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 /** Used for opcode modifiers - They multiply the result by the specified amount */
491 static const char * const shift_glsl_tab[] = {
493 "2.0 * ", /* 1 (x2) */
494 "4.0 * ", /* 2 (x4) */
495 "8.0 * ", /* 3 (x8) */
496 "16.0 * ", /* 4 (x16) */
497 "32.0 * ", /* 5 (x32) */
504 "0.0625 * ", /* 12 (d16) */
505 "0.125 * ", /* 13 (d8) */
506 "0.25 * ", /* 14 (d4) */
507 "0.5 * " /* 15 (d2) */
510 /* Generate a GLSL parameter that does the input modifier computation and return the input register/mask to use */
511 static void shader_glsl_gen_modifier (
514 const char *in_regswizzle,
519 if (instr == WINED3DSIO_TEXKILL)
522 switch (instr & WINED3DSP_SRCMOD_MASK) {
523 case WINED3DSPSM_DZ: /* Need to handle this in the instructions itself (texld & texcrd). */
525 case WINED3DSPSM_NONE:
526 sprintf(out_str, "%s%s", in_reg, in_regswizzle);
528 case WINED3DSPSM_NEG:
529 sprintf(out_str, "-%s%s", in_reg, in_regswizzle);
531 case WINED3DSPSM_NOT:
532 sprintf(out_str, "!%s%s", in_reg, in_regswizzle);
534 case WINED3DSPSM_BIAS:
535 sprintf(out_str, "(%s%s - vec4(0.5)%s)", in_reg, in_regswizzle, in_regswizzle);
537 case WINED3DSPSM_BIASNEG:
538 sprintf(out_str, "-(%s%s - vec4(0.5)%s)", in_reg, in_regswizzle, in_regswizzle);
540 case WINED3DSPSM_SIGN:
541 sprintf(out_str, "(2.0 * (%s%s - 0.5))", in_reg, in_regswizzle);
543 case WINED3DSPSM_SIGNNEG:
544 sprintf(out_str, "-(2.0 * (%s%s - 0.5))", in_reg, in_regswizzle);
546 case WINED3DSPSM_COMP:
547 sprintf(out_str, "(1.0 - %s%s)", in_reg, in_regswizzle);
550 sprintf(out_str, "(2.0 * %s%s)", in_reg, in_regswizzle);
552 case WINED3DSPSM_X2NEG:
553 sprintf(out_str, "-(2.0 * %s%s)", in_reg, in_regswizzle);
555 case WINED3DSPSM_ABS:
556 sprintf(out_str, "abs(%s%s)", in_reg, in_regswizzle);
558 case WINED3DSPSM_ABSNEG:
559 sprintf(out_str, "-abs(%s%s)", in_reg, in_regswizzle);
562 FIXME("Unhandled modifier %u\n", (instr & WINED3DSP_SRCMOD_MASK));
563 sprintf(out_str, "%s%s", in_reg, in_regswizzle);
567 /** Writes the GLSL variable name that corresponds to the register that the
568 * DX opcode parameter is trying to access */
569 static void shader_glsl_get_register_name(
571 const DWORD addr_token,
574 SHADER_OPCODE_ARG* arg) {
576 /* oPos, oFog and oPts in D3D */
577 static const char * const hwrastout_reg_names[] = { "gl_Position", "gl_FogFragCoord", "gl_PointSize" };
579 DWORD reg = param & WINED3DSP_REGNUM_MASK;
580 DWORD regtype = shader_get_regtype(param);
581 IWineD3DBaseShaderImpl* This = (IWineD3DBaseShaderImpl*) arg->shader;
582 IWineD3DDeviceImpl* deviceImpl = (IWineD3DDeviceImpl*) This->baseShader.device;
583 WineD3D_GL_Info* gl_info = &((IWineD3DImpl*)deviceImpl->wineD3D)->gl_info;
585 char pshader = shader_is_pshader_version(This->baseShader.hex_version);
591 case WINED3DSPR_TEMP:
592 sprintf(tmpStr, "R%u", reg);
594 case WINED3DSPR_INPUT:
596 /* Pixel shaders >= 3.0 */
597 if (WINED3DSHADER_VERSION_MAJOR(This->baseShader.hex_version) >= 3)
598 sprintf(tmpStr, "IN%u", reg);
601 strcpy(tmpStr, "gl_Color");
603 strcpy(tmpStr, "gl_SecondaryColor");
606 if (vshader_input_is_color((IWineD3DVertexShader*) This, reg))
608 sprintf(tmpStr, "attrib%u", reg);
611 case WINED3DSPR_CONST:
613 const char* prefix = pshader? "PC":"VC";
615 /* Relative addressing */
616 if (param & WINED3DSHADER_ADDRMODE_RELATIVE) {
618 /* Relative addressing on shaders 2.0+ have a relative address token,
619 * prior to that, it was hard-coded as "A0.x" because there's only 1 register */
620 if (WINED3DSHADER_VERSION_MAJOR(This->baseShader.hex_version) >= 2) {
621 char relStr[100], relReg[50], relMask[6];
622 shader_glsl_add_src_param(arg, addr_token, 0, WINED3DSP_WRITEMASK_0, relReg, relMask, relStr);
623 sprintf(tmpStr, "%s[%s + %u]", prefix, relStr, reg);
625 sprintf(tmpStr, "%s[A0.x + %u]", prefix, reg);
628 sprintf(tmpStr, "%s[%u]", prefix, reg);
632 case WINED3DSPR_CONSTINT:
634 sprintf(tmpStr, "PI[%u]", reg);
636 sprintf(tmpStr, "VI[%u]", reg);
638 case WINED3DSPR_CONSTBOOL:
640 sprintf(tmpStr, "PB[%u]", reg);
642 sprintf(tmpStr, "VB[%u]", reg);
644 case WINED3DSPR_TEXTURE: /* case WINED3DSPR_ADDR: */
646 sprintf(tmpStr, "T%u", reg);
648 sprintf(tmpStr, "A%u", reg);
651 case WINED3DSPR_LOOP:
652 sprintf(tmpStr, "aL");
654 case WINED3DSPR_SAMPLER:
656 sprintf(tmpStr, "Psampler%u", reg);
658 sprintf(tmpStr, "Vsampler%u", reg);
660 case WINED3DSPR_COLOROUT:
661 if (reg >= GL_LIMITS(buffers)) {
662 WARN("Write to render target %u, only %d supported\n", reg, 4);
664 if (GL_SUPPORT(ARB_DRAW_BUFFERS)) {
665 sprintf(tmpStr, "gl_FragData[%u]", reg);
666 } else { /* On older cards with GLSL support like the GeforceFX there's only one buffer. */
667 sprintf(tmpStr, "gl_FragColor");
670 case WINED3DSPR_RASTOUT:
671 sprintf(tmpStr, "%s", hwrastout_reg_names[reg]);
673 case WINED3DSPR_DEPTHOUT:
674 sprintf(tmpStr, "gl_FragDepth");
676 case WINED3DSPR_ATTROUT:
678 sprintf(tmpStr, "gl_FrontColor");
680 sprintf(tmpStr, "gl_FrontSecondaryColor");
683 case WINED3DSPR_TEXCRDOUT:
684 /* Vertex shaders >= 3.0: WINED3DSPR_OUTPUT */
685 if (WINED3DSHADER_VERSION_MAJOR(This->baseShader.hex_version) >= 3)
686 sprintf(tmpStr, "OUT%u", reg);
688 sprintf(tmpStr, "gl_TexCoord[%u]", reg);
691 FIXME("Unhandled register name Type(%d)\n", regtype);
692 sprintf(tmpStr, "unrecognized_register");
696 strcat(regstr, tmpStr);
699 /* Get the GLSL write mask for the destination register */
700 static DWORD shader_glsl_get_write_mask(const DWORD param, char *write_mask) {
701 char *ptr = write_mask;
702 DWORD mask = param & WINED3DSP_WRITEMASK_ALL;
704 /* gl_FogFragCoord and glPointSize are floats, fixup the write mask. */
705 if ((shader_get_regtype(param) == WINED3DSPR_RASTOUT) && ((param & WINED3DSP_REGNUM_MASK) != 0)) {
706 mask = WINED3DSP_WRITEMASK_0;
708 if (mask != WINED3DSP_WRITEMASK_ALL) {
710 if (param & WINED3DSP_WRITEMASK_0) *ptr++ = 'x';
711 if (param & WINED3DSP_WRITEMASK_1) *ptr++ = 'y';
712 if (param & WINED3DSP_WRITEMASK_2) *ptr++ = 'z';
713 if (param & WINED3DSP_WRITEMASK_3) *ptr++ = 'w';
722 static size_t shader_glsl_get_write_mask_size(DWORD write_mask) {
725 if (write_mask & WINED3DSP_WRITEMASK_0) ++size;
726 if (write_mask & WINED3DSP_WRITEMASK_1) ++size;
727 if (write_mask & WINED3DSP_WRITEMASK_2) ++size;
728 if (write_mask & WINED3DSP_WRITEMASK_3) ++size;
733 static void shader_glsl_get_swizzle(const DWORD param, BOOL fixup, DWORD mask, char *swizzle_str) {
734 /* For registers of type WINED3DDECLTYPE_D3DCOLOR, data is stored as "bgra",
735 * but addressed as "rgba". To fix this we need to swap the register's x
736 * and z components. */
737 const char *swizzle_chars = fixup ? "zyxw" : "xyzw";
738 char *ptr = swizzle_str;
740 /* swizzle bits fields: wwzzyyxx */
741 DWORD swizzle = (param & WINED3DSP_SWIZZLE_MASK) >> WINED3DSP_SWIZZLE_SHIFT;
742 DWORD swizzle_x = swizzle & 0x03;
743 DWORD swizzle_y = (swizzle >> 2) & 0x03;
744 DWORD swizzle_z = (swizzle >> 4) & 0x03;
745 DWORD swizzle_w = (swizzle >> 6) & 0x03;
748 if (mask & WINED3DSP_WRITEMASK_0) *ptr++ = swizzle_chars[swizzle_x];
749 if (mask & WINED3DSP_WRITEMASK_1) *ptr++ = swizzle_chars[swizzle_y];
750 if (mask & WINED3DSP_WRITEMASK_2) *ptr++ = swizzle_chars[swizzle_z];
751 if (mask & WINED3DSP_WRITEMASK_3) *ptr++ = swizzle_chars[swizzle_w];
756 /* From a given parameter token, generate the corresponding GLSL string.
757 * Also, return the actual register name and swizzle in case the
758 * caller needs this information as well. */
759 static void shader_glsl_add_src_param(SHADER_OPCODE_ARG* arg, const DWORD param,
760 const DWORD addr_token, DWORD mask, char *reg_name, char *swizzle, char *out_str) {
761 BOOL is_color = FALSE;
762 swizzle[0] = reg_name[0] = out_str[0] = 0;
764 shader_glsl_get_register_name(param, addr_token, reg_name, &is_color, arg);
766 shader_glsl_get_swizzle(param, is_color, mask, swizzle);
767 shader_glsl_gen_modifier(param, reg_name, swizzle, out_str);
770 /* From a given parameter token, generate the corresponding GLSL string.
771 * Also, return the actual register name and swizzle in case the
772 * caller needs this information as well. */
773 static DWORD shader_glsl_add_dst_param(SHADER_OPCODE_ARG* arg, const DWORD param,
774 const DWORD addr_token, char *reg_name, char *write_mask, char *out_str) {
775 BOOL is_color = FALSE;
777 write_mask[0] = reg_name[0] = out_str[0] = 0;
779 shader_glsl_get_register_name(param, addr_token, reg_name, &is_color, arg);
781 mask = shader_glsl_get_write_mask(param, write_mask);
782 sprintf(out_str, "%s%s", reg_name, write_mask);
787 /* Append the destination part of the instruction to the buffer, return the effective write mask */
788 static DWORD shader_glsl_append_dst(SHADER_BUFFER *buffer, SHADER_OPCODE_ARG *arg) {
789 char reg_name[50], write_mask[6], reg_str[100];
793 shift = (arg->dst & WINED3DSP_DSTSHIFT_MASK) >> WINED3DSP_DSTSHIFT_SHIFT;
794 mask = shader_glsl_add_dst_param(arg, arg->dst, arg->dst_addr, reg_name, write_mask, reg_str);
795 shader_addline(buffer, "%s = %s(", reg_str, shift_glsl_tab[shift]);
800 /** Process GLSL instruction modifiers */
801 void shader_glsl_add_instruction_modifiers(SHADER_OPCODE_ARG* arg) {
803 DWORD mask = arg->dst & WINED3DSP_DSTMOD_MASK;
805 if (arg->opcode->dst_token && mask != 0) {
810 shader_glsl_add_dst_param(arg, arg->dst, 0, dst_reg, dst_mask, dst_str);
812 if (mask & WINED3DSPDM_SATURATE) {
813 /* _SAT means to clamp the value of the register to between 0 and 1 */
814 shader_addline(arg->buffer, "%s%s = clamp(%s%s, 0.0, 1.0);\n", dst_reg, dst_mask, dst_reg, dst_mask);
816 if (mask & WINED3DSPDM_MSAMPCENTROID) {
817 FIXME("_centroid modifier not handled\n");
819 if (mask & WINED3DSPDM_PARTIALPRECISION) {
820 /* MSDN says this modifier can be safely ignored, so that's what we'll do. */
825 static inline const char* shader_get_comp_op(
826 const DWORD opcode) {
828 DWORD op = (opcode & INST_CONTROLS_MASK) >> INST_CONTROLS_SHIFT;
830 case COMPARISON_GT: return ">";
831 case COMPARISON_EQ: return "==";
832 case COMPARISON_GE: return ">=";
833 case COMPARISON_LT: return "<";
834 case COMPARISON_NE: return "!=";
835 case COMPARISON_LE: return "<=";
837 FIXME("Unrecognized comparison value: %u\n", op);
842 static void shader_glsl_get_sample_function(DWORD sampler_type, BOOL projected, glsl_sample_function_t *sample_function) {
843 /* Note that there's no such thing as a projected cube texture. */
844 switch(sampler_type) {
846 sample_function->name = projected ? "texture1DProj" : "texture1D";
847 sample_function->coord_mask = WINED3DSP_WRITEMASK_0;
850 sample_function->name = projected ? "texture2DProj" : "texture2D";
851 sample_function->coord_mask = WINED3DSP_WRITEMASK_0 | WINED3DSP_WRITEMASK_1;
853 case WINED3DSTT_CUBE:
854 sample_function->name = "textureCube";
855 sample_function->coord_mask = WINED3DSP_WRITEMASK_0 | WINED3DSP_WRITEMASK_1 | WINED3DSP_WRITEMASK_2;
857 case WINED3DSTT_VOLUME:
858 sample_function->name = projected ? "texture3DProj" : "texture3D";
859 sample_function->coord_mask = WINED3DSP_WRITEMASK_0 | WINED3DSP_WRITEMASK_1 | WINED3DSP_WRITEMASK_2;
862 sample_function->name = "";
863 FIXME("Unrecognized sampler type: %#x;\n", sampler_type);
869 /*****************************************************************************
871 * Begin processing individual instruction opcodes
873 ****************************************************************************/
875 /* Generate GLSL arithmetic functions (dst = src1 + src2) */
876 void shader_glsl_arith(SHADER_OPCODE_ARG* arg) {
877 CONST SHADER_OPCODE* curOpcode = arg->opcode;
878 SHADER_BUFFER* buffer = arg->buffer;
879 char src0_reg[50], src1_reg[50];
880 char src0_mask[6], src1_mask[6];
881 char src0_str[100], src1_str[100];
885 /* Determine the GLSL operator to use based on the opcode */
886 switch (curOpcode->opcode) {
887 case WINED3DSIO_MUL: op = '*'; break;
888 case WINED3DSIO_ADD: op = '+'; break;
889 case WINED3DSIO_SUB: op = '-'; break;
892 FIXME("Opcode %s not yet handled in GLSL\n", curOpcode->name);
896 write_mask = shader_glsl_append_dst(buffer, arg);
897 shader_glsl_add_src_param(arg, arg->src[0], arg->src_addr[0], write_mask, src0_reg, src0_mask, src0_str);
898 shader_glsl_add_src_param(arg, arg->src[1], arg->src_addr[1], write_mask, src1_reg, src1_mask, src1_str);
899 shader_addline(buffer, "%s %c %s);\n", src0_str, op, src1_str);
902 /* Process the WINED3DSIO_MOV opcode using GLSL (dst = src) */
903 void shader_glsl_mov(SHADER_OPCODE_ARG* arg) {
904 IWineD3DBaseShaderImpl* shader = (IWineD3DBaseShaderImpl*) arg->shader;
905 SHADER_BUFFER* buffer = arg->buffer;
911 write_mask = shader_glsl_append_dst(buffer, arg);
912 shader_glsl_add_src_param(arg, arg->src[0], arg->src_addr[0], write_mask, src0_reg, src0_mask, src0_str);
914 /* In vs_1_1 WINED3DSIO_MOV can write to the adress register. In later
915 * shader versions WINED3DSIO_MOVA is used for this. */
916 if ((WINED3DSHADER_VERSION_MAJOR(shader->baseShader.hex_version) == 1 &&
917 !shader_is_pshader_version(shader->baseShader.hex_version) &&
918 shader_get_regtype(arg->dst) == WINED3DSPR_ADDR) ||
919 arg->opcode->opcode == WINED3DSIO_MOVA) {
920 /* We need to *round* to the nearest int here. */
921 size_t mask_size = shader_glsl_get_write_mask_size(write_mask);
923 shader_addline(buffer, "ivec%d(floor(%s + vec%d(0.5))));\n", mask_size, src0_str, mask_size);
925 shader_addline(buffer, "int(floor(%s + 0.5)));\n", src0_str);
928 shader_addline(buffer, "%s);\n", src0_str);
932 /* Process the dot product operators DP3 and DP4 in GLSL (dst = dot(src0, src1)) */
933 void shader_glsl_dot(SHADER_OPCODE_ARG* arg) {
934 CONST SHADER_OPCODE* curOpcode = arg->opcode;
935 SHADER_BUFFER* buffer = arg->buffer;
936 char src0_str[100], src1_str[100];
937 char src0_reg[50], src1_reg[50];
938 char src0_mask[6], src1_mask[6];
939 DWORD dst_write_mask, src_write_mask;
942 dst_write_mask = shader_glsl_append_dst(buffer, arg);
943 dst_size = shader_glsl_get_write_mask_size(dst_write_mask);
945 /* dp3 works on vec3, dp4 on vec4 */
946 if (curOpcode->opcode == WINED3DSIO_DP4) {
947 src_write_mask = WINED3DSP_WRITEMASK_ALL;
949 src_write_mask = WINED3DSP_WRITEMASK_0 | WINED3DSP_WRITEMASK_1 | WINED3DSP_WRITEMASK_2;
952 shader_glsl_add_src_param(arg, arg->src[0], arg->src_addr[0], src_write_mask, src0_reg, src0_mask, src0_str);
953 shader_glsl_add_src_param(arg, arg->src[1], arg->src_addr[1], src_write_mask, src1_reg, src1_mask, src1_str);
956 shader_addline(buffer, "vec%d(dot(%s, %s)));\n", dst_size, src0_str, src1_str);
958 shader_addline(buffer, "dot(%s, %s));\n", src0_str, src1_str);
962 /* Note that this instruction has some restrictions. The destination write mask
963 * can't contain the w component, and the source swizzles have to be .xyzw */
964 void shader_glsl_cross(SHADER_OPCODE_ARG *arg) {
965 char src0_reg[50], src0_mask[6], src0_str[100];
966 char src1_reg[50], src1_mask[6], src1_str[100];
967 DWORD src_mask = WINED3DSP_WRITEMASK_0 | WINED3DSP_WRITEMASK_1 | WINED3DSP_WRITEMASK_2;
970 shader_glsl_get_write_mask(arg->dst, dst_mask);
971 shader_glsl_append_dst(arg->buffer, arg);
972 shader_glsl_add_src_param(arg, arg->src[0], arg->src_addr[0], src_mask, src0_reg, src0_mask, src0_str);
973 shader_glsl_add_src_param(arg, arg->src[1], arg->src_addr[1], src_mask, src1_reg, src1_mask, src1_str);
974 shader_addline(arg->buffer, "cross(%s, %s).%s);\n", src0_str, src1_str, dst_mask);
977 /* Map the opcode 1-to-1 to the GL code (arg->dst = instruction(src0, src1, ...) */
978 void shader_glsl_map2gl(SHADER_OPCODE_ARG* arg) {
979 CONST SHADER_OPCODE* curOpcode = arg->opcode;
980 SHADER_BUFFER* buffer = arg->buffer;
981 const char *instruction;
989 /* Determine the GLSL function to use based on the opcode */
990 /* TODO: Possibly make this a table for faster lookups */
991 switch (curOpcode->opcode) {
992 case WINED3DSIO_MIN: instruction = "min"; break;
993 case WINED3DSIO_MAX: instruction = "max"; break;
994 case WINED3DSIO_RSQ: instruction = "inversesqrt"; break;
995 case WINED3DSIO_ABS: instruction = "abs"; break;
996 case WINED3DSIO_FRC: instruction = "fract"; break;
997 case WINED3DSIO_POW: instruction = "pow"; break;
998 case WINED3DSIO_NRM: instruction = "normalize"; break;
999 case WINED3DSIO_LOGP:
1000 case WINED3DSIO_LOG: instruction = "log2"; break;
1001 case WINED3DSIO_EXP: instruction = "exp2"; break;
1002 case WINED3DSIO_SGN: instruction = "sign"; break;
1003 default: instruction = "";
1004 FIXME("Opcode %s not yet handled in GLSL\n", curOpcode->name);
1008 write_mask = shader_glsl_append_dst(buffer, arg);
1010 arguments[0] = '\0';
1011 if (curOpcode->num_params > 0) {
1012 shader_glsl_add_src_param(arg, arg->src[0], arg->src_addr[0], write_mask, src_reg, src_mask, src_str);
1013 strcat(arguments, src_str);
1014 for (i = 2; i < curOpcode->num_params; ++i) {
1015 strcat(arguments, ", ");
1016 shader_glsl_add_src_param(arg, arg->src[i-1], arg->src_addr[i-1], write_mask, src_reg, src_mask, src_str);
1017 strcat(arguments, src_str);
1021 shader_addline(buffer, "%s(%s));\n", instruction, arguments);
1024 /** Process the WINED3DSIO_EXPP instruction in GLSL:
1025 * For shader model 1.x, do the following (and honor the writemask, so use a temporary variable):
1026 * dst.x = 2^(floor(src))
1027 * dst.y = src - floor(src)
1028 * dst.z = 2^src (partial precision is allowed, but optional)
1030 * For 2.0 shaders, just do this (honoring writemask and swizzle):
1031 * dst = 2^src; (partial precision is allowed, but optional)
1033 void shader_glsl_expp(SHADER_OPCODE_ARG* arg) {
1034 IWineD3DBaseShaderImpl *shader = (IWineD3DBaseShaderImpl *)arg->shader;
1039 shader_glsl_add_src_param(arg, arg->src[0], arg->src_addr[0], WINED3DSP_WRITEMASK_0, src_reg, src_mask, src_str);
1041 if (shader->baseShader.hex_version < WINED3DPS_VERSION(2,0)) {
1044 shader_addline(arg->buffer, "tmp0.x = exp2(floor(%s));\n", src_str);
1045 shader_addline(arg->buffer, "tmp0.y = %s - floor(%s);\n", src_str, src_str);
1046 shader_addline(arg->buffer, "tmp0.z = exp2(%s);\n", src_str);
1047 shader_addline(arg->buffer, "tmp0.w = 1.0;\n");
1049 shader_glsl_append_dst(arg->buffer, arg);
1050 shader_glsl_get_write_mask(arg->dst, dst_mask);
1051 shader_addline(arg->buffer, "tmp0%s);\n", dst_mask);
1056 write_mask = shader_glsl_append_dst(arg->buffer, arg);
1057 mask_size = shader_glsl_get_write_mask_size(write_mask);
1059 if (mask_size > 1) {
1060 shader_addline(arg->buffer, "vec%d(exp2(%s)));\n", mask_size, src_str);
1062 shader_addline(arg->buffer, "exp2(%s));\n", src_str);
1067 /** Process the RCP (reciprocal or inverse) opcode in GLSL (dst = 1 / src) */
1068 void shader_glsl_rcp(SHADER_OPCODE_ARG* arg) {
1075 write_mask = shader_glsl_append_dst(arg->buffer, arg);
1076 mask_size = shader_glsl_get_write_mask_size(write_mask);
1077 shader_glsl_add_src_param(arg, arg->src[0], arg->src_addr[0], WINED3DSP_WRITEMASK_0, src_reg, src_mask, src_str);
1079 if (mask_size > 1) {
1080 shader_addline(arg->buffer, "vec%d(1.0 / %s));\n", mask_size, src_str);
1082 shader_addline(arg->buffer, "1.0 / %s);\n", src_str);
1086 /** Process signed comparison opcodes in GLSL. */
1087 void shader_glsl_compare(SHADER_OPCODE_ARG* arg) {
1088 char src0_str[100], src1_str[100];
1089 char src0_reg[50], src1_reg[50];
1090 char src0_mask[6], src1_mask[6];
1094 write_mask = shader_glsl_append_dst(arg->buffer, arg);
1095 mask_size = shader_glsl_get_write_mask_size(write_mask);
1096 shader_glsl_add_src_param(arg, arg->src[0], arg->src_addr[0], write_mask, src0_reg, src0_mask, src0_str);
1097 shader_glsl_add_src_param(arg, arg->src[1], arg->src_addr[1], write_mask, src1_reg, src1_mask, src1_str);
1099 if (mask_size > 1) {
1100 const char *compare;
1102 switch(arg->opcode->opcode) {
1103 case WINED3DSIO_SLT: compare = "lessThan"; break;
1104 case WINED3DSIO_SGE: compare = "greaterThanEqual"; break;
1105 default: compare = "";
1106 FIXME("Can't handle opcode %s\n", arg->opcode->name);
1109 shader_addline(arg->buffer, "vec%d(%s(%s, %s)));\n", mask_size, compare, src0_str, src1_str);
1111 const char *compare;
1113 switch(arg->opcode->opcode) {
1114 case WINED3DSIO_SLT: compare = "<"; break;
1115 case WINED3DSIO_SGE: compare = ">="; break;
1116 default: compare = "";
1117 FIXME("Can't handle opcode %s\n", arg->opcode->name);
1120 shader_addline(arg->buffer, "(%s %s %s) ? 1.0 : 0.0);\n", src0_str, compare, src1_str);
1124 /** Process CMP instruction in GLSL (dst = src0 >= 0.0 ? src1 : src2), per channel */
1125 void shader_glsl_cmp(SHADER_OPCODE_ARG* arg) {
1126 char src0_str[100], src1_str[100], src2_str[100];
1127 char src0_reg[50], src1_reg[50], src2_reg[50];
1128 char src0_mask[6], src1_mask[6], src2_mask[6];
1132 write_mask = shader_glsl_append_dst(arg->buffer, arg);
1133 mask_size = shader_glsl_get_write_mask_size(write_mask);
1135 shader_glsl_add_src_param(arg, arg->src[0], arg->src_addr[0], write_mask, src0_reg, src0_mask, src0_str);
1136 shader_glsl_add_src_param(arg, arg->src[1], arg->src_addr[1], write_mask, src1_reg, src1_mask, src1_str);
1137 shader_glsl_add_src_param(arg, arg->src[2], arg->src_addr[2], write_mask, src2_reg, src2_mask, src2_str);
1139 if (mask_size > 1) {
1140 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);
1142 shader_addline(arg->buffer, "%s >= 0.0 ? %s : %s);\n", src0_str, src1_str, src2_str);
1146 /** Process the CND opcode in GLSL (dst = (src0 < 0.5) ? src1 : src2) */
1147 /* For ps 1.1-1.3, only a single component of src0 is used. For ps 1.4
1148 * the compare is done per component of src0. */
1149 void shader_glsl_cnd(SHADER_OPCODE_ARG* arg) {
1150 IWineD3DBaseShaderImpl* shader = (IWineD3DBaseShaderImpl*) arg->shader;
1151 char src0_str[100], src1_str[100], src2_str[100];
1152 char src0_reg[50], src1_reg[50], src2_reg[50];
1153 char src0_mask[6], src1_mask[6], src2_mask[6];
1157 write_mask = shader_glsl_append_dst(arg->buffer, arg);
1159 if (shader->baseShader.hex_version < WINED3DPS_VERSION(1, 4)) {
1161 shader_glsl_add_src_param(arg, arg->src[0], arg->src_addr[0], WINED3DSP_WRITEMASK_0, src0_reg, src0_mask, src0_str);
1163 mask_size = shader_glsl_get_write_mask_size(write_mask);
1164 shader_glsl_add_src_param(arg, arg->src[0], arg->src_addr[0], write_mask, src0_reg, src0_mask, src0_str);
1167 shader_glsl_add_src_param(arg, arg->src[1], arg->src_addr[1], write_mask, src1_reg, src1_mask, src1_str);
1168 shader_glsl_add_src_param(arg, arg->src[2], arg->src_addr[2], write_mask, src2_reg, src2_mask, src2_str);
1170 if (mask_size > 1) {
1171 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);
1173 shader_addline(arg->buffer, "%s < 0.5 ? %s : %s);\n", src0_str, src1_str, src2_str);
1177 /** GLSL code generation for WINED3DSIO_MAD: Multiply the first 2 opcodes, then add the last */
1178 void shader_glsl_mad(SHADER_OPCODE_ARG* arg) {
1179 char src0_str[100], src1_str[100], src2_str[100];
1180 char src0_reg[50], src1_reg[50], src2_reg[50];
1181 char src0_mask[6], src1_mask[6], src2_mask[6];
1184 write_mask = shader_glsl_append_dst(arg->buffer, arg);
1185 shader_glsl_add_src_param(arg, arg->src[0], arg->src_addr[0], write_mask, src0_reg, src0_mask, src0_str);
1186 shader_glsl_add_src_param(arg, arg->src[1], arg->src_addr[1], write_mask, src1_reg, src1_mask, src1_str);
1187 shader_glsl_add_src_param(arg, arg->src[2], arg->src_addr[2], write_mask, src2_reg, src2_mask, src2_str);
1188 shader_addline(arg->buffer, "(%s * %s) + %s);\n", src0_str, src1_str, src2_str);
1191 /** Handles transforming all WINED3DSIO_M?x? opcodes for
1192 Vertex shaders to GLSL codes */
1193 void shader_glsl_mnxn(SHADER_OPCODE_ARG* arg) {
1195 int nComponents = 0;
1196 SHADER_OPCODE_ARG tmpArg;
1198 memset(&tmpArg, 0, sizeof(SHADER_OPCODE_ARG));
1200 /* Set constants for the temporary argument */
1201 tmpArg.shader = arg->shader;
1202 tmpArg.buffer = arg->buffer;
1203 tmpArg.src[0] = arg->src[0];
1204 tmpArg.src_addr[0] = arg->src_addr[0];
1205 tmpArg.src_addr[1] = arg->src_addr[1];
1206 tmpArg.reg_maps = arg->reg_maps;
1208 switch(arg->opcode->opcode) {
1209 case WINED3DSIO_M4x4:
1211 tmpArg.opcode = shader_get_opcode(arg->shader, WINED3DSIO_DP4);
1213 case WINED3DSIO_M4x3:
1215 tmpArg.opcode = shader_get_opcode(arg->shader, WINED3DSIO_DP4);
1217 case WINED3DSIO_M3x4:
1219 tmpArg.opcode = shader_get_opcode(arg->shader, WINED3DSIO_DP3);
1221 case WINED3DSIO_M3x3:
1223 tmpArg.opcode = shader_get_opcode(arg->shader, WINED3DSIO_DP3);
1225 case WINED3DSIO_M3x2:
1227 tmpArg.opcode = shader_get_opcode(arg->shader, WINED3DSIO_DP3);
1233 for (i = 0; i < nComponents; i++) {
1234 tmpArg.dst = ((arg->dst) & ~WINED3DSP_WRITEMASK_ALL)|(WINED3DSP_WRITEMASK_0<<i);
1235 tmpArg.src[1] = arg->src[1]+i;
1236 shader_glsl_dot(&tmpArg);
1241 The LRP instruction performs a component-wise linear interpolation
1242 between the second and third operands using the first operand as the
1243 blend factor. Equation: (dst = src2 + src0 * (src1 - src2))
1244 This is equivalent to mix(src2, src1, src0);
1246 void shader_glsl_lrp(SHADER_OPCODE_ARG* arg) {
1247 char src0_str[100], src1_str[100], src2_str[100];
1248 char src0_reg[50], src1_reg[50], src2_reg[50];
1249 char src0_mask[6], src1_mask[6], src2_mask[6];
1252 write_mask = shader_glsl_append_dst(arg->buffer, arg);
1254 shader_glsl_add_src_param(arg, arg->src[0], arg->src_addr[0], write_mask, src0_reg, src0_mask, src0_str);
1255 shader_glsl_add_src_param(arg, arg->src[1], arg->src_addr[1], write_mask, src1_reg, src1_mask, src1_str);
1256 shader_glsl_add_src_param(arg, arg->src[2], arg->src_addr[2], write_mask, src2_reg, src2_mask, src2_str);
1258 shader_addline(arg->buffer, "mix(%s, %s, %s));\n", src2_str, src1_str, src0_str);
1261 /** Process the WINED3DSIO_LIT instruction in GLSL:
1262 * dst.x = dst.w = 1.0
1263 * dst.y = (src0.x > 0) ? src0.x
1264 * dst.z = (src0.x > 0) ? ((src0.y > 0) ? pow(src0.y, src.w) : 0) : 0
1265 * where src.w is clamped at +- 128
1267 void shader_glsl_lit(SHADER_OPCODE_ARG* arg) {
1279 shader_glsl_append_dst(arg->buffer, arg);
1280 shader_glsl_get_write_mask(arg->dst, dst_mask);
1282 shader_glsl_add_src_param(arg, arg->src[0], arg->src_addr[0], WINED3DSP_WRITEMASK_0, src0_reg, src0_mask, src0_str);
1283 shader_glsl_add_src_param(arg, arg->src[0], arg->src_addr[0], WINED3DSP_WRITEMASK_1, src1_reg, src1_mask, src1_str);
1284 shader_glsl_add_src_param(arg, arg->src[0], arg->src_addr[0], WINED3DSP_WRITEMASK_3, src3_reg, src3_mask, src3_str);
1286 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",
1287 src0_str, src0_str, src0_str, src1_str, src1_str, src3_str, dst_mask);
1290 /** Process the WINED3DSIO_DST instruction in GLSL:
1292 * dst.y = src0.x * src0.y
1296 void shader_glsl_dst(SHADER_OPCODE_ARG* arg) {
1297 char src0y_str[100], src0z_str[100], src1y_str[100], src1w_str[100];
1298 char src0y_reg[50], src0z_reg[50], src1y_reg[50], src1w_reg[50];
1299 char dst_mask[6], src0y_mask[6], src0z_mask[6], src1y_mask[6], src1w_mask[6];
1301 shader_glsl_append_dst(arg->buffer, arg);
1302 shader_glsl_get_write_mask(arg->dst, dst_mask);
1304 shader_glsl_add_src_param(arg, arg->src[0], arg->src_addr[0], WINED3DSP_WRITEMASK_1, src0y_reg, src0y_mask, src0y_str);
1305 shader_glsl_add_src_param(arg, arg->src[0], arg->src_addr[0], WINED3DSP_WRITEMASK_2, src0z_reg, src0z_mask, src0z_str);
1306 shader_glsl_add_src_param(arg, arg->src[1], arg->src_addr[1], WINED3DSP_WRITEMASK_1, src1y_reg, src1y_mask, src1y_str);
1307 shader_glsl_add_src_param(arg, arg->src[1], arg->src_addr[1], WINED3DSP_WRITEMASK_3, src1w_reg, src1w_mask, src1w_str);
1309 shader_addline(arg->buffer, "vec4(1.0, %s * %s, %s, %s))%s;\n",
1310 src0y_str, src1y_str, src0z_str, src1w_str, dst_mask);
1313 /** Process the WINED3DSIO_SINCOS instruction in GLSL:
1314 * VS 2.0 requires that specific cosine and sine constants be passed to this instruction so the hardware
1315 * can handle it. But, these functions are built-in for GLSL, so we can just ignore the last 2 params.
1317 * dst.x = cos(src0.?)
1318 * dst.y = sin(src0.?)
1322 void shader_glsl_sincos(SHADER_OPCODE_ARG* arg) {
1328 write_mask = shader_glsl_append_dst(arg->buffer, arg);
1329 shader_glsl_add_src_param(arg, arg->src[0], arg->src_addr[0], WINED3DSP_WRITEMASK_0, src0_reg, src0_mask, src0_str);
1331 switch (write_mask) {
1332 case WINED3DSP_WRITEMASK_0:
1333 shader_addline(arg->buffer, "cos(%s));\n", src0_str);
1336 case WINED3DSP_WRITEMASK_1:
1337 shader_addline(arg->buffer, "sin(%s));\n", src0_str);
1340 case (WINED3DSP_WRITEMASK_0 | WINED3DSP_WRITEMASK_1):
1341 shader_addline(arg->buffer, "vec2(cos(%s), sin(%s)));\n", src0_str, src0_str);
1345 ERR("Write mask should be .x, .y or .xy\n");
1350 /** Process the WINED3DSIO_LOOP instruction in GLSL:
1351 * Start a for() loop where src1.y is the initial value of aL,
1352 * increment aL by src1.z for a total of src1.x iterations.
1353 * Need to use a temporary variable for this operation.
1355 /* FIXME: I don't think nested loops will work correctly this way. */
1356 void shader_glsl_loop(SHADER_OPCODE_ARG* arg) {
1362 shader_glsl_add_src_param(arg, arg->src[1], arg->src_addr[1], WINED3DSP_WRITEMASK_ALL, src1_reg, src1_mask, src1_str);
1364 shader_addline(arg->buffer, "for (tmpInt = 0, aL = %s.y; tmpInt < %s.x; tmpInt++, aL += %s.z) {\n",
1365 src1_reg, src1_reg, src1_reg);
1368 void shader_glsl_end(SHADER_OPCODE_ARG* arg) {
1369 shader_addline(arg->buffer, "}\n");
1372 void shader_glsl_rep(SHADER_OPCODE_ARG* arg) {
1378 shader_glsl_add_src_param(arg, arg->src[0], arg->src_addr[0], WINED3DSP_WRITEMASK_0, src0_reg, src0_mask, src0_str);
1379 shader_addline(arg->buffer, "for (tmpInt = 0; tmpInt < %s; tmpInt++) {\n", src0_str);
1382 void shader_glsl_if(SHADER_OPCODE_ARG* arg) {
1388 shader_glsl_add_src_param(arg, arg->src[0], arg->src_addr[0], WINED3DSP_WRITEMASK_0, src0_reg, src0_mask, src0_str);
1389 shader_addline(arg->buffer, "if (%s) {\n", src0_str);
1392 void shader_glsl_ifc(SHADER_OPCODE_ARG* arg) {
1394 char src0_str[100], src1_str[100];
1395 char src0_reg[50], src1_reg[50];
1396 char src0_mask[6], src1_mask[6];
1398 shader_glsl_add_src_param(arg, arg->src[0], arg->src_addr[0], WINED3DSP_WRITEMASK_0, src0_reg, src0_mask, src0_str);
1399 shader_glsl_add_src_param(arg, arg->src[1], arg->src_addr[1], WINED3DSP_WRITEMASK_0, src1_reg, src1_mask, src1_str);
1401 shader_addline(arg->buffer, "if (%s %s %s) {\n",
1402 src0_str, shader_get_comp_op(arg->opcode_token), src1_str);
1405 void shader_glsl_else(SHADER_OPCODE_ARG* arg) {
1406 shader_addline(arg->buffer, "} else {\n");
1409 void shader_glsl_break(SHADER_OPCODE_ARG* arg) {
1410 shader_addline(arg->buffer, "break;\n");
1413 /* FIXME: According to MSDN the compare is done per component. */
1414 void shader_glsl_breakc(SHADER_OPCODE_ARG* arg) {
1416 char src0_str[100], src1_str[100];
1417 char src0_reg[50], src1_reg[50];
1418 char src0_mask[6], src1_mask[6];
1420 shader_glsl_add_src_param(arg, arg->src[0], arg->src_addr[0], WINED3DSP_WRITEMASK_0, src0_reg, src0_mask, src0_str);
1421 shader_glsl_add_src_param(arg, arg->src[1], arg->src_addr[1], WINED3DSP_WRITEMASK_0, src1_reg, src1_mask, src1_str);
1423 shader_addline(arg->buffer, "if (%s %s %s) break;\n",
1424 src0_str, shader_get_comp_op(arg->opcode_token), src1_str);
1427 void shader_glsl_label(SHADER_OPCODE_ARG* arg) {
1429 DWORD snum = (arg->src[0]) & WINED3DSP_REGNUM_MASK;
1430 shader_addline(arg->buffer, "}\n");
1431 shader_addline(arg->buffer, "void subroutine%lu () {\n", snum);
1434 void shader_glsl_call(SHADER_OPCODE_ARG* arg) {
1435 DWORD snum = (arg->src[0]) & WINED3DSP_REGNUM_MASK;
1436 shader_addline(arg->buffer, "subroutine%lu();\n", snum);
1439 void shader_glsl_callnz(SHADER_OPCODE_ARG* arg) {
1445 DWORD snum = (arg->src[0]) & WINED3DSP_REGNUM_MASK;
1446 shader_glsl_add_src_param(arg, arg->src[1], arg->src_addr[1], WINED3DSP_WRITEMASK_0, src1_reg, src1_mask, src1_str);
1447 shader_addline(arg->buffer, "if (%s) subroutine%lu();\n", src1_str, snum);
1450 /*********************************************
1451 * Pixel Shader Specific Code begins here
1452 ********************************************/
1453 void pshader_glsl_tex(SHADER_OPCODE_ARG* arg) {
1454 IWineD3DPixelShaderImpl* This = (IWineD3DPixelShaderImpl*) arg->shader;
1455 DWORD hex_version = This->baseShader.hex_version;
1456 char coord_str[100], coord_reg[50], coord_mask[6];
1457 char dst_swizzle[6];
1458 glsl_sample_function_t sample_function;
1464 /* All versions have a destination register */
1465 shader_glsl_append_dst(arg->buffer, arg);
1467 /* 1.0-1.4: Use destination register as sampler source.
1468 * 2.0+: Use provided sampler source. */
1469 if (hex_version < WINED3DPS_VERSION(1,4)) {
1470 IWineD3DDeviceImpl* deviceImpl = (IWineD3DDeviceImpl*) This->baseShader.device;
1473 sampler_idx = arg->dst & WINED3DSP_REGNUM_MASK;
1474 flags = deviceImpl->stateBlock->textureState[sampler_idx][WINED3DTSS_TEXTURETRANSFORMFLAGS];
1476 if (flags & WINED3DTTFF_PROJECTED) {
1478 switch (flags & ~WINED3DTTFF_PROJECTED) {
1479 case WINED3DTTFF_COUNT1: FIXME("WINED3DTTFF_PROJECTED with WINED3DTTFF_COUNT1?\n"); break;
1480 case WINED3DTTFF_COUNT2: mask = WINED3DSP_WRITEMASK_1; break;
1481 case WINED3DTTFF_COUNT3: mask = WINED3DSP_WRITEMASK_2; break;
1482 case WINED3DTTFF_COUNT4:
1483 case WINED3DTTFF_DISABLE: mask = WINED3DSP_WRITEMASK_3; break;
1488 } else if (hex_version < WINED3DPS_VERSION(2,0)) {
1489 DWORD src_mod = arg->src[0] & WINED3DSP_SRCMOD_MASK;
1490 sampler_idx = arg->dst & WINED3DSP_REGNUM_MASK;
1492 if (src_mod == WINED3DSPSM_DZ) {
1494 mask = WINED3DSP_WRITEMASK_2;
1495 } else if (src_mod == WINED3DSPSM_DW) {
1497 mask = WINED3DSP_WRITEMASK_3;
1502 sampler_idx = arg->src[1] & WINED3DSP_REGNUM_MASK;
1503 /* TODO: Handle D3DSI_TEXLD_PROJECTED... */
1507 sampler_type = arg->reg_maps->samplers[sampler_idx] & WINED3DSP_TEXTURETYPE_MASK;
1508 shader_glsl_get_sample_function(sampler_type, projected, &sample_function);
1509 mask |= sample_function.coord_mask;
1511 if (hex_version < WINED3DPS_VERSION(2,0)) {
1512 shader_glsl_get_write_mask(arg->dst, dst_swizzle);
1514 shader_glsl_get_swizzle(arg->src[1], FALSE, arg->dst, dst_swizzle);
1517 /* 1.0-1.3: Use destination register as coordinate source.
1518 1.4+: Use provided coordinate source register. */
1519 if (hex_version < WINED3DPS_VERSION(1,4)) {
1520 shader_glsl_get_write_mask(mask, coord_mask);
1521 shader_addline(arg->buffer, "%s(Psampler%u, T%u%s)%s);\n",
1522 sample_function.name, sampler_idx, sampler_idx, coord_mask, dst_swizzle);
1524 shader_glsl_add_src_param(arg, arg->src[0], arg->src_addr[0], mask, coord_reg, coord_mask, coord_str);
1525 shader_addline(arg->buffer, "%s(Psampler%u, %s)%s);\n",
1526 sample_function.name, sampler_idx, coord_str, dst_swizzle);
1530 void pshader_glsl_texcoord(SHADER_OPCODE_ARG* arg) {
1532 /* FIXME: Make this work for more than just 2D textures */
1534 IWineD3DPixelShaderImpl* This = (IWineD3DPixelShaderImpl*) arg->shader;
1535 SHADER_BUFFER* buffer = arg->buffer;
1536 DWORD hex_version = This->baseShader.hex_version;
1540 write_mask = shader_glsl_append_dst(arg->buffer, arg);
1541 shader_glsl_get_write_mask(write_mask, dst_mask);
1543 if (hex_version != WINED3DPS_VERSION(1,4)) {
1544 DWORD reg = arg->dst & WINED3DSP_REGNUM_MASK;
1545 shader_addline(buffer, "clamp(gl_TexCoord[%u], 0.0, 1.0)%s);\n", reg, dst_mask);
1547 DWORD reg = arg->src[0] & WINED3DSP_REGNUM_MASK;
1548 DWORD src_mod = arg->src[0] & WINED3DSP_SRCMOD_MASK;
1549 char dst_swizzle[6];
1551 shader_glsl_get_swizzle(arg->src[0], FALSE, write_mask, dst_swizzle);
1553 if (src_mod == WINED3DSPSM_DZ) {
1554 char div_str[100], div_reg[50], div_mask[6];
1555 size_t mask_size = shader_glsl_get_write_mask_size(write_mask);
1556 shader_glsl_add_src_param(arg, arg->src[0], arg->src_addr[0], WINED3DSP_WRITEMASK_2, div_reg, div_mask, div_str);
1558 if (mask_size > 1) {
1559 shader_addline(buffer, "gl_TexCoord[%u]%s / vec%d(%s));\n", reg, dst_swizzle, mask_size, div_str);
1561 shader_addline(buffer, "gl_TexCoord[%u]%s / %s);\n", reg, dst_swizzle, div_str);
1563 } else if (src_mod == WINED3DSPSM_DW) {
1564 char div_str[100], div_reg[50], div_mask[6];
1565 size_t mask_size = shader_glsl_get_write_mask_size(write_mask);
1566 shader_glsl_add_src_param(arg, arg->src[0], arg->src_addr[0], WINED3DSP_WRITEMASK_3, div_reg, div_mask, div_str);
1568 if (mask_size > 1) {
1569 shader_addline(buffer, "gl_TexCoord[%u]%s / vec%d(%s));\n", reg, dst_swizzle, mask_size, div_str);
1571 shader_addline(buffer, "gl_TexCoord[%u]%s / %s);\n", reg, dst_swizzle, div_str);
1574 shader_addline(buffer, "gl_TexCoord[%u]%s);\n", reg, dst_swizzle);
1579 /** Process the WINED3DSIO_TEXDP3TEX instruction in GLSL:
1580 * Take a 3-component dot product of the TexCoord[dstreg] and src,
1581 * then perform a 1D texture lookup from stage dstregnum, place into dst. */
1582 void pshader_glsl_texdp3tex(SHADER_OPCODE_ARG* arg) {
1587 DWORD sampler_idx = arg->dst & WINED3DSP_REGNUM_MASK;
1588 DWORD src_mask = WINED3DSP_WRITEMASK_0 | WINED3DSP_WRITEMASK_1 | WINED3DSP_WRITEMASK_2;
1590 shader_glsl_add_src_param(arg, arg->src[0], arg->src_addr[0], src_mask, src0_name, src0_mask, src0_str);
1592 shader_glsl_append_dst(arg->buffer, arg);
1593 shader_glsl_get_write_mask(arg->dst, dst_mask);
1594 shader_addline(arg->buffer, "texture1D(Psampler%u, dot(gl_TexCoord[%u].xyz, %s))%s);\n",
1595 sampler_idx, sampler_idx, src0_str, dst_mask);
1598 /** Process the WINED3DSIO_TEXDP3 instruction in GLSL:
1599 * Take a 3-component dot product of the TexCoord[dstreg] and src. */
1600 void pshader_glsl_texdp3(SHADER_OPCODE_ARG* arg) {
1604 DWORD dstreg = arg->dst & WINED3DSP_REGNUM_MASK;
1605 DWORD src_mask = WINED3DSP_WRITEMASK_0 | WINED3DSP_WRITEMASK_1 | WINED3DSP_WRITEMASK_2;
1609 dst_mask = shader_glsl_append_dst(arg->buffer, arg);
1610 mask_size = shader_glsl_get_write_mask_size(dst_mask);
1611 shader_glsl_add_src_param(arg, arg->src[0], arg->src_addr[0], src_mask, src0_name, src0_mask, src0_str);
1613 if (mask_size > 1) {
1614 shader_addline(arg->buffer, "vec%d(dot(T%u.xyz, %s)));\n", mask_size, dstreg, src0_str);
1616 shader_addline(arg->buffer, "dot(T%u.xyz, %s));\n", dstreg, src0_str);
1620 /** Process the WINED3DSIO_TEXDEPTH instruction in GLSL:
1621 * Calculate the depth as dst.x / dst.y */
1622 void pshader_glsl_texdepth(SHADER_OPCODE_ARG* arg) {
1628 shader_glsl_add_dst_param(arg, arg->dst, 0, dst_reg, dst_mask, dst_str);
1630 shader_addline(arg->buffer, "gl_FragDepth = %s.x / %s.y;\n", dst_reg, dst_reg);
1633 /** Process the WINED3DSIO_TEXM3X2DEPTH instruction in GLSL:
1634 * Last row of a 3x2 matrix multiply, use the result to calculate the depth:
1635 * Calculate tmp0.y = TexCoord[dstreg] . src.xyz; (tmp0.x has already been calculated)
1636 * depth = (tmp0.y == 0.0) ? 1.0 : tmp0.x / tmp0.y
1638 void pshader_glsl_texm3x2depth(SHADER_OPCODE_ARG* arg) {
1639 DWORD src_mask = WINED3DSP_WRITEMASK_0 | WINED3DSP_WRITEMASK_1 | WINED3DSP_WRITEMASK_2;
1640 DWORD dstreg = arg->dst & WINED3DSP_REGNUM_MASK;
1641 char src0_str[100], src0_name[50], src0_mask[6];
1643 shader_glsl_add_src_param(arg, arg->src[0], arg->src_addr[0], src_mask, src0_name, src0_mask, src0_str);
1645 shader_addline(arg->buffer, "tmp0.y = dot(T%u.xyz, %s);\n", dstreg, src0_str);
1646 shader_addline(arg->buffer, "gl_FragDepth = (tmp0.y == 0.0) ? 1.0 : clamp(tmp0.x / tmp0.y, 0.0, 1.0);\n");
1649 /** Process the WINED3DSIO_TEXM3X2PAD instruction in GLSL
1650 * Calculate the 1st of a 2-row matrix multiplication. */
1651 void pshader_glsl_texm3x2pad(SHADER_OPCODE_ARG* arg) {
1652 DWORD src_mask = WINED3DSP_WRITEMASK_0 | WINED3DSP_WRITEMASK_1 | WINED3DSP_WRITEMASK_2;
1653 DWORD reg = arg->dst & WINED3DSP_REGNUM_MASK;
1654 SHADER_BUFFER* buffer = arg->buffer;
1659 shader_glsl_add_src_param(arg, arg->src[0], arg->src_addr[0], src_mask, src0_name, src0_mask, src0_str);
1660 shader_addline(buffer, "tmp0.x = dot(T%u.xyz, %s);\n", reg, src0_str);
1663 /** Process the WINED3DSIO_TEXM3X3PAD instruction in GLSL
1664 * Calculate the 1st or 2nd row of a 3-row matrix multiplication. */
1665 void pshader_glsl_texm3x3pad(SHADER_OPCODE_ARG* arg) {
1667 IWineD3DPixelShaderImpl* shader = (IWineD3DPixelShaderImpl*) arg->shader;
1668 DWORD src_mask = WINED3DSP_WRITEMASK_0 | WINED3DSP_WRITEMASK_1 | WINED3DSP_WRITEMASK_2;
1669 DWORD reg = arg->dst & WINED3DSP_REGNUM_MASK;
1670 SHADER_BUFFER* buffer = arg->buffer;
1671 SHADER_PARSE_STATE* current_state = &shader->baseShader.parse_state;
1676 shader_glsl_add_src_param(arg, arg->src[0], arg->src_addr[0], src_mask, src0_name, src0_mask, src0_str);
1677 shader_addline(buffer, "tmp0.%c = dot(T%u.xyz, %s);\n", 'x' + current_state->current_row, reg, src0_str);
1678 current_state->texcoord_w[current_state->current_row++] = reg;
1681 void pshader_glsl_texm3x2tex(SHADER_OPCODE_ARG* arg) {
1682 DWORD src_mask = WINED3DSP_WRITEMASK_0 | WINED3DSP_WRITEMASK_1 | WINED3DSP_WRITEMASK_2;
1683 DWORD reg = arg->dst & WINED3DSP_REGNUM_MASK;
1684 SHADER_BUFFER* buffer = arg->buffer;
1690 shader_glsl_add_src_param(arg, arg->src[0], arg->src_addr[0], src_mask, src0_name, src0_mask, src0_str);
1691 shader_addline(buffer, "tmp0.y = dot(T%u.xyz, %s);\n", reg, src0_str);
1693 shader_glsl_append_dst(buffer, arg);
1694 shader_glsl_get_write_mask(arg->dst, dst_mask);
1696 /* Sample the texture using the calculated coordinates */
1697 shader_addline(buffer, "texture2D(Psampler%u, tmp0.xy)%s);\n", reg, dst_mask);
1700 /** Process the WINED3DSIO_TEXM3X3TEX instruction in GLSL
1701 * Perform the 3rd row of a 3x3 matrix multiply, then sample the texture using the calculated coordinates */
1702 void pshader_glsl_texm3x3tex(SHADER_OPCODE_ARG* arg) {
1703 DWORD src_mask = WINED3DSP_WRITEMASK_0 | WINED3DSP_WRITEMASK_1 | WINED3DSP_WRITEMASK_2;
1708 DWORD reg = arg->dst & WINED3DSP_REGNUM_MASK;
1709 IWineD3DPixelShaderImpl* This = (IWineD3DPixelShaderImpl*) arg->shader;
1710 SHADER_PARSE_STATE* current_state = &This->baseShader.parse_state;
1711 DWORD sampler_type = arg->reg_maps->samplers[reg] & WINED3DSP_TEXTURETYPE_MASK;
1712 glsl_sample_function_t sample_function;
1714 shader_glsl_add_src_param(arg, arg->src[0], arg->src_addr[0], src_mask, src0_name, src0_mask, src0_str);
1715 shader_addline(arg->buffer, "tmp0.z = dot(T%u.xyz, %s);\n", reg, src0_str);
1717 shader_glsl_append_dst(arg->buffer, arg);
1718 shader_glsl_get_write_mask(arg->dst, dst_mask);
1719 shader_glsl_get_sample_function(sampler_type, FALSE, &sample_function);
1721 /* Sample the texture using the calculated coordinates */
1722 shader_addline(arg->buffer, "%s(Psampler%u, tmp0.xyz)%s);\n", sample_function.name, reg, dst_mask);
1724 current_state->current_row = 0;
1727 /** Process the WINED3DSIO_TEXM3X3 instruction in GLSL
1728 * Perform the 3rd row of a 3x3 matrix multiply */
1729 void pshader_glsl_texm3x3(SHADER_OPCODE_ARG* arg) {
1730 DWORD src_mask = WINED3DSP_WRITEMASK_0 | WINED3DSP_WRITEMASK_1 | WINED3DSP_WRITEMASK_2;
1735 DWORD reg = arg->dst & WINED3DSP_REGNUM_MASK;
1736 IWineD3DPixelShaderImpl* This = (IWineD3DPixelShaderImpl*) arg->shader;
1737 SHADER_PARSE_STATE* current_state = &This->baseShader.parse_state;
1739 shader_glsl_add_src_param(arg, arg->src[0], arg->src_addr[0], src_mask, src0_name, src0_mask, src0_str);
1741 shader_glsl_append_dst(arg->buffer, arg);
1742 shader_glsl_get_write_mask(arg->dst, dst_mask);
1743 shader_addline(arg->buffer, "vec4(tmp.xy, dot(T%u.xyz, %s), 1.0)%s);\n", reg, src0_str, dst_mask);
1745 current_state->current_row = 0;
1748 /** Process the WINED3DSIO_TEXM3X3SPEC instruction in GLSL
1749 * Peform the final texture lookup based on the previous 2 3x3 matrix multiplies */
1750 void pshader_glsl_texm3x3spec(SHADER_OPCODE_ARG* arg) {
1752 IWineD3DPixelShaderImpl* shader = (IWineD3DPixelShaderImpl*) arg->shader;
1753 DWORD reg = arg->dst & WINED3DSP_REGNUM_MASK;
1754 char src0_str[100], src0_name[50], src0_mask[6];
1755 char src1_str[100], src1_name[50], src1_mask[6];
1757 SHADER_BUFFER* buffer = arg->buffer;
1758 SHADER_PARSE_STATE* current_state = &shader->baseShader.parse_state;
1759 DWORD stype = arg->reg_maps->samplers[reg] & WINED3DSP_TEXTURETYPE_MASK;
1760 DWORD src_mask = WINED3DSP_WRITEMASK_0 | WINED3DSP_WRITEMASK_1 | WINED3DSP_WRITEMASK_2;
1761 glsl_sample_function_t sample_function;
1763 shader_glsl_add_src_param(arg, arg->src[0], arg->src_addr[0], src_mask, src0_name, src0_mask, src0_str);
1764 shader_glsl_add_src_param(arg, arg->src[1], arg->src_addr[1], src_mask, src1_name, src1_mask, src1_str);
1766 /* Perform the last matrix multiply operation */
1767 shader_addline(buffer, "tmp0.z = dot(T%u.xyz, %s);\n", reg, src0_str);
1769 /* Calculate reflection vector, 2*(tmp0.src1)*tmp0-src1
1770 * This is equavalent to reflect(-src1, tmp0); */
1771 shader_addline(buffer, "tmp0.xyz = reflect(-(%s), tmp0.xyz);\n", src1_str);
1773 shader_glsl_append_dst(buffer, arg);
1774 shader_glsl_get_write_mask(arg->dst, dst_mask);
1775 shader_glsl_get_sample_function(stype, FALSE, &sample_function);
1777 /* Sample the texture */
1778 shader_addline(buffer, "%s(Psampler%u, tmp0.xyz)%s);\n", sample_function.name, reg, dst_mask);
1780 current_state->current_row = 0;
1783 /** Process the WINED3DSIO_TEXM3X3VSPEC instruction in GLSL
1784 * Peform the final texture lookup based on the previous 2 3x3 matrix multiplies */
1785 void pshader_glsl_texm3x3vspec(SHADER_OPCODE_ARG* arg) {
1787 IWineD3DPixelShaderImpl* shader = (IWineD3DPixelShaderImpl*) arg->shader;
1788 DWORD reg = arg->dst & WINED3DSP_REGNUM_MASK;
1789 SHADER_BUFFER* buffer = arg->buffer;
1790 SHADER_PARSE_STATE* current_state = &shader->baseShader.parse_state;
1791 char src0_str[100], src0_name[50], src0_mask[6], dst_mask[6];
1792 DWORD src_mask = WINED3DSP_WRITEMASK_0 | WINED3DSP_WRITEMASK_1 | WINED3DSP_WRITEMASK_2;
1793 DWORD sampler_type = arg->reg_maps->samplers[reg] & WINED3DSP_TEXTURETYPE_MASK;
1794 glsl_sample_function_t sample_function;
1796 shader_glsl_add_src_param(arg, arg->src[0], arg->src_addr[0], src_mask, src0_name, src0_mask, src0_str);
1798 /* Perform the last matrix multiply operation */
1799 shader_addline(buffer, "tmp0.z = dot(vec3(T%u), vec3(%s));\n", reg, src0_str);
1801 /* Construct the eye-ray vector from w coordinates */
1802 shader_addline(buffer, "tmp1.xyz = normalize(vec3(gl_TexCoord[%u].w, gl_TexCoord[%u].w, gl_TexCoord[%u].w));\n",
1803 current_state->texcoord_w[0], current_state->texcoord_w[1], reg);
1805 /* Calculate reflection vector (Assume normal is normalized): RF = 2*(tmp0.tmp1)*tmp0-tmp1
1806 * This is equavalent to reflect(-tmp1, tmp0); */
1807 shader_addline(buffer, "tmp0.xyz = reflect(-tmp1.xyz, tmp0.xyz);\n");
1809 shader_glsl_append_dst(buffer, arg);
1810 shader_glsl_get_write_mask(arg->dst, dst_mask);
1811 shader_glsl_get_sample_function(sampler_type, FALSE, &sample_function);
1813 /* Sample the texture using the calculated coordinates */
1814 shader_addline(buffer, "%s(Psampler%u, tmp0.xyz)%s);\n", sample_function.name, reg, dst_mask);
1816 current_state->current_row = 0;
1819 /** Process the WINED3DSIO_TEXBEM instruction in GLSL.
1820 * Apply a fake bump map transform.
1821 * FIXME: Should apply the BUMPMAPENV matrix. For now, just sample the texture */
1822 void pshader_glsl_texbem(SHADER_OPCODE_ARG* arg) {
1824 DWORD reg1 = arg->dst & WINED3DSP_REGNUM_MASK;
1825 DWORD reg2 = arg->src[0] & WINED3DSP_REGNUM_MASK;
1827 FIXME("Not applying the BUMPMAPENV matrix for pixel shader instruction texbem.\n");
1828 shader_addline(arg->buffer, "T%u = texture2D(Psampler%u, gl_TexCoord[%u].xy + T%u.xy);\n",
1829 reg1, reg1, reg1, reg2);
1832 /** Process the WINED3DSIO_TEXREG2AR instruction in GLSL
1833 * Sample 2D texture at dst using the alpha & red (wx) components of src as texture coordinates */
1834 void pshader_glsl_texreg2ar(SHADER_OPCODE_ARG* arg) {
1838 DWORD sampler_idx = arg->dst & WINED3DSP_REGNUM_MASK;
1841 shader_glsl_append_dst(arg->buffer, arg);
1842 shader_glsl_get_write_mask(arg->dst, dst_mask);
1843 shader_glsl_add_src_param(arg, arg->src[0], arg->src_addr[0], WINED3DSP_WRITEMASK_ALL, src0_reg, src0_mask, src0_str);
1845 shader_addline(arg->buffer, "texture2D(Psampler%u, %s.wx)%s);\n", sampler_idx, src0_reg, dst_mask);
1848 /** Process the WINED3DSIO_TEXREG2GB instruction in GLSL
1849 * Sample 2D texture at dst using the green & blue (yz) components of src as texture coordinates */
1850 void pshader_glsl_texreg2gb(SHADER_OPCODE_ARG* arg) {
1854 DWORD sampler_idx = arg->dst & WINED3DSP_REGNUM_MASK;
1857 shader_glsl_append_dst(arg->buffer, arg);
1858 shader_glsl_get_write_mask(arg->dst, dst_mask);
1859 shader_glsl_add_src_param(arg, arg->src[0], arg->src_addr[0], WINED3DSP_WRITEMASK_ALL, src0_reg, src0_mask, src0_str);
1861 shader_addline(arg->buffer, "texture2D(Psampler%u, %s.yz)%s);\n", sampler_idx, src0_reg, dst_mask);
1864 /** Process the WINED3DSIO_TEXREG2RGB instruction in GLSL
1865 * Sample texture at dst using the rgb (xyz) components of src as texture coordinates */
1866 void pshader_glsl_texreg2rgb(SHADER_OPCODE_ARG* arg) {
1871 DWORD sampler_idx = arg->dst & WINED3DSP_REGNUM_MASK;
1872 DWORD sampler_type = arg->reg_maps->samplers[sampler_idx] & WINED3DSP_TEXTURETYPE_MASK;
1873 glsl_sample_function_t sample_function;
1875 shader_glsl_append_dst(arg->buffer, arg);
1876 shader_glsl_get_write_mask(arg->dst, dst_mask);
1877 shader_glsl_get_sample_function(sampler_type, FALSE, &sample_function);
1878 shader_glsl_add_src_param(arg, arg->src[0], arg->src_addr[0], sample_function.coord_mask, src0_reg, src0_mask, src0_str);
1880 shader_addline(arg->buffer, "%s(Psampler%u, %s)%s);\n", sample_function.name, sampler_idx, src0_str, dst_mask);
1883 /** Process the WINED3DSIO_TEXKILL instruction in GLSL.
1884 * If any of the first 3 components are < 0, discard this pixel */
1885 void pshader_glsl_texkill(SHADER_OPCODE_ARG* arg) {
1887 char dst_str[100], dst_name[50], dst_mask[6];
1889 shader_glsl_add_dst_param(arg, arg->dst, 0, dst_name, dst_mask, dst_str);
1890 shader_addline(arg->buffer, "if (any(lessThan(%s.xyz, vec3(0.0)))) discard;\n", dst_name);
1893 /** Process the WINED3DSIO_DP2ADD instruction in GLSL.
1894 * dst = dot2(src0, src1) + src2 */
1895 void pshader_glsl_dp2add(SHADER_OPCODE_ARG* arg) {
1896 char src0_str[100], src1_str[100], src2_str[100];
1897 char src0_reg[50], src1_reg[50], src2_reg[50];
1898 char src0_mask[6], src1_mask[6], src2_mask[6];
1902 write_mask = shader_glsl_append_dst(arg->buffer, arg);
1903 mask_size = shader_glsl_get_write_mask_size(write_mask);
1905 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);
1906 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);
1907 shader_glsl_add_src_param(arg, arg->src[2], arg->src_addr[2], WINED3DSP_WRITEMASK_0, src2_reg, src2_mask, src2_str);
1909 shader_addline(arg->buffer, "dot(%s, %s) + %s);\n", src0_str, src1_str, src2_str);
1912 void pshader_glsl_input_pack(
1913 SHADER_BUFFER* buffer,
1914 semantic* semantics_in) {
1918 for (i = 0; i < MAX_REG_INPUT; i++) {
1920 DWORD usage_token = semantics_in[i].usage;
1921 DWORD register_token = semantics_in[i].reg;
1922 DWORD usage, usage_idx;
1926 if (!usage_token) continue;
1927 usage = (usage_token & WINED3DSP_DCL_USAGE_MASK) >> WINED3DSP_DCL_USAGE_SHIFT;
1928 usage_idx = (usage_token & WINED3DSP_DCL_USAGEINDEX_MASK) >> WINED3DSP_DCL_USAGEINDEX_SHIFT;
1929 shader_glsl_get_write_mask(register_token, reg_mask);
1933 case D3DDECLUSAGE_COLOR:
1935 shader_addline(buffer, "IN%u%s = vec4(gl_Color)%s;\n",
1936 i, reg_mask, reg_mask);
1937 else if (usage_idx == 1)
1938 shader_addline(buffer, "IN%u%s = vec4(gl_SecondaryColor)%s;\n",
1939 i, reg_mask, reg_mask);
1941 shader_addline(buffer, "IN%u%s = vec4(unsupported_color_input)%s;\n",
1942 i, reg_mask, reg_mask);
1945 case D3DDECLUSAGE_TEXCOORD:
1946 shader_addline(buffer, "IN%u%s = vec4(gl_TexCoord[%u])%s;\n",
1947 i, reg_mask, usage_idx, reg_mask );
1950 case D3DDECLUSAGE_FOG:
1951 shader_addline(buffer, "IN%u%s = vec4(gl_FogFragCoord)%s;\n",
1952 i, reg_mask, reg_mask);
1956 shader_addline(buffer, "IN%u%s = vec4(unsupported_input)%s;\n",
1957 i, reg_mask, reg_mask);
1962 /*********************************************
1963 * Vertex Shader Specific Code begins here
1964 ********************************************/
1966 void vshader_glsl_output_unpack(
1967 SHADER_BUFFER* buffer,
1968 semantic* semantics_out) {
1972 for (i = 0; i < MAX_REG_OUTPUT; i++) {
1974 DWORD usage_token = semantics_out[i].usage;
1975 DWORD register_token = semantics_out[i].reg;
1976 DWORD usage, usage_idx;
1980 if (!usage_token) continue;
1982 usage = (usage_token & WINED3DSP_DCL_USAGE_MASK) >> WINED3DSP_DCL_USAGE_SHIFT;
1983 usage_idx = (usage_token & WINED3DSP_DCL_USAGEINDEX_MASK) >> WINED3DSP_DCL_USAGEINDEX_SHIFT;
1984 shader_glsl_get_write_mask(register_token, reg_mask);
1988 case D3DDECLUSAGE_COLOR:
1990 shader_addline(buffer, "gl_FrontColor%s = OUT%u%s;\n", reg_mask, i, reg_mask);
1991 else if (usage_idx == 1)
1992 shader_addline(buffer, "gl_FrontSecondaryColor%s = OUT%u%s;\n", reg_mask, i, reg_mask);
1994 shader_addline(buffer, "unsupported_color_output%s = OUT%u%s;\n", reg_mask, i, reg_mask);
1997 case D3DDECLUSAGE_POSITION:
1998 shader_addline(buffer, "gl_Position%s = OUT%u%s;\n", reg_mask, i, reg_mask);
2001 case D3DDECLUSAGE_TEXCOORD:
2002 shader_addline(buffer, "gl_TexCoord[%u]%s = OUT%u%s;\n",
2003 usage_idx, reg_mask, i, reg_mask);
2006 case WINED3DSHADERDECLUSAGE_PSIZE:
2007 shader_addline(buffer, "gl_PointSize = OUT%u.x;\n", i);
2010 case WINED3DSHADERDECLUSAGE_FOG:
2011 shader_addline(buffer, "gl_FogFragCoord%s = OUT%u%s;\n", reg_mask, i, reg_mask);
2015 shader_addline(buffer, "unsupported_output%s = OUT%u%s;\n", reg_mask, i, reg_mask);
2020 /** Attach a GLSL pixel or vertex shader object to the shader program */
2021 static void attach_glsl_shader(IWineD3DDevice *iface, IWineD3DBaseShader* shader) {
2022 IWineD3DDeviceImpl *This = (IWineD3DDeviceImpl *)iface;
2023 WineD3D_GL_Info *gl_info = &((IWineD3DImpl *)(This->wineD3D))->gl_info;
2024 GLhandleARB shaderObj = ((IWineD3DBaseShaderImpl*)shader)->baseShader.prgId;
2025 if (This->stateBlock->glsl_program && shaderObj != 0) {
2026 TRACE("Attaching GLSL shader object %u to program %u\n", shaderObj, This->stateBlock->glsl_program->programId);
2027 GL_EXTCALL(glAttachObjectARB(This->stateBlock->glsl_program->programId, shaderObj));
2028 checkGLcall("glAttachObjectARB");
2032 /** Sets the GLSL program ID for the given pixel and vertex shader combination.
2033 * It sets the programId on the current StateBlock (because it should be called
2034 * inside of the DrawPrimitive() part of the render loop).
2036 * If a program for the given combination does not exist, create one, and store
2037 * the program in the list. If it creates a program, it will link the given
2040 * We keep the shader programs around on a list because linking
2041 * shader objects together is an expensive operation. It's much
2042 * faster to loop through a list of pre-compiled & linked programs
2043 * each time that the application sets a new pixel or vertex shader
2044 * than it is to re-link them together at that time.
2046 * The list will be deleted in IWineD3DDevice::Release().
2048 static void set_glsl_shader_program(IWineD3DDevice *iface) {
2049 IWineD3DDeviceImpl *This = (IWineD3DDeviceImpl *)iface;
2050 WineD3D_GL_Info *gl_info = &((IWineD3DImpl *)(This->wineD3D))->gl_info;
2051 IWineD3DPixelShader *pshader = This->stateBlock->pixelShader;
2052 IWineD3DVertexShader *vshader = This->stateBlock->vertexShader;
2053 struct glsl_shader_prog_link *curLink = NULL;
2054 struct glsl_shader_prog_link *newLink = NULL;
2055 struct list *ptr = NULL;
2056 GLhandleARB programId = 0;
2060 ptr = list_head( &This->glsl_shader_progs );
2062 /* At least one program exists - see if it matches our ps/vs combination */
2063 curLink = LIST_ENTRY( ptr, struct glsl_shader_prog_link, entry );
2064 if (vshader == curLink->vertexShader && pshader == curLink->pixelShader) {
2065 /* Existing Program found, use it */
2066 TRACE("Found existing program (%u) for this vertex/pixel shader combination\n",
2067 curLink->programId);
2068 This->stateBlock->glsl_program = curLink;
2071 /* This isn't the entry we need - try the next one */
2072 ptr = list_next( &This->glsl_shader_progs, ptr );
2075 /* If we get to this point, then no matching program exists, so we create one */
2076 programId = GL_EXTCALL(glCreateProgramObjectARB());
2077 TRACE("Created new GLSL shader program %u\n", programId);
2079 /* Allocate a new link for the list of programs */
2080 newLink = HeapAlloc(GetProcessHeap(), 0, sizeof(struct glsl_shader_prog_link));
2081 newLink->programId = programId;
2082 This->stateBlock->glsl_program = newLink;
2084 /* Attach GLSL vshader */
2085 if (NULL != vshader && This->vs_selected_mode == SHADER_GLSL) {
2087 int max_attribs = 16; /* TODO: Will this always be the case? It is at the moment... */
2090 TRACE("Attaching vertex shader to GLSL program\n");
2091 attach_glsl_shader(iface, (IWineD3DBaseShader*)vshader);
2093 /* Bind vertex attributes to a corresponding index number to match
2094 * the same index numbers as ARB_vertex_programs (makes loading
2095 * vertex attributes simpler). With this method, we can use the
2096 * exact same code to load the attributes later for both ARB and
2099 * We have to do this here because we need to know the Program ID
2100 * in order to make the bindings work, and it has to be done prior
2101 * to linking the GLSL program. */
2102 for (i = 0; i < max_attribs; ++i) {
2103 snprintf(tmp_name, sizeof(tmp_name), "attrib%i", i);
2104 GL_EXTCALL(glBindAttribLocationARB(programId, i, tmp_name));
2106 checkGLcall("glBindAttribLocationARB");
2107 newLink->vertexShader = vshader;
2110 /* Attach GLSL pshader */
2111 if (NULL != pshader && This->ps_selected_mode == SHADER_GLSL) {
2112 TRACE("Attaching pixel shader to GLSL program\n");
2113 attach_glsl_shader(iface, (IWineD3DBaseShader*)pshader);
2114 newLink->pixelShader = pshader;
2117 /* Link the program */
2118 TRACE("Linking GLSL shader program %u\n", programId);
2119 GL_EXTCALL(glLinkProgramARB(programId));
2120 print_glsl_info_log(&GLINFO_LOCATION, programId);
2121 list_add_head( &This->glsl_shader_progs, &newLink->entry);
2123 newLink->vuniformF_locations = HeapAlloc(GetProcessHeap(), 0, sizeof(GLhandleARB) * GL_LIMITS(vshader_constantsF));
2124 for (i = 0; i < GL_LIMITS(vshader_constantsF); ++i) {
2125 snprintf(glsl_name, sizeof(glsl_name), "VC[%i]", i);
2126 newLink->vuniformF_locations[i] = GL_EXTCALL(glGetUniformLocationARB(programId, glsl_name));
2128 newLink->puniformF_locations = HeapAlloc(GetProcessHeap(), 0, sizeof(GLhandleARB) * GL_LIMITS(pshader_constantsF));
2129 for (i = 0; i < GL_LIMITS(pshader_constantsF); ++i) {
2130 snprintf(glsl_name, sizeof(glsl_name), "PC[%i]", i);
2131 newLink->puniformF_locations[i] = GL_EXTCALL(glGetUniformLocationARB(programId, glsl_name));
2137 static GLhandleARB create_glsl_blt_shader(WineD3D_GL_Info *gl_info) {
2138 GLhandleARB program_id;
2139 GLhandleARB vshader_id, pshader_id;
2140 const char *blt_vshader[] = {
2143 " gl_Position = gl_Vertex;\n"
2144 " gl_FrontColor = vec4(1.0);\n"
2145 " gl_TexCoord[0].x = (gl_Vertex.x * 0.5) + 0.5;\n"
2146 " gl_TexCoord[0].y = (-gl_Vertex.y * 0.5) + 0.5;\n"
2150 const char *blt_pshader[] = {
2151 "uniform sampler2D sampler;\n"
2154 " gl_FragDepth = texture2D(sampler, gl_TexCoord[0].xy).x;\n"
2158 vshader_id = GL_EXTCALL(glCreateShaderObjectARB(GL_VERTEX_SHADER_ARB));
2159 GL_EXTCALL(glShaderSourceARB(vshader_id, 1, blt_vshader, NULL));
2160 GL_EXTCALL(glCompileShaderARB(vshader_id));
2162 pshader_id = GL_EXTCALL(glCreateShaderObjectARB(GL_FRAGMENT_SHADER_ARB));
2163 GL_EXTCALL(glShaderSourceARB(pshader_id, 1, blt_pshader, NULL));
2164 GL_EXTCALL(glCompileShaderARB(pshader_id));
2166 program_id = GL_EXTCALL(glCreateProgramObjectARB());
2167 GL_EXTCALL(glAttachObjectARB(program_id, vshader_id));
2168 GL_EXTCALL(glAttachObjectARB(program_id, pshader_id));
2169 GL_EXTCALL(glLinkProgramARB(program_id));
2171 print_glsl_info_log(&GLINFO_LOCATION, program_id);
2176 static void shader_glsl_select(IWineD3DDevice *iface, BOOL usePS, BOOL useVS) {
2177 IWineD3DDeviceImpl *This = (IWineD3DDeviceImpl *)iface;
2178 WineD3D_GL_Info *gl_info = &((IWineD3DImpl *)(This->wineD3D))->gl_info;
2179 GLhandleARB program_id = 0;
2181 if (useVS || usePS) set_glsl_shader_program(iface);
2182 else This->stateBlock->glsl_program = NULL;
2184 program_id = This->stateBlock->glsl_program ? This->stateBlock->glsl_program->programId : 0;
2185 if (program_id) TRACE("Using GLSL program %u\n", program_id);
2186 GL_EXTCALL(glUseProgramObjectARB(program_id));
2187 checkGLcall("glUseProgramObjectARB");
2190 static void shader_glsl_select_depth_blt(IWineD3DDevice *iface) {
2191 IWineD3DDeviceImpl *This = (IWineD3DDeviceImpl *)iface;
2192 WineD3D_GL_Info *gl_info = &((IWineD3DImpl *)(This->wineD3D))->gl_info;
2193 static GLhandleARB program_id = 0;
2194 static GLhandleARB loc = -1;
2197 program_id = create_glsl_blt_shader(gl_info);
2198 loc = GL_EXTCALL(glGetUniformLocationARB(program_id, "sampler"));
2201 GL_EXTCALL(glUseProgramObjectARB(program_id));
2202 GL_EXTCALL(glUniform1iARB(loc, 0));
2205 static void shader_glsl_cleanup(BOOL usePS, BOOL useVS) {
2209 const shader_backend_t glsl_shader_backend = {
2210 &shader_glsl_select,
2211 &shader_glsl_select_depth_blt,
2212 &shader_glsl_load_constants,
2213 &shader_glsl_cleanup