t.tbw = gs.srcbuf.bw;
t.psm = gs.srcbuf.psm;
SaveTex(&t, 0);
}
#endif
}
}

View File

@ -100,6 +100,7 @@
<Unit filename="../../GSmain.cpp" /> <Unit filename="../../GSmain.cpp" />
<Unit filename="../../GifTransfer.cpp" /> <Unit filename="../../GifTransfer.cpp" />
<Unit filename="../../GifTransfer.h" /> <Unit filename="../../GifTransfer.h" />
<Unit filename="../../HostMemory.cpp" />
<Unit filename="../Conf.cpp" /> <Unit filename="../Conf.cpp" />
<Unit filename="../Linux.cpp" /> <Unit filename="../Linux.cpp" />
<Unit filename="../Linux.h" /> <Unit filename="../Linux.h" />

View File

@ -1,5 +1,5 @@
/* ZeroGS KOSMOS /* ZeroGS KOSMOS
* Copyright (C) 2005-2006 zerorog@gmail.com * Copyright (C) 2005-2006 zerofrog@gmail.com
* *
* This program is free software; you can redistribute it and/or modify * This program is free software; you can redistribute it and/or modify
* it under the terms of the GNU General Public License as published by * it under the terms of the GNU General Public License as published by
@ -201,16 +201,8 @@ void tex0Write(int i, u32 *data)
// kh and others // kh and others
return; return;
} }
//ZeroGS::vb[i].uCurTex0.i64 = r->i64;
// ZeroGS::vb[i].uNextTex0Data[0] = r->ai32[0];
// ZeroGS::vb[i].uNextTex0Data[1] = r->ai32[1];
ZeroGS::vb[i].uNextTex0Data[0] = data[0];
ZeroGS::vb[i].uNextTex0Data[1] = data[1];
ZeroGS::vb[i].bNeedTexCheck = 1;
// don't update unless necessary // don't update unless necessary
if (PSMT_ISCLUT(psm)) if (PSMT_ISCLUT(psm))
{ {
if (ZeroGS::CheckChangeInClut(data[1], psm)) if (ZeroGS::CheckChangeInClut(data[1], psm))
@ -218,13 +210,17 @@ void tex0Write(int i, u32 *data)
// loading clut, so flush whole texture // loading clut, so flush whole texture
ZeroGS::vb[i].FlushTexData(); ZeroGS::vb[i].FlushTexData();
} }
// else if (r->CSA != (ZeroGS::vb[i].uCurTex0.CSA)) // else if ((data[1] & 0x1f780000) != (ZeroGS::vb[i].uCurTex0Data[1] & 0x1f780000))
else if ((data[1] & 0x1f780000) != (ZeroGS::vb[i].uCurTex0Data[1] & 0x1f780000)) else if (r->CSA != (ZeroGS::vb[i].uCurTex0.CSA))
{ {
// check if csa is the same!! (ffx bisaid island, grass) // check if csa is the same!! (ffx bisaid island, grass)
ZeroGS::Flush(i); // flush any previous entries ZeroGS::Flush(i); // flush any previous entries
} }
} }
ZeroGS::vb[i].uCurTex0.i64 = r->i64;
ZeroGS::vb[i].bNeedTexCheck = 1;
} }
void tex2Write(int i, u32 *data) void tex2Write(int i, u32 *data)

View File

@ -1,5 +1,5 @@
/* ZeroGS KOSMOS /* ZeroGS KOSMOS
* Copyright (C) 2005-2006 zerorog@gmail.com * Copyright (C) 2005-2006 zerofrog@gmail.com
* *
* This program is free software; you can redistribute it and/or modify * This program is free software; you can redistribute it and/or modify
* it under the terms of the GNU General Public License as published by * it under the terms of the GNU General Public License as published by

View File

@ -142,6 +142,7 @@
<ExcludedFromBuild Condition="'$(Configuration)|$(Platform)'=='Release|Win32'">true</ExcludedFromBuild> <ExcludedFromBuild Condition="'$(Configuration)|$(Platform)'=='Release|Win32'">true</ExcludedFromBuild>
</ClCompile> </ClCompile>
<ClCompile Include="..\GSmain.cpp" /> <ClCompile Include="..\GSmain.cpp" />
<ClCompile Include="..\HostMemory.cpp" />
<ClCompile Include="..\Mem.cpp" /> <ClCompile Include="..\Mem.cpp" />
<ClCompile Include="..\Mem_Swizzle.cpp" /> <ClCompile Include="..\Mem_Swizzle.cpp" />
<ClCompile Include="..\Mem_Tables.cpp" /> <ClCompile Include="..\Mem_Tables.cpp" />

View File

@ -269,6 +269,10 @@
RelativePath="..\GSmain.cpp" RelativePath="..\GSmain.cpp"
> >
</File> </File>
<File
RelativePath="..\HostMemory.cpp"
>
</File>
<File <File
RelativePath="..\Mem.cpp" RelativePath="..\Mem.cpp"
> >

View File

@ -2752,7 +2752,6 @@ namespace ZeroGS
{ {
CRangeManager s_RangeMngr; // manages overwritten memory CRangeManager s_RangeMngr; // manages overwritten memory
static int gs_imageEnd = 0;
void ResolveInRange(int start, int end) void ResolveInRange(int start, int end)
{ {
@ -2928,834 +2927,6 @@ void FlushTransferRanges(const tex0Info* ptex)
s_RangeMngr.Clear(); s_RangeMngr.Clear();
} }
static vector<u8> s_vTempBuffer, s_vTransferCache;
void InitTransferHostLocal()
{
FUNCLOG
//if (g_bIsLost) return;
#if defined(ZEROGS_DEVBUILD)
if (gs.trxpos.dx + gs.imageWnew > gs.dstbuf.bw)
ZZLog::Warn_Log("Transfer error, width exceeded.");
#endif
//bool bHasFlushed = false;
gs.imageX = gs.trxpos.dx;
gs.imageY = gs.trxpos.dy;
gs.imageEndX = gs.imageX + gs.imageWnew;
gs.imageEndY = gs.imageY + gs.imageHnew;
assert(gs.imageEndX < 2048 && gs.imageEndY < 2048);
// hack! viewful joe
if (gs.dstbuf.psm == 63) gs.dstbuf.psm = 0;
int start, end;
GetRectMemAddress(start, end, gs.dstbuf.psm, gs.trxpos.dx, gs.trxpos.dy, gs.imageWnew, gs.imageHnew, gs.dstbuf.bp, gs.dstbuf.bw);
if (end > 0x00400000)
{
ZZLog::Warn_Log("Host local out of bounds!");
//gs.imageTransfer = -1;
end = 0x00400000;
}
gs_imageEnd = end;
if (vb[0].nCount > 0) Flush(0);
if (vb[1].nCount > 0) Flush(1);
//ZZLog::Prim_Log("trans: bp:%x x:%x y:%x w:%x h:%x\n", gs.dstbuf.bp, gs.trxpos.dx, gs.trxpos.dy, gs.imageWnew, gs.imageHnew);
// if( !bHasFlushed && (vb[0].bNeedFrameCheck || vb[0].bNeedZCheck || vb[1].bNeedFrameCheck || vb[1].bNeedZCheck)) {
// FlushBoth();
// bHasFlushed = 1;
// }
//
// // for all ranges, flush the targets
// // check if new rect intersects with current rendering texture, if so, flush
// if( vb[0].nCount > 0 && vb[0].curprim.tme ) {
// int tstart, tend;
// GetRectMemAddress(tstart, tend, vb[0].tex0.psm, 0, 0, vb[0].tex0.tw, vb[0].tex0.th, vb[0].tex0.tbp0, vb[0].tex0.tbw);
//
// if( start < tend && end > tstart ) {
// FlushBoth();
// bHasFlushed = 1;
// }
// }
//
// if( !bHasFlushed && vb[1].nCount > 0 && vb[1].curprim.tme ) {
// int tstart, tend;
// GetRectMemAddress(tstart, tend, vb[1].tex0.psm, 0, 0, vb[1].tex0.tw, vb[1].tex0.th, vb[1].tex0.tbp0, vb[1].tex0.tbw);
//
// if( start < tend && end > tstart ) {
// FlushBoth();
// bHasFlushed = 1;
// }
// }
//ZeroGS::g_MemTargs.ClearRange(start, end);
//s_RangeMngr.Insert(start, end);
}
void TransferHostLocal(const void* pbyMem, u32 nQWordSize)
{
FUNCLOG
// if (g_bIsLost) return;
int start, end;
GetRectMemAddress(start, end, gs.dstbuf.psm, gs.imageX, gs.imageY, gs.imageWnew, gs.imageHnew, gs.dstbuf.bp, gs.dstbuf.bw);
assert(start < gs_imageEnd);
end = gs_imageEnd;
// sometimes games can decompress to alpha channel of render target only, in this case
// do a resolve right away. wolverine x2
if (((gs.dstbuf.psm == PSMT8H) || (gs.dstbuf.psm == PSMT4HL) || (gs.dstbuf.psm == PSMT4HH)) && !(conf.settings().gust))
{
list<CRenderTarget*> listTransmissionUpdateTargs;
s_RTs.GetTargs(start, end, listTransmissionUpdateTargs);
for (list<CRenderTarget*>::iterator it = listTransmissionUpdateTargs.begin(); it != listTransmissionUpdateTargs.end(); ++it)
{
CRenderTarget* ptarg = *it;
if ((ptarg->status & CRenderTarget::TS_Virtual)) continue;
//ZZLog::Error_Log("Resolving to alpha channel.");
ptarg->Resolve();
}
}
s_RangeMngr.Insert(start, min(end, start + (int)nQWordSize*16));
const u8* porgend = (const u8*)pbyMem + 4 * nQWordSize;
if (s_vTransferCache.size() > 0)
{
int imagecache = s_vTransferCache.size();
s_vTempBuffer.resize(imagecache + nQWordSize*4);
memcpy(&s_vTempBuffer[0], &s_vTransferCache[0], imagecache);
memcpy(&s_vTempBuffer[imagecache], pbyMem, nQWordSize*4);
pbyMem = (const void*) & s_vTempBuffer[0];
porgend = &s_vTempBuffer[0] + s_vTempBuffer.size();
int wordinc = imagecache / 4;
if ((nQWordSize * 4 + imagecache) / 3 == ((nQWordSize + wordinc) * 4) / 3)
{
// can use the data
nQWordSize += wordinc;
}
}
int leftover = m_Blocks[gs.dstbuf.psm].TransferHostLocal(pbyMem, nQWordSize);
if (leftover > 0)
{
// copy the last gs.image24bitOffset to the cache
s_vTransferCache.resize(leftover);
memcpy(&s_vTransferCache[0], porgend - leftover, leftover);
}
else
{
s_vTransferCache.resize(0);
}
#if defined(_DEBUG)
if (g_bSaveTrans)
{
tex0Info t;
t.tbp0 = gs.dstbuf.bp;
t.tw = gs.imageWnew;
t.th = gs.imageHnew;
t.tbw = gs.dstbuf.bw;
t.psm = gs.dstbuf.psm;
SaveTex(&t, 0);
}
#endif
}
#if 0
// left/right, top/down
//void TransferHostLocal(const void* pbyMem, u32 nQWordSize)
//{
// assert( gs.imageTransfer == 0 );
// u8* pstart = g_pbyGSMemory + gs.dstbuf.bp*256;
//
// const u8* pendbuf = (const u8*)pbyMem + nQWordSize*4;
// int i = gs.imageY, j = gs.imageX;
//
//#define DSTPSM gs.dstbuf.psm
//
//#define TRANSFERHOSTLOCAL(psm, T, widthlimit) { \
// const T* pbuf = (const T*)pbyMem; \
// u32 nSize = nQWordSize*(4/sizeof(T)); \
// assert( (nSize%widthlimit) == 0 && widthlimit <= 4 ); \
// if( ((gs.imageEndX-gs.trxpos.dx)%widthlimit) ) ZZLog::Error_Log("Bad Transmission! %d %d, psm: %d.", gs.trxpos.dx, gs.imageEndX, DSTPSM); \
// for(; i < gs.imageEndY; ++i) { \
// for(; j < gs.imageEndX && nSize > 0; j += widthlimit, nSize -= widthlimit, pbuf += widthlimit) { \
// /* write as many pixel at one time as possible */ \
// writePixel##psm##_0(pstart, j%2048, i%2048, pbuf[0], gs.dstbuf.bw); \
// \
// if( widthlimit > 1 ) { \
// writePixel##psm##_0(pstart, (j+1)%2048, i%2048, pbuf[1], gs.dstbuf.bw); \
// \
// if( widthlimit > 2 ) { \
// writePixel##psm##_0(pstart, (j+2)%2048, i%2048, pbuf[2], gs.dstbuf.bw); \
// \
// if( widthlimit > 3 ) { \
// writePixel##psm##_0(pstart, (j+3)%2048, i%2048, pbuf[3], gs.dstbuf.bw); \
// } \
// } \
// } \
// } \
// \
// if( j >= gs.imageEndX ) { assert(j == gs.imageEndX); j = gs.trxpos.dx; } \
// else { assert( nSize == 0 ); goto End; } \
// } \
//} \
//
//#define TRANSFERHOSTLOCAL_4(psm) { \
// const u8* pbuf = (const u8*)pbyMem; \
// u32 nSize = nQWordSize*8; \
// for(; i < gs.imageEndY; ++i) { \
// for(; j < gs.imageEndX && nSize > 0; j += 8, nSize -= 8) { \
// /* write as many pixel at one time as possible */ \
// writePixel##psm##_0(pstart, j%2048, i%2048, *pbuf&0x0f, gs.dstbuf.bw); \
// writePixel##psm##_0(pstart, (j+1)%2048, i%2048, *pbuf>>4, gs.dstbuf.bw); \
// pbuf++; \
// writePixel##psm##_0(pstart, (j+2)%2048, i%2048, *pbuf&0x0f, gs.dstbuf.bw); \
// writePixel##psm##_0(pstart, (j+3)%2048, i%2048, *pbuf>>4, gs.dstbuf.bw); \
// pbuf++; \
// writePixel##psm##_0(pstart, (j+4)%2048, i%2048, *pbuf&0x0f, gs.dstbuf.bw); \
// writePixel##psm##_0(pstart, (j+5)%2048, i%2048, *pbuf>>4, gs.dstbuf.bw); \
// pbuf++; \
// writePixel##psm##_0(pstart, (j+6)%2048, i%2048, *pbuf&0x0f, gs.dstbuf.bw); \
// writePixel##psm##_0(pstart, (j+7)%2048, i%2048, *pbuf>>4, gs.dstbuf.bw); \
// pbuf++; \
// } \
// \
// if( j >= gs.imageEndX ) { /*assert(j == gs.imageEndX);*/ j = gs.trxpos.dx; } \
// else { assert( nSize == 0 ); goto End; } \
// } \
//} \
//
// switch (gs.dstbuf.psm) {
// case 0x0: TRANSFERHOSTLOCAL(32, u32, 2); break;
// case 0x1: TRANSFERHOSTLOCAL(24, u32, 4); break;
// case 0x2: TRANSFERHOSTLOCAL(16, u16, 4); break;
// case 0xA: TRANSFERHOSTLOCAL(16S, u16, 4); break;
// case 0x13:
// if( ((gs.imageEndX-gs.trxpos.dx)%4) ) {
// TRANSFERHOSTLOCAL(8, u8, 1);
// }
// else {
// TRANSFERHOSTLOCAL(8, u8, 4);
// }
// break;
//
// case 0x14:
//// if( (gs.imageEndX-gs.trxpos.dx)%8 ) {
//// // hack
//// if( abs((int)nQWordSize*8 - (gs.imageEndY-i)*(gs.imageEndX-gs.trxpos.dx)+(j-gs.trxpos.dx)) <= 8 ) {
//// // don't transfer
//// ZZLog::Error_Log("bad texture 4: %d %d %d.", gs.trxpos.dx, gs.imageEndX, nQWordSize);
//// gs.imageEndX = gs.trxpos.dx + (gs.imageEndX-gs.trxpos.dx)&~7;
//// //i = gs.imageEndY;
//// //goto End;
//// gs.imageTransfer = -1;
//// }
//// }
// TRANSFERHOSTLOCAL_4(4);
// break;
// case 0x1B: TRANSFERHOSTLOCAL(8H, u8, 4); break;
// case 0x24: TRANSFERHOSTLOCAL_4(4HL); break;
// case 0x2C: TRANSFERHOSTLOCAL_4(4HH); break;
// case 0x30: TRANSFERHOSTLOCAL(32Z, u32, 2); break;
// case 0x31: TRANSFERHOSTLOCAL(24Z, u32, 4); break;
// case 0x32: TRANSFERHOSTLOCAL(16Z, u16, 4); break;
// case 0x3A: TRANSFERHOSTLOCAL(16SZ, u16, 4); break;
// }
//
//End:
// if( i >= gs.imageEndY ) {
// assert( i == gs.imageEndY );
// gs.imageTransfer = -1;
//
// if( g_bSaveTrans ) {
// tex0Info t;
// t.tbp0 = gs.dstbuf.bp;
// t.tw = gs.imageWnew;
// t.th = gs.imageHnew;
// t.tbw = gs.dstbuf.bw;
// t.psm = gs.dstbuf.psm;
// SaveTex(&t, 0);
// }
// }
// else {
// /* update new params */
// gs.imageY = i;
// gs.imageX = j;
// }
//}
#endif //if 0
void InitTransferLocalHost()
{
FUNCLOG
assert(gs.trxpos.sx + gs.imageWnew <= 2048 && gs.trxpos.sy + gs.imageHnew <= 2048);
#if defined(ZEROGS_DEVBUILD)
if (gs.trxpos.sx + gs.imageWnew > gs.srcbuf.bw)
ZZLog::Warn_Log("Transfer error, width exceeded.");
#endif
gs.imageX = gs.trxpos.sx;
gs.imageY = gs.trxpos.sy;
gs.imageEndX = gs.imageX + gs.imageWnew;
gs.imageEndY = gs.imageY + gs.imageHnew;
s_vTransferCache.resize(0);
int start, end;
GetRectMemAddress(start, end, gs.srcbuf.psm, gs.trxpos.sx, gs.trxpos.sy, gs.imageWnew, gs.imageHnew, gs.srcbuf.bp, gs.srcbuf.bw);
ResolveInRange(start, end);
}
template <class T>
void TransferLocalHost(void* pbyMem, u32 nQWordSize, int& x, int& y, u8 *pstart, _readPixel_0 rp)
{
int i = x, j = y;
T* pbuf = (T*)pbyMem;
u32 nSize = nQWordSize * 16 / sizeof(T);
for (; i < gs.imageEndY; ++i)
{
for (; j < gs.imageEndX && nSize > 0; ++j, --nSize)
{
*pbuf++ = rp(pstart, j % 2048, i % 2048, gs.srcbuf.bw);
}
if (j >= gs.imageEndX)
{
assert(j == gs.imageEndX);
j = gs.trxpos.sx;
}
else
{
assert(nSize == 0);
break;
}
}
}
void TransferLocalHost_24(void* pbyMem, u32 nQWordSize, int& x, int& y, u8 *pstart, _readPixel_0 rp)
{
int i = x, j = y;
u8* pbuf = (u8*)pbyMem;
u32 nSize = nQWordSize * 16 / 3;
for (; i < gs.imageEndY; ++i)
{
for (; j < gs.imageEndX && nSize > 0; ++j, --nSize)
{
u32 p = rp(pstart, j % 2048, i % 2048, gs.srcbuf.bw);
pbuf[0] = (u8)p;
pbuf[1] = (u8)(p >> 8);
pbuf[2] = (u8)(p >> 16);
pbuf += 3;
}
if (j >= gs.imageEndX)
{
assert(j == gs.imageEndX);
j = gs.trxpos.sx;
}
else
{
assert(nSize == 0);
break;
}
}
}
// left/right, top/down
void TransferLocalHost(void* pbyMem, u32 nQWordSize)
{
FUNCLOG
assert(gs.imageTransfer == 1);
u8* pstart = g_pbyGSMemory + 256 * gs.srcbuf.bp;
int i = gs.imageY, j = gs.imageX;
switch (gs.srcbuf.psm)
{
case PSMCT32:
TransferLocalHost<u32>(pbyMem, nQWordSize, i, j, pstart, readPixel32_0);
break;
case PSMCT24:
TransferLocalHost_24(pbyMem, nQWordSize, i, j, pstart, readPixel24_0);
break;
case PSMCT16:
TransferLocalHost<u16>(pbyMem, nQWordSize, i, j, pstart, readPixel16_0);
break;
case PSMCT16S:
TransferLocalHost<u16>(pbyMem, nQWordSize, i, j, pstart, readPixel16S_0);
break;
case PSMT8:
TransferLocalHost<u8>(pbyMem, nQWordSize, i, j, pstart, readPixel8_0);
break;
case PSMT8H:
TransferLocalHost<u8>(pbyMem, nQWordSize, i, j, pstart, readPixel8H_0);
break;
case PSMT32Z:
TransferLocalHost<u32>(pbyMem, nQWordSize, i, j, pstart, readPixel32Z_0);
break;
case PSMT24Z:
TransferLocalHost_24(pbyMem, nQWordSize, i, j, pstart, readPixel24Z_0);
break;
case PSMT16Z:
TransferLocalHost<u16>(pbyMem, nQWordSize, i, j, pstart, readPixel16Z_0);
break;
case PSMT16SZ:
TransferLocalHost<u16>(pbyMem, nQWordSize, i, j, pstart, readPixel16SZ_0);
break;
default:
assert(0);
}
gs.imageY = i;
gs.imageX = j;
if (gs.imageY >= gs.imageEndY)
{
assert(gs.imageY == gs.imageEndY);
gs.imageTransfer = -1;
}
}
// dir depends on trxpos.dirx & trxpos.diry
void TransferLocalLocal()
{
FUNCLOG
assert(gs.imageTransfer == 2);
assert(gs.trxpos.sx + gs.imageWnew < 2048 && gs.trxpos.sy + gs.imageHnew < 2048);
assert(gs.trxpos.dx + gs.imageWnew < 2048 && gs.trxpos.dy + gs.imageHnew < 2048);
assert((gs.srcbuf.psm&0x7) == (gs.dstbuf.psm&0x7));
if (gs.trxpos.sx + gs.imageWnew > gs.srcbuf.bw)
ZZLog::Warn_Log("Transfer error, src width exceeded.");
if (gs.trxpos.dx + gs.imageWnew > gs.dstbuf.bw)
ZZLog::Warn_Log("Transfer error, dst width exceeded.");
int srcstart, srcend, dststart, dstend;
GetRectMemAddress(srcstart, srcend, gs.srcbuf.psm, gs.trxpos.sx, gs.trxpos.sy, gs.imageWnew, gs.imageHnew, gs.srcbuf.bp, gs.srcbuf.bw);
GetRectMemAddress(dststart, dstend, gs.dstbuf.psm, gs.trxpos.dx, gs.trxpos.dy, gs.imageWnew, gs.imageHnew, gs.dstbuf.bp, gs.dstbuf.bw);
// resolve the targs
ResolveInRange(srcstart, srcend);
list<CRenderTarget*> listTargs;
s_RTs.GetTargs(dststart, dstend, listTargs);
for (list<CRenderTarget*>::iterator it = listTargs.begin(); it != listTargs.end(); ++it)
{
if (!((*it)->status & CRenderTarget::TS_Virtual))
{
(*it)->Resolve();
//(*it)->status |= CRenderTarget::TS_NeedUpdate;
}
}
u8* pSrcBuf = g_pbyGSMemory + gs.srcbuf.bp * 256;
u8* pDstBuf = g_pbyGSMemory + gs.dstbuf.bp * 256;
#define TRANSFERLOCALLOCAL(srcpsm, dstpsm, widthlimit) { \
if( (gs.imageWnew&widthlimit)!=0 ) break; \
assert( (gs.imageWnew&widthlimit)==0 && widthlimit <= 4); \
for(int i = gs.trxpos.sy, i2 = gs.trxpos.dy; i < gs.trxpos.sy+gs.imageHnew; i++, i2++) { \
for(int j = gs.trxpos.sx, j2 = gs.trxpos.dx; j < gs.trxpos.sx+gs.imageWnew; j+=widthlimit, j2+=widthlimit) { \
\
writePixel##dstpsm##_0(pDstBuf, j2%2048, i2%2048, \
readPixel##srcpsm##_0(pSrcBuf, j%2048, i%2048, gs.srcbuf.bw), gs.dstbuf.bw); \
\
if( widthlimit > 1 ) { \
writePixel##dstpsm##_0(pDstBuf, (j2+1)%2048, i2%2048, \
readPixel##srcpsm##_0(pSrcBuf, (j+1)%2048, i%2048, gs.srcbuf.bw), gs.dstbuf.bw); \
\
if( widthlimit > 2 ) { \
writePixel##dstpsm##_0(pDstBuf, (j2+2)%2048, i2%2048, \
readPixel##srcpsm##_0(pSrcBuf, (j+2)%2048, i%2048, gs.srcbuf.bw), gs.dstbuf.bw); \
\
if( widthlimit > 3 ) { \
writePixel##dstpsm##_0(pDstBuf, (j2+3)%2048, i2%2048, \
readPixel##srcpsm##_0(pSrcBuf, (j+3)%2048, i%2048, gs.srcbuf.bw), gs.dstbuf.bw); \
} \
} \
} \
} \
} \
} \
#define TRANSFERLOCALLOCAL_4(srcpsm, dstpsm) { \
assert( (gs.imageWnew%8) == 0 ); \
for(int i = gs.trxpos.sy, i2 = gs.trxpos.dy; i < gs.trxpos.sy+gs.imageHnew; ++i, ++i2) { \
for(int j = gs.trxpos.sx, j2 = gs.trxpos.dx; j < gs.trxpos.sx+gs.imageWnew; j+=8, j2+=8) { \
/* NOTE: the 2 conseq 4bit values are in NOT in the same byte */ \
u32 read = getPixelAddress##srcpsm##_0(j%2048, i%2048, gs.srcbuf.bw); \
u32 write = getPixelAddress##dstpsm##_0(j2%2048, i2%2048, gs.dstbuf.bw); \
pDstBuf[write] = (pDstBuf[write]&0xf0)|(pSrcBuf[read]&0x0f); \
\
read = getPixelAddress##srcpsm##_0((j+1)%2048, i%2048, gs.srcbuf.bw); \
write = getPixelAddress##dstpsm##_0((j2+1)%2048, i2%2048, gs.dstbuf.bw); \
pDstBuf[write] = (pDstBuf[write]&0x0f)|(pSrcBuf[read]&0xf0); \
\
read = getPixelAddress##srcpsm##_0((j+2)%2048, i%2048, gs.srcbuf.bw); \
write = getPixelAddress##dstpsm##_0((j2+2)%2048, i2%2048, gs.dstbuf.bw); \
pDstBuf[write] = (pDstBuf[write]&0xf0)|(pSrcBuf[read]&0x0f); \
\
read = getPixelAddress##srcpsm##_0((j+3)%2048, i%2048, gs.srcbuf.bw); \
write = getPixelAddress##dstpsm##_0((j2+3)%2048, i2%2048, gs.dstbuf.bw); \
pDstBuf[write] = (pDstBuf[write]&0x0f)|(pSrcBuf[read]&0xf0); \
\
read = getPixelAddress##srcpsm##_0((j+2)%2048, i%2048, gs.srcbuf.bw); \
write = getPixelAddress##dstpsm##_0((j2+2)%2048, i2%2048, gs.dstbuf.bw); \
pDstBuf[write] = (pDstBuf[write]&0xf0)|(pSrcBuf[read]&0x0f); \
\
read = getPixelAddress##srcpsm##_0((j+3)%2048, i%2048, gs.srcbuf.bw); \
write = getPixelAddress##dstpsm##_0((j2+3)%2048, i2%2048, gs.dstbuf.bw); \
pDstBuf[write] = (pDstBuf[write]&0x0f)|(pSrcBuf[read]&0xf0); \
\
read = getPixelAddress##srcpsm##_0((j+2)%2048, i%2048, gs.srcbuf.bw); \
write = getPixelAddress##dstpsm##_0((j2+2)%2048, i2%2048, gs.dstbuf.bw); \
pDstBuf[write] = (pDstBuf[write]&0xf0)|(pSrcBuf[read]&0x0f); \
\
read = getPixelAddress##srcpsm##_0((j+3)%2048, i%2048, gs.srcbuf.bw); \
write = getPixelAddress##dstpsm##_0((j2+3)%2048, i2%2048, gs.dstbuf.bw); \
pDstBuf[write] = (pDstBuf[write]&0x0f)|(pSrcBuf[read]&0xf0); \
} \
} \
} \
switch (gs.srcbuf.psm)
{
case PSMCT32:
if (gs.dstbuf.psm == PSMCT32)
{
TRANSFERLOCALLOCAL(32, 32, 2);
}
else
{
TRANSFERLOCALLOCAL(32, 32Z, 2);
}
break;
case PSMCT24:
if (gs.dstbuf.psm == PSMCT24)
{
TRANSFERLOCALLOCAL(24, 24, 4);
}
else
{
TRANSFERLOCALLOCAL(24, 24Z, 4);
}
break;
case PSMCT16:
switch (gs.dstbuf.psm)
{
case PSMCT16:
TRANSFERLOCALLOCAL(16, 16, 4);
break;
case PSMCT16S:
TRANSFERLOCALLOCAL(16, 16S, 4);
break;
case PSMT16Z:
TRANSFERLOCALLOCAL(16, 16Z, 4);
break;
case PSMT16SZ:
TRANSFERLOCALLOCAL(16, 16SZ, 4);
break;
}
break;
case PSMCT16S:
switch (gs.dstbuf.psm)
{
case PSMCT16:
TRANSFERLOCALLOCAL(16S, 16, 4);
break;
case PSMCT16S:
TRANSFERLOCALLOCAL(16S, 16S, 4);
break;
case PSMT16Z:
TRANSFERLOCALLOCAL(16S, 16Z, 4);
break;
case PSMT16SZ:
TRANSFERLOCALLOCAL(16S, 16SZ, 4);
break;
}
break;
case PSMT8:
if (gs.dstbuf.psm == PSMT8)
{
TRANSFERLOCALLOCAL(8, 8, 4);
}
else
{
TRANSFERLOCALLOCAL(8, 8H, 4);
}
break;
case PSMT4:
switch (gs.dstbuf.psm)
{
case PSMT4:
TRANSFERLOCALLOCAL_4(4, 4);
break;
case PSMT4HL:
TRANSFERLOCALLOCAL_4(4, 4HL);
break;
case PSMT4HH:
TRANSFERLOCALLOCAL_4(4, 4HH);
break;
}
break;
case PSMT8H:
if (gs.dstbuf.psm == PSMT8)
{
TRANSFERLOCALLOCAL(8H, 8, 4);
}
else
{
TRANSFERLOCALLOCAL(8H, 8H, 4);
}
break;
case PSMT4HL:
switch (gs.dstbuf.psm)
{
case PSMT4:
TRANSFERLOCALLOCAL_4(4HL, 4);
break;
case PSMT4HL:
TRANSFERLOCALLOCAL_4(4HL, 4HL);
break;
case PSMT4HH:
TRANSFERLOCALLOCAL_4(4HL, 4HH);
break;
}
break;
case PSMT4HH:
switch (gs.dstbuf.psm)
{
case PSMT4:
TRANSFERLOCALLOCAL_4(4HH, 4);
break;
case PSMT4HL:
TRANSFERLOCALLOCAL_4(4HH, 4HL);
break;
case PSMT4HH:
TRANSFERLOCALLOCAL_4(4HH, 4HH);
break;
}
break;
case PSMT32Z:
if (gs.dstbuf.psm == PSMCT32)
{
TRANSFERLOCALLOCAL(32Z, 32, 2);
}
else
{
TRANSFERLOCALLOCAL(32Z, 32Z, 2);
}
break;
case PSMT24Z:
if (gs.dstbuf.psm == PSMCT24)
{
TRANSFERLOCALLOCAL(24Z, 24, 4);
}
else
{
TRANSFERLOCALLOCAL(24Z, 24Z, 4);
}
break;
case PSMT16Z:
switch (gs.dstbuf.psm)
{
case PSMCT16:
TRANSFERLOCALLOCAL(16Z, 16, 4);
break;
case PSMCT16S:
TRANSFERLOCALLOCAL(16Z, 16S, 4);
break;
case PSMT16Z:
TRANSFERLOCALLOCAL(16Z, 16Z, 4);
break;
case PSMT16SZ:
TRANSFERLOCALLOCAL(16Z, 16SZ, 4);
break;
}
break;
case PSMT16SZ:
switch (gs.dstbuf.psm)
{
case PSMCT16:
TRANSFERLOCALLOCAL(16SZ, 16, 4);
break;
case PSMCT16S:
TRANSFERLOCALLOCAL(16SZ, 16S, 4);
break;
case PSMT16Z:
TRANSFERLOCALLOCAL(16SZ, 16Z, 4);
break;
case PSMT16SZ:
TRANSFERLOCALLOCAL(16SZ, 16SZ, 4);
break;
}
break;
}
g_MemTargs.ClearRange(dststart, dstend);
#ifdef DEVBUILD
if (g_bSaveTrans)
{
tex0Info t;
t.tbp0 = gs.dstbuf.bp;
t.tw = gs.imageWnew;
t.th = gs.imageHnew;
t.tbw = gs.dstbuf.bw;
t.psm = gs.dstbuf.psm;
SaveTex(&t, 0);
t.tbp0 = gs.srcbuf.bp;
t.tw = gs.imageWnew;
t.th = gs.imageHnew;
t.tbw = gs.srcbuf.bw;
t.psm = gs.srcbuf.psm;
SaveTex(&t, 0);
}
#endif
}
void GetRectMemAddress(int& start, int& end, int psm, int x, int y, int w, int h, int bp, int bw)
{
FUNCLOG
if (m_Blocks[psm].bpp == 0)
{
ZZLog::Error_Log("ZeroGS: Bad psm 0x%x.", psm);
start = 0;
end = 0x00400000;
return;
}
if (PSMT_ISZTEX(psm) || psm == PSMCT16S)
{
const BLOCK& b = m_Blocks[psm];
bw = (bw + b.width - 1) / b.width;
start = bp * 256 + ((y / b.height) * bw + (x / b.width)) * 0x2000;
end = bp * 256 + (((y + h - 1) / b.height) * bw + (x + w + b.width - 1) / b.width) * 0x2000;
}
else
{
// just take the addresses
switch (psm)
{
case PSMCT32:
case PSMCT24:
case PSMT8H:
case PSMT4HL:
case PSMT4HH:
start = 4 * getPixelAddress32(x, y, bp, bw);
end = 4 * getPixelAddress32(x + w - 1, y + h - 1, bp, bw) + 4;
break;
case PSMCT16:
start = 2 * getPixelAddress16(x, y, bp, bw);
end = 2 * getPixelAddress16(x + w - 1, y + h - 1, bp, bw) + 2;
break;
case PSMT8:
start = getPixelAddress8(x, y, bp, bw);
end = getPixelAddress8(x + w - 1, y + h - 1, bp, bw) + 1;
break;
case PSMT4:
{
start = getPixelAddress4(x, y, bp, bw) / 2;
int newx = ((x + w - 1 + 31) & ~31) - 1;
int newy = ((y + h - 1 + 15) & ~15) - 1;
end = (getPixelAddress4(max(newx, x), max(newy, y), bp, bw) + 2) / 2;
break;
}
}
}
}
// I removed some code here that wasn't getting called. The old versions #if'ed out below this. // I removed some code here that wasn't getting called. The old versions #if'ed out below this.
#define RESOLVE_32_BIT(PSM, T, Tsrc, convfn) \ #define RESOLVE_32_BIT(PSM, T, Tsrc, convfn) \