Sweden-Number/dlls/wined3d/ati_fragment_shader.c

1046 lines
51 KiB
C

/*
* Fixed function pipeline replacement using GL_ATI_fragment_shader
*
* Copyright 2008 Stefan Dösinger(for CodeWeavers)
*
* This library is free software; you can redistribute it and/or
* modify it under the terms of the GNU Lesser General Public
* License as published by the Free Software Foundation; either
* version 2.1 of the License, or (at your option) any later version.
*
* This library 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
* Lesser General Public License for more details.
*
* You should have received a copy of the GNU Lesser General Public
* License along with this library; if not, write to the Free Software
* Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301, USA
*/
#include "config.h"
#include <math.h>
#include <stdio.h>
#include "wined3d_private.h"
WINE_DEFAULT_DEBUG_CHANNEL(d3d_shader);
/* Some private defines, Constant associations, etc
* Env bump matrix and per stage constant should be independent,
* a stage that bumpmaps can't read the per state constant
*/
#define ATI_FFP_CONST_BUMPMAT(i) (GL_CON_0_ATI + i)
#define ATI_FFP_CONST_CONSTANT0 GL_CON_0_ATI
#define ATI_FFP_CONST_CONSTANT1 GL_CON_1_ATI
#define ATI_FFP_CONST_CONSTANT2 GL_CON_2_ATI
#define ATI_FFP_CONST_CONSTANT3 GL_CON_3_ATI
#define ATI_FFP_CONST_CONSTANT4 GL_CON_4_ATI
#define ATI_FFP_CONST_CONSTANT5 GL_CON_5_ATI
#define ATI_FFP_CONST_TFACTOR GL_CON_6_ATI
/* GL_ATI_fragment_shader specific fixed function pipeline description. "Inherits" from the common one */
struct atifs_ffp_desc
{
struct ffp_desc parent;
GLuint shader;
};
struct atifs_private_data
{
struct shader_arb_priv parent;
struct list fragment_shaders; /* A linked list to track fragment pipeline replacement shaders */
};
static const char *debug_dstmod(GLuint mod) {
switch(mod) {
case GL_NONE: return "GL_NONE";
case GL_2X_BIT_ATI: return "GL_2X_BIT_ATI";
case GL_4X_BIT_ATI: return "GL_4X_BIT_ATI";
case GL_8X_BIT_ATI: return "GL_8X_BIT_ATI";
case GL_HALF_BIT_ATI: return "GL_HALF_BIT_ATI";
case GL_QUARTER_BIT_ATI: return "GL_QUARTER_BIT_ATI";
case GL_EIGHTH_BIT_ATI: return "GL_EIGHTH_BIT_ATI";
case GL_SATURATE_BIT_ATI: return "GL_SATURATE_BIT_ATI";
default: return "Unexpected modifier\n";
}
}
static const char *debug_argmod(GLuint mod) {
switch(mod) {
case GL_NONE:
return "GL_NONE";
case GL_2X_BIT_ATI:
return "GL_2X_BIT_ATI";
case GL_COMP_BIT_ATI:
return "GL_COMP_BIT_ATI";
case GL_NEGATE_BIT_ATI:
return "GL_NEGATE_BIT_ATI";
case GL_BIAS_BIT_ATI:
return "GL_BIAS_BIT_ATI";
case GL_2X_BIT_ATI | GL_COMP_BIT_ATI:
return "GL_2X_BIT_ATI | GL_COMP_BIT_ATI";
case GL_2X_BIT_ATI | GL_NEGATE_BIT_ATI:
return "GL_2X_BIT_ATI | GL_NEGATE_BIT_ATI";
case GL_2X_BIT_ATI | GL_BIAS_BIT_ATI:
return "GL_2X_BIT_ATI | GL_BIAS_BIT_ATI";
case GL_COMP_BIT_ATI | GL_NEGATE_BIT_ATI:
return "GL_COMP_BIT_ATI | GL_NEGATE_BIT_ATI";
case GL_COMP_BIT_ATI | GL_BIAS_BIT_ATI:
return "GL_COMP_BIT_ATI | GL_BIAS_BIT_ATI";
case GL_NEGATE_BIT_ATI | GL_BIAS_BIT_ATI:
return "GL_NEGATE_BIT_ATI | GL_BIAS_BIT_ATI";
case GL_COMP_BIT_ATI | GL_NEGATE_BIT_ATI | GL_BIAS_BIT_ATI:
return "GL_COMP_BIT_ATI | GL_NEGATE_BIT_ATI | GL_BIAS_BIT_ATI";
case GL_2X_BIT_ATI | GL_NEGATE_BIT_ATI | GL_BIAS_BIT_ATI:
return "GL_2X_BIT_ATI | GL_NEGATE_BIT_ATI | GL_BIAS_BIT_ATI";
case GL_2X_BIT_ATI | GL_COMP_BIT_ATI | GL_BIAS_BIT_ATI:
return "GL_2X_BIT_ATI | GL_COMP_BIT_ATI | GL_BIAS_BIT_ATI";
case GL_2X_BIT_ATI | GL_COMP_BIT_ATI | GL_NEGATE_BIT_ATI:
return "GL_2X_BIT_ATI | GL_COMP_BIT_ATI | GL_NEGATE_BIT_ATI";
case GL_2X_BIT_ATI | GL_COMP_BIT_ATI | GL_NEGATE_BIT_ATI | GL_BIAS_BIT_ATI:
return "GL_2X_BIT_ATI | GL_COMP_BIT_ATI | GL_NEGATE_BIT_ATI | GL_BIAS_BIT_ATI";
default:
return "Unexpected argmod combination\n";
}
}
static const char *debug_register(GLuint reg) {
switch(reg) {
case GL_REG_0_ATI: return "GL_REG_0_ATI";
case GL_REG_1_ATI: return "GL_REG_1_ATI";
case GL_REG_2_ATI: return "GL_REG_2_ATI";
case GL_REG_3_ATI: return "GL_REG_3_ATI";
case GL_REG_4_ATI: return "GL_REG_4_ATI";
case GL_REG_5_ATI: return "GL_REG_5_ATI";
case GL_CON_0_ATI: return "GL_CON_0_ATI";
case GL_CON_1_ATI: return "GL_CON_1_ATI";
case GL_CON_2_ATI: return "GL_CON_2_ATI";
case GL_CON_3_ATI: return "GL_CON_3_ATI";
case GL_CON_4_ATI: return "GL_CON_4_ATI";
case GL_CON_5_ATI: return "GL_CON_5_ATI";
case GL_CON_6_ATI: return "GL_CON_6_ATI";
case GL_CON_7_ATI: return "GL_CON_7_ATI";
case GL_ZERO: return "GL_ZERO";
case GL_ONE: return "GL_ONE";
case GL_PRIMARY_COLOR: return "GL_PRIMARY_COLOR";
case GL_SECONDARY_INTERPOLATOR_ATI: return "GL_SECONDARY_INTERPOLATOR_ATI";
default: return "Unknown register\n";
}
}
static const char *debug_swizzle(GLuint swizzle) {
switch(swizzle) {
case GL_SWIZZLE_STR_ATI: return "GL_SWIZZLE_STR_ATI";
case GL_SWIZZLE_STQ_ATI: return "GL_SWIZZLE_STQ_ATI";
case GL_SWIZZLE_STR_DR_ATI: return "GL_SWIZZLE_STR_DR_ATI";
case GL_SWIZZLE_STQ_DQ_ATI: return "GL_SWIZZLE_STQ_DQ_ATI";
default: return "unknown swizzle";
}
}
#define GLINFO_LOCATION (*gl_info)
static GLuint register_for_arg(DWORD arg, WineD3D_GL_Info *gl_info, unsigned int stage, GLuint *mod, GLuint tmparg) {
GLenum ret;
if(mod) *mod = GL_NONE;
if(arg == 0xFFFFFFFF) return -1; /* This is the marker for unused registers */
switch(arg & WINED3DTA_SELECTMASK) {
case WINED3DTA_DIFFUSE:
ret = GL_PRIMARY_COLOR;
break;
case WINED3DTA_CURRENT:
/* Note that using GL_REG_0_ATI for the passed on register is safe because
* texture0 is read at stage0, so in the worst case it is read in the
* instruction writing to reg0. Afterwards texture0 is not used any longer.
* If we're reading from current
*/
if(stage == 0) {
ret = GL_PRIMARY_COLOR;
} else {
ret = GL_REG_0_ATI;
}
break;
case WINED3DTA_TEXTURE:
ret = GL_REG_0_ATI + stage;
break;
case WINED3DTA_TFACTOR:
ret = ATI_FFP_CONST_TFACTOR;
break;
case WINED3DTA_SPECULAR:
ret = GL_SECONDARY_INTERPOLATOR_ATI;
break;
case WINED3DTA_TEMP:
ret = tmparg;
break;
case WINED3DTA_CONSTANT:
FIXME("Unhandled source argument WINED3DTA_TEMP\n");
ret = GL_CON_0_ATI;
break;
default:
FIXME("Unknown source argument %d\n", arg);
ret = GL_ZERO;
}
if(arg & WINED3DTA_COMPLEMENT) {
if(mod) *mod |= GL_COMP_BIT_ATI;
}
if(arg & WINED3DTA_ALPHAREPLICATE) {
FIXME("Unhandled read modifier WINED3DTA_ALPHAREPLICATE\n");
}
return ret;
}
static GLuint find_tmpreg(struct texture_stage_op op[MAX_TEXTURES]) {
int lowest_read = -1;
int lowest_write = -1;
int i;
BOOL tex_used[MAX_TEXTURES];
memset(tex_used, 0, sizeof(tex_used));
for(i = 0; i < MAX_TEXTURES; i++) {
if(op[i].cop == WINED3DTOP_DISABLE) {
break;
}
if(lowest_read == -1 &&
(op[i].carg1 == WINED3DTA_TEMP || op[i].carg2 == WINED3DTA_TEMP || op[i].carg0 == WINED3DTA_TEMP ||
op[i].aarg1 == WINED3DTA_TEMP || op[i].aarg2 == WINED3DTA_TEMP || op[i].aarg0 == WINED3DTA_TEMP)) {
lowest_read = i;
}
if(lowest_write == -1 && op[i].dst == WINED3DTA_TEMP) {
lowest_write = i;
}
if(op[i].carg1 == WINED3DTA_TEXTURE || op[i].carg2 == WINED3DTA_TEXTURE || op[i].carg0 == WINED3DTA_TEXTURE ||
op[i].aarg1 == WINED3DTA_TEXTURE || op[i].aarg2 == WINED3DTA_TEXTURE || op[i].aarg0 == WINED3DTA_TEXTURE) {
tex_used[i] = TRUE;
}
}
/* Temp reg not read? We don't need it, return GL_NONE */
if(lowest_read == -1) return GL_NONE;
if(lowest_write >= lowest_read) {
FIXME("Temp register read before beeing written\n");
}
if(lowest_write == -1) {
/* This needs a test. Maybe we are supposed to return 0.0/0.0/0.0/0.0, or fail drawprim, or whatever */
FIXME("Temp register read without beeing written\n");
return GL_REG_1_ATI;
} else if(lowest_write >= 1) {
/* If we're writing to the temp reg at earliest in stage 1, we can use register 1 for the temp result.
* there may be texture data stored in reg 1, but we do not need it any longer since stage 1 already
* read it
*/
return GL_REG_1_ATI;
} else {
/* Search for a free texture register. We have 6 registers available. GL_REG_0_ATI is already used
* for the regular result
*/
for(i = 1; i < 6; i++) {
if(!tex_used[i]) {
return GL_REG_0_ATI + i;
}
}
/* What to do here? Report it in ValidateDevice? */
FIXME("Could not find a register for the temporary register\n");
return 0;
}
}
static GLuint gen_ati_shader(struct texture_stage_op op[MAX_TEXTURES], WineD3D_GL_Info *gl_info) {
GLuint ret = GL_EXTCALL(glGenFragmentShadersATI(1));
unsigned int stage;
GLuint arg0, arg1, arg2, extrarg;
GLuint dstmod, argmod0, argmod1, argmod2, argmodextra;
GLuint swizzle;
GLuint tmparg = find_tmpreg(op);
GLuint dstreg;
if(!ret) {
ERR("Failed to generate a GL_ATI_fragment_shader shader id\n");
return 0;
}
GL_EXTCALL(glBindFragmentShaderATI(ret));
checkGLcall("GL_EXTCALL(glBindFragmentShaderATI(ret))");
TRACE("glBeginFragmentShaderATI()\n");
GL_EXTCALL(glBeginFragmentShaderATI());
checkGLcall("GL_EXTCALL(glBeginFragmentShaderATI())");
/* Pass 1: Generate sampling instructions for perturbation maps */
for(stage = 0; stage < GL_LIMITS(textures); stage++) {
if(op[stage].cop == WINED3DTOP_DISABLE) break;
if(op[stage].cop != WINED3DTOP_BUMPENVMAP &&
op[stage].cop != WINED3DTOP_BUMPENVMAPLUMINANCE) continue;
TRACE("glSampleMapATI(GL_REG_%d_ATI, GL_TEXTURE_%d_ARB, GL_SWIZZLE_STR_ATI)\n",
stage, stage);
GL_EXTCALL(glSampleMapATI(GL_REG_0_ATI + stage,
GL_TEXTURE0_ARB + stage,
GL_SWIZZLE_STR_ATI));
TRACE("glPassTexCoordATI(GL_REG_%d_ATI, GL_TEXTURE_%d_ARB, GL_SWIZZLE_STR_ATI)\n",
stage + 1, stage + 1);
GL_EXTCALL(glPassTexCoordATI(GL_REG_0_ATI + stage + 1,
GL_TEXTURE0_ARB + stage + 1,
GL_SWIZZLE_STR_ATI));
/* We need GL_REG_5_ATI as a temporary register to swizzle the bump matrix. So we run into
* issues if we're bump mapping on stage 4 or 5
*/
if(stage >= 4) {
FIXME("Bump mapping in stage %d\n", stage);
}
}
/* Pass 2: Generate perturbation calculations */
for(stage = 0; stage < GL_LIMITS(textures); stage++) {
if(op[stage].cop == WINED3DTOP_DISABLE) break;
if(op[stage].cop != WINED3DTOP_BUMPENVMAP &&
op[stage].cop != WINED3DTOP_BUMPENVMAPLUMINANCE) continue;
/* Nice thing, we get the color correction for free :-) */
if(op[stage].color_correction == WINED3DFMT_V8U8) {
argmodextra = GL_2X_BIT_ATI | GL_BIAS_BIT_ATI;
} else {
argmodextra = 0;
}
TRACE("glColorFragmentOp3ATI(GL_DOT2_ADD_ATI, GL_REG_%d_ATI, GL_RED_BIT_ATI, GL_NONE, GL_REG_%d_ATI, GL_NONE, %s, ATI_FFP_CONST_BUMPMAT(%d), GL_NONE, GL_NONE, GL_REG_%d_ATI, GL_RED, GL_NONE)\n",
stage + 1, stage, debug_argmod(argmodextra), stage, stage + 1);
GL_EXTCALL(glColorFragmentOp3ATI(GL_DOT2_ADD_ATI, GL_REG_0_ATI + stage + 1, GL_RED_BIT_ATI, GL_NONE,
GL_REG_0_ATI + stage, GL_NONE, argmodextra,
ATI_FFP_CONST_BUMPMAT(stage), GL_NONE, GL_2X_BIT_ATI | GL_BIAS_BIT_ATI,
GL_REG_0_ATI + stage + 1, GL_RED, GL_NONE));
/* FIXME: How can I make GL_DOT2_ADD_ATI read the factors from blue and alpha? It defaults to red and green,
* and it is fairly easy to make it read GL_BLUE or BL_ALPHA, but I can't get an R * B + G * A. So we're wasting
* one register and two instructions in this pass for a simple swizzling operation.
* For starters it might be good enough to merge the two movs into one, but even that isn't possible :-(
*
* NOTE: GL_BLUE | GL_ALPHA is not possible. It doesn't throw a compilation error, but an OR operation on the
* constants doesn't make sense, considering their values.
*/
TRACE("glColorFragmentOp1ATI(GL_MOV_ATI, GL_REG_5_ATI, GL_RED_BIT_ATI, GL_NONE, ATI_FFP_CONST_BUMPMAT(%d), GL_BLUE, GL_NONE)\n", stage);
GL_EXTCALL(glColorFragmentOp1ATI(GL_MOV_ATI, GL_REG_5_ATI, GL_RED_BIT_ATI, GL_NONE,
ATI_FFP_CONST_BUMPMAT(stage), GL_BLUE, GL_NONE));
TRACE("glColorFragmentOp1ATI(GL_MOV_ATI, GL_REG_5_ATI, GL_GREEN_BIT_ATI, GL_NONE, ATI_FFP_CONST_BUMPMAT(%d), GL_ALPHA, GL_NONE)\n", stage);
GL_EXTCALL(glColorFragmentOp1ATI(GL_MOV_ATI, GL_REG_5_ATI, GL_GREEN_BIT_ATI, GL_NONE,
ATI_FFP_CONST_BUMPMAT(stage), GL_ALPHA, GL_NONE));
TRACE("glColorFragmentOp3ATI(GL_DOT2_ADD_ATI, GL_REG_%d_ATI, GL_GREEN_BIT_ATI, GL_NONE, GL_REG_%d_ATI, GL_NONE, %s, GL_REG_5_ATI, GL_NONE, GL_NONE, GL_REG_%d_ATI, GL_GREEN, GL_NONE)\n",
stage + 1, stage, debug_argmod(argmodextra), stage + 1);
GL_EXTCALL(glColorFragmentOp3ATI(GL_DOT2_ADD_ATI, GL_REG_0_ATI + stage + 1, GL_GREEN_BIT_ATI, GL_NONE,
GL_REG_0_ATI + stage, GL_NONE, argmodextra,
GL_REG_5_ATI, GL_NONE, GL_2X_BIT_ATI | GL_BIAS_BIT_ATI,
GL_REG_0_ATI + stage + 1, GL_GREEN, GL_NONE));
}
/* Pass 3: Generate sampling instructions for regular textures */
for(stage = 0; stage < GL_LIMITS(textures); stage++) {
if(op[stage].cop == WINED3DTOP_DISABLE) {
break;
}
if(op[stage].projected == proj_none) {
swizzle = GL_SWIZZLE_STR_ATI;
} else if(op[stage].projected == proj_count3) {
/* TODO: D3DTTFF_COUNT3 | D3DTTFF_PROJECTED would be GL_SWIZZLE_STR_DR_ATI.
* However, the FFP vertex processing texture transform matrix handler does
* some transformations in the texture matrix which makes the 3rd coordinate
* arrive in Q, not R in that case. This is needed for opengl fixed function
* fragment processing which always divides by Q. In this backend we can
* handle that properly and be compatible with vertex shader output and avoid
* side effects of the texture matrix games
*/
swizzle = GL_SWIZZLE_STQ_DQ_ATI;
} else {
swizzle = GL_SWIZZLE_STQ_DQ_ATI;
}
if((op[stage].carg0 & WINED3DTA_SELECTMASK) == WINED3DTA_TEXTURE ||
(op[stage].carg1 & WINED3DTA_SELECTMASK) == WINED3DTA_TEXTURE ||
(op[stage].carg2 & WINED3DTA_SELECTMASK) == WINED3DTA_TEXTURE ||
(op[stage].aarg0 & WINED3DTA_SELECTMASK) == WINED3DTA_TEXTURE ||
(op[stage].aarg1 & WINED3DTA_SELECTMASK) == WINED3DTA_TEXTURE ||
(op[stage].aarg2 & WINED3DTA_SELECTMASK) == WINED3DTA_TEXTURE ||
op[stage].cop == WINED3DTOP_BLENDTEXTUREALPHA) {
if(stage > 0 &&
(op[stage - 1].cop == WINED3DTOP_BUMPENVMAP ||
op[stage - 1].cop == WINED3DTOP_BUMPENVMAPLUMINANCE)) {
TRACE("glSampleMapATI(GL_REG_%d_ATI, GL_REG_%d_ATI, GL_SWIZZLE_STR_ATI)\n",
stage, stage);
GL_EXTCALL(glSampleMapATI(GL_REG_0_ATI + stage,
GL_REG_0_ATI + stage,
GL_SWIZZLE_STR_ATI));
} else {
TRACE("glSampleMapATI(GL_REG_%d_ATI, GL_TEXTURE_%d_ARB, %s)\n",
stage, stage, debug_swizzle(swizzle));
GL_EXTCALL(glSampleMapATI(GL_REG_0_ATI + stage,
GL_TEXTURE0_ARB + stage,
swizzle));
}
}
}
/* Pass 4: Generate the arithmetic instructions */
for(stage = 0; stage < MAX_TEXTURES; stage++) {
if(op[stage].cop == WINED3DTOP_DISABLE) {
if(stage == 0) {
/* Handle complete texture disabling gracefully */
TRACE("glColorFragmentOp1ATI(GL_MOV_ATI, GL_REG_0_ATI, GL_NONE, GL_NONE, GL_PRIMARY_COLOR, GL_NONE, GL_NONE)\n");
GL_EXTCALL(glColorFragmentOp1ATI(GL_MOV_ATI, GL_REG_0_ATI, GL_NONE, GL_NONE,
GL_PRIMARY_COLOR, GL_NONE, GL_NONE));
TRACE("glAlphaFragmentOp1ATI(GL_MOV_ATI, GL_REG_0_ATI, GL_NONE, GL_PRIMARY_COLOR, GL_NONE, GL_NONE)\n");
GL_EXTCALL(glAlphaFragmentOp1ATI(GL_MOV_ATI, GL_REG_0_ATI, GL_NONE,
GL_PRIMARY_COLOR, GL_NONE, GL_NONE));
}
break;
}
if(op[stage].dst == WINED3DTA_TEMP) {
/* If we're writing to D3DTA_TEMP, but never reading from it we don't have to write there in the first place.
* skip the entire stage, this saves some GPU time
*/
if(tmparg == GL_NONE) continue;
dstreg = tmparg;
} else {
dstreg = GL_REG_0_ATI;
}
arg0 = register_for_arg(op[stage].carg0, gl_info, stage, &argmod0, tmparg);
arg1 = register_for_arg(op[stage].carg1, gl_info, stage, &argmod1, tmparg);
arg2 = register_for_arg(op[stage].carg2, gl_info, stage, &argmod2, tmparg);
dstmod = GL_NONE;
argmodextra = GL_NONE;
extrarg = GL_NONE;
switch(op[stage].cop) {
case WINED3DTOP_SELECTARG2:
arg1 = arg2;
argmod1 = argmod2;
case WINED3DTOP_SELECTARG1:
TRACE("glColorFragmentOp1ATI(GL_MOV_ATI, %s, GL_NONE, GL_NONE, %s, GL_NONE, %s)\n",
debug_register(dstreg), debug_register(arg1), debug_argmod(argmod1));
GL_EXTCALL(glColorFragmentOp1ATI(GL_MOV_ATI, dstreg, GL_NONE, GL_NONE,
arg1, GL_NONE, argmod1));
break;
case WINED3DTOP_MODULATE4X:
if(dstmod == GL_NONE) dstmod = GL_4X_BIT_ATI;
case WINED3DTOP_MODULATE2X:
if(dstmod == GL_NONE) dstmod = GL_2X_BIT_ATI;
case WINED3DTOP_MODULATE:
TRACE("glColorFragmentOp2ATI(GL_MUL_ATI, %s, GL_NONE, %s, %s, GL_NONE, %s, %s, GL_NONE, %s)\n",
debug_register(dstreg), debug_dstmod(dstmod),
debug_register(arg1), debug_argmod(argmod1),
debug_register(arg2), debug_argmod(argmod2));
GL_EXTCALL(glColorFragmentOp2ATI(GL_MUL_ATI, dstreg, GL_NONE, dstmod,
arg1, GL_NONE, argmod1,
arg2, GL_NONE, argmod2));
break;
case WINED3DTOP_ADDSIGNED2X:
dstmod = GL_2X_BIT_ATI;
case WINED3DTOP_ADDSIGNED:
argmodextra = GL_BIAS_BIT_ATI;
case WINED3DTOP_ADD:
TRACE("glColorFragmentOp2ATI(GL_ADD_ATI, %s, GL_NONE, %s, %s, GL_NONE, %s, %s, GL_NONE, %s)\n",
debug_register(dstreg), debug_dstmod(dstmod),
debug_register(arg1), debug_argmod(argmod1),
debug_register(arg2), debug_argmod(argmodextra | argmod2));
GL_EXTCALL(glColorFragmentOp2ATI(GL_ADD_ATI, GL_REG_0_ATI, GL_NONE, dstmod,
arg1, GL_NONE, argmod1,
arg2, GL_NONE, argmodextra | argmod2));
break;
case WINED3DTOP_SUBTRACT:
TRACE("glColorFragmentOp2ATI(GL_SUB_ATI, %s, GL_NONE, %s, %s, GL_NONE, %s, %s, GL_NONE, %s)\n",
debug_register(dstreg), debug_dstmod(dstmod),
debug_register(arg1), debug_argmod(argmod1),
debug_register(arg2), debug_argmod(argmod2));
GL_EXTCALL(glColorFragmentOp2ATI(GL_SUB_ATI, dstreg, GL_NONE, dstmod,
arg1, GL_NONE, argmod1,
arg2, GL_NONE, argmod2));
break;
case WINED3DTOP_ADDSMOOTH:
argmodextra = argmod1 & GL_COMP_BIT_ATI ? argmod1 & ~GL_COMP_BIT_ATI : argmod1 | GL_COMP_BIT_ATI;
TRACE("glColorFragmentOp3ATI(GL_MAD_ATI, %s, GL_NONE, GL_NONE, %s, GL_NONE, %s, %s, GL_NONE, %s, %s, GL_NONE, %s)\n",
debug_register(dstreg),
debug_register(arg2), debug_argmod(argmod2),
debug_register(arg1), debug_argmod(argmodextra),
debug_register(arg1), debug_argmod(argmod1));
/* Dst = arg1 + * arg2(1 -arg 1)
* = arg2 * (1 - arg1) + arg1
*/
GL_EXTCALL(glColorFragmentOp3ATI(GL_MAD_ATI, dstreg, GL_NONE, GL_NONE,
arg2, GL_NONE, argmod2,
arg1, GL_NONE, argmodextra,
arg1, GL_NONE, argmod1));
break;
case WINED3DTOP_BLENDCURRENTALPHA:
if(extrarg == GL_NONE) extrarg = register_for_arg(WINED3DTA_CURRENT, gl_info, stage, NULL, -1);
case WINED3DTOP_BLENDFACTORALPHA:
if(extrarg == GL_NONE) extrarg = register_for_arg(WINED3DTA_TFACTOR, gl_info, stage, NULL, -1);
case WINED3DTOP_BLENDTEXTUREALPHA:
if(extrarg == GL_NONE) extrarg = register_for_arg(WINED3DTA_TEXTURE, gl_info, stage, NULL, -1);
case WINED3DTOP_BLENDDIFFUSEALPHA:
if(extrarg == GL_NONE) extrarg = register_for_arg(WINED3DTA_DIFFUSE, gl_info, stage, NULL, -1);
TRACE("glColorFragmentOp3ATI(GL_LERP_ATI, %s, GL_NONE, GL_NONE, %s, GL_ALPHA, GL_NONE, %s, GL_NONE, %s, %s, GL_NONE, %s)\n",
debug_register(dstreg),
debug_register(extrarg),
debug_register(arg1), debug_argmod(argmod1),
debug_register(arg2), debug_argmod(argmod2));
GL_EXTCALL(glColorFragmentOp3ATI(GL_LERP_ATI, dstreg, GL_NONE, GL_NONE,
extrarg, GL_ALPHA, GL_NONE,
arg1, GL_NONE, argmod1,
arg2, GL_NONE, argmod2));
break;
case WINED3DTOP_BLENDTEXTUREALPHAPM:
arg0 = register_for_arg(WINED3DTA_TEXTURE, gl_info, stage, NULL, -1);
TRACE("glColorFragmentOp3ATI(GL_MAD_ATI, %s, GL_NONE, GL_NONE, %s, GL_NONE, %s, %s, GL_ALPHA, GL_COMP_BIT_ATI, %s, GL_NONE, %s)\n",
debug_register(dstreg),
debug_register(arg2), debug_argmod(argmod2),
debug_register(arg0),
debug_register(arg1), debug_argmod(argmod1));
GL_EXTCALL(glColorFragmentOp3ATI(GL_MAD_ATI, dstreg, GL_NONE, GL_NONE,
arg2, GL_NONE, argmod2,
arg0, GL_ALPHA, GL_COMP_BIT_ATI,
arg1, GL_NONE, argmod1));
break;
/* D3DTOP_PREMODULATE ???? */
case WINED3DTOP_MODULATEINVALPHA_ADDCOLOR:
argmodextra = argmod1 & GL_COMP_BIT_ATI ? argmod1 & ~GL_COMP_BIT_ATI : argmod1 | GL_COMP_BIT_ATI;
case WINED3DTOP_MODULATEALPHA_ADDCOLOR:
if(!argmodextra) argmodextra = argmod1;
TRACE("glColorFragmentOp3ATI(GL_MAD_ATI, %s, GL_NONE, GL_NONE, %s, GL_NONE, %s, %s, GL_ALPHA, %s, %s, GL_NONE, %s)\n",
debug_register(dstreg),
debug_register(arg2), debug_argmod(argmod2),
debug_register(arg1), debug_argmod(argmodextra), debug_register(arg1), debug_argmod(arg1));
GL_EXTCALL(glColorFragmentOp3ATI(GL_MAD_ATI, dstreg, GL_NONE, GL_NONE,
arg2, GL_NONE, argmod2,
arg1, GL_ALPHA, argmodextra,
arg1, GL_NONE, argmod1));
break;
case WINED3DTOP_MODULATEINVCOLOR_ADDALPHA:
argmodextra = argmod1 & GL_COMP_BIT_ATI ? argmod1 & ~GL_COMP_BIT_ATI : argmod1 | GL_COMP_BIT_ATI;
case WINED3DTOP_MODULATECOLOR_ADDALPHA:
if(!argmodextra) argmodextra = argmod1;
TRACE("glColorFragmentOp3ATI(GL_MAD_ATI, %s, GL_NONE, GL_NONE, %s, GL_NONE, %s, %s, GL_NONE, %s, %s, GL_ALPHA, %s)\n",
debug_register(dstreg),
debug_register(arg2), debug_argmod(argmod2),
debug_register(arg1), debug_argmod(argmodextra),
debug_register(arg1), debug_argmod(argmod1));
GL_EXTCALL(glColorFragmentOp3ATI(GL_MAD_ATI, dstreg, GL_NONE, GL_NONE,
arg2, GL_NONE, argmod2,
arg1, GL_NONE, argmodextra,
arg1, GL_ALPHA, argmod1));
break;
case WINED3DTOP_DOTPRODUCT3:
TRACE("glColorFragmentOp2ATI(GL_DOT3_ATI, %s, GL_NONE, GL_4X_BIT_ATI, %s, GL_NONE, %s, %s, GL_NONE, %s)\n",
debug_register(dstreg),
debug_register(arg1), debug_argmod(argmod1 | GL_BIAS_BIT_ATI),
debug_register(arg2), debug_argmod(argmod2 | GL_BIAS_BIT_ATI));
GL_EXTCALL(glColorFragmentOp2ATI(GL_DOT3_ATI, dstreg, GL_NONE, GL_4X_BIT_ATI,
arg1, GL_NONE, argmod1 | GL_BIAS_BIT_ATI,
arg2, GL_NONE, argmod2 | GL_BIAS_BIT_ATI));
break;
case WINED3DTOP_MULTIPLYADD:
TRACE("glColorFragmentOp3ATI(GL_MAD_ATI, %s, GL_NONE, GL_NONE, %s, GL_NONE, %s, %s, GL_NONE, %s, %s, GL_NONE, %s)\n",
debug_register(dstreg),
debug_register(arg0), debug_argmod(argmod0),
debug_register(arg2), debug_argmod(argmod2),
debug_register(arg1), debug_argmod(argmod1));
GL_EXTCALL(glColorFragmentOp3ATI(GL_MAD_ATI, dstreg, GL_NONE, GL_NONE,
arg0, GL_NONE, argmod0,
arg2, GL_NONE, argmod2,
arg1, GL_NONE, argmod1));
break;
case WINED3DTOP_LERP:
TRACE("glColorFragmentOp3ATI(GL_LERP_ATI, %s, GL_NONE, GL_NONE, %s, GL_NONE, %s, %s, GL_NONE, %s, %s, GL_NONE, %s)\n",
debug_register(dstreg),
debug_register(arg1), debug_argmod(argmod1),
debug_register(arg2), debug_argmod(argmod2),
debug_register(arg0), debug_argmod(argmod0));
GL_EXTCALL(glColorFragmentOp3ATI(GL_LERP_ATI, dstreg, GL_NONE, GL_NONE,
arg1, GL_NONE, argmod1,
arg2, GL_NONE, argmod2,
arg0, GL_NONE, argmod0));
break;
case WINED3DTOP_BUMPENVMAP:
case WINED3DTOP_BUMPENVMAPLUMINANCE:
/* Those are handled in the first pass of the shader(generation pass 1 and 2) alraedy */
break;
default: FIXME("Unhandled color operation %d on stage %d\n", op[stage].cop, stage);
}
arg0 = register_for_arg(op[stage].aarg0, gl_info, stage, &argmod0, tmparg);
arg1 = register_for_arg(op[stage].aarg1, gl_info, stage, &argmod1, tmparg);
arg2 = register_for_arg(op[stage].aarg2, gl_info, stage, &argmod2, tmparg);
dstmod = GL_NONE;
argmodextra = GL_NONE;
extrarg = GL_NONE;
switch(op[stage].aop) {
case WINED3DTOP_DISABLE:
/* Get the primary color to the output if on stage 0, otherwise leave register 0 untouched */
if(stage == 0) {
TRACE("glAlphaFragmentOp1ATI(GL_MOV_ATI, GL_REG_0_ATI, GL_NONE, GL_PRIMARY_COLOR, GL_NONE, GL_NONE)\n");
GL_EXTCALL(glAlphaFragmentOp1ATI(GL_MOV_ATI, GL_REG_0_ATI, GL_NONE,
GL_PRIMARY_COLOR, GL_NONE, GL_NONE));
}
break;
case WINED3DTOP_SELECTARG2:
arg1 = arg2;
argmod1 = argmod2;
case WINED3DTOP_SELECTARG1:
TRACE("glAlphaFragmentOp1ATI(GL_MOV_ATI, %s, GL_NONE, %s, GL_NONE, %s)\n",
debug_register(dstreg),
debug_register(arg1), debug_argmod(argmod1));
GL_EXTCALL(glAlphaFragmentOp1ATI(GL_MOV_ATI, dstreg, GL_NONE,
arg1, GL_NONE, argmod1));
break;
case WINED3DTOP_MODULATE4X:
if(dstmod == GL_NONE) dstmod = GL_4X_BIT_ATI;
case WINED3DTOP_MODULATE2X:
if(dstmod == GL_NONE) dstmod = GL_2X_BIT_ATI;
case WINED3DTOP_MODULATE:
TRACE("glAlphaFragmentOp2ATI(GL_MUL_ATI, %s, %s, %s, GL_NONE, %s, %s, GL_NONE, %s)\n",
debug_register(dstreg), debug_dstmod(dstmod),
debug_register(arg1), debug_argmod(argmod1),
debug_register(arg2), debug_argmod(argmod2));
GL_EXTCALL(glAlphaFragmentOp2ATI(GL_MUL_ATI, dstreg, dstmod,
arg1, GL_NONE, argmod1,
arg2, GL_NONE, argmod2));
break;
case WINED3DTOP_ADDSIGNED2X:
dstmod = GL_2X_BIT_ATI;
case WINED3DTOP_ADDSIGNED:
argmodextra = GL_BIAS_BIT_ATI;
case WINED3DTOP_ADD:
TRACE("glAlphaFragmentOp2ATI(GL_ADD_ATI, %s, %s, %s, GL_NONE, %s, %s, GL_NONE, %s)\n",
debug_register(dstreg), debug_dstmod(dstmod),
debug_register(arg1), debug_argmod(argmod1),
debug_register(arg2), debug_argmod(argmodextra | argmod2));
GL_EXTCALL(glAlphaFragmentOp2ATI(GL_ADD_ATI, dstreg, dstmod,
arg1, GL_NONE, argmod1,
arg2, GL_NONE, argmodextra | argmod2));
break;
case WINED3DTOP_SUBTRACT:
TRACE("glAlphaFragmentOp2ATI(GL_SUB_ATI, %s, GL_NONE, %s, %s, GL_NONE, %s, %s, GL_NONE, %s)\n",
debug_register(dstreg), debug_dstmod(dstmod),
debug_register(arg1), debug_argmod(argmod1),
debug_register(arg2), debug_argmod(argmod2));
GL_EXTCALL(glAlphaFragmentOp2ATI(GL_SUB_ATI, dstreg, dstmod,
arg1, GL_NONE, argmod1,
arg2, GL_NONE, argmod2));
break;
case WINED3DTOP_ADDSMOOTH:
argmodextra = argmod1 & GL_COMP_BIT_ATI ? argmod1 & ~GL_COMP_BIT_ATI : argmod1 | GL_COMP_BIT_ATI;
TRACE("glAlphaFragmentOp3ATI(GL_MAD_ATI, %s, GL_NONE, %s, GL_NONE, %s, %s, GL_NONE, %s, %s, GL_NONE, %s)\n",
debug_register(dstreg),
debug_register(arg2), debug_argmod(argmod2),
debug_register(arg1), debug_argmod(argmodextra),
debug_register(arg1), debug_argmod(argmod1));
/* Dst = arg1 + * arg2(1 -arg 1)
* = arg2 * (1 - arg1) + arg1
*/
GL_EXTCALL(glAlphaFragmentOp3ATI(GL_MAD_ATI, dstreg, GL_NONE,
arg2, GL_NONE, argmod2,
arg1, GL_NONE, argmodextra,
arg1, GL_NONE, argmod1));
break;
case WINED3DTOP_BLENDCURRENTALPHA:
if(extrarg == GL_NONE) extrarg = register_for_arg(WINED3DTA_CURRENT, gl_info, stage, NULL, -1);
case WINED3DTOP_BLENDFACTORALPHA:
if(extrarg == GL_NONE) extrarg = register_for_arg(WINED3DTA_TFACTOR, gl_info, stage, NULL, -1);
case WINED3DTOP_BLENDTEXTUREALPHA:
if(extrarg == GL_NONE) extrarg = register_for_arg(WINED3DTA_TEXTURE, gl_info, stage, NULL, -1);
case WINED3DTOP_BLENDDIFFUSEALPHA:
if(extrarg == GL_NONE) extrarg = register_for_arg(WINED3DTA_DIFFUSE, gl_info, stage, NULL, -1);
TRACE("glAlphaFragmentOp3ATI(GL_LERP_ATI, %s, GL_NONE, %s, GL_ALPHA, GL_NONE, %s, GL_NONE, %s, %s, GL_NONE, %s)\n",
debug_register(dstreg),
debug_register(extrarg),
debug_register(arg1), debug_argmod(argmod1),
debug_register(arg2), debug_argmod(argmod2));
GL_EXTCALL(glAlphaFragmentOp3ATI(GL_LERP_ATI, dstreg, GL_NONE,
extrarg, GL_ALPHA, GL_NONE,
arg1, GL_NONE, argmod1,
arg2, GL_NONE, argmod2));
break;
case WINED3DTOP_BLENDTEXTUREALPHAPM:
arg0 = register_for_arg(WINED3DTA_TEXTURE, gl_info, stage, NULL, -1);
TRACE("glAlphaFragmentOp3ATI(GL_MAD_ATI, %s, GL_NONE, %s, GL_NONE, %s, %s, GL_ALPHA, GL_COMP_BIT_ATI, %s, GL_NONE, %s)\n",
debug_register(dstreg),
debug_register(arg2), debug_argmod(argmod2),
debug_register(arg0),
debug_register(arg1), debug_argmod(argmod1));
GL_EXTCALL(glAlphaFragmentOp3ATI(GL_MAD_ATI, dstreg, GL_NONE,
arg2, GL_NONE, argmod2,
arg0, GL_ALPHA, GL_COMP_BIT_ATI,
arg1, GL_NONE, argmod1));
break;
/* D3DTOP_PREMODULATE ???? */
case WINED3DTOP_DOTPRODUCT3:
TRACE("glAlphaFragmentOp2ATI(GL_DOT3_ATI, %s, GL_NONE, GL_4X_BIT_ATI, %s, GL_NONE, %s, %s, GL_NONE, %s)\n",
debug_register(dstreg),
debug_register(arg1), debug_argmod(argmod1 | GL_BIAS_BIT_ATI),
debug_register(arg2), debug_argmod(argmod2 | GL_BIAS_BIT_ATI));
GL_EXTCALL(glAlphaFragmentOp2ATI(GL_DOT3_ATI, dstreg, GL_4X_BIT_ATI,
arg1, GL_NONE, argmod1 | GL_BIAS_BIT_ATI,
arg2, GL_NONE, argmod2 | GL_BIAS_BIT_ATI));
break;
case WINED3DTOP_MULTIPLYADD:
TRACE("glAlphaFragmentOp3ATI(GL_MAD_ATI, %s, GL_NONE, %s, GL_NONE, %s, %s, GL_NONE, %s, %s, GL_NONE, %s)\n",
debug_register(dstreg),
debug_register(arg0), debug_argmod(argmod0),
debug_register(arg2), debug_argmod(argmod2),
debug_register(arg1), debug_argmod(argmod1));
GL_EXTCALL(glAlphaFragmentOp3ATI(GL_MAD_ATI, dstreg, GL_NONE,
arg0, GL_NONE, argmod0,
arg2, GL_NONE, argmod2,
arg1, GL_NONE, argmod1));
break;
case WINED3DTOP_LERP:
TRACE("glAlphaFragmentOp3ATI(GL_LERP_ATI, %s, GL_NONE, %s, GL_NONE, %s, %s, GL_NONE, %s, %s, GL_NONE, %s)\n",
debug_register(dstreg),
debug_register(arg1), debug_argmod(argmod1),
debug_register(arg2), debug_argmod(argmod2),
debug_register(arg0), debug_argmod(argmod0));
GL_EXTCALL(glAlphaFragmentOp3ATI(GL_LERP_ATI, dstreg, GL_NONE,
arg1, GL_NONE, argmod1,
arg2, GL_NONE, argmod2,
arg0, GL_NONE, argmod0));
break;
case WINED3DTOP_MODULATEINVALPHA_ADDCOLOR:
case WINED3DTOP_MODULATEALPHA_ADDCOLOR:
case WINED3DTOP_MODULATECOLOR_ADDALPHA:
case WINED3DTOP_MODULATEINVCOLOR_ADDALPHA:
case WINED3DTOP_BUMPENVMAP:
case WINED3DTOP_BUMPENVMAPLUMINANCE:
ERR("Application uses an invalid alpha operation\n");
break;
default: FIXME("Unhandled alpha operation %d on stage %d\n", op[stage].aop, stage);
}
}
TRACE("glEndFragmentShaderATI()\n");
GL_EXTCALL(glEndFragmentShaderATI());
checkGLcall("GL_EXTCALL(glEndFragmentShaderATI())");
return ret;
}
#undef GLINFO_LOCATION
#define GLINFO_LOCATION stateblock->wineD3DDevice->adapter->gl_info
static void set_tex_op_atifs(DWORD state, IWineD3DStateBlockImpl *stateblock, WineD3DContext *context) {
IWineD3DDeviceImpl *This = stateblock->wineD3DDevice;
struct atifs_ffp_desc *desc;
struct texture_stage_op op[MAX_TEXTURES];
struct atifs_private_data *priv = (struct atifs_private_data *) This->shader_priv;
gen_ffp_op(stateblock, op);
desc = (struct atifs_ffp_desc *) find_ffp_shader(&priv->fragment_shaders, op);
if(!desc) {
desc = HeapAlloc(GetProcessHeap(), 0, sizeof(*desc));
if(!desc) {
ERR("Out of memory\n");
return;
}
memcpy(desc->parent.op, op, sizeof(op));
desc->shader = gen_ati_shader(op, &GLINFO_LOCATION);
add_ffp_shader(&priv->fragment_shaders, &desc->parent);
TRACE("Allocated fixed function replacement shader descriptor %p\n", desc);
}
GL_EXTCALL(glBindFragmentShaderATI(desc->shader));
}
static void state_texfactor_atifs(DWORD state, IWineD3DStateBlockImpl *stateblock, WineD3DContext *context) {
float col[4];
D3DCOLORTOGLFLOAT4(stateblock->renderState[WINED3DRS_TEXTUREFACTOR], col);
GL_EXTCALL(glSetFragmentShaderConstantATI(ATI_FFP_CONST_TFACTOR, col));
checkGLcall("glSetFragmentShaderConstantATI(ATI_FFP_CONST_TFACTOR, col)");
}
static void set_bumpmat(DWORD state, IWineD3DStateBlockImpl *stateblock, WineD3DContext *context) {
DWORD stage = (state - STATE_TEXTURESTAGE(0, 0)) / WINED3D_HIGHEST_TEXTURE_STATE;
float mat[2][2];
mat[0][0] = *((float *) &stateblock->textureState[stage][WINED3DTSS_BUMPENVMAT00]);
mat[1][0] = *((float *) &stateblock->textureState[stage][WINED3DTSS_BUMPENVMAT01]);
mat[0][1] = *((float *) &stateblock->textureState[stage][WINED3DTSS_BUMPENVMAT10]);
mat[1][1] = *((float *) &stateblock->textureState[stage][WINED3DTSS_BUMPENVMAT11]);
/* GL_ATI_fragment_shader allows only constants from 0.0 to 1.0, but the bumpmat
* constants can be in any range. While they should stay between [-1.0 and 1.0] because
* Shader Model 1.x pixel shaders are clamped to that range negative values are used occasionally,
* for example by our d3d9 test. So to get negative values scale -1;1 to 0;1 and undo that in the
* shader(it is free). This might potentially reduce precision. However, if the hardware does
* support proper floats it shouldn't, and if it doesn't we can't get anything better anyway
*/
mat[0][0] = (mat[0][0] + 1.0) * 0.5;
mat[1][0] = (mat[1][0] + 1.0) * 0.5;
mat[0][1] = (mat[0][1] + 1.0) * 0.5;
mat[1][1] = (mat[1][1] + 1.0) * 0.5;
GL_EXTCALL(glSetFragmentShaderConstantATI(ATI_FFP_CONST_BUMPMAT(stage), (float *) mat));
checkGLcall("glSetFragmentShaderConstantATI(ATI_FFP_CONST_BUMPMAT(stage), mat)");
/* FIXME: This should go away
* This is currently needed because atifs borrows a pixel shader implementation
* from somewhere else, but consumes bump map matrix change events. The other pixel
* shader implementation may need notification about the change to update the texbem
* constants. Once ATIFS supports real shaders on its own, and GLSL/ARB have a replacement
* pipeline this call can go away
*
* FIXME2: Even considering this workaround calling FFPStateTable directly isn't nice
* as well. Better would be to call the model's table we inherit from, but currently
* it is always the FFP table, and as soon as this changes we can remove the call anyway
*/
FFPStateTable[state].apply(state, stateblock, context);
}
#undef GLINFO_LOCATION
/* our state table. Borrows lots of stuff from the base implementation */
struct StateEntry ATIFSStateTable[STATE_HIGHEST + 1];
static void init_state_table() {
unsigned int i;
const DWORD rep = STATE_TEXTURESTAGE(0, WINED3DTSS_COLOROP);
memcpy(ATIFSStateTable, arb_program_shader_backend.StateTable, sizeof(ATIFSStateTable));
for(i = 0; i < MAX_TEXTURES; i++) {
ATIFSStateTable[STATE_TEXTURESTAGE(i, WINED3DTSS_COLOROP)].apply = set_tex_op_atifs;
ATIFSStateTable[STATE_TEXTURESTAGE(i, WINED3DTSS_COLOROP)].representative = rep;
ATIFSStateTable[STATE_TEXTURESTAGE(i, WINED3DTSS_COLORARG1)].apply = set_tex_op_atifs;
ATIFSStateTable[STATE_TEXTURESTAGE(i, WINED3DTSS_COLORARG1)].representative = rep;
ATIFSStateTable[STATE_TEXTURESTAGE(i, WINED3DTSS_COLORARG2)].apply = set_tex_op_atifs;
ATIFSStateTable[STATE_TEXTURESTAGE(i, WINED3DTSS_COLORARG2)].representative = rep;
ATIFSStateTable[STATE_TEXTURESTAGE(i, WINED3DTSS_COLORARG0)].apply = set_tex_op_atifs;
ATIFSStateTable[STATE_TEXTURESTAGE(i, WINED3DTSS_COLORARG0)].representative = rep;
ATIFSStateTable[STATE_TEXTURESTAGE(i, WINED3DTSS_ALPHAOP)].apply = set_tex_op_atifs;
ATIFSStateTable[STATE_TEXTURESTAGE(i, WINED3DTSS_ALPHAOP)].representative = rep;
ATIFSStateTable[STATE_TEXTURESTAGE(i, WINED3DTSS_ALPHAARG1)].apply = set_tex_op_atifs;
ATIFSStateTable[STATE_TEXTURESTAGE(i, WINED3DTSS_ALPHAARG1)].representative = rep;
ATIFSStateTable[STATE_TEXTURESTAGE(i, WINED3DTSS_ALPHAARG2)].apply = set_tex_op_atifs;
ATIFSStateTable[STATE_TEXTURESTAGE(i, WINED3DTSS_ALPHAARG2)].representative = rep;
ATIFSStateTable[STATE_TEXTURESTAGE(i, WINED3DTSS_ALPHAARG0)].apply = set_tex_op_atifs;
ATIFSStateTable[STATE_TEXTURESTAGE(i, WINED3DTSS_ALPHAARG0)].representative = rep;
ATIFSStateTable[STATE_TEXTURESTAGE(i, WINED3DTSS_RESULTARG)].apply = set_tex_op_atifs;
ATIFSStateTable[STATE_TEXTURESTAGE(i, WINED3DTSS_RESULTARG)].representative = rep;
ATIFSStateTable[STATE_TEXTURESTAGE(i, WINED3DTSS_BUMPENVMAT00)].apply = set_bumpmat;
ATIFSStateTable[STATE_TEXTURESTAGE(i, WINED3DTSS_BUMPENVMAT01)].apply = set_bumpmat;
ATIFSStateTable[STATE_TEXTURESTAGE(i, WINED3DTSS_BUMPENVMAT10)].apply = set_bumpmat;
ATIFSStateTable[STATE_TEXTURESTAGE(i, WINED3DTSS_BUMPENVMAT11)].apply = set_bumpmat;
}
ATIFSStateTable[STATE_RENDER(WINED3DRS_TEXTUREFACTOR)].apply = state_texfactor_atifs;
ATIFSStateTable[STATE_RENDER(WINED3DRS_TEXTUREFACTOR)].representative = STATE_RENDER(WINED3DRS_TEXTUREFACTOR);
}
/* GL_ATI_fragment_shader backend.It borrows a lot from a the
* ARB shader backend, currently the whole vertex processing
* code. This code would also forward pixel shaders, but if
* GL_ARB_fragment_program is supported, the atifs shader backend
* is not used.
*/
static void shader_atifs_select(IWineD3DDevice *iface, BOOL usePS, BOOL useVS) {
arb_program_shader_backend.shader_select(iface, usePS, useVS);
}
static void shader_atifs_select_depth_blt(IWineD3DDevice *iface) {
arb_program_shader_backend.shader_select_depth_blt(iface);
}
static void shader_atifs_destroy_depth_blt(IWineD3DDevice *iface) {
arb_program_shader_backend.shader_destroy_depth_blt(iface);
}
static void shader_atifs_load_constants(IWineD3DDevice *iface, char usePS, char useVS) {
arb_program_shader_backend.shader_load_constants(iface, usePS, useVS);
}
static void shader_atifs_cleanup(IWineD3DDevice *iface) {
arb_program_shader_backend.shader_cleanup(iface);
}
static void shader_atifs_color_correction(SHADER_OPCODE_ARG* arg) {
arb_program_shader_backend.shader_color_correction(arg);
}
static void shader_atifs_destroy(IWineD3DBaseShader *iface) {
arb_program_shader_backend.shader_destroy(iface);
}
static HRESULT shader_atifs_alloc(IWineD3DDevice *iface) {
IWineD3DDeviceImpl *This = (IWineD3DDeviceImpl *) iface;
HRESULT hr;
struct atifs_private_data *priv;
hr = arb_program_shader_backend.shader_alloc_private(iface);
if(FAILED(hr)) return hr;
This->shader_priv = HeapReAlloc(GetProcessHeap(), 0, This->shader_priv,
sizeof(struct atifs_private_data));
priv = (struct atifs_private_data *) This->shader_priv;
list_init(&priv->fragment_shaders);
return WINED3D_OK;
}
#define GLINFO_LOCATION This->adapter->gl_info
static void shader_atifs_free(IWineD3DDevice *iface) {
IWineD3DDeviceImpl *This = (IWineD3DDeviceImpl *) iface;
struct atifs_private_data *priv = (struct atifs_private_data *) This->shader_priv;
struct ffp_desc *entry, *entry2;
struct atifs_ffp_desc *entry_ati;
ENTER_GL();
LIST_FOR_EACH_ENTRY_SAFE(entry, entry2, &priv->fragment_shaders, struct ffp_desc, entry) {
entry_ati = (struct atifs_ffp_desc *) entry;
GL_EXTCALL(glDeleteFragmentShaderATI(entry_ati->shader));
checkGLcall("glDeleteFragmentShaderATI(entry->shader)");
list_remove(&entry->entry);
HeapFree(GetProcessHeap(), 0, entry);
}
LEAVE_GL();
/* Not actually needed, but revert what we've done before */
This->shader_priv = HeapReAlloc(GetProcessHeap(), 0, This->shader_priv,
sizeof(struct shader_arb_priv));
arb_program_shader_backend.shader_free_private(iface);
}
#undef GLINFO_LOCATION
static BOOL shader_atifs_dirty_const(IWineD3DDevice *iface) {
return arb_program_shader_backend.shader_dirtifyable_constants(iface);
}
static void shader_atifs_load_init(void) {
arb_program_shader_backend.shader_dll_load_init();
init_state_table();
}
static void shader_atifs_get_caps(WINED3DDEVTYPE devtype, WineD3D_GL_Info *gl_info, struct shader_caps *caps) {
arb_program_shader_backend.shader_get_caps(devtype, gl_info, caps);
caps->TextureOpCaps = WINED3DTEXOPCAPS_DISABLE |
WINED3DTEXOPCAPS_SELECTARG1 |
WINED3DTEXOPCAPS_SELECTARG2 |
WINED3DTEXOPCAPS_MODULATE4X |
WINED3DTEXOPCAPS_MODULATE2X |
WINED3DTEXOPCAPS_MODULATE |
WINED3DTEXOPCAPS_ADDSIGNED2X |
WINED3DTEXOPCAPS_ADDSIGNED |
WINED3DTEXOPCAPS_ADD |
WINED3DTEXOPCAPS_SUBTRACT |
WINED3DTEXOPCAPS_ADDSMOOTH |
WINED3DTEXOPCAPS_BLENDCURRENTALPHA |
WINED3DTEXOPCAPS_BLENDFACTORALPHA |
WINED3DTEXOPCAPS_BLENDTEXTUREALPHA |
WINED3DTEXOPCAPS_BLENDDIFFUSEALPHA |
WINED3DTEXOPCAPS_BLENDTEXTUREALPHAPM |
WINED3DTEXOPCAPS_MODULATEALPHA_ADDCOLOR |
WINED3DTEXOPCAPS_MODULATECOLOR_ADDALPHA |
WINED3DTEXOPCAPS_MODULATEINVCOLOR_ADDALPHA |
WINED3DTEXOPCAPS_MODULATEINVALPHA_ADDCOLOR |
WINED3DTEXOPCAPS_DOTPRODUCT3 |
WINED3DTEXOPCAPS_MULTIPLYADD |
WINED3DTEXOPCAPS_LERP |
WINED3DTEXOPCAPS_BUMPENVMAP;
/* TODO: Implement WINED3DTEXOPCAPS_BUMPENVMAPLUMINANCE
and WINED3DTEXOPCAPS_PREMODULATE */
/* GL_ATI_fragment_shader only supports up to 6 textures, which was the limit on r200 cards
* which this extension is exclusively focused on(later cards have GL_ARB_fragment_program).
* If the current card has more than 8 fixed function textures in OpenGL's regular fixed
* function pipeline then the ATI_fragment_shader backend imposes a stricter limit. This
* shouldn't be too hard since Nvidia cards have a limit of 4 textures with the default ffp
* pipeline, and almost all games are happy with that. We can however support up to 8
* texture stages because we have a 2nd pass limit of 8 instructions, and per stage we use
* only 1 instruction.
*
* The proper fix for this is not to use GL_ATI_fragment_shader on cards newer than the
* r200 series and use an ARB or GLSL shader instead
*/
if(caps->MaxSimultaneousTextures > 6) {
WARN("OpenGL fixed function supports %d simultaneous textures,\n", caps->MaxSimultaneousTextures);
WARN("but GL_ATI_fragment_shader limits this to 6\n");
caps->MaxSimultaneousTextures = 6;
}
caps->PrimitiveMiscCaps |= WINED3DPMISCCAPS_TSSARGTEMP;
}
static void shader_atifs_generate_pshader(IWineD3DPixelShader *iface, SHADER_BUFFER *buffer) {
ERR("Should not get here\n");
}
static void shader_atifs_generate_vshader(IWineD3DVertexShader *iface, SHADER_BUFFER *buffer) {
arb_program_shader_backend.shader_generate_vshader(iface, buffer);
}
const shader_backend_t atifs_shader_backend = {
shader_atifs_select,
shader_atifs_select_depth_blt,
shader_atifs_destroy_depth_blt,
shader_atifs_load_constants,
shader_atifs_cleanup,
shader_atifs_color_correction,
shader_atifs_destroy,
shader_atifs_alloc,
shader_atifs_free,
shader_atifs_dirty_const,
shader_atifs_generate_pshader,
shader_atifs_generate_vshader,
shader_atifs_get_caps,
shader_atifs_load_init,
ATIFSStateTable
};