libretro/pcsx2
1
0
Fork 0
mirror of https://github.com/libretro/pcsx2.git synced 2024-11-23 17:29:46 +00:00
Code Issues Actions Packages Projects Releases Wiki Activity

tweaked FPU compare opcode clamping.

Browse Source 
this fixes a bug with Digimon rumble arena 2.
thanks to Nneeve for figuring out the problem.


git-svn-id: http://pcsx2.googlecode.com/svn/trunk@547 96395faa-99c1-11dd-bbfe-3dabce05a288
This commit is contained in:
cottonvibes 2009-02-20 05:37:01 +00:00
parent e56eac1862
commit 7e065afb4d
1 changed files with 29 additions and 20 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

49
pcsx2/x86/iFPU.cpp
View File

@ -404,6 +404,16 @@ void fpuFloat2(int regd) { // +NaN -> +fMax, -NaN -> -fMax, +Inf -> +fMax, -Inf
}
}
__forceinline void fpuFloat3(int regd) {
// This clamp function used in the recC_xx opcodes
// Rule of Rose needs clamping or else it crashes (minss or maxss both fix the crash)
// Digimon Rumble Arena 2 needs MAXSS clamping (if you only use minss, it spins on the intro-menus;
// it also doesn't like preserving NaN sign with fpuFloat2, so the only way to make Digimon work
// is by calling MAXSS first)
SSE_MAXSS_M32_to_XMM(regd, (uptr)&g_minvals[0]);
//SSE_MINSS_M32_to_XMM(regd, (uptr)&g_maxvals[0]);
}
void ClampValues(int regd) {
fpuFloat(regd);
}
@ -698,30 +708,30 @@ void recC_EQ_xmm(int info)
switch(info & (PROCESS_EE_S|PROCESS_EE_T) ) {
case PROCESS_EE_S:
SSE_MINSS_M32_to_XMM(EEREC_S, (uptr)&g_maxvals[0]);
fpuFloat3(EEREC_S);
t0reg = _allocTempXMMreg(XMMT_FPS, -1);
if (t0reg >= 0) {
SSE_MOVSS_M32_to_XMM(t0reg, (uptr)&fpuRegs.fpr[_Ft_]);
SSE_MINSS_M32_to_XMM(t0reg, (uptr)&g_maxvals[0]);
fpuFloat3(t0reg);
SSE_UCOMISS_XMM_to_XMM(EEREC_S, t0reg);
_freeXMMreg(t0reg);
}
else SSE_UCOMISS_M32_to_XMM(EEREC_S, (uptr)&fpuRegs.fpr[_Ft_]);
break;
case PROCESS_EE_T:
SSE_MINSS_M32_to_XMM(EEREC_T, (uptr)&g_maxvals[0]);
fpuFloat3(EEREC_T);
t0reg = _allocTempXMMreg(XMMT_FPS, -1);
if (t0reg >= 0) {
SSE_MOVSS_M32_to_XMM(t0reg, (uptr)&fpuRegs.fpr[_Fs_]);
SSE_MINSS_M32_to_XMM(t0reg, (uptr)&g_maxvals[0]);
fpuFloat3(t0reg);
SSE_UCOMISS_XMM_to_XMM(t0reg, EEREC_T);
_freeXMMreg(t0reg);
}
else SSE_UCOMISS_M32_to_XMM(EEREC_T, (uptr)&fpuRegs.fpr[_Fs_]);
break;
case (PROCESS_EE_S|PROCESS_EE_T):
SSE_MINSS_M32_to_XMM(EEREC_S, (uptr)&g_maxvals[0]);
SSE_MINSS_M32_to_XMM(EEREC_T, (uptr)&g_maxvals[0]);
fpuFloat3(EEREC_S);
fpuFloat3(EEREC_T);
SSE_UCOMISS_XMM_to_XMM(EEREC_S, EEREC_T);
break;
default:
@ -768,22 +778,22 @@ void recC_LE_xmm(int info )
switch(info & (PROCESS_EE_S|PROCESS_EE_T) ) {
case PROCESS_EE_S:
SSE_MINSS_M32_to_XMM(EEREC_S, (uptr)&g_maxvals[0]);
fpuFloat3(EEREC_S);
t0reg = _allocTempXMMreg(XMMT_FPS, -1);
if (t0reg >= 0) {
SSE_MOVSS_M32_to_XMM(t0reg, (uptr)&fpuRegs.fpr[_Ft_]);
SSE_MINSS_M32_to_XMM(t0reg, (uptr)&g_maxvals[0]);
fpuFloat3(t0reg);
SSE_UCOMISS_XMM_to_XMM(EEREC_S, t0reg);
_freeXMMreg(t0reg);
}
else SSE_UCOMISS_M32_to_XMM(EEREC_S, (uptr)&fpuRegs.fpr[_Ft_]);
break;
case PROCESS_EE_T:
SSE_MINSS_M32_to_XMM(EEREC_T, (uptr)&g_maxvals[0]);
fpuFloat3(EEREC_T);
t0reg = _allocTempXMMreg(XMMT_FPS, -1);
if (t0reg >= 0) {
SSE_MOVSS_M32_to_XMM(t0reg, (uptr)&fpuRegs.fpr[_Fs_]);
SSE_MINSS_M32_to_XMM(t0reg, (uptr)&g_maxvals[0]);
fpuFloat3(t0reg);
SSE_UCOMISS_XMM_to_XMM(t0reg, EEREC_T);
_freeXMMreg(t0reg);
}
@ -800,8 +810,8 @@ void recC_LE_xmm(int info )
}
break;
case (PROCESS_EE_S|PROCESS_EE_T):
SSE_MINSS_M32_to_XMM(EEREC_S, (uptr)&g_maxvals[0]);
SSE_MINSS_M32_to_XMM(EEREC_T, (uptr)&g_maxvals[0]);
fpuFloat3(EEREC_S);
fpuFloat3(EEREC_T);
SSE_UCOMISS_XMM_to_XMM(EEREC_S, EEREC_T);
break;
default: // Untested and incorrect, but this case is never reached AFAIK (cottonvibes)
@ -842,22 +852,22 @@ void recC_LT_xmm(int info)
switch(info & (PROCESS_EE_S|PROCESS_EE_T) ) {
case PROCESS_EE_S:
SSE_MINSS_M32_to_XMM(EEREC_S, (uptr)&g_maxvals[0]);
fpuFloat3(EEREC_S);
t0reg = _allocTempXMMreg(XMMT_FPS, -1);
if (t0reg >= 0) {
SSE_MOVSS_M32_to_XMM(t0reg, (uptr)&fpuRegs.fpr[_Ft_]);
SSE_MINSS_M32_to_XMM(t0reg, (uptr)&g_maxvals[0]);
fpuFloat3(t0reg);
SSE_UCOMISS_XMM_to_XMM(EEREC_S, t0reg);
_freeXMMreg(t0reg);
}
else SSE_UCOMISS_M32_to_XMM(EEREC_S, (uptr)&fpuRegs.fpr[_Ft_]);
break;
case PROCESS_EE_T:
SSE_MINSS_M32_to_XMM(EEREC_T, (uptr)&g_maxvals[0]);
fpuFloat3(EEREC_T);
t0reg = _allocTempXMMreg(XMMT_FPS, -1);
if (t0reg >= 0) {
SSE_MOVSS_M32_to_XMM(t0reg, (uptr)&fpuRegs.fpr[_Fs_]);
SSE_MINSS_M32_to_XMM(t0reg, (uptr)&g_maxvals[0]);
fpuFloat3(t0reg);
SSE_UCOMISS_XMM_to_XMM(t0reg, EEREC_T);
_freeXMMreg(t0reg);
}
@ -874,11 +884,10 @@ void recC_LT_xmm(int info)
}
break;
case (PROCESS_EE_S|PROCESS_EE_T):
// Makes NaNs and +Infinity be +maximum; -Infinity stays
// the same, but this is okay for a Compare operation.
// Clamp NaNs
// Note: This fixes a crash in Rule of Rose.
SSE_MINSS_M32_to_XMM(EEREC_S, (uptr)&g_maxvals[0]);
SSE_MINSS_M32_to_XMM(EEREC_T, (uptr)&g_maxvals[0]);
fpuFloat3(EEREC_S);
fpuFloat3(EEREC_T);
SSE_UCOMISS_XMM_to_XMM(EEREC_S, EEREC_T);
break;
default: