chacha/glide
1
0
Fork 0
Code Issues Pull Requests Actions Packages Releases Activity

Cleaned up win32 makefiles. Added work around for v5 6000 dac problem. Completed merging screen shot routines.

Browse Source
This commit is contained in:
koolsmoky
2003-06-05 08:24:00 +00:00
parent ba980646b9
commit 9a64ace60c
21 changed files with 1137 additions and 458 deletions
Download Patch File Download Diff File Expand all files Collapse all files

402
glide3x/config.mak
View File

@@ -5,172 +5,148 @@
# 1 ve3d 1.0 KoolSmoky global config
#
# moved this to win32mak.bat
# BUILD_DRIVE = $(HOMEDRIVE)
# use this for 95 DDK compatiblity
#!include $(BUILD_DRIVE)\98DDK\inc\master.mk
# moved this to win32mak.bat
# BUILD_ROOT=$(BUILD_DRIVE)\work\Glide3x
INCLUDE=$(INCLUDE)
# moved this to win32mak.bat
# BUILD_ROOT_SWLIBS=$(BUILD_ROOT)\swlibs
# moved this to win32mak.bat
# DIRECTXSDK=$(BUILD_DRIVE)\DXSDK
#
# Variables local to the build.
#
# moved this to win32mak.bat
# FX_HW_PROJECTS=glide
#
# --enable-fx-glide-hw chooses which ASIC to build for.
#
# FX_GLIDE_HW=h5
#
INSTALL_DESTINATION=$(BUILD_ROOT)\$(FX_GLIDE_HW)
#
# --enable-fx-dri-build chooses to build with or without DRI.
#
#DRI_BUILD=0
#
# This is makefile.linux behavior. I don't know exactly
# what this does.
#
# if voodoo2 use HAL_HW=true
#HAL_HW=1
#
# Define debuging
# FX_GLIDE_SYMBOLICS=1 enables us to get stuff from SoftIce etc even when
# building an optimized version. Supports minimal debug info on default. if you don't
# want any, CEASE_ALL_GDBG=1 comments out #define GDBG_INFO_ON in
# minihwc.c and win_mode.c .
#FX_GLIDE_SYMBOLICS=1
#DEBUG = 1
#CEASE_ALL_GDBG=1
#BETA = 1
!if "$(CEASE_ALL_GDBG)"=="1" && "$(DEBUG)"!="1"
LCDEFS = $(LCDEFS) -DCEASE_ALL_GDBG=1
!endif
!if "$(BETA)"=="1"
LCDEFS = $(LCDEFS) -DBETA=1
!endif
#
# More debuging
# use output to nt kernel debugger
#KERNEL_NT=1
# Yet more debuging
#FX_GLIDE_VTUNE=
#
# Define --enable-amd3d
#
#enable_amd3d=false
#GL_AMD3D=1
#
# --enable-fx-texlib=texus2 enables new texus library.
#
TEXTURE_UTILITIES_DIR=texus2
#
# These are conditional variables whose value is set by
# one of the other enable macros.
#
#FX_GLIDE_H3=0
FX_GLIDE_H5=1
#FX_GLIDE_CVG=0
#
# Architecture
# current, i386,i486,i586,i685,alpha
# default=current
FX_GLIDE_BUILD_ARCHITECTURE=i686
#
# Various tests needed at points in the build
# First, we set defaults.
#
#FX_GLIDE_PACKET_FIFO=1
#FX_GLIDE3=1
#
# There is no entry for this for H5
#FX_CHRIS_DENIS_ANTHONY_HACK=1
#
# FX_GLIDE_ALT_TAB already in makefile of h5\glide3x\src but commented
# out there. Need IS_ALT_TAB=1 to complete alt-tab in NT5.1
GLIDE_ALT_TAB=1
#FX_GLIDE_ALT_TAB=1
#IS_ALT_TAB=1
#!if "$(FX_GLIDE_ALT_TAB)"=="1" && "$(IS_ALT_TAB)"=="1"
#LCDEFS = $(LCDEFS) -DIS_ALT_TAB=1
#!endif
# Colourless
# Remove this line to disable the WinXP Alt Tab fix. There will be a speed up
# Shouldn't use this anymore. Faster generally works better, in particular for SDL
# app. Doing it this way doesn't always detect vid mode switches
#LCDEFS = $(LCDEFS) -DWINXP_ALT_TAB_FIX=1
# Colourless
# Remove this line to disable the faster WinXP Alt Tab fix. There will be a minor speed up
LCDEFS = $(LCDEFS) -DWINXP_FASTER_ALT_TAB_FIX=1
#
# use direct i/o writes
#FX_GLIDE_DIRECT_WRITE=1
#
# HAL_CSIM has to be 1 or NULL (Simulator=1, Hardware=NULL)
HAL_CSIM=
#
# 1 turns on cpu optimizations.
# HAL_CSIM must be NULL for this to work.
FX_GLIDE_CPUOPTS=1
#
#FX_GLIDE_NO_FIFO=1
#
!if "$(DEBUG)"=="1"
#FX_GLIDE_DEBUG_FIFO=1
!endif
#
#LCDEFS = $(LCDEFS) -D_FIFODUMP=1
#
# disable 3dfx watermark
FX_GLIDE_NO_PLUG=0
#
# disable 3dfx splashscreen
FX_GLIDE_NO_SPLASH=0
#
#
# Make a static link library.
FX_DLL_BUILD=1
#
# HW setup
# it's now faster to cull backfaced/zero area triangles in sw before the fifo
# forced to sw when using amd asm optimization.
#FX_GLIDE_HW_CULL=1
# Stick one more slot in the dataList so that the clamping loop does
# not go one more iteration.
#GLIDE_FP_CLAMP=1
# if we're going to stick one more 0 in the texture datalist we need to
# have GLIDE_FP_CLAMP=1 ofcourse
#GLIDE_FP_CLAMP_TEX=1
#
# use C trisetup instead of asm setup.
#FX_GLIDE_CTRISETUP=1
#
#setting this to 1 will disable hw TSU for triangle rendering.
#FX_GLIDE_SW_SETUP=1
#
#
# compiler defs
FX_COMPILER=MICROSOFT
FX_TARGET=WIN32
FX_HOST=WIN32
FX_PCI_TARGET=NT
#FX_TARGET_MINOR=WIN95
# FX_TARGET_MINOR=WIN95
INCLUDE=$(INCLUDE)
INSTALL_DESTINATION=$(BUILD_ROOT)\$(FX_GLIDE_HW)
# Make a static link library.
FX_DLL_BUILD=1
# Define beta release
# LCDEFS = $(LCDEFS) -DBETA=1
# Define debuging
# DEBUG = 1
# Enables us to get stuff from SoftIce etc even when building
# an optimized version.
# FX_GLIDE_SYMBOLICS=1
# Supports minimal debug info on default even when building an
# optimized version. Use this to disable all debug info.
# CEASE_ALL_GDBG=1
# Use output to nt kernel debugger
# KERNEL_NT=1
# Use Intel VTune
# FX_GLIDE_VTUNE=1
# HAL_CSIM has to be 1 or NULL (Simulator=1, Hardware=NULL)
HAL_CSIM=
# Use CPU optimizations.
# HAL_CSIM must be NULL for this to work.
FX_GLIDE_CPUOPTS=1
# disable 3dfx watermark
FX_GLIDE_NO_PLUG=0
# disable 3dfx splashscreen
FX_GLIDE_NO_SPLASH=0
# it's now faster to cull backfaced/zero area triangles in sw before the fifo
# forced to sw when using amd asm optimization.
# FX_GLIDE_HW_CULL=1
# Stick one more slot in the dataList so that the clamping loop does
# not go one more iteration.
# GLIDE_FP_CLAMP=1
# if we're going to stick one more 0 in the texture datalist we need to
# have GLIDE_FP_CLAMP=1 ofcourse
# GLIDE_FP_CLAMP_TEX=1
# use C trisetup instead of asm setup.
# FX_GLIDE_CTRISETUP=1
# setting this to 1 will disable hw TSU for triangle rendering.
# FX_GLIDE_SW_SETUP=1
# use direct i/o writes
# FX_GLIDE_DIRECT_WRITE=1
# FX_GLIDE_NO_FIFO=1
# FX_GLIDE_DEBUG_FIFO=1
# LCDEFS = $(LCDEFS) -D_FIFODUMP=1
# are we compiling for glide3x?
!if "$(FX_HW_PROJECTS)"=="glide3"
FX_GLIDE3 = 1
!endif
# use dxdrvr.c instead of win_mode.c for minihwc.lib
# HWC_DXDRVR = 1
# if 1, HWC_EXT_INIT is not used
# HWC_INIT_VXD = 1
# VXD stuff: need the LCDEFS options for win95 DDK compatiblity, just in case.
HWC_WIN40COMPAT = 1
# ABJ- hack
# use the minivdd to setup SLI/AA mode for hwcRestoreVideo().
# This is need till the not lazy stuff is properly done on a multichip comp.
HWC_MINIVDD_HACK = 1
# Obsolete. We now use Colourless's Alt Tab fix for winnt5.1
# LCDEFS = $(LCDEFS) -DIS_ALT_TAB=1
# Colourless
# Remove this line to disable the WinXP Alt Tab fix. There will be a speed up
# Shouldn't use this anymore. Faster generally works better, in particular for SDL
# app. Doing it this way doesn't always detect vid mode switches
# LCDEFS = $(LCDEFS) -DWINXP_ALT_TAB_FIX=1
# Colourless
# Remove this line to disable the faster WinXP Alt Tab fix. There will be a minor speed up
LCDEFS = $(LCDEFS) -DWINXP_FASTER_ALT_TAB_FIX=1
# profiling
# LLDOPTS = /PROFILE
#
# misc options
#
LCDEFS = $(LCDEFS) -DSTRICT -DWIN32_LEAN_AND_MEAN
LADEFS = $(LADEFS) -DSTRICT
LCDEFS = $(LCDEFS) -D_MSC_VER=1200 -DWINVER=0x0400
# LCDEFS = $(LCDEFS) -DGLIDE3_VERTEX_LAYOUT=1
# LCDEFS = $(LCDEFS) -DGLIDE_VERTEX_TABLE=1
# LCDEFS = $(LCDEFS) -DGLIDE3_SCALER=1
# LCDEFS = $(LCDEFS) -DGLIDE_TEST_TEXTURE_ALIGNMENT=1
# LCDEFS = $(LCDEFS) -DLFB_DISABLE_SLAVE_FIFO=1
# LCDEFS = $(LCDEFS) -DGLIDE_MULTIPLATFORM=1
# LCDEFS = $(LCDEFS) -DPCI_BUMP_N_GRIND=1 machintosh?
# LCDEFS = $(LCDEFS) -DGLIDE_USE_ALT_REGMAP=1
#
#THISDIR = devel
# Sandro's fast C clipping routine. GLIDE_USE_C_TRISETUP=1 must be set.
# LCDEFS = $(LCDEFS) -DFAST_C_CLIP=1
#
# Check for trilinearness
# LCDEFS = $(LCDEFS) -DGLIDE_CHECK_TRILINEAR=1
#
# Protected hacks for Glide/Win. Working 1-window command fifo.
# enables 1 window glide in window. removed! we'll never use it.
# LCDEFS = $(LCDEFS) -DTACO_MEMORY_FIFO_HACK=1
#
# for WINNT Global structure used to store app state, used by the display driver
# LCDEFS = $(LCDEFS) -DWINNT=1 -D_WIN32_WINNT=0x0500
#
# Called once when we start using the command fifo in order to tell
# NT / W2K to hold off on unmapping that view of the frame buffer and
# then called again once Glide knows it's lost its context so that
# the driver knows that it's now ok to unmap the memory.
# need WINNT=1 to work.
# LCDEFS = $(LCDEFS) -DENABLE_V3_W2K_GLIDE_CHANGES=1
# LCDEFS = $(LCDEFS) -DSLI_AA=1
#
# We branched glide starting w/ Voodoo2(aka CVG) so
# make sure that the right thing happens. If this sort
@@ -178,114 +154,18 @@ FX_PCI_TARGET=NT
#
!if "$(FX_GLIDE_HW)" == "sst1"
SUBDIRS = swlibs sst1
!else
!if "$(FX_GLIDE_HW)" == "sst96"
!else if "$(FX_GLIDE_HW)" == "sst96"
SUBDIRS = swlibs sst1
!else
!if "$(FX_GLIDE_HW)" == "cvg"
!else if "$(FX_GLIDE_HW)" == "cvg"
FX_GLIDE_CVG=1
SUBDIRS = swlibs cvg
!ifndef HAL_HW
HAL_HW = 1
!endif
!else
!if "$(FX_GLIDE_HW)" == "h3"
!else if "$(FX_GLIDE_HW)" == "h3"
FX_GLIDE_H3=1
SUBDIRS = swlibs h3
!else
!if "$(FX_GLIDE_HW)" == "h5"
!else if "$(FX_GLIDE_HW)" == "h5"
FX_GLIDE_H5=1
SUBDIRS = swlibs h5
!else
FX_GLIDE_HW = h3
SUBDIRS = swlibs h3
!endif
!endif
!endif
!endif
!endif
#
# Other variables we need to set
#
!if "$(FX_HW_PROJECTS)"=="glide3"
FX_GLIDE3 = 1
!endif
############################
#
# variables added
#
############################
# use dxdrvr.c instead of win_mode.c for minihwc.lib
#HWC_DXDRVR = 1
#
#
# VXD stuff: need the LCDEFS options for win95 DDK compatiblity, just in case.
HWC_WIN40COMPAT = 1
#
#
# ABJ- hack
# use the minivdd to setup SLI/AA mode for hwcRestoreVideo().
# This is need till the not lazy stuff is properly done on a multichip comp.
HWC_MINIVDD_HACK = 1
#
#
# Save the initial vidScreenSize, lfbMemoryConfig, and bpp at hwcinitvideo
# and restore them at resetVideo().
HWC_WINXP_MINIVDD_HACK = 0
!if "$(HWC_MINIVDD_HACK)"!="1"
HWC_WINXP_MINIVDD_HACK = 0
!endif
#
#
# if 1, HWC_EXT_INIT is not used
#HWC_INIT_VXD = 1
#
#
# STRICT
STRICT=1
!if "$(STRICT)"=="1"
LCDEFS = $(LCDEFS) -DSTRICT -DWIN32_LEAN_AND_MEAN
#LADEFS = $(LADEFS) -DSTRICT
!endif
#
#
# other stuff
LCDEFS = $(LCDEFS) -D_MSC_VER=1200 -DWINVER=0x0400
#
!if "$(FX_GLIDE3)"=="1"
LCDEFS = $(LCDEFS) -DGLIDE3_VERTEX_LAYOUT=1 -DGLIDE3_SCALER=1
!endif
#
#-DGLIDE_VERTEX_TABLE=1
#-DGLIDE3_SCALER=1
#-D_MSC_VER=1200
#-DWINVER=0x0501
#-DGLIDE_TEST_TEXTURE_ALIGNMENT=1
#-DGLIDE_CHECK_TRILINEAR=1 use this if we want to check for trilinearness
#-DGLIDE3_VERTEX_LAYOUT=1 this was removed.
#-DLFB_DISABLE_SLAVE_FIFO=1
#-DGLIDE_MULTIPLATFORM=1
#-DPCI_BUMP_N_GRIND=1 machintosh?
#-DGLIDE_USE_ALT_REGMAP=1
#
# Protected hacks for Glide/Win. Working 1-window command fifo.
# enables 1 window glide in window. removed! we'll never use it.
#-DTACO_MEMORY_FIFO_HACK=1
#
# for WINNT Global structure used to store app state, used by the display driver
#LCDEFS = $(LCDEFS) -DWINNT=1 -D_WIN32_WINNT=0x0500
# Called once when we start using the command fifo in order to tell
# NT / W2K to hold off on unmapping that view of the frame buffer and
# then called again once Glide knows it's lost its context so that
# the driver knows that it's now ok to unmap the memory.
# need WINNT=1 to work.
#ENABLE_V3_W2K_GLIDE_CHANGES = 1
!if "$(ENABLE_V3_W2K_GLIDE_CHANGES)"=="1"
LCDEFS = $(LCDEFS) -DENABLE_V3_W2K_GLIDE_CHANGES=1
!endif
# need WINNT=1 to work.
#SLI_AA = 1
!if "$(SLI_AA)"=="1"
LCDEFS = $(LCDEFS) -DSLI_AA=1
!endif
#
#LLDOPTS = /PROFILE

5
glide3x/g3_mak.bat
View File

@@ -37,7 +37,6 @@ rem +-include---nmake
rem |
rem +-newpci---pcilib
rem
rem ///////////////////////////////////////////////////////////////////
set BUILD_ROOT=%cd%
set BUILD_ROOT_SWLIBS=%BUILD_ROOT%\swlibs
set FX_GLIDE_HW=h5
@@ -46,11 +45,13 @@ rem
rem ///////////////////////////////////////////////////////////////////
rem Compiler Environment
rem These are usually preconfigured in the command prompt environment.
rem ///////////////////////////////////////////////////////////////////
rem
rem set DXSDKROOT=c:\dxsdk
rem set CC=%C32_ROOT%\bin\cl.exe
rem set AS=%MASM_ROOT%\bin\ml.exe
rem
rem ///////////////////////////////////////////////////////////////////
rem
if "%1" == "" goto error
goto %1
REM ----------build routine from here -------------

21
glide3x/h5/glide3/src/diglide.c
View File

@@ -350,8 +350,6 @@ GR_DIENTRY(grGlideInit, void, (void))
GDBG_INFO(80,"grGlideInit()\n");
FXUNUSED(*glideIdent);
/* dBorca - play safe */
grErrorSetCallback(_grErrorDefaultCallback);
@@ -370,23 +368,16 @@ GR_DIENTRY(grGlideInit, void, (void))
#endif
}
// _GlideInitEnvironment(); /* the main init code */
// FXUNUSED(*glideIdent);
#if GDBG_INFO_ON
GDBG_ERROR_SET_CALLBACK(_grErrorCallback);
#endif
// if (_GlideRoot.initialized) {
// initThreadStorage();
// initCriticalSection();
/* NB: We need to select the default device here so that grGetXXX
* routines work before grSstWinOpen or the surface attachment
* routines are called.
*/
grSstSelect(0);
// }
/* NB: We need to select the default device here so that grGetXXX
* routines work before grSstWinOpen or the surface attachment
* routines are called.
*/
grSstSelect(0);
FXUNUSED(*glideIdent);
_grResetTriStats();
GDBG_INFO(281,"grGlideInit --done---------------------------------------\n");

11
glide3x/h5/glide3/src/disst.c
View File

@@ -178,6 +178,7 @@ GR_DIENTRY(grSstQueryHardware, FxBool, ( GrHwConfiguration *hwc ))
GR_DIENTRY(grSstSelect, void, ( int which ))
{
static FxBool initialized[HWC_MAX_BOARDS];
FxU32 i;
/* NB: We cannot use GR_DCL_GC here because we may be setting
* a context into tls.
*/
@@ -186,15 +187,19 @@ GR_DIENTRY(grSstSelect, void, ( int which ))
if ( which >= _GlideRoot.hwConfig.num_sst )
GrErrorCallback( "grSstSelect: non-existent SST", FXTRUE );
if(!initialized[which]) _GlideRoot.initialized = FXFALSE;
_GlideInitEnvironment(which); /* the main init code */
initialized[which] = FXTRUE;
if(!initialized[which]) {
_GlideRoot.initialized = FXFALSE;
_GlideInitEnvironment(which); /* the main init code */
}
for(i = 0; i < HWC_MAX_BOARDS; i++) initialized[i] = FXFALSE;
#if GDBG_INFO_ON
GDBG_ERROR_SET_CALLBACK(_grErrorCallback);
#endif
if (_GlideRoot.initialized) {
initialized[which] = FXTRUE;
initThreadStorage();
initCriticalSection();
_GlideRoot.current_sst = which;

21
glide3x/h5/glide3/src/distate.c
View File

@@ -1176,9 +1176,11 @@ GR_DIENTRY(grRenderBuffer, void , (GrBuffer_t buffer) )
#ifdef FX_GLIDE_NAPALM
if (gc->grPixelSample > 1)
{
_grAAOffsetValue(_GlideRoot.environment.aaXOffset[gc->sampleOffsetIndex],
_GlideRoot.environment.aaYOffset[gc->sampleOffsetIndex],
0, gc->chipCount - 1, FXTRUE, gc->enableSecondaryBuffer) ;
if (!gc->state.grEnableArgs.aaMultisampleDisableCount) {
_grAAOffsetValue(_GlideRoot.environment.aaXOffset[gc->sampleOffsetIndex],
_GlideRoot.environment.aaYOffset[gc->sampleOffsetIndex],
0, gc->chipCount - 1, FXTRUE, gc->enableSecondaryBuffer) ;
}
}
#endif
@@ -2046,6 +2048,7 @@ _grValidateState()
/* Check for alpha test optimization */
if (NOTVALID(alphaMode) || NOTVALID(fbzMode) || NOTVALID(stencilMode)) {
updateAlphaMode = FXTRUE;
// KoolSmoky - need to recheck this.
/*if((LOADARG(grAlphaBlendFunction, rgb_sf) == GR_BLEND_SRC_ALPHA) &&
(LOADARG(grAlphaBlendFunction, rgb_df) == GR_BLEND_ONE_MINUS_SRC_ALPHA) &&
(LOADARG(grAlphaBlendFunction, rgb_op) == GR_BLEND_OP_ADD) &&
@@ -2787,11 +2790,15 @@ GR_DIENTRY(grDisable, void , (GrEnableMode_t mode) )
#ifdef FX_GLIDE_NAPALM
case GR_AA_MULTI_SAMPLE:
{
gc->state.grEnableArgs.aaMultisampleDisableCount++;
if (!gc->state.grEnableArgs.aaMultisampleDisableCount)
{
_grAAOffsetValue(_GlideRoot.environment.aaXOffset[0],
_GlideRoot.environment.aaYOffset[0],
0, gc->chipCount - 1, FXTRUE, gc->enableSecondaryBuffer) ;
}
gc->state.grEnableArgs.aaMultisampleDisableCount++;
_grAAOffsetValue(_GlideRoot.environment.aaXOffset[0],
_GlideRoot.environment.aaYOffset[0],
0, gc->chipCount - 1, FXTRUE, gc->enableSecondaryBuffer) ;
}
break;
#endif

84
glide3x/h5/glide3/src/fxglide.h
View File

@@ -777,6 +777,44 @@ typedef struct {
#define VRETRACEMASK 0x00000fff
#define HRETRACEPOS 16
struct tmuState_s {
#define OOOO_OOOO_OIII_IIII 0x007f
#define SR_MASK_4 OOOO_OOOO_OIII_IIII
FxU32 texPkt4Hdr_0;
FxU32 textureMode; /* 0x300 ( 0 ) */
FxU32 tLOD; /* 0x304 ( 1 ) */
FxU32 tDetail; /* 0x308 ( 2 ) */
FxU32 texBaseAddr; /* 0x30C ( 3 ) */
FxU32 texBaseAddr_1; /* 0x310 ( 4 ) */
FxU32 texBaseAddr_2; /* 0x314 ( 5 ) */
FxU32 texBaseAddr_3_8; /* 0x318 ( 6 ) */
/* ----------- end packet -------------*/
#define OOOO_OOOO_OOOO_OOII 0x0002
#define SR_MASK_5 OOOO_OOOO_OOOO_OOII
FxU32 texPkt4Hdr_1;
FxU32 texchromaKey;
FxU32 texchromaRange;
/* ----------- end packet -------------*/
#define SR_WORDS_6 24
FxU32 texPkt1Hdr_2;
FxU32 nccTable0[12];
FxU32 nccTable1[12];
/* ----------- end packet -------------*/
#ifdef FX_GLIDE_NAPALM
#define OOOO_OOOO_OOOO_OOOI 0x0001
#define SR_MASK_7 OOOO_OOOO_OOOO_OOOI
FxU32 texPkt4Hdr_3;
FxU32 combineMode;
/* ----------- end packet -------------*/
#endif /* FX_GLIDE_NAPALM */
};
/*--------------------------------------------------------------------------
State Restoration Buffer
@@ -852,44 +890,7 @@ typedef struct {
/* ----------- end packet -------------*/
#endif /* FX_GLIDE_NAPALM */
struct tmuState {
#define OOOO_OOOO_OIII_IIII 0x007f
#define SR_MASK_4 OOOO_OOOO_OIII_IIII
FxU32 texPkt4Hdr_0;
FxU32 textureMode; /* 0x300 ( 0 ) */
FxU32 tLOD; /* 0x304 ( 1 ) */
FxU32 tDetail; /* 0x308 ( 2 ) */
FxU32 texBaseAddr; /* 0x30C ( 3 ) */
FxU32 texBaseAddr_1; /* 0x310 ( 4 ) */
FxU32 texBaseAddr_2; /* 0x314 ( 5 ) */
FxU32 texBaseAddr_3_8; /* 0x318 ( 6 ) */
/* ----------- end packet -------------*/
#define OOOO_OOOO_OOOO_OOII 0x0002
#define SR_MASK_5 OOOO_OOOO_OOOO_OOII
FxU32 texPkt4Hdr_1;
FxU32 texchromaKey;
FxU32 texchromaRange;
/* ----------- end packet -------------*/
#define SR_WORDS_6 24
FxU32 texPkt1Hdr_2;
FxU32 nccTable0[12];
FxU32 nccTable1[12];
/* ----------- end packet -------------*/
#ifdef FX_GLIDE_NAPALM
#define OOOO_OOOO_OOOO_OOOI 0x0001
#define SR_MASK_7 OOOO_OOOO_OOOO_OOOI
FxU32 texPkt4Hdr_3;
FxU32 combineMode;
/* ----------- end packet -------------*/
#endif /* FX_GLIDE_NAPALM */
} tmuState[GLIDE_NUM_TMU];
struct tmuState_s tmuState[GLIDE_NUM_TMU];
struct PaletteRow {
#define SR_WORDS_P 8
@@ -939,7 +940,7 @@ typedef struct {
FxU32 tmuAlphaPassthrough; /* TMU alpha combine is in passthrough */
GrStateBuffer shadow; /* shadow of all hw state registers */
struct tmuState tmuShadow[GLIDE_NUM_TMU]; /* shadow of TMU registers */
struct tmuState_s tmuShadow[GLIDE_NUM_TMU]; /* shadow of TMU registers */
GrColor_t tmuColor[GLIDE_NUM_TMU]; /* Shadow of TMU constant color values */
FxU32 tmuMaskShadow; /* Intermediate tmuMask value before grValidateTMUState() */
@@ -2140,9 +2141,10 @@ struct _GlideRoot_s {
FxBool memFIFOHack; /* flush FIFO as much as possible */
#endif
FxU32 oglLfbLockHack; /* Enables disable hack to get around forced 32bit problems in OpenGL */
FxU32 oglLfbLockHack; /* Enables disable hack to get around forced 32bit problems in OpenGL */
FxU32 useHwcAAforLfbRead; /* Specifies whether to use HwcAAReadRegion for read Locks and LfbReadRegion calls */
FxU32 ditherHwcAA; /* Specifies whether to use HwcAAReadRegion should dither */
FxU32 ditherHwcAA; /* Specifies whether to use HwcAAReadRegion should dither */
FxI32 lockCounter;
} environment;
GrHwConfiguration hwConfig;

17
glide3x/h5/glide3/src/gdraw.c
View File

@@ -236,9 +236,18 @@ GR_ENTRY(grDrawPoint, void, (const void *p))
GR_BEGIN_NOFIFOCHECK(FN_NAME, 90);
GDBG_INFO_MORE(gc->myLevel, "(p = 0x%x)\n", p);
(gc->state.grEnableArgs.primitive_smooth_mode & GR_AA_ORDERED_POINTS_MASK
? _grAADrawPoints
: _grDrawPoints)(GR_VTX_PTR_ARRAY, 1, (void *)&p);
#if defined(__linux__) || defined(__DJGPP__)
if (gc->state.grEnableArgs.primitive_smooth_mode & GR_AA_ORDERED_POINTS_MASK)
_grAADrawPoints(GR_VTX_PTR_ARRAY, 1, (void *)&p);
else
_grDrawPoints(GR_VTX_PTR_ARRAY, 1, (void *)&p);
#else
if (gc->state.grEnableArgs.primitive_smooth_mode & GR_AA_ORDERED_POINTS_MASK)
_grAADrawPoints(GR_VTX_PTR_ARRAY, 1, &(void *)p);
else
_grDrawPoints(GR_VTX_PTR_ARRAY, 1, &(void *)p);
#endif
#undef FN_NAME
} /* grDrawPoint */
@@ -281,7 +290,7 @@ GR_ENTRY(grDrawLine, void, (const void *a, const void *b))
_grAADrawLineStrip(GR_VTX_PTR_ARRAY, GR_LINES, 2, verts);
else
_grDrawLineStrip(GR_VTX_PTR_ARRAY, GR_LINES, 2, verts);
}
}
#else
#if defined(__linux__) || defined(__DJGPP__)
if (gc->state.grEnableArgs.primitive_smooth_mode & GR_AA_ORDERED_LINES_MASK)

14
glide3x/h5/glide3/src/gglide.c
View File

@@ -2621,7 +2621,21 @@ GR_ENTRY(grBufferSwap, void, (FxU32 swapInterval))
GR_BEGIN_NOFIFOCHECK(FN_NAME,86);
GDBG_INFO_MORE(gc->myLevel,"(%d)\n",swapInterval);
// First thing first. Decrement the lockCounter
if (_GlideRoot.environment.lockCounter > -10) _GlideRoot.environment.lockCounter--;
#ifdef FX_GLIDE_NAPALM
#if !defined(__linux__) && !defined(__DJGPP__)
/* Window hacky stuff */
if (gc->windowed)
{
extern void _grFlipWindowSurface();
grFinish();
_grFlipWindowSurface();
gc->stats.bufferSwaps++;
return;
}
#endif
// 4x FSAA or greater FSAA
if (0 && gc->grPixelSample >= 4) {

181
glide3x/h5/glide3/src/glfb.c
View File

@@ -343,11 +343,13 @@ static FxBool _grLfbLock (GrLock_t type, GrBuffer_t buffer,
GrLfbWriteMode_t writeMode, GrOriginLocation_t origin,
FxBool pixelPipeline, GrLfbInfo_t *info)
{
#define FN_NAME "grLfbLock"
#define FN_NAME "_grLfbLock"
FxBool
rv = FXTRUE;
const FxBool
idleLockP = ((type & GR_LFB_NOIDLE) == 0);
FxBool
noRead = FXFALSE;
FxU32
lfbMode,
zaColor,
@@ -357,7 +359,7 @@ static FxBool _grLfbLock (GrLock_t type, GrBuffer_t buffer,
GrLfbWriteMode_t fbMode = 0;
GrLfbWriteMode_t depthMode = 0;
GR_BEGIN_NOFIFOCHECK_RET("grLfbLock", 82);
GR_BEGIN_NOFIFOCHECK_RET("_grLfbLock", 82);
GDBG_INFO_MORE(gc->myLevel,"(%d,%d,0x%x)\n", type, buffer, info);
GR_CHECK_COMPATABILITY(FN_NAME, !info,
@@ -368,6 +370,13 @@ static FxBool _grLfbLock (GrLock_t type, GrBuffer_t buffer,
type = type & ~(GR_LFB_NOIDLE);
// LFBWOEXP Extension
if (type == GR_LFB_WRITE_ONLY_EXPLICIT_EXT)
{
type = GR_LFB_WRITE_ONLY;
noRead = FXTRUE;
}
/* Pray that no one has made any glide calls that touch the hardware... */
#ifdef FX_GLIDE_NAPALM
if((gc->sliCount > 1) &&
@@ -724,7 +733,7 @@ static FxBool _grLfbLock (GrLock_t type, GrBuffer_t buffer,
we just return the current buffer lfbPtr as the write ptr.
This fixes those games that use the lfb write pointer to do
lfb reads. --mikec */
if((writeMode == (FxI32)fbMode) &&
if((writeMode == (FxI32)fbMode) && (!noRead) &&
(!pixelPipeline) &&
/* Origin must be upper left since we will return raw lfb */
(origin != GR_ORIGIN_LOWER_LEFT)){
@@ -771,7 +780,7 @@ static FxBool _grLfbLock (GrLock_t type, GrBuffer_t buffer,
* take both a read and write lock, but only save off the stride value
* from the latter one. So if we return different strides OpenGL's lfb
* accesses will be whacked. -- KCD */
if((writeMode == fbMode) &&
if((writeMode == fbMode) && (!noRead) &&
(!pixelPipeline) &&
/* Origin must be upper left since we will return raw lfb */
(origin != GR_ORIGIN_LOWER_LEFT)){
@@ -917,7 +926,7 @@ static FxBool _grLfbLock (GrLock_t type, GrBuffer_t buffer,
GR_RETURN(rv);
#undef FN_NAME
} /* grLfbLock */
} /* _grLfbLock */
/* Hack to correct locks of forced 32 bit surfaces */
static FxU16 *forced_32bpp_lock_buffer = 0;
@@ -934,6 +943,7 @@ GR_ENTRY(grLfbLock, FxBool,(GrLock_t _type, GrBuffer_t buffer,
#define FN_NAME "grLfbLock"
FxBool rv = FXTRUE;
FxBool wantHwc;
FxBool forbidden = FXFALSE;
GrLock_t type;
@@ -951,17 +961,41 @@ GR_ENTRY(grLfbLock, FxBool,(GrLock_t _type, GrBuffer_t buffer,
GDBG_INFO(79, "Read lock failure due to existing lock");
return FXFALSE;
}
/* Ok, we will disallow all Hwc reads (regardless of what the app might want) */
/* If we detect that the buffer is being locked far too much (aka a lock every frame) */
if (_GlideRoot.environment.lockCounter > 3)
{
/* Disable HwcAAforLfbRead */
if (_GlideRoot.environment.useHwcAAforLfbRead & 2)
{
_GlideRoot.environment.useHwcAAforLfbRead |= 4;
_GlideRoot.environment.useHwcAAforLfbRead &= ~2;
}
forbidden = FXTRUE;
}
/* Re-allow it if we drop back to 0 */
else if ((_GlideRoot.environment.lockCounter) == -10 && (_GlideRoot.environment.useHwcAAforLfbRead & 4))
{
/* Re-enable HwcAAforLfbRead */
_GlideRoot.environment.useHwcAAforLfbRead |= 2;
_GlideRoot.environment.useHwcAAforLfbRead &= ~4;
forbidden = FXFALSE;
}
/* We want to read using HWC if we using FSAA with 4 chips or want dithering */
/* or we are using forced 32 bit mode and want dithering */
wantHwc = (_GlideRoot.environment.useHwcAAforLfbRead & 2) &&
((gc->state.forced32BPP != 0 && _GlideRoot.environment.ditherHwcAA) ||
(gc->bInfo->h3pixelSample >= 2 && (gc->chipCount == 4 || _GlideRoot.environment.ditherHwcAA)));
/* Incrememt the lockCounter */
if (type == GR_LFB_READ_ONLY) _GlideRoot.environment.lockCounter++;
/* If we are using forced 32 bpp mode, the app is expecting 16 bit data */
/* Or we want to be using HwcAA for the Lfb read lock */
/* We need to use a hack for reading in OpenGL since they do 2 locks. Why, oh why, oh why...*/
if ((gc->state.forced32BPP || wantHwc) &&
if (!forbidden && (gc->state.forced32BPP || wantHwc) &&
(!_GlideRoot.environment.is_opengl || _GlideRoot.environment.oglLfbLockHack) &&
(buffer == GR_BUFFER_FRONTBUFFER || buffer == GR_BUFFER_BACKBUFFER))
{
@@ -1000,6 +1034,9 @@ GR_ENTRY(grLfbLock, FxBool,(GrLock_t _type, GrBuffer_t buffer,
/* Reset the useHwcAAforLfbRead setting */
_GlideRoot.environment.useHwcAAforLfbRead = old_useHwcAAforLfbRead;
/* Increment the lock counter again */
_GlideRoot.environment.lockCounter++;
/* Failed to read */
if (!rv)
@@ -1045,10 +1082,10 @@ GR_ENTRY(grLfbLock, FxBool,(GrLock_t _type, GrBuffer_t buffer,
static FxBool _grLfbUnlock (GrLock_t type, GrBuffer_t buffer)
{
#define FN_NAME "grLfbUnlock"
#define FN_NAME "_grLfbUnlock"
FxBool rval = FXFALSE;
GR_BEGIN_NOFIFOCHECK_RET("grLfbUnLock", 83);
GR_BEGIN_NOFIFOCHECK_RET("_grLfbUnLock", 83);
GDBG_INFO_MORE(gc->myLevel,"(%d, %d)\n", type, buffer);
type = type & ~(GR_LFB_NOIDLE);
@@ -1148,7 +1185,7 @@ static FxBool _grLfbUnlock (GrLock_t type, GrBuffer_t buffer)
GR_RETURN(rval);
#undef FN_NAME
} /* grLfbUnlock */
} /* _grLfbUnlock */
/* Hack to correct locks of forced 32 bit surfaces */
GR_ENTRY(grLfbUnlock, FxBool, (GrLock_t _type, GrBuffer_t buffer))
@@ -1157,7 +1194,7 @@ GR_ENTRY(grLfbUnlock, FxBool, (GrLock_t _type, GrBuffer_t buffer))
FxBool rval = FXFALSE;
GrLock_t type;
GR_BEGIN_NOFIFOCHECK_RET("_grLfbUnLock", 83);
GR_BEGIN_NOFIFOCHECK_RET("grLfbUnLock", 83);
GDBG_INFO_MORE(gc->myLevel,"(%d, %d)\n", _type, buffer);
type = _type & ~(GR_LFB_NOIDLE);
@@ -1312,7 +1349,7 @@ _grLfbWriteRegion(FxBool pixPipelineP,
info.size = sizeof(info);
if (grLfbLock(GR_LFB_WRITE_ONLY,
if (_grLfbLock(GR_LFB_WRITE_ONLY_EXPLICIT_EXT,
dst_buffer,
writeMode,
GR_ORIGIN_UPPER_LEFT,
@@ -1327,7 +1364,6 @@ _grLfbWriteRegion(FxBool pixPipelineP,
FxU32 scanline; /* scanline number */
int aligned; /* word aligned? */
srcData = (FxU32 *) src_data;
dstData = (FxU32 *) (((char*)info.lfbPtr)+ (dst_y*info.strideInBytes));
scanline = src_height;
@@ -1417,7 +1453,7 @@ _grLfbWriteRegion(FxBool pixPipelineP,
rv = FXFALSE;
break;
}
grLfbUnlock(GR_LFB_WRITE_ONLY, dst_buffer);
_grLfbUnlock(GR_LFB_WRITE_ONLY, dst_buffer);
} else {
rv = FXFALSE;
}
@@ -1514,6 +1550,7 @@ static FxBool grLfbReadRegionOrigin (GrBuffer_t src_buffer, GrOriginLocation_t o
#define FN_NAME "grLfbReadRegion"
FxU32 bpp;
FxBool rv;
FxBool wantHwc;
GrLfbInfo_t info;
GR_BEGIN_NOFIFOCHECK_RET("grLfbReadRegion",82);
@@ -1531,7 +1568,61 @@ static FxBool grLfbReadRegionOrigin (GrBuffer_t src_buffer, GrOriginLocation_t o
}
rv=FXFALSE;
if (grLfbLock(GR_LFB_READ_ONLY,
/* We want to read using HWC if we using FSAA with 4 chips or want dithering */
/* or we are using forced 32 bit mode and want dithering */
wantHwc = (_GlideRoot.environment.useHwcAAforLfbRead & 1) &&
((gc->state.forced32BPP != 0 && _GlideRoot.environment.ditherHwcAA) ||
(gc->bInfo->h3pixelSample >= 2 && (gc->chipCount == 4 || _GlideRoot.environment.ditherHwcAA)));
/* We want to use the 'advanced' and slow capture method */
if (wantHwc)
{
FxU32 colBufferIndex = 0;
FxU32 bpp = 0;
if (gc->state.forced32BPP) bpp = gc->state.forced32BPP;
else switch(gc->grPixelFormat)
{
case GR_PIXFMT_ARGB_1555:
case GR_PIXFMT_AA_2_ARGB_1555:
case GR_PIXFMT_AA_4_ARGB_1555:
case GR_PIXFMT_AA_8_ARGB_1555: /* 8xaa */
bpp = 15;
break;
case GR_PIXFMT_ARGB_8888:
case GR_PIXFMT_AA_2_ARGB_8888:
case GR_PIXFMT_AA_4_ARGB_8888:
case GR_PIXFMT_AA_8_ARGB_8888: /* 8xaa */
bpp = 32;
break;
case GR_PIXFMT_RGB_565:
case GR_PIXFMT_AA_2_RGB_565:
case GR_PIXFMT_AA_4_RGB_565:
case GR_PIXFMT_AA_8_RGB_565: /* 8xaa */
default:
bpp = 16;
break;
}
switch(src_buffer)
{
case GR_BUFFER_FRONTBUFFER:
colBufferIndex = gc->frontBuffer;
break;
case GR_BUFFER_BACKBUFFER:
colBufferIndex = gc->backBuffer;
break;
}
hwcAAReadRegion(gc->bInfo, colBufferIndex, src_x, src_y,
src_width, src_height, dst_stride, dst_data,
bpp, _GlideRoot.environment.ditherHwcAA);
rv=FXTRUE;
goto done;
}
if (_grLfbLock(GR_LFB_READ_ONLY,
src_buffer,
GR_LFBWRITEMODE_ANY,
GR_ORIGIN_UPPER_LEFT,
@@ -1554,9 +1645,9 @@ static FxBool grLfbReadRegionOrigin (GrBuffer_t src_buffer, GrOriginLocation_t o
dst_adjust=dst_stride - tmp;
/* should be endian and pixel size safe */
/* it would be nice to test if quad blocks were faster */
/* like mmx loads and stores */
while(src_height--)
/* it would be nice to test if quad blocks were faster */
/* like mmx loads and stores */
if (!gc->state.forced32BPP) while(src_height--)
{
/* adjust starting alignment */
if (((FxU32)src)&3)
@@ -1582,9 +1673,57 @@ static FxBool grLfbReadRegionOrigin (GrBuffer_t src_buffer, GrOriginLocation_t o
((FxU8 *)src)+=src_adjust;
((FxU8 *)dst)+=dst_adjust;
}
/* Nice I've got to convert it from 32 bit to 16 bit */
else if (gc->state.forced32BPP == 16) while(src_height--) {
/* read in dwords of pixels */
if(length)
{
FxU32 byte_index=0;
FxU32 byte_index2=0;
/* copies aligned dwords */
do
{
FxU32 s =*((FxU32 *)(((FxU32)src) + byte_index));
FxU16 d = (FxU16) (s & 0xF8) >> 3;
d |= (s & 0xFC00) >> 5;
d |= (s & 0xF80000) >> 8;
*((FxU16 *)(((FxU32)dst) + (byte_index2))) = d;
byte_index +=4;
}while((byte_index2+=2)<(src_width*2));
}
/* adjust for next line */
((FxU8 *)src)+=info.strideInBytes;
((FxU8 *)dst)+=dst_stride;
}
else if (gc->state.forced32BPP == 15) while(src_height--) {
/* read in dwords of pixels */
if(length)
{
FxU32 byte_index=0;
FxU32 byte_index2=0;
/* copies aligned dwords */
do
{
FxU32 s =*((FxU32 *)(((FxU32)src) + byte_index));
FxU16 d = (FxU16) (s & 0xF8) >> 3;
d |= (s & 0xF800) >> 6;
d |= (s & 0xF80000) >> 9;
*((FxU16 *)(((FxU32)dst) + (byte_index2))) = d;
byte_index +=4;
}while((byte_index2+=2)<(src_width*2));
}
/* adjust for next line */
((FxU8 *)src)+=info.strideInBytes;
((FxU8 *)dst)+=dst_stride;
}
rv=FXTRUE;
/* unlock buffer */
grLfbUnlock(GR_LFB_READ_ONLY,src_buffer);
/* unlock buffer */
_grLfbUnlock(GR_LFB_READ_ONLY,src_buffer);
}
done:
GR_RETURN(rv);
@@ -1617,7 +1756,7 @@ GR_ENTRY(grLfbReadRegion, FxBool, (GrBuffer_t src_buffer,
src_width, src_height, dst_stride, dst_data);
info.size = sizeof(info);
if (grLfbLock(GR_LFB_READ_ONLY,
if (_grLfbLock(GR_LFB_READ_ONLY,
src_buffer,
GR_LFBWRITEMODE_ANY,
GR_ORIGIN_UPPER_LEFT,
@@ -1729,7 +1868,7 @@ GR_ENTRY(grLfbReadRegion, FxBool, (GrBuffer_t src_buffer,
}
}
#endif
grLfbUnlock(GR_LFB_READ_ONLY, src_buffer);
_grLfbUnlock(GR_LFB_READ_ONLY, src_buffer);
} else {
rv = FXFALSE;
}

2
glide3x/h5/glide3/src/glide.h
View File

@@ -266,6 +266,8 @@ typedef FxU32 GrLock_t;
#define GR_LFB_IDLE 0x00
#define GR_LFB_NOIDLE 0x10
#define GR_LFB_WRITE_ONLY_EXPLICIT_EXT 0x02 // explicitly not allow reading from the lfb pointer
typedef FxI32 GrLfbBypassMode_t;
#define GR_LFBBYPASS_DISABLE 0x0
#define GR_LFBBYPASS_ENABLE 0x1

13
glide3x/h5/glide3/src/glide.rc
View File

@@ -42,8 +42,8 @@
#endif
VS_VERSION_INFO VERSIONINFO
FILEVERSION MANVERSION, MANREVISION, MINVERSION, BUILD_NUMBER
PRODUCTVERSION MANVERSION, MANREVISION, MINVERSION, BUILD_NUMBER
FILEVERSION MANVERSION, MANREVISION, 0, BUILD_NUMBER
PRODUCTVERSION MANVERSION, MANREVISION, 0, BUILD_NUMBER
FILEFLAGSMASK 0x0030003FL
FILEFLAGS (VER_PRIVATEBUILD|VER_PRERELEASE|VER_DEBUG)
@@ -56,13 +56,22 @@ BEGIN
BLOCK "040904E4"
BEGIN
VALUE "CompanyName", "3dfx Interactive, Inc.\0"
#ifdef BETA
VALUE "FileDescription", "*BETA* do NOT distribute without permission.\0"
VALUE "FileVersion", VERSIONSTR " BETA"
#else
VALUE "FileDescription", "3dfx Interactive, Inc. Glide\256 DLL\0"
VALUE "FileVersion", VERSIONSTR
#endif
VALUE "InternalName", VERSIONNAME
VALUE "LegalCopyright", "Copyright \251 3dfx Interactive, Inc. 2003\0"
VALUE "OriginalFilename", VERSIONNAME
VALUE "ProductName", PRODNAME
#ifdef BETA
VALUE "ProductVersion", VERSIONSTR " BETA"
#else
VALUE "ProductVersion", VERSIONSTR
#endif
VALUE "Graphics Subsystem", HWSTR
VALUE "Contact", CONTACTSTR
END

16
glide3x/h5/glide3/src/gpci.c
View File

@@ -1095,8 +1095,8 @@ _grSstDetectResources(void)
hwcInfo
*hInfo; /* Info about all the relavent boards */
int tmu;
char* envChipNum;
FxU32 chipCount = 1;
//char* envChipNum; /* unused */
//FxU32 chipCount = 1; /* unused */
if ((hInfo = hwcInit(0x121a, 0x9)) == NULL)
if ((hInfo = hwcInit(0x121a, 0x5)) == NULL) /* Voodoo3 */
@@ -1116,10 +1116,10 @@ _grSstDetectResources(void)
** KoolSmoky - chipCount should be after hwcInit(0x....) because
** it calls GETENV.
*/
envChipNum = GETENV("FX_GLIDE_NUM_CHIPS", GC.bInfo->RegPath);
/*envChipNum = GETENV("FX_GLIDE_NUM_CHIPS", GC.bInfo->RegPath);
if (envChipNum)
chipCount = atoi(envChipNum);
chipCount = atoi(envChipNum);*/ /* unused */
if (bInfo->pciInfo.deviceID == SST_DEVICE_ID_H3)
SST.type = GR_SSTTYPE_Banshee;
@@ -1876,15 +1876,15 @@ _GlideInitEnvironment(int which)
GDBG_INFO(80," snapshot: %d\n",_GlideRoot.environment.snapshot);
_GlideRoot.environment.guardbandclipping = GLIDE_GETENV("FX_GLIDE_GBC", GC.bInfo->RegPath, 1L);
GDBG_INFO(80," guardbandclipping: %d\n",_GlideRoot.environment.guardbandclipping);
/* KoolSmoky - enable 2ppc only in certain condition. -1=disable 1=enable 0=glide desides. enabled for now */
_GlideRoot.environment.do2ppc = GLIDE_GETENV("FX_GLIDE_2PPC", GC.bInfo->RegPath, 1L);
/* KoolSmoky - enable 2ppc only in certain condition. -1=disable 1=enable 0=glide desides. disabled for now */
_GlideRoot.environment.do2ppc = GLIDE_GETENV("FX_GLIDE_2PPC", GC.bInfo->RegPath, 0L);
GDBG_INFO(80," do2ppc : %d\n",_GlideRoot.environment.do2ppc);
_GlideRoot.environment.band2ppc = GLIDE_GETENV("FX_GLIDE_2PPC_BAND", GC.bInfo->RegPath, 2L);
GDBG_INFO(80," band2ppc : %d\n",_GlideRoot.environment.band2ppc);
_GlideRoot.environment.sliBandHeight = GLIDE_GETENV("FX_GLIDE_SLI_BAND_HEIGHT", GC.bInfo->RegPath, 0L);
GDBG_INFO(80," sliBandHeight : %d\n",_GlideRoot.environment.sliBandHeight);
_GlideRoot.environment.aaClip = GLIDE_GETENV("FX_GLIDE_AA_CLIP", GC.bInfo->RegPath, 1L);
GDBG_INFO(80," aaClip : %d\n",_GlideRoot.environment.aaClip);
_GlideRoot.environment.aaClip = GLIDE_GETENV("FX_GLIDE_AA_CLIP", GC.bInfo->RegPath, 1L);
GDBG_INFO(80," aaClip : %d\n",_GlideRoot.environment.aaClip);
/* KoolSmoky - there is a possibility that grEnable(GR_OPENGL_MODE_EXT) is called once
* but grSstSelect is called multiple times, so we won't retreive the sliBandHeightForce

32
glide3x/h5/glide3/src/gsst.c
View File

@@ -1572,6 +1572,18 @@ GR_EXT_ENTRY(grSstWinOpenExt, GrContext_t, ( FxU32 hWnd,
GDBG_INFO(80, "Setting hwnd to foreground.\n");
SetForegroundWindow((HWND)hWnd);
*/
/* Want Windowed Mode */
if (resolution == GR_RESOLUTION_NONE)
{
extern GrContext_t _grCreateWindowSurface(FxU32 hWnd,
GrColorFormat_t format,
GrOriginLocation_t origin,
GrPixelFormat_t pixelformat,
int nAuxBuffer);
return _grCreateWindowSurface(hWnd, format, origin, pixelformat, nAuxBuffers);
}
#endif /* defined(__linux__) || defined(__DJGPP__) */
/* NB: TLS must be setup before the 'declaration' which grabs the
@@ -2011,17 +2023,14 @@ GR_EXT_ENTRY(grSstWinOpenExt, GrContext_t, ( FxU32 hWnd,
{
switch (gc->grSstRez)
{
case GR_RESOLUTION_1600x1024:
gc->do2ppc = FXTRUE;
break ;
case GR_RESOLUTION_1600x1200:
break;
case GR_RESOLUTION_1792x1344:
case GR_RESOLUTION_1856x1392:
case GR_RESOLUTION_1920x1440:
case GR_RESOLUTION_2048x1536:
case GR_RESOLUTION_2048x2048:
gc->bInfo->h3analogSli = 1 ;
gc->do2ppc = FXTRUE;
break;
case GR_RESOLUTION_400x300:
case GR_RESOLUTION_320x200:
@@ -2043,8 +2052,6 @@ GR_EXT_ENTRY(grSstWinOpenExt, GrContext_t, ( FxU32 hWnd,
switch (gc->grSstRez)
{
case GR_RESOLUTION_1600x1024:
gc->bInfo->h3analogSli = 1 ;
break ;
case GR_RESOLUTION_1600x1200:
case GR_RESOLUTION_1792x1344:
case GR_RESOLUTION_1856x1392:
@@ -2064,6 +2071,7 @@ GR_EXT_ENTRY(grSstWinOpenExt, GrContext_t, ( FxU32 hWnd,
gc->chipCount = 1 ;
gc->grPixelSample = 1 ;
default:
gc->do2ppc = FXTRUE;
break;
}
}
@@ -3075,10 +3083,21 @@ GR_ENTRY(grSstWinClose, FxBool, (GrContext_t context))
if (!gc)
return 0;
#if !defined(__linux__) && !defined(__DJGPP__)
/* We are in Windowed Mode */
if (gc->windowed)
{
extern void _grReleaseWindowSurface(GrContext_t ctx);
_grReleaseWindowSurface(context);
return FXTRUE;
}
#endif
/* If we are OpenGL, we need to release Exclusive mode so other
** OpenGL fullscreen apps can run. If not, we will cause a lot
** of problems.
*/
#if 0
if (_GlideRoot.environment.is_opengl == FXTRUE) {
/* KoolSmoky- don't release Exclusive mode if we're running
* in NT5.1. This may cause probems. but, ohwell.
@@ -3086,6 +3105,7 @@ GR_ENTRY(grSstWinClose, FxBool, (GrContext_t context))
if( !gc->bInfo->osNT51 )
hwcRestoreVideo(gc->bInfo);
}
#endif
#ifndef __linux__
if (gc->lostContext) {

73
glide3x/h5/glide3/src/gtexdl.c
View File

@@ -732,10 +732,57 @@ _grTexDownloadMipMapLevelPartialTiled(GrChipID_t tmu,
const FxU32
texelSize = _grBitsPerTexel[format],
texStrideBytes = memInfo->texStrideBytes;
FxU32 maxS;
FxU32 minS = 0, maxS = 0;
FxU32 texOffset = 0x00UL;
if (thisLod < largeLod) {
switch(format) {
case GR_TEXFMT_ARGB_CMP_FXT1:
minS = 0;
if (!maxS) maxS = WIDTH_BY_ASPECT_LOD_FXT1(aspectRatio, thisLod);
else if (maxS < 8) maxS = 8;
break;
case GR_TEXFMT_ARGB_CMP_DXT1:
minS = 0;
if (!maxS) maxS = WIDTH_BY_ASPECT_LOD_DXT(aspectRatio, thisLod);
else if (maxS < 8) maxS = 4;
break;
case GR_TEXFMT_ARGB_CMP_DXT2:
case GR_TEXFMT_ARGB_CMP_DXT3:
case GR_TEXFMT_ARGB_CMP_DXT4:
case GR_TEXFMT_ARGB_CMP_DXT5:
minS = 0;
if (!maxS) maxS = WIDTH_BY_ASPECT_LOD_DXT(aspectRatio, thisLod);
else if (maxS < 8) maxS = 4;
break;
default:
if (!maxS) maxS = WIDTH_BY_ASPECT_LOD(aspectRatio, thisLod);
// 32bit Align minS
if (texelSize == 8) minS &= 4;
else if (texelSize == 16) minS &= 2;
maxS -= minS;
// 32bit Align maxS, and add minS to offset
if (texelSize == 8)
{
if (maxS>4) maxS = (maxS+3)&4;
texOffset+=minS;
}
else if (texelSize == 16)
{
if (maxS>2) maxS = (maxS+1)&2;
texOffset+=minS*2;
}
else
{
texOffset+=minS*4;
}
break;
}
data = texOffset+(char*)data;
if (thisLod < largeLod) {
texOffset = _grTexCalcMipmapLevelOffsetTiled(tmu,
thisLod, largeLod,
aspectRatio,
@@ -748,22 +795,6 @@ _grTexDownloadMipMapLevelPartialTiled(GrChipID_t tmu,
GR_CHECK_F(FN_NAME, texelSize == 0, "invalid texture format");
switch(format) {
case GR_TEXFMT_ARGB_CMP_FXT1:
maxS = WIDTH_BY_ASPECT_LOD_FXT1(aspectRatio, thisLod);
break;
case GR_TEXFMT_ARGB_CMP_DXT1:
case GR_TEXFMT_ARGB_CMP_DXT2:
case GR_TEXFMT_ARGB_CMP_DXT3:
case GR_TEXFMT_ARGB_CMP_DXT4:
case GR_TEXFMT_ARGB_CMP_DXT5:
maxS = WIDTH_BY_ASPECT_LOD_DXT(aspectRatio, thisLod);
break;
default:
maxS = WIDTH_BY_ASPECT_LOD(aspectRatio, thisLod);
break;
}
switch(texelSize) {
case 4: /* 4-bit textures */
{
@@ -880,7 +911,7 @@ _grTexDownloadMipMapLevelPartialTiled(GrChipID_t tmu,
case 16: /* 16-bit textures */
{
const FxU16
*src16 = (const FxU16*)data;
*src16 = minS + (const FxU16*)data;
switch(maxS) {
case 1:
@@ -1158,7 +1189,9 @@ GR_ENTRY(grTexDownloadMipMapLevelPartial,
* restriction.
* same for DXT1 ((4x4x1/2)x2) minimum level size is 16 bytes
*/
if((format < GR_TEXFMT_ARGB_CMP_DXT1) && (format != GR_TEXFMT_ARGB_CMP_FXT1)) {
if(format != GR_TEXFMT_ARGB_CMP_FXT1 && format != GR_TEXFMT_ARGB_CMP_DXT1 &&
format != GR_TEXFMT_ARGB_CMP_DXT2 && format != GR_TEXFMT_ARGB_CMP_DXT3 &&
format != GR_TEXFMT_ARGB_CMP_DXT4 && format != GR_TEXFMT_ARGB_CMP_DXT5 ) {
const FxU32
aspectIndex = ((aspectRatio < GR_ASPECT_LOG2_1x1)
? -aspectRatio

13
glide3x/h5/glide3/src/gthread.c
View File

@@ -78,16 +78,11 @@ FxU32 getThreadValueSLOW( void ) {
#elif 1
__GR_GET_TLSC_VALUE();
#else
/* __asm {
__asm mov esi, DWORD PTR fs:[WNT_TEB_PTR]
__asm add esi, DWORD PTR _GlideRoot.tlsOffset \
__asm mov eax, DWORD PTR [esi] \
}
*/
__asm {
__asm mov esi, DWORD PTR fs:[WNT_TEB_PTR];
__asm add esi, DWORD PTR _GlideRoot.tlsOffset;
__asm mov eax, DWORD PTR [esi];
mov esi, DWORD PTR fs:[WNT_TEB_PTR]
add esi, DWORD PTR _GlideRoot.tlsOffset
mov eax, DWORD PTR [esi]
}
#endif

4
glide3x/h5/glide3/src/makefile
View File

@@ -347,8 +347,10 @@ CFILES = $(CFILES) gsplash.c g3df.c gu.c gthread.c \
gpci.c diglide.c disst.c ditex.c gsfc.c\
gbanner.c gerror.c gaa.c gdraw.c gglide.c $(GLIDE3FILES) \
glfb.c gsst.c gtex.c gtexdl.c $(HWSPEC) cpuid.c
CPPFILES = $(CPPFILES) winsurf.cpp
OFILES = $(CFILES:.c=.obj)
OFILES = $(CFILES:.c=.obj) $(CPPFILES:.cpp=.obj)
OTHER_CFILES = fxgasm.c fxbldno.c fxinline.h

8
glide3x/h5/glide3/src/rcver.h
View File

@@ -23,14 +23,14 @@
#define MANVERSION 3
#define MANREVISION 10
#define MINVERSION 01
#define MINVERSION 00
#ifdef __WIN32__
#define BUILD_NUMBER 0001
#define BUILD_NUMBER 3001 /* hmm, must be at least 674 for the 3dfx OpenGL ICD to work correctly */
#endif
#define VERSIONSTR "3.10.01.0001"
#define CONTACTSTR "The Glide Open Source Project http://glide.sourceforge.net/\0"
#define VERSIONSTR "3.10.00.3001"
#define CONTACTSTR "The Glide Open Source Project http://glide.sourceforge.net/"
#if defined(__WIN32__)
#define _TRADEMARK_ "(tm)"

450
glide3x/h5/glide3/src/winsurf.cpp Normal file
View File

@@ -0,0 +1,450 @@
//
// This sets up Windowed rendering
//
#include <math.h>
extern "C" {
// All this is C Linked
#include <3dfx.h>
#define FX_DLL_DEFINITION
#include <fxdll.h>
#include <glide.h>
#include "fxglide.h"
#include "fxcmd.h"
#include "gsfc.h"
};
/* Begin hacky stuff to enable windowed Glide rendering in Windows */
#include <ddraw.h>
static LPDIRECTDRAW lpDDraw1 = 0;
static LPDIRECTDRAW2 lpDDraw2 = 0;
static LPDIRECTDRAWSURFACE lpPrimSurf = 0;
static LPDIRECTDRAWSURFACE lpColSurf = 0;
static LPDIRECTDRAWSURFACE lpAuxSurf = 0;
static LPDIRECTDRAWSURFACE lpTexSurf0 = 0;
static LPDIRECTDRAWSURFACE lpTexSurf1 = 0;
static HWND thehWnd = 0;
static DWORD surfWidth = 0;
static DWORD surfHeight = 0;
static DWORD screenWidth = 0;
static DWORD screenHeight = 0;
static DWORD bpp = 0;
static BOOL useDDOverlay = FALSE;
static INT WindowPosX = 0;
static INT WindowPosY = 0;
extern "C" void doSplash( void );
extern "C" void _grReleaseWindowSurface(GrContext_t ctx);
#define OVERLAY_COLOUR 0x0D0E0F
/* This will create our DirectDraw surfaces and create the context */
extern "C" GrContext_t _grCreateWindowSurface(FxU32 hWnd,
GrColorFormat_t format,
GrOriginLocation_t origin,
GrPixelFormat_t pixelformat,
int nAuxBuffer)
{
/* Disable this for now */
// return 0;
/* Allocate a context */
GrContext_t ctx = grSurfaceCreateContext(GR_SURFACECONTEXT_WINDOWED);
GrGC *gc = (GrGC *)ctx;
thehWnd = (HWND) hWnd;
screenWidth = GetSystemMetrics(SM_CXSCREEN);
screenHeight = GetSystemMetrics(SM_CYSCREEN);
RECT clientRect;
GetClientRect(thehWnd, &clientRect);
surfWidth = clientRect.right - clientRect.left;
surfHeight = clientRect.bottom - clientRect.top;
switch(pixelformat)
{
case GR_PIXFMT_ARGB_8888:
case GR_PIXFMT_AA_2_ARGB_8888:
case GR_PIXFMT_AA_4_ARGB_8888:
case GR_PIXFMT_AA_8_ARGB_8888:
bpp = 32;
break;
default:
if (0 && _GlideRoot.environment.outputBpp == 32)
{
gc->state.forced32BPP = 16;
bpp = 32;
}
else
bpp = 16;
break;
}
bpp = 16;
/* Now, create DirectDraw */
if ( DirectDrawCreate( NULL, &lpDDraw1, NULL ) != DD_OK)
{
GDBG_INFO(80, "DDraw Obj Create Failed!\n");
_grReleaseWindowSurface(ctx);
return 0;
}
GDBG_INFO(80, "DDraw Obj created!\n");
/* Get us a DirectDraw2 interface */
if ( lpDDraw1->QueryInterface (IID_IDirectDraw2, (LPVOID*)&lpDDraw2 ) != DD_OK )
{
GDBG_INFO(80, "DDraw2 Obj Create Failed!\n");
_grReleaseWindowSurface(ctx);
return 0;
}
GDBG_INFO(80, "DDraw2 Obj created!\n");
/* Set the Coop level to normal */
lpDDraw2->SetCooperativeLevel((HWND) hWnd, DDSCL_NORMAL);
/* Now create us a Primary surface */
DDSURFACEDESC ddsd;
memset(&ddsd, 0, sizeof(ddsd));
ddsd.dwSize = sizeof(ddsd);
ddsd.dwFlags = DDSD_CAPS;
ddsd.ddsCaps.dwCaps = DDSCAPS_PRIMARYSURFACE|DDSCAPS_VIDEOMEMORY;
if (lpDDraw2->CreateSurface(&ddsd, &lpPrimSurf, NULL) != DD_OK)
{
GDBG_INFO(80, "Primary Surface Create Failed!\n");
_grReleaseWindowSurface(ctx);
return 0;
}
/* Now create us a Colour surface */
if (useDDOverlay) do
{
memset(&ddsd, 0, sizeof(ddsd));
ddsd.dwSize = sizeof(ddsd);
ddsd.dwFlags = DDSD_CAPS|DDSD_HEIGHT|DDSD_WIDTH|DDSD_PIXELFORMAT;
ddsd.ddsCaps.dwCaps = DDSCAPS_OVERLAY|DDSCAPS_VIDEOMEMORY;
ddsd.dwWidth = surfWidth;
ddsd.dwHeight = screenHeight;
ddsd.ddpfPixelFormat.dwSize = sizeof(DDPIXELFORMAT);
ddsd.ddpfPixelFormat.dwFlags = DDPF_RGB;
ddsd.ddpfPixelFormat.dwRGBBitCount = bpp;
ddsd.ddpfPixelFormat.dwRGBBitCount = 16;
ddsd.ddpfPixelFormat.dwRBitMask = 0xF800;
ddsd.ddpfPixelFormat.dwGBitMask = 0x07E0;
ddsd.ddpfPixelFormat.dwBBitMask = 0x001F;
if (lpDDraw2->CreateSurface(&ddsd, &lpColSurf, NULL) != DD_OK)
{
GDBG_INFO(80, "Overlay Colour Surface Create Failed!\n");
lpColSurf = 0;
break;
}
RECT rectOverlay = {0,0,surfWidth,surfHeight};
RECT rectClient;
DDOVERLAYFX ddofx;
POINT p = {0, 0};
GetClientRect(thehWnd, &rectClient);
ClientToScreen(thehWnd, &p);
rectClient.bottom += p.y;
rectClient.top = p.y;
rectClient.left = p.x;
rectClient.right += p.x;
WindowPosX = p.x;
WindowPosY = p.y;
memset(&ddofx, 0, sizeof(ddofx));
ddofx.dwSize = sizeof(ddofx);
ddofx.dckDestColorkey.dwColorSpaceLowValue = OVERLAY_COLOUR;
ddofx.dckDestColorkey.dwColorSpaceHighValue = OVERLAY_COLOUR;
HRESULT ddrval = lpColSurf->UpdateOverlay(&rectOverlay, lpPrimSurf,
&rectClient,
DDOVER_SHOW|DDOVER_DDFX | DDOVER_KEYDESTOVERRIDE,
&ddofx);
if(ddrval != DD_OK)
{
GDBG_INFO(80, "Unable to update overlay!\n");
lpColSurf->Release();
lpColSurf = 0;
break;
}
/* Now Fill Primary with the col */
DDBLTFX ddbfx;
memset(&ddbfx, 0, sizeof(ddbfx));
ddbfx.dwSize = sizeof(DDBLTFX);
ddbfx.dwFillColor = OVERLAY_COLOUR;
lpPrimSurf->Blt(&rectClient, NULL, &rectClient, DDBLT_COLORFILL|DDBLT_WAIT, &ddbfx);
//lpColSurf->Release();
//lpColSurf = 0;
}
while(0);
if (!lpColSurf)
{
useDDOverlay = FALSE;
memset(&ddsd, 0, sizeof(ddsd));
ddsd.dwSize = sizeof(ddsd);
ddsd.dwFlags = DDSD_CAPS|DDSD_HEIGHT|DDSD_WIDTH|DDSD_PIXELFORMAT;
ddsd.ddsCaps.dwCaps = DDSCAPS_OFFSCREENPLAIN|DDSCAPS_VIDEOMEMORY|DDSCAPS_LOCALVIDMEM;
ddsd.dwWidth = surfWidth;
ddsd.dwHeight = screenHeight;
ddsd.ddpfPixelFormat.dwSize = sizeof(DDPIXELFORMAT);
ddsd.ddpfPixelFormat.dwFlags = DDPF_RGB;
ddsd.ddpfPixelFormat.dwRGBBitCount = bpp;
if (lpDDraw2->CreateSurface(&ddsd, &lpColSurf, NULL) != DD_OK)
{
GDBG_INFO(80, "Colour Surface Create Failed!\n");
_grReleaseWindowSurface(ctx);
return 0;
}
}
/* Now create us a Aux surface */
if (nAuxBuffer)
{
memset(&ddsd, 0, sizeof(ddsd));
ddsd.dwSize = sizeof(ddsd);
ddsd.dwFlags = DDSD_CAPS|DDSD_HEIGHT|DDSD_WIDTH|DDSD_PIXELFORMAT;
ddsd.ddsCaps.dwCaps = DDSCAPS_OFFSCREENPLAIN|DDSCAPS_VIDEOMEMORY|DDSCAPS_LOCALVIDMEM;
ddsd.dwWidth = surfWidth;
ddsd.dwHeight = screenHeight;
ddsd.ddpfPixelFormat.dwSize = sizeof(DDPIXELFORMAT);
ddsd.ddpfPixelFormat.dwFlags = DDPF_RGB;
ddsd.ddpfPixelFormat.dwRGBBitCount = bpp;
if (lpDDraw2->CreateSurface(&ddsd, &lpAuxSurf, NULL) != DD_OK)
{
GDBG_INFO(80, "Aux Surface Create Failed!\n");
_grReleaseWindowSurface(ctx);
return 0;
}
}
/* Now create us a Tex surface */
memset(&ddsd, 0, sizeof(ddsd));
ddsd.dwSize = sizeof(ddsd);
ddsd.dwFlags = DDSD_CAPS|DDSD_HEIGHT|DDSD_WIDTH|DDSD_PIXELFORMAT;
ddsd.ddsCaps.dwCaps = DDSCAPS_OFFSCREENPLAIN|DDSCAPS_VIDEOMEMORY|DDSCAPS_LOCALVIDMEM;
ddsd.dwWidth = 2048;
ddsd.dwHeight = 2048;
ddsd.ddpfPixelFormat.dwSize = sizeof(DDPIXELFORMAT);
ddsd.ddpfPixelFormat.dwFlags = DDPF_RGB;
ddsd.ddpfPixelFormat.dwRGBBitCount = 16;
if (lpDDraw2->CreateSurface(&ddsd, &lpTexSurf0, NULL) != DD_OK)
{
GDBG_INFO(80, "Tex Surface Create Failed!\n");
_grReleaseWindowSurface(ctx);
return 0;
}
if (gc->num_tmu == 2)
{
if (lpDDraw2->CreateSurface(&ddsd, &lpTexSurf1, NULL) != DD_OK)
{
GDBG_INFO(80, "Tex Surface Create Failed!\n");
_grReleaseWindowSurface(ctx);
return 0;
}
}
/* Select the Context */
grSelectContext(ctx);
/* Set the Render Buffer */
grSurfaceSetRenderingSurface((GrSurface_t*) lpColSurf, FXFALSE);
/* Set the Aux Buffer */
if (nAuxBuffer) grSurfaceSetAuxSurface((GrSurface_t*) lpAuxSurf);
/* Now create the texture buffer0 and 1 */
grSurfaceSetTextureSurface(GR_TMU0, (GrSurface_t*) lpTexSurf0);
if (gc->num_tmu == 2) grSurfaceSetTextureSurface(GR_TMU1, (GrSurface_t*) lpTexSurf1);
/* Set some 'crap' */
gc->grHwnd = hWnd;
gc->grColBuf = gc->state.num_buffers = 2;
gc->grAuxBuf = nAuxBuffer;
gc->grPixelFormat = (int) pixelformat;
gc->chipmask = SST_CHIP_MASK_ALL_CHIPS;
gc->state.color_format = format;
gc->sliCount = 1;
grSstOrigin(origin);
/* Clear the buffers */
grBufferClear( 0x0, 0, 0 );
return ctx;
}
/* This will blit/flip our surfaces */
extern "C" void _grFlipWindowSurface()
{
// Blit from col to prim
if (lpPrimSurf && lpColSurf)
{
POINT ptClient;
ptClient.x = 0;
ptClient.y = 0;
ClientToScreen(thehWnd, &ptClient);
if (!useDDOverlay)
{
RECT region;
region.bottom = min(surfHeight,screenHeight-ptClient.y);
region.right = min(surfWidth,screenWidth-ptClient.x);
region.top = 0;
region.left = 0;
if (ptClient.y < 0)
{
region.top = -ptClient.y;
ptClient.y = 0;
}
if (ptClient.x < 0)
{
region.left = -ptClient.x;
ptClient.x = 0;
}
if (region.bottom != region.top && region.left != region.right)
lpPrimSurf->BltFast(ptClient.x, ptClient.y, lpColSurf, &region, 0);
}
else
{
/* Need to update the overlay */
if (ptClient.x != WindowPosX || ptClient.y != WindowPosY)
{
WindowPosX = ptClient.x;
WindowPosY = ptClient.y;
RECT region;
RECT rectClient;
region.bottom = min(surfHeight,screenHeight-ptClient.y);
region.right = min(surfWidth,screenWidth-ptClient.x);
region.top = 0;
region.left = 0;
if (ptClient.y < 0)
{
region.top = -ptClient.y;
rectClient.top = 0;
}
else
{
rectClient.top = ptClient.y;
}
if (ptClient.x < 0)
{
region.left = -ptClient.x;
rectClient.left = 0;
}
else
{
rectClient.left = ptClient.x;
}
// Now we need to work out dest width and height
rectClient.right = (region.right-region.left)+rectClient.left;
rectClient.bottom = (region.bottom-region.top)+rectClient.top;
DDOVERLAYFX ddofx;
memset(&ddofx, 0, sizeof(ddofx));
ddofx.dwSize = sizeof(ddofx);
ddofx.dckDestColorkey.dwColorSpaceLowValue = OVERLAY_COLOUR;
ddofx.dckDestColorkey.dwColorSpaceHighValue = OVERLAY_COLOUR;
HRESULT ddrval = lpColSurf->UpdateOverlay(&region, lpPrimSurf,
&rectClient,
DDOVER_SHOW|DDOVER_DDFX|DDOVER_KEYDESTOVERRIDE,
&ddofx);
if(ddrval != DD_OK)
{
GDBG_INFO(80, "Unable to update overlay!\n");
lpColSurf->Release();
lpColSurf = 0;
}
else
{
/* Now Fill Primary with the col */
DDBLTFX ddbfx;
memset(&ddbfx, 0, sizeof(ddbfx));
ddbfx.dwSize = sizeof(DDBLTFX);
ddbfx.dwFillColor = OVERLAY_COLOUR;
lpPrimSurf->Blt(&rectClient, NULL, &rectClient, DDBLT_COLORFILL|DDBLT_WAIT, &ddbfx);
}
}
}
}
}
/* This will release the context, and release the surfaces, and destroy the DDRAW objects */
extern "C" void _grReleaseWindowSurface(GrContext_t ctx)
{
// Release Context
grSurfaceReleaseContext(ctx);
GDBG_INFO(80, "Released context!\n");
// Now release all the surfaces
if (lpTexSurf1)
{
lpTexSurf1->Release();
lpTexSurf1 = 0;
}
if (lpTexSurf0)
{
lpTexSurf0->Release();
lpTexSurf0 = 0;
}
if (lpAuxSurf)
{
lpAuxSurf->Release();
lpAuxSurf = 0;
}
if (lpColSurf)
{
lpColSurf->Release();
lpColSurf = 0;
}
if (lpPrimSurf)
{
lpPrimSurf->Release();
lpPrimSurf = 0;
}
GDBG_INFO(80, "All surfaces released\n");
// And the DirectDraw objects
if (lpDDraw2)
{
lpDDraw2->Release();
lpDDraw2 = NULL;
}
if (lpDDraw1)
{
lpDDraw1->Release();
lpDDraw1 = NULL;
}
GDBG_INFO(80, "Direct Draw released\n");
}

2
glide3x/h5/glide3/src/xdraw3.asm
View File

@@ -3042,7 +3042,7 @@ $clip_setup_fog_oow:
add dataElem, 4 ; dataElem++
movss xmm3,[gc + depth_range] ;
pfmul xmm3,xmm0 ;
mulss xmm3,xmm0 ;
movss xmm4,[gc + depth_range] ; depth range
subss xmm4,xmm3 ;

3
glide3x/h5/minihwc/makefile
View File

@@ -27,6 +27,9 @@
!if "$(DEBUG)" == "1"
VCDEFS = $(VCDEFS) -DGDBG_INFO_ON=1 -DGLIDE_DEBUG=1
!if "$(CEASE_ALL_GDBG)"=="1"
VCDEFS = $(VCDEFS) -DCEASE_ALL_GDBG=1
!endif
!endif
!if "$(FX_TARGET)" == "DOS"

223
glide3x/h5/minihwc/minihwc.c
View File

@@ -19,6 +19,7 @@
**
** $Header$
** $Log:
** 100 1.72.8 05/30/03 KoolSmoky fixed up asm routines
** 99 GlideXP 1.72.7 12/28/01 Ryan Nunn Removing stupid debug
** comments.
** 98 GlideXP 1.72.6 12/28/01 Ryan Nunn Improving faster WinXP
@@ -3237,6 +3238,13 @@ hwcGetSurfaceInfo(const hwcBoardInfo* bInfo,
ret->height = desc.dwHeight;
ret->width = desc.dwWidth;
ret->bitdepth = _hwcPixelFormatToBitDepth(&desc.ddpfPixelFormat);
GDBG_INFO(80, "h %d\n", desc.dwHeight);
GDBG_INFO(80, "w %d\n", desc.dwWidth);
GDBG_INFO(80, "f %X\n", desc.dwFlags);
GDBG_INFO(80, "p %X\n", desc.ddpfPixelFormat.dwFlags);
GDBG_INFO(80, "c %X\n", desc.ddsCaps.dwCaps);
retVal = (ret->bitdepth != 0x00UL);
if (!retVal) {
sprintf(errorString, "%s: Invalid surface pixel format (0x%X)\n",
@@ -3530,6 +3538,12 @@ hwcAllocWinFifo(hwcBoardInfo* bInfo,
goto __errSurfaceFifo;
}
GDBG_INFO(80, "ph %d\n", primaryDesc.dwHeight);
GDBG_INFO(80, "pw %d\n", primaryDesc.dwWidth);
GDBG_INFO(80, "pf %X\n", primaryDesc.dwFlags);
GDBG_INFO(80, "pp %X\n", primaryDesc.ddpfPixelFormat.dwFlags);
GDBG_INFO(80, "pc %X\n", primaryDesc.ddsCaps.dwCaps);
/* If the DirectDraw driver supports wide surfaces then we can
* specify what will align well w/ the underlying hw's page
* size. Otherwise we need to check the primary surface's
@@ -4774,10 +4788,14 @@ hwcInitVideo(hwcBoardInfo *bInfo, FxBool tiled, FxVideoTimingInfo *vidTiming,
if (NULL == GETENV("SSTH3_OVERLAYMODE", bInfo->RegPath))
{
/* We are in optimal mode by default */
if(bInfo->vidInfo.xRes < 1024)
vidProcCfg |= SST_OVERLAY_FILTER_2X2;
else
vidProcCfg |= SST_OVERLAY_FILTER_4X4;
if(bpp == 32 && !((bInfo->pciInfo.numChips == 4) && (bInfo->h3pixelSample >= 4))) { /* 32bpp and not 4x,8xfsaa on v56k */
vidProcCfg |= SST_OVERLAY_FILTER_POINT;
} else {
if(bInfo->vidInfo.xRes < 1024)
vidProcCfg |= SST_OVERLAY_FILTER_2X2;
else
vidProcCfg |= SST_OVERLAY_FILTER_4X4;
}
}
else
{
@@ -4785,10 +4803,14 @@ hwcInitVideo(hwcBoardInfo *bInfo, FxBool tiled, FxVideoTimingInfo *vidTiming,
{
default:
case 1: /* Optimal */
if(bInfo->vidInfo.xRes < 1024)
vidProcCfg |= SST_OVERLAY_FILTER_2X2;
else
vidProcCfg |= SST_OVERLAY_FILTER_4X4;
if(bpp == 32 && !((bInfo->pciInfo.numChips == 4) && (bInfo->h3pixelSample >= 4))) { /* 32bpp and not 4x,8xfsaa on v56k */
vidProcCfg |= SST_OVERLAY_FILTER_POINT;
} else {
if(bInfo->vidInfo.xRes < 1024)
vidProcCfg |= SST_OVERLAY_FILTER_2X2;
else
vidProcCfg |= SST_OVERLAY_FILTER_4X4;
}
break;
case 2: /* Normal */
vidProcCfg |= SST_OVERLAY_FILTER_4X4;
@@ -4864,13 +4886,13 @@ hwcInitVideo(hwcBoardInfo *bInfo, FxBool tiled, FxVideoTimingInfo *vidTiming,
/* Sorry, can only use 4x1 filter in this mode. */
if ( bInfo->h3nwaySli > 1 ) {
if( ((vidProcCfg & SST_OVERLAY_FILTER_BILINEAR) == SST_OVERLAY_FILTER_BILINEAR) ||
((vidProcCfg & SST_OVERLAY_FILTER_2X2) == SST_OVERLAY_FILTER_2X2) ) {
if(vidProcCfg & SST_OVERLAY_FILTER_MODE) {
vidProcCfg &= ~SST_OVERLAY_FILTER_MODE;
vidProcCfg |= SST_OVERLAY_FILTER_4X4;
}
}
#if 0 /* use optimal or let the user decide */
/* Disable video filter in 32-bit mode */
#if 1 /* looks better with filter when in 16bpp 2xfsaa */
if(bpp == 32 || bInfo->h3pixelSample > 2) {
@@ -4880,6 +4902,7 @@ hwcInitVideo(hwcBoardInfo *bInfo, FxBool tiled, FxVideoTimingInfo *vidTiming,
vidProcCfg &= ~SST_OVERLAY_FILTER_MODE;
vidProcCfg |= SST_OVERLAY_FILTER_POINT;
}
#endif
HWC_IO_STORE(bInfo->regInfo, vidProcCfg, vidProcCfg);
@@ -6499,7 +6522,9 @@ static void hwcReadRegion565(hwcBoardInfo *bInfo, FxU32 src, FxU32 src_x, FxU32
end_x = src_x + src_width;
end_y = src_y + src_height;
s = (FxU8 *)(src + src_x*2 + src_y*strideInBytes);
s = (FxU8 *)(src + src_x*4);
for (y = 0; y < src_y; y++) if((y & renderMask) == compareMask) s += strideInBytes;
stride_diff = strideInBytes - (src_width*2);
if (bInfo->CPUInfo.os_support & _CPU_FEATURE_MMX)
@@ -6515,6 +6540,8 @@ static void hwcReadRegion565(hwcBoardInfo *bInfo, FxU32 src, FxU32 src_x, FxU32
);
#else
__asm {
emms /* mmx */
/* mm7 = zero */
pXor mm7, mm7;
@@ -6535,10 +6562,11 @@ static void hwcReadRegion565(hwcBoardInfo *bInfo, FxU32 src, FxU32 src_x, FxU32
push 0x10401080; /* g=10000010 b=10000100 */
MovQ mm4, [esp]
Add esp, 8
}
#endif
for(y = 0; y < end_y; y++)
for(y = src_y; y < end_y; y++)
{
if((y & renderMask) == compareMask)
{
@@ -6551,10 +6579,10 @@ static void hwcReadRegion565(hwcBoardInfo *bInfo, FxU32 src, FxU32 src_x, FxU32
/* Scanline Setup */
Mov eax, s
Mov edx, dst
Mov edi, end_x
Mov edi, src_width
Shl edi, 3
Add edi, edx
Xor ebx, ebx
Xor ebx, ebx
align 8
@@ -6597,6 +6625,7 @@ static void hwcReadRegion565(hwcBoardInfo *bInfo, FxU32 src, FxU32 src_x, FxU32
add eax, stride_diff
mov s, eax
mov dst, edx
};
#endif
}
@@ -6604,20 +6633,25 @@ static void hwcReadRegion565(hwcBoardInfo *bInfo, FxU32 src, FxU32 src_x, FxU32
{
dst += src_width*4;
}
}
}
#ifndef __DJGPP__
__asm {
emms /* mmx */
}
#endif
/* Reset MMX State */
_m_empty();
}
else
{
/* Standard Unoptimized Loop */
for(y = 0; y < end_y; y++)
for(y = src_y; y < end_y; y++)
{
if((y & renderMask) == compareMask)
{
/* This chip owns this scanline. */
for(x = 0; x < end_x; x++) {
for(x = src_x; x < end_x; x++)
{
*dst++ += (FxU16) (((*(FxU16*) s) >> 00) & 0x1F) << 3;
*dst++ += (FxU16) (((*(FxU16*) s) >> 05) & 0x3F) << 2;
*dst++ += (FxU16) (((*(FxU16*) s) >> 11) & 0x1F) << 3;
@@ -6648,7 +6682,9 @@ static void hwcReadRegion1555(hwcBoardInfo *bInfo, FxU32 src, FxU32 src_x, FxU32
end_x = src_x + src_width;
end_y = src_y + src_height;
s = (FxU8 *)(src + src_x*2 + src_y*strideInBytes);
s = (FxU8 *)(src + src_x*4);
for (y = 0; y < src_y; y++) if((y & renderMask) == compareMask) s += strideInBytes;
stride_diff = strideInBytes - (src_width*2);
if (bInfo->CPUInfo.os_support & _CPU_FEATURE_MMX)
@@ -6664,6 +6700,8 @@ static void hwcReadRegion1555(hwcBoardInfo *bInfo, FxU32 src, FxU32 src_x, FxU32
);
#else
__asm {
emms /* mmx */
/* mm7 = zero */
pXor mm7, mm7;
@@ -6684,10 +6722,11 @@ static void hwcReadRegion1555(hwcBoardInfo *bInfo, FxU32 src, FxU32 src_x, FxU32
push 0x10801080; /* g=1000010000 b=1000010000 */
MovQ mm4, [esp]
Add esp, 8
}
#endif
for(y = 0; y < end_y; y++)
for(y = src_y; y < end_y; y++)
{
if((y & renderMask) == compareMask)
{
@@ -6700,7 +6739,7 @@ static void hwcReadRegion1555(hwcBoardInfo *bInfo, FxU32 src, FxU32 src_x, FxU32
/* Scanline Setup */
Mov eax, s
Mov edx, dst
Mov edi, end_x
Mov edi, src_width
Shl edi, 3
Add edi, edx
Xor ebx, ebx
@@ -6745,6 +6784,7 @@ static void hwcReadRegion1555(hwcBoardInfo *bInfo, FxU32 src, FxU32 src_x, FxU32
add eax, stride_diff
mov s, eax
mov dst, edx
};
#endif
}
@@ -6752,20 +6792,24 @@ static void hwcReadRegion1555(hwcBoardInfo *bInfo, FxU32 src, FxU32 src_x, FxU32
{
dst += src_width*4;
}
}
}
/* Reset MMX State */
_m_empty();
#ifndef __DJGPP__
__asm {
emms /* mmx */
}
#endif
}
else
{
/* Standard Unoptimized Loop */
for(y = 0; y < end_y; y++)
for(y = src_y; y < end_y; y++)
{
if((y & renderMask) == compareMask)
{
/* This chip owns this scanline. */
for(x = 0; x < end_x; x++) {
for(x = src_x; x < end_x; x++)
{
*dst++ += (FxU16) (((*(FxU16*) s) >> 00) & 0x1F) << 3;
*dst++ += (FxU16) (((*(FxU16*) s) >> 05) & 0x1F) << 3;
*dst++ += (FxU16) (((*(FxU16*) s) >> 10) & 0x1F) << 3;
@@ -6797,7 +6841,9 @@ static void hwcReadRegion8888(hwcBoardInfo *bInfo, FxU32 src, FxU32 src_x, FxU32
end_x = src_x + src_width;
end_y = src_y + src_height;
s = (FxU8 *)(src + src_x*4 + src_y*strideInBytes);
s = (FxU8 *)(src + src_x*4);
for (y = 0; y < src_y; y++) if((y & renderMask) == compareMask) s += strideInBytes;
stride_diff = strideInBytes - (src_width*4);
if (bInfo->CPUInfo.os_support & _CPU_FEATURE_MMX)
@@ -6806,10 +6852,14 @@ static void hwcReadRegion8888(hwcBoardInfo *bInfo, FxU32 src, FxU32 src_x, FxU32
#ifdef __DJGPP__
MMX_SETUP4();
#else
__asm pXor mm7, mm7;
__asm {
emms /* mmx */
pXor mm7, mm7
}
#endif
for(y = 0; y < end_y; y++)
for(y = src_y; y < end_y; y++)
{
if((y & renderMask) == compareMask)
{
@@ -6822,7 +6872,7 @@ static void hwcReadRegion8888(hwcBoardInfo *bInfo, FxU32 src, FxU32 src_x, FxU32
/* Scanline Setup */
Mov eax, s
Mov edx, dst
Mov edi, end_x
Mov edi, src_width
Shl edi, 3
Add edi, edx
@@ -6847,6 +6897,7 @@ static void hwcReadRegion8888(hwcBoardInfo *bInfo, FxU32 src, FxU32 src_x, FxU32
add eax, stride_diff
mov s, eax
mov dst, edx
};
#endif
}
@@ -6856,18 +6907,21 @@ static void hwcReadRegion8888(hwcBoardInfo *bInfo, FxU32 src, FxU32 src_x, FxU32
}
}
/* Reset MMX State */
_m_empty();
#ifndef __DJGPP__
__asm {
emms /* mmx */
}
#endif
}
else
{
/* Standard Unoptimized Loop */
for(y = 0; y < end_y; y++)
for(y = src_y; y < end_y; y++)
{
if((y & renderMask) == compareMask)
{
/* This chip owns this scanline. */
for(x = 0; x < end_x; x++)
for(x = src_x; x < end_x; x++)
{
*dst++ += (FxU16) ((*(FxU32*) s) >> 0) & 0xFF;
*dst++ += (FxU16) ((*(FxU32*) s) >> 8) & 0xFF;
@@ -6951,7 +7005,7 @@ FxU32 hwcAAReadRegion16(hwcBoardInfo *bInfo, FxU32 colBufNum,
* or 4 chip boards */
if ((bInfo->h3pixelSize == 2) ||
(bInfo->h3pixelSample == 2) ||
(numChips == 4)) {
(numChips == 4 && bInfo->h3pixelSample != 1)) {
/* Figure out framebuffer format (1555 or 565) */
HWC_IO_LOAD(bInfo->regInfo, vidProcCfg, vidProcCfg);
@@ -7135,6 +7189,8 @@ static void hwcCopyBuffer8888Flipped(hwcBoardInfo *bInfo, FxU16 *source, int w,
/* MMX Optimized Loop */
__asm
{
emms /* mmx */
Mov eax, src /* eax = source */
Mov edx, dst /* edx = dest */
Mov edi, endline /* edi = endline */
@@ -7181,9 +7237,9 @@ static void hwcCopyBuffer8888Flipped(hwcBoardInfo *bInfo, FxU16 *source, int w,
Sub eax, ebx /* src -= w*2; */
cmp edx, end /* if (dst!=end) */
jne loop_begin /* goto loop_begin; */
emms /* mmx */
/* Reset MMX State */
emms;
}
}
else
@@ -7217,6 +7273,8 @@ static void hwcCopyBuffer8888FlippedShifted(hwcBoardInfo *bInfo, FxU16 *source,
/* MMX Optimized Loop */
__asm
{
emms /* mmx */
Mov eax, src /* eax = source */
Mov edx, dst /* edx = dest */
Mov edi, endline /* edi = endline */
@@ -7265,9 +7323,9 @@ static void hwcCopyBuffer8888FlippedShifted(hwcBoardInfo *bInfo, FxU16 *source,
Sub eax, ebx /* src -= w*2; */
cmp edx, end /* if (dst!=end) */
jne loop_begin /* goto loop_begin; */
emms /* mmx */
/* Reset MMX State */
emms;
}
}
else
@@ -7312,6 +7370,8 @@ static void hwcCopyBuffer8888FlippedDithered(hwcBoardInfo *bInfo, FxU16 *source,
/* MMX Optimized Loop */
__asm
{
emms /* mmx */
/* mm7 = all ones */
Mov eax, 0xFFFFFFFF
MovD mm7, eax
@@ -7484,8 +7544,8 @@ static void hwcCopyBuffer8888FlippedDithered(hwcBoardInfo *bInfo, FxU16 *source,
jne mmx_loop_begin /* goto mmx_loop_begin; */
finished:
/* Reset MMX State */
emms;
emms /* mmx */
}
}
else
@@ -7569,10 +7629,7 @@ void hwcAAScreenShot(hwcBoardInfo *bInfo, FxU32 colBufNum, FxBool dither)
{
hwcCopyBuffer8888Flipped(bInfo, buffer, bInfo->vidInfo.xRes, bInfo->vidInfo.yRes, out, 0);
}
/* MMX Stuff */
_m_empty();
/* Write buffer to disk */
#ifdef _WIN32
GetLocalTime(&curtime);
@@ -7739,7 +7796,63 @@ static void hwcCopyBuffer565Shifted(hwcBoardInfo *bInfo, FxU16 *src, int w, int
gshift = 3 - aaShift;
rshift = 8 - aaShift;
while (dst<end)
if (bInfo->CPUInfo.os_support & _CPU_FEATURE_MMX)
{
/* MMX Optimized Loop */
__asm
{
emms /* mmx */
Mov eax, src /* eax = source */
Mov edx, dst /* edx = dest */
Mov edi, endline /* edi = endline */
/* mm6 = aaShift */
MovD mm6, aaShift
/* mm5 = mask */
push 0x0000F800; /* a r */
push 0xFC00F800; /* g b */
MovQ mm5, [esp]
Add esp, 8
align 8
loop_begin:
/* Read Pixel, Reduce to 32 bits */
MovQ mm0, mmword ptr [eax]
pSRLW mm0, mm6 /* AA Shift to 32 bit */
//pAnd mm0, mm5 /* 565 Mask */
PackUSWB mm0, mm0
MovD ebx, mm0
/* Convert to 565 */
Mov ecx, ebx
And ebx, 0x0000FCF8
And ecx, 0x00F80000
Shr bh, 2
Shr ecx, 8
Shr ebx, 3
Or ebx, ecx
/* Write */
Mov word ptr [edx], bx
/* Next Pixel */
Add eax, 8
Add edx, 2
cmp edx, edi /* if (dst!=endline) */
jne loop_begin /* goto loop_begin; */
/* Next Scanline */
Add edi, stride_dest/* endline += stride_dest; */
Add edx, stride_diff/* dest += stride_diff; */
cmp edx, end /* if (dst!=end) */
jne loop_begin /* goto loop_begin; */
emms /* mmx */
}
}
else while (dst<end)
{
while (dst<endline)
{
@@ -8041,7 +8154,7 @@ void hwcAAReadRegion(hwcBoardInfo *bInfo, FxU32 colBufNum,
{
hwcCopyBuffer1555(bInfo, buffer, src_width, src_height, dst_stride, dst_data, 0);
}
}
}
/* Free memory */
_aligned_free (buffer);
@@ -8052,7 +8165,7 @@ void hwcAAReadRegion(hwcBoardInfo *bInfo, FxU32 colBufNum,
#define BLUE_SHIFT 0
#define CLAMP(val, min, max) if (val > max) val = max; /*else if (val < min) val = min;*//* KoolSmoky */
#define CLAMP(val, min, max) if (val > max) val = max; else if (val < min) val = min
#define ADJUST(val, lowest, low, high, typ) if (high < lowest) val=(typ)(low); else val = (typ)(high)
#define GETFLOATENV(s, r, v) if (GETENV(s, r)) v = (FxFloat)(atof(GETENV(s, r)))
@@ -8159,11 +8272,15 @@ hwcGammaTable(hwcBoardInfo *bInfo, FxU32 nEntries, FxU32 *r, FxU32 *g, FxU32 *b)
GDBG_INFO(69,": gRamp[%d] = %d\n", i, gRamp[i]);
}
/* Colourless - Hack for V5 6000 4x and 8x FSAA*/
/* Colourless - Hack for V5 6000 4x and 8x FSAA */
/* KoolSmoky - Since the DAC is shared between 2 chips, the input to gamma look up
table is divided by extra 2 and the MSB is lost. To correct this the table should
be only 7 bits. The resulting image will have only 7.5 bits per prime */
if (bInfo->pciInfo.numChips == 4 && bInfo->h3pixelSample >= 4) {
/* Go through 0 to 127 */
for (i = 0; i < 128; i++) gRamp[i] = gRamp[(i*255)/127];
gRamp[0] = 0; /* KoolSmoky - if row 0 is not 0 on napalm, we get strange banding effects on exit to desktop */
/* Go through 1 to 127 */
for (i = 1; i < 128; i++) gRamp[i] = ((gRamp[(i*255)/127] >> 1) & 0x007F7F7F); /* KoolSmoky - dac output is doubled in 4x, 8xfsaa */
/* Go through 128 to 255 */
for (; i < 256; i++) gRamp[i] = gRamp[127];