2 * GLSL pixel and vertex shader implementation
4 * Copyright 2006 Jason Green
5 * Copyright 2006-2007 Henri Verbeet
7 * This library is free software; you can redistribute it and/or
8 * modify it under the terms of the GNU Lesser General Public
9 * License as published by the Free Software Foundation; either
10 * version 2.1 of the License, or (at your option) any later version.
12 * This library is distributed in the hope that it will be useful,
13 * but WITHOUT ANY WARRANTY; without even the implied warranty of
14 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
15 * Lesser General Public License for more details.
17 * You should have received a copy of the GNU Lesser General Public
18 * License along with this library; if not, write to the Free Software
19 * Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301, USA
23 * D3D shader asm has swizzles on source parameters, and write masks for
24 * destination parameters. GLSL uses swizzles for both. The result of this is
25 * that for example "mov dst.xw, src.zyxw" becomes "dst.xw = src.zw" in GLSL.
26 * Ie, to generate a proper GLSL source swizzle, we need to take the D3D write
27 * mask for the destination parameter into account.
32 #include "wined3d_private.h"
34 WINE_DEFAULT_DEBUG_CHANNEL(d3d_shader
);
36 #define GLINFO_LOCATION (*gl_info)
41 } glsl_sample_function_t
;
43 /** Prints the GLSL info log which will contain error messages if they exist */
44 void print_glsl_info_log(WineD3D_GL_Info
*gl_info
, GLhandleARB obj
) {
46 int infologLength
= 0;
49 GL_EXTCALL(glGetObjectParameterivARB(obj
,
50 GL_OBJECT_INFO_LOG_LENGTH_ARB
,
53 /* A size of 1 is just a null-terminated string, so the log should be bigger than
54 * that if there are errors. */
55 if (infologLength
> 1)
57 infoLog
= (char *)HeapAlloc(GetProcessHeap(), 0, infologLength
);
58 GL_EXTCALL(glGetInfoLogARB(obj
, infologLength
, NULL
, infoLog
));
59 FIXME("Error received from GLSL shader #%u: %s\n", obj
, debugstr_a(infoLog
));
60 HeapFree(GetProcessHeap(), 0, infoLog
);
65 * Loads (pixel shader) samplers
67 void shader_glsl_load_psamplers(
68 WineD3D_GL_Info
*gl_info
,
69 IWineD3DStateBlock
* iface
) {
71 IWineD3DStateBlockImpl
* stateBlock
= (IWineD3DStateBlockImpl
*) iface
;
72 GLhandleARB programId
= stateBlock
->glsl_program
->programId
;
75 char sampler_name
[20];
77 for (i
=0; i
< GL_LIMITS(samplers
); ++i
) {
78 if (stateBlock
->textures
[i
] != NULL
) {
79 snprintf(sampler_name
, sizeof(sampler_name
), "Psampler%d", i
);
80 name_loc
= GL_EXTCALL(glGetUniformLocationARB(programId
, sampler_name
));
82 TRACE("Loading %s for texture %d\n", sampler_name
, i
);
83 GL_EXTCALL(glUniform1iARB(name_loc
, i
));
84 checkGLcall("glUniform1iARB");
91 * Loads floating point constants (aka uniforms) into the currently set GLSL program.
92 * When constant_list == NULL, it will load all the constants.
94 static void shader_glsl_load_constantsF(IWineD3DBaseShaderImpl
* This
, WineD3D_GL_Info
*gl_info
,
95 unsigned int max_constants
, float* constants
, GLhandleARB
*constant_locations
,
96 struct list
*constant_list
) {
97 local_constant
* lconst
;
101 if (!constant_list
) {
102 if (TRACE_ON(d3d_shader
)) {
103 for (i
= 0; i
< max_constants
; ++i
) {
104 tmp_loc
= constant_locations
[i
];
106 TRACE("Loading constants %i: %f, %f, %f, %f\n", i
,
107 constants
[i
* 4 + 0], constants
[i
* 4 + 1],
108 constants
[i
* 4 + 2], constants
[i
* 4 + 3]);
112 for (i
= 0; i
< max_constants
; ++i
) {
113 tmp_loc
= constant_locations
[i
];
115 /* We found this uniform name in the program - go ahead and send the data */
116 GL_EXTCALL(glUniform4fvARB(tmp_loc
, 1, constants
+ (i
* 4)));
119 checkGLcall("glUniform4fvARB()");
121 constant_entry
*constant
;
122 if (TRACE_ON(d3d_shader
)) {
123 LIST_FOR_EACH_ENTRY(constant
, constant_list
, constant_entry
, entry
) {
125 tmp_loc
= constant_locations
[i
];
127 TRACE("Loading constants %i: %f, %f, %f, %f\n", i
,
128 constants
[i
* 4 + 0], constants
[i
* 4 + 1],
129 constants
[i
* 4 + 2], constants
[i
* 4 + 3]);
133 LIST_FOR_EACH_ENTRY(constant
, constant_list
, constant_entry
, entry
) {
135 tmp_loc
= constant_locations
[i
];
137 /* We found this uniform name in the program - go ahead and send the data */
138 GL_EXTCALL(glUniform4fvARB(tmp_loc
, 1, constants
+ (i
* 4)));
141 checkGLcall("glUniform4fvARB()");
144 /* Load immediate constants */
145 if (TRACE_ON(d3d_shader
)) {
146 LIST_FOR_EACH_ENTRY(lconst
, &This
->baseShader
.constantsF
, local_constant
, entry
) {
147 tmp_loc
= constant_locations
[lconst
->idx
];
149 GLfloat
* values
= (GLfloat
*)lconst
->value
;
150 TRACE("Loading local constants %i: %f, %f, %f, %f\n", lconst
->idx
,
151 values
[0], values
[1], values
[2], values
[3]);
155 LIST_FOR_EACH_ENTRY(lconst
, &This
->baseShader
.constantsF
, local_constant
, entry
) {
156 tmp_loc
= constant_locations
[lconst
->idx
];
158 /* We found this uniform name in the program - go ahead and send the data */
159 GL_EXTCALL(glUniform4fvARB(tmp_loc
, 1, (GLfloat
*)lconst
->value
));
162 checkGLcall("glUniform4fvARB()");
166 * Loads integer constants (aka uniforms) into the currently set GLSL program.
167 * When @constants_set == NULL, it will load all the constants.
169 void shader_glsl_load_constantsI(
170 IWineD3DBaseShaderImpl
* This
,
171 WineD3D_GL_Info
*gl_info
,
172 GLhandleARB programId
,
173 unsigned max_constants
,
175 BOOL
* constants_set
) {
180 char is_pshader
= shader_is_pshader_version(This
->baseShader
.hex_version
);
181 const char* prefix
= is_pshader
? "PI":"VI";
184 for (i
=0; i
<max_constants
; ++i
) {
185 if (NULL
== constants_set
|| constants_set
[i
]) {
187 TRACE("Loading constants %i: %i, %i, %i, %i\n",
188 i
, constants
[i
*4], constants
[i
*4+1], constants
[i
*4+2], constants
[i
*4+3]);
190 /* TODO: Benchmark and see if it would be beneficial to store the
191 * locations of the constants to avoid looking up each time */
192 snprintf(tmp_name
, sizeof(tmp_name
), "%s[%i]", prefix
, i
);
193 tmp_loc
= GL_EXTCALL(glGetUniformLocationARB(programId
, tmp_name
));
195 /* We found this uniform name in the program - go ahead and send the data */
196 GL_EXTCALL(glUniform4ivARB(tmp_loc
, 1, &constants
[i
*4]));
197 checkGLcall("glUniform4ivARB");
202 /* Load immediate constants */
203 ptr
= list_head(&This
->baseShader
.constantsI
);
205 local_constant
* lconst
= LIST_ENTRY(ptr
, struct local_constant
, entry
);
206 unsigned int idx
= lconst
->idx
;
207 GLint
* values
= (GLint
*) lconst
->value
;
209 TRACE("Loading local constants %i: %i, %i, %i, %i\n", idx
,
210 values
[0], values
[1], values
[2], values
[3]);
212 snprintf(tmp_name
, sizeof(tmp_name
), "%s[%i]", prefix
, idx
);
213 tmp_loc
= GL_EXTCALL(glGetUniformLocationARB(programId
, tmp_name
));
215 /* We found this uniform name in the program - go ahead and send the data */
216 GL_EXTCALL(glUniform4ivARB(tmp_loc
, 1, values
));
217 checkGLcall("glUniform4ivARB");
219 ptr
= list_next(&This
->baseShader
.constantsI
, ptr
);
224 * Loads boolean constants (aka uniforms) into the currently set GLSL program.
225 * When @constants_set == NULL, it will load all the constants.
227 void shader_glsl_load_constantsB(
228 IWineD3DBaseShaderImpl
* This
,
229 WineD3D_GL_Info
*gl_info
,
230 GLhandleARB programId
,
231 unsigned max_constants
,
233 BOOL
* constants_set
) {
238 char is_pshader
= shader_is_pshader_version(This
->baseShader
.hex_version
);
239 const char* prefix
= is_pshader
? "PB":"VB";
242 for (i
=0; i
<max_constants
; ++i
) {
243 if (NULL
== constants_set
|| constants_set
[i
]) {
245 TRACE("Loading constants %i: %i;\n", i
, constants
[i
*4]);
247 /* TODO: Benchmark and see if it would be beneficial to store the
248 * locations of the constants to avoid looking up each time */
249 snprintf(tmp_name
, sizeof(tmp_name
), "%s[%i]", prefix
, i
);
250 tmp_loc
= GL_EXTCALL(glGetUniformLocationARB(programId
, tmp_name
));
252 /* We found this uniform name in the program - go ahead and send the data */
253 GL_EXTCALL(glUniform1ivARB(tmp_loc
, 1, &constants
[i
*4]));
254 checkGLcall("glUniform1ivARB");
259 /* Load immediate constants */
260 ptr
= list_head(&This
->baseShader
.constantsB
);
262 local_constant
* lconst
= LIST_ENTRY(ptr
, struct local_constant
, entry
);
263 unsigned int idx
= lconst
->idx
;
264 GLint
* values
= (GLint
*) lconst
->value
;
266 TRACE("Loading local constants %i: %i\n", idx
, values
[0]);
268 snprintf(tmp_name
, sizeof(tmp_name
), "%s[%i]", prefix
, idx
);
269 tmp_loc
= GL_EXTCALL(glGetUniformLocationARB(programId
, tmp_name
));
271 /* We found this uniform name in the program - go ahead and send the data */
272 GL_EXTCALL(glUniform1ivARB(tmp_loc
, 1, values
));
273 checkGLcall("glUniform1ivARB");
275 ptr
= list_next(&This
->baseShader
.constantsB
, ptr
);
282 * Loads the app-supplied constants into the currently set GLSL program.
284 void shader_glsl_load_constants(
285 IWineD3DDevice
* device
,
287 char useVertexShader
) {
289 IWineD3DDeviceImpl
* deviceImpl
= (IWineD3DDeviceImpl
*) device
;
290 IWineD3DStateBlockImpl
* stateBlock
= deviceImpl
->stateBlock
;
291 WineD3D_GL_Info
*gl_info
= &((IWineD3DImpl
*) deviceImpl
->wineD3D
)->gl_info
;
293 GLhandleARB
*constant_locations
;
294 struct list
*constant_list
;
295 GLhandleARB programId
;
297 if (!stateBlock
->glsl_program
) {
298 /* No GLSL program set - nothing to do. */
301 programId
= stateBlock
->glsl_program
->programId
;
303 if (useVertexShader
) {
304 IWineD3DBaseShaderImpl
* vshader
= (IWineD3DBaseShaderImpl
*) stateBlock
->vertexShader
;
305 IWineD3DVertexShaderImpl
* vshader_impl
= (IWineD3DVertexShaderImpl
*) vshader
;
308 IWineD3DVertexDeclarationImpl
* vertexDeclaration
=
309 (IWineD3DVertexDeclarationImpl
*) vshader_impl
->vertexDeclaration
;
311 constant_locations
= stateBlock
->glsl_program
->vuniformF_locations
;
312 constant_list
= &stateBlock
->set_vconstantsF
;
314 if (NULL
!= vertexDeclaration
&& NULL
!= vertexDeclaration
->constants
) {
315 /* Load DirectX 8 float constants/uniforms for vertex shader */
316 shader_glsl_load_constantsF(vshader
, gl_info
, GL_LIMITS(vshader_constantsF
),
317 vertexDeclaration
->constants
, constant_locations
, NULL
);
320 /* Load DirectX 9 float constants/uniforms for vertex shader */
321 shader_glsl_load_constantsF(vshader
, gl_info
, GL_LIMITS(vshader_constantsF
),
322 stateBlock
->vertexShaderConstantF
, constant_locations
, constant_list
);
324 /* Load DirectX 9 integer constants/uniforms for vertex shader */
325 shader_glsl_load_constantsI(vshader
, gl_info
, programId
, MAX_CONST_I
,
326 stateBlock
->vertexShaderConstantI
,
327 stateBlock
->set
.vertexShaderConstantsI
);
329 /* Load DirectX 9 boolean constants/uniforms for vertex shader */
330 shader_glsl_load_constantsB(vshader
, gl_info
, programId
, MAX_CONST_B
,
331 stateBlock
->vertexShaderConstantB
,
332 stateBlock
->set
.vertexShaderConstantsB
);
334 /* Upload the position fixup params */
335 pos
= GL_EXTCALL(glGetUniformLocationARB(programId
, "posFixup"));
336 checkGLcall("glGetUniformLocationARB");
337 GL_EXTCALL(glUniform4fvARB(pos
, 1, &deviceImpl
->posFixup
[0]));
338 checkGLcall("glUniform4fvARB");
341 if (usePixelShader
) {
343 IWineD3DBaseShaderImpl
* pshader
= (IWineD3DBaseShaderImpl
*) stateBlock
->pixelShader
;
345 constant_locations
= stateBlock
->glsl_program
->puniformF_locations
;
346 constant_list
= &stateBlock
->set_pconstantsF
;
348 /* Load pixel shader samplers */
349 shader_glsl_load_psamplers(gl_info
, (IWineD3DStateBlock
*) stateBlock
);
351 /* Load DirectX 9 float constants/uniforms for pixel shader */
352 shader_glsl_load_constantsF(pshader
, gl_info
, GL_LIMITS(pshader_constantsF
),
353 stateBlock
->pixelShaderConstantF
, constant_locations
, constant_list
);
355 /* Load DirectX 9 integer constants/uniforms for pixel shader */
356 shader_glsl_load_constantsI(pshader
, gl_info
, programId
, MAX_CONST_I
,
357 stateBlock
->pixelShaderConstantI
,
358 stateBlock
->set
.pixelShaderConstantsI
);
360 /* Load DirectX 9 boolean constants/uniforms for pixel shader */
361 shader_glsl_load_constantsB(pshader
, gl_info
, programId
, MAX_CONST_B
,
362 stateBlock
->pixelShaderConstantB
,
363 stateBlock
->set
.pixelShaderConstantsB
);
367 /** Generate the variable & register declarations for the GLSL output target */
368 void shader_generate_glsl_declarations(
369 IWineD3DBaseShader
*iface
,
370 shader_reg_maps
* reg_maps
,
371 SHADER_BUFFER
* buffer
,
372 WineD3D_GL_Info
* gl_info
) {
374 IWineD3DBaseShaderImpl
* This
= (IWineD3DBaseShaderImpl
*) iface
;
377 /* There are some minor differences between pixel and vertex shaders */
378 char pshader
= shader_is_pshader_version(This
->baseShader
.hex_version
);
379 char prefix
= pshader
? 'P' : 'V';
381 /* Prototype the subroutines */
382 for (i
= 0; i
< This
->baseShader
.limits
.label
; i
++) {
383 if (reg_maps
->labels
[i
])
384 shader_addline(buffer
, "void subroutine%lu();\n", i
);
387 /* Declare the constants (aka uniforms) */
388 if (This
->baseShader
.limits
.constant_float
> 0) {
389 unsigned max_constantsF
= min(This
->baseShader
.limits
.constant_float
,
390 (pshader
? GL_LIMITS(pshader_constantsF
) : GL_LIMITS(vshader_constantsF
)));
391 shader_addline(buffer
, "uniform vec4 %cC[%u];\n", prefix
, max_constantsF
);
394 if (This
->baseShader
.limits
.constant_int
> 0)
395 shader_addline(buffer
, "uniform ivec4 %cI[%u];\n", prefix
, This
->baseShader
.limits
.constant_int
);
397 if (This
->baseShader
.limits
.constant_bool
> 0)
398 shader_addline(buffer
, "uniform bool %cB[%u];\n", prefix
, This
->baseShader
.limits
.constant_bool
);
401 shader_addline(buffer
, "uniform vec4 posFixup;\n");
403 /* Declare texture samplers */
404 for (i
= 0; i
< This
->baseShader
.limits
.sampler
; i
++) {
405 if (reg_maps
->samplers
[i
]) {
407 DWORD stype
= reg_maps
->samplers
[i
] & WINED3DSP_TEXTURETYPE_MASK
;
411 shader_addline(buffer
, "uniform sampler1D %csampler%lu;\n", prefix
, i
);
414 shader_addline(buffer
, "uniform sampler2D %csampler%lu;\n", prefix
, i
);
416 case WINED3DSTT_CUBE
:
417 shader_addline(buffer
, "uniform samplerCube %csampler%lu;\n", prefix
, i
);
419 case WINED3DSTT_VOLUME
:
420 shader_addline(buffer
, "uniform sampler3D %csampler%lu;\n", prefix
, i
);
423 shader_addline(buffer
, "uniform unsupported_sampler %csampler%lu;\n", prefix
, i
);
424 FIXME("Unrecognized sampler type: %#x\n", stype
);
430 /* Declare address variables */
431 for (i
= 0; i
< This
->baseShader
.limits
.address
; i
++) {
432 if (reg_maps
->address
[i
])
433 shader_addline(buffer
, "ivec4 A%d;\n", i
);
436 /* Declare texture coordinate temporaries and initialize them */
437 for (i
= 0; i
< This
->baseShader
.limits
.texcoord
; i
++) {
438 if (reg_maps
->texcoord
[i
])
439 shader_addline(buffer
, "vec4 T%lu = gl_TexCoord[%lu];\n", i
, i
);
442 /* Declare input register temporaries */
443 for (i
=0; i
< This
->baseShader
.limits
.packed_input
; i
++) {
444 if (reg_maps
->packed_input
[i
])
445 shader_addline(buffer
, "vec4 IN%lu;\n", i
);
448 /* Declare output register temporaries */
449 for (i
= 0; i
< This
->baseShader
.limits
.packed_output
; i
++) {
450 if (reg_maps
->packed_output
[i
])
451 shader_addline(buffer
, "vec4 OUT%lu;\n", i
);
454 /* Declare temporary variables */
455 for(i
= 0; i
< This
->baseShader
.limits
.temporary
; i
++) {
456 if (reg_maps
->temporary
[i
])
457 shader_addline(buffer
, "vec4 R%lu;\n", i
);
460 /* Declare attributes */
461 for (i
= 0; i
< This
->baseShader
.limits
.attributes
; i
++) {
462 if (reg_maps
->attributes
[i
])
463 shader_addline(buffer
, "attribute vec4 attrib%i;\n", i
);
466 /* Declare loop register aL */
467 if (reg_maps
->loop
) {
468 shader_addline(buffer
, "int aL;\n");
469 shader_addline(buffer
, "int tmpInt;\n");
472 /* Temporary variables for matrix operations */
473 shader_addline(buffer
, "vec4 tmp0;\n");
474 shader_addline(buffer
, "vec4 tmp1;\n");
476 /* Start the main program */
477 shader_addline(buffer
, "void main() {\n");
480 /*****************************************************************************
481 * Functions to generate GLSL strings from DirectX Shader bytecode begin here.
483 * For more information, see http://wiki.winehq.org/DirectX-Shaders
484 ****************************************************************************/
487 static void shader_glsl_add_src_param(SHADER_OPCODE_ARG
* arg
, const DWORD param
,
488 const DWORD addr_token
, DWORD mask
, char *reg_name
, char *swizzle
, char *out_str
);
490 /* FIXME: This should be removed */
491 static void shader_glsl_add_src_param_old(SHADER_OPCODE_ARG
* arg
, const DWORD param
,
492 const DWORD addr_token
, char *reg_name
, char *swizzle
, char *out_str
) {
493 shader_glsl_add_src_param(arg
, param
, addr_token
, 0, reg_name
, swizzle
, out_str
);
496 /** Used for opcode modifiers - They multiply the result by the specified amount */
497 static const char * const shift_glsl_tab
[] = {
499 "2.0 * ", /* 1 (x2) */
500 "4.0 * ", /* 2 (x4) */
501 "8.0 * ", /* 3 (x8) */
502 "16.0 * ", /* 4 (x16) */
503 "32.0 * ", /* 5 (x32) */
510 "0.0625 * ", /* 12 (d16) */
511 "0.125 * ", /* 13 (d8) */
512 "0.25 * ", /* 14 (d4) */
513 "0.5 * " /* 15 (d2) */
516 /* Generate a GLSL parameter that does the input modifier computation and return the input register/mask to use */
517 static void shader_glsl_gen_modifier (
520 const char *in_regswizzle
,
525 if (instr
== WINED3DSIO_TEXKILL
)
528 switch (instr
& WINED3DSP_SRCMOD_MASK
) {
529 case WINED3DSPSM_NONE
:
530 sprintf(out_str
, "%s%s", in_reg
, in_regswizzle
);
532 case WINED3DSPSM_NEG
:
533 sprintf(out_str
, "-%s%s", in_reg
, in_regswizzle
);
535 case WINED3DSPSM_NOT
:
536 sprintf(out_str
, "!%s%s", in_reg
, in_regswizzle
);
538 case WINED3DSPSM_BIAS
:
539 sprintf(out_str
, "(%s%s - vec4(0.5)%s)", in_reg
, in_regswizzle
, in_regswizzle
);
541 case WINED3DSPSM_BIASNEG
:
542 sprintf(out_str
, "-(%s%s - vec4(0.5)%s)", in_reg
, in_regswizzle
, in_regswizzle
);
544 case WINED3DSPSM_SIGN
:
545 sprintf(out_str
, "(2.0 * (%s%s - 0.5))", in_reg
, in_regswizzle
);
547 case WINED3DSPSM_SIGNNEG
:
548 sprintf(out_str
, "-(2.0 * (%s%s - 0.5))", in_reg
, in_regswizzle
);
550 case WINED3DSPSM_COMP
:
551 sprintf(out_str
, "(1.0 - %s%s)", in_reg
, in_regswizzle
);
554 sprintf(out_str
, "(2.0 * %s%s)", in_reg
, in_regswizzle
);
556 case WINED3DSPSM_X2NEG
:
557 sprintf(out_str
, "-(2.0 * %s%s)", in_reg
, in_regswizzle
);
559 case WINED3DSPSM_DZ
: /* reg1_db = { reg1.r/b, reg1.g/b, ...} The g & a components are undefined, so we'll leave them alone */
560 sprintf(out_str
, "vec4(%s.r / %s.b, %s.g / %s.b, %s.b, %s.a)", in_reg
, in_reg
, in_reg
, in_reg
, in_reg
, in_reg
);
563 sprintf(out_str
, "vec4(%s.r / %s.a, %s.g / %s.a, %s.b, %s.a)", in_reg
, in_reg
, in_reg
, in_reg
, in_reg
, in_reg
);
565 case WINED3DSPSM_ABS
:
566 sprintf(out_str
, "abs(%s%s)", in_reg
, in_regswizzle
);
568 case WINED3DSPSM_ABSNEG
:
569 sprintf(out_str
, "-abs(%s%s)", in_reg
, in_regswizzle
);
572 FIXME("Unhandled modifier %u\n", (instr
& WINED3DSP_SRCMOD_MASK
));
573 sprintf(out_str
, "%s%s", in_reg
, in_regswizzle
);
577 /** Writes the GLSL variable name that corresponds to the register that the
578 * DX opcode parameter is trying to access */
579 static void shader_glsl_get_register_name(
581 const DWORD addr_token
,
584 SHADER_OPCODE_ARG
* arg
) {
586 /* oPos, oFog and oPts in D3D */
587 static const char * const hwrastout_reg_names
[] = { "gl_Position", "gl_FogFragCoord", "gl_PointSize" };
589 DWORD reg
= param
& WINED3DSP_REGNUM_MASK
;
590 DWORD regtype
= shader_get_regtype(param
);
591 IWineD3DBaseShaderImpl
* This
= (IWineD3DBaseShaderImpl
*) arg
->shader
;
592 IWineD3DDeviceImpl
* deviceImpl
= (IWineD3DDeviceImpl
*) This
->baseShader
.device
;
593 WineD3D_GL_Info
* gl_info
= &((IWineD3DImpl
*)deviceImpl
->wineD3D
)->gl_info
;
595 char pshader
= shader_is_pshader_version(This
->baseShader
.hex_version
);
601 case WINED3DSPR_TEMP
:
602 sprintf(tmpStr
, "R%u", reg
);
604 case WINED3DSPR_INPUT
:
606 /* Pixel shaders >= 3.0 */
607 if (WINED3DSHADER_VERSION_MAJOR(This
->baseShader
.hex_version
) >= 3)
608 sprintf(tmpStr
, "IN%u", reg
);
611 strcpy(tmpStr
, "gl_Color");
613 strcpy(tmpStr
, "gl_SecondaryColor");
616 if (vshader_input_is_color((IWineD3DVertexShader
*) This
, reg
))
618 sprintf(tmpStr
, "attrib%u", reg
);
621 case WINED3DSPR_CONST
:
623 const char* prefix
= pshader
? "PC":"VC";
625 /* Relative addressing */
626 if (param
& WINED3DSHADER_ADDRMODE_RELATIVE
) {
628 /* Relative addressing on shaders 2.0+ have a relative address token,
629 * prior to that, it was hard-coded as "A0.x" because there's only 1 register */
630 if (WINED3DSHADER_VERSION_MAJOR(This
->baseShader
.hex_version
) >= 2) {
631 char relStr
[100], relReg
[50], relMask
[6];
632 shader_glsl_add_src_param(arg
, addr_token
, 0, WINED3DSP_WRITEMASK_0
, relReg
, relMask
, relStr
);
633 sprintf(tmpStr
, "%s[%s + %u]", prefix
, relStr
, reg
);
635 sprintf(tmpStr
, "%s[A0.x + %u]", prefix
, reg
);
638 sprintf(tmpStr
, "%s[%u]", prefix
, reg
);
642 case WINED3DSPR_CONSTINT
:
644 sprintf(tmpStr
, "PI[%u]", reg
);
646 sprintf(tmpStr
, "VI[%u]", reg
);
648 case WINED3DSPR_CONSTBOOL
:
650 sprintf(tmpStr
, "PB[%u]", reg
);
652 sprintf(tmpStr
, "VB[%u]", reg
);
654 case WINED3DSPR_TEXTURE
: /* case WINED3DSPR_ADDR: */
656 sprintf(tmpStr
, "T%u", reg
);
658 sprintf(tmpStr
, "A%u", reg
);
661 case WINED3DSPR_LOOP
:
662 sprintf(tmpStr
, "aL");
664 case WINED3DSPR_SAMPLER
:
666 sprintf(tmpStr
, "Psampler%u", reg
);
668 sprintf(tmpStr
, "Vsampler%u", reg
);
670 case WINED3DSPR_COLOROUT
:
671 if (reg
>= GL_LIMITS(buffers
)) {
672 WARN("Write to render target %u, only %d supported\n", reg
, 4);
674 if (GL_SUPPORT(ARB_DRAW_BUFFERS
)) {
675 sprintf(tmpStr
, "gl_FragData[%u]", reg
);
676 } else { /* On older cards with GLSL support like the GeforceFX there's only one buffer. */
677 sprintf(tmpStr
, "gl_FragColor");
680 case WINED3DSPR_RASTOUT
:
681 sprintf(tmpStr
, "%s", hwrastout_reg_names
[reg
]);
683 case WINED3DSPR_DEPTHOUT
:
684 sprintf(tmpStr
, "gl_FragDepth");
686 case WINED3DSPR_ATTROUT
:
688 sprintf(tmpStr
, "gl_FrontColor");
690 sprintf(tmpStr
, "gl_FrontSecondaryColor");
693 case WINED3DSPR_TEXCRDOUT
:
694 /* Vertex shaders >= 3.0: WINED3DSPR_OUTPUT */
695 if (WINED3DSHADER_VERSION_MAJOR(This
->baseShader
.hex_version
) >= 3)
696 sprintf(tmpStr
, "OUT%u", reg
);
698 sprintf(tmpStr
, "gl_TexCoord[%u]", reg
);
701 FIXME("Unhandled register name Type(%d)\n", regtype
);
702 sprintf(tmpStr
, "unrecognized_register");
706 strcat(regstr
, tmpStr
);
709 /* Get the GLSL write mask for the destination register */
710 static DWORD
shader_glsl_get_write_mask(const DWORD param
, char *write_mask
) {
711 char *ptr
= write_mask
;
712 DWORD mask
= param
& WINED3DSP_WRITEMASK_ALL
;
714 /* gl_FogFragCoord and glPointSize are floats, fixup the write mask. */
715 if ((shader_get_regtype(param
) == WINED3DSPR_RASTOUT
) && ((param
& WINED3DSP_REGNUM_MASK
) != 0)) {
716 mask
= WINED3DSP_WRITEMASK_0
;
718 if (mask
!= WINED3DSP_WRITEMASK_ALL
) {
720 if (param
& WINED3DSP_WRITEMASK_0
) *ptr
++ = 'x';
721 if (param
& WINED3DSP_WRITEMASK_1
) *ptr
++ = 'y';
722 if (param
& WINED3DSP_WRITEMASK_2
) *ptr
++ = 'z';
723 if (param
& WINED3DSP_WRITEMASK_3
) *ptr
++ = 'w';
732 static size_t shader_glsl_get_write_mask_size(DWORD write_mask
) {
735 if (write_mask
& WINED3DSP_WRITEMASK_0
) ++size
;
736 if (write_mask
& WINED3DSP_WRITEMASK_1
) ++size
;
737 if (write_mask
& WINED3DSP_WRITEMASK_2
) ++size
;
738 if (write_mask
& WINED3DSP_WRITEMASK_3
) ++size
;
743 static void shader_glsl_get_swizzle(const DWORD param
, BOOL fixup
, DWORD mask
, char *swizzle_str
) {
744 /* For registers of type WINED3DDECLTYPE_D3DCOLOR, data is stored as "bgra",
745 * but addressed as "rgba". To fix this we need to swap the register's x
746 * and z components. */
747 const char *swizzle_chars
= fixup
? "zyxw" : "xyzw";
748 char *ptr
= swizzle_str
;
750 /* swizzle bits fields: wwzzyyxx */
751 DWORD swizzle
= (param
& WINED3DSP_SWIZZLE_MASK
) >> WINED3DSP_SWIZZLE_SHIFT
;
752 DWORD swizzle_x
= swizzle
& 0x03;
753 DWORD swizzle_y
= (swizzle
>> 2) & 0x03;
754 DWORD swizzle_z
= (swizzle
>> 4) & 0x03;
755 DWORD swizzle_w
= (swizzle
>> 6) & 0x03;
757 /* FIXME: This is here to keep shader_glsl_add_src_param_old happy.
758 * As soon as that is removed, this needs to go as well. */
760 /* If the swizzle is the default swizzle (ie, "xyzw"), we don't need to
761 * generate a swizzle string. Unless we need to our own swizzling. */
762 if ((WINED3DSP_NOSWIZZLE
>> WINED3DSP_SWIZZLE_SHIFT
) != swizzle
|| fixup
) {
764 if (swizzle_x
== swizzle_y
&& swizzle_x
== swizzle_z
&& swizzle_x
== swizzle_w
) {
765 *ptr
++ = swizzle_chars
[swizzle_x
];
767 *ptr
++ = swizzle_chars
[swizzle_x
];
768 *ptr
++ = swizzle_chars
[swizzle_y
];
769 *ptr
++ = swizzle_chars
[swizzle_z
];
770 *ptr
++ = swizzle_chars
[swizzle_w
];
775 if (mask
& WINED3DSP_WRITEMASK_0
) *ptr
++ = swizzle_chars
[swizzle_x
];
776 if (mask
& WINED3DSP_WRITEMASK_1
) *ptr
++ = swizzle_chars
[swizzle_y
];
777 if (mask
& WINED3DSP_WRITEMASK_2
) *ptr
++ = swizzle_chars
[swizzle_z
];
778 if (mask
& WINED3DSP_WRITEMASK_3
) *ptr
++ = swizzle_chars
[swizzle_w
];
784 /* From a given parameter token, generate the corresponding GLSL string.
785 * Also, return the actual register name and swizzle in case the
786 * caller needs this information as well. */
787 static void shader_glsl_add_src_param(SHADER_OPCODE_ARG
* arg
, const DWORD param
,
788 const DWORD addr_token
, DWORD mask
, char *reg_name
, char *swizzle
, char *out_str
) {
789 BOOL is_color
= FALSE
;
790 swizzle
[0] = reg_name
[0] = out_str
[0] = 0;
792 shader_glsl_get_register_name(param
, addr_token
, reg_name
, &is_color
, arg
);
794 shader_glsl_get_swizzle(param
, is_color
, mask
, swizzle
);
795 shader_glsl_gen_modifier(param
, reg_name
, swizzle
, out_str
);
798 /* From a given parameter token, generate the corresponding GLSL string.
799 * Also, return the actual register name and swizzle in case the
800 * caller needs this information as well. */
801 static DWORD
shader_glsl_add_dst_param(SHADER_OPCODE_ARG
* arg
, const DWORD param
,
802 const DWORD addr_token
, char *reg_name
, char *write_mask
, char *out_str
) {
803 BOOL is_color
= FALSE
;
805 write_mask
[0] = reg_name
[0] = out_str
[0] = 0;
807 shader_glsl_get_register_name(param
, addr_token
, reg_name
, &is_color
, arg
);
809 mask
= shader_glsl_get_write_mask(param
, write_mask
);
810 sprintf(out_str
, "%s%s", reg_name
, write_mask
);
815 /* Append the destination part of the instruction to the buffer, return the effective write mask */
816 static DWORD
shader_glsl_append_dst(SHADER_BUFFER
*buffer
, SHADER_OPCODE_ARG
*arg
) {
817 char reg_name
[50], write_mask
[6], reg_str
[100];
821 shift
= (arg
->dst
& WINED3DSP_DSTSHIFT_MASK
) >> WINED3DSP_DSTSHIFT_SHIFT
;
822 mask
= shader_glsl_add_dst_param(arg
, arg
->dst
, arg
->dst_addr
, reg_name
, write_mask
, reg_str
);
823 shader_addline(buffer
, "%s = %s(", reg_str
, shift_glsl_tab
[shift
]);
828 /** Process GLSL instruction modifiers */
829 void shader_glsl_add_instruction_modifiers(SHADER_OPCODE_ARG
* arg
) {
831 DWORD mask
= arg
->dst
& WINED3DSP_DSTMOD_MASK
;
833 if (arg
->opcode
->dst_token
&& mask
!= 0) {
838 shader_glsl_add_dst_param(arg
, arg
->dst
, 0, dst_reg
, dst_mask
, dst_str
);
840 if (mask
& WINED3DSPDM_SATURATE
) {
841 /* _SAT means to clamp the value of the register to between 0 and 1 */
842 shader_addline(arg
->buffer
, "%s%s = clamp(%s%s, 0.0, 1.0);\n", dst_reg
, dst_mask
, dst_reg
, dst_mask
);
844 if (mask
& WINED3DSPDM_MSAMPCENTROID
) {
845 FIXME("_centroid modifier not handled\n");
847 if (mask
& WINED3DSPDM_PARTIALPRECISION
) {
848 /* MSDN says this modifier can be safely ignored, so that's what we'll do. */
853 static inline const char* shader_get_comp_op(
854 const DWORD opcode
) {
856 DWORD op
= (opcode
& INST_CONTROLS_MASK
) >> INST_CONTROLS_SHIFT
;
858 case COMPARISON_GT
: return ">";
859 case COMPARISON_EQ
: return "==";
860 case COMPARISON_GE
: return ">=";
861 case COMPARISON_LT
: return "<";
862 case COMPARISON_NE
: return "!=";
863 case COMPARISON_LE
: return "<=";
865 FIXME("Unrecognized comparison value: %u\n", op
);
870 static void shader_glsl_get_sample_function(DWORD sampler_type
, BOOL projected
, glsl_sample_function_t
*sample_function
) {
871 /* Note that there's no such thing as a projected cube texture. */
872 switch(sampler_type
) {
874 sample_function
->name
= projected
? "texture1DProj" : "texture1D";
875 sample_function
->coord_mask
= WINED3DSP_WRITEMASK_0
;
878 sample_function
->name
= projected
? "texture2DProj" : "texture2D";
879 sample_function
->coord_mask
= WINED3DSP_WRITEMASK_0
| WINED3DSP_WRITEMASK_1
;
881 case WINED3DSTT_CUBE
:
882 sample_function
->name
= "textureCube";
883 sample_function
->coord_mask
= WINED3DSP_WRITEMASK_0
| WINED3DSP_WRITEMASK_1
| WINED3DSP_WRITEMASK_2
;
885 case WINED3DSTT_VOLUME
:
886 sample_function
->name
= projected
? "texture3DProj" : "texture3D";
887 sample_function
->coord_mask
= WINED3DSP_WRITEMASK_0
| WINED3DSP_WRITEMASK_1
| WINED3DSP_WRITEMASK_2
;
890 sample_function
->name
= "";
891 FIXME("Unrecognized sampler type: %#x;\n", sampler_type
);
896 static void shader_glsl_sample(SHADER_OPCODE_ARG
* arg
, DWORD sampler_idx
, const char *dst_str
, const char *coord_reg
) {
897 IWineD3DPixelShaderImpl
* This
= (IWineD3DPixelShaderImpl
*) arg
->shader
;
898 IWineD3DDeviceImpl
* deviceImpl
= (IWineD3DDeviceImpl
*) This
->baseShader
.device
;
899 DWORD sampler_type
= arg
->reg_maps
->samplers
[sampler_idx
] & WINED3DSP_TEXTURETYPE_MASK
;
900 const char sampler_prefix
= shader_is_pshader_version(This
->baseShader
.hex_version
) ? 'P' : 'V';
901 SHADER_BUFFER
* buffer
= arg
->buffer
;
902 glsl_sample_function_t sample_function
;
904 if(deviceImpl
->stateBlock
->textureState
[sampler_idx
][WINED3DTSS_TEXTURETRANSFORMFLAGS
] & WINED3DTTFF_PROJECTED
) {
905 shader_glsl_get_sample_function(sampler_type
, TRUE
, &sample_function
);
906 shader_addline(buffer
, "%s = %s(%csampler%u, %s);\n", dst_str
, sample_function
.name
, sampler_prefix
, sampler_idx
, coord_reg
);
908 char coord_swizzle
[6];
910 shader_glsl_get_sample_function(sampler_type
, FALSE
, &sample_function
);
911 shader_glsl_get_write_mask(sample_function
.coord_mask
, coord_swizzle
);
913 shader_addline(buffer
, "%s = %s(%csampler%u, %s%s);\n", dst_str
, sample_function
.name
, sampler_prefix
, sampler_idx
, coord_reg
, coord_swizzle
);
918 /*****************************************************************************
920 * Begin processing individual instruction opcodes
922 ****************************************************************************/
924 /* Generate GLSL arithmetic functions (dst = src1 + src2) */
925 void shader_glsl_arith(SHADER_OPCODE_ARG
* arg
) {
926 CONST SHADER_OPCODE
* curOpcode
= arg
->opcode
;
927 SHADER_BUFFER
* buffer
= arg
->buffer
;
928 char src0_reg
[50], src1_reg
[50];
929 char src0_mask
[6], src1_mask
[6];
930 char src0_str
[100], src1_str
[100];
934 /* Determine the GLSL operator to use based on the opcode */
935 switch (curOpcode
->opcode
) {
936 case WINED3DSIO_MUL
: op
= '*'; break;
937 case WINED3DSIO_ADD
: op
= '+'; break;
938 case WINED3DSIO_SUB
: op
= '-'; break;
941 FIXME("Opcode %s not yet handled in GLSL\n", curOpcode
->name
);
945 write_mask
= shader_glsl_append_dst(buffer
, arg
);
946 shader_glsl_add_src_param(arg
, arg
->src
[0], arg
->src_addr
[0], write_mask
, src0_reg
, src0_mask
, src0_str
);
947 shader_glsl_add_src_param(arg
, arg
->src
[1], arg
->src_addr
[1], write_mask
, src1_reg
, src1_mask
, src1_str
);
948 shader_addline(buffer
, "%s %c %s);\n", src0_str
, op
, src1_str
);
951 /* Process the WINED3DSIO_MOV opcode using GLSL (dst = src) */
952 void shader_glsl_mov(SHADER_OPCODE_ARG
* arg
) {
953 IWineD3DBaseShaderImpl
* shader
= (IWineD3DBaseShaderImpl
*) arg
->shader
;
954 SHADER_BUFFER
* buffer
= arg
->buffer
;
960 write_mask
= shader_glsl_append_dst(buffer
, arg
);
961 shader_glsl_add_src_param(arg
, arg
->src
[0], arg
->src_addr
[0], write_mask
, src0_reg
, src0_mask
, src0_str
);
963 /* In vs_1_1 WINED3DSIO_MOV can write to the adress register. In later
964 * shader versions WINED3DSIO_MOVA is used for this. */
965 if ((WINED3DSHADER_VERSION_MAJOR(shader
->baseShader
.hex_version
) == 1 &&
966 !shader_is_pshader_version(shader
->baseShader
.hex_version
) &&
967 shader_get_regtype(arg
->dst
) == WINED3DSPR_ADDR
) ||
968 arg
->opcode
->opcode
== WINED3DSIO_MOVA
) {
969 /* We need to *round* to the nearest int here. */
970 size_t mask_size
= shader_glsl_get_write_mask_size(write_mask
);
972 shader_addline(buffer
, "ivec%d(floor(%s + vec%d(0.5))));\n", mask_size
, src0_str
, mask_size
);
974 shader_addline(buffer
, "int(floor(%s + 0.5)));\n", src0_str
);
977 shader_addline(buffer
, "%s);\n", src0_str
);
981 /* Process the dot product operators DP3 and DP4 in GLSL (dst = dot(src0, src1)) */
982 void shader_glsl_dot(SHADER_OPCODE_ARG
* arg
) {
983 CONST SHADER_OPCODE
* curOpcode
= arg
->opcode
;
984 SHADER_BUFFER
* buffer
= arg
->buffer
;
985 char src0_str
[100], src1_str
[100];
986 char src0_reg
[50], src1_reg
[50];
987 char src0_mask
[6], src1_mask
[6];
988 DWORD dst_write_mask
, src_write_mask
;
991 dst_write_mask
= shader_glsl_append_dst(buffer
, arg
);
992 dst_size
= shader_glsl_get_write_mask_size(dst_write_mask
);
994 /* dp3 works on vec3, dp4 on vec4 */
995 if (curOpcode
->opcode
== WINED3DSIO_DP4
) {
996 src_write_mask
= WINED3DSP_WRITEMASK_ALL
;
998 src_write_mask
= WINED3DSP_WRITEMASK_0
| WINED3DSP_WRITEMASK_1
| WINED3DSP_WRITEMASK_2
;
1001 shader_glsl_add_src_param(arg
, arg
->src
[0], arg
->src_addr
[0], src_write_mask
, src0_reg
, src0_mask
, src0_str
);
1002 shader_glsl_add_src_param(arg
, arg
->src
[1], arg
->src_addr
[1], src_write_mask
, src1_reg
, src1_mask
, src1_str
);
1005 shader_addline(buffer
, "vec%d(dot(%s, %s)));\n", dst_size
, src0_str
, src1_str
);
1007 shader_addline(buffer
, "dot(%s, %s));\n", src0_str
, src1_str
);
1011 /* Note that this instruction has some restrictions. The destination write mask
1012 * can't contain the w component, and the source swizzles have to be .xyzw */
1013 void shader_glsl_cross(SHADER_OPCODE_ARG
*arg
) {
1014 char src0_reg
[50], src0_mask
[6], src0_str
[100];
1015 char src1_reg
[50], src1_mask
[6], src1_str
[100];
1016 DWORD src_mask
= WINED3DSP_WRITEMASK_0
| WINED3DSP_WRITEMASK_1
| WINED3DSP_WRITEMASK_2
;
1019 shader_glsl_get_write_mask(arg
->dst
, dst_mask
);
1020 shader_glsl_append_dst(arg
->buffer
, arg
);
1021 shader_glsl_add_src_param(arg
, arg
->src
[0], arg
->src_addr
[0], src_mask
, src0_reg
, src0_mask
, src0_str
);
1022 shader_glsl_add_src_param(arg
, arg
->src
[1], arg
->src_addr
[1], src_mask
, src1_reg
, src1_mask
, src1_str
);
1023 shader_addline(arg
->buffer
, "cross(%s, %s).%s);\n", src0_str
, src1_str
, dst_mask
);
1026 /* Map the opcode 1-to-1 to the GL code (arg->dst = instruction(src0, src1, ...) */
1027 void shader_glsl_map2gl(SHADER_OPCODE_ARG
* arg
) {
1028 CONST SHADER_OPCODE
* curOpcode
= arg
->opcode
;
1029 SHADER_BUFFER
* buffer
= arg
->buffer
;
1030 const char *instruction
;
1031 char arguments
[256];
1038 /* Determine the GLSL function to use based on the opcode */
1039 /* TODO: Possibly make this a table for faster lookups */
1040 switch (curOpcode
->opcode
) {
1041 case WINED3DSIO_MIN
: instruction
= "min"; break;
1042 case WINED3DSIO_MAX
: instruction
= "max"; break;
1043 case WINED3DSIO_RSQ
: instruction
= "inversesqrt"; break;
1044 case WINED3DSIO_ABS
: instruction
= "abs"; break;
1045 case WINED3DSIO_FRC
: instruction
= "fract"; break;
1046 case WINED3DSIO_POW
: instruction
= "pow"; break;
1047 case WINED3DSIO_NRM
: instruction
= "normalize"; break;
1048 case WINED3DSIO_LOGP
:
1049 case WINED3DSIO_LOG
: instruction
= "log2"; break;
1050 case WINED3DSIO_EXP
: instruction
= "exp2"; break;
1051 case WINED3DSIO_SGN
: instruction
= "sign"; break;
1052 default: instruction
= "";
1053 FIXME("Opcode %s not yet handled in GLSL\n", curOpcode
->name
);
1057 write_mask
= shader_glsl_append_dst(buffer
, arg
);
1059 arguments
[0] = '\0';
1060 if (curOpcode
->num_params
> 0) {
1061 shader_glsl_add_src_param(arg
, arg
->src
[0], arg
->src_addr
[0], write_mask
, src_reg
, src_mask
, src_str
);
1062 strcat(arguments
, src_str
);
1063 for (i
= 2; i
< curOpcode
->num_params
; ++i
) {
1064 strcat(arguments
, ", ");
1065 shader_glsl_add_src_param(arg
, arg
->src
[i
-1], arg
->src_addr
[i
-1], write_mask
, src_reg
, src_mask
, src_str
);
1066 strcat(arguments
, src_str
);
1070 shader_addline(buffer
, "%s(%s));\n", instruction
, arguments
);
1073 /** Process the WINED3DSIO_EXPP instruction in GLSL:
1074 * For shader model 1.x, do the following (and honor the writemask, so use a temporary variable):
1075 * dst.x = 2^(floor(src))
1076 * dst.y = src - floor(src)
1077 * dst.z = 2^src (partial precision is allowed, but optional)
1079 * For 2.0 shaders, just do this (honoring writemask and swizzle):
1080 * dst = 2^src; (partial precision is allowed, but optional)
1082 void shader_glsl_expp(SHADER_OPCODE_ARG
* arg
) {
1083 IWineD3DBaseShaderImpl
*shader
= (IWineD3DBaseShaderImpl
*)arg
->shader
;
1088 shader_glsl_add_src_param(arg
, arg
->src
[0], arg
->src_addr
[0], WINED3DSP_WRITEMASK_0
, src_reg
, src_mask
, src_str
);
1090 if (shader
->baseShader
.hex_version
< WINED3DPS_VERSION(2,0)) {
1093 shader_addline(arg
->buffer
, "tmp0.x = exp2(floor(%s));\n", src_str
);
1094 shader_addline(arg
->buffer
, "tmp0.y = %s - floor(%s);\n", src_str
, src_str
);
1095 shader_addline(arg
->buffer
, "tmp0.z = exp2(%s);\n", src_str
);
1096 shader_addline(arg
->buffer
, "tmp0.w = 1.0;\n");
1098 shader_glsl_append_dst(arg
->buffer
, arg
);
1099 shader_glsl_get_write_mask(arg
->dst
, dst_mask
);
1100 shader_addline(arg
->buffer
, "tmp0%s);\n", dst_mask
);
1105 write_mask
= shader_glsl_append_dst(arg
->buffer
, arg
);
1106 mask_size
= shader_glsl_get_write_mask_size(write_mask
);
1108 if (mask_size
> 1) {
1109 shader_addline(arg
->buffer
, "vec%d(exp2(%s)));\n", mask_size
, src_str
);
1111 shader_addline(arg
->buffer
, "exp2(%s));\n", src_str
);
1116 /** Process the RCP (reciprocal or inverse) opcode in GLSL (dst = 1 / src) */
1117 void shader_glsl_rcp(SHADER_OPCODE_ARG
* arg
) {
1124 write_mask
= shader_glsl_append_dst(arg
->buffer
, arg
);
1125 mask_size
= shader_glsl_get_write_mask_size(write_mask
);
1126 shader_glsl_add_src_param(arg
, arg
->src
[0], arg
->src_addr
[0], WINED3DSP_WRITEMASK_0
, src_reg
, src_mask
, src_str
);
1128 if (mask_size
> 1) {
1129 shader_addline(arg
->buffer
, "vec%d(1.0 / %s));\n", mask_size
, src_str
);
1131 shader_addline(arg
->buffer
, "1.0 / %s);\n", src_str
);
1135 /** Process signed comparison opcodes in GLSL. */
1136 void shader_glsl_compare(SHADER_OPCODE_ARG
* arg
) {
1137 char src0_str
[100], src1_str
[100];
1138 char src0_reg
[50], src1_reg
[50];
1139 char src0_mask
[6], src1_mask
[6];
1143 write_mask
= shader_glsl_append_dst(arg
->buffer
, arg
);
1144 mask_size
= shader_glsl_get_write_mask_size(write_mask
);
1145 shader_glsl_add_src_param(arg
, arg
->src
[0], arg
->src_addr
[0], write_mask
, src0_reg
, src0_mask
, src0_str
);
1146 shader_glsl_add_src_param(arg
, arg
->src
[1], arg
->src_addr
[1], write_mask
, src1_reg
, src1_mask
, src1_str
);
1148 if (mask_size
> 1) {
1149 const char *compare
;
1151 switch(arg
->opcode
->opcode
) {
1152 case WINED3DSIO_SLT
: compare
= "lessThan"; break;
1153 case WINED3DSIO_SGE
: compare
= "greaterThanEqual"; break;
1154 default: compare
= "";
1155 FIXME("Can't handle opcode %s\n", arg
->opcode
->name
);
1158 shader_addline(arg
->buffer
, "vec%d(%s(%s, %s)));\n", mask_size
, compare
, src0_str
, src1_str
);
1160 const char *compare
;
1162 switch(arg
->opcode
->opcode
) {
1163 case WINED3DSIO_SLT
: compare
= "<"; break;
1164 case WINED3DSIO_SGE
: compare
= ">="; break;
1165 default: compare
= "";
1166 FIXME("Can't handle opcode %s\n", arg
->opcode
->name
);
1169 shader_addline(arg
->buffer
, "(%s %s %s) ? 1.0 : 0.0);\n", src0_str
, compare
, src1_str
);
1173 /** Process CMP instruction in GLSL (dst = src0 >= 0.0 ? src1 : src2), per channel */
1174 void shader_glsl_cmp(SHADER_OPCODE_ARG
* arg
) {
1175 char src0_str
[100], src1_str
[100], src2_str
[100];
1176 char src0_reg
[50], src1_reg
[50], src2_reg
[50];
1177 char src0_mask
[6], src1_mask
[6], src2_mask
[6];
1181 write_mask
= shader_glsl_append_dst(arg
->buffer
, arg
);
1182 mask_size
= shader_glsl_get_write_mask_size(write_mask
);
1184 shader_glsl_add_src_param(arg
, arg
->src
[0], arg
->src_addr
[0], write_mask
, src0_reg
, src0_mask
, src0_str
);
1185 shader_glsl_add_src_param(arg
, arg
->src
[1], arg
->src_addr
[1], write_mask
, src1_reg
, src1_mask
, src1_str
);
1186 shader_glsl_add_src_param(arg
, arg
->src
[2], arg
->src_addr
[2], write_mask
, src2_reg
, src2_mask
, src2_str
);
1188 if (mask_size
> 1) {
1189 shader_addline(arg
->buffer
, "mix(%s, %s, vec%d(lessThan(%s, vec%d(0.0)))));\n", src1_str
, src2_str
, mask_size
, src0_str
, mask_size
);
1191 shader_addline(arg
->buffer
, "%s >= 0.0 ? %s : %s);\n", src0_str
, src1_str
, src2_str
);
1195 /** Process the CND opcode in GLSL (dst = (src0 < 0.5) ? src1 : src2) */
1196 /* For ps 1.1-1.3, only a single component of src0 is used. For ps 1.4
1197 * the compare is done per component of src0. */
1198 void shader_glsl_cnd(SHADER_OPCODE_ARG
* arg
) {
1199 IWineD3DBaseShaderImpl
* shader
= (IWineD3DBaseShaderImpl
*) arg
->shader
;
1200 char src0_str
[100], src1_str
[100], src2_str
[100];
1201 char src0_reg
[50], src1_reg
[50], src2_reg
[50];
1202 char src0_mask
[6], src1_mask
[6], src2_mask
[6];
1206 write_mask
= shader_glsl_append_dst(arg
->buffer
, arg
);
1208 if (shader
->baseShader
.hex_version
< WINED3DPS_VERSION(1, 4)) {
1210 shader_glsl_add_src_param(arg
, arg
->src
[0], arg
->src_addr
[0], WINED3DSP_WRITEMASK_0
, src0_reg
, src0_mask
, src0_str
);
1212 mask_size
= shader_glsl_get_write_mask_size(write_mask
);
1213 shader_glsl_add_src_param(arg
, arg
->src
[0], arg
->src_addr
[0], write_mask
, src0_reg
, src0_mask
, src0_str
);
1216 shader_glsl_add_src_param(arg
, arg
->src
[1], arg
->src_addr
[1], write_mask
, src1_reg
, src1_mask
, src1_str
);
1217 shader_glsl_add_src_param(arg
, arg
->src
[2], arg
->src_addr
[2], write_mask
, src2_reg
, src2_mask
, src2_str
);
1219 if (mask_size
> 1) {
1220 shader_addline(arg
->buffer
, "mix(%s, %s, vec%d(lessThan(%s, vec%d(0.5)))));\n", src2_str
, src1_str
, mask_size
, src0_str
, mask_size
);
1222 shader_addline(arg
->buffer
, "%s < 0.5 ? %s : %s);\n", src0_str
, src1_str
, src2_str
);
1226 /** GLSL code generation for WINED3DSIO_MAD: Multiply the first 2 opcodes, then add the last */
1227 void shader_glsl_mad(SHADER_OPCODE_ARG
* arg
) {
1228 char src0_str
[100], src1_str
[100], src2_str
[100];
1229 char src0_reg
[50], src1_reg
[50], src2_reg
[50];
1230 char src0_mask
[6], src1_mask
[6], src2_mask
[6];
1233 write_mask
= shader_glsl_append_dst(arg
->buffer
, arg
);
1234 shader_glsl_add_src_param(arg
, arg
->src
[0], arg
->src_addr
[0], write_mask
, src0_reg
, src0_mask
, src0_str
);
1235 shader_glsl_add_src_param(arg
, arg
->src
[1], arg
->src_addr
[1], write_mask
, src1_reg
, src1_mask
, src1_str
);
1236 shader_glsl_add_src_param(arg
, arg
->src
[2], arg
->src_addr
[2], write_mask
, src2_reg
, src2_mask
, src2_str
);
1237 shader_addline(arg
->buffer
, "(%s * %s) + %s);\n", src0_str
, src1_str
, src2_str
);
1240 /** Handles transforming all WINED3DSIO_M?x? opcodes for
1241 Vertex shaders to GLSL codes */
1242 void shader_glsl_mnxn(SHADER_OPCODE_ARG
* arg
) {
1244 int nComponents
= 0;
1245 SHADER_OPCODE_ARG tmpArg
;
1247 memset(&tmpArg
, 0, sizeof(SHADER_OPCODE_ARG
));
1249 /* Set constants for the temporary argument */
1250 tmpArg
.shader
= arg
->shader
;
1251 tmpArg
.buffer
= arg
->buffer
;
1252 tmpArg
.src
[0] = arg
->src
[0];
1253 tmpArg
.src_addr
[0] = arg
->src_addr
[0];
1254 tmpArg
.src_addr
[1] = arg
->src_addr
[1];
1255 tmpArg
.reg_maps
= arg
->reg_maps
;
1257 switch(arg
->opcode
->opcode
) {
1258 case WINED3DSIO_M4x4
:
1260 tmpArg
.opcode
= shader_get_opcode(arg
->shader
, WINED3DSIO_DP4
);
1262 case WINED3DSIO_M4x3
:
1264 tmpArg
.opcode
= shader_get_opcode(arg
->shader
, WINED3DSIO_DP4
);
1266 case WINED3DSIO_M3x4
:
1268 tmpArg
.opcode
= shader_get_opcode(arg
->shader
, WINED3DSIO_DP3
);
1270 case WINED3DSIO_M3x3
:
1272 tmpArg
.opcode
= shader_get_opcode(arg
->shader
, WINED3DSIO_DP3
);
1274 case WINED3DSIO_M3x2
:
1276 tmpArg
.opcode
= shader_get_opcode(arg
->shader
, WINED3DSIO_DP3
);
1282 for (i
= 0; i
< nComponents
; i
++) {
1283 tmpArg
.dst
= ((arg
->dst
) & ~WINED3DSP_WRITEMASK_ALL
)|(WINED3DSP_WRITEMASK_0
<<i
);
1284 tmpArg
.src
[1] = arg
->src
[1]+i
;
1285 shader_glsl_dot(&tmpArg
);
1290 The LRP instruction performs a component-wise linear interpolation
1291 between the second and third operands using the first operand as the
1292 blend factor. Equation: (dst = src2 + src0 * (src1 - src2))
1293 This is equivalent to mix(src2, src1, src0);
1295 void shader_glsl_lrp(SHADER_OPCODE_ARG
* arg
) {
1296 char src0_str
[100], src1_str
[100], src2_str
[100];
1297 char src0_reg
[50], src1_reg
[50], src2_reg
[50];
1298 char src0_mask
[6], src1_mask
[6], src2_mask
[6];
1301 write_mask
= shader_glsl_append_dst(arg
->buffer
, arg
);
1303 shader_glsl_add_src_param(arg
, arg
->src
[0], arg
->src_addr
[0], write_mask
, src0_reg
, src0_mask
, src0_str
);
1304 shader_glsl_add_src_param(arg
, arg
->src
[1], arg
->src_addr
[1], write_mask
, src1_reg
, src1_mask
, src1_str
);
1305 shader_glsl_add_src_param(arg
, arg
->src
[2], arg
->src_addr
[2], write_mask
, src2_reg
, src2_mask
, src2_str
);
1307 shader_addline(arg
->buffer
, "mix(%s, %s, %s));\n", src2_str
, src1_str
, src0_str
);
1310 /** Process the WINED3DSIO_LIT instruction in GLSL:
1311 * dst.x = dst.w = 1.0
1312 * dst.y = (src0.x > 0) ? src0.x
1313 * dst.z = (src0.x > 0) ? ((src0.y > 0) ? pow(src0.y, src.w) : 0) : 0
1314 * where src.w is clamped at +- 128
1316 void shader_glsl_lit(SHADER_OPCODE_ARG
* arg
) {
1328 shader_glsl_append_dst(arg
->buffer
, arg
);
1329 shader_glsl_get_write_mask(arg
->dst
, dst_mask
);
1331 shader_glsl_add_src_param(arg
, arg
->src
[0], arg
->src_addr
[0], WINED3DSP_WRITEMASK_0
, src0_reg
, src0_mask
, src0_str
);
1332 shader_glsl_add_src_param(arg
, arg
->src
[0], arg
->src_addr
[0], WINED3DSP_WRITEMASK_1
, src1_reg
, src1_mask
, src1_str
);
1333 shader_glsl_add_src_param(arg
, arg
->src
[0], arg
->src_addr
[0], WINED3DSP_WRITEMASK_3
, src3_reg
, src3_mask
, src3_str
);
1335 shader_addline(arg
->buffer
, "vec4(1.0, (%s > 0.0 ? %s : 0.0), (%s > 0.0 ? ((%s > 0.0) ? pow(%s, clamp(%s, -128.0, 128.0)) : 0.0) : 0.0), 1.0)%s);\n",
1336 src0_str
, src0_str
, src0_str
, src1_str
, src1_str
, src3_str
, dst_mask
);
1339 /** Process the WINED3DSIO_DST instruction in GLSL:
1341 * dst.y = src0.x * src0.y
1345 void shader_glsl_dst(SHADER_OPCODE_ARG
* arg
) {
1346 char src0y_str
[100], src0z_str
[100], src1y_str
[100], src1w_str
[100];
1347 char src0y_reg
[50], src0z_reg
[50], src1y_reg
[50], src1w_reg
[50];
1348 char dst_mask
[6], src0y_mask
[6], src0z_mask
[6], src1y_mask
[6], src1w_mask
[6];
1350 shader_glsl_append_dst(arg
->buffer
, arg
);
1351 shader_glsl_get_write_mask(arg
->dst
, dst_mask
);
1353 shader_glsl_add_src_param(arg
, arg
->src
[0], arg
->src_addr
[0], WINED3DSP_WRITEMASK_1
, src0y_reg
, src0y_mask
, src0y_str
);
1354 shader_glsl_add_src_param(arg
, arg
->src
[0], arg
->src_addr
[0], WINED3DSP_WRITEMASK_2
, src0z_reg
, src0z_mask
, src0z_str
);
1355 shader_glsl_add_src_param(arg
, arg
->src
[1], arg
->src_addr
[1], WINED3DSP_WRITEMASK_1
, src1y_reg
, src1y_mask
, src1y_str
);
1356 shader_glsl_add_src_param(arg
, arg
->src
[1], arg
->src_addr
[1], WINED3DSP_WRITEMASK_3
, src1w_reg
, src1w_mask
, src1w_str
);
1358 shader_addline(arg
->buffer
, "vec4(1.0, %s * %s, %s, %s))%s;\n",
1359 src0y_str
, src1y_str
, src0z_str
, src1w_str
, dst_mask
);
1362 /** Process the WINED3DSIO_SINCOS instruction in GLSL:
1363 * VS 2.0 requires that specific cosine and sine constants be passed to this instruction so the hardware
1364 * can handle it. But, these functions are built-in for GLSL, so we can just ignore the last 2 params.
1366 * dst.x = cos(src0.?)
1367 * dst.y = sin(src0.?)
1371 void shader_glsl_sincos(SHADER_OPCODE_ARG
* arg
) {
1377 write_mask
= shader_glsl_append_dst(arg
->buffer
, arg
);
1378 shader_glsl_add_src_param(arg
, arg
->src
[0], arg
->src_addr
[0], WINED3DSP_WRITEMASK_0
, src0_reg
, src0_mask
, src0_str
);
1380 switch (write_mask
) {
1381 case WINED3DSP_WRITEMASK_0
:
1382 shader_addline(arg
->buffer
, "cos(%s));\n", src0_str
);
1385 case WINED3DSP_WRITEMASK_1
:
1386 shader_addline(arg
->buffer
, "sin(%s));\n", src0_str
);
1389 case (WINED3DSP_WRITEMASK_0
| WINED3DSP_WRITEMASK_1
):
1390 shader_addline(arg
->buffer
, "vec2(cos(%s), sin(%s)));\n", src0_str
, src0_str
);
1394 ERR("Write mask should be .x, .y or .xy\n");
1399 /** Process the WINED3DSIO_LOOP instruction in GLSL:
1400 * Start a for() loop where src0.y is the initial value of aL,
1401 * increment aL by src0.z for a total of src0.x iterations.
1402 * Need to use a temporary variable for this operation.
1404 void shader_glsl_loop(SHADER_OPCODE_ARG
* arg
) {
1410 shader_glsl_add_src_param_old(arg
, arg
->src
[1], arg
->src_addr
[1], src1_reg
, src1_mask
, src1_str
);
1412 shader_addline(arg
->buffer
, "for (tmpInt = 0, aL = %s.y; tmpInt < %s.x; tmpInt++, aL += %s.z) {\n",
1413 src1_reg
, src1_reg
, src1_reg
);
1416 void shader_glsl_end(SHADER_OPCODE_ARG
* arg
) {
1417 shader_addline(arg
->buffer
, "}\n");
1420 void shader_glsl_rep(SHADER_OPCODE_ARG
* arg
) {
1426 shader_glsl_add_src_param_old(arg
, arg
->src
[0], arg
->src_addr
[0], src0_reg
, src0_mask
, src0_str
);
1427 shader_addline(arg
->buffer
, "for (tmpInt = 0; tmpInt < %s.x; tmpInt++) {\n", src0_reg
);
1430 void shader_glsl_if(SHADER_OPCODE_ARG
* arg
) {
1436 shader_glsl_add_src_param_old(arg
, arg
->src
[0], arg
->src_addr
[0], src0_reg
, src0_mask
, src0_str
);
1437 shader_addline(arg
->buffer
, "if (%s) {\n", src0_str
);
1440 void shader_glsl_ifc(SHADER_OPCODE_ARG
* arg
) {
1442 char src0_str
[100], src1_str
[100];
1443 char src0_reg
[50], src1_reg
[50];
1444 char src0_mask
[6], src1_mask
[6];
1446 shader_glsl_add_src_param_old(arg
, arg
->src
[0], arg
->src_addr
[0], src0_reg
, src0_mask
, src0_str
);
1447 shader_glsl_add_src_param_old(arg
, arg
->src
[1], arg
->src_addr
[1], src1_reg
, src1_mask
, src1_str
);
1449 shader_addline(arg
->buffer
, "if (%s %s %s) {\n",
1450 src0_str
, shader_get_comp_op(arg
->opcode_token
), src1_str
);
1453 void shader_glsl_else(SHADER_OPCODE_ARG
* arg
) {
1454 shader_addline(arg
->buffer
, "} else {\n");
1457 void shader_glsl_break(SHADER_OPCODE_ARG
* arg
) {
1458 shader_addline(arg
->buffer
, "break;\n");
1461 void shader_glsl_breakc(SHADER_OPCODE_ARG
* arg
) {
1463 char src0_str
[100], src1_str
[100];
1464 char src0_reg
[50], src1_reg
[50];
1465 char src0_mask
[6], src1_mask
[6];
1467 shader_glsl_add_src_param_old(arg
, arg
->src
[0], arg
->src_addr
[0], src0_reg
, src0_mask
, src0_str
);
1468 shader_glsl_add_src_param_old(arg
, arg
->src
[1], arg
->src_addr
[1], src1_reg
, src1_mask
, src1_str
);
1470 shader_addline(arg
->buffer
, "if (%s %s %s) break;\n",
1471 src0_str
, shader_get_comp_op(arg
->opcode_token
), src1_str
);
1474 void shader_glsl_label(SHADER_OPCODE_ARG
* arg
) {
1476 DWORD snum
= (arg
->src
[0]) & WINED3DSP_REGNUM_MASK
;
1477 shader_addline(arg
->buffer
, "}\n");
1478 shader_addline(arg
->buffer
, "void subroutine%lu () {\n", snum
);
1481 void shader_glsl_call(SHADER_OPCODE_ARG
* arg
) {
1482 DWORD snum
= (arg
->src
[0]) & WINED3DSP_REGNUM_MASK
;
1483 shader_addline(arg
->buffer
, "subroutine%lu();\n", snum
);
1486 void shader_glsl_callnz(SHADER_OPCODE_ARG
* arg
) {
1492 DWORD snum
= (arg
->src
[0]) & WINED3DSP_REGNUM_MASK
;
1493 shader_glsl_add_src_param_old(arg
, arg
->src
[1], arg
->src_addr
[1], src1_reg
, src1_mask
, src1_str
);
1494 shader_addline(arg
->buffer
, "if (%s) subroutine%lu();\n", src1_str
, snum
);
1497 /*********************************************
1498 * Pixel Shader Specific Code begins here
1499 ********************************************/
1500 void pshader_glsl_tex(SHADER_OPCODE_ARG
* arg
) {
1501 IWineD3DPixelShaderImpl
* This
= (IWineD3DPixelShaderImpl
*) arg
->shader
;
1503 DWORD hex_version
= This
->baseShader
.hex_version
;
1505 char dst_str
[100], dst_reg
[50], dst_mask
[6];
1506 char coord_str
[100], coord_reg
[50], coord_mask
[6];
1508 /* All versions have a destination register */
1509 shader_glsl_add_dst_param(arg
, arg
->dst
, 0, dst_reg
, dst_mask
, dst_str
);
1511 /* 1.0-1.3: Use destination register as coordinate source.
1512 1.4+: Use provided coordinate source register. */
1513 if (hex_version
< WINED3DPS_VERSION(1,4))
1514 strcpy(coord_reg
, dst_reg
);
1516 shader_glsl_add_src_param_old(arg
, arg
->src
[0], arg
->src_addr
[0], coord_reg
, coord_mask
, coord_str
);
1518 /* 1.0-1.4: Use destination register as sampler source.
1519 * 2.0+: Use provided sampler source. */
1520 if (hex_version
< WINED3DPS_VERSION(2,0)) {
1521 shader_glsl_sample(arg
, arg
->dst
& WINED3DSP_REGNUM_MASK
, dst_str
, coord_reg
);
1523 shader_glsl_sample(arg
, arg
->src
[1] & WINED3DSP_REGNUM_MASK
, dst_str
, coord_reg
);
1528 void pshader_glsl_texcoord(SHADER_OPCODE_ARG
* arg
) {
1530 /* FIXME: Make this work for more than just 2D textures */
1532 IWineD3DPixelShaderImpl
* This
= (IWineD3DPixelShaderImpl
*) arg
->shader
;
1533 SHADER_BUFFER
* buffer
= arg
->buffer
;
1534 DWORD hex_version
= This
->baseShader
.hex_version
;
1541 shader_glsl_add_dst_param(arg
, arg
->dst
, 0, tmpReg
, tmpMask
, tmpStr
);
1543 if (hex_version
!= WINED3DPS_VERSION(1,4)) {
1544 DWORD reg
= arg
->dst
& WINED3DSP_REGNUM_MASK
;
1545 shader_addline(buffer
, "%s = clamp(gl_TexCoord[%u], 0.0, 1.0);\n", tmpReg
, reg
);
1547 DWORD reg2
= arg
->src
[0] & WINED3DSP_REGNUM_MASK
;
1548 shader_addline(buffer
, "%s = gl_TexCoord[%u]%s;\n", tmpStr
, reg2
, tmpMask
);
1552 /** Process the WINED3DSIO_TEXDP3TEX instruction in GLSL:
1553 * Take a 3-component dot product of the TexCoord[dstreg] and src,
1554 * then perform a 1D texture lookup from stage dstregnum, place into dst. */
1555 void pshader_glsl_texdp3tex(SHADER_OPCODE_ARG
* arg
) {
1560 DWORD sampler_idx
= arg
->dst
& WINED3DSP_REGNUM_MASK
;
1561 DWORD src_mask
= WINED3DSP_WRITEMASK_0
| WINED3DSP_WRITEMASK_1
| WINED3DSP_WRITEMASK_2
;
1563 shader_glsl_add_src_param(arg
, arg
->src
[0], arg
->src_addr
[0], src_mask
, src0_name
, src0_mask
, src0_str
);
1565 shader_glsl_append_dst(arg
->buffer
, arg
);
1566 shader_glsl_get_write_mask(arg
->dst
, dst_mask
);
1567 shader_addline(arg
->buffer
, "texture1D(Psampler%u, dot(gl_TexCoord[%u].xyz, %s))%s);\n",
1568 sampler_idx
, sampler_idx
, src0_str
, dst_mask
);
1571 /** Process the WINED3DSIO_TEXDP3 instruction in GLSL:
1572 * Take a 3-component dot product of the TexCoord[dstreg] and src. */
1573 void pshader_glsl_texdp3(SHADER_OPCODE_ARG
* arg
) {
1575 DWORD dstreg
= arg
->dst
& WINED3DSP_REGNUM_MASK
;
1576 char src0_str
[100], dst_str
[100];
1577 char src0_name
[50], dst_name
[50];
1578 char src0_mask
[6], dst_mask
[6];
1580 shader_glsl_add_dst_param(arg
, arg
->dst
, 0, dst_name
, dst_mask
, dst_str
);
1581 shader_glsl_add_src_param_old(arg
, arg
->src
[0], arg
->src_addr
[0], src0_name
, src0_mask
, src0_str
);
1583 shader_addline(arg
->buffer
, "%s = vec4(dot(vec3(T%u), vec3(%s)))%s;\n",
1584 dst_str
, dstreg
, src0_str
, dst_mask
);
1587 /** Process the WINED3DSIO_TEXDEPTH instruction in GLSL:
1588 * Calculate the depth as dst.x / dst.y */
1589 void pshader_glsl_texdepth(SHADER_OPCODE_ARG
* arg
) {
1595 shader_glsl_add_dst_param(arg
, arg
->dst
, 0, dst_reg
, dst_mask
, dst_str
);
1597 shader_addline(arg
->buffer
, "gl_FragDepth = %s.x / %s.y;\n", dst_reg
, dst_reg
);
1600 /** Process the WINED3DSIO_TEXM3X2DEPTH instruction in GLSL:
1601 * Last row of a 3x2 matrix multiply, use the result to calculate the depth:
1602 * Calculate tmp0.y = TexCoord[dstreg] . src.xyz; (tmp0.x has already been calculated)
1603 * depth = (tmp0.y == 0.0) ? 1.0 : tmp0.x / tmp0.y
1605 void pshader_glsl_texm3x2depth(SHADER_OPCODE_ARG
* arg
) {
1607 DWORD dstreg
= arg
->dst
& WINED3DSP_REGNUM_MASK
;
1608 char src0_str
[100], dst_str
[100];
1609 char src0_name
[50], dst_name
[50];
1610 char src0_mask
[6], dst_mask
[6];
1612 shader_glsl_add_dst_param(arg
, arg
->dst
, 0, dst_name
, dst_mask
, dst_str
);
1613 shader_glsl_add_src_param_old(arg
, arg
->src
[0], arg
->src_addr
[0], src0_name
, src0_mask
, src0_str
);
1615 shader_addline(arg
->buffer
, "tmp0.y = dot(vec3(T%u), vec3(%s));\n", dstreg
, src0_str
);
1616 shader_addline(arg
->buffer
, "gl_FragDepth = vec4((tmp0.y == 0.0) ? 1.0 : tmp0.x / tmp0.y)%s;\n", dst_str
, dst_name
);
1619 /** Process the WINED3DSIO_TEXM3X2PAD instruction in GLSL
1620 * Calculate the 1st of a 2-row matrix multiplication. */
1621 void pshader_glsl_texm3x2pad(SHADER_OPCODE_ARG
* arg
) {
1623 DWORD reg
= arg
->dst
& WINED3DSP_REGNUM_MASK
;
1624 SHADER_BUFFER
* buffer
= arg
->buffer
;
1629 shader_glsl_add_src_param_old(arg
, arg
->src
[0], arg
->src_addr
[0], src0_name
, src0_mask
, src0_str
);
1630 shader_addline(buffer
, "tmp0.x = dot(vec3(T%u), vec3(%s));\n", reg
, src0_str
);
1633 /** Process the WINED3DSIO_TEXM3X3PAD instruction in GLSL
1634 * Calculate the 1st or 2nd row of a 3-row matrix multiplication. */
1635 void pshader_glsl_texm3x3pad(SHADER_OPCODE_ARG
* arg
) {
1637 IWineD3DPixelShaderImpl
* shader
= (IWineD3DPixelShaderImpl
*) arg
->shader
;
1638 DWORD reg
= arg
->dst
& WINED3DSP_REGNUM_MASK
;
1639 SHADER_BUFFER
* buffer
= arg
->buffer
;
1640 SHADER_PARSE_STATE
* current_state
= &shader
->baseShader
.parse_state
;
1645 shader_glsl_add_src_param_old(arg
, arg
->src
[0], arg
->src_addr
[0], src0_name
, src0_mask
, src0_str
);
1646 shader_addline(buffer
, "tmp0.%c = dot(vec3(T%u), vec3(%s));\n", 'x' + current_state
->current_row
, reg
, src0_str
);
1647 current_state
->texcoord_w
[current_state
->current_row
++] = reg
;
1650 void pshader_glsl_texm3x2tex(SHADER_OPCODE_ARG
* arg
) {
1652 DWORD reg
= arg
->dst
& WINED3DSP_REGNUM_MASK
;
1653 SHADER_BUFFER
* buffer
= arg
->buffer
;
1658 shader_glsl_add_src_param_old(arg
, arg
->src
[0], arg
->src_addr
[0], src0_name
, src0_mask
, src0_str
);
1659 shader_addline(buffer
, "tmp0.y = dot(vec3(T%u), vec3(%s));\n", reg
, src0_str
);
1661 /* Sample the texture using the calculated coordinates */
1662 sprintf(dst_str
, "T%u", reg
);
1663 shader_glsl_sample(arg
, reg
, dst_str
, "tmp0");
1666 /** Process the WINED3DSIO_TEXM3X3TEX instruction in GLSL
1667 * Perform the 3rd row of a 3x3 matrix multiply, then sample the texture using the calculated coordinates */
1668 void pshader_glsl_texm3x3tex(SHADER_OPCODE_ARG
* arg
) {
1673 DWORD reg
= arg
->dst
& WINED3DSP_REGNUM_MASK
;
1674 IWineD3DPixelShaderImpl
* This
= (IWineD3DPixelShaderImpl
*) arg
->shader
;
1675 SHADER_PARSE_STATE
* current_state
= &This
->baseShader
.parse_state
;
1677 shader_glsl_add_src_param_old(arg
, arg
->src
[0], arg
->src_addr
[0], src0_name
, src0_mask
, src0_str
);
1678 shader_addline(arg
->buffer
, "tmp0.z = dot(vec3(T%u), vec3(%s));\n", reg
, src0_str
);
1680 /* Sample the texture using the calculated coordinates */
1681 sprintf(dst_str
, "T%u", reg
);
1682 shader_glsl_sample(arg
, reg
, dst_str
, "tmp0");
1683 current_state
->current_row
= 0;
1686 /** Process the WINED3DSIO_TEXM3X3 instruction in GLSL
1687 * Perform the 3rd row of a 3x3 matrix multiply */
1688 void pshader_glsl_texm3x3(SHADER_OPCODE_ARG
* arg
) {
1693 DWORD reg
= arg
->dst
& WINED3DSP_REGNUM_MASK
;
1694 IWineD3DPixelShaderImpl
* This
= (IWineD3DPixelShaderImpl
*) arg
->shader
;
1695 SHADER_PARSE_STATE
* current_state
= &This
->baseShader
.parse_state
;
1697 shader_glsl_add_src_param_old(arg
, arg
->src
[0], arg
->src_addr
[0], src0_name
, src0_mask
, src0_str
);
1699 shader_addline(arg
->buffer
, "tmp0.z = dot(vec3(T%u), vec3(%s));\n", reg
, src0_str
);
1700 shader_addline(arg
->buffer
, "T%u = vec4(tmp0.x, tmp0.y, tmp0.z, 1.0);\n", reg
);
1701 current_state
->current_row
= 0;
1704 /** Process the WINED3DSIO_TEXM3X3SPEC instruction in GLSL
1705 * Peform the final texture lookup based on the previous 2 3x3 matrix multiplies */
1706 void pshader_glsl_texm3x3spec(SHADER_OPCODE_ARG
* arg
) {
1708 IWineD3DPixelShaderImpl
* shader
= (IWineD3DPixelShaderImpl
*) arg
->shader
;
1709 DWORD reg
= arg
->dst
& WINED3DSP_REGNUM_MASK
;
1712 char src0_str
[100], src0_name
[50], src0_mask
[6];
1713 char src1_str
[100], src1_name
[50], src1_mask
[6];
1714 SHADER_BUFFER
* buffer
= arg
->buffer
;
1715 SHADER_PARSE_STATE
* current_state
= &shader
->baseShader
.parse_state
;
1716 DWORD stype
= arg
->reg_maps
->samplers
[reg
] & WINED3DSP_TEXTURETYPE_MASK
;
1719 case WINED3DSTT_2D
: strcpy(dimensions
, "2D"); break;
1720 case WINED3DSTT_CUBE
: strcpy(dimensions
, "Cube"); break;
1721 case WINED3DSTT_VOLUME
: strcpy(dimensions
, "3D"); break;
1723 strcpy(dimensions
, "");
1724 FIXME("Unrecognized sampler type: %#x\n", stype
);
1728 shader_glsl_add_src_param_old(arg
, arg
->src
[0], arg
->src_addr
[0], src0_name
, src0_mask
, src0_str
);
1729 shader_glsl_add_src_param_old(arg
, arg
->src
[1], arg
->src_addr
[1], src1_name
, src1_mask
, src1_str
);
1731 /* Perform the last matrix multiply operation */
1732 shader_addline(buffer
, "tmp0.z = dot(vec3(T%u), vec3(%s));\n", reg
, src0_str
);
1734 /* Calculate reflection vector */
1735 shader_addline(buffer
, "tmp0.xyz = reflect(-vec3(%s), vec3(tmp0));\n", src1_str
);
1737 /* Sample the texture */
1738 sprintf(dst_str
, "T%u", reg
);
1739 shader_glsl_sample(arg
, reg
, dst_str
, "tmp0");
1740 current_state
->current_row
= 0;
1743 /** Process the WINED3DSIO_TEXM3X3VSPEC instruction in GLSL
1744 * Peform the final texture lookup based on the previous 2 3x3 matrix multiplies */
1745 void pshader_glsl_texm3x3vspec(SHADER_OPCODE_ARG
* arg
) {
1747 IWineD3DPixelShaderImpl
* shader
= (IWineD3DPixelShaderImpl
*) arg
->shader
;
1748 DWORD reg
= arg
->dst
& WINED3DSP_REGNUM_MASK
;
1749 SHADER_BUFFER
* buffer
= arg
->buffer
;
1750 SHADER_PARSE_STATE
* current_state
= &shader
->baseShader
.parse_state
;
1752 char src0_str
[100], src0_name
[50], src0_mask
[6];
1754 shader_glsl_add_src_param_old(arg
, arg
->src
[0], arg
->src_addr
[0], src0_name
, src0_mask
, src0_str
);
1756 /* Perform the last matrix multiply operation */
1757 shader_addline(buffer
, "tmp0.z = dot(vec3(T%u), vec3(%s));\n", reg
, src0_str
);
1759 /* Construct the eye-ray vector from w coordinates */
1760 shader_addline(buffer
, "tmp1.x = gl_TexCoord[%u].w;\n", current_state
->texcoord_w
[0]);
1761 shader_addline(buffer
, "tmp1.y = gl_TexCoord[%u].w;\n", current_state
->texcoord_w
[1]);
1762 shader_addline(buffer
, "tmp1.z = gl_TexCoord[%u].w;\n", reg
);
1764 /* Calculate reflection vector (Assume normal is normalized): RF = 2*(N.E)*N -E */
1765 shader_addline(buffer
, "tmp0.x = dot(vec3(tmp0), vec3(tmp1));\n");
1766 shader_addline(buffer
, "tmp0 = tmp0.w * tmp0;\n");
1767 shader_addline(buffer
, "tmp0 = (2.0 * tmp0) - tmp1;\n");
1769 /* Sample the texture using the calculated coordinates */
1770 sprintf(dst_str
, "T%u", reg
);
1771 shader_glsl_sample(arg
, reg
, dst_str
, "tmp0");
1772 current_state
->current_row
= 0;
1775 /** Process the WINED3DSIO_TEXBEM instruction in GLSL.
1776 * Apply a fake bump map transform.
1777 * FIXME: Should apply the BUMPMAPENV matrix. For now, just sample the texture */
1778 void pshader_glsl_texbem(SHADER_OPCODE_ARG
* arg
) {
1780 DWORD reg1
= arg
->dst
& WINED3DSP_REGNUM_MASK
;
1781 DWORD reg2
= arg
->src
[0] & WINED3DSP_REGNUM_MASK
;
1783 FIXME("Not applying the BUMPMAPENV matrix for pixel shader instruction texbem.\n");
1784 shader_addline(arg
->buffer
, "T%u = texture2D(Psampler%u, gl_TexCoord[%u].xy + T%u.xy);\n",
1785 reg1
, reg1
, reg1
, reg2
);
1788 /** Process the WINED3DSIO_TEXREG2AR instruction in GLSL
1789 * Sample 2D texture at dst using the alpha & red (wx) components of src as texture coordinates */
1790 void pshader_glsl_texreg2ar(SHADER_OPCODE_ARG
* arg
) {
1794 DWORD sampler_idx
= arg
->dst
& WINED3DSP_REGNUM_MASK
;
1797 shader_glsl_append_dst(arg
->buffer
, arg
);
1798 shader_glsl_get_write_mask(arg
->dst
, dst_mask
);
1799 shader_glsl_add_src_param(arg
, arg
->src
[0], arg
->src_addr
[0], WINED3DSP_WRITEMASK_ALL
, src0_reg
, src0_mask
, src0_str
);
1801 shader_addline(arg
->buffer
, "texture2D(Psampler%u, %s.wx)%s);\n", sampler_idx
, src0_reg
, dst_mask
);
1804 /** Process the WINED3DSIO_TEXREG2GB instruction in GLSL
1805 * Sample 2D texture at dst using the green & blue (yz) components of src as texture coordinates */
1806 void pshader_glsl_texreg2gb(SHADER_OPCODE_ARG
* arg
) {
1810 DWORD sampler_idx
= arg
->dst
& WINED3DSP_REGNUM_MASK
;
1813 shader_glsl_append_dst(arg
->buffer
, arg
);
1814 shader_glsl_get_write_mask(arg
->dst
, dst_mask
);
1815 shader_glsl_add_src_param(arg
, arg
->src
[0], arg
->src_addr
[0], WINED3DSP_WRITEMASK_ALL
, src0_reg
, src0_mask
, src0_str
);
1817 shader_addline(arg
->buffer
, "texture2D(Psampler%u, %s.yz)%s);\n", sampler_idx
, src0_reg
, dst_mask
);
1820 /** Process the WINED3DSIO_TEXREG2RGB instruction in GLSL
1821 * Sample texture at dst using the rgb (xyz) components of src as texture coordinates */
1822 void pshader_glsl_texreg2rgb(SHADER_OPCODE_ARG
* arg
) {
1827 DWORD sampler_idx
= arg
->dst
& WINED3DSP_REGNUM_MASK
;
1828 DWORD sampler_type
= arg
->reg_maps
->samplers
[sampler_idx
] & WINED3DSP_TEXTURETYPE_MASK
;
1829 glsl_sample_function_t sample_function
;
1831 shader_glsl_append_dst(arg
->buffer
, arg
);
1832 shader_glsl_get_write_mask(arg
->dst
, dst_mask
);
1833 shader_glsl_get_sample_function(sampler_type
, FALSE
, &sample_function
);
1834 shader_glsl_add_src_param(arg
, arg
->src
[0], arg
->src_addr
[0], sample_function
.coord_mask
, src0_reg
, src0_mask
, src0_str
);
1836 shader_addline(arg
->buffer
, "%s(Psampler%u, %s)%s);\n", sample_function
.name
, sampler_idx
, src0_str
, dst_mask
);
1839 /** Process the WINED3DSIO_TEXKILL instruction in GLSL.
1840 * If any of the first 3 components are < 0, discard this pixel */
1841 void pshader_glsl_texkill(SHADER_OPCODE_ARG
* arg
) {
1843 char dst_str
[100], dst_name
[50], dst_mask
[6];
1845 shader_glsl_add_dst_param(arg
, arg
->dst
, 0, dst_name
, dst_mask
, dst_str
);
1846 shader_addline(arg
->buffer
, "if (any(lessThan(%s.xyz, vec3(0.0)))) discard;\n", dst_name
);
1849 /** Process the WINED3DSIO_DP2ADD instruction in GLSL.
1850 * dst = dot2(src0, src1) + src2 */
1851 void pshader_glsl_dp2add(SHADER_OPCODE_ARG
* arg
) {
1852 char src0_str
[100], src1_str
[100], src2_str
[100];
1853 char src0_reg
[50], src1_reg
[50], src2_reg
[50];
1854 char src0_mask
[6], src1_mask
[6], src2_mask
[6];
1858 write_mask
= shader_glsl_append_dst(arg
->buffer
, arg
);
1859 mask_size
= shader_glsl_get_write_mask_size(write_mask
);
1861 shader_glsl_add_src_param(arg
, arg
->src
[0], arg
->src_addr
[0], WINED3DSP_WRITEMASK_0
| WINED3DSP_WRITEMASK_1
, src0_reg
, src0_mask
, src0_str
);
1862 shader_glsl_add_src_param(arg
, arg
->src
[1], arg
->src_addr
[1], WINED3DSP_WRITEMASK_0
| WINED3DSP_WRITEMASK_1
, src1_reg
, src1_mask
, src1_str
);
1863 shader_glsl_add_src_param(arg
, arg
->src
[2], arg
->src_addr
[2], WINED3DSP_WRITEMASK_0
, src2_reg
, src2_mask
, src2_str
);
1865 shader_addline(arg
->buffer
, "dot(%s, %s) + %s);\n", src0_str
, src1_str
, src2_str
);
1868 void pshader_glsl_input_pack(
1869 SHADER_BUFFER
* buffer
,
1870 semantic
* semantics_in
) {
1874 for (i
= 0; i
< MAX_REG_INPUT
; i
++) {
1876 DWORD usage_token
= semantics_in
[i
].usage
;
1877 DWORD register_token
= semantics_in
[i
].reg
;
1878 DWORD usage
, usage_idx
;
1882 if (!usage_token
) continue;
1883 usage
= (usage_token
& WINED3DSP_DCL_USAGE_MASK
) >> WINED3DSP_DCL_USAGE_SHIFT
;
1884 usage_idx
= (usage_token
& WINED3DSP_DCL_USAGEINDEX_MASK
) >> WINED3DSP_DCL_USAGEINDEX_SHIFT
;
1885 shader_glsl_get_write_mask(register_token
, reg_mask
);
1889 case D3DDECLUSAGE_COLOR
:
1891 shader_addline(buffer
, "IN%u%s = vec4(gl_Color)%s;\n",
1892 i
, reg_mask
, reg_mask
);
1893 else if (usage_idx
== 1)
1894 shader_addline(buffer
, "IN%u%s = vec4(gl_SecondaryColor)%s;\n",
1895 i
, reg_mask
, reg_mask
);
1897 shader_addline(buffer
, "IN%u%s = vec4(unsupported_color_input)%s;\n",
1898 i
, reg_mask
, reg_mask
);
1901 case D3DDECLUSAGE_TEXCOORD
:
1902 shader_addline(buffer
, "IN%u%s = vec4(gl_TexCoord[%u])%s;\n",
1903 i
, reg_mask
, usage_idx
, reg_mask
);
1906 case D3DDECLUSAGE_FOG
:
1907 shader_addline(buffer
, "IN%u%s = vec4(gl_FogFragCoord)%s;\n",
1908 i
, reg_mask
, reg_mask
);
1912 shader_addline(buffer
, "IN%u%s = vec4(unsupported_input)%s;\n",
1913 i
, reg_mask
, reg_mask
);
1918 /*********************************************
1919 * Vertex Shader Specific Code begins here
1920 ********************************************/
1922 void vshader_glsl_output_unpack(
1923 SHADER_BUFFER
* buffer
,
1924 semantic
* semantics_out
) {
1928 for (i
= 0; i
< MAX_REG_OUTPUT
; i
++) {
1930 DWORD usage_token
= semantics_out
[i
].usage
;
1931 DWORD register_token
= semantics_out
[i
].reg
;
1932 DWORD usage
, usage_idx
;
1936 if (!usage_token
) continue;
1938 usage
= (usage_token
& WINED3DSP_DCL_USAGE_MASK
) >> WINED3DSP_DCL_USAGE_SHIFT
;
1939 usage_idx
= (usage_token
& WINED3DSP_DCL_USAGEINDEX_MASK
) >> WINED3DSP_DCL_USAGEINDEX_SHIFT
;
1940 shader_glsl_get_write_mask(register_token
, reg_mask
);
1944 case D3DDECLUSAGE_COLOR
:
1946 shader_addline(buffer
, "gl_FrontColor%s = OUT%u%s;\n", reg_mask
, i
, reg_mask
);
1947 else if (usage_idx
== 1)
1948 shader_addline(buffer
, "gl_FrontSecondaryColor%s = OUT%u%s;\n", reg_mask
, i
, reg_mask
);
1950 shader_addline(buffer
, "unsupported_color_output%s = OUT%u%s;\n", reg_mask
, i
, reg_mask
);
1953 case D3DDECLUSAGE_POSITION
:
1954 shader_addline(buffer
, "gl_Position%s = OUT%u%s;\n", reg_mask
, i
, reg_mask
);
1957 case D3DDECLUSAGE_TEXCOORD
:
1958 shader_addline(buffer
, "gl_TexCoord[%u]%s = OUT%u%s;\n",
1959 usage_idx
, reg_mask
, i
, reg_mask
);
1962 case WINED3DSHADERDECLUSAGE_PSIZE
:
1963 shader_addline(buffer
, "gl_PointSize = OUT%u.x;\n", i
);
1966 case WINED3DSHADERDECLUSAGE_FOG
:
1967 shader_addline(buffer
, "gl_FogFragCoord%s = OUT%u%s;\n", reg_mask
, i
, reg_mask
);
1971 shader_addline(buffer
, "unsupported_output%s = OUT%u%s;\n", reg_mask
, i
, reg_mask
);
1976 /** Attach a GLSL pixel or vertex shader object to the shader program */
1977 static void attach_glsl_shader(IWineD3DDevice
*iface
, IWineD3DBaseShader
* shader
) {
1978 IWineD3DDeviceImpl
*This
= (IWineD3DDeviceImpl
*)iface
;
1979 WineD3D_GL_Info
*gl_info
= &((IWineD3DImpl
*)(This
->wineD3D
))->gl_info
;
1980 GLhandleARB shaderObj
= ((IWineD3DBaseShaderImpl
*)shader
)->baseShader
.prgId
;
1981 if (This
->stateBlock
->glsl_program
&& shaderObj
!= 0) {
1982 TRACE("Attaching GLSL shader object %u to program %u\n", shaderObj
, This
->stateBlock
->glsl_program
->programId
);
1983 GL_EXTCALL(glAttachObjectARB(This
->stateBlock
->glsl_program
->programId
, shaderObj
));
1984 checkGLcall("glAttachObjectARB");
1988 /** Sets the GLSL program ID for the given pixel and vertex shader combination.
1989 * It sets the programId on the current StateBlock (because it should be called
1990 * inside of the DrawPrimitive() part of the render loop).
1992 * If a program for the given combination does not exist, create one, and store
1993 * the program in the list. If it creates a program, it will link the given
1996 * We keep the shader programs around on a list because linking
1997 * shader objects together is an expensive operation. It's much
1998 * faster to loop through a list of pre-compiled & linked programs
1999 * each time that the application sets a new pixel or vertex shader
2000 * than it is to re-link them together at that time.
2002 * The list will be deleted in IWineD3DDevice::Release().
2004 static void set_glsl_shader_program(IWineD3DDevice
*iface
) {
2005 IWineD3DDeviceImpl
*This
= (IWineD3DDeviceImpl
*)iface
;
2006 WineD3D_GL_Info
*gl_info
= &((IWineD3DImpl
*)(This
->wineD3D
))->gl_info
;
2007 IWineD3DPixelShader
*pshader
= This
->stateBlock
->pixelShader
;
2008 IWineD3DVertexShader
*vshader
= This
->stateBlock
->vertexShader
;
2009 struct glsl_shader_prog_link
*curLink
= NULL
;
2010 struct glsl_shader_prog_link
*newLink
= NULL
;
2011 struct list
*ptr
= NULL
;
2012 GLhandleARB programId
= 0;
2016 ptr
= list_head( &This
->glsl_shader_progs
);
2018 /* At least one program exists - see if it matches our ps/vs combination */
2019 curLink
= LIST_ENTRY( ptr
, struct glsl_shader_prog_link
, entry
);
2020 if (vshader
== curLink
->vertexShader
&& pshader
== curLink
->pixelShader
) {
2021 /* Existing Program found, use it */
2022 TRACE("Found existing program (%u) for this vertex/pixel shader combination\n",
2023 curLink
->programId
);
2024 This
->stateBlock
->glsl_program
= curLink
;
2027 /* This isn't the entry we need - try the next one */
2028 ptr
= list_next( &This
->glsl_shader_progs
, ptr
);
2031 /* If we get to this point, then no matching program exists, so we create one */
2032 programId
= GL_EXTCALL(glCreateProgramObjectARB());
2033 TRACE("Created new GLSL shader program %u\n", programId
);
2035 /* Allocate a new link for the list of programs */
2036 newLink
= HeapAlloc(GetProcessHeap(), 0, sizeof(struct glsl_shader_prog_link
));
2037 newLink
->programId
= programId
;
2038 This
->stateBlock
->glsl_program
= newLink
;
2040 /* Attach GLSL vshader */
2041 if (NULL
!= vshader
&& This
->vs_selected_mode
== SHADER_GLSL
) {
2043 int max_attribs
= 16; /* TODO: Will this always be the case? It is at the moment... */
2046 TRACE("Attaching vertex shader to GLSL program\n");
2047 attach_glsl_shader(iface
, (IWineD3DBaseShader
*)vshader
);
2049 /* Bind vertex attributes to a corresponding index number to match
2050 * the same index numbers as ARB_vertex_programs (makes loading
2051 * vertex attributes simpler). With this method, we can use the
2052 * exact same code to load the attributes later for both ARB and
2055 * We have to do this here because we need to know the Program ID
2056 * in order to make the bindings work, and it has to be done prior
2057 * to linking the GLSL program. */
2058 for (i
= 0; i
< max_attribs
; ++i
) {
2059 snprintf(tmp_name
, sizeof(tmp_name
), "attrib%i", i
);
2060 GL_EXTCALL(glBindAttribLocationARB(programId
, i
, tmp_name
));
2062 checkGLcall("glBindAttribLocationARB");
2063 newLink
->vertexShader
= vshader
;
2066 /* Attach GLSL pshader */
2067 if (NULL
!= pshader
&& This
->ps_selected_mode
== SHADER_GLSL
) {
2068 TRACE("Attaching pixel shader to GLSL program\n");
2069 attach_glsl_shader(iface
, (IWineD3DBaseShader
*)pshader
);
2070 newLink
->pixelShader
= pshader
;
2073 /* Link the program */
2074 TRACE("Linking GLSL shader program %u\n", programId
);
2075 GL_EXTCALL(glLinkProgramARB(programId
));
2076 print_glsl_info_log(&GLINFO_LOCATION
, programId
);
2077 list_add_head( &This
->glsl_shader_progs
, &newLink
->entry
);
2079 newLink
->vuniformF_locations
= HeapAlloc(GetProcessHeap(), 0, sizeof(GLhandleARB
) * GL_LIMITS(vshader_constantsF
));
2080 for (i
= 0; i
< GL_LIMITS(vshader_constantsF
); ++i
) {
2081 snprintf(glsl_name
, sizeof(glsl_name
), "VC[%i]", i
);
2082 newLink
->vuniformF_locations
[i
] = GL_EXTCALL(glGetUniformLocationARB(programId
, glsl_name
));
2084 newLink
->puniformF_locations
= HeapAlloc(GetProcessHeap(), 0, sizeof(GLhandleARB
) * GL_LIMITS(pshader_constantsF
));
2085 for (i
= 0; i
< GL_LIMITS(pshader_constantsF
); ++i
) {
2086 snprintf(glsl_name
, sizeof(glsl_name
), "PC[%i]", i
);
2087 newLink
->puniformF_locations
[i
] = GL_EXTCALL(glGetUniformLocationARB(programId
, glsl_name
));
2093 static GLhandleARB
create_glsl_blt_shader(WineD3D_GL_Info
*gl_info
) {
2094 GLhandleARB program_id
;
2095 GLhandleARB vshader_id
, pshader_id
;
2096 const char *blt_vshader
[] = {
2099 " gl_Position = gl_Vertex;\n"
2100 " gl_FrontColor = vec4(1.0);\n"
2101 " gl_TexCoord[0].x = (gl_Vertex.x * 0.5) + 0.5;\n"
2102 " gl_TexCoord[0].y = (-gl_Vertex.y * 0.5) + 0.5;\n"
2106 const char *blt_pshader
[] = {
2107 "uniform sampler2D sampler;\n"
2110 " gl_FragDepth = texture2D(sampler, gl_TexCoord[0].xy).x;\n"
2114 vshader_id
= GL_EXTCALL(glCreateShaderObjectARB(GL_VERTEX_SHADER_ARB
));
2115 GL_EXTCALL(glShaderSourceARB(vshader_id
, 1, blt_vshader
, NULL
));
2116 GL_EXTCALL(glCompileShaderARB(vshader_id
));
2118 pshader_id
= GL_EXTCALL(glCreateShaderObjectARB(GL_FRAGMENT_SHADER_ARB
));
2119 GL_EXTCALL(glShaderSourceARB(pshader_id
, 1, blt_pshader
, NULL
));
2120 GL_EXTCALL(glCompileShaderARB(pshader_id
));
2122 program_id
= GL_EXTCALL(glCreateProgramObjectARB());
2123 GL_EXTCALL(glAttachObjectARB(program_id
, vshader_id
));
2124 GL_EXTCALL(glAttachObjectARB(program_id
, pshader_id
));
2125 GL_EXTCALL(glLinkProgramARB(program_id
));
2127 print_glsl_info_log(&GLINFO_LOCATION
, program_id
);
2132 static void shader_glsl_select(IWineD3DDevice
*iface
, BOOL usePS
, BOOL useVS
) {
2133 IWineD3DDeviceImpl
*This
= (IWineD3DDeviceImpl
*)iface
;
2134 WineD3D_GL_Info
*gl_info
= &((IWineD3DImpl
*)(This
->wineD3D
))->gl_info
;
2135 GLhandleARB program_id
= 0;
2137 if (useVS
|| usePS
) set_glsl_shader_program(iface
);
2138 else This
->stateBlock
->glsl_program
= NULL
;
2140 program_id
= This
->stateBlock
->glsl_program
? This
->stateBlock
->glsl_program
->programId
: 0;
2141 if (program_id
) TRACE("Using GLSL program %u\n", program_id
);
2142 GL_EXTCALL(glUseProgramObjectARB(program_id
));
2143 checkGLcall("glUseProgramObjectARB");
2146 static void shader_glsl_select_depth_blt(IWineD3DDevice
*iface
) {
2147 IWineD3DDeviceImpl
*This
= (IWineD3DDeviceImpl
*)iface
;
2148 WineD3D_GL_Info
*gl_info
= &((IWineD3DImpl
*)(This
->wineD3D
))->gl_info
;
2149 static GLhandleARB program_id
= 0;
2150 static GLhandleARB loc
= -1;
2153 program_id
= create_glsl_blt_shader(gl_info
);
2154 loc
= GL_EXTCALL(glGetUniformLocationARB(program_id
, "sampler"));
2157 GL_EXTCALL(glUseProgramObjectARB(program_id
));
2158 GL_EXTCALL(glUniform1iARB(loc
, 0));
2161 static void shader_glsl_cleanup(BOOL usePS
, BOOL useVS
) {
2165 const shader_backend_t glsl_shader_backend
= {
2166 &shader_glsl_select
,
2167 &shader_glsl_select_depth_blt
,
2168 &shader_glsl_load_constants
,
2169 &shader_glsl_cleanup