2 * GLSL pixel and vertex shader implementation
4 * Copyright 2006 Jason Green
5 * Copyright 2006-2007 Henri Verbeet
6 * Copyright 2007-2008 Stefan Dösinger for CodeWeavers
7 * Copyright 2009-2011 Henri Verbeet for CodeWeavers
9 * This library is free software; you can redistribute it and/or
10 * modify it under the terms of the GNU Lesser General Public
11 * License as published by the Free Software Foundation; either
12 * version 2.1 of the License, or (at your option) any later version.
14 * This library is distributed in the hope that it will be useful,
15 * but WITHOUT ANY WARRANTY; without even the implied warranty of
16 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
17 * Lesser General Public License for more details.
19 * You should have received a copy of the GNU Lesser General Public
20 * License along with this library; if not, write to the Free Software
21 * Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301, USA
25 * D3D shader asm has swizzles on source parameters, and write masks for
26 * destination parameters. GLSL uses swizzles for both. The result of this is
27 * that for example "mov dst.xw, src.zyxw" becomes "dst.xw = src.zw" in GLSL.
28 * Ie, to generate a proper GLSL source swizzle, we need to take the D3D write
29 * mask for the destination parameter into account.
33 #include "wine/port.h"
38 #include "wined3d_private.h"
40 WINE_DEFAULT_DEBUG_CHANNEL(d3d_shader
);
41 WINE_DECLARE_DEBUG_CHANNEL(d3d_constants
);
42 WINE_DECLARE_DEBUG_CHANNEL(d3d
);
43 WINE_DECLARE_DEBUG_CHANNEL(winediag
);
45 #define WINED3D_GLSL_SAMPLE_PROJECTED 0x1
46 #define WINED3D_GLSL_SAMPLE_NPOT 0x2
47 #define WINED3D_GLSL_SAMPLE_LOD 0x4
48 #define WINED3D_GLSL_SAMPLE_GRAD 0x8
62 struct glsl_sample_function
70 HEAP_NODE_TRAVERSE_LEFT
,
71 HEAP_NODE_TRAVERSE_RIGHT
,
83 struct constant_entry
*entries
;
84 unsigned int *positions
;
88 /* GLSL shader private data */
89 struct shader_glsl_priv
{
90 struct wined3d_shader_buffer shader_buffer
;
91 struct wine_rb_tree program_lookup
;
92 struct glsl_shader_prog_link
*glsl_program
;
93 struct constant_heap vconst_heap
;
94 struct constant_heap pconst_heap
;
96 GLhandleARB depth_blt_program_full
[tex_type_count
];
97 GLhandleARB depth_blt_program_masked
[tex_type_count
];
98 UINT next_constant_version
;
100 const struct wined3d_vertex_pipe_ops
*vertex_pipe
;
101 const struct fragment_pipeline
*fragment_pipe
;
102 struct wine_rb_tree ffp_fragment_shaders
;
103 BOOL ffp_proj_control
;
106 struct glsl_vs_program
108 struct list shader_entry
;
110 GLenum vertex_color_clamp
;
111 GLint
*uniform_f_locations
;
112 GLint uniform_i_locations
[MAX_CONST_I
];
113 GLint pos_fixup_location
;
116 struct glsl_gs_program
118 struct list shader_entry
;
122 struct glsl_ps_program
124 struct list shader_entry
;
126 GLint
*uniform_f_locations
;
127 GLint uniform_i_locations
[MAX_CONST_I
];
128 GLint bumpenv_mat_location
[MAX_TEXTURES
];
129 GLint bumpenv_lum_scale_location
[MAX_TEXTURES
];
130 GLint bumpenv_lum_offset_location
[MAX_TEXTURES
];
131 GLint tex_factor_location
;
132 GLint specular_enable_location
;
133 GLint ycorrection_location
;
134 GLint np2_fixup_location
;
135 const struct ps_np2fixup_info
*np2_fixup_info
;
138 /* Struct to maintain data about a linked GLSL program */
139 struct glsl_shader_prog_link
141 struct wine_rb_entry program_lookup_entry
;
142 struct glsl_vs_program vs
;
143 struct glsl_gs_program gs
;
144 struct glsl_ps_program ps
;
145 GLhandleARB programId
;
146 UINT constant_version
;
149 struct glsl_program_key
156 struct shader_glsl_ctx_priv
{
157 const struct vs_compile_args
*cur_vs_args
;
158 const struct ps_compile_args
*cur_ps_args
;
159 struct ps_np2fixup_info
*cur_np2fixup_info
;
162 struct glsl_ps_compiled_shader
164 struct ps_compile_args args
;
165 struct ps_np2fixup_info np2fixup
;
169 struct glsl_vs_compiled_shader
171 struct vs_compile_args args
;
175 struct glsl_gs_compiled_shader
180 struct glsl_shader_private
184 struct glsl_vs_compiled_shader
*vs
;
185 struct glsl_gs_compiled_shader
*gs
;
186 struct glsl_ps_compiled_shader
*ps
;
188 UINT num_gl_shaders
, shader_array_size
;
191 struct glsl_ffp_fragment_shader
193 struct ffp_frag_desc entry
;
195 struct list linked_programs
;
198 static const char *debug_gl_shader_type(GLenum type
)
202 #define WINED3D_TO_STR(u) case u: return #u
203 WINED3D_TO_STR(GL_VERTEX_SHADER_ARB
);
204 WINED3D_TO_STR(GL_GEOMETRY_SHADER_ARB
);
205 WINED3D_TO_STR(GL_FRAGMENT_SHADER_ARB
);
206 #undef WINED3D_TO_STR
208 return wine_dbg_sprintf("UNKNOWN(%#x)", type
);
212 static const char *shader_glsl_get_prefix(enum wined3d_shader_type type
)
216 case WINED3D_SHADER_TYPE_VERTEX
:
219 case WINED3D_SHADER_TYPE_GEOMETRY
:
222 case WINED3D_SHADER_TYPE_PIXEL
:
226 FIXME("Unhandled shader type %#x.\n", type
);
231 /* Extract a line from the info log.
232 * Note that this modifies the source string. */
233 static char *get_info_log_line(char **ptr
)
238 if (!(q
= strstr(p
, "\n")))
240 if (!*p
) return NULL
;
250 /* Context activation is done by the caller. */
251 static void print_glsl_info_log(const struct wined3d_gl_info
*gl_info
, GLhandleARB obj
)
253 int infologLength
= 0;
256 if (!WARN_ON(d3d_shader
) && !FIXME_ON(d3d_shader
))
259 GL_EXTCALL(glGetObjectParameterivARB(obj
,
260 GL_OBJECT_INFO_LOG_LENGTH_ARB
,
263 /* A size of 1 is just a null-terminated string, so the log should be bigger than
264 * that if there are errors. */
265 if (infologLength
> 1)
269 infoLog
= HeapAlloc(GetProcessHeap(), 0, infologLength
);
270 /* The info log is supposed to be zero-terminated, but at least some
271 * versions of fglrx don't terminate the string properly. The reported
272 * length does include the terminator, so explicitly set it to zero
274 infoLog
[infologLength
- 1] = 0;
275 GL_EXTCALL(glGetInfoLogARB(obj
, infologLength
, NULL
, infoLog
));
278 if (gl_info
->quirks
& WINED3D_QUIRK_INFO_LOG_SPAM
)
280 WARN("Info log received from GLSL shader #%u:\n", obj
);
281 while ((line
= get_info_log_line(&ptr
))) WARN(" %s\n", line
);
285 FIXME("Info log received from GLSL shader #%u:\n", obj
);
286 while ((line
= get_info_log_line(&ptr
))) FIXME(" %s\n", line
);
288 HeapFree(GetProcessHeap(), 0, infoLog
);
292 /* Context activation is done by the caller. */
293 static void shader_glsl_compile(const struct wined3d_gl_info
*gl_info
, GLhandleARB shader
, const char *src
)
295 TRACE("Compiling shader object %u.\n", shader
);
296 GL_EXTCALL(glShaderSourceARB(shader
, 1, &src
, NULL
));
297 checkGLcall("glShaderSourceARB");
298 GL_EXTCALL(glCompileShaderARB(shader
));
299 checkGLcall("glCompileShaderARB");
300 print_glsl_info_log(gl_info
, shader
);
303 /* Context activation is done by the caller. */
304 static void shader_glsl_dump_program_source(const struct wined3d_gl_info
*gl_info
, GLhandleARB program
)
306 GLint i
, object_count
, source_size
= -1;
307 GLhandleARB
*objects
;
310 GL_EXTCALL(glGetObjectParameterivARB(program
, GL_OBJECT_ATTACHED_OBJECTS_ARB
, &object_count
));
311 objects
= HeapAlloc(GetProcessHeap(), 0, object_count
* sizeof(*objects
));
314 ERR("Failed to allocate object array memory.\n");
318 GL_EXTCALL(glGetAttachedObjectsARB(program
, object_count
, NULL
, objects
));
319 for (i
= 0; i
< object_count
; ++i
)
324 GL_EXTCALL(glGetObjectParameterivARB(objects
[i
], GL_OBJECT_SHADER_SOURCE_LENGTH_ARB
, &tmp
));
326 if (source_size
< tmp
)
328 HeapFree(GetProcessHeap(), 0, source
);
330 source
= HeapAlloc(GetProcessHeap(), HEAP_ZERO_MEMORY
, tmp
);
333 ERR("Failed to allocate %d bytes for shader source.\n", tmp
);
334 HeapFree(GetProcessHeap(), 0, objects
);
340 FIXME("Object %u:\n", objects
[i
]);
341 GL_EXTCALL(glGetObjectParameterivARB(objects
[i
], GL_OBJECT_SUBTYPE_ARB
, &tmp
));
342 FIXME(" GL_OBJECT_SUBTYPE_ARB: %s.\n", debug_gl_shader_type(tmp
));
343 GL_EXTCALL(glGetObjectParameterivARB(objects
[i
], GL_OBJECT_COMPILE_STATUS_ARB
, &tmp
));
344 FIXME(" GL_OBJECT_COMPILE_STATUS_ARB: %d.\n", tmp
);
348 GL_EXTCALL(glGetShaderSourceARB(objects
[i
], source_size
, NULL
, source
));
349 while ((line
= get_info_log_line(&ptr
))) FIXME(" %s\n", line
);
353 HeapFree(GetProcessHeap(), 0, source
);
354 HeapFree(GetProcessHeap(), 0, objects
);
357 /* Context activation is done by the caller. */
358 static void shader_glsl_validate_link(const struct wined3d_gl_info
*gl_info
, GLhandleARB program
)
362 if (!TRACE_ON(d3d_shader
) && !FIXME_ON(d3d_shader
)) return;
364 GL_EXTCALL(glGetObjectParameterivARB(program
, GL_OBJECT_TYPE_ARB
, &tmp
));
365 if (tmp
== GL_PROGRAM_OBJECT_ARB
)
367 GL_EXTCALL(glGetObjectParameterivARB(program
, GL_OBJECT_LINK_STATUS_ARB
, &tmp
));
370 FIXME("Program %u link status invalid.\n", program
);
371 shader_glsl_dump_program_source(gl_info
, program
);
375 print_glsl_info_log(gl_info
, program
);
378 /* Context activation is done by the caller. */
379 static void shader_glsl_load_psamplers(const struct wined3d_gl_info
*gl_info
,
380 const DWORD
*tex_unit_map
, GLhandleARB programId
)
383 char sampler_name
[20];
386 for (i
= 0; i
< MAX_FRAGMENT_SAMPLERS
; ++i
)
388 snprintf(sampler_name
, sizeof(sampler_name
), "ps_sampler%u", i
);
389 name_loc
= GL_EXTCALL(glGetUniformLocationARB(programId
, sampler_name
));
390 if (name_loc
!= -1) {
391 DWORD mapped_unit
= tex_unit_map
[i
];
392 if (mapped_unit
!= WINED3D_UNMAPPED_STAGE
&& mapped_unit
< gl_info
->limits
.fragment_samplers
)
394 TRACE("Loading %s for texture %d\n", sampler_name
, mapped_unit
);
395 GL_EXTCALL(glUniform1iARB(name_loc
, mapped_unit
));
396 checkGLcall("glUniform1iARB");
398 ERR("Trying to load sampler %s on unsupported unit %d\n", sampler_name
, mapped_unit
);
404 /* Context activation is done by the caller. */
405 static void shader_glsl_load_vsamplers(const struct wined3d_gl_info
*gl_info
,
406 const DWORD
*tex_unit_map
, GLhandleARB programId
)
409 char sampler_name
[20];
412 for (i
= 0; i
< MAX_VERTEX_SAMPLERS
; ++i
)
414 snprintf(sampler_name
, sizeof(sampler_name
), "vs_sampler%u", i
);
415 name_loc
= GL_EXTCALL(glGetUniformLocationARB(programId
, sampler_name
));
416 if (name_loc
!= -1) {
417 DWORD mapped_unit
= tex_unit_map
[MAX_FRAGMENT_SAMPLERS
+ i
];
418 if (mapped_unit
!= WINED3D_UNMAPPED_STAGE
&& mapped_unit
< gl_info
->limits
.combined_samplers
)
420 TRACE("Loading %s for texture %d\n", sampler_name
, mapped_unit
);
421 GL_EXTCALL(glUniform1iARB(name_loc
, mapped_unit
));
422 checkGLcall("glUniform1iARB");
424 ERR("Trying to load sampler %s on unsupported unit %d\n", sampler_name
, mapped_unit
);
430 /* Context activation is done by the caller. */
431 static inline void walk_constant_heap(const struct wined3d_gl_info
*gl_info
, const float *constants
,
432 const GLint
*constant_locations
, const struct constant_heap
*heap
, unsigned char *stack
, DWORD version
)
435 unsigned int heap_idx
= 1;
438 if (heap
->entries
[heap_idx
].version
<= version
) return;
440 idx
= heap
->entries
[heap_idx
].idx
;
441 if (constant_locations
[idx
] != -1) GL_EXTCALL(glUniform4fvARB(constant_locations
[idx
], 1, &constants
[idx
* 4]));
442 stack
[stack_idx
] = HEAP_NODE_TRAVERSE_LEFT
;
444 while (stack_idx
>= 0)
446 /* Note that we fall through to the next case statement. */
447 switch(stack
[stack_idx
])
449 case HEAP_NODE_TRAVERSE_LEFT
:
451 unsigned int left_idx
= heap_idx
<< 1;
452 if (left_idx
< heap
->size
&& heap
->entries
[left_idx
].version
> version
)
455 idx
= heap
->entries
[heap_idx
].idx
;
456 if (constant_locations
[idx
] != -1)
457 GL_EXTCALL(glUniform4fvARB(constant_locations
[idx
], 1, &constants
[idx
* 4]));
459 stack
[stack_idx
++] = HEAP_NODE_TRAVERSE_RIGHT
;
460 stack
[stack_idx
] = HEAP_NODE_TRAVERSE_LEFT
;
465 case HEAP_NODE_TRAVERSE_RIGHT
:
467 unsigned int right_idx
= (heap_idx
<< 1) + 1;
468 if (right_idx
< heap
->size
&& heap
->entries
[right_idx
].version
> version
)
470 heap_idx
= right_idx
;
471 idx
= heap
->entries
[heap_idx
].idx
;
472 if (constant_locations
[idx
] != -1)
473 GL_EXTCALL(glUniform4fvARB(constant_locations
[idx
], 1, &constants
[idx
* 4]));
475 stack
[stack_idx
++] = HEAP_NODE_POP
;
476 stack
[stack_idx
] = HEAP_NODE_TRAVERSE_LEFT
;
487 checkGLcall("walk_constant_heap()");
490 /* Context activation is done by the caller. */
491 static inline void apply_clamped_constant(const struct wined3d_gl_info
*gl_info
, GLint location
, const GLfloat
*data
)
493 GLfloat clamped_constant
[4];
495 if (location
== -1) return;
497 clamped_constant
[0] = data
[0] < -1.0f
? -1.0f
: data
[0] > 1.0f
? 1.0f
: data
[0];
498 clamped_constant
[1] = data
[1] < -1.0f
? -1.0f
: data
[1] > 1.0f
? 1.0f
: data
[1];
499 clamped_constant
[2] = data
[2] < -1.0f
? -1.0f
: data
[2] > 1.0f
? 1.0f
: data
[2];
500 clamped_constant
[3] = data
[3] < -1.0f
? -1.0f
: data
[3] > 1.0f
? 1.0f
: data
[3];
502 GL_EXTCALL(glUniform4fvARB(location
, 1, clamped_constant
));
505 /* Context activation is done by the caller. */
506 static inline void walk_constant_heap_clamped(const struct wined3d_gl_info
*gl_info
, const float *constants
,
507 const GLint
*constant_locations
, const struct constant_heap
*heap
, unsigned char *stack
, DWORD version
)
510 unsigned int heap_idx
= 1;
513 if (heap
->entries
[heap_idx
].version
<= version
) return;
515 idx
= heap
->entries
[heap_idx
].idx
;
516 apply_clamped_constant(gl_info
, constant_locations
[idx
], &constants
[idx
* 4]);
517 stack
[stack_idx
] = HEAP_NODE_TRAVERSE_LEFT
;
519 while (stack_idx
>= 0)
521 /* Note that we fall through to the next case statement. */
522 switch(stack
[stack_idx
])
524 case HEAP_NODE_TRAVERSE_LEFT
:
526 unsigned int left_idx
= heap_idx
<< 1;
527 if (left_idx
< heap
->size
&& heap
->entries
[left_idx
].version
> version
)
530 idx
= heap
->entries
[heap_idx
].idx
;
531 apply_clamped_constant(gl_info
, constant_locations
[idx
], &constants
[idx
* 4]);
533 stack
[stack_idx
++] = HEAP_NODE_TRAVERSE_RIGHT
;
534 stack
[stack_idx
] = HEAP_NODE_TRAVERSE_LEFT
;
539 case HEAP_NODE_TRAVERSE_RIGHT
:
541 unsigned int right_idx
= (heap_idx
<< 1) + 1;
542 if (right_idx
< heap
->size
&& heap
->entries
[right_idx
].version
> version
)
544 heap_idx
= right_idx
;
545 idx
= heap
->entries
[heap_idx
].idx
;
546 apply_clamped_constant(gl_info
, constant_locations
[idx
], &constants
[idx
* 4]);
548 stack
[stack_idx
++] = HEAP_NODE_POP
;
549 stack
[stack_idx
] = HEAP_NODE_TRAVERSE_LEFT
;
560 checkGLcall("walk_constant_heap_clamped()");
563 /* Context activation is done by the caller. */
564 static void shader_glsl_load_constantsF(const struct wined3d_shader
*shader
, const struct wined3d_gl_info
*gl_info
,
565 const float *constants
, const GLint
*constant_locations
, const struct constant_heap
*heap
,
566 unsigned char *stack
, UINT version
)
568 const struct wined3d_shader_lconst
*lconst
;
570 /* 1.X pshaders have the constants clamped to [-1;1] implicitly. */
571 if (shader
->reg_maps
.shader_version
.major
== 1
572 && shader
->reg_maps
.shader_version
.type
== WINED3D_SHADER_TYPE_PIXEL
)
573 walk_constant_heap_clamped(gl_info
, constants
, constant_locations
, heap
, stack
, version
);
575 walk_constant_heap(gl_info
, constants
, constant_locations
, heap
, stack
, version
);
577 if (!shader
->load_local_constsF
)
579 TRACE("No need to load local float constants for this shader\n");
583 /* Immediate constants are clamped to [-1;1] at shader creation time if needed */
584 LIST_FOR_EACH_ENTRY(lconst
, &shader
->constantsF
, struct wined3d_shader_lconst
, entry
)
586 GL_EXTCALL(glUniform4fvARB(constant_locations
[lconst
->idx
], 1, (const GLfloat
*)lconst
->value
));
588 checkGLcall("glUniform4fvARB()");
591 /* Context activation is done by the caller. */
592 static void shader_glsl_load_constantsI(const struct wined3d_shader
*shader
, const struct wined3d_gl_info
*gl_info
,
593 const GLint locations
[MAX_CONST_I
], const int *constants
, WORD constants_set
)
598 for (i
= 0; constants_set
; constants_set
>>= 1, ++i
)
600 if (!(constants_set
& 1)) continue;
602 TRACE_(d3d_constants
)("Loading constants %u: %i, %i, %i, %i\n",
603 i
, constants
[i
*4], constants
[i
*4+1], constants
[i
*4+2], constants
[i
*4+3]);
605 /* We found this uniform name in the program - go ahead and send the data */
606 GL_EXTCALL(glUniform4ivARB(locations
[i
], 1, &constants
[i
*4]));
607 checkGLcall("glUniform4ivARB");
610 /* Load immediate constants */
611 ptr
= list_head(&shader
->constantsI
);
614 const struct wined3d_shader_lconst
*lconst
= LIST_ENTRY(ptr
, const struct wined3d_shader_lconst
, entry
);
615 unsigned int idx
= lconst
->idx
;
616 const GLint
*values
= (const GLint
*)lconst
->value
;
618 TRACE_(d3d_constants
)("Loading local constants %i: %i, %i, %i, %i\n", idx
,
619 values
[0], values
[1], values
[2], values
[3]);
621 /* We found this uniform name in the program - go ahead and send the data */
622 GL_EXTCALL(glUniform4ivARB(locations
[idx
], 1, values
));
623 checkGLcall("glUniform4ivARB");
624 ptr
= list_next(&shader
->constantsI
, ptr
);
628 /* Context activation is done by the caller. */
629 static void shader_glsl_load_constantsB(const struct wined3d_shader
*shader
, const struct wined3d_gl_info
*gl_info
,
630 GLhandleARB programId
, const BOOL
*constants
, WORD constants_set
)
638 prefix
= shader_glsl_get_prefix(shader
->reg_maps
.shader_version
.type
);
640 /* TODO: Benchmark and see if it would be beneficial to store the
641 * locations of the constants to avoid looking up each time */
642 for (i
= 0; constants_set
; constants_set
>>= 1, ++i
)
644 if (!(constants_set
& 1)) continue;
646 TRACE_(d3d_constants
)("Loading constants %i: %i;\n", i
, constants
[i
]);
648 /* TODO: Benchmark and see if it would be beneficial to store the
649 * locations of the constants to avoid looking up each time */
650 snprintf(tmp_name
, sizeof(tmp_name
), "%s_b[%i]", prefix
, i
);
651 tmp_loc
= GL_EXTCALL(glGetUniformLocationARB(programId
, tmp_name
));
652 GL_EXTCALL(glUniform1ivARB(tmp_loc
, 1, &constants
[i
]));
655 /* Load immediate constants */
656 ptr
= list_head(&shader
->constantsB
);
659 const struct wined3d_shader_lconst
*lconst
= LIST_ENTRY(ptr
, const struct wined3d_shader_lconst
, entry
);
660 unsigned int idx
= lconst
->idx
;
661 const GLint
*values
= (const GLint
*)lconst
->value
;
663 TRACE_(d3d_constants
)("Loading local constants %i: %i\n", idx
, values
[0]);
665 snprintf(tmp_name
, sizeof(tmp_name
), "%s_b[%i]", prefix
, idx
);
666 tmp_loc
= GL_EXTCALL(glGetUniformLocationARB(programId
, tmp_name
));
667 GL_EXTCALL(glUniform1ivARB(tmp_loc
, 1, values
));
668 ptr
= list_next(&shader
->constantsB
, ptr
);
671 checkGLcall("shader_glsl_load_constantsB()");
674 static void reset_program_constant_version(struct wine_rb_entry
*entry
, void *context
)
676 WINE_RB_ENTRY_VALUE(entry
, struct glsl_shader_prog_link
, program_lookup_entry
)->constant_version
= 0;
679 /* Context activation is done by the caller (state handler). */
680 static void shader_glsl_load_np2fixup_constants(void *shader_priv
,
681 const struct wined3d_gl_info
*gl_info
, const struct wined3d_state
*state
)
683 struct shader_glsl_priv
*glsl_priv
= shader_priv
;
684 const struct glsl_shader_prog_link
*prog
= glsl_priv
->glsl_program
;
686 /* No GLSL program set - nothing to do. */
689 /* NP2 texcoord fixup is (currently) only done for pixelshaders. */
690 if (!use_ps(state
)) return;
692 if (prog
->ps
.np2_fixup_info
&& prog
->ps
.np2_fixup_location
!= -1)
695 UINT fixup
= prog
->ps
.np2_fixup_info
->active
;
696 GLfloat np2fixup_constants
[4 * MAX_FRAGMENT_SAMPLERS
];
698 for (i
= 0; fixup
; fixup
>>= 1, ++i
)
700 const struct wined3d_texture
*tex
= state
->textures
[i
];
701 const unsigned char idx
= prog
->ps
.np2_fixup_info
->idx
[i
];
702 GLfloat
*tex_dim
= &np2fixup_constants
[(idx
>> 1) * 4];
706 ERR("Nonexistent texture is flagged for NP2 texcoord fixup.\n");
712 tex_dim
[2] = tex
->pow2_matrix
[0];
713 tex_dim
[3] = tex
->pow2_matrix
[5];
717 tex_dim
[0] = tex
->pow2_matrix
[0];
718 tex_dim
[1] = tex
->pow2_matrix
[5];
722 GL_EXTCALL(glUniform4fvARB(prog
->ps
.np2_fixup_location
,
723 prog
->ps
.np2_fixup_info
->num_consts
, np2fixup_constants
));
727 /* Context activation is done by the caller (state handler). */
728 static void shader_glsl_load_constants(const struct wined3d_context
*context
,
729 BOOL usePixelShader
, BOOL useVertexShader
)
731 const struct wined3d_gl_info
*gl_info
= context
->gl_info
;
732 struct wined3d_device
*device
= context
->swapchain
->device
;
733 struct wined3d_stateblock
*stateBlock
= device
->stateBlock
;
734 const struct wined3d_state
*state
= &stateBlock
->state
;
735 struct shader_glsl_priv
*priv
= device
->shader_priv
;
736 float position_fixup
[4];
738 GLhandleARB programId
;
739 struct glsl_shader_prog_link
*prog
= priv
->glsl_program
;
740 UINT constant_version
;
744 /* No GLSL program set - nothing to do. */
747 programId
= prog
->programId
;
748 constant_version
= prog
->constant_version
;
752 const struct wined3d_shader
*vshader
= state
->vertex_shader
;
754 /* Load DirectX 9 float constants/uniforms for vertex shader */
755 shader_glsl_load_constantsF(vshader
, gl_info
, state
->vs_consts_f
,
756 prog
->vs
.uniform_f_locations
, &priv
->vconst_heap
, priv
->stack
, constant_version
);
758 /* Load DirectX 9 integer constants/uniforms for vertex shader */
759 shader_glsl_load_constantsI(vshader
, gl_info
, prog
->vs
.uniform_i_locations
, state
->vs_consts_i
,
760 stateBlock
->changed
.vertexShaderConstantsI
& vshader
->reg_maps
.integer_constants
);
762 /* Load DirectX 9 boolean constants/uniforms for vertex shader */
763 shader_glsl_load_constantsB(vshader
, gl_info
, programId
, state
->vs_consts_b
,
764 stateBlock
->changed
.vertexShaderConstantsB
& vshader
->reg_maps
.boolean_constants
);
766 /* Upload the position fixup params */
767 shader_get_position_fixup(context
, state
, position_fixup
);
768 GL_EXTCALL(glUniform4fvARB(prog
->vs
.pos_fixup_location
, 1, position_fixup
));
769 checkGLcall("glUniform4fvARB");
774 const struct wined3d_shader
*pshader
= state
->pixel_shader
;
776 /* Load DirectX 9 float constants/uniforms for pixel shader */
777 shader_glsl_load_constantsF(pshader
, gl_info
, state
->ps_consts_f
,
778 prog
->ps
.uniform_f_locations
, &priv
->pconst_heap
, priv
->stack
, constant_version
);
780 /* Load DirectX 9 integer constants/uniforms for pixel shader */
781 shader_glsl_load_constantsI(pshader
, gl_info
, prog
->ps
.uniform_i_locations
, state
->ps_consts_i
,
782 stateBlock
->changed
.pixelShaderConstantsI
& pshader
->reg_maps
.integer_constants
);
784 /* Load DirectX 9 boolean constants/uniforms for pixel shader */
785 shader_glsl_load_constantsB(pshader
, gl_info
, programId
, state
->ps_consts_b
,
786 stateBlock
->changed
.pixelShaderConstantsB
& pshader
->reg_maps
.boolean_constants
);
788 /* Upload the environment bump map matrix if needed. The needsbumpmat
789 * member specifies the texture stage to load the matrix from. It
790 * can't be 0 for a valid texbem instruction. */
791 for (i
= 0; i
< MAX_TEXTURES
; ++i
)
795 if (prog
->ps
.bumpenv_mat_location
[i
] == -1)
798 data
= (const float *)&state
->texture_states
[i
][WINED3D_TSS_BUMPENV_MAT00
];
799 GL_EXTCALL(glUniformMatrix2fvARB(prog
->ps
.bumpenv_mat_location
[i
], 1, 0, data
));
800 checkGLcall("glUniformMatrix2fvARB");
802 /* texbeml needs the luminance scale and offset too. If texbeml
803 * is used, needsbumpmat is set too, so we can check that in the
804 * needsbumpmat check. */
805 if (prog
->ps
.bumpenv_lum_scale_location
[i
] != -1)
807 const GLfloat
*scale
= (const GLfloat
*)&state
->texture_states
[i
][WINED3D_TSS_BUMPENV_LSCALE
];
808 const GLfloat
*offset
= (const GLfloat
*)&state
->texture_states
[i
][WINED3D_TSS_BUMPENV_LOFFSET
];
810 GL_EXTCALL(glUniform1fvARB(prog
->ps
.bumpenv_lum_scale_location
[i
], 1, scale
));
811 checkGLcall("glUniform1fvARB");
812 GL_EXTCALL(glUniform1fvARB(prog
->ps
.bumpenv_lum_offset_location
[i
], 1, offset
));
813 checkGLcall("glUniform1fvARB");
817 if (prog
->ps
.ycorrection_location
!= -1)
819 float correction_params
[4];
821 if (context
->render_offscreen
)
823 correction_params
[0] = 0.0f
;
824 correction_params
[1] = 1.0f
;
826 /* position is window relative, not viewport relative */
827 correction_params
[0] = (float) context
->current_rt
->resource
.height
;
828 correction_params
[1] = -1.0f
;
830 GL_EXTCALL(glUniform4fvARB(prog
->ps
.ycorrection_location
, 1, correction_params
));
833 else if (priv
->fragment_pipe
== &glsl_fragment_pipe
)
837 for (i
= 0; i
< MAX_TEXTURES
; ++i
)
839 GL_EXTCALL(glUniformMatrix2fvARB(prog
->ps
.bumpenv_mat_location
[i
], 1, 0,
840 (const float *)&state
->texture_states
[i
][WINED3D_TSS_BUMPENV_MAT00
]));
841 GL_EXTCALL(glUniform1fARB(prog
->ps
.bumpenv_lum_scale_location
[i
],
842 *(const float *)&state
->texture_states
[i
][WINED3D_TSS_BUMPENV_LSCALE
]));
843 GL_EXTCALL(glUniform1fARB(prog
->ps
.bumpenv_lum_offset_location
[i
],
844 *(const float *)&state
->texture_states
[i
][WINED3D_TSS_BUMPENV_LOFFSET
]));
847 D3DCOLORTOGLFLOAT4(state
->render_states
[WINED3D_RS_TEXTUREFACTOR
], col
);
848 GL_EXTCALL(glUniform4fARB(prog
->ps
.tex_factor_location
, col
[0], col
[1], col
[2], col
[3]));
850 if (state
->render_states
[WINED3D_RS_SPECULARENABLE
])
851 GL_EXTCALL(glUniform4fARB(prog
->ps
.specular_enable_location
, 1.0f
, 1.0f
, 1.0f
, 0.0f
));
853 GL_EXTCALL(glUniform4fARB(prog
->ps
.specular_enable_location
, 0.0f
, 0.0f
, 0.0f
, 0.0f
));
855 checkGLcall("fixed function uniforms");
858 if (priv
->next_constant_version
== UINT_MAX
)
860 TRACE("Max constant version reached, resetting to 0.\n");
861 wine_rb_for_each_entry(&priv
->program_lookup
, reset_program_constant_version
, NULL
);
862 priv
->next_constant_version
= 1;
866 prog
->constant_version
= priv
->next_constant_version
++;
870 static void update_heap_entry(const struct constant_heap
*heap
, unsigned int idx
,
871 unsigned int heap_idx
, DWORD new_version
)
873 struct constant_entry
*entries
= heap
->entries
;
874 unsigned int *positions
= heap
->positions
;
875 unsigned int parent_idx
;
879 parent_idx
= heap_idx
>> 1;
881 if (new_version
<= entries
[parent_idx
].version
) break;
883 entries
[heap_idx
] = entries
[parent_idx
];
884 positions
[entries
[parent_idx
].idx
] = heap_idx
;
885 heap_idx
= parent_idx
;
888 entries
[heap_idx
].version
= new_version
;
889 entries
[heap_idx
].idx
= idx
;
890 positions
[idx
] = heap_idx
;
893 static void shader_glsl_update_float_vertex_constants(struct wined3d_device
*device
, UINT start
, UINT count
)
895 struct shader_glsl_priv
*priv
= device
->shader_priv
;
896 struct constant_heap
*heap
= &priv
->vconst_heap
;
899 for (i
= start
; i
< count
+ start
; ++i
)
901 if (!device
->stateBlock
->changed
.vertexShaderConstantsF
[i
])
902 update_heap_entry(heap
, i
, heap
->size
++, priv
->next_constant_version
);
904 update_heap_entry(heap
, i
, heap
->positions
[i
], priv
->next_constant_version
);
908 static void shader_glsl_update_float_pixel_constants(struct wined3d_device
*device
, UINT start
, UINT count
)
910 struct shader_glsl_priv
*priv
= device
->shader_priv
;
911 struct constant_heap
*heap
= &priv
->pconst_heap
;
914 for (i
= start
; i
< count
+ start
; ++i
)
916 if (!device
->stateBlock
->changed
.pixelShaderConstantsF
[i
])
917 update_heap_entry(heap
, i
, heap
->size
++, priv
->next_constant_version
);
919 update_heap_entry(heap
, i
, heap
->positions
[i
], priv
->next_constant_version
);
923 static unsigned int vec4_varyings(DWORD shader_major
, const struct wined3d_gl_info
*gl_info
)
925 unsigned int ret
= gl_info
->limits
.glsl_varyings
/ 4;
926 /* 4.0 shaders do not write clip coords because d3d10 does not support user clipplanes */
927 if(shader_major
> 3) return ret
;
929 /* 3.0 shaders may need an extra varying for the clip coord on some cards(mostly dx10 ones) */
930 if (gl_info
->quirks
& WINED3D_QUIRK_GLSL_CLIP_VARYING
) ret
-= 1;
934 /** Generate the variable & register declarations for the GLSL output target */
935 static void shader_generate_glsl_declarations(const struct wined3d_context
*context
,
936 struct wined3d_shader_buffer
*buffer
, const struct wined3d_shader
*shader
,
937 const struct wined3d_shader_reg_maps
*reg_maps
, const struct shader_glsl_ctx_priv
*ctx_priv
)
939 const struct wined3d_shader_version
*version
= ®_maps
->shader_version
;
940 const struct wined3d_state
*state
= &shader
->device
->stateBlock
->state
;
941 const struct ps_compile_args
*ps_args
= ctx_priv
->cur_ps_args
;
942 const struct wined3d_gl_info
*gl_info
= context
->gl_info
;
943 const struct wined3d_fb_state
*fb
= &shader
->device
->fb
;
944 unsigned int i
, extra_constants_needed
= 0;
945 const struct wined3d_shader_lconst
*lconst
;
949 prefix
= shader_glsl_get_prefix(version
->type
);
951 /* Prototype the subroutines */
952 for (i
= 0, map
= reg_maps
->labels
; map
; map
>>= 1, ++i
)
954 if (map
& 1) shader_addline(buffer
, "void subroutine%u();\n", i
);
957 /* Declare the constants (aka uniforms) */
958 if (shader
->limits
.constant_float
> 0)
960 unsigned max_constantsF
;
962 /* Unless the shader uses indirect addressing, always declare the
963 * maximum array size and ignore that we need some uniforms privately.
964 * E.g. if GL supports 256 uniforms, and we need 2 for the pos fixup
965 * and immediate values, still declare VC[256]. If the shader needs
966 * more uniforms than we have it won't work in any case. If it uses
967 * less, the compiler will figure out which uniforms are really used
968 * and strip them out. This allows a shader to use c255 on a dx9 card,
969 * as long as it doesn't also use all the other constants.
971 * If the shader uses indirect addressing the compiler must assume
972 * that all declared uniforms are used. In this case, declare only the
973 * amount that we're assured to have.
975 * Thus we run into problems in these two cases:
976 * 1) The shader really uses more uniforms than supported.
977 * 2) The shader uses indirect addressing, less constants than
978 * supported, but uses a constant index > #supported consts. */
979 if (version
->type
== WINED3D_SHADER_TYPE_PIXEL
)
981 /* No indirect addressing here. */
982 max_constantsF
= gl_info
->limits
.glsl_ps_float_constants
;
986 if (reg_maps
->usesrelconstF
)
988 /* Subtract the other potential uniforms from the max
989 * available (bools, ints, and 1 row of projection matrix).
990 * Subtract another uniform for immediate values, which have
991 * to be loaded via uniform by the driver as well. The shader
992 * code only uses 0.5, 2.0, 1.0, 128 and -128 in vertex
993 * shader code, so one vec4 should be enough. (Unfortunately
994 * the Nvidia driver doesn't store 128 and -128 in one float).
996 * Writing gl_ClipVertex requires one uniform for each
997 * clipplane as well. */
998 max_constantsF
= gl_info
->limits
.glsl_vs_float_constants
- 3;
999 if(ctx_priv
->cur_vs_args
->clip_enabled
)
1001 max_constantsF
-= gl_info
->limits
.clipplanes
;
1003 max_constantsF
-= count_bits(reg_maps
->integer_constants
);
1004 /* Strictly speaking a bool only uses one scalar, but the nvidia(Linux) compiler doesn't pack them properly,
1005 * so each scalar requires a full vec4. We could work around this by packing the booleans ourselves, but
1006 * for now take this into account when calculating the number of available constants
1008 max_constantsF
-= count_bits(reg_maps
->boolean_constants
);
1009 /* Set by driver quirks in directx.c */
1010 max_constantsF
-= gl_info
->reserved_glsl_constants
;
1012 if (max_constantsF
< shader
->limits
.constant_float
)
1014 static unsigned int once
;
1017 ERR_(winediag
)("The hardware does not support enough uniform components to run this shader,"
1018 " it may not render correctly.\n");
1020 WARN("The hardware does not support enough uniform components to run this shader.\n");
1025 max_constantsF
= gl_info
->limits
.glsl_vs_float_constants
;
1028 max_constantsF
= min(shader
->limits
.constant_float
, max_constantsF
);
1029 shader_addline(buffer
, "uniform vec4 %s_c[%u];\n", prefix
, max_constantsF
);
1032 /* Always declare the full set of constants, the compiler can remove the
1033 * unused ones because d3d doesn't (yet) support indirect int and bool
1034 * constant addressing. This avoids problems if the app uses e.g. i0 and i9. */
1035 if (shader
->limits
.constant_int
> 0 && reg_maps
->integer_constants
)
1036 shader_addline(buffer
, "uniform ivec4 %s_i[%u];\n", prefix
, shader
->limits
.constant_int
);
1038 if (shader
->limits
.constant_bool
> 0 && reg_maps
->boolean_constants
)
1039 shader_addline(buffer
, "uniform bool %s_b[%u];\n", prefix
, shader
->limits
.constant_bool
);
1041 for (i
= 0; i
< WINED3D_MAX_CBS
; ++i
)
1043 if (reg_maps
->cb_sizes
[i
])
1044 shader_addline(buffer
, "uniform vec4 %s_cb%u[%u];\n", prefix
, i
, reg_maps
->cb_sizes
[i
]);
1047 /* Declare texture samplers */
1048 for (i
= 0; i
< shader
->limits
.sampler
; ++i
)
1050 if (reg_maps
->sampler_type
[i
])
1052 BOOL shadow_sampler
= version
->type
== WINED3D_SHADER_TYPE_PIXEL
&& (ps_args
->shadow
& (1 << i
));
1055 switch (reg_maps
->sampler_type
[i
])
1059 shader_addline(buffer
, "uniform sampler1DShadow %s_sampler%u;\n", prefix
, i
);
1061 shader_addline(buffer
, "uniform sampler1D %s_sampler%u;\n", prefix
, i
);
1064 tex_rect
= version
->type
== WINED3D_SHADER_TYPE_PIXEL
&& (ps_args
->np2_fixup
& (1 << i
));
1065 tex_rect
= tex_rect
&& gl_info
->supported
[ARB_TEXTURE_RECTANGLE
];
1069 shader_addline(buffer
, "uniform sampler2DRectShadow %s_sampler%u;\n", prefix
, i
);
1071 shader_addline(buffer
, "uniform sampler2DShadow %s_sampler%u;\n", prefix
, i
);
1076 shader_addline(buffer
, "uniform sampler2DRect %s_sampler%u;\n", prefix
, i
);
1078 shader_addline(buffer
, "uniform sampler2D %s_sampler%u;\n", prefix
, i
);
1081 case WINED3DSTT_CUBE
:
1083 FIXME("Unsupported Cube shadow sampler.\n");
1084 shader_addline(buffer
, "uniform samplerCube %s_sampler%u;\n", prefix
, i
);
1086 case WINED3DSTT_VOLUME
:
1088 FIXME("Unsupported 3D shadow sampler.\n");
1089 shader_addline(buffer
, "uniform sampler3D %s_sampler%u;\n", prefix
, i
);
1092 shader_addline(buffer
, "uniform unsupported_sampler %s_sampler%u;\n", prefix
, i
);
1093 FIXME("Unrecognized sampler type: %#x\n", reg_maps
->sampler_type
[i
]);
1099 /* Declare uniforms for NP2 texcoord fixup:
1100 * This is NOT done inside the loop that declares the texture samplers
1101 * since the NP2 fixup code is currently only used for the GeforceFX
1102 * series and when forcing the ARB_npot extension off. Modern cards just
1103 * skip the code anyway, so put it inside a separate loop. */
1104 if (version
->type
== WINED3D_SHADER_TYPE_PIXEL
&& ps_args
->np2_fixup
)
1106 struct ps_np2fixup_info
*fixup
= ctx_priv
->cur_np2fixup_info
;
1109 /* NP2/RECT textures in OpenGL use texcoords in the range [0,width]x[0,height]
1110 * while D3D has them in the (normalized) [0,1]x[0,1] range.
1111 * samplerNP2Fixup stores texture dimensions and is updated through
1112 * shader_glsl_load_np2fixup_constants when the sampler changes. */
1114 for (i
= 0; i
< shader
->limits
.sampler
; ++i
)
1116 if (reg_maps
->sampler_type
[i
])
1118 if (!(ps_args
->np2_fixup
& (1 << i
))) continue;
1120 if (WINED3DSTT_2D
!= reg_maps
->sampler_type
[i
]) {
1121 FIXME("Non-2D texture is flagged for NP2 texcoord fixup.\n");
1125 fixup
->idx
[i
] = cur
++;
1129 fixup
->num_consts
= (cur
+ 1) >> 1;
1130 fixup
->active
= ps_args
->np2_fixup
;
1131 shader_addline(buffer
, "uniform vec4 %s_samplerNP2Fixup[%u];\n", prefix
, fixup
->num_consts
);
1134 /* Declare address variables */
1135 for (i
= 0, map
= reg_maps
->address
; map
; map
>>= 1, ++i
)
1137 if (map
& 1) shader_addline(buffer
, "ivec4 A%u;\n", i
);
1140 /* Declare texture coordinate temporaries and initialize them */
1141 for (i
= 0, map
= reg_maps
->texcoord
; map
; map
>>= 1, ++i
)
1143 if (map
& 1) shader_addline(buffer
, "vec4 T%u = gl_TexCoord[%u];\n", i
, i
);
1146 if (version
->type
== WINED3D_SHADER_TYPE_VERTEX
)
1148 /* Declare attributes. */
1149 for (i
= 0, map
= reg_maps
->input_registers
; map
; map
>>= 1, ++i
)
1152 shader_addline(buffer
, "attribute vec4 %s_in%u;\n", prefix
, i
);
1155 shader_addline(buffer
, "uniform vec4 posFixup;\n");
1156 shader_addline(buffer
, "void order_ps_input(in vec4[%u]);\n", shader
->limits
.packed_output
);
1158 else if (version
->type
== WINED3D_SHADER_TYPE_GEOMETRY
)
1160 shader_addline(buffer
, "varying in vec4 gs_in[][%u];\n", shader
->limits
.packed_input
);
1162 else if (version
->type
== WINED3D_SHADER_TYPE_PIXEL
)
1164 if (version
->major
>= 3)
1166 UINT in_count
= min(vec4_varyings(version
->major
, gl_info
), shader
->limits
.packed_input
);
1169 shader_addline(buffer
, "varying vec4 %s_in[%u];\n", prefix
, in_count
);
1171 /* TODO: Write a replacement shader for the fixed function
1172 * vertex pipeline, so this isn't needed. For fixed function
1173 * vertex processing + 3.0 pixel shader we need a separate
1174 * function in the pixel shader that reads the fixed function
1175 * color into the packed input registers. */
1176 shader_addline(buffer
, "vec4 %s_in[%u];\n", prefix
, in_count
);
1179 for (i
= 0, map
= reg_maps
->bumpmat
; map
; map
>>= 1, ++i
)
1184 shader_addline(buffer
, "uniform mat2 bumpenv_mat%u;\n", i
);
1186 if (reg_maps
->luminanceparams
& (1 << i
))
1188 shader_addline(buffer
, "uniform float bumpenv_lum_scale%u;\n", i
);
1189 shader_addline(buffer
, "uniform float bumpenv_lum_offset%u;\n", i
);
1190 extra_constants_needed
++;
1193 extra_constants_needed
++;
1196 if (ps_args
->srgb_correction
)
1198 shader_addline(buffer
, "const vec4 srgb_const0 = vec4(%.8e, %.8e, %.8e, %.8e);\n",
1199 srgb_pow
, srgb_mul_high
, srgb_sub_high
, srgb_mul_low
);
1200 shader_addline(buffer
, "const vec4 srgb_const1 = vec4(%.8e, 0.0, 0.0, 0.0);\n",
1203 if (reg_maps
->vpos
|| reg_maps
->usesdsy
)
1205 if (shader
->limits
.constant_float
+ extra_constants_needed
1206 + 1 < gl_info
->limits
.glsl_ps_float_constants
)
1208 shader_addline(buffer
, "uniform vec4 ycorrection;\n");
1209 extra_constants_needed
++;
1213 /* This happens because we do not have proper tracking of the constant registers that are
1214 * actually used, only the max limit of the shader version
1216 FIXME("Cannot find a free uniform for vpos correction params\n");
1217 shader_addline(buffer
, "const vec4 ycorrection = vec4(%f, %f, 0.0, 0.0);\n",
1218 context
->render_offscreen
? 0.0f
: fb
->render_targets
[0]->resource
.height
,
1219 context
->render_offscreen
? 1.0f
: -1.0f
);
1221 shader_addline(buffer
, "vec4 vpos;\n");
1225 /* Declare output register temporaries */
1226 if (shader
->limits
.packed_output
)
1227 shader_addline(buffer
, "vec4 %s_out[%u];\n", prefix
, shader
->limits
.packed_output
);
1229 /* Declare temporary variables */
1230 for (i
= 0, map
= reg_maps
->temporary
; map
; map
>>= 1, ++i
)
1232 if (map
& 1) shader_addline(buffer
, "vec4 R%u;\n", i
);
1235 /* Declare loop registers aLx */
1236 if (version
->major
< 4)
1238 for (i
= 0; i
< reg_maps
->loop_depth
; ++i
)
1240 shader_addline(buffer
, "int aL%u;\n", i
);
1241 shader_addline(buffer
, "int tmpInt%u;\n", i
);
1245 /* Temporary variables for matrix operations */
1246 shader_addline(buffer
, "vec4 tmp0;\n");
1247 shader_addline(buffer
, "vec4 tmp1;\n");
1249 if (!shader
->load_local_constsF
)
1251 LIST_FOR_EACH_ENTRY(lconst
, &shader
->constantsF
, struct wined3d_shader_lconst
, entry
)
1254 value
= (const float *)lconst
->value
;
1255 shader_addline(buffer
, "const vec4 %s_lc%u = vec4(%.8e, %.8e, %.8e, %.8e);\n",
1256 prefix
, lconst
->idx
, value
[0], value
[1], value
[2], value
[3]);
1260 /* Start the main program. */
1261 shader_addline(buffer
, "void main()\n{\n");
1263 /* Direct3D applications expect integer vPos values, while OpenGL drivers
1264 * add approximately 0.5. This causes off-by-one problems as spotted by
1265 * the vPos d3d9 visual test. Unfortunately ATI cards do not add exactly
1266 * 0.5, but rather something like 0.49999999 or 0.50000001, which still
1267 * causes precision troubles when we just subtract 0.5.
1269 * To deal with that, just floor() the position. This will eliminate the
1270 * fraction on all cards.
1272 * TODO: Test how this behaves with multisampling.
1274 * An advantage of floor is that it works even if the driver doesn't add
1275 * 0.5. It is somewhat questionable if 1.5, 2.5, ... are the proper values
1276 * to return in gl_FragCoord, even though coordinates specify the pixel
1277 * centers instead of the pixel corners. This code will behave correctly
1278 * on drivers that returns integer values. */
1279 if (version
->type
== WINED3D_SHADER_TYPE_PIXEL
&& reg_maps
->vpos
)
1280 shader_addline(buffer
,
1281 "vpos = floor(vec4(0, ycorrection[0], 0, 0) + gl_FragCoord * vec4(1, ycorrection[1], 1, 1));\n");
1284 /*****************************************************************************
1285 * Functions to generate GLSL strings from DirectX Shader bytecode begin here.
1287 * For more information, see http://wiki.winehq.org/DirectX-Shaders
1288 ****************************************************************************/
1291 static void shader_glsl_add_src_param(const struct wined3d_shader_instruction
*ins
,
1292 const struct wined3d_shader_src_param
*wined3d_src
, DWORD mask
, struct glsl_src_param
*glsl_src
);
1294 /** Used for opcode modifiers - They multiply the result by the specified amount */
1295 static const char * const shift_glsl_tab
[] = {
1297 "2.0 * ", /* 1 (x2) */
1298 "4.0 * ", /* 2 (x4) */
1299 "8.0 * ", /* 3 (x8) */
1300 "16.0 * ", /* 4 (x16) */
1301 "32.0 * ", /* 5 (x32) */
1308 "0.0625 * ", /* 12 (d16) */
1309 "0.125 * ", /* 13 (d8) */
1310 "0.25 * ", /* 14 (d4) */
1311 "0.5 * " /* 15 (d2) */
1314 /* Generate a GLSL parameter that does the input modifier computation and return the input register/mask to use */
1315 static void shader_glsl_gen_modifier(enum wined3d_shader_src_modifier src_modifier
,
1316 const char *in_reg
, const char *in_regswizzle
, char *out_str
)
1320 switch (src_modifier
)
1322 case WINED3DSPSM_DZ
: /* Need to handle this in the instructions itself (texld & texcrd). */
1323 case WINED3DSPSM_DW
:
1324 case WINED3DSPSM_NONE
:
1325 sprintf(out_str
, "%s%s", in_reg
, in_regswizzle
);
1327 case WINED3DSPSM_NEG
:
1328 sprintf(out_str
, "-%s%s", in_reg
, in_regswizzle
);
1330 case WINED3DSPSM_NOT
:
1331 sprintf(out_str
, "!%s%s", in_reg
, in_regswizzle
);
1333 case WINED3DSPSM_BIAS
:
1334 sprintf(out_str
, "(%s%s - vec4(0.5)%s)", in_reg
, in_regswizzle
, in_regswizzle
);
1336 case WINED3DSPSM_BIASNEG
:
1337 sprintf(out_str
, "-(%s%s - vec4(0.5)%s)", in_reg
, in_regswizzle
, in_regswizzle
);
1339 case WINED3DSPSM_SIGN
:
1340 sprintf(out_str
, "(2.0 * (%s%s - 0.5))", in_reg
, in_regswizzle
);
1342 case WINED3DSPSM_SIGNNEG
:
1343 sprintf(out_str
, "-(2.0 * (%s%s - 0.5))", in_reg
, in_regswizzle
);
1345 case WINED3DSPSM_COMP
:
1346 sprintf(out_str
, "(1.0 - %s%s)", in_reg
, in_regswizzle
);
1348 case WINED3DSPSM_X2
:
1349 sprintf(out_str
, "(2.0 * %s%s)", in_reg
, in_regswizzle
);
1351 case WINED3DSPSM_X2NEG
:
1352 sprintf(out_str
, "-(2.0 * %s%s)", in_reg
, in_regswizzle
);
1354 case WINED3DSPSM_ABS
:
1355 sprintf(out_str
, "abs(%s%s)", in_reg
, in_regswizzle
);
1357 case WINED3DSPSM_ABSNEG
:
1358 sprintf(out_str
, "-abs(%s%s)", in_reg
, in_regswizzle
);
1361 FIXME("Unhandled modifier %u\n", src_modifier
);
1362 sprintf(out_str
, "%s%s", in_reg
, in_regswizzle
);
1366 /** Writes the GLSL variable name that corresponds to the register that the
1367 * DX opcode parameter is trying to access */
1368 static void shader_glsl_get_register_name(const struct wined3d_shader_register
*reg
,
1369 char *register_name
, BOOL
*is_color
, const struct wined3d_shader_instruction
*ins
)
1371 /* oPos, oFog and oPts in D3D */
1372 static const char * const hwrastout_reg_names
[] = {"vs_out[10]", "vs_out[11].x", "vs_out[11].y"};
1374 const struct wined3d_shader
*shader
= ins
->ctx
->shader
;
1375 const struct wined3d_shader_reg_maps
*reg_maps
= ins
->ctx
->reg_maps
;
1376 const struct wined3d_shader_version
*version
= ®_maps
->shader_version
;
1377 const struct wined3d_gl_info
*gl_info
= ins
->ctx
->gl_info
;
1378 const char *prefix
= shader_glsl_get_prefix(version
->type
);
1379 struct glsl_src_param rel_param0
, rel_param1
;
1381 if (reg
->idx
[0].offset
!= ~0U && reg
->idx
[0].rel_addr
)
1382 shader_glsl_add_src_param(ins
, reg
->idx
[0].rel_addr
, WINED3DSP_WRITEMASK_0
, &rel_param0
);
1383 if (reg
->idx
[1].offset
!= ~0U && reg
->idx
[1].rel_addr
)
1384 shader_glsl_add_src_param(ins
, reg
->idx
[1].rel_addr
, WINED3DSP_WRITEMASK_0
, &rel_param1
);
1389 case WINED3DSPR_TEMP
:
1390 sprintf(register_name
, "R%u", reg
->idx
[0].offset
);
1393 case WINED3DSPR_INPUT
:
1394 /* vertex shaders */
1395 if (version
->type
== WINED3D_SHADER_TYPE_VERTEX
)
1397 struct shader_glsl_ctx_priv
*priv
= ins
->ctx
->backend_data
;
1398 if (priv
->cur_vs_args
->swizzle_map
& (1 << reg
->idx
[0].offset
))
1400 sprintf(register_name
, "%s_in%u", prefix
, reg
->idx
[0].offset
);
1404 if (version
->type
== WINED3D_SHADER_TYPE_GEOMETRY
)
1406 if (reg
->idx
[0].rel_addr
)
1408 if (reg
->idx
[1].rel_addr
)
1409 sprintf(register_name
, "gs_in[%s + %u][%s + %u]",
1410 rel_param0
.param_str
, reg
->idx
[0].offset
, rel_param1
.param_str
, reg
->idx
[1].offset
);
1412 sprintf(register_name
, "gs_in[%s + %u][%u]",
1413 rel_param0
.param_str
, reg
->idx
[0].offset
, reg
->idx
[1].offset
);
1415 else if (reg
->idx
[1].rel_addr
)
1416 sprintf(register_name
, "gs_in[%u][%s + %u]",
1417 reg
->idx
[0].offset
, rel_param1
.param_str
, reg
->idx
[1].offset
);
1419 sprintf(register_name
, "gs_in[%u][%u]", reg
->idx
[0].offset
, reg
->idx
[1].offset
);
1423 /* pixel shaders >= 3.0 */
1424 if (version
->major
>= 3)
1426 DWORD idx
= shader
->u
.ps
.input_reg_map
[reg
->idx
[0].offset
];
1427 unsigned int in_count
= vec4_varyings(version
->major
, gl_info
);
1429 if (reg
->idx
[0].rel_addr
)
1431 /* Removing a + 0 would be an obvious optimization, but
1432 * OS X doesn't see the NOP operation there. */
1435 if (shader
->u
.ps
.declared_in_count
> in_count
)
1437 sprintf(register_name
,
1438 "((%s + %u) > %u ? (%s + %u) > %u ? gl_SecondaryColor : gl_Color : %s_in[%s + %u])",
1439 rel_param0
.param_str
, idx
, in_count
- 1, rel_param0
.param_str
, idx
, in_count
,
1440 prefix
, rel_param0
.param_str
, idx
);
1444 sprintf(register_name
, "%s_in[%s + %u]", prefix
, rel_param0
.param_str
, idx
);
1449 if (shader
->u
.ps
.declared_in_count
> in_count
)
1451 sprintf(register_name
, "((%s) > %u ? (%s) > %u ? gl_SecondaryColor : gl_Color : %s_in[%s])",
1452 rel_param0
.param_str
, in_count
- 1, rel_param0
.param_str
, in_count
,
1453 prefix
, rel_param0
.param_str
);
1457 sprintf(register_name
, "%s_in[%s]", prefix
, rel_param0
.param_str
);
1463 if (idx
== in_count
) sprintf(register_name
, "gl_Color");
1464 else if (idx
== in_count
+ 1) sprintf(register_name
, "gl_SecondaryColor");
1465 else sprintf(register_name
, "%s_in[%u]", prefix
, idx
);
1470 if (!reg
->idx
[0].offset
)
1471 strcpy(register_name
, "gl_Color");
1473 strcpy(register_name
, "gl_SecondaryColor");
1478 case WINED3DSPR_CONST
:
1480 /* Relative addressing */
1481 if (reg
->idx
[0].rel_addr
)
1483 if (reg
->idx
[0].offset
)
1484 sprintf(register_name
, "%s_c[%s + %u]", prefix
, rel_param0
.param_str
, reg
->idx
[0].offset
);
1486 sprintf(register_name
, "%s_c[%s]", prefix
, rel_param0
.param_str
);
1490 if (shader_constant_is_local(shader
, reg
->idx
[0].offset
))
1491 sprintf(register_name
, "%s_lc%u", prefix
, reg
->idx
[0].offset
);
1493 sprintf(register_name
, "%s_c[%u]", prefix
, reg
->idx
[0].offset
);
1498 case WINED3DSPR_CONSTINT
:
1499 sprintf(register_name
, "%s_i[%u]", prefix
, reg
->idx
[0].offset
);
1502 case WINED3DSPR_CONSTBOOL
:
1503 sprintf(register_name
, "%s_b[%u]", prefix
, reg
->idx
[0].offset
);
1506 case WINED3DSPR_TEXTURE
: /* case WINED3DSPR_ADDR: */
1507 if (version
->type
== WINED3D_SHADER_TYPE_PIXEL
)
1508 sprintf(register_name
, "T%u", reg
->idx
[0].offset
);
1510 sprintf(register_name
, "A%u", reg
->idx
[0].offset
);
1513 case WINED3DSPR_LOOP
:
1514 sprintf(register_name
, "aL%u", ins
->ctx
->loop_state
->current_reg
- 1);
1517 case WINED3DSPR_SAMPLER
:
1518 sprintf(register_name
, "%s_sampler%u", prefix
, reg
->idx
[0].offset
);
1521 case WINED3DSPR_COLOROUT
:
1522 if (reg
->idx
[0].offset
>= gl_info
->limits
.buffers
)
1523 WARN("Write to render target %u, only %d supported.\n",
1524 reg
->idx
[0].offset
, gl_info
->limits
.buffers
);
1526 sprintf(register_name
, "gl_FragData[%u]", reg
->idx
[0].offset
);
1529 case WINED3DSPR_RASTOUT
:
1530 sprintf(register_name
, "%s", hwrastout_reg_names
[reg
->idx
[0].offset
]);
1533 case WINED3DSPR_DEPTHOUT
:
1534 sprintf(register_name
, "gl_FragDepth");
1537 case WINED3DSPR_ATTROUT
:
1538 if (!reg
->idx
[0].offset
)
1539 sprintf(register_name
, "%s_out[8]", prefix
);
1541 sprintf(register_name
, "%s_out[9]", prefix
);
1544 case WINED3DSPR_TEXCRDOUT
:
1545 /* Vertex shaders >= 3.0: WINED3DSPR_OUTPUT */
1546 sprintf(register_name
, "%s_out[%u]", prefix
, reg
->idx
[0].offset
);
1549 case WINED3DSPR_MISCTYPE
:
1550 if (!reg
->idx
[0].offset
)
1553 sprintf(register_name
, "vpos");
1555 else if (reg
->idx
[0].offset
== 1)
1557 /* Note that gl_FrontFacing is a bool, while vFace is
1558 * a float for which the sign determines front/back */
1559 sprintf(register_name
, "(gl_FrontFacing ? 1.0 : -1.0)");
1563 FIXME("Unhandled misctype register %u.\n", reg
->idx
[0].offset
);
1564 sprintf(register_name
, "unrecognized_register");
1568 case WINED3DSPR_IMMCONST
:
1569 switch (reg
->immconst_type
)
1571 case WINED3D_IMMCONST_SCALAR
:
1572 switch (reg
->data_type
)
1574 case WINED3D_DATA_FLOAT
:
1575 sprintf(register_name
, "%.8e", *(const float *)reg
->immconst_data
);
1577 case WINED3D_DATA_INT
:
1578 sprintf(register_name
, "%#x", reg
->immconst_data
[0]);
1580 case WINED3D_DATA_RESOURCE
:
1581 case WINED3D_DATA_SAMPLER
:
1582 case WINED3D_DATA_UINT
:
1583 sprintf(register_name
, "%#xu", reg
->immconst_data
[0]);
1586 sprintf(register_name
, "<unhandled data type %#x>", reg
->data_type
);
1591 case WINED3D_IMMCONST_VEC4
:
1592 switch (reg
->data_type
)
1594 case WINED3D_DATA_FLOAT
:
1595 sprintf(register_name
, "vec4(%.8e, %.8e, %.8e, %.8e)",
1596 *(const float *)®
->immconst_data
[0], *(const float *)®
->immconst_data
[1],
1597 *(const float *)®
->immconst_data
[2], *(const float *)®
->immconst_data
[3]);
1599 case WINED3D_DATA_INT
:
1600 sprintf(register_name
, "ivec4(%#x, %#x, %#x, %#x)",
1601 reg
->immconst_data
[0], reg
->immconst_data
[1],
1602 reg
->immconst_data
[2], reg
->immconst_data
[3]);
1604 case WINED3D_DATA_RESOURCE
:
1605 case WINED3D_DATA_SAMPLER
:
1606 case WINED3D_DATA_UINT
:
1607 sprintf(register_name
, "uvec4(%#xu, %#xu, %#xu, %#xu)",
1608 reg
->immconst_data
[0], reg
->immconst_data
[1],
1609 reg
->immconst_data
[2], reg
->immconst_data
[3]);
1612 sprintf(register_name
, "<unhandled data type %#x>", reg
->data_type
);
1618 FIXME("Unhandled immconst type %#x\n", reg
->immconst_type
);
1619 sprintf(register_name
, "<unhandled_immconst_type %#x>", reg
->immconst_type
);
1623 case WINED3DSPR_CONSTBUFFER
:
1624 if (reg
->idx
[1].rel_addr
)
1625 sprintf(register_name
, "%s_cb%u[%s + %u]",
1626 prefix
, reg
->idx
[0].offset
, rel_param1
.param_str
, reg
->idx
[1].offset
);
1628 sprintf(register_name
, "%s_cb%u[%u]", prefix
, reg
->idx
[0].offset
, reg
->idx
[1].offset
);
1631 case WINED3DSPR_PRIMID
:
1632 sprintf(register_name
, "uint(gl_PrimitiveIDIn)");
1636 FIXME("Unhandled register type %#x.\n", reg
->type
);
1637 sprintf(register_name
, "unrecognized_register");
1642 static void shader_glsl_write_mask_to_str(DWORD write_mask
, char *str
)
1645 if (write_mask
& WINED3DSP_WRITEMASK_0
) *str
++ = 'x';
1646 if (write_mask
& WINED3DSP_WRITEMASK_1
) *str
++ = 'y';
1647 if (write_mask
& WINED3DSP_WRITEMASK_2
) *str
++ = 'z';
1648 if (write_mask
& WINED3DSP_WRITEMASK_3
) *str
++ = 'w';
1652 /* Get the GLSL write mask for the destination register */
1653 static DWORD
shader_glsl_get_write_mask(const struct wined3d_shader_dst_param
*param
, char *write_mask
)
1655 DWORD mask
= param
->write_mask
;
1657 if (shader_is_scalar(¶m
->reg
))
1659 mask
= WINED3DSP_WRITEMASK_0
;
1664 shader_glsl_write_mask_to_str(mask
, write_mask
);
1670 static unsigned int shader_glsl_get_write_mask_size(DWORD write_mask
) {
1671 unsigned int size
= 0;
1673 if (write_mask
& WINED3DSP_WRITEMASK_0
) ++size
;
1674 if (write_mask
& WINED3DSP_WRITEMASK_1
) ++size
;
1675 if (write_mask
& WINED3DSP_WRITEMASK_2
) ++size
;
1676 if (write_mask
& WINED3DSP_WRITEMASK_3
) ++size
;
1681 static void shader_glsl_swizzle_to_str(const DWORD swizzle
, BOOL fixup
, DWORD mask
, char *str
)
1683 /* For registers of type WINED3DDECLTYPE_D3DCOLOR, data is stored as "bgra",
1684 * but addressed as "rgba". To fix this we need to swap the register's x
1685 * and z components. */
1686 const char *swizzle_chars
= fixup
? "zyxw" : "xyzw";
1689 /* swizzle bits fields: wwzzyyxx */
1690 if (mask
& WINED3DSP_WRITEMASK_0
) *str
++ = swizzle_chars
[swizzle
& 0x03];
1691 if (mask
& WINED3DSP_WRITEMASK_1
) *str
++ = swizzle_chars
[(swizzle
>> 2) & 0x03];
1692 if (mask
& WINED3DSP_WRITEMASK_2
) *str
++ = swizzle_chars
[(swizzle
>> 4) & 0x03];
1693 if (mask
& WINED3DSP_WRITEMASK_3
) *str
++ = swizzle_chars
[(swizzle
>> 6) & 0x03];
1697 static void shader_glsl_get_swizzle(const struct wined3d_shader_src_param
*param
,
1698 BOOL fixup
, DWORD mask
, char *swizzle_str
)
1700 if (shader_is_scalar(¶m
->reg
))
1701 *swizzle_str
= '\0';
1703 shader_glsl_swizzle_to_str(param
->swizzle
, fixup
, mask
, swizzle_str
);
1706 /* From a given parameter token, generate the corresponding GLSL string.
1707 * Also, return the actual register name and swizzle in case the
1708 * caller needs this information as well. */
1709 static void shader_glsl_add_src_param(const struct wined3d_shader_instruction
*ins
,
1710 const struct wined3d_shader_src_param
*wined3d_src
, DWORD mask
, struct glsl_src_param
*glsl_src
)
1712 BOOL is_color
= FALSE
;
1713 char swizzle_str
[6];
1715 glsl_src
->reg_name
[0] = '\0';
1716 glsl_src
->param_str
[0] = '\0';
1717 swizzle_str
[0] = '\0';
1719 shader_glsl_get_register_name(&wined3d_src
->reg
, glsl_src
->reg_name
, &is_color
, ins
);
1720 shader_glsl_get_swizzle(wined3d_src
, is_color
, mask
, swizzle_str
);
1722 if (wined3d_src
->reg
.type
== WINED3DSPR_IMMCONST
|| wined3d_src
->reg
.type
== WINED3DSPR_PRIMID
)
1724 shader_glsl_gen_modifier(wined3d_src
->modifiers
, glsl_src
->reg_name
, swizzle_str
, glsl_src
->param_str
);
1728 char param_str
[200];
1730 shader_glsl_gen_modifier(wined3d_src
->modifiers
, glsl_src
->reg_name
, swizzle_str
, param_str
);
1732 switch (wined3d_src
->reg
.data_type
)
1734 case WINED3D_DATA_FLOAT
:
1735 sprintf(glsl_src
->param_str
, "%s", param_str
);
1737 case WINED3D_DATA_INT
:
1738 sprintf(glsl_src
->param_str
, "floatBitsToInt(%s)", param_str
);
1740 case WINED3D_DATA_RESOURCE
:
1741 case WINED3D_DATA_SAMPLER
:
1742 case WINED3D_DATA_UINT
:
1743 sprintf(glsl_src
->param_str
, "floatBitsToUint(%s)", param_str
);
1746 FIXME("Unhandled data type %#x.\n", wined3d_src
->reg
.data_type
);
1747 sprintf(glsl_src
->param_str
, "%s", param_str
);
1753 /* From a given parameter token, generate the corresponding GLSL string.
1754 * Also, return the actual register name and swizzle in case the
1755 * caller needs this information as well. */
1756 static DWORD
shader_glsl_add_dst_param(const struct wined3d_shader_instruction
*ins
,
1757 const struct wined3d_shader_dst_param
*wined3d_dst
, struct glsl_dst_param
*glsl_dst
)
1759 BOOL is_color
= FALSE
;
1761 glsl_dst
->mask_str
[0] = '\0';
1762 glsl_dst
->reg_name
[0] = '\0';
1764 shader_glsl_get_register_name(&wined3d_dst
->reg
, glsl_dst
->reg_name
, &is_color
, ins
);
1765 return shader_glsl_get_write_mask(wined3d_dst
, glsl_dst
->mask_str
);
1768 /* Append the destination part of the instruction to the buffer, return the effective write mask */
1769 static DWORD
shader_glsl_append_dst_ext(struct wined3d_shader_buffer
*buffer
,
1770 const struct wined3d_shader_instruction
*ins
, const struct wined3d_shader_dst_param
*dst
)
1772 struct glsl_dst_param glsl_dst
;
1775 if ((mask
= shader_glsl_add_dst_param(ins
, dst
, &glsl_dst
)))
1777 switch (dst
->reg
.data_type
)
1779 case WINED3D_DATA_FLOAT
:
1780 shader_addline(buffer
, "%s%s = %s(",
1781 glsl_dst
.reg_name
, glsl_dst
.mask_str
, shift_glsl_tab
[dst
->shift
]);
1783 case WINED3D_DATA_INT
:
1784 shader_addline(buffer
, "%s%s = %sintBitsToFloat(",
1785 glsl_dst
.reg_name
, glsl_dst
.mask_str
, shift_glsl_tab
[dst
->shift
]);
1787 case WINED3D_DATA_RESOURCE
:
1788 case WINED3D_DATA_SAMPLER
:
1789 case WINED3D_DATA_UINT
:
1790 shader_addline(buffer
, "%s%s = %suintBitsToFloat(",
1791 glsl_dst
.reg_name
, glsl_dst
.mask_str
, shift_glsl_tab
[dst
->shift
]);
1794 FIXME("Unhandled data type %#x.\n", dst
->reg
.data_type
);
1795 shader_addline(buffer
, "%s%s = %s(",
1796 glsl_dst
.reg_name
, glsl_dst
.mask_str
, shift_glsl_tab
[dst
->shift
]);
1804 /* Append the destination part of the instruction to the buffer, return the effective write mask */
1805 static DWORD
shader_glsl_append_dst(struct wined3d_shader_buffer
*buffer
, const struct wined3d_shader_instruction
*ins
)
1807 return shader_glsl_append_dst_ext(buffer
, ins
, &ins
->dst
[0]);
1810 /** Process GLSL instruction modifiers */
1811 static void shader_glsl_add_instruction_modifiers(const struct wined3d_shader_instruction
*ins
)
1813 struct glsl_dst_param dst_param
;
1816 if (!ins
->dst_count
) return;
1818 modifiers
= ins
->dst
[0].modifiers
;
1819 if (!modifiers
) return;
1821 shader_glsl_add_dst_param(ins
, &ins
->dst
[0], &dst_param
);
1823 if (modifiers
& WINED3DSPDM_SATURATE
)
1825 /* _SAT means to clamp the value of the register to between 0 and 1 */
1826 shader_addline(ins
->ctx
->buffer
, "%s%s = clamp(%s%s, 0.0, 1.0);\n", dst_param
.reg_name
,
1827 dst_param
.mask_str
, dst_param
.reg_name
, dst_param
.mask_str
);
1830 if (modifiers
& WINED3DSPDM_MSAMPCENTROID
)
1832 FIXME("_centroid modifier not handled\n");
1835 if (modifiers
& WINED3DSPDM_PARTIALPRECISION
)
1837 /* MSDN says this modifier can be safely ignored, so that's what we'll do. */
1841 static const char *shader_glsl_get_rel_op(enum wined3d_shader_rel_op op
)
1845 case WINED3D_SHADER_REL_OP_GT
: return ">";
1846 case WINED3D_SHADER_REL_OP_EQ
: return "==";
1847 case WINED3D_SHADER_REL_OP_GE
: return ">=";
1848 case WINED3D_SHADER_REL_OP_LT
: return "<";
1849 case WINED3D_SHADER_REL_OP_NE
: return "!=";
1850 case WINED3D_SHADER_REL_OP_LE
: return "<=";
1852 FIXME("Unrecognized operator %#x.\n", op
);
1857 static void shader_glsl_get_sample_function(const struct wined3d_shader_context
*ctx
,
1858 DWORD sampler_idx
, DWORD flags
, struct glsl_sample_function
*sample_function
)
1860 enum wined3d_sampler_texture_type sampler_type
= ctx
->reg_maps
->sampler_type
[sampler_idx
];
1861 const struct wined3d_gl_info
*gl_info
= ctx
->gl_info
;
1862 BOOL shadow
= ctx
->reg_maps
->shader_version
.type
== WINED3D_SHADER_TYPE_PIXEL
1863 && (((const struct shader_glsl_ctx_priv
*)ctx
->backend_data
)->cur_ps_args
->shadow
& (1 << sampler_idx
));
1864 BOOL projected
= flags
& WINED3D_GLSL_SAMPLE_PROJECTED
;
1865 BOOL texrect
= flags
& WINED3D_GLSL_SAMPLE_NPOT
&& gl_info
->supported
[ARB_TEXTURE_RECTANGLE
];
1866 BOOL lod
= flags
& WINED3D_GLSL_SAMPLE_LOD
;
1867 BOOL grad
= flags
& WINED3D_GLSL_SAMPLE_GRAD
;
1869 /* Note that there's no such thing as a projected cube texture. */
1870 switch(sampler_type
) {
1876 sample_function
->name
= projected
? "shadow1DProjLod" : "shadow1DLod";
1880 if (gl_info
->supported
[EXT_GPU_SHADER4
])
1881 sample_function
->name
= projected
? "shadow1DProjGrad" : "shadow1DGrad";
1882 else if (gl_info
->supported
[ARB_SHADER_TEXTURE_LOD
])
1883 sample_function
->name
= projected
? "shadow1DProjGradARB" : "shadow1DGradARB";
1886 FIXME("Unsupported 1D shadow grad function.\n");
1887 sample_function
->name
= "unsupported1DGrad";
1892 sample_function
->name
= projected
? "shadow1DProj" : "shadow1D";
1894 sample_function
->coord_mask
= WINED3DSP_WRITEMASK_0
| WINED3DSP_WRITEMASK_1
;
1900 sample_function
->name
= projected
? "texture1DProjLod" : "texture1DLod";
1904 if (gl_info
->supported
[EXT_GPU_SHADER4
])
1905 sample_function
->name
= projected
? "texture1DProjGrad" : "texture1DGrad";
1906 else if (gl_info
->supported
[ARB_SHADER_TEXTURE_LOD
])
1907 sample_function
->name
= projected
? "texture1DProjGradARB" : "texture1DGradARB";
1910 FIXME("Unsupported 1D grad function.\n");
1911 sample_function
->name
= "unsupported1DGrad";
1916 sample_function
->name
= projected
? "texture1DProj" : "texture1D";
1918 sample_function
->coord_mask
= WINED3DSP_WRITEMASK_0
;
1929 sample_function
->name
= projected
? "shadow2DRectProjLod" : "shadow2DRectLod";
1933 if (gl_info
->supported
[EXT_GPU_SHADER4
])
1934 sample_function
->name
= projected
? "shadow2DRectProjGrad" : "shadow2DRectGrad";
1935 else if (gl_info
->supported
[ARB_SHADER_TEXTURE_LOD
])
1936 sample_function
->name
= projected
? "shadow2DRectProjGradARB" : "shadow2DRectGradARB";
1939 FIXME("Unsupported RECT shadow grad function.\n");
1940 sample_function
->name
= "unsupported2DRectGrad";
1945 sample_function
->name
= projected
? "shadow2DRectProj" : "shadow2DRect";
1952 sample_function
->name
= projected
? "shadow2DProjLod" : "shadow2DLod";
1956 if (gl_info
->supported
[EXT_GPU_SHADER4
])
1957 sample_function
->name
= projected
? "shadow2DProjGrad" : "shadow2DGrad";
1958 else if (gl_info
->supported
[ARB_SHADER_TEXTURE_LOD
])
1959 sample_function
->name
= projected
? "shadow2DProjGradARB" : "shadow2DGradARB";
1962 FIXME("Unsupported 2D shadow grad function.\n");
1963 sample_function
->name
= "unsupported2DGrad";
1968 sample_function
->name
= projected
? "shadow2DProj" : "shadow2D";
1971 sample_function
->coord_mask
= WINED3DSP_WRITEMASK_0
| WINED3DSP_WRITEMASK_1
| WINED3DSP_WRITEMASK_2
;
1979 sample_function
->name
= projected
? "texture2DRectProjLod" : "texture2DRectLod";
1983 if (gl_info
->supported
[EXT_GPU_SHADER4
])
1984 sample_function
->name
= projected
? "texture2DRectProjGrad" : "texture2DRectGrad";
1985 else if (gl_info
->supported
[ARB_SHADER_TEXTURE_LOD
])
1986 sample_function
->name
= projected
? "texture2DRectProjGradARB" : "texture2DRectGradARB";
1989 FIXME("Unsupported RECT grad function.\n");
1990 sample_function
->name
= "unsupported2DRectGrad";
1995 sample_function
->name
= projected
? "texture2DRectProj" : "texture2DRect";
2002 sample_function
->name
= projected
? "texture2DProjLod" : "texture2DLod";
2006 if (gl_info
->supported
[EXT_GPU_SHADER4
])
2007 sample_function
->name
= projected
? "texture2DProjGrad" : "texture2DGrad";
2008 else if (gl_info
->supported
[ARB_SHADER_TEXTURE_LOD
])
2009 sample_function
->name
= projected
? "texture2DProjGradARB" : "texture2DGradARB";
2012 FIXME("Unsupported 2D grad function.\n");
2013 sample_function
->name
= "unsupported2DGrad";
2018 sample_function
->name
= projected
? "texture2DProj" : "texture2D";
2021 sample_function
->coord_mask
= WINED3DSP_WRITEMASK_0
| WINED3DSP_WRITEMASK_1
;
2025 case WINED3DSTT_CUBE
:
2028 FIXME("Unsupported Cube shadow function.\n");
2029 sample_function
->name
= "unsupportedCubeShadow";
2030 sample_function
->coord_mask
= 0;
2036 sample_function
->name
= "textureCubeLod";
2040 if (gl_info
->supported
[EXT_GPU_SHADER4
])
2041 sample_function
->name
= "textureCubeGrad";
2042 else if (gl_info
->supported
[ARB_SHADER_TEXTURE_LOD
])
2043 sample_function
->name
= "textureCubeGradARB";
2046 FIXME("Unsupported Cube grad function.\n");
2047 sample_function
->name
= "unsupportedCubeGrad";
2052 sample_function
->name
= "textureCube";
2054 sample_function
->coord_mask
= WINED3DSP_WRITEMASK_0
| WINED3DSP_WRITEMASK_1
| WINED3DSP_WRITEMASK_2
;
2058 case WINED3DSTT_VOLUME
:
2061 FIXME("Unsupported 3D shadow function.\n");
2062 sample_function
->name
= "unsupported3DShadow";
2063 sample_function
->coord_mask
= 0;
2069 sample_function
->name
= projected
? "texture3DProjLod" : "texture3DLod";
2073 if (gl_info
->supported
[EXT_GPU_SHADER4
])
2074 sample_function
->name
= projected
? "texture3DProjGrad" : "texture3DGrad";
2075 else if (gl_info
->supported
[ARB_SHADER_TEXTURE_LOD
])
2076 sample_function
->name
= projected
? "texture3DProjGradARB" : "texture3DGradARB";
2079 FIXME("Unsupported 3D grad function.\n");
2080 sample_function
->name
= "unsupported3DGrad";
2085 sample_function
->name
= projected
? "texture3DProj" : "texture3D";
2087 sample_function
->coord_mask
= WINED3DSP_WRITEMASK_0
| WINED3DSP_WRITEMASK_1
| WINED3DSP_WRITEMASK_2
;
2092 sample_function
->name
= "";
2093 sample_function
->coord_mask
= 0;
2094 FIXME("Unrecognized sampler type: %#x;\n", sampler_type
);
2099 static void shader_glsl_append_fixup_arg(char *arguments
, const char *reg_name
,
2100 BOOL sign_fixup
, enum fixup_channel_source channel_source
)
2102 switch(channel_source
)
2104 case CHANNEL_SOURCE_ZERO
:
2105 strcat(arguments
, "0.0");
2108 case CHANNEL_SOURCE_ONE
:
2109 strcat(arguments
, "1.0");
2112 case CHANNEL_SOURCE_X
:
2113 strcat(arguments
, reg_name
);
2114 strcat(arguments
, ".x");
2117 case CHANNEL_SOURCE_Y
:
2118 strcat(arguments
, reg_name
);
2119 strcat(arguments
, ".y");
2122 case CHANNEL_SOURCE_Z
:
2123 strcat(arguments
, reg_name
);
2124 strcat(arguments
, ".z");
2127 case CHANNEL_SOURCE_W
:
2128 strcat(arguments
, reg_name
);
2129 strcat(arguments
, ".w");
2133 FIXME("Unhandled channel source %#x\n", channel_source
);
2134 strcat(arguments
, "undefined");
2138 if (sign_fixup
) strcat(arguments
, " * 2.0 - 1.0");
2141 static void shader_glsl_color_correction_ext(struct wined3d_shader_buffer
*buffer
,
2142 const char *reg_name
, DWORD mask
, struct color_fixup_desc fixup
)
2144 unsigned int mask_size
, remaining
;
2145 DWORD fixup_mask
= 0;
2146 char arguments
[256];
2149 if (fixup
.x_sign_fixup
|| fixup
.x_source
!= CHANNEL_SOURCE_X
) fixup_mask
|= WINED3DSP_WRITEMASK_0
;
2150 if (fixup
.y_sign_fixup
|| fixup
.y_source
!= CHANNEL_SOURCE_Y
) fixup_mask
|= WINED3DSP_WRITEMASK_1
;
2151 if (fixup
.z_sign_fixup
|| fixup
.z_source
!= CHANNEL_SOURCE_Z
) fixup_mask
|= WINED3DSP_WRITEMASK_2
;
2152 if (fixup
.w_sign_fixup
|| fixup
.w_source
!= CHANNEL_SOURCE_W
) fixup_mask
|= WINED3DSP_WRITEMASK_3
;
2153 if (!(mask
&= fixup_mask
))
2156 if (is_complex_fixup(fixup
))
2158 enum complex_fixup complex_fixup
= get_complex_fixup(fixup
);
2159 FIXME("Complex fixup (%#x) not supported\n",complex_fixup
);
2163 shader_glsl_write_mask_to_str(mask
, mask_str
);
2164 mask_size
= shader_glsl_get_write_mask_size(mask
);
2166 arguments
[0] = '\0';
2167 remaining
= mask_size
;
2168 if (mask
& WINED3DSP_WRITEMASK_0
)
2170 shader_glsl_append_fixup_arg(arguments
, reg_name
, fixup
.x_sign_fixup
, fixup
.x_source
);
2171 if (--remaining
) strcat(arguments
, ", ");
2173 if (mask
& WINED3DSP_WRITEMASK_1
)
2175 shader_glsl_append_fixup_arg(arguments
, reg_name
, fixup
.y_sign_fixup
, fixup
.y_source
);
2176 if (--remaining
) strcat(arguments
, ", ");
2178 if (mask
& WINED3DSP_WRITEMASK_2
)
2180 shader_glsl_append_fixup_arg(arguments
, reg_name
, fixup
.z_sign_fixup
, fixup
.z_source
);
2181 if (--remaining
) strcat(arguments
, ", ");
2183 if (mask
& WINED3DSP_WRITEMASK_3
)
2185 shader_glsl_append_fixup_arg(arguments
, reg_name
, fixup
.w_sign_fixup
, fixup
.w_source
);
2186 if (--remaining
) strcat(arguments
, ", ");
2190 shader_addline(buffer
, "%s%s = vec%u(%s);\n", reg_name
, mask_str
, mask_size
, arguments
);
2192 shader_addline(buffer
, "%s%s = %s;\n", reg_name
, mask_str
, arguments
);
2195 static void shader_glsl_color_correction(const struct wined3d_shader_instruction
*ins
, struct color_fixup_desc fixup
)
2200 shader_glsl_get_register_name(&ins
->dst
[0].reg
, reg_name
, &is_color
, ins
);
2201 shader_glsl_color_correction_ext(ins
->ctx
->buffer
, reg_name
, ins
->dst
[0].write_mask
, fixup
);
2204 static void PRINTF_ATTR(8, 9) shader_glsl_gen_sample_code(const struct wined3d_shader_instruction
*ins
,
2205 DWORD sampler
, const struct glsl_sample_function
*sample_function
, DWORD swizzle
,
2206 const char *dx
, const char *dy
, const char *bias
, const char *coord_reg_fmt
, ...)
2208 const struct wined3d_shader_version
*version
= &ins
->ctx
->reg_maps
->shader_version
;
2209 char dst_swizzle
[6];
2210 struct color_fixup_desc fixup
;
2211 BOOL np2_fixup
= FALSE
;
2214 shader_glsl_swizzle_to_str(swizzle
, FALSE
, ins
->dst
[0].write_mask
, dst_swizzle
);
2216 if (version
->type
== WINED3D_SHADER_TYPE_PIXEL
)
2218 const struct shader_glsl_ctx_priv
*priv
= ins
->ctx
->backend_data
;
2219 fixup
= priv
->cur_ps_args
->color_fixup
[sampler
];
2221 if(priv
->cur_ps_args
->np2_fixup
& (1 << sampler
)) {
2223 FIXME("Biased sampling from NP2 textures is unsupported\n");
2231 fixup
= COLOR_FIXUP_IDENTITY
; /* FIXME: Vshader color fixup */
2234 shader_glsl_append_dst(ins
->ctx
->buffer
, ins
);
2236 shader_addline(ins
->ctx
->buffer
, "%s(%s_sampler%u, ",
2237 sample_function
->name
, shader_glsl_get_prefix(version
->type
), sampler
);
2239 va_start(args
, coord_reg_fmt
);
2240 shader_vaddline(ins
->ctx
->buffer
, coord_reg_fmt
, args
);
2244 shader_addline(ins
->ctx
->buffer
, ", %s)%s);\n", bias
, dst_swizzle
);
2247 const struct shader_glsl_ctx_priv
*priv
= ins
->ctx
->backend_data
;
2248 const unsigned char idx
= priv
->cur_np2fixup_info
->idx
[sampler
];
2250 shader_addline(ins
->ctx
->buffer
, " * ps_samplerNP2Fixup[%u].%s)%s);\n", idx
>> 1,
2251 (idx
% 2) ? "zw" : "xy", dst_swizzle
);
2252 } else if(dx
&& dy
) {
2253 shader_addline(ins
->ctx
->buffer
, ", %s, %s)%s);\n", dx
, dy
, dst_swizzle
);
2255 shader_addline(ins
->ctx
->buffer
, ")%s);\n", dst_swizzle
);
2259 if(!is_identity_fixup(fixup
)) {
2260 shader_glsl_color_correction(ins
, fixup
);
2264 /*****************************************************************************
2265 * Begin processing individual instruction opcodes
2266 ****************************************************************************/
2268 static void shader_glsl_binop(const struct wined3d_shader_instruction
*ins
)
2270 struct wined3d_shader_buffer
*buffer
= ins
->ctx
->buffer
;
2271 struct glsl_src_param src0_param
;
2272 struct glsl_src_param src1_param
;
2276 /* Determine the GLSL operator to use based on the opcode */
2277 switch (ins
->handler_idx
)
2279 case WINED3DSIH_ADD
: op
= "+"; break;
2280 case WINED3DSIH_AND
: op
= "&"; break;
2281 case WINED3DSIH_DIV
: op
= "/"; break;
2282 case WINED3DSIH_IADD
: op
= "+"; break;
2283 case WINED3DSIH_MUL
: op
= "*"; break;
2284 case WINED3DSIH_SUB
: op
= "-"; break;
2285 case WINED3DSIH_USHR
: op
= ">>"; break;
2286 case WINED3DSIH_XOR
: op
= "^"; break;
2288 op
= "<unhandled operator>";
2289 FIXME("Opcode %#x not yet handled in GLSL\n", ins
->handler_idx
);
2293 write_mask
= shader_glsl_append_dst(buffer
, ins
);
2294 shader_glsl_add_src_param(ins
, &ins
->src
[0], write_mask
, &src0_param
);
2295 shader_glsl_add_src_param(ins
, &ins
->src
[1], write_mask
, &src1_param
);
2296 shader_addline(buffer
, "%s %s %s);\n", src0_param
.param_str
, op
, src1_param
.param_str
);
2299 static void shader_glsl_relop(const struct wined3d_shader_instruction
*ins
)
2301 struct wined3d_shader_buffer
*buffer
= ins
->ctx
->buffer
;
2302 struct glsl_src_param src0_param
;
2303 struct glsl_src_param src1_param
;
2304 unsigned int mask_size
;
2308 write_mask
= shader_glsl_append_dst(buffer
, ins
);
2309 mask_size
= shader_glsl_get_write_mask_size(write_mask
);
2310 shader_glsl_add_src_param(ins
, &ins
->src
[0], write_mask
, &src0_param
);
2311 shader_glsl_add_src_param(ins
, &ins
->src
[1], write_mask
, &src1_param
);
2315 switch (ins
->handler_idx
)
2317 case WINED3DSIH_EQ
: op
= "equal"; break;
2318 case WINED3DSIH_GE
: op
= "greaterThanEqual"; break;
2319 case WINED3DSIH_IGE
: op
= "greaterThanEqual"; break;
2320 case WINED3DSIH_LT
: op
= "lessThan"; break;
2322 op
= "<unhandled operator>";
2323 ERR("Unhandled opcode %#x.\n", ins
->handler_idx
);
2327 shader_addline(buffer
, "uvec%u(%s(%s, %s)) * 0xffffffffu);\n",
2328 mask_size
, op
, src0_param
.param_str
, src1_param
.param_str
);
2332 switch (ins
->handler_idx
)
2334 case WINED3DSIH_EQ
: op
= "=="; break;
2335 case WINED3DSIH_GE
: op
= ">="; break;
2336 case WINED3DSIH_IGE
: op
= ">="; break;
2337 case WINED3DSIH_LT
: op
= "<"; break;
2339 op
= "<unhandled operator>";
2340 ERR("Unhandled opcode %#x.\n", ins
->handler_idx
);
2344 shader_addline(buffer
, "%s %s %s ? 0xffffffffu : 0u);\n",
2345 src0_param
.param_str
, op
, src1_param
.param_str
);
2349 static void shader_glsl_imul(const struct wined3d_shader_instruction
*ins
)
2351 struct wined3d_shader_buffer
*buffer
= ins
->ctx
->buffer
;
2352 struct glsl_src_param src0_param
;
2353 struct glsl_src_param src1_param
;
2356 /* If we have ARB_gpu_shader5 or GLSL 4.0, we can use imulExtended(). If
2357 * not, we can emulate it. */
2358 if (ins
->dst
[0].reg
.type
!= WINED3DSPR_NULL
)
2359 FIXME("64-bit integer multiplies not implemented.\n");
2361 if (ins
->dst
[1].reg
.type
!= WINED3DSPR_NULL
)
2363 write_mask
= shader_glsl_append_dst_ext(buffer
, ins
, &ins
->dst
[1]);
2364 shader_glsl_add_src_param(ins
, &ins
->src
[0], write_mask
, &src0_param
);
2365 shader_glsl_add_src_param(ins
, &ins
->src
[1], write_mask
, &src1_param
);
2367 shader_addline(ins
->ctx
->buffer
, "%s * %s);\n",
2368 src0_param
.param_str
, src1_param
.param_str
);
2372 static void shader_glsl_udiv(const struct wined3d_shader_instruction
*ins
)
2374 struct wined3d_shader_buffer
*buffer
= ins
->ctx
->buffer
;
2375 struct glsl_src_param src0_param
, src1_param
;
2378 if (ins
->dst
[0].reg
.type
!= WINED3DSPR_NULL
)
2381 if (ins
->dst
[1].reg
.type
!= WINED3DSPR_NULL
)
2385 write_mask
= shader_glsl_get_write_mask(&ins
->dst
[0], dst_mask
);
2386 shader_glsl_add_src_param(ins
, &ins
->src
[0], write_mask
, &src0_param
);
2387 shader_glsl_add_src_param(ins
, &ins
->src
[1], write_mask
, &src1_param
);
2388 shader_addline(buffer
, "tmp0%s = %s / %s;\n",
2389 dst_mask
, src0_param
.param_str
, src1_param
.param_str
);
2391 write_mask
= shader_glsl_append_dst_ext(buffer
, ins
, &ins
->dst
[1]);
2392 shader_glsl_add_src_param(ins
, &ins
->src
[0], write_mask
, &src0_param
);
2393 shader_glsl_add_src_param(ins
, &ins
->src
[1], write_mask
, &src1_param
);
2394 shader_addline(buffer
, "%s %% %s));\n", src0_param
.param_str
, src1_param
.param_str
);
2396 shader_glsl_append_dst_ext(buffer
, ins
, &ins
->dst
[0]);
2397 shader_addline(buffer
, "tmp0%s);\n", dst_mask
);
2401 write_mask
= shader_glsl_append_dst_ext(buffer
, ins
, &ins
->dst
[0]);
2402 shader_glsl_add_src_param(ins
, &ins
->src
[0], write_mask
, &src0_param
);
2403 shader_glsl_add_src_param(ins
, &ins
->src
[1], write_mask
, &src1_param
);
2404 shader_addline(buffer
, "%s / %s);\n", src0_param
.param_str
, src1_param
.param_str
);
2407 else if (ins
->dst
[1].reg
.type
!= WINED3DSPR_NULL
)
2409 write_mask
= shader_glsl_append_dst_ext(buffer
, ins
, &ins
->dst
[1]);
2410 shader_glsl_add_src_param(ins
, &ins
->src
[0], write_mask
, &src0_param
);
2411 shader_glsl_add_src_param(ins
, &ins
->src
[1], write_mask
, &src1_param
);
2412 shader_addline(buffer
, "%s %% %s);\n", src0_param
.param_str
, src1_param
.param_str
);
2416 /* Process the WINED3DSIO_MOV opcode using GLSL (dst = src) */
2417 static void shader_glsl_mov(const struct wined3d_shader_instruction
*ins
)
2419 const struct wined3d_gl_info
*gl_info
= ins
->ctx
->gl_info
;
2420 struct wined3d_shader_buffer
*buffer
= ins
->ctx
->buffer
;
2421 struct glsl_src_param src0_param
;
2424 write_mask
= shader_glsl_append_dst(buffer
, ins
);
2425 shader_glsl_add_src_param(ins
, &ins
->src
[0], write_mask
, &src0_param
);
2427 /* In vs_1_1 WINED3DSIO_MOV can write to the address register. In later
2428 * shader versions WINED3DSIO_MOVA is used for this. */
2429 if (ins
->ctx
->reg_maps
->shader_version
.major
== 1
2430 && ins
->ctx
->reg_maps
->shader_version
.type
== WINED3D_SHADER_TYPE_VERTEX
2431 && ins
->dst
[0].reg
.type
== WINED3DSPR_ADDR
)
2433 /* This is a simple floor() */
2434 unsigned int mask_size
= shader_glsl_get_write_mask_size(write_mask
);
2435 if (mask_size
> 1) {
2436 shader_addline(buffer
, "ivec%d(floor(%s)));\n", mask_size
, src0_param
.param_str
);
2438 shader_addline(buffer
, "int(floor(%s)));\n", src0_param
.param_str
);
2441 else if(ins
->handler_idx
== WINED3DSIH_MOVA
)
2443 /* We need to *round* to the nearest int here. */
2444 unsigned int mask_size
= shader_glsl_get_write_mask_size(write_mask
);
2446 if (gl_info
->supported
[EXT_GPU_SHADER4
])
2449 shader_addline(buffer
, "ivec%d(round(%s)));\n", mask_size
, src0_param
.param_str
);
2451 shader_addline(buffer
, "int(round(%s)));\n", src0_param
.param_str
);
2456 shader_addline(buffer
, "ivec%d(floor(abs(%s) + vec%d(0.5)) * sign(%s)));\n",
2457 mask_size
, src0_param
.param_str
, mask_size
, src0_param
.param_str
);
2459 shader_addline(buffer
, "int(floor(abs(%s) + 0.5) * sign(%s)));\n",
2460 src0_param
.param_str
, src0_param
.param_str
);
2465 shader_addline(buffer
, "%s);\n", src0_param
.param_str
);
2469 /* Process the dot product operators DP3 and DP4 in GLSL (dst = dot(src0, src1)) */
2470 static void shader_glsl_dot(const struct wined3d_shader_instruction
*ins
)
2472 struct wined3d_shader_buffer
*buffer
= ins
->ctx
->buffer
;
2473 struct glsl_src_param src0_param
;
2474 struct glsl_src_param src1_param
;
2475 DWORD dst_write_mask
, src_write_mask
;
2476 unsigned int dst_size
= 0;
2478 dst_write_mask
= shader_glsl_append_dst(buffer
, ins
);
2479 dst_size
= shader_glsl_get_write_mask_size(dst_write_mask
);
2481 /* dp3 works on vec3, dp4 on vec4 */
2482 if (ins
->handler_idx
== WINED3DSIH_DP4
)
2484 src_write_mask
= WINED3DSP_WRITEMASK_ALL
;
2486 src_write_mask
= WINED3DSP_WRITEMASK_0
| WINED3DSP_WRITEMASK_1
| WINED3DSP_WRITEMASK_2
;
2489 shader_glsl_add_src_param(ins
, &ins
->src
[0], src_write_mask
, &src0_param
);
2490 shader_glsl_add_src_param(ins
, &ins
->src
[1], src_write_mask
, &src1_param
);
2493 shader_addline(buffer
, "vec%d(dot(%s, %s)));\n", dst_size
, src0_param
.param_str
, src1_param
.param_str
);
2495 shader_addline(buffer
, "dot(%s, %s));\n", src0_param
.param_str
, src1_param
.param_str
);
2499 /* Note that this instruction has some restrictions. The destination write mask
2500 * can't contain the w component, and the source swizzles have to be .xyzw */
2501 static void shader_glsl_cross(const struct wined3d_shader_instruction
*ins
)
2503 DWORD src_mask
= WINED3DSP_WRITEMASK_0
| WINED3DSP_WRITEMASK_1
| WINED3DSP_WRITEMASK_2
;
2504 struct glsl_src_param src0_param
;
2505 struct glsl_src_param src1_param
;
2508 shader_glsl_get_write_mask(&ins
->dst
[0], dst_mask
);
2509 shader_glsl_append_dst(ins
->ctx
->buffer
, ins
);
2510 shader_glsl_add_src_param(ins
, &ins
->src
[0], src_mask
, &src0_param
);
2511 shader_glsl_add_src_param(ins
, &ins
->src
[1], src_mask
, &src1_param
);
2512 shader_addline(ins
->ctx
->buffer
, "cross(%s, %s)%s);\n", src0_param
.param_str
, src1_param
.param_str
, dst_mask
);
2515 static void shader_glsl_cut(const struct wined3d_shader_instruction
*ins
)
2517 shader_addline(ins
->ctx
->buffer
, "EndPrimitive();\n");
2520 /* Process the WINED3DSIO_POW instruction in GLSL (dst = |src0|^src1)
2521 * Src0 and src1 are scalars. Note that D3D uses the absolute of src0, while
2522 * GLSL uses the value as-is. */
2523 static void shader_glsl_pow(const struct wined3d_shader_instruction
*ins
)
2525 struct wined3d_shader_buffer
*buffer
= ins
->ctx
->buffer
;
2526 struct glsl_src_param src0_param
;
2527 struct glsl_src_param src1_param
;
2528 DWORD dst_write_mask
;
2529 unsigned int dst_size
;
2531 dst_write_mask
= shader_glsl_append_dst(buffer
, ins
);
2532 dst_size
= shader_glsl_get_write_mask_size(dst_write_mask
);
2534 shader_glsl_add_src_param(ins
, &ins
->src
[0], WINED3DSP_WRITEMASK_0
, &src0_param
);
2535 shader_glsl_add_src_param(ins
, &ins
->src
[1], WINED3DSP_WRITEMASK_0
, &src1_param
);
2539 shader_addline(buffer
, "vec%u(%s == 0.0 ? 1.0 : pow(abs(%s), %s)));\n",
2540 dst_size
, src1_param
.param_str
, src0_param
.param_str
, src1_param
.param_str
);
2544 shader_addline(buffer
, "%s == 0.0 ? 1.0 : pow(abs(%s), %s));\n",
2545 src1_param
.param_str
, src0_param
.param_str
, src1_param
.param_str
);
2549 /* Process the WINED3DSIO_LOG instruction in GLSL (dst = log2(|src0|))
2550 * Src0 is a scalar. Note that D3D uses the absolute of src0, while
2551 * GLSL uses the value as-is. */
2552 static void shader_glsl_log(const struct wined3d_shader_instruction
*ins
)
2554 struct wined3d_shader_buffer
*buffer
= ins
->ctx
->buffer
;
2555 struct glsl_src_param src0_param
;
2556 DWORD dst_write_mask
;
2557 unsigned int dst_size
;
2559 dst_write_mask
= shader_glsl_append_dst(buffer
, ins
);
2560 dst_size
= shader_glsl_get_write_mask_size(dst_write_mask
);
2562 shader_glsl_add_src_param(ins
, &ins
->src
[0], WINED3DSP_WRITEMASK_0
, &src0_param
);
2566 shader_addline(buffer
, "vec%u(log2(abs(%s))));\n",
2567 dst_size
, src0_param
.param_str
);
2571 shader_addline(buffer
, "log2(abs(%s)));\n",
2572 src0_param
.param_str
);
2576 /* Map the opcode 1-to-1 to the GL code (arg->dst = instruction(src0, src1, ...) */
2577 static void shader_glsl_map2gl(const struct wined3d_shader_instruction
*ins
)
2579 struct wined3d_shader_buffer
*buffer
= ins
->ctx
->buffer
;
2580 struct glsl_src_param src_param
;
2581 const char *instruction
;
2585 /* Determine the GLSL function to use based on the opcode */
2586 /* TODO: Possibly make this a table for faster lookups */
2587 switch (ins
->handler_idx
)
2589 case WINED3DSIH_MIN
: instruction
= "min"; break;
2590 case WINED3DSIH_MAX
: instruction
= "max"; break;
2591 case WINED3DSIH_ABS
: instruction
= "abs"; break;
2592 case WINED3DSIH_FRC
: instruction
= "fract"; break;
2593 case WINED3DSIH_EXP
: instruction
= "exp2"; break;
2594 case WINED3DSIH_DSX
: instruction
= "dFdx"; break;
2595 case WINED3DSIH_DSY
: instruction
= "ycorrection.y * dFdy"; break;
2596 case WINED3DSIH_ROUND_NI
: instruction
= "floor"; break;
2597 default: instruction
= "";
2598 FIXME("Opcode %#x not yet handled in GLSL\n", ins
->handler_idx
);
2602 write_mask
= shader_glsl_append_dst(buffer
, ins
);
2604 shader_addline(buffer
, "%s(", instruction
);
2608 shader_glsl_add_src_param(ins
, &ins
->src
[0], write_mask
, &src_param
);
2609 shader_addline(buffer
, "%s", src_param
.param_str
);
2610 for (i
= 1; i
< ins
->src_count
; ++i
)
2612 shader_glsl_add_src_param(ins
, &ins
->src
[i
], write_mask
, &src_param
);
2613 shader_addline(buffer
, ", %s", src_param
.param_str
);
2617 shader_addline(buffer
, "));\n");
2620 static void shader_glsl_nop(const struct wined3d_shader_instruction
*ins
) {}
2622 static void shader_glsl_nrm(const struct wined3d_shader_instruction
*ins
)
2624 struct wined3d_shader_buffer
*buffer
= ins
->ctx
->buffer
;
2625 struct glsl_src_param src_param
;
2626 unsigned int mask_size
;
2630 write_mask
= shader_glsl_get_write_mask(ins
->dst
, dst_mask
);
2631 mask_size
= shader_glsl_get_write_mask_size(write_mask
);
2632 shader_glsl_add_src_param(ins
, &ins
->src
[0], write_mask
, &src_param
);
2634 shader_addline(buffer
, "tmp0.x = dot(%s, %s);\n",
2635 src_param
.param_str
, src_param
.param_str
);
2636 shader_glsl_append_dst(buffer
, ins
);
2640 shader_addline(buffer
, "tmp0.x == 0.0 ? vec%u(0.0) : (%s * inversesqrt(tmp0.x)));\n",
2641 mask_size
, src_param
.param_str
);
2645 shader_addline(buffer
, "tmp0.x == 0.0 ? 0.0 : (%s * inversesqrt(tmp0.x)));\n",
2646 src_param
.param_str
);
2650 /** Process the WINED3DSIO_EXPP instruction in GLSL:
2651 * For shader model 1.x, do the following (and honor the writemask, so use a temporary variable):
2652 * dst.x = 2^(floor(src))
2653 * dst.y = src - floor(src)
2654 * dst.z = 2^src (partial precision is allowed, but optional)
2656 * For 2.0 shaders, just do this (honoring writemask and swizzle):
2657 * dst = 2^src; (partial precision is allowed, but optional)
2659 static void shader_glsl_expp(const struct wined3d_shader_instruction
*ins
)
2661 struct glsl_src_param src_param
;
2663 shader_glsl_add_src_param(ins
, &ins
->src
[0], WINED3DSP_WRITEMASK_0
, &src_param
);
2665 if (ins
->ctx
->reg_maps
->shader_version
.major
< 2)
2669 shader_addline(ins
->ctx
->buffer
, "tmp0.x = exp2(floor(%s));\n", src_param
.param_str
);
2670 shader_addline(ins
->ctx
->buffer
, "tmp0.y = %s - floor(%s);\n", src_param
.param_str
, src_param
.param_str
);
2671 shader_addline(ins
->ctx
->buffer
, "tmp0.z = exp2(%s);\n", src_param
.param_str
);
2672 shader_addline(ins
->ctx
->buffer
, "tmp0.w = 1.0;\n");
2674 shader_glsl_append_dst(ins
->ctx
->buffer
, ins
);
2675 shader_glsl_get_write_mask(&ins
->dst
[0], dst_mask
);
2676 shader_addline(ins
->ctx
->buffer
, "tmp0%s);\n", dst_mask
);
2679 unsigned int mask_size
;
2681 write_mask
= shader_glsl_append_dst(ins
->ctx
->buffer
, ins
);
2682 mask_size
= shader_glsl_get_write_mask_size(write_mask
);
2684 if (mask_size
> 1) {
2685 shader_addline(ins
->ctx
->buffer
, "vec%d(exp2(%s)));\n", mask_size
, src_param
.param_str
);
2687 shader_addline(ins
->ctx
->buffer
, "exp2(%s));\n", src_param
.param_str
);
2692 static void shader_glsl_to_int(const struct wined3d_shader_instruction
*ins
)
2694 struct wined3d_shader_buffer
*buffer
= ins
->ctx
->buffer
;
2695 struct glsl_src_param src_param
;
2696 unsigned int mask_size
;
2699 write_mask
= shader_glsl_append_dst(buffer
, ins
);
2700 mask_size
= shader_glsl_get_write_mask_size(write_mask
);
2701 shader_glsl_add_src_param(ins
, &ins
->src
[0], write_mask
, &src_param
);
2704 shader_addline(buffer
, "ivec%u(%s));\n", mask_size
, src_param
.param_str
);
2706 shader_addline(buffer
, "int(%s));\n", src_param
.param_str
);
2709 static void shader_glsl_to_float(const struct wined3d_shader_instruction
*ins
)
2711 struct wined3d_shader_buffer
*buffer
= ins
->ctx
->buffer
;
2712 struct glsl_src_param src_param
;
2713 unsigned int mask_size
;
2716 write_mask
= shader_glsl_append_dst(buffer
, ins
);
2717 mask_size
= shader_glsl_get_write_mask_size(write_mask
);
2718 shader_glsl_add_src_param(ins
, &ins
->src
[0], write_mask
, &src_param
);
2721 shader_addline(buffer
, "vec%u(%s));\n", mask_size
, src_param
.param_str
);
2723 shader_addline(buffer
, "float(%s));\n", src_param
.param_str
);
2726 /** Process the RCP (reciprocal or inverse) opcode in GLSL (dst = 1 / src) */
2727 static void shader_glsl_rcp(const struct wined3d_shader_instruction
*ins
)
2729 struct glsl_src_param src_param
;
2731 unsigned int mask_size
;
2733 write_mask
= shader_glsl_append_dst(ins
->ctx
->buffer
, ins
);
2734 mask_size
= shader_glsl_get_write_mask_size(write_mask
);
2735 shader_glsl_add_src_param(ins
, &ins
->src
[0], WINED3DSP_WRITEMASK_3
, &src_param
);
2739 shader_addline(ins
->ctx
->buffer
, "vec%u(1.0 / %s));\n",
2740 mask_size
, src_param
.param_str
);
2744 shader_addline(ins
->ctx
->buffer
, "1.0 / %s);\n",
2745 src_param
.param_str
);
2749 static void shader_glsl_rsq(const struct wined3d_shader_instruction
*ins
)
2751 struct wined3d_shader_buffer
*buffer
= ins
->ctx
->buffer
;
2752 struct glsl_src_param src_param
;
2754 unsigned int mask_size
;
2756 write_mask
= shader_glsl_append_dst(buffer
, ins
);
2757 mask_size
= shader_glsl_get_write_mask_size(write_mask
);
2759 shader_glsl_add_src_param(ins
, &ins
->src
[0], WINED3DSP_WRITEMASK_3
, &src_param
);
2763 shader_addline(buffer
, "vec%u(inversesqrt(abs(%s))));\n",
2764 mask_size
, src_param
.param_str
);
2768 shader_addline(buffer
, "inversesqrt(abs(%s)));\n",
2769 src_param
.param_str
);
2773 /** Process signed comparison opcodes in GLSL. */
2774 static void shader_glsl_compare(const struct wined3d_shader_instruction
*ins
)
2776 struct glsl_src_param src0_param
;
2777 struct glsl_src_param src1_param
;
2779 unsigned int mask_size
;
2781 write_mask
= shader_glsl_append_dst(ins
->ctx
->buffer
, ins
);
2782 mask_size
= shader_glsl_get_write_mask_size(write_mask
);
2783 shader_glsl_add_src_param(ins
, &ins
->src
[0], write_mask
, &src0_param
);
2784 shader_glsl_add_src_param(ins
, &ins
->src
[1], write_mask
, &src1_param
);
2786 if (mask_size
> 1) {
2787 const char *compare
;
2789 switch(ins
->handler_idx
)
2791 case WINED3DSIH_SLT
: compare
= "lessThan"; break;
2792 case WINED3DSIH_SGE
: compare
= "greaterThanEqual"; break;
2793 default: compare
= "";
2794 FIXME("Can't handle opcode %#x\n", ins
->handler_idx
);
2797 shader_addline(ins
->ctx
->buffer
, "vec%d(%s(%s, %s)));\n", mask_size
, compare
,
2798 src0_param
.param_str
, src1_param
.param_str
);
2800 switch(ins
->handler_idx
)
2802 case WINED3DSIH_SLT
:
2803 /* Step(src0, src1) is not suitable here because if src0 == src1 SLT is supposed,
2804 * to return 0.0 but step returns 1.0 because step is not < x
2805 * An alternative is a bvec compare padded with an unused second component.
2806 * step(src1 * -1.0, src0 * -1.0) is not an option because it suffers from the same
2807 * issue. Playing with not() is not possible either because not() does not accept
2810 shader_addline(ins
->ctx
->buffer
, "(%s < %s) ? 1.0 : 0.0);\n",
2811 src0_param
.param_str
, src1_param
.param_str
);
2813 case WINED3DSIH_SGE
:
2814 /* Here we can use the step() function and safe a conditional */
2815 shader_addline(ins
->ctx
->buffer
, "step(%s, %s));\n", src1_param
.param_str
, src0_param
.param_str
);
2818 FIXME("Can't handle opcode %#x\n", ins
->handler_idx
);
2824 static void shader_glsl_conditional_move(const struct wined3d_shader_instruction
*ins
)
2826 const char *condition_prefix
, *condition_suffix
;
2827 struct wined3d_shader_dst_param dst
;
2828 struct glsl_src_param src0_param
;
2829 struct glsl_src_param src1_param
;
2830 struct glsl_src_param src2_param
;
2831 BOOL temp_destination
= FALSE
;
2832 DWORD cmp_channel
= 0;
2837 switch (ins
->handler_idx
)
2839 case WINED3DSIH_CMP
:
2840 condition_prefix
= "";
2841 condition_suffix
= " >= 0.0";
2844 case WINED3DSIH_CND
:
2845 condition_prefix
= "";
2846 condition_suffix
= " > 0.5";
2849 case WINED3DSIH_MOVC
:
2850 condition_prefix
= "bool(";
2851 condition_suffix
= ")";
2855 FIXME("Unhandled instruction %#x.\n", ins
->handler_idx
);
2856 condition_prefix
= "<unhandled prefix>";
2857 condition_suffix
= "<unhandled suffix>";
2861 if (shader_is_scalar(&ins
->dst
[0].reg
) || shader_is_scalar(&ins
->src
[0].reg
))
2863 write_mask
= shader_glsl_append_dst(ins
->ctx
->buffer
, ins
);
2864 shader_glsl_add_src_param(ins
, &ins
->src
[0], write_mask
, &src0_param
);
2865 shader_glsl_add_src_param(ins
, &ins
->src
[1], write_mask
, &src1_param
);
2866 shader_glsl_add_src_param(ins
, &ins
->src
[2], write_mask
, &src2_param
);
2868 shader_addline(ins
->ctx
->buffer
, "%s%s%s ? %s : %s);\n",
2869 condition_prefix
, src0_param
.param_str
, condition_suffix
,
2870 src1_param
.param_str
, src2_param
.param_str
);
2876 /* Splitting the instruction up in multiple lines imposes a problem:
2877 * The first lines may overwrite source parameters of the following lines.
2878 * Deal with that by using a temporary destination register if needed. */
2879 if ((ins
->src
[0].reg
.idx
[0].offset
== dst
.reg
.idx
[0].offset
2880 && ins
->src
[0].reg
.type
== dst
.reg
.type
)
2881 || (ins
->src
[1].reg
.idx
[0].offset
== dst
.reg
.idx
[0].offset
2882 && ins
->src
[1].reg
.type
== dst
.reg
.type
)
2883 || (ins
->src
[2].reg
.idx
[0].offset
== dst
.reg
.idx
[0].offset
2884 && ins
->src
[2].reg
.type
== dst
.reg
.type
))
2885 temp_destination
= TRUE
;
2887 /* Cycle through all source0 channels. */
2888 for (i
= 0; i
< 4; ++i
)
2891 /* Find the destination channels which use the current source0 channel. */
2892 for (j
= 0; j
< 4; ++j
)
2894 if (((ins
->src
[0].swizzle
>> (2 * j
)) & 0x3) == i
)
2896 write_mask
|= WINED3DSP_WRITEMASK_0
<< j
;
2897 cmp_channel
= WINED3DSP_WRITEMASK_0
<< j
;
2900 dst
.write_mask
= ins
->dst
[0].write_mask
& write_mask
;
2902 if (temp_destination
)
2904 if (!(write_mask
= shader_glsl_get_write_mask(&dst
, mask_char
)))
2906 shader_addline(ins
->ctx
->buffer
, "tmp0%s = (", mask_char
);
2908 else if (!(write_mask
= shader_glsl_append_dst_ext(ins
->ctx
->buffer
, ins
, &dst
)))
2911 shader_glsl_add_src_param(ins
, &ins
->src
[0], cmp_channel
, &src0_param
);
2912 shader_glsl_add_src_param(ins
, &ins
->src
[1], write_mask
, &src1_param
);
2913 shader_glsl_add_src_param(ins
, &ins
->src
[2], write_mask
, &src2_param
);
2915 shader_addline(ins
->ctx
->buffer
, "%s%s%s ? %s : %s);\n",
2916 condition_prefix
, src0_param
.param_str
, condition_suffix
,
2917 src1_param
.param_str
, src2_param
.param_str
);
2920 if (temp_destination
)
2922 shader_glsl_get_write_mask(&ins
->dst
[0], mask_char
);
2923 shader_glsl_append_dst(ins
->ctx
->buffer
, ins
);
2924 shader_addline(ins
->ctx
->buffer
, "tmp0%s);\n", mask_char
);
2928 /** Process the CND opcode in GLSL (dst = (src0 > 0.5) ? src1 : src2) */
2929 /* For ps 1.1-1.3, only a single component of src0 is used. For ps 1.4
2930 * the compare is done per component of src0. */
2931 static void shader_glsl_cnd(const struct wined3d_shader_instruction
*ins
)
2933 struct glsl_src_param src0_param
;
2934 struct glsl_src_param src1_param
;
2935 struct glsl_src_param src2_param
;
2937 DWORD shader_version
= WINED3D_SHADER_VERSION(ins
->ctx
->reg_maps
->shader_version
.major
,
2938 ins
->ctx
->reg_maps
->shader_version
.minor
);
2940 if (shader_version
< WINED3D_SHADER_VERSION(1, 4))
2942 write_mask
= shader_glsl_append_dst(ins
->ctx
->buffer
, ins
);
2943 shader_glsl_add_src_param(ins
, &ins
->src
[0], WINED3DSP_WRITEMASK_0
, &src0_param
);
2944 shader_glsl_add_src_param(ins
, &ins
->src
[1], write_mask
, &src1_param
);
2945 shader_glsl_add_src_param(ins
, &ins
->src
[2], write_mask
, &src2_param
);
2947 /* Fun: The D3DSI_COISSUE flag changes the semantic of the cnd instruction for < 1.4 shaders */
2950 shader_addline(ins
->ctx
->buffer
, "%s /* COISSUE! */);\n", src1_param
.param_str
);
2952 shader_addline(ins
->ctx
->buffer
, "%s > 0.5 ? %s : %s);\n",
2953 src0_param
.param_str
, src1_param
.param_str
, src2_param
.param_str
);
2958 shader_glsl_conditional_move(ins
);
2961 /** GLSL code generation for WINED3DSIO_MAD: Multiply the first 2 opcodes, then add the last */
2962 static void shader_glsl_mad(const struct wined3d_shader_instruction
*ins
)
2964 struct glsl_src_param src0_param
;
2965 struct glsl_src_param src1_param
;
2966 struct glsl_src_param src2_param
;
2969 write_mask
= shader_glsl_append_dst(ins
->ctx
->buffer
, ins
);
2970 shader_glsl_add_src_param(ins
, &ins
->src
[0], write_mask
, &src0_param
);
2971 shader_glsl_add_src_param(ins
, &ins
->src
[1], write_mask
, &src1_param
);
2972 shader_glsl_add_src_param(ins
, &ins
->src
[2], write_mask
, &src2_param
);
2973 shader_addline(ins
->ctx
->buffer
, "(%s * %s) + %s);\n",
2974 src0_param
.param_str
, src1_param
.param_str
, src2_param
.param_str
);
2977 /* Handles transforming all WINED3DSIO_M?x? opcodes for
2978 Vertex shaders to GLSL codes */
2979 static void shader_glsl_mnxn(const struct wined3d_shader_instruction
*ins
)
2982 int nComponents
= 0;
2983 struct wined3d_shader_dst_param tmp_dst
= {{0}};
2984 struct wined3d_shader_src_param tmp_src
[2] = {{{0}}};
2985 struct wined3d_shader_instruction tmp_ins
;
2987 memset(&tmp_ins
, 0, sizeof(tmp_ins
));
2989 /* Set constants for the temporary argument */
2990 tmp_ins
.ctx
= ins
->ctx
;
2991 tmp_ins
.dst_count
= 1;
2992 tmp_ins
.dst
= &tmp_dst
;
2993 tmp_ins
.src_count
= 2;
2994 tmp_ins
.src
= tmp_src
;
2996 switch(ins
->handler_idx
)
2998 case WINED3DSIH_M4x4
:
3000 tmp_ins
.handler_idx
= WINED3DSIH_DP4
;
3002 case WINED3DSIH_M4x3
:
3004 tmp_ins
.handler_idx
= WINED3DSIH_DP4
;
3006 case WINED3DSIH_M3x4
:
3008 tmp_ins
.handler_idx
= WINED3DSIH_DP3
;
3010 case WINED3DSIH_M3x3
:
3012 tmp_ins
.handler_idx
= WINED3DSIH_DP3
;
3014 case WINED3DSIH_M3x2
:
3016 tmp_ins
.handler_idx
= WINED3DSIH_DP3
;
3022 tmp_dst
= ins
->dst
[0];
3023 tmp_src
[0] = ins
->src
[0];
3024 tmp_src
[1] = ins
->src
[1];
3025 for (i
= 0; i
< nComponents
; ++i
)
3027 tmp_dst
.write_mask
= WINED3DSP_WRITEMASK_0
<< i
;
3028 shader_glsl_dot(&tmp_ins
);
3029 ++tmp_src
[1].reg
.idx
[0].offset
;
3034 The LRP instruction performs a component-wise linear interpolation
3035 between the second and third operands using the first operand as the
3036 blend factor. Equation: (dst = src2 + src0 * (src1 - src2))
3037 This is equivalent to mix(src2, src1, src0);
3039 static void shader_glsl_lrp(const struct wined3d_shader_instruction
*ins
)
3041 struct glsl_src_param src0_param
;
3042 struct glsl_src_param src1_param
;
3043 struct glsl_src_param src2_param
;
3046 write_mask
= shader_glsl_append_dst(ins
->ctx
->buffer
, ins
);
3048 shader_glsl_add_src_param(ins
, &ins
->src
[0], write_mask
, &src0_param
);
3049 shader_glsl_add_src_param(ins
, &ins
->src
[1], write_mask
, &src1_param
);
3050 shader_glsl_add_src_param(ins
, &ins
->src
[2], write_mask
, &src2_param
);
3052 shader_addline(ins
->ctx
->buffer
, "mix(%s, %s, %s));\n",
3053 src2_param
.param_str
, src1_param
.param_str
, src0_param
.param_str
);
3056 /** Process the WINED3DSIO_LIT instruction in GLSL:
3057 * dst.x = dst.w = 1.0
3058 * dst.y = (src0.x > 0) ? src0.x
3059 * dst.z = (src0.x > 0) ? ((src0.y > 0) ? pow(src0.y, src.w) : 0) : 0
3060 * where src.w is clamped at +- 128
3062 static void shader_glsl_lit(const struct wined3d_shader_instruction
*ins
)
3064 struct glsl_src_param src0_param
;
3065 struct glsl_src_param src1_param
;
3066 struct glsl_src_param src3_param
;
3069 shader_glsl_append_dst(ins
->ctx
->buffer
, ins
);
3070 shader_glsl_get_write_mask(&ins
->dst
[0], dst_mask
);
3072 shader_glsl_add_src_param(ins
, &ins
->src
[0], WINED3DSP_WRITEMASK_0
, &src0_param
);
3073 shader_glsl_add_src_param(ins
, &ins
->src
[0], WINED3DSP_WRITEMASK_1
, &src1_param
);
3074 shader_glsl_add_src_param(ins
, &ins
->src
[0], WINED3DSP_WRITEMASK_3
, &src3_param
);
3076 /* The sdk specifies the instruction like this
3078 * if(src.x > 0.0) dst.y = src.x
3080 * if(src.x > 0.0 && src.y > 0.0) dst.z = pow(src.y, power);
3083 * (where power = src.w clamped between -128 and 128)
3085 * Obviously that has quite a few conditionals in it which we don't like. So the first step is this:
3086 * dst.x = 1.0 ... No further explanation needed
3087 * dst.y = max(src.y, 0.0); ... If x < 0.0, use 0.0, otherwise x. Same as the conditional
3088 * dst.z = x > 0.0 ? pow(max(y, 0.0), p) : 0; ... 0 ^ power is 0, and otherwise we use y anyway
3089 * dst.w = 1.0. ... Nothing fancy.
3091 * So we still have one conditional in there. So do this:
3092 * dst.z = pow(max(0.0, src.y) * step(0.0, src.x), power);
3094 * step(0.0, x) will return 1 if src.x > 0.0, and 0 otherwise. So if y is 0 we get pow(0.0 * 1.0, power),
3095 * which sets dst.z to 0. If y > 0, but x = 0.0, we get pow(y * 0.0, power), which results in 0 too.
3096 * if both x and y are > 0, we get pow(y * 1.0, power), as it is supposed to.
3098 * Unfortunately pow(0.0 ^ 0.0) returns NaN on most GPUs, but lit with src.y = 0 and src.w = 0 returns
3099 * a non-NaN value in dst.z. What we return doesn't matter, as long as it is not NaN. Return 0, which is
3100 * what all Windows HW drivers and GL_ARB_vertex_program's LIT do.
3102 shader_addline(ins
->ctx
->buffer
,
3103 "vec4(1.0, max(%s, 0.0), %s == 0.0 ? 0.0 : "
3104 "pow(max(0.0, %s) * step(0.0, %s), clamp(%s, -128.0, 128.0)), 1.0)%s);\n",
3105 src0_param
.param_str
, src3_param
.param_str
, src1_param
.param_str
,
3106 src0_param
.param_str
, src3_param
.param_str
, dst_mask
);
3109 /** Process the WINED3DSIO_DST instruction in GLSL:
3111 * dst.y = src0.x * src0.y
3115 static void shader_glsl_dst(const struct wined3d_shader_instruction
*ins
)
3117 struct glsl_src_param src0y_param
;
3118 struct glsl_src_param src0z_param
;
3119 struct glsl_src_param src1y_param
;
3120 struct glsl_src_param src1w_param
;
3123 shader_glsl_append_dst(ins
->ctx
->buffer
, ins
);
3124 shader_glsl_get_write_mask(&ins
->dst
[0], dst_mask
);
3126 shader_glsl_add_src_param(ins
, &ins
->src
[0], WINED3DSP_WRITEMASK_1
, &src0y_param
);
3127 shader_glsl_add_src_param(ins
, &ins
->src
[0], WINED3DSP_WRITEMASK_2
, &src0z_param
);
3128 shader_glsl_add_src_param(ins
, &ins
->src
[1], WINED3DSP_WRITEMASK_1
, &src1y_param
);
3129 shader_glsl_add_src_param(ins
, &ins
->src
[1], WINED3DSP_WRITEMASK_3
, &src1w_param
);
3131 shader_addline(ins
->ctx
->buffer
, "vec4(1.0, %s * %s, %s, %s))%s;\n",
3132 src0y_param
.param_str
, src1y_param
.param_str
, src0z_param
.param_str
, src1w_param
.param_str
, dst_mask
);
3135 /** Process the WINED3DSIO_SINCOS instruction in GLSL:
3136 * VS 2.0 requires that specific cosine and sine constants be passed to this instruction so the hardware
3137 * can handle it. But, these functions are built-in for GLSL, so we can just ignore the last 2 params.
3139 * dst.x = cos(src0.?)
3140 * dst.y = sin(src0.?)
3144 static void shader_glsl_sincos(const struct wined3d_shader_instruction
*ins
)
3146 struct wined3d_shader_buffer
*buffer
= ins
->ctx
->buffer
;
3147 struct glsl_src_param src0_param
;
3150 if (ins
->ctx
->reg_maps
->shader_version
.major
< 4)
3152 shader_glsl_add_src_param(ins
, &ins
->src
[0], WINED3DSP_WRITEMASK_0
, &src0_param
);
3154 write_mask
= shader_glsl_append_dst(buffer
, ins
);
3157 case WINED3DSP_WRITEMASK_0
:
3158 shader_addline(buffer
, "cos(%s));\n", src0_param
.param_str
);
3161 case WINED3DSP_WRITEMASK_1
:
3162 shader_addline(buffer
, "sin(%s));\n", src0_param
.param_str
);
3165 case (WINED3DSP_WRITEMASK_0
| WINED3DSP_WRITEMASK_1
):
3166 shader_addline(buffer
, "vec2(cos(%s), sin(%s)));\n",
3167 src0_param
.param_str
, src0_param
.param_str
);
3171 ERR("Write mask should be .x, .y or .xy\n");
3178 if (ins
->dst
[0].reg
.type
!= WINED3DSPR_NULL
)
3181 if (ins
->dst
[1].reg
.type
!= WINED3DSPR_NULL
)
3185 write_mask
= shader_glsl_get_write_mask(&ins
->dst
[0], dst_mask
);
3186 shader_glsl_add_src_param(ins
, &ins
->src
[0], write_mask
, &src0_param
);
3187 shader_addline(buffer
, "tmp0%s = sin(%s);\n", dst_mask
, src0_param
.param_str
);
3189 write_mask
= shader_glsl_append_dst_ext(buffer
, ins
, &ins
->dst
[1]);
3190 shader_glsl_add_src_param(ins
, &ins
->src
[0], write_mask
, &src0_param
);
3191 shader_addline(buffer
, "cos(%s));\n", src0_param
.param_str
);
3193 shader_glsl_append_dst_ext(buffer
, ins
, &ins
->dst
[0]);
3194 shader_addline(buffer
, "tmp0%s);\n", dst_mask
);
3198 write_mask
= shader_glsl_append_dst_ext(buffer
, ins
, &ins
->dst
[0]);
3199 shader_glsl_add_src_param(ins
, &ins
->src
[0], write_mask
, &src0_param
);
3200 shader_addline(buffer
, "sin(%s));\n", src0_param
.param_str
);
3203 else if (ins
->dst
[1].reg
.type
!= WINED3DSPR_NULL
)
3205 write_mask
= shader_glsl_append_dst_ext(buffer
, ins
, &ins
->dst
[1]);
3206 shader_glsl_add_src_param(ins
, &ins
->src
[0], write_mask
, &src0_param
);
3207 shader_addline(buffer
, "cos(%s));\n", src0_param
.param_str
);
3211 /* sgn in vs_2_0 has 2 extra parameters(registers for temporary storage) which we don't use
3212 * here. But those extra parameters require a dedicated function for sgn, since map2gl would
3213 * generate invalid code
3215 static void shader_glsl_sgn(const struct wined3d_shader_instruction
*ins
)
3217 struct glsl_src_param src0_param
;
3220 write_mask
= shader_glsl_append_dst(ins
->ctx
->buffer
, ins
);
3221 shader_glsl_add_src_param(ins
, &ins
->src
[0], write_mask
, &src0_param
);
3223 shader_addline(ins
->ctx
->buffer
, "sign(%s));\n", src0_param
.param_str
);
3226 /** Process the WINED3DSIO_LOOP instruction in GLSL:
3227 * Start a for() loop where src1.y is the initial value of aL,
3228 * increment aL by src1.z for a total of src1.x iterations.
3229 * Need to use a temporary variable for this operation.
3231 /* FIXME: I don't think nested loops will work correctly this way. */
3232 static void shader_glsl_loop(const struct wined3d_shader_instruction
*ins
)
3234 struct wined3d_shader_loop_state
*loop_state
= ins
->ctx
->loop_state
;
3235 struct wined3d_shader_buffer
*buffer
= ins
->ctx
->buffer
;
3236 const struct wined3d_shader
*shader
= ins
->ctx
->shader
;
3237 const struct wined3d_shader_lconst
*constant
;
3238 struct glsl_src_param src1_param
;
3239 const DWORD
*control_values
= NULL
;
3241 if (ins
->ctx
->reg_maps
->shader_version
.major
< 4)
3243 shader_glsl_add_src_param(ins
, &ins
->src
[1], WINED3DSP_WRITEMASK_ALL
, &src1_param
);
3245 /* Try to hardcode the loop control parameters if possible. Direct3D 9
3246 * class hardware doesn't support real varying indexing, but Microsoft
3247 * designed this feature for Shader model 2.x+. If the loop control is
3248 * known at compile time, the GLSL compiler can unroll the loop, and
3249 * replace indirect addressing with direct addressing. */
3250 if (ins
->src
[1].reg
.type
== WINED3DSPR_CONSTINT
)
3252 LIST_FOR_EACH_ENTRY(constant
, &shader
->constantsI
, struct wined3d_shader_lconst
, entry
)
3254 if (constant
->idx
== ins
->src
[1].reg
.idx
[0].offset
)
3256 control_values
= constant
->value
;
3264 struct wined3d_shader_loop_control loop_control
;
3265 loop_control
.count
= control_values
[0];
3266 loop_control
.start
= control_values
[1];
3267 loop_control
.step
= (int)control_values
[2];
3269 if (loop_control
.step
> 0)
3271 shader_addline(buffer
, "for (aL%u = %u; aL%u < (%u * %d + %u); aL%u += %d)\n{\n",
3272 loop_state
->current_depth
, loop_control
.start
,
3273 loop_state
->current_depth
, loop_control
.count
, loop_control
.step
, loop_control
.start
,
3274 loop_state
->current_depth
, loop_control
.step
);
3276 else if (loop_control
.step
< 0)
3278 shader_addline(buffer
, "for (aL%u = %u; aL%u > (%u * %d + %u); aL%u += %d)\n{\n",
3279 loop_state
->current_depth
, loop_control
.start
,
3280 loop_state
->current_depth
, loop_control
.count
, loop_control
.step
, loop_control
.start
,
3281 loop_state
->current_depth
, loop_control
.step
);
3285 shader_addline(buffer
, "for (aL%u = %u, tmpInt%u = 0; tmpInt%u < %u; tmpInt%u++)\n{\n",
3286 loop_state
->current_depth
, loop_control
.start
, loop_state
->current_depth
,
3287 loop_state
->current_depth
, loop_control
.count
,
3288 loop_state
->current_depth
);
3293 shader_addline(buffer
, "for (tmpInt%u = 0, aL%u = %s.y; tmpInt%u < %s.x; tmpInt%u++, aL%u += %s.z)\n{\n",
3294 loop_state
->current_depth
, loop_state
->current_reg
,
3295 src1_param
.reg_name
, loop_state
->current_depth
, src1_param
.reg_name
,
3296 loop_state
->current_depth
, loop_state
->current_reg
, src1_param
.reg_name
);
3299 ++loop_state
->current_reg
;
3303 shader_addline(buffer
, "for (;;)\n{\n");
3306 ++loop_state
->current_depth
;
3309 static void shader_glsl_end(const struct wined3d_shader_instruction
*ins
)
3311 struct wined3d_shader_loop_state
*loop_state
= ins
->ctx
->loop_state
;
3313 shader_addline(ins
->ctx
->buffer
, "}\n");
3315 if (ins
->handler_idx
== WINED3DSIH_ENDLOOP
)
3317 --loop_state
->current_depth
;
3318 --loop_state
->current_reg
;
3321 if (ins
->handler_idx
== WINED3DSIH_ENDREP
)
3323 --loop_state
->current_depth
;
3327 static void shader_glsl_rep(const struct wined3d_shader_instruction
*ins
)
3329 const struct wined3d_shader
*shader
= ins
->ctx
->shader
;
3330 struct wined3d_shader_loop_state
*loop_state
= ins
->ctx
->loop_state
;
3331 const struct wined3d_shader_lconst
*constant
;
3332 struct glsl_src_param src0_param
;
3333 const DWORD
*control_values
= NULL
;
3335 /* Try to hardcode local values to help the GLSL compiler to unroll and optimize the loop */
3336 if (ins
->src
[0].reg
.type
== WINED3DSPR_CONSTINT
)
3338 LIST_FOR_EACH_ENTRY(constant
, &shader
->constantsI
, struct wined3d_shader_lconst
, entry
)
3340 if (constant
->idx
== ins
->src
[0].reg
.idx
[0].offset
)
3342 control_values
= constant
->value
;
3350 shader_addline(ins
->ctx
->buffer
, "for (tmpInt%d = 0; tmpInt%d < %d; tmpInt%d++) {\n",
3351 loop_state
->current_depth
, loop_state
->current_depth
,
3352 control_values
[0], loop_state
->current_depth
);
3356 shader_glsl_add_src_param(ins
, &ins
->src
[0], WINED3DSP_WRITEMASK_0
, &src0_param
);
3357 shader_addline(ins
->ctx
->buffer
, "for (tmpInt%d = 0; tmpInt%d < %s; tmpInt%d++) {\n",
3358 loop_state
->current_depth
, loop_state
->current_depth
,
3359 src0_param
.param_str
, loop_state
->current_depth
);
3362 ++loop_state
->current_depth
;
3365 static void shader_glsl_if(const struct wined3d_shader_instruction
*ins
)
3367 struct glsl_src_param src0_param
;
3369 shader_glsl_add_src_param(ins
, &ins
->src
[0], WINED3DSP_WRITEMASK_0
, &src0_param
);
3370 shader_addline(ins
->ctx
->buffer
, "if (%s) {\n", src0_param
.param_str
);
3373 static void shader_glsl_ifc(const struct wined3d_shader_instruction
*ins
)
3375 struct glsl_src_param src0_param
;
3376 struct glsl_src_param src1_param
;
3378 shader_glsl_add_src_param(ins
, &ins
->src
[0], WINED3DSP_WRITEMASK_0
, &src0_param
);
3379 shader_glsl_add_src_param(ins
, &ins
->src
[1], WINED3DSP_WRITEMASK_0
, &src1_param
);
3381 shader_addline(ins
->ctx
->buffer
, "if (%s %s %s) {\n",
3382 src0_param
.param_str
, shader_glsl_get_rel_op(ins
->flags
), src1_param
.param_str
);
3385 static void shader_glsl_else(const struct wined3d_shader_instruction
*ins
)
3387 shader_addline(ins
->ctx
->buffer
, "} else {\n");
3390 static void shader_glsl_emit(const struct wined3d_shader_instruction
*ins
)
3392 shader_addline(ins
->ctx
->buffer
, "EmitVertex();\n");
3395 static void shader_glsl_break(const struct wined3d_shader_instruction
*ins
)
3397 shader_addline(ins
->ctx
->buffer
, "break;\n");
3400 /* FIXME: According to MSDN the compare is done per component. */
3401 static void shader_glsl_breakc(const struct wined3d_shader_instruction
*ins
)
3403 struct glsl_src_param src0_param
;
3404 struct glsl_src_param src1_param
;
3406 shader_glsl_add_src_param(ins
, &ins
->src
[0], WINED3DSP_WRITEMASK_0
, &src0_param
);
3407 shader_glsl_add_src_param(ins
, &ins
->src
[1], WINED3DSP_WRITEMASK_0
, &src1_param
);
3409 shader_addline(ins
->ctx
->buffer
, "if (%s %s %s) break;\n",
3410 src0_param
.param_str
, shader_glsl_get_rel_op(ins
->flags
), src1_param
.param_str
);
3413 static void shader_glsl_breakp(const struct wined3d_shader_instruction
*ins
)
3415 struct glsl_src_param src_param
;
3417 shader_glsl_add_src_param(ins
, &ins
->src
[0], WINED3DSP_WRITEMASK_0
, &src_param
);
3418 shader_addline(ins
->ctx
->buffer
, "if (bool(%s)) break;\n", src_param
.param_str
);
3421 static void shader_glsl_label(const struct wined3d_shader_instruction
*ins
)
3423 shader_addline(ins
->ctx
->buffer
, "}\n");
3424 shader_addline(ins
->ctx
->buffer
, "void subroutine%u()\n{\n", ins
->src
[0].reg
.idx
[0].offset
);
3427 static void shader_glsl_call(const struct wined3d_shader_instruction
*ins
)
3429 shader_addline(ins
->ctx
->buffer
, "subroutine%u();\n", ins
->src
[0].reg
.idx
[0].offset
);
3432 static void shader_glsl_callnz(const struct wined3d_shader_instruction
*ins
)
3434 struct glsl_src_param src1_param
;
3436 shader_glsl_add_src_param(ins
, &ins
->src
[1], WINED3DSP_WRITEMASK_0
, &src1_param
);
3437 shader_addline(ins
->ctx
->buffer
, "if (%s) subroutine%u();\n",
3438 src1_param
.param_str
, ins
->src
[0].reg
.idx
[0].offset
);
3441 static void shader_glsl_ret(const struct wined3d_shader_instruction
*ins
)
3443 /* No-op. The closing } is written when a new function is started, and at the end of the shader. This
3444 * function only suppresses the unhandled instruction warning
3448 /*********************************************
3449 * Pixel Shader Specific Code begins here
3450 ********************************************/
3451 static void shader_glsl_tex(const struct wined3d_shader_instruction
*ins
)
3453 DWORD shader_version
= WINED3D_SHADER_VERSION(ins
->ctx
->reg_maps
->shader_version
.major
,
3454 ins
->ctx
->reg_maps
->shader_version
.minor
);
3455 struct glsl_sample_function sample_function
;
3456 DWORD sample_flags
= 0;
3458 DWORD mask
= 0, swizzle
;
3459 const struct shader_glsl_ctx_priv
*priv
= ins
->ctx
->backend_data
;
3461 /* 1.0-1.4: Use destination register as sampler source.
3462 * 2.0+: Use provided sampler source. */
3463 if (shader_version
< WINED3D_SHADER_VERSION(2,0))
3464 sampler_idx
= ins
->dst
[0].reg
.idx
[0].offset
;
3466 sampler_idx
= ins
->src
[1].reg
.idx
[0].offset
;
3468 if (shader_version
< WINED3D_SHADER_VERSION(1,4))
3470 DWORD flags
= (priv
->cur_ps_args
->tex_transform
>> sampler_idx
* WINED3D_PSARGS_TEXTRANSFORM_SHIFT
)
3471 & WINED3D_PSARGS_TEXTRANSFORM_MASK
;
3472 enum wined3d_sampler_texture_type sampler_type
= ins
->ctx
->reg_maps
->sampler_type
[sampler_idx
];
3474 /* Projected cube textures don't make a lot of sense, the resulting coordinates stay the same. */
3475 if (flags
& WINED3D_PSARGS_PROJECTED
&& sampler_type
!= WINED3DSTT_CUBE
)
3477 sample_flags
|= WINED3D_GLSL_SAMPLE_PROJECTED
;
3478 switch (flags
& ~WINED3D_PSARGS_PROJECTED
)
3480 case WINED3D_TTFF_COUNT1
:
3481 FIXME("WINED3D_TTFF_PROJECTED with WINED3D_TTFF_COUNT1?\n");
3483 case WINED3D_TTFF_COUNT2
:
3484 mask
= WINED3DSP_WRITEMASK_1
;
3486 case WINED3D_TTFF_COUNT3
:
3487 mask
= WINED3DSP_WRITEMASK_2
;
3489 case WINED3D_TTFF_COUNT4
:
3490 case WINED3D_TTFF_DISABLE
:
3491 mask
= WINED3DSP_WRITEMASK_3
;
3496 else if (shader_version
< WINED3D_SHADER_VERSION(2,0))
3498 enum wined3d_shader_src_modifier src_mod
= ins
->src
[0].modifiers
;
3500 if (src_mod
== WINED3DSPSM_DZ
) {
3501 sample_flags
|= WINED3D_GLSL_SAMPLE_PROJECTED
;
3502 mask
= WINED3DSP_WRITEMASK_2
;
3503 } else if (src_mod
== WINED3DSPSM_DW
) {
3504 sample_flags
|= WINED3D_GLSL_SAMPLE_PROJECTED
;
3505 mask
= WINED3DSP_WRITEMASK_3
;
3508 if (ins
->flags
& WINED3DSI_TEXLD_PROJECT
)
3510 /* ps 2.0 texldp instruction always divides by the fourth component. */
3511 sample_flags
|= WINED3D_GLSL_SAMPLE_PROJECTED
;
3512 mask
= WINED3DSP_WRITEMASK_3
;
3516 if (priv
->cur_ps_args
->np2_fixup
& (1 << sampler_idx
))
3517 sample_flags
|= WINED3D_GLSL_SAMPLE_NPOT
;
3519 shader_glsl_get_sample_function(ins
->ctx
, sampler_idx
, sample_flags
, &sample_function
);
3520 mask
|= sample_function
.coord_mask
;
3522 if (shader_version
< WINED3D_SHADER_VERSION(2,0)) swizzle
= WINED3DSP_NOSWIZZLE
;
3523 else swizzle
= ins
->src
[1].swizzle
;
3525 /* 1.0-1.3: Use destination register as coordinate source.
3526 1.4+: Use provided coordinate source register. */
3527 if (shader_version
< WINED3D_SHADER_VERSION(1,4))
3530 shader_glsl_write_mask_to_str(mask
, coord_mask
);
3531 shader_glsl_gen_sample_code(ins
, sampler_idx
, &sample_function
, swizzle
, NULL
, NULL
, NULL
,
3532 "T%u%s", sampler_idx
, coord_mask
);
3536 struct glsl_src_param coord_param
;
3537 shader_glsl_add_src_param(ins
, &ins
->src
[0], mask
, &coord_param
);
3538 if (ins
->flags
& WINED3DSI_TEXLD_BIAS
)
3540 struct glsl_src_param bias
;
3541 shader_glsl_add_src_param(ins
, &ins
->src
[0], WINED3DSP_WRITEMASK_3
, &bias
);
3542 shader_glsl_gen_sample_code(ins
, sampler_idx
, &sample_function
, swizzle
, NULL
, NULL
, bias
.param_str
,
3543 "%s", coord_param
.param_str
);
3545 shader_glsl_gen_sample_code(ins
, sampler_idx
, &sample_function
, swizzle
, NULL
, NULL
, NULL
,
3546 "%s", coord_param
.param_str
);
3551 static void shader_glsl_texldd(const struct wined3d_shader_instruction
*ins
)
3553 const struct wined3d_gl_info
*gl_info
= ins
->ctx
->gl_info
;
3554 struct glsl_src_param coord_param
, dx_param
, dy_param
;
3555 DWORD sample_flags
= WINED3D_GLSL_SAMPLE_GRAD
;
3556 struct glsl_sample_function sample_function
;
3558 DWORD swizzle
= ins
->src
[1].swizzle
;
3559 const struct shader_glsl_ctx_priv
*priv
= ins
->ctx
->backend_data
;
3561 if (!gl_info
->supported
[ARB_SHADER_TEXTURE_LOD
] && !gl_info
->supported
[EXT_GPU_SHADER4
])
3563 FIXME("texldd used, but not supported by hardware. Falling back to regular tex\n");
3564 shader_glsl_tex(ins
);
3568 sampler_idx
= ins
->src
[1].reg
.idx
[0].offset
;
3569 if (priv
->cur_ps_args
->np2_fixup
& (1 << sampler_idx
))
3570 sample_flags
|= WINED3D_GLSL_SAMPLE_NPOT
;
3572 shader_glsl_get_sample_function(ins
->ctx
, sampler_idx
, sample_flags
, &sample_function
);
3573 shader_glsl_add_src_param(ins
, &ins
->src
[0], sample_function
.coord_mask
, &coord_param
);
3574 shader_glsl_add_src_param(ins
, &ins
->src
[2], sample_function
.coord_mask
, &dx_param
);
3575 shader_glsl_add_src_param(ins
, &ins
->src
[3], sample_function
.coord_mask
, &dy_param
);
3577 shader_glsl_gen_sample_code(ins
, sampler_idx
, &sample_function
, swizzle
, dx_param
.param_str
, dy_param
.param_str
, NULL
,
3578 "%s", coord_param
.param_str
);
3581 static void shader_glsl_texldl(const struct wined3d_shader_instruction
*ins
)
3583 const struct wined3d_gl_info
*gl_info
= ins
->ctx
->gl_info
;
3584 struct glsl_src_param coord_param
, lod_param
;
3585 DWORD sample_flags
= WINED3D_GLSL_SAMPLE_LOD
;
3586 struct glsl_sample_function sample_function
;
3588 DWORD swizzle
= ins
->src
[1].swizzle
;
3589 const struct shader_glsl_ctx_priv
*priv
= ins
->ctx
->backend_data
;
3591 sampler_idx
= ins
->src
[1].reg
.idx
[0].offset
;
3592 if (ins
->ctx
->reg_maps
->shader_version
.type
== WINED3D_SHADER_TYPE_PIXEL
3593 && priv
->cur_ps_args
->np2_fixup
& (1 << sampler_idx
))
3594 sample_flags
|= WINED3D_GLSL_SAMPLE_NPOT
;
3596 shader_glsl_get_sample_function(ins
->ctx
, sampler_idx
, sample_flags
, &sample_function
);
3597 shader_glsl_add_src_param(ins
, &ins
->src
[0], sample_function
.coord_mask
, &coord_param
);
3599 shader_glsl_add_src_param(ins
, &ins
->src
[0], WINED3DSP_WRITEMASK_3
, &lod_param
);
3601 if (!gl_info
->supported
[ARB_SHADER_TEXTURE_LOD
] && !gl_info
->supported
[EXT_GPU_SHADER4
]
3602 && ins
->ctx
->reg_maps
->shader_version
.type
== WINED3D_SHADER_TYPE_PIXEL
)
3604 /* Plain GLSL only supports Lod sampling functions in vertex shaders.
3605 * However, the NVIDIA drivers allow them in fragment shaders as well,
3606 * even without the appropriate extension. */
3607 WARN("Using %s in fragment shader.\n", sample_function
.name
);
3609 shader_glsl_gen_sample_code(ins
, sampler_idx
, &sample_function
, swizzle
, NULL
, NULL
, lod_param
.param_str
,
3610 "%s", coord_param
.param_str
);
3613 static void shader_glsl_texcoord(const struct wined3d_shader_instruction
*ins
)
3615 /* FIXME: Make this work for more than just 2D textures */
3616 struct wined3d_shader_buffer
*buffer
= ins
->ctx
->buffer
;
3617 DWORD write_mask
= shader_glsl_append_dst(ins
->ctx
->buffer
, ins
);
3619 if (!(ins
->ctx
->reg_maps
->shader_version
.major
== 1 && ins
->ctx
->reg_maps
->shader_version
.minor
== 4))
3623 shader_glsl_get_write_mask(&ins
->dst
[0], dst_mask
);
3624 shader_addline(buffer
, "clamp(gl_TexCoord[%u], 0.0, 1.0)%s);\n",
3625 ins
->dst
[0].reg
.idx
[0].offset
, dst_mask
);
3629 enum wined3d_shader_src_modifier src_mod
= ins
->src
[0].modifiers
;
3630 DWORD reg
= ins
->src
[0].reg
.idx
[0].offset
;
3631 char dst_swizzle
[6];
3633 shader_glsl_get_swizzle(&ins
->src
[0], FALSE
, write_mask
, dst_swizzle
);
3635 if (src_mod
== WINED3DSPSM_DZ
)
3637 unsigned int mask_size
= shader_glsl_get_write_mask_size(write_mask
);
3638 struct glsl_src_param div_param
;
3640 shader_glsl_add_src_param(ins
, &ins
->src
[0], WINED3DSP_WRITEMASK_2
, &div_param
);
3642 if (mask_size
> 1) {
3643 shader_addline(buffer
, "gl_TexCoord[%u]%s / vec%d(%s));\n", reg
, dst_swizzle
, mask_size
, div_param
.param_str
);
3645 shader_addline(buffer
, "gl_TexCoord[%u]%s / %s);\n", reg
, dst_swizzle
, div_param
.param_str
);
3648 else if (src_mod
== WINED3DSPSM_DW
)
3650 unsigned int mask_size
= shader_glsl_get_write_mask_size(write_mask
);
3651 struct glsl_src_param div_param
;
3653 shader_glsl_add_src_param(ins
, &ins
->src
[0], WINED3DSP_WRITEMASK_3
, &div_param
);
3655 if (mask_size
> 1) {
3656 shader_addline(buffer
, "gl_TexCoord[%u]%s / vec%d(%s));\n", reg
, dst_swizzle
, mask_size
, div_param
.param_str
);
3658 shader_addline(buffer
, "gl_TexCoord[%u]%s / %s);\n", reg
, dst_swizzle
, div_param
.param_str
);
3661 shader_addline(buffer
, "gl_TexCoord[%u]%s);\n", reg
, dst_swizzle
);
3666 /** Process the WINED3DSIO_TEXDP3TEX instruction in GLSL:
3667 * Take a 3-component dot product of the TexCoord[dstreg] and src,
3668 * then perform a 1D texture lookup from stage dstregnum, place into dst. */
3669 static void shader_glsl_texdp3tex(const struct wined3d_shader_instruction
*ins
)
3671 DWORD src_mask
= WINED3DSP_WRITEMASK_0
| WINED3DSP_WRITEMASK_1
| WINED3DSP_WRITEMASK_2
;
3672 DWORD sampler_idx
= ins
->dst
[0].reg
.idx
[0].offset
;
3673 struct glsl_sample_function sample_function
;
3674 struct glsl_src_param src0_param
;
3677 shader_glsl_add_src_param(ins
, &ins
->src
[0], src_mask
, &src0_param
);
3679 /* Do I have to take care about the projected bit? I don't think so, since the dp3 returns only one
3680 * scalar, and projected sampling would require 4.
3682 * It is a dependent read - not valid with conditional NP2 textures
3684 shader_glsl_get_sample_function(ins
->ctx
, sampler_idx
, 0, &sample_function
);
3685 mask_size
= shader_glsl_get_write_mask_size(sample_function
.coord_mask
);
3690 shader_glsl_gen_sample_code(ins
, sampler_idx
, &sample_function
, WINED3DSP_NOSWIZZLE
, NULL
, NULL
, NULL
,
3691 "dot(gl_TexCoord[%u].xyz, %s)", sampler_idx
, src0_param
.param_str
);
3695 shader_glsl_gen_sample_code(ins
, sampler_idx
, &sample_function
, WINED3DSP_NOSWIZZLE
, NULL
, NULL
, NULL
,
3696 "vec2(dot(gl_TexCoord[%u].xyz, %s), 0.0)", sampler_idx
, src0_param
.param_str
);
3700 shader_glsl_gen_sample_code(ins
, sampler_idx
, &sample_function
, WINED3DSP_NOSWIZZLE
, NULL
, NULL
, NULL
,
3701 "vec3(dot(gl_TexCoord[%u].xyz, %s), 0.0, 0.0)", sampler_idx
, src0_param
.param_str
);
3705 FIXME("Unexpected mask size %u\n", mask_size
);
3710 /** Process the WINED3DSIO_TEXDP3 instruction in GLSL:
3711 * Take a 3-component dot product of the TexCoord[dstreg] and src. */
3712 static void shader_glsl_texdp3(const struct wined3d_shader_instruction
*ins
)
3714 DWORD src_mask
= WINED3DSP_WRITEMASK_0
| WINED3DSP_WRITEMASK_1
| WINED3DSP_WRITEMASK_2
;
3715 DWORD dstreg
= ins
->dst
[0].reg
.idx
[0].offset
;
3716 struct glsl_src_param src0_param
;
3718 unsigned int mask_size
;
3720 dst_mask
= shader_glsl_append_dst(ins
->ctx
->buffer
, ins
);
3721 mask_size
= shader_glsl_get_write_mask_size(dst_mask
);
3722 shader_glsl_add_src_param(ins
, &ins
->src
[0], src_mask
, &src0_param
);
3724 if (mask_size
> 1) {
3725 shader_addline(ins
->ctx
->buffer
, "vec%d(dot(T%u.xyz, %s)));\n", mask_size
, dstreg
, src0_param
.param_str
);
3727 shader_addline(ins
->ctx
->buffer
, "dot(T%u.xyz, %s));\n", dstreg
, src0_param
.param_str
);
3731 /** Process the WINED3DSIO_TEXDEPTH instruction in GLSL:
3732 * Calculate the depth as dst.x / dst.y */
3733 static void shader_glsl_texdepth(const struct wined3d_shader_instruction
*ins
)
3735 struct glsl_dst_param dst_param
;
3737 shader_glsl_add_dst_param(ins
, &ins
->dst
[0], &dst_param
);
3739 /* Tests show that texdepth never returns anything below 0.0, and that r5.y is clamped to 1.0.
3740 * Negative input is accepted, -0.25 / -0.5 returns 0.5. GL should clamp gl_FragDepth to [0;1], but
3741 * this doesn't always work, so clamp the results manually. Whether or not the x value is clamped at 1
3742 * too is irrelevant, since if x = 0, any y value < 1.0 (and > 1.0 is not allowed) results in a result
3745 shader_addline(ins
->ctx
->buffer
, "gl_FragDepth = clamp((%s.x / min(%s.y, 1.0)), 0.0, 1.0);\n",
3746 dst_param
.reg_name
, dst_param
.reg_name
);
3749 /** Process the WINED3DSIO_TEXM3X2DEPTH instruction in GLSL:
3750 * Last row of a 3x2 matrix multiply, use the result to calculate the depth:
3751 * Calculate tmp0.y = TexCoord[dstreg] . src.xyz; (tmp0.x has already been calculated)
3752 * depth = (tmp0.y == 0.0) ? 1.0 : tmp0.x / tmp0.y
3754 static void shader_glsl_texm3x2depth(const struct wined3d_shader_instruction
*ins
)
3756 DWORD src_mask
= WINED3DSP_WRITEMASK_0
| WINED3DSP_WRITEMASK_1
| WINED3DSP_WRITEMASK_2
;
3757 DWORD dstreg
= ins
->dst
[0].reg
.idx
[0].offset
;
3758 struct glsl_src_param src0_param
;
3760 shader_glsl_add_src_param(ins
, &ins
->src
[0], src_mask
, &src0_param
);
3762 shader_addline(ins
->ctx
->buffer
, "tmp0.y = dot(T%u.xyz, %s);\n", dstreg
, src0_param
.param_str
);
3763 shader_addline(ins
->ctx
->buffer
, "gl_FragDepth = (tmp0.y == 0.0) ? 1.0 : clamp(tmp0.x / tmp0.y, 0.0, 1.0);\n");
3766 /** Process the WINED3DSIO_TEXM3X2PAD instruction in GLSL
3767 * Calculate the 1st of a 2-row matrix multiplication. */
3768 static void shader_glsl_texm3x2pad(const struct wined3d_shader_instruction
*ins
)
3770 DWORD src_mask
= WINED3DSP_WRITEMASK_0
| WINED3DSP_WRITEMASK_1
| WINED3DSP_WRITEMASK_2
;
3771 DWORD reg
= ins
->dst
[0].reg
.idx
[0].offset
;
3772 struct wined3d_shader_buffer
*buffer
= ins
->ctx
->buffer
;
3773 struct glsl_src_param src0_param
;
3775 shader_glsl_add_src_param(ins
, &ins
->src
[0], src_mask
, &src0_param
);
3776 shader_addline(buffer
, "tmp0.x = dot(T%u.xyz, %s);\n", reg
, src0_param
.param_str
);
3779 /** Process the WINED3DSIO_TEXM3X3PAD instruction in GLSL
3780 * Calculate the 1st or 2nd row of a 3-row matrix multiplication. */
3781 static void shader_glsl_texm3x3pad(const struct wined3d_shader_instruction
*ins
)
3783 DWORD src_mask
= WINED3DSP_WRITEMASK_0
| WINED3DSP_WRITEMASK_1
| WINED3DSP_WRITEMASK_2
;
3784 struct wined3d_shader_buffer
*buffer
= ins
->ctx
->buffer
;
3785 struct wined3d_shader_tex_mx
*tex_mx
= ins
->ctx
->tex_mx
;
3786 DWORD reg
= ins
->dst
[0].reg
.idx
[0].offset
;
3787 struct glsl_src_param src0_param
;
3789 shader_glsl_add_src_param(ins
, &ins
->src
[0], src_mask
, &src0_param
);
3790 shader_addline(buffer
, "tmp0.%c = dot(T%u.xyz, %s);\n", 'x' + tex_mx
->current_row
, reg
, src0_param
.param_str
);
3791 tex_mx
->texcoord_w
[tex_mx
->current_row
++] = reg
;
3794 static void shader_glsl_texm3x2tex(const struct wined3d_shader_instruction
*ins
)
3796 DWORD src_mask
= WINED3DSP_WRITEMASK_0
| WINED3DSP_WRITEMASK_1
| WINED3DSP_WRITEMASK_2
;
3797 struct wined3d_shader_buffer
*buffer
= ins
->ctx
->buffer
;
3798 struct glsl_sample_function sample_function
;
3799 DWORD reg
= ins
->dst
[0].reg
.idx
[0].offset
;
3800 struct glsl_src_param src0_param
;
3802 shader_glsl_add_src_param(ins
, &ins
->src
[0], src_mask
, &src0_param
);
3803 shader_addline(buffer
, "tmp0.y = dot(T%u.xyz, %s);\n", reg
, src0_param
.param_str
);
3805 shader_glsl_get_sample_function(ins
->ctx
, reg
, 0, &sample_function
);
3807 /* Sample the texture using the calculated coordinates */
3808 shader_glsl_gen_sample_code(ins
, reg
, &sample_function
, WINED3DSP_NOSWIZZLE
, NULL
, NULL
, NULL
, "tmp0.xy");
3811 /** Process the WINED3DSIO_TEXM3X3TEX instruction in GLSL
3812 * Perform the 3rd row of a 3x3 matrix multiply, then sample the texture using the calculated coordinates */
3813 static void shader_glsl_texm3x3tex(const struct wined3d_shader_instruction
*ins
)
3815 DWORD src_mask
= WINED3DSP_WRITEMASK_0
| WINED3DSP_WRITEMASK_1
| WINED3DSP_WRITEMASK_2
;
3816 struct wined3d_shader_tex_mx
*tex_mx
= ins
->ctx
->tex_mx
;
3817 struct glsl_sample_function sample_function
;
3818 DWORD reg
= ins
->dst
[0].reg
.idx
[0].offset
;
3819 struct glsl_src_param src0_param
;
3821 shader_glsl_add_src_param(ins
, &ins
->src
[0], src_mask
, &src0_param
);
3822 shader_addline(ins
->ctx
->buffer
, "tmp0.z = dot(T%u.xyz, %s);\n", reg
, src0_param
.param_str
);
3824 /* Dependent read, not valid with conditional NP2 */
3825 shader_glsl_get_sample_function(ins
->ctx
, reg
, 0, &sample_function
);
3827 /* Sample the texture using the calculated coordinates */
3828 shader_glsl_gen_sample_code(ins
, reg
, &sample_function
, WINED3DSP_NOSWIZZLE
, NULL
, NULL
, NULL
, "tmp0.xyz");
3830 tex_mx
->current_row
= 0;
3833 /** Process the WINED3DSIO_TEXM3X3 instruction in GLSL
3834 * Perform the 3rd row of a 3x3 matrix multiply */
3835 static void shader_glsl_texm3x3(const struct wined3d_shader_instruction
*ins
)
3837 DWORD src_mask
= WINED3DSP_WRITEMASK_0
| WINED3DSP_WRITEMASK_1
| WINED3DSP_WRITEMASK_2
;
3838 struct wined3d_shader_tex_mx
*tex_mx
= ins
->ctx
->tex_mx
;
3839 DWORD reg
= ins
->dst
[0].reg
.idx
[0].offset
;
3840 struct glsl_src_param src0_param
;
3843 shader_glsl_add_src_param(ins
, &ins
->src
[0], src_mask
, &src0_param
);
3845 shader_glsl_append_dst(ins
->ctx
->buffer
, ins
);
3846 shader_glsl_get_write_mask(&ins
->dst
[0], dst_mask
);
3847 shader_addline(ins
->ctx
->buffer
, "vec4(tmp0.xy, dot(T%u.xyz, %s), 1.0)%s);\n", reg
, src0_param
.param_str
, dst_mask
);
3849 tex_mx
->current_row
= 0;
3852 /* Process the WINED3DSIO_TEXM3X3SPEC instruction in GLSL
3853 * Perform the final texture lookup based on the previous 2 3x3 matrix multiplies */
3854 static void shader_glsl_texm3x3spec(const struct wined3d_shader_instruction
*ins
)
3856 struct glsl_src_param src0_param
;
3857 struct glsl_src_param src1_param
;
3858 struct wined3d_shader_buffer
*buffer
= ins
->ctx
->buffer
;
3859 struct wined3d_shader_tex_mx
*tex_mx
= ins
->ctx
->tex_mx
;
3860 DWORD src_mask
= WINED3DSP_WRITEMASK_0
| WINED3DSP_WRITEMASK_1
| WINED3DSP_WRITEMASK_2
;
3861 struct glsl_sample_function sample_function
;
3862 DWORD reg
= ins
->dst
[0].reg
.idx
[0].offset
;
3865 shader_glsl_add_src_param(ins
, &ins
->src
[0], src_mask
, &src0_param
);
3866 shader_glsl_add_src_param(ins
, &ins
->src
[1], src_mask
, &src1_param
);
3868 /* Perform the last matrix multiply operation */
3869 shader_addline(buffer
, "tmp0.z = dot(T%u.xyz, %s);\n", reg
, src0_param
.param_str
);
3870 /* Reflection calculation */
3871 shader_addline(buffer
, "tmp0.xyz = -reflect((%s), normalize(tmp0.xyz));\n", src1_param
.param_str
);
3873 /* Dependent read, not valid with conditional NP2 */
3874 shader_glsl_get_sample_function(ins
->ctx
, reg
, 0, &sample_function
);
3875 shader_glsl_write_mask_to_str(sample_function
.coord_mask
, coord_mask
);
3877 /* Sample the texture */
3878 shader_glsl_gen_sample_code(ins
, reg
, &sample_function
, WINED3DSP_NOSWIZZLE
,
3879 NULL
, NULL
, NULL
, "tmp0%s", coord_mask
);
3881 tex_mx
->current_row
= 0;
3884 /* Process the WINED3DSIO_TEXM3X3VSPEC instruction in GLSL
3885 * Perform the final texture lookup based on the previous 2 3x3 matrix multiplies */
3886 static void shader_glsl_texm3x3vspec(const struct wined3d_shader_instruction
*ins
)
3888 struct wined3d_shader_buffer
*buffer
= ins
->ctx
->buffer
;
3889 struct wined3d_shader_tex_mx
*tex_mx
= ins
->ctx
->tex_mx
;
3890 DWORD src_mask
= WINED3DSP_WRITEMASK_0
| WINED3DSP_WRITEMASK_1
| WINED3DSP_WRITEMASK_2
;
3891 struct glsl_sample_function sample_function
;
3892 DWORD reg
= ins
->dst
[0].reg
.idx
[0].offset
;
3893 struct glsl_src_param src0_param
;
3896 shader_glsl_add_src_param(ins
, &ins
->src
[0], src_mask
, &src0_param
);
3898 /* Perform the last matrix multiply operation */
3899 shader_addline(buffer
, "tmp0.z = dot(vec3(T%u), vec3(%s));\n", reg
, src0_param
.param_str
);
3901 /* Construct the eye-ray vector from w coordinates */
3902 shader_addline(buffer
, "tmp1.xyz = normalize(vec3(gl_TexCoord[%u].w, gl_TexCoord[%u].w, gl_TexCoord[%u].w));\n",
3903 tex_mx
->texcoord_w
[0], tex_mx
->texcoord_w
[1], reg
);
3904 shader_addline(buffer
, "tmp0.xyz = -reflect(tmp1.xyz, normalize(tmp0.xyz));\n");
3906 /* Dependent read, not valid with conditional NP2 */
3907 shader_glsl_get_sample_function(ins
->ctx
, reg
, 0, &sample_function
);
3908 shader_glsl_write_mask_to_str(sample_function
.coord_mask
, coord_mask
);
3910 /* Sample the texture using the calculated coordinates */
3911 shader_glsl_gen_sample_code(ins
, reg
, &sample_function
, WINED3DSP_NOSWIZZLE
,
3912 NULL
, NULL
, NULL
, "tmp0%s", coord_mask
);
3914 tex_mx
->current_row
= 0;
3917 /** Process the WINED3DSIO_TEXBEM instruction in GLSL.
3918 * Apply a fake bump map transform.
3919 * texbem is pshader <= 1.3 only, this saves a few version checks
3921 static void shader_glsl_texbem(const struct wined3d_shader_instruction
*ins
)
3923 const struct shader_glsl_ctx_priv
*priv
= ins
->ctx
->backend_data
;
3924 struct glsl_sample_function sample_function
;
3925 struct glsl_src_param coord_param
;
3931 sampler_idx
= ins
->dst
[0].reg
.idx
[0].offset
;
3932 flags
= (priv
->cur_ps_args
->tex_transform
>> sampler_idx
* WINED3D_PSARGS_TEXTRANSFORM_SHIFT
)
3933 & WINED3D_PSARGS_TEXTRANSFORM_MASK
;
3935 /* Dependent read, not valid with conditional NP2 */
3936 shader_glsl_get_sample_function(ins
->ctx
, sampler_idx
, 0, &sample_function
);
3937 mask
= sample_function
.coord_mask
;
3939 shader_glsl_write_mask_to_str(mask
, coord_mask
);
3941 /* With projected textures, texbem only divides the static texture coord,
3942 * not the displacement, so we can't let GL handle this. */
3943 if (flags
& WINED3D_PSARGS_PROJECTED
)
3946 char coord_div_mask
[3];
3947 switch (flags
& ~WINED3D_PSARGS_PROJECTED
)
3949 case WINED3D_TTFF_COUNT1
:
3950 FIXME("WINED3D_TTFF_PROJECTED with WINED3D_TTFF_COUNT1?\n");
3952 case WINED3D_TTFF_COUNT2
:
3953 div_mask
= WINED3DSP_WRITEMASK_1
;
3955 case WINED3D_TTFF_COUNT3
:
3956 div_mask
= WINED3DSP_WRITEMASK_2
;
3958 case WINED3D_TTFF_COUNT4
:
3959 case WINED3D_TTFF_DISABLE
:
3960 div_mask
= WINED3DSP_WRITEMASK_3
;
3963 shader_glsl_write_mask_to_str(div_mask
, coord_div_mask
);
3964 shader_addline(ins
->ctx
->buffer
, "T%u%s /= T%u%s;\n", sampler_idx
, coord_mask
, sampler_idx
, coord_div_mask
);
3967 shader_glsl_add_src_param(ins
, &ins
->src
[0], WINED3DSP_WRITEMASK_0
| WINED3DSP_WRITEMASK_1
, &coord_param
);
3969 shader_glsl_gen_sample_code(ins
, sampler_idx
, &sample_function
, WINED3DSP_NOSWIZZLE
, NULL
, NULL
, NULL
,
3970 "T%u%s + vec4(bumpenv_mat%u * %s, 0.0, 0.0)%s", sampler_idx
, coord_mask
, sampler_idx
,
3971 coord_param
.param_str
, coord_mask
);
3973 if (ins
->handler_idx
== WINED3DSIH_TEXBEML
)
3975 struct glsl_src_param luminance_param
;
3976 struct glsl_dst_param dst_param
;
3978 shader_glsl_add_src_param(ins
, &ins
->src
[0], WINED3DSP_WRITEMASK_2
, &luminance_param
);
3979 shader_glsl_add_dst_param(ins
, &ins
->dst
[0], &dst_param
);
3981 shader_addline(ins
->ctx
->buffer
, "%s%s *= (%s * bumpenv_lum_scale%u + bumpenv_lum_offset%u);\n",
3982 dst_param
.reg_name
, dst_param
.mask_str
,
3983 luminance_param
.param_str
, sampler_idx
, sampler_idx
);
3987 static void shader_glsl_bem(const struct wined3d_shader_instruction
*ins
)
3989 DWORD sampler_idx
= ins
->dst
[0].reg
.idx
[0].offset
;
3990 struct glsl_src_param src0_param
, src1_param
;
3992 shader_glsl_add_src_param(ins
, &ins
->src
[0], WINED3DSP_WRITEMASK_0
| WINED3DSP_WRITEMASK_1
, &src0_param
);
3993 shader_glsl_add_src_param(ins
, &ins
->src
[1], WINED3DSP_WRITEMASK_0
| WINED3DSP_WRITEMASK_1
, &src1_param
);
3995 shader_glsl_append_dst(ins
->ctx
->buffer
, ins
);
3996 shader_addline(ins
->ctx
->buffer
, "%s + bumpenv_mat%u * %s);\n",
3997 src0_param
.param_str
, sampler_idx
, src1_param
.param_str
);
4000 /** Process the WINED3DSIO_TEXREG2AR instruction in GLSL
4001 * Sample 2D texture at dst using the alpha & red (wx) components of src as texture coordinates */
4002 static void shader_glsl_texreg2ar(const struct wined3d_shader_instruction
*ins
)
4004 DWORD sampler_idx
= ins
->dst
[0].reg
.idx
[0].offset
;
4005 struct glsl_sample_function sample_function
;
4006 struct glsl_src_param src0_param
;
4008 shader_glsl_add_src_param(ins
, &ins
->src
[0], WINED3DSP_WRITEMASK_ALL
, &src0_param
);
4010 shader_glsl_get_sample_function(ins
->ctx
, sampler_idx
, 0, &sample_function
);
4011 shader_glsl_gen_sample_code(ins
, sampler_idx
, &sample_function
, WINED3DSP_NOSWIZZLE
, NULL
, NULL
, NULL
,
4012 "%s.wx", src0_param
.reg_name
);
4015 /** Process the WINED3DSIO_TEXREG2GB instruction in GLSL
4016 * Sample 2D texture at dst using the green & blue (yz) components of src as texture coordinates */
4017 static void shader_glsl_texreg2gb(const struct wined3d_shader_instruction
*ins
)
4019 DWORD sampler_idx
= ins
->dst
[0].reg
.idx
[0].offset
;
4020 struct glsl_sample_function sample_function
;
4021 struct glsl_src_param src0_param
;
4023 shader_glsl_add_src_param(ins
, &ins
->src
[0], WINED3DSP_WRITEMASK_ALL
, &src0_param
);
4025 shader_glsl_get_sample_function(ins
->ctx
, sampler_idx
, 0, &sample_function
);
4026 shader_glsl_gen_sample_code(ins
, sampler_idx
, &sample_function
, WINED3DSP_NOSWIZZLE
, NULL
, NULL
, NULL
,
4027 "%s.yz", src0_param
.reg_name
);
4030 /** Process the WINED3DSIO_TEXREG2RGB instruction in GLSL
4031 * Sample texture at dst using the rgb (xyz) components of src as texture coordinates */
4032 static void shader_glsl_texreg2rgb(const struct wined3d_shader_instruction
*ins
)
4034 DWORD sampler_idx
= ins
->dst
[0].reg
.idx
[0].offset
;
4035 struct glsl_sample_function sample_function
;
4036 struct glsl_src_param src0_param
;
4038 /* Dependent read, not valid with conditional NP2 */
4039 shader_glsl_get_sample_function(ins
->ctx
, sampler_idx
, 0, &sample_function
);
4040 shader_glsl_add_src_param(ins
, &ins
->src
[0], sample_function
.coord_mask
, &src0_param
);
4042 shader_glsl_gen_sample_code(ins
, sampler_idx
, &sample_function
, WINED3DSP_NOSWIZZLE
, NULL
, NULL
, NULL
,
4043 "%s", src0_param
.param_str
);
4046 /** Process the WINED3DSIO_TEXKILL instruction in GLSL.
4047 * If any of the first 3 components are < 0, discard this pixel */
4048 static void shader_glsl_texkill(const struct wined3d_shader_instruction
*ins
)
4050 struct glsl_dst_param dst_param
;
4052 /* The argument is a destination parameter, and no writemasks are allowed */
4053 shader_glsl_add_dst_param(ins
, &ins
->dst
[0], &dst_param
);
4054 if (ins
->ctx
->reg_maps
->shader_version
.major
>= 2)
4056 /* 2.0 shaders compare all 4 components in texkill */
4057 shader_addline(ins
->ctx
->buffer
, "if (any(lessThan(%s.xyzw, vec4(0.0)))) discard;\n", dst_param
.reg_name
);
4059 /* 1.X shaders only compare the first 3 components, probably due to the nature of the texkill
4060 * instruction as a tex* instruction, and phase, which kills all a / w components. Even if all
4061 * 4 components are defined, only the first 3 are used
4063 shader_addline(ins
->ctx
->buffer
, "if (any(lessThan(%s.xyz, vec3(0.0)))) discard;\n", dst_param
.reg_name
);
4067 /** Process the WINED3DSIO_DP2ADD instruction in GLSL.
4068 * dst = dot2(src0, src1) + src2 */
4069 static void shader_glsl_dp2add(const struct wined3d_shader_instruction
*ins
)
4071 struct glsl_src_param src0_param
;
4072 struct glsl_src_param src1_param
;
4073 struct glsl_src_param src2_param
;
4075 unsigned int mask_size
;
4077 write_mask
= shader_glsl_append_dst(ins
->ctx
->buffer
, ins
);
4078 mask_size
= shader_glsl_get_write_mask_size(write_mask
);
4080 shader_glsl_add_src_param(ins
, &ins
->src
[0], WINED3DSP_WRITEMASK_0
| WINED3DSP_WRITEMASK_1
, &src0_param
);
4081 shader_glsl_add_src_param(ins
, &ins
->src
[1], WINED3DSP_WRITEMASK_0
| WINED3DSP_WRITEMASK_1
, &src1_param
);
4082 shader_glsl_add_src_param(ins
, &ins
->src
[2], WINED3DSP_WRITEMASK_0
, &src2_param
);
4084 if (mask_size
> 1) {
4085 shader_addline(ins
->ctx
->buffer
, "vec%d(dot(%s, %s) + %s));\n",
4086 mask_size
, src0_param
.param_str
, src1_param
.param_str
, src2_param
.param_str
);
4088 shader_addline(ins
->ctx
->buffer
, "dot(%s, %s) + %s);\n",
4089 src0_param
.param_str
, src1_param
.param_str
, src2_param
.param_str
);
4093 static void shader_glsl_input_pack(const struct wined3d_shader
*shader
, struct wined3d_shader_buffer
*buffer
,
4094 const struct wined3d_shader_signature_element
*input_signature
,
4095 const struct wined3d_shader_reg_maps
*reg_maps
,
4096 enum vertexprocessing_mode vertexprocessing
)
4098 WORD map
= reg_maps
->input_registers
;
4101 for (i
= 0; map
; map
>>= 1, ++i
)
4103 const char *semantic_name
;
4108 if (!(map
& 1)) continue;
4110 semantic_name
= input_signature
[i
].semantic_name
;
4111 semantic_idx
= input_signature
[i
].semantic_idx
;
4112 shader_glsl_write_mask_to_str(input_signature
[i
].mask
, reg_mask
);
4114 if (shader_match_semantic(semantic_name
, WINED3D_DECL_USAGE_TEXCOORD
))
4116 if (semantic_idx
< 8 && vertexprocessing
== pretransformed
)
4117 shader_addline(buffer
, "ps_in[%u]%s = gl_TexCoord[%u]%s;\n",
4118 shader
->u
.ps
.input_reg_map
[i
], reg_mask
, semantic_idx
, reg_mask
);
4120 shader_addline(buffer
, "ps_in[%u]%s = vec4(0.0, 0.0, 0.0, 0.0)%s;\n",
4121 shader
->u
.ps
.input_reg_map
[i
], reg_mask
, reg_mask
);
4123 else if (shader_match_semantic(semantic_name
, WINED3D_DECL_USAGE_COLOR
))
4126 shader_addline(buffer
, "ps_in[%u]%s = vec4(gl_Color)%s;\n",
4127 shader
->u
.ps
.input_reg_map
[i
], reg_mask
, reg_mask
);
4128 else if (semantic_idx
== 1)
4129 shader_addline(buffer
, "ps_in[%u]%s = vec4(gl_SecondaryColor)%s;\n",
4130 shader
->u
.ps
.input_reg_map
[i
], reg_mask
, reg_mask
);
4132 shader_addline(buffer
, "ps_in[%u]%s = vec4(0.0, 0.0, 0.0, 0.0)%s;\n",
4133 shader
->u
.ps
.input_reg_map
[i
], reg_mask
, reg_mask
);
4137 shader_addline(buffer
, "ps_in[%u]%s = vec4(0.0, 0.0, 0.0, 0.0)%s;\n",
4138 shader
->u
.ps
.input_reg_map
[i
], reg_mask
, reg_mask
);
4143 /*********************************************
4144 * Vertex Shader Specific Code begins here
4145 ********************************************/
4147 static void add_glsl_program_entry(struct shader_glsl_priv
*priv
, struct glsl_shader_prog_link
*entry
)
4149 struct glsl_program_key key
;
4151 key
.vs_id
= entry
->vs
.id
;
4152 key
.gs_id
= entry
->gs
.id
;
4153 key
.ps_id
= entry
->ps
.id
;
4155 if (wine_rb_put(&priv
->program_lookup
, &key
, &entry
->program_lookup_entry
) == -1)
4157 ERR("Failed to insert program entry.\n");
4161 static struct glsl_shader_prog_link
*get_glsl_program_entry(const struct shader_glsl_priv
*priv
,
4162 GLhandleARB vs_id
, GLhandleARB gs_id
, GLhandleARB ps_id
)
4164 struct wine_rb_entry
*entry
;
4165 struct glsl_program_key key
;
4171 entry
= wine_rb_get(&priv
->program_lookup
, &key
);
4172 return entry
? WINE_RB_ENTRY_VALUE(entry
, struct glsl_shader_prog_link
, program_lookup_entry
) : NULL
;
4175 /* Context activation is done by the caller. */
4176 static void delete_glsl_program_entry(struct shader_glsl_priv
*priv
, const struct wined3d_gl_info
*gl_info
,
4177 struct glsl_shader_prog_link
*entry
)
4179 struct glsl_program_key key
;
4181 key
.vs_id
= entry
->vs
.id
;
4182 key
.gs_id
= entry
->gs
.id
;
4183 key
.ps_id
= entry
->ps
.id
;
4184 wine_rb_remove(&priv
->program_lookup
, &key
);
4186 GL_EXTCALL(glDeleteObjectARB(entry
->programId
));
4188 list_remove(&entry
->vs
.shader_entry
);
4190 list_remove(&entry
->gs
.shader_entry
);
4192 list_remove(&entry
->ps
.shader_entry
);
4193 HeapFree(GetProcessHeap(), 0, entry
->vs
.uniform_f_locations
);
4194 HeapFree(GetProcessHeap(), 0, entry
->ps
.uniform_f_locations
);
4195 HeapFree(GetProcessHeap(), 0, entry
);
4198 static void handle_ps3_input(struct wined3d_shader_buffer
*buffer
,
4199 const struct wined3d_gl_info
*gl_info
, const DWORD
*map
,
4200 const struct wined3d_shader_signature_element
*input_signature
,
4201 const struct wined3d_shader_reg_maps
*reg_maps_in
,
4202 const struct wined3d_shader_signature_element
*output_signature
,
4203 const struct wined3d_shader_reg_maps
*reg_maps_out
)
4206 const char *semantic_name_in
;
4207 UINT semantic_idx_in
;
4210 unsigned int in_count
= vec4_varyings(3, gl_info
);
4212 char destination
[50];
4213 WORD input_map
, output_map
;
4215 set
= HeapAlloc(GetProcessHeap(), HEAP_ZERO_MEMORY
, sizeof(*set
) * (in_count
+ 2));
4217 input_map
= reg_maps_in
->input_registers
;
4218 for (i
= 0; input_map
; input_map
>>= 1, ++i
)
4220 if (!(input_map
& 1)) continue;
4223 /* Declared, but not read register */
4224 if (in_idx
== ~0U) continue;
4225 if (in_idx
>= (in_count
+ 2))
4227 FIXME("More input varyings declared than supported, expect issues.\n");
4231 if (in_idx
== in_count
)
4232 sprintf(destination
, "gl_FrontColor");
4233 else if (in_idx
== in_count
+ 1)
4234 sprintf(destination
, "gl_FrontSecondaryColor");
4236 sprintf(destination
, "ps_in[%u]", in_idx
);
4238 semantic_name_in
= input_signature
[i
].semantic_name
;
4239 semantic_idx_in
= input_signature
[i
].semantic_idx
;
4242 output_map
= reg_maps_out
->output_registers
;
4243 for (j
= 0; output_map
; output_map
>>= 1, ++j
)
4247 if (!(output_map
& 1)
4248 || semantic_idx_in
!= output_signature
[j
].semantic_idx
4249 || strcmp(semantic_name_in
, output_signature
[j
].semantic_name
)
4250 || !(mask
= input_signature
[i
].mask
& output_signature
[j
].mask
))
4254 shader_glsl_write_mask_to_str(mask
, reg_mask
);
4256 shader_addline(buffer
, "%s%s = vs_out[%u]%s;\n",
4257 destination
, reg_mask
, j
, reg_mask
);
4261 for (i
= 0; i
< in_count
+ 2; ++i
)
4265 if (!set
[i
] || set
[i
] == WINED3DSP_WRITEMASK_ALL
)
4268 if (set
[i
] == ~0U) set
[i
] = 0;
4271 if (!(set
[i
] & WINED3DSP_WRITEMASK_0
)) reg_mask
[size
++] = 'x';
4272 if (!(set
[i
] & WINED3DSP_WRITEMASK_1
)) reg_mask
[size
++] = 'y';
4273 if (!(set
[i
] & WINED3DSP_WRITEMASK_2
)) reg_mask
[size
++] = 'z';
4274 if (!(set
[i
] & WINED3DSP_WRITEMASK_3
)) reg_mask
[size
++] = 'w';
4275 reg_mask
[size
] = '\0';
4278 sprintf(destination
, "gl_FrontColor");
4279 else if (i
== in_count
+ 1)
4280 sprintf(destination
, "gl_FrontSecondaryColor");
4282 sprintf(destination
, "ps_in[%u]", i
);
4284 if (size
== 1) shader_addline(buffer
, "%s.%s = 0.0;\n", destination
, reg_mask
);
4285 else shader_addline(buffer
, "%s.%s = vec%u(0.0);\n", destination
, reg_mask
, size
);
4288 HeapFree(GetProcessHeap(), 0, set
);
4291 /* Context activation is done by the caller. */
4292 static GLhandleARB
generate_param_reorder_function(struct wined3d_shader_buffer
*buffer
,
4293 const struct wined3d_shader
*vs
, const struct wined3d_shader
*ps
,
4294 const struct wined3d_gl_info
*gl_info
)
4296 GLhandleARB ret
= 0;
4297 DWORD ps_major
= ps
? ps
->reg_maps
.shader_version
.major
: 0;
4299 const char *semantic_name
;
4302 const struct wined3d_shader_signature_element
*output_signature
= vs
->output_signature
;
4303 WORD map
= vs
->reg_maps
.output_registers
;
4305 shader_buffer_clear(buffer
);
4307 shader_addline(buffer
, "#version 120\n");
4311 shader_addline(buffer
, "void order_ps_input(in vec4 vs_out[%u])\n{\n", vs
->limits
.packed_output
);
4313 for (i
= 0; map
; map
>>= 1, ++i
)
4317 if (!(map
& 1)) continue;
4319 semantic_name
= output_signature
[i
].semantic_name
;
4320 semantic_idx
= output_signature
[i
].semantic_idx
;
4321 write_mask
= output_signature
[i
].mask
;
4322 shader_glsl_write_mask_to_str(write_mask
, reg_mask
);
4324 if (shader_match_semantic(semantic_name
, WINED3D_DECL_USAGE_COLOR
))
4327 shader_addline(buffer
, "gl_FrontColor%s = vs_out[%u]%s;\n",
4328 reg_mask
, i
, reg_mask
);
4329 else if (semantic_idx
== 1)
4330 shader_addline(buffer
, "gl_FrontSecondaryColor%s = vs_out[%u]%s;\n",
4331 reg_mask
, i
, reg_mask
);
4333 else if (shader_match_semantic(semantic_name
, WINED3D_DECL_USAGE_POSITION
))
4335 shader_addline(buffer
, "gl_Position%s = vs_out[%u]%s;\n",
4336 reg_mask
, i
, reg_mask
);
4338 else if (shader_match_semantic(semantic_name
, WINED3D_DECL_USAGE_TEXCOORD
))
4340 if (semantic_idx
< 8)
4342 if (!(gl_info
->quirks
& WINED3D_QUIRK_SET_TEXCOORD_W
) || ps_major
> 0)
4343 write_mask
|= WINED3DSP_WRITEMASK_3
;
4345 shader_addline(buffer
, "gl_TexCoord[%u]%s = vs_out[%u]%s;\n",
4346 semantic_idx
, reg_mask
, i
, reg_mask
);
4347 if (!(write_mask
& WINED3DSP_WRITEMASK_3
))
4348 shader_addline(buffer
, "gl_TexCoord[%u].w = 1.0;\n", semantic_idx
);
4351 else if (shader_match_semantic(semantic_name
, WINED3D_DECL_USAGE_PSIZE
))
4353 shader_addline(buffer
, "gl_PointSize = vs_out[%u].%c;\n", i
, reg_mask
[1]);
4355 else if (shader_match_semantic(semantic_name
, WINED3D_DECL_USAGE_FOG
))
4357 shader_addline(buffer
, "gl_FogFragCoord = clamp(vs_out[%u].%c, 0.0, 1.0);\n", i
, reg_mask
[1]);
4360 shader_addline(buffer
, "}\n");
4364 UINT in_count
= min(vec4_varyings(ps_major
, gl_info
), ps
->limits
.packed_input
);
4365 /* This one is tricky: a 3.0 pixel shader reads from a 3.0 vertex shader */
4366 shader_addline(buffer
, "varying vec4 ps_in[%u];\n", in_count
);
4367 shader_addline(buffer
, "void order_ps_input(in vec4 vs_out[%u])\n{\n", vs
->limits
.packed_output
);
4369 /* First, sort out position and point size. Those are not passed to the pixel shader */
4370 for (i
= 0; map
; map
>>= 1, ++i
)
4372 if (!(map
& 1)) continue;
4374 semantic_name
= output_signature
[i
].semantic_name
;
4375 shader_glsl_write_mask_to_str(output_signature
[i
].mask
, reg_mask
);
4377 if (shader_match_semantic(semantic_name
, WINED3D_DECL_USAGE_POSITION
))
4379 shader_addline(buffer
, "gl_Position%s = vs_out[%u]%s;\n",
4380 reg_mask
, i
, reg_mask
);
4382 else if (shader_match_semantic(semantic_name
, WINED3D_DECL_USAGE_PSIZE
))
4384 shader_addline(buffer
, "gl_PointSize = vs_out[%u].%c;\n", i
, reg_mask
[1]);
4388 /* Then, fix the pixel shader input */
4389 handle_ps3_input(buffer
, gl_info
, ps
->u
.ps
.input_reg_map
, ps
->input_signature
,
4390 &ps
->reg_maps
, output_signature
, &vs
->reg_maps
);
4392 shader_addline(buffer
, "}\n");
4395 ret
= GL_EXTCALL(glCreateShaderObjectARB(GL_VERTEX_SHADER_ARB
));
4396 checkGLcall("glCreateShaderObjectARB(GL_VERTEX_SHADER_ARB)");
4397 shader_glsl_compile(gl_info
, ret
, buffer
->buffer
);
4402 static void shader_glsl_generate_srgb_write_correction(struct wined3d_shader_buffer
*buffer
)
4404 shader_addline(buffer
, "tmp0.xyz = pow(gl_FragData[0].xyz, vec3(srgb_const0.x));\n");
4405 shader_addline(buffer
, "tmp0.xyz = tmp0.xyz * vec3(srgb_const0.y) - vec3(srgb_const0.z);\n");
4406 shader_addline(buffer
, "tmp1.xyz = gl_FragData[0].xyz * vec3(srgb_const0.w);\n");
4407 shader_addline(buffer
, "bvec3 srgb_compare = lessThan(gl_FragData[0].xyz, vec3(srgb_const1.x));\n");
4408 shader_addline(buffer
, "gl_FragData[0].xyz = mix(tmp0.xyz, tmp1.xyz, vec3(srgb_compare));\n");
4409 shader_addline(buffer
, "gl_FragData[0] = clamp(gl_FragData[0], 0.0, 1.0);\n");
4412 static void shader_glsl_generate_fog_code(struct wined3d_shader_buffer
*buffer
, enum fogmode mode
)
4420 /* Fog = (gl_Fog.end - gl_FogFragCoord) / (gl_Fog.end - gl_Fog.start) */
4421 shader_addline(buffer
, "float Fog = (gl_Fog.end - gl_FogFragCoord) / (gl_Fog.end - gl_Fog.start);\n");
4425 /* Fog = e^-(gl_Fog.density * gl_FogFragCoord) */
4426 shader_addline(buffer
, "float Fog = exp(-gl_Fog.density * gl_FogFragCoord);\n");
4430 /* Fog = e^-((gl_Fog.density * gl_FogFragCoord)^2) */
4431 shader_addline(buffer
, "float Fog = exp(-gl_Fog.density * gl_Fog.density * gl_FogFragCoord * gl_FogFragCoord);\n");
4435 ERR("Invalid fog mode %#x.\n", mode
);
4439 shader_addline(buffer
, "gl_FragData[0].xyz = mix(gl_Fog.color.xyz, gl_FragData[0].xyz, clamp(Fog, 0.0, 1.0));\n");
4442 /* Context activation is done by the caller. */
4443 static GLuint
shader_glsl_generate_pshader(const struct wined3d_context
*context
,
4444 struct wined3d_shader_buffer
*buffer
, const struct wined3d_shader
*shader
,
4445 const struct ps_compile_args
*args
, struct ps_np2fixup_info
*np2fixup_info
)
4447 const struct wined3d_shader_reg_maps
*reg_maps
= &shader
->reg_maps
;
4448 const struct wined3d_gl_info
*gl_info
= context
->gl_info
;
4449 const DWORD
*function
= shader
->function
;
4450 struct shader_glsl_ctx_priv priv_ctx
;
4452 /* Create the hw GLSL shader object and assign it as the shader->prgId */
4453 GLhandleARB shader_obj
= GL_EXTCALL(glCreateShaderObjectARB(GL_FRAGMENT_SHADER_ARB
));
4455 memset(&priv_ctx
, 0, sizeof(priv_ctx
));
4456 priv_ctx
.cur_ps_args
= args
;
4457 priv_ctx
.cur_np2fixup_info
= np2fixup_info
;
4459 shader_addline(buffer
, "#version 120\n");
4461 if (gl_info
->supported
[ARB_SHADER_BIT_ENCODING
])
4462 shader_addline(buffer
, "#extension GL_ARB_shader_bit_encoding : enable\n");
4463 if (gl_info
->supported
[ARB_SHADER_TEXTURE_LOD
])
4464 shader_addline(buffer
, "#extension GL_ARB_shader_texture_lod : enable\n");
4465 /* The spec says that it doesn't have to be explicitly enabled, but the
4466 * nvidia drivers write a warning if we don't do so. */
4467 if (gl_info
->supported
[ARB_TEXTURE_RECTANGLE
])
4468 shader_addline(buffer
, "#extension GL_ARB_texture_rectangle : enable\n");
4469 if (gl_info
->supported
[EXT_GPU_SHADER4
])
4470 shader_addline(buffer
, "#extension GL_EXT_gpu_shader4 : enable\n");
4472 /* Base Declarations */
4473 shader_generate_glsl_declarations(context
, buffer
, shader
, reg_maps
, &priv_ctx
);
4475 /* Pack 3.0 inputs */
4476 if (reg_maps
->shader_version
.major
>= 3 && args
->vp_mode
!= vertexshader
)
4477 shader_glsl_input_pack(shader
, buffer
, shader
->input_signature
, reg_maps
, args
->vp_mode
);
4479 /* Base Shader Body */
4480 shader_generate_main(shader
, buffer
, reg_maps
, function
, &priv_ctx
);
4482 /* Pixel shaders < 2.0 place the resulting color in R0 implicitly */
4483 if (reg_maps
->shader_version
.major
< 2)
4485 /* Some older cards like GeforceFX ones don't support multiple buffers, so also not gl_FragData */
4486 shader_addline(buffer
, "gl_FragData[0] = R0;\n");
4489 if (args
->srgb_correction
)
4490 shader_glsl_generate_srgb_write_correction(buffer
);
4492 /* SM < 3 does not replace the fog stage. */
4493 if (reg_maps
->shader_version
.major
< 3)
4494 shader_glsl_generate_fog_code(buffer
, args
->fog
);
4496 shader_addline(buffer
, "}\n");
4498 TRACE("Compiling shader object %u\n", shader_obj
);
4499 shader_glsl_compile(gl_info
, shader_obj
, buffer
->buffer
);
4501 /* Store the shader object */
4505 /* Context activation is done by the caller. */
4506 static GLuint
shader_glsl_generate_vshader(const struct wined3d_context
*context
,
4507 struct wined3d_shader_buffer
*buffer
, const struct wined3d_shader
*shader
,
4508 const struct vs_compile_args
*args
)
4510 const struct wined3d_shader_reg_maps
*reg_maps
= &shader
->reg_maps
;
4511 const struct wined3d_gl_info
*gl_info
= context
->gl_info
;
4512 const DWORD
*function
= shader
->function
;
4513 struct shader_glsl_ctx_priv priv_ctx
;
4515 /* Create the hw GLSL shader program and assign it as the shader->prgId */
4516 GLhandleARB shader_obj
= GL_EXTCALL(glCreateShaderObjectARB(GL_VERTEX_SHADER_ARB
));
4518 shader_addline(buffer
, "#version 120\n");
4520 if (gl_info
->supported
[ARB_SHADER_BIT_ENCODING
])
4521 shader_addline(buffer
, "#extension GL_ARB_shader_bit_encoding : enable\n");
4522 if (gl_info
->supported
[EXT_GPU_SHADER4
])
4523 shader_addline(buffer
, "#extension GL_EXT_gpu_shader4 : enable\n");
4525 memset(&priv_ctx
, 0, sizeof(priv_ctx
));
4526 priv_ctx
.cur_vs_args
= args
;
4528 /* Base Declarations */
4529 shader_generate_glsl_declarations(context
, buffer
, shader
, reg_maps
, &priv_ctx
);
4531 /* Base Shader Body */
4532 shader_generate_main(shader
, buffer
, reg_maps
, function
, &priv_ctx
);
4534 /* Unpack outputs */
4535 shader_addline(buffer
, "order_ps_input(vs_out);\n");
4537 /* The D3DRS_FOGTABLEMODE render state defines if the shader-generated fog coord is used
4538 * or if the fragment depth is used. If the fragment depth is used(FOGTABLEMODE != NONE),
4539 * the fog frag coord is thrown away. If the fog frag coord is used, but not written by
4540 * the shader, it is set to 0.0(fully fogged, since start = 1.0, end = 0.0)
4542 if (args
->fog_src
== VS_FOG_Z
)
4543 shader_addline(buffer
, "gl_FogFragCoord = gl_Position.z;\n");
4544 else if (!reg_maps
->fog
)
4545 shader_addline(buffer
, "gl_FogFragCoord = 0.0;\n");
4547 /* We always store the clipplanes without y inversion */
4548 if (args
->clip_enabled
)
4549 shader_addline(buffer
, "gl_ClipVertex = gl_Position;\n");
4551 /* Write the final position.
4553 * OpenGL coordinates specify the center of the pixel while d3d coords specify
4554 * the corner. The offsets are stored in z and w in posFixup. posFixup.y contains
4555 * 1.0 or -1.0 to turn the rendering upside down for offscreen rendering. PosFixup.x
4556 * contains 1.0 to allow a mad.
4558 shader_addline(buffer
, "gl_Position.y = gl_Position.y * posFixup.y;\n");
4559 shader_addline(buffer
, "gl_Position.xy += posFixup.zw * gl_Position.ww;\n");
4561 /* Z coord [0;1]->[-1;1] mapping, see comment in transform_projection in state.c
4563 * Basically we want (in homogeneous coordinates) z = z * 2 - 1. However, shaders are run
4564 * before the homogeneous divide, so we have to take the w into account: z = ((z / w) * 2 - 1) * w,
4565 * which is the same as z = z * 2 - w.
4567 shader_addline(buffer
, "gl_Position.z = gl_Position.z * 2.0 - gl_Position.w;\n");
4569 shader_addline(buffer
, "}\n");
4571 TRACE("Compiling shader object %u\n", shader_obj
);
4572 shader_glsl_compile(gl_info
, shader_obj
, buffer
->buffer
);
4577 /* Context activation is done by the caller. */
4578 static GLhandleARB
shader_glsl_generate_geometry_shader(const struct wined3d_context
*context
,
4579 struct wined3d_shader_buffer
*buffer
, const struct wined3d_shader
*shader
)
4581 const struct wined3d_shader_reg_maps
*reg_maps
= &shader
->reg_maps
;
4582 const struct wined3d_gl_info
*gl_info
= context
->gl_info
;
4583 const DWORD
*function
= shader
->function
;
4584 struct shader_glsl_ctx_priv priv_ctx
;
4585 GLhandleARB shader_id
;
4587 shader_id
= GL_EXTCALL(glCreateShaderObjectARB(GL_GEOMETRY_SHADER_ARB
));
4589 shader_addline(buffer
, "#version 120\n");
4591 if (gl_info
->supported
[ARB_GEOMETRY_SHADER4
])
4592 shader_addline(buffer
, "#extension GL_ARB_geometry_shader4 : enable\n");
4593 if (gl_info
->supported
[ARB_SHADER_BIT_ENCODING
])
4594 shader_addline(buffer
, "#extension GL_ARB_shader_bit_encoding : enable\n");
4595 if (gl_info
->supported
[EXT_GPU_SHADER4
])
4596 shader_addline(buffer
, "#extension GL_EXT_gpu_shader4 : enable\n");
4598 memset(&priv_ctx
, 0, sizeof(priv_ctx
));
4599 shader_generate_glsl_declarations(context
, buffer
, shader
, reg_maps
, &priv_ctx
);
4600 shader_generate_main(shader
, buffer
, reg_maps
, function
, &priv_ctx
);
4601 shader_addline(buffer
, "}\n");
4603 TRACE("Compiling shader object %u.\n", shader_id
);
4604 shader_glsl_compile(gl_info
, shader_id
, buffer
->buffer
);
4609 static GLhandleARB
find_glsl_pshader(const struct wined3d_context
*context
,
4610 struct wined3d_shader_buffer
*buffer
, struct wined3d_shader
*shader
,
4611 const struct ps_compile_args
*args
, const struct ps_np2fixup_info
**np2fixup_info
)
4613 struct glsl_ps_compiled_shader
*gl_shaders
, *new_array
;
4614 struct glsl_shader_private
*shader_data
;
4615 struct ps_np2fixup_info
*np2fixup
;
4620 if (!shader
->backend_data
)
4622 shader
->backend_data
= HeapAlloc(GetProcessHeap(), HEAP_ZERO_MEMORY
, sizeof(*shader_data
));
4623 if (!shader
->backend_data
)
4625 ERR("Failed to allocate backend data.\n");
4629 shader_data
= shader
->backend_data
;
4630 gl_shaders
= shader_data
->gl_shaders
.ps
;
4632 /* Usually we have very few GL shaders for each d3d shader(just 1 or maybe 2),
4633 * so a linear search is more performant than a hashmap or a binary search
4634 * (cache coherency etc)
4636 for (i
= 0; i
< shader_data
->num_gl_shaders
; ++i
)
4638 if (!memcmp(&gl_shaders
[i
].args
, args
, sizeof(*args
)))
4640 if (args
->np2_fixup
)
4641 *np2fixup_info
= &gl_shaders
[i
].np2fixup
;
4642 return gl_shaders
[i
].prgId
;
4646 TRACE("No matching GL shader found for shader %p, compiling a new shader.\n", shader
);
4647 if(shader_data
->shader_array_size
== shader_data
->num_gl_shaders
) {
4648 if (shader_data
->num_gl_shaders
)
4650 new_size
= shader_data
->shader_array_size
+ max(1, shader_data
->shader_array_size
/ 2);
4651 new_array
= HeapReAlloc(GetProcessHeap(), 0, shader_data
->gl_shaders
.ps
,
4652 new_size
* sizeof(*gl_shaders
));
4656 new_array
= HeapAlloc(GetProcessHeap(), 0, sizeof(*gl_shaders
));
4661 ERR("Out of memory\n");
4664 shader_data
->gl_shaders
.ps
= new_array
;
4665 shader_data
->shader_array_size
= new_size
;
4666 gl_shaders
= new_array
;
4669 gl_shaders
[shader_data
->num_gl_shaders
].args
= *args
;
4671 np2fixup
= &gl_shaders
[shader_data
->num_gl_shaders
].np2fixup
;
4672 memset(np2fixup
, 0, sizeof(*np2fixup
));
4673 *np2fixup_info
= args
->np2_fixup
? np2fixup
: NULL
;
4675 pixelshader_update_samplers(shader
, args
->tex_types
);
4677 shader_buffer_clear(buffer
);
4678 ret
= shader_glsl_generate_pshader(context
, buffer
, shader
, args
, np2fixup
);
4679 gl_shaders
[shader_data
->num_gl_shaders
++].prgId
= ret
;
4684 static inline BOOL
vs_args_equal(const struct vs_compile_args
*stored
, const struct vs_compile_args
*new,
4685 const DWORD use_map
) {
4686 if((stored
->swizzle_map
& use_map
) != new->swizzle_map
) return FALSE
;
4687 if((stored
->clip_enabled
) != new->clip_enabled
) return FALSE
;
4688 return stored
->fog_src
== new->fog_src
;
4691 static GLhandleARB
find_glsl_vshader(const struct wined3d_context
*context
,
4692 struct wined3d_shader_buffer
*buffer
, struct wined3d_shader
*shader
,
4693 const struct vs_compile_args
*args
)
4697 DWORD use_map
= shader
->device
->stream_info
.use_map
;
4698 struct glsl_vs_compiled_shader
*gl_shaders
, *new_array
;
4699 struct glsl_shader_private
*shader_data
;
4702 if (!shader
->backend_data
)
4704 shader
->backend_data
= HeapAlloc(GetProcessHeap(), HEAP_ZERO_MEMORY
, sizeof(*shader_data
));
4705 if (!shader
->backend_data
)
4707 ERR("Failed to allocate backend data.\n");
4711 shader_data
= shader
->backend_data
;
4712 gl_shaders
= shader_data
->gl_shaders
.vs
;
4714 /* Usually we have very few GL shaders for each d3d shader(just 1 or maybe 2),
4715 * so a linear search is more performant than a hashmap or a binary search
4716 * (cache coherency etc)
4718 for (i
= 0; i
< shader_data
->num_gl_shaders
; ++i
)
4720 if (vs_args_equal(&gl_shaders
[i
].args
, args
, use_map
))
4721 return gl_shaders
[i
].prgId
;
4724 TRACE("No matching GL shader found for shader %p, compiling a new shader.\n", shader
);
4726 if(shader_data
->shader_array_size
== shader_data
->num_gl_shaders
) {
4727 if (shader_data
->num_gl_shaders
)
4729 new_size
= shader_data
->shader_array_size
+ max(1, shader_data
->shader_array_size
/ 2);
4730 new_array
= HeapReAlloc(GetProcessHeap(), 0, shader_data
->gl_shaders
.vs
,
4731 new_size
* sizeof(*gl_shaders
));
4735 new_array
= HeapAlloc(GetProcessHeap(), 0, sizeof(*gl_shaders
));
4740 ERR("Out of memory\n");
4743 shader_data
->gl_shaders
.vs
= new_array
;
4744 shader_data
->shader_array_size
= new_size
;
4745 gl_shaders
= new_array
;
4748 gl_shaders
[shader_data
->num_gl_shaders
].args
= *args
;
4750 shader_buffer_clear(buffer
);
4751 ret
= shader_glsl_generate_vshader(context
, buffer
, shader
, args
);
4752 gl_shaders
[shader_data
->num_gl_shaders
++].prgId
= ret
;
4757 static GLhandleARB
find_glsl_geometry_shader(const struct wined3d_context
*context
,
4758 struct wined3d_shader_buffer
*buffer
, struct wined3d_shader
*shader
)
4760 struct glsl_gs_compiled_shader
*gl_shaders
;
4761 struct glsl_shader_private
*shader_data
;
4764 if (!shader
->backend_data
)
4766 if (!(shader
->backend_data
= HeapAlloc(GetProcessHeap(), HEAP_ZERO_MEMORY
, sizeof(*shader_data
))))
4768 ERR("Failed to allocate backend data.\n");
4772 shader_data
= shader
->backend_data
;
4773 gl_shaders
= shader_data
->gl_shaders
.gs
;
4775 if (shader_data
->num_gl_shaders
)
4776 return gl_shaders
[0].id
;
4778 TRACE("No matching GL shader found for shader %p, compiling a new shader.\n", shader
);
4780 if (!(shader_data
->gl_shaders
.gs
= HeapAlloc(GetProcessHeap(), 0, sizeof(*gl_shaders
))))
4782 ERR("Failed to allocate GL shader array.\n");
4785 shader_data
->shader_array_size
= 1;
4786 gl_shaders
= shader_data
->gl_shaders
.gs
;
4788 shader_buffer_clear(buffer
);
4789 ret
= shader_glsl_generate_geometry_shader(context
, buffer
, shader
);
4790 gl_shaders
[shader_data
->num_gl_shaders
++].id
= ret
;
4795 static const char *shader_glsl_get_ffp_fragment_op_arg(struct wined3d_shader_buffer
*buffer
,
4796 DWORD argnum
, unsigned int stage
, DWORD arg
)
4800 if (arg
== ARG_UNUSED
)
4801 return "<unused arg>";
4803 switch (arg
& WINED3DTA_SELECTMASK
)
4805 case WINED3DTA_DIFFUSE
:
4809 case WINED3DTA_CURRENT
:
4816 case WINED3DTA_TEXTURE
:
4819 case 0: ret
= "tex0"; break;
4820 case 1: ret
= "tex1"; break;
4821 case 2: ret
= "tex2"; break;
4822 case 3: ret
= "tex3"; break;
4823 case 4: ret
= "tex4"; break;
4824 case 5: ret
= "tex5"; break;
4825 case 6: ret
= "tex6"; break;
4826 case 7: ret
= "tex7"; break;
4828 ret
= "<invalid texture>";
4833 case WINED3DTA_TFACTOR
:
4837 case WINED3DTA_SPECULAR
:
4838 ret
= "gl_SecondaryColor";
4841 case WINED3DTA_TEMP
:
4845 case WINED3DTA_CONSTANT
:
4846 FIXME("Per-stage constants not implemented.\n");
4849 case 0: ret
= "const0"; break;
4850 case 1: ret
= "const1"; break;
4851 case 2: ret
= "const2"; break;
4852 case 3: ret
= "const3"; break;
4853 case 4: ret
= "const4"; break;
4854 case 5: ret
= "const5"; break;
4855 case 6: ret
= "const6"; break;
4856 case 7: ret
= "const7"; break;
4858 ret
= "<invalid constant>";
4864 return "<unhandled arg>";
4867 if (arg
& WINED3DTA_COMPLEMENT
)
4869 shader_addline(buffer
, "arg%u = vec4(1.0) - %s;\n", argnum
, ret
);
4872 else if (argnum
== 1)
4874 else if (argnum
== 2)
4878 if (arg
& WINED3DTA_ALPHAREPLICATE
)
4880 shader_addline(buffer
, "arg%u = vec4(%s.w);\n", argnum
, ret
);
4883 else if (argnum
== 1)
4885 else if (argnum
== 2)
4892 static void shader_glsl_ffp_fragment_op(struct wined3d_shader_buffer
*buffer
, unsigned int stage
, BOOL color
,
4893 BOOL alpha
, DWORD dst
, DWORD op
, DWORD dw_arg0
, DWORD dw_arg1
, DWORD dw_arg2
)
4895 const char *dstmask
, *dstreg
, *arg0
, *arg1
, *arg2
;
4905 dstreg
= "temp_reg";
4909 arg0
= shader_glsl_get_ffp_fragment_op_arg(buffer
, 0, stage
, dw_arg0
);
4910 arg1
= shader_glsl_get_ffp_fragment_op_arg(buffer
, 1, stage
, dw_arg1
);
4911 arg2
= shader_glsl_get_ffp_fragment_op_arg(buffer
, 2, stage
, dw_arg2
);
4915 case WINED3D_TOP_DISABLE
:
4917 shader_addline(buffer
, "%s%s = gl_Color%s;\n", dstreg
, dstmask
, dstmask
);
4920 case WINED3D_TOP_SELECT_ARG1
:
4921 shader_addline(buffer
, "%s%s = %s%s;\n", dstreg
, dstmask
, arg1
, dstmask
);
4924 case WINED3D_TOP_SELECT_ARG2
:
4925 shader_addline(buffer
, "%s%s = %s%s;\n", dstreg
, dstmask
, arg2
, dstmask
);
4928 case WINED3D_TOP_MODULATE
:
4929 shader_addline(buffer
, "%s%s = %s%s * %s%s;\n", dstreg
, dstmask
, arg1
, dstmask
, arg2
, dstmask
);
4932 case WINED3D_TOP_MODULATE_4X
:
4933 shader_addline(buffer
, "%s%s = clamp(%s%s * %s%s * 4.0, 0.0, 1.0);\n",
4934 dstreg
, dstmask
, arg1
, dstmask
, arg2
, dstmask
);
4937 case WINED3D_TOP_MODULATE_2X
:
4938 shader_addline(buffer
, "%s%s = clamp(%s%s * %s%s * 2.0, 0.0, 1.0);\n",
4939 dstreg
, dstmask
, arg1
, dstmask
, arg2
, dstmask
);
4942 case WINED3D_TOP_ADD
:
4943 shader_addline(buffer
, "%s%s = clamp(%s%s + %s%s, 0.0, 1.0);\n",
4944 dstreg
, dstmask
, arg1
, dstmask
, arg2
, dstmask
);
4947 case WINED3D_TOP_ADD_SIGNED
:
4948 shader_addline(buffer
, "%s%s = clamp(%s%s + (%s - vec4(0.5))%s, 0.0, 1.0);\n",
4949 dstreg
, dstmask
, arg1
, dstmask
, arg2
, dstmask
);
4952 case WINED3D_TOP_ADD_SIGNED_2X
:
4953 shader_addline(buffer
, "%s%s = clamp((%s%s + (%s - vec4(0.5))%s) * 2.0, 0.0, 1.0);\n",
4954 dstreg
, dstmask
, arg1
, dstmask
, arg2
, dstmask
);
4957 case WINED3D_TOP_SUBTRACT
:
4958 shader_addline(buffer
, "%s%s = clamp(%s%s - %s%s, 0.0, 1.0);\n",
4959 dstreg
, dstmask
, arg1
, dstmask
, arg2
, dstmask
);
4962 case WINED3D_TOP_ADD_SMOOTH
:
4963 shader_addline(buffer
, "%s%s = clamp((vec4(1.0) - %s)%s * %s%s + %s%s, 0.0, 1.0);\n",
4964 dstreg
, dstmask
, arg1
, dstmask
, arg2
, dstmask
, arg1
, dstmask
);
4967 case WINED3D_TOP_BLEND_DIFFUSE_ALPHA
:
4968 arg0
= shader_glsl_get_ffp_fragment_op_arg(buffer
, 0, stage
, WINED3DTA_DIFFUSE
);
4969 shader_addline(buffer
, "%s%s = mix(%s%s, %s%s, %s.w);\n",
4970 dstreg
, dstmask
, arg2
, dstmask
, arg1
, dstmask
, arg0
);
4973 case WINED3D_TOP_BLEND_TEXTURE_ALPHA
:
4974 arg0
= shader_glsl_get_ffp_fragment_op_arg(buffer
, 0, stage
, WINED3DTA_TEXTURE
);
4975 shader_addline(buffer
, "%s%s = mix(%s%s, %s%s, %s.w);\n",
4976 dstreg
, dstmask
, arg2
, dstmask
, arg1
, dstmask
, arg0
);
4979 case WINED3D_TOP_BLEND_FACTOR_ALPHA
:
4980 arg0
= shader_glsl_get_ffp_fragment_op_arg(buffer
, 0, stage
, WINED3DTA_TFACTOR
);
4981 shader_addline(buffer
, "%s%s = mix(%s%s, %s%s, %s.w);\n",
4982 dstreg
, dstmask
, arg2
, dstmask
, arg1
, dstmask
, arg0
);
4985 case WINED3D_TOP_BLEND_TEXTURE_ALPHA_PM
:
4986 arg0
= shader_glsl_get_ffp_fragment_op_arg(buffer
, 0, stage
, WINED3DTA_TEXTURE
);
4987 shader_addline(buffer
, "%s%s = clamp(%s%s * (1.0 - %s.w) + %s%s, 0.0, 1.0);\n",
4988 dstreg
, dstmask
, arg2
, dstmask
, arg0
, arg1
, dstmask
);
4991 case WINED3D_TOP_BLEND_CURRENT_ALPHA
:
4992 arg0
= shader_glsl_get_ffp_fragment_op_arg(buffer
, 0, stage
, WINED3DTA_CURRENT
);
4993 shader_addline(buffer
, "%s%s = mix(%s%s, %s%s, %s.w);\n",
4994 dstreg
, dstmask
, arg2
, dstmask
, arg1
, dstmask
, arg0
);
4997 case WINED3D_TOP_MODULATE_ALPHA_ADD_COLOR
:
4998 shader_addline(buffer
, "%s%s = clamp(%s%s * %s.w + %s%s, 0.0, 1.0);\n",
4999 dstreg
, dstmask
, arg2
, dstmask
, arg1
, arg1
, dstmask
);
5002 case WINED3D_TOP_MODULATE_COLOR_ADD_ALPHA
:
5003 shader_addline(buffer
, "%s%s = clamp(%s%s * %s%s + %s.w, 0.0, 1.0);\n",
5004 dstreg
, dstmask
, arg1
, dstmask
, arg2
, dstmask
, arg1
);
5007 case WINED3D_TOP_MODULATE_INVALPHA_ADD_COLOR
:
5008 shader_addline(buffer
, "%s%s = clamp(%s%s * (1.0 - %s.w) + %s%s, 0.0, 1.0);\n",
5009 dstreg
, dstmask
, arg2
, dstmask
, arg1
, arg1
, dstmask
);
5011 case WINED3D_TOP_MODULATE_INVCOLOR_ADD_ALPHA
:
5012 shader_addline(buffer
, "%s%s = clamp((vec4(1.0) - %s)%s * %s%s + %s.w, 0.0, 1.0);\n",
5013 dstreg
, dstmask
, arg1
, dstmask
, arg2
, dstmask
, arg1
);
5016 case WINED3D_TOP_BUMPENVMAP
:
5017 case WINED3D_TOP_BUMPENVMAP_LUMINANCE
:
5018 /* These are handled in the first pass, nothing to do. */
5021 case WINED3D_TOP_DOTPRODUCT3
:
5022 shader_addline(buffer
, "%s%s = vec4(clamp(dot(%s.xyz - 0.5, %s.xyz - 0.5) * 4.0, 0.0, 1.0))%s;\n",
5023 dstreg
, dstmask
, arg1
, arg2
, dstmask
);
5026 case WINED3D_TOP_MULTIPLY_ADD
:
5027 shader_addline(buffer
, "%s%s = clamp(%s%s * %s%s + %s%s, 0.0, 1.0);\n",
5028 dstreg
, dstmask
, arg1
, dstmask
, arg2
, dstmask
, arg0
, dstmask
);
5031 case WINED3D_TOP_LERP
:
5032 /* MSDN isn't quite right here. */
5033 shader_addline(buffer
, "%s%s = mix(%s%s, %s%s, %s%s);\n",
5034 dstreg
, dstmask
, arg2
, dstmask
, arg1
, dstmask
, arg0
, dstmask
);
5038 FIXME("Unhandled operation %#x.\n", op
);
5043 /* Context activation is done by the caller. */
5044 static GLuint
shader_glsl_generate_ffp_fragment_shader(struct wined3d_shader_buffer
*buffer
,
5045 const struct ffp_frag_settings
*settings
, const struct wined3d_gl_info
*gl_info
)
5047 BOOL tempreg_used
= FALSE
, tfactor_used
= FALSE
;
5048 BYTE lum_map
= 0, bump_map
= 0, tex_map
= 0;
5049 const char *final_combiner_src
= "ret";
5050 UINT lowest_disabled_stage
;
5051 GLhandleARB shader_obj
;
5052 DWORD arg0
, arg1
, arg2
;
5055 shader_buffer_clear(buffer
);
5057 /* Find out which textures are read */
5058 for (stage
= 0; stage
< MAX_TEXTURES
; ++stage
)
5060 if (settings
->op
[stage
].cop
== WINED3D_TOP_DISABLE
)
5063 arg0
= settings
->op
[stage
].carg0
& WINED3DTA_SELECTMASK
;
5064 arg1
= settings
->op
[stage
].carg1
& WINED3DTA_SELECTMASK
;
5065 arg2
= settings
->op
[stage
].carg2
& WINED3DTA_SELECTMASK
;
5067 if (arg0
== WINED3DTA_TEXTURE
|| arg1
== WINED3DTA_TEXTURE
|| arg2
== WINED3DTA_TEXTURE
)
5068 tex_map
|= 1 << stage
;
5069 if (arg0
== WINED3DTA_TFACTOR
|| arg1
== WINED3DTA_TFACTOR
|| arg2
== WINED3DTA_TFACTOR
)
5070 tfactor_used
= TRUE
;
5071 if (arg0
== WINED3DTA_TEMP
|| arg1
== WINED3DTA_TEMP
|| arg2
== WINED3DTA_TEMP
)
5072 tempreg_used
= TRUE
;
5073 if (settings
->op
[stage
].dst
== tempreg
)
5074 tempreg_used
= TRUE
;
5076 switch (settings
->op
[stage
].cop
)
5078 case WINED3D_TOP_BUMPENVMAP_LUMINANCE
:
5079 lum_map
|= 1 << stage
;
5081 case WINED3D_TOP_BUMPENVMAP
:
5082 bump_map
|= 1 << stage
;
5084 case WINED3D_TOP_BLEND_TEXTURE_ALPHA
:
5085 case WINED3D_TOP_BLEND_TEXTURE_ALPHA_PM
:
5086 tex_map
|= 1 << stage
;
5089 case WINED3D_TOP_BLEND_FACTOR_ALPHA
:
5090 tfactor_used
= TRUE
;
5097 if (settings
->op
[stage
].aop
== WINED3D_TOP_DISABLE
)
5100 arg0
= settings
->op
[stage
].aarg0
& WINED3DTA_SELECTMASK
;
5101 arg1
= settings
->op
[stage
].aarg1
& WINED3DTA_SELECTMASK
;
5102 arg2
= settings
->op
[stage
].aarg2
& WINED3DTA_SELECTMASK
;
5104 if (arg0
== WINED3DTA_TEXTURE
|| arg1
== WINED3DTA_TEXTURE
|| arg2
== WINED3DTA_TEXTURE
)
5105 tex_map
|= 1 << stage
;
5106 if (arg0
== WINED3DTA_TFACTOR
|| arg1
== WINED3DTA_TFACTOR
|| arg2
== WINED3DTA_TFACTOR
)
5107 tfactor_used
= TRUE
;
5108 if (arg0
== WINED3DTA_TEMP
|| arg1
== WINED3DTA_TEMP
|| arg2
== WINED3DTA_TEMP
)
5109 tempreg_used
= TRUE
;
5111 lowest_disabled_stage
= stage
;
5113 shader_addline(buffer
, "#version 120\n");
5115 if (gl_info
->supported
[ARB_TEXTURE_RECTANGLE
])
5116 shader_addline(buffer
, "#extension GL_ARB_texture_rectangle : enable\n");
5118 shader_addline(buffer
, "vec4 tmp0, tmp1;\n");
5119 shader_addline(buffer
, "vec4 ret;\n");
5120 if (tempreg_used
|| settings
->sRGB_write
)
5121 shader_addline(buffer
, "vec4 temp_reg;\n");
5122 shader_addline(buffer
, "vec4 arg0, arg1, arg2;\n");
5124 for (stage
= 0; stage
< MAX_TEXTURES
; ++stage
)
5126 if (!(tex_map
& (1 << stage
)))
5129 switch (settings
->op
[stage
].tex_type
)
5132 shader_addline(buffer
, "uniform sampler1D ps_sampler%u;\n", stage
);
5135 shader_addline(buffer
, "uniform sampler2D ps_sampler%u;\n", stage
);
5138 shader_addline(buffer
, "uniform sampler3D ps_sampler%u;\n", stage
);
5141 shader_addline(buffer
, "uniform samplerCube ps_sampler%u;\n", stage
);
5144 shader_addline(buffer
, "uniform sampler2DRect ps_sampler%u;\n", stage
);
5147 FIXME("Unhandled sampler type %#x.\n", settings
->op
[stage
].tex_type
);
5151 shader_addline(buffer
, "vec4 tex%u;\n", stage
);
5153 if (!(bump_map
& (1 << stage
)))
5155 shader_addline(buffer
, "uniform mat2 bumpenv_mat%u;\n", stage
);
5157 if (!(lum_map
& (1 << stage
)))
5159 shader_addline(buffer
, "uniform float bumpenv_lum_scale%u;\n", stage
);
5160 shader_addline(buffer
, "uniform float bumpenv_lum_offset%u;\n", stage
);
5163 shader_addline(buffer
, "uniform vec4 tex_factor;\n");
5164 shader_addline(buffer
, "uniform vec4 specular_enable;\n");
5166 if (settings
->sRGB_write
)
5168 shader_addline(buffer
, "const vec4 srgb_const0 = vec4(%.8e, %.8e, %.8e, %.8e);\n",
5169 srgb_pow
, srgb_mul_high
, srgb_sub_high
, srgb_mul_low
);
5170 shader_addline(buffer
, "const vec4 srgb_const1 = vec4(%.8e, 0.0, 0.0, 0.0);\n",
5174 shader_addline(buffer
, "void main()\n{\n");
5176 if (lowest_disabled_stage
< 7 && settings
->emul_clipplanes
)
5177 shader_addline(buffer
, "if (any(lessThan(gl_texCoord[7], vec4(0.0)))) discard;\n");
5179 /* Generate texture sampling instructions) */
5180 for (stage
= 0; stage
< MAX_TEXTURES
&& settings
->op
[stage
].cop
!= WINED3D_TOP_DISABLE
; ++stage
)
5182 const char *texture_function
, *coord_mask
;
5183 char tex_reg_name
[8];
5186 if (!(tex_map
& (1 << stage
)))
5189 if (settings
->op
[stage
].projected
== proj_none
)
5193 else if (settings
->op
[stage
].projected
== proj_count4
5194 || settings
->op
[stage
].projected
== proj_count3
)
5200 FIXME("Unexpected projection mode %d\n", settings
->op
[stage
].projected
);
5204 if (settings
->op
[stage
].cop
== WINED3D_TOP_BUMPENVMAP
5205 || settings
->op
[stage
].cop
== WINED3D_TOP_BUMPENVMAP_LUMINANCE
)
5210 switch (settings
->op
[stage
].tex_type
)
5215 texture_function
= "texture1DProj";
5220 texture_function
= "texture1D";
5227 texture_function
= "texture2DProj";
5232 texture_function
= "texture2D";
5239 texture_function
= "texture3DProj";
5240 coord_mask
= "xyzw";
5244 texture_function
= "texture3D";
5249 texture_function
= "textureCube";
5255 texture_function
= "texture2DRectProj";
5260 texture_function
= "texture2DRect";
5265 FIXME("Unhandled texture type %#x.\n", settings
->op
[stage
].tex_type
);
5266 texture_function
= "";
5267 coord_mask
= "xyzw";
5272 && (settings
->op
[stage
- 1].cop
== WINED3D_TOP_BUMPENVMAP
5273 || settings
->op
[stage
- 1].cop
== WINED3D_TOP_BUMPENVMAP_LUMINANCE
))
5275 shader_addline(buffer
, "ret.xy = bumpenv_mat%u * tex%u.xy;\n", stage
- 1, stage
- 1);
5277 /* With projective textures, texbem only divides the static
5278 * texture coord, not the displacement, so multiply the
5279 * displacement with the dividing parameter before passing it to
5281 if (settings
->op
[stage
].projected
!= proj_none
)
5283 if (settings
->op
[stage
].projected
== proj_count4
)
5285 shader_addline(buffer
, "ret.xy = (ret.xy * gl_TexCoord[%u].w) + gl_TexCoord[%u].xy;\n",
5287 shader_addline(buffer
, "ret.zw = gl_TexCoord[%u].ww;\n", stage
);
5291 shader_addline(buffer
, "ret.xy = (ret.xy * gl_TexCoord[%u].z) + gl_TexCoord[%u].xy;\n",
5293 shader_addline(buffer
, "ret.zw = gl_TexCoord[%u].zz;\n", stage
);
5298 shader_addline(buffer
, "ret = gl_TexCoord[%u] + ret.xyxy;\n", stage
);
5302 shader_addline(buffer
, "tex%u = clamp(%s(ps_sampler%u, ret.%s), 0.0, 1.0);\n",
5303 stage
, texture_function
, stage
, coord_mask
);
5305 shader_addline(buffer
, "tex%u = %s(ps_sampler%u, ret.%s);\n",
5306 stage
, texture_function
, stage
, coord_mask
);
5308 if (settings
->op
[stage
- 1].cop
== WINED3D_TOP_BUMPENVMAP_LUMINANCE
)
5309 shader_addline(buffer
, "tex%u *= clamp(tex%u.z * bumpenv_lum_scale%u + bumpenv_lum_offset%u, 0.0, 1.0);\n",
5310 stage
, stage
- 1, stage
- 1, stage
- 1);
5312 else if (settings
->op
[stage
].projected
== proj_count3
)
5315 shader_addline(buffer
, "tex%u = clamp(%s(ps_sampler%u, gl_TexCoord[%u].xyz), 0.0, 1.0);\n",
5316 stage
, texture_function
, stage
, stage
);
5318 shader_addline(buffer
, "tex%u = %s(ps_sampler%u, gl_TexCoord[%u].xyz);\n",
5319 stage
, texture_function
, stage
, stage
);
5324 shader_addline(buffer
, "tex%u = clamp(%s(ps_sampler%u, gl_TexCoord[%u].%s), 0.0, 1.0);\n",
5325 stage
, texture_function
, stage
, stage
, coord_mask
);
5327 shader_addline(buffer
, "tex%u = %s(ps_sampler%u, gl_TexCoord[%u].%s);\n",
5328 stage
, texture_function
, stage
, stage
, coord_mask
);
5331 sprintf(tex_reg_name
, "tex%u", stage
);
5332 shader_glsl_color_correction_ext(buffer
, tex_reg_name
, WINED3DSP_WRITEMASK_ALL
,
5333 settings
->op
[stage
].color_fixup
);
5336 /* Generate the main shader */
5337 for (stage
= 0; stage
< MAX_TEXTURES
; ++stage
)
5341 if (settings
->op
[stage
].cop
== WINED3D_TOP_DISABLE
)
5344 final_combiner_src
= "gl_Color";
5348 if (settings
->op
[stage
].cop
== WINED3D_TOP_SELECT_ARG1
5349 && settings
->op
[stage
].aop
== WINED3D_TOP_SELECT_ARG1
)
5350 op_equal
= settings
->op
[stage
].carg1
== settings
->op
[stage
].aarg1
;
5351 else if (settings
->op
[stage
].cop
== WINED3D_TOP_SELECT_ARG1
5352 && settings
->op
[stage
].aop
== WINED3D_TOP_SELECT_ARG2
)
5353 op_equal
= settings
->op
[stage
].carg1
== settings
->op
[stage
].aarg2
;
5354 else if (settings
->op
[stage
].cop
== WINED3D_TOP_SELECT_ARG2
5355 && settings
->op
[stage
].aop
== WINED3D_TOP_SELECT_ARG1
)
5356 op_equal
= settings
->op
[stage
].carg2
== settings
->op
[stage
].aarg1
;
5357 else if (settings
->op
[stage
].cop
== WINED3D_TOP_SELECT_ARG2
5358 && settings
->op
[stage
].aop
== WINED3D_TOP_SELECT_ARG2
)
5359 op_equal
= settings
->op
[stage
].carg2
== settings
->op
[stage
].aarg2
;
5361 op_equal
= settings
->op
[stage
].aop
== settings
->op
[stage
].cop
5362 && settings
->op
[stage
].carg0
== settings
->op
[stage
].aarg0
5363 && settings
->op
[stage
].carg1
== settings
->op
[stage
].aarg1
5364 && settings
->op
[stage
].carg2
== settings
->op
[stage
].aarg2
;
5366 if (settings
->op
[stage
].aop
== WINED3D_TOP_DISABLE
)
5368 shader_glsl_ffp_fragment_op(buffer
, stage
, TRUE
, FALSE
, settings
->op
[stage
].dst
,
5369 settings
->op
[stage
].cop
, settings
->op
[stage
].carg0
,
5370 settings
->op
[stage
].carg1
, settings
->op
[stage
].carg2
);
5372 shader_addline(buffer
, "ret.w = gl_Color.w;\n");
5376 shader_glsl_ffp_fragment_op(buffer
, stage
, TRUE
, TRUE
, settings
->op
[stage
].dst
,
5377 settings
->op
[stage
].cop
, settings
->op
[stage
].carg0
,
5378 settings
->op
[stage
].carg1
, settings
->op
[stage
].carg2
);
5382 shader_glsl_ffp_fragment_op(buffer
, stage
, TRUE
, FALSE
, settings
->op
[stage
].dst
,
5383 settings
->op
[stage
].cop
, settings
->op
[stage
].carg0
,
5384 settings
->op
[stage
].carg1
, settings
->op
[stage
].carg2
);
5385 shader_glsl_ffp_fragment_op(buffer
, stage
, FALSE
, TRUE
, settings
->op
[stage
].dst
,
5386 settings
->op
[stage
].aop
, settings
->op
[stage
].aarg0
,
5387 settings
->op
[stage
].aarg1
, settings
->op
[stage
].aarg2
);
5391 shader_addline(buffer
, "gl_FragData[0] = gl_SecondaryColor * specular_enable + %s;\n", final_combiner_src
);
5393 if (settings
->sRGB_write
)
5394 shader_glsl_generate_srgb_write_correction(buffer
);
5396 shader_glsl_generate_fog_code(buffer
, settings
->fog
);
5398 shader_addline(buffer
, "}\n");
5400 shader_obj
= GL_EXTCALL(glCreateShaderObjectARB(GL_FRAGMENT_SHADER_ARB
));
5401 shader_glsl_compile(gl_info
, shader_obj
, buffer
->buffer
);
5405 static struct glsl_ffp_fragment_shader
*shader_glsl_find_ffp_fragment_shader(struct shader_glsl_priv
*priv
,
5406 const struct wined3d_gl_info
*gl_info
, const struct ffp_frag_settings
*args
)
5408 struct glsl_ffp_fragment_shader
*glsl_desc
;
5409 const struct ffp_frag_desc
*desc
;
5411 if ((desc
= find_ffp_frag_shader(&priv
->ffp_fragment_shaders
, args
)))
5412 return CONTAINING_RECORD(desc
, struct glsl_ffp_fragment_shader
, entry
);
5414 if (!(glsl_desc
= HeapAlloc(GetProcessHeap(), 0, sizeof(*glsl_desc
))))
5417 glsl_desc
->entry
.settings
= *args
;
5418 glsl_desc
->id
= shader_glsl_generate_ffp_fragment_shader(&priv
->shader_buffer
, args
, gl_info
);
5419 list_init(&glsl_desc
->linked_programs
);
5420 add_ffp_frag_shader(&priv
->ffp_fragment_shaders
, &glsl_desc
->entry
);
5426 static void shader_glsl_init_vs_uniform_locations(const struct wined3d_gl_info
*gl_info
,
5427 GLhandleARB program_id
, struct glsl_vs_program
*vs
)
5432 vs
->uniform_f_locations
= HeapAlloc(GetProcessHeap(), 0,
5433 sizeof(GLhandleARB
) * gl_info
->limits
.glsl_vs_float_constants
);
5434 for (i
= 0; i
< gl_info
->limits
.glsl_vs_float_constants
; ++i
)
5436 snprintf(name
, sizeof(name
), "vs_c[%u]", i
);
5437 vs
->uniform_f_locations
[i
] = GL_EXTCALL(glGetUniformLocationARB(program_id
, name
));
5440 for (i
= 0; i
< MAX_CONST_I
; ++i
)
5442 snprintf(name
, sizeof(name
), "vs_i[%u]", i
);
5443 vs
->uniform_i_locations
[i
] = GL_EXTCALL(glGetUniformLocationARB(program_id
, name
));
5446 vs
->pos_fixup_location
= GL_EXTCALL(glGetUniformLocationARB(program_id
, "posFixup"));
5449 static void shader_glsl_init_ps_uniform_locations(const struct wined3d_gl_info
*gl_info
,
5450 GLhandleARB program_id
, struct glsl_ps_program
*ps
)
5455 ps
->uniform_f_locations
= HeapAlloc(GetProcessHeap(), 0,
5456 sizeof(GLhandleARB
) * gl_info
->limits
.glsl_ps_float_constants
);
5457 for (i
= 0; i
< gl_info
->limits
.glsl_ps_float_constants
; ++i
)
5459 snprintf(name
, sizeof(name
), "ps_c[%u]", i
);
5460 ps
->uniform_f_locations
[i
] = GL_EXTCALL(glGetUniformLocationARB(program_id
, name
));
5463 for (i
= 0; i
< MAX_CONST_I
; ++i
)
5465 snprintf(name
, sizeof(name
), "ps_i[%u]", i
);
5466 ps
->uniform_i_locations
[i
] = GL_EXTCALL(glGetUniformLocationARB(program_id
, name
));
5469 for (i
= 0; i
< MAX_TEXTURES
; ++i
)
5471 snprintf(name
, sizeof(name
), "bumpenv_mat%u", i
);
5472 ps
->bumpenv_mat_location
[i
] = GL_EXTCALL(glGetUniformLocationARB(program_id
, name
));
5473 snprintf(name
, sizeof(name
), "bumpenv_lum_scale%u", i
);
5474 ps
->bumpenv_lum_scale_location
[i
] = GL_EXTCALL(glGetUniformLocationARB(program_id
, name
));
5475 snprintf(name
, sizeof(name
), "bumpenv_lum_offset%u", i
);
5476 ps
->bumpenv_lum_offset_location
[i
] = GL_EXTCALL(glGetUniformLocationARB(program_id
, name
));
5479 ps
->tex_factor_location
= GL_EXTCALL(glGetUniformLocationARB(program_id
, "tex_factor"));
5480 ps
->specular_enable_location
= GL_EXTCALL(glGetUniformLocationARB(program_id
, "specular_enable"));
5481 ps
->np2_fixup_location
= GL_EXTCALL(glGetUniformLocationARB(program_id
, "ps_samplerNP2Fixup"));
5482 ps
->ycorrection_location
= GL_EXTCALL(glGetUniformLocationARB(program_id
, "ycorrection"));
5485 /* Context activation is done by the caller. */
5486 static void set_glsl_shader_program(const struct wined3d_context
*context
, struct wined3d_device
*device
,
5487 enum wined3d_shader_mode vertex_mode
, enum wined3d_shader_mode fragment_mode
)
5489 const struct wined3d_state
*state
= &device
->stateBlock
->state
;
5490 const struct wined3d_gl_info
*gl_info
= context
->gl_info
;
5491 const struct ps_np2fixup_info
*np2fixup_info
= NULL
;
5492 struct shader_glsl_priv
*priv
= device
->shader_priv
;
5493 struct glsl_shader_prog_link
*entry
= NULL
;
5494 struct wined3d_shader
*vshader
= NULL
;
5495 struct wined3d_shader
*gshader
= NULL
;
5496 struct wined3d_shader
*pshader
= NULL
;
5497 GLhandleARB programId
= 0;
5498 GLhandleARB reorder_shader_id
= 0;
5500 struct ps_compile_args ps_compile_args
;
5501 struct vs_compile_args vs_compile_args
;
5502 GLhandleARB vs_id
, gs_id
, ps_id
;
5503 struct list
*ps_list
;
5505 if (vertex_mode
== WINED3D_SHADER_MODE_SHADER
)
5507 vshader
= state
->vertex_shader
;
5508 find_vs_compile_args(state
, vshader
, &vs_compile_args
);
5509 vs_id
= find_glsl_vshader(context
, &priv
->shader_buffer
, vshader
, &vs_compile_args
);
5511 if ((gshader
= state
->geometry_shader
))
5512 gs_id
= find_glsl_geometry_shader(context
, &priv
->shader_buffer
, gshader
);
5522 if (fragment_mode
== WINED3D_SHADER_MODE_SHADER
)
5524 pshader
= state
->pixel_shader
;
5525 find_ps_compile_args(state
, pshader
, &ps_compile_args
);
5526 ps_id
= find_glsl_pshader(context
, &priv
->shader_buffer
,
5527 pshader
, &ps_compile_args
, &np2fixup_info
);
5528 ps_list
= &pshader
->linked_programs
;
5530 else if (fragment_mode
== WINED3D_SHADER_MODE_FFP
&& priv
->fragment_pipe
== &glsl_fragment_pipe
)
5532 struct glsl_ffp_fragment_shader
*ffp_shader
;
5533 struct ffp_frag_settings settings
;
5535 gen_ffp_frag_op(context
, state
, &settings
, FALSE
);
5536 ffp_shader
= shader_glsl_find_ffp_fragment_shader(priv
, gl_info
, &settings
);
5537 ps_id
= ffp_shader
->id
;
5538 ps_list
= &ffp_shader
->linked_programs
;
5545 if ((!vs_id
&& !gs_id
&& !ps_id
) || (entry
= get_glsl_program_entry(priv
, vs_id
, gs_id
, ps_id
)))
5547 priv
->glsl_program
= entry
;
5551 /* If we get to this point, then no matching program exists, so we create one */
5552 programId
= GL_EXTCALL(glCreateProgramObjectARB());
5553 TRACE("Created new GLSL shader program %u\n", programId
);
5555 /* Create the entry */
5556 entry
= HeapAlloc(GetProcessHeap(), 0, sizeof(struct glsl_shader_prog_link
));
5557 entry
->programId
= programId
;
5558 entry
->vs
.id
= vs_id
;
5559 entry
->gs
.id
= gs_id
;
5560 entry
->ps
.id
= ps_id
;
5561 entry
->constant_version
= 0;
5562 entry
->ps
.np2_fixup_info
= np2fixup_info
;
5563 /* Add the hash table entry */
5564 add_glsl_program_entry(priv
, entry
);
5566 /* Set the current program */
5567 priv
->glsl_program
= entry
;
5569 /* Attach GLSL vshader */
5572 WORD map
= vshader
->reg_maps
.input_registers
;
5575 reorder_shader_id
= generate_param_reorder_function(&priv
->shader_buffer
, vshader
, pshader
, gl_info
);
5576 TRACE("Attaching GLSL shader object %u to program %u\n", reorder_shader_id
, programId
);
5577 GL_EXTCALL(glAttachObjectARB(programId
, reorder_shader_id
));
5578 checkGLcall("glAttachObjectARB");
5579 /* Flag the reorder function for deletion, then it will be freed automatically when the program
5582 GL_EXTCALL(glDeleteObjectARB(reorder_shader_id
));
5584 TRACE("Attaching GLSL shader object %u to program %u.\n", vs_id
, programId
);
5585 GL_EXTCALL(glAttachObjectARB(programId
, vs_id
));
5586 checkGLcall("glAttachObjectARB");
5588 /* Bind vertex attributes to a corresponding index number to match
5589 * the same index numbers as ARB_vertex_programs (makes loading
5590 * vertex attributes simpler). With this method, we can use the
5591 * exact same code to load the attributes later for both ARB and
5594 * We have to do this here because we need to know the Program ID
5595 * in order to make the bindings work, and it has to be done prior
5596 * to linking the GLSL program. */
5597 for (i
= 0; map
; map
>>= 1, ++i
)
5599 if (!(map
& 1)) continue;
5601 snprintf(tmp_name
, sizeof(tmp_name
), "vs_in%u", i
);
5602 GL_EXTCALL(glBindAttribLocationARB(programId
, i
, tmp_name
));
5604 checkGLcall("glBindAttribLocationARB");
5606 list_add_head(&vshader
->linked_programs
, &entry
->vs
.shader_entry
);
5611 TRACE("Attaching GLSL geometry shader object %u to program %u.\n", gs_id
, programId
);
5612 GL_EXTCALL(glAttachObjectARB(programId
, gs_id
));
5613 checkGLcall("glAttachObjectARB");
5615 TRACE("input type %s, output type %s, vertices out %u.\n",
5616 debug_d3dprimitivetype(gshader
->u
.gs
.input_type
),
5617 debug_d3dprimitivetype(gshader
->u
.gs
.output_type
),
5618 gshader
->u
.gs
.vertices_out
);
5619 GL_EXTCALL(glProgramParameteriARB(programId
, GL_GEOMETRY_INPUT_TYPE_ARB
,
5620 gl_primitive_type_from_d3d(gshader
->u
.gs
.input_type
)));
5621 GL_EXTCALL(glProgramParameteriARB(programId
, GL_GEOMETRY_OUTPUT_TYPE_ARB
,
5622 gl_primitive_type_from_d3d(gshader
->u
.gs
.output_type
)));
5623 GL_EXTCALL(glProgramParameteriARB(programId
, GL_GEOMETRY_VERTICES_OUT_ARB
,
5624 gshader
->u
.gs
.vertices_out
));
5625 checkGLcall("glProgramParameteriARB");
5627 list_add_head(&gshader
->linked_programs
, &entry
->gs
.shader_entry
);
5630 /* Attach GLSL pshader */
5633 TRACE("Attaching GLSL shader object %u to program %u.\n", ps_id
, programId
);
5634 GL_EXTCALL(glAttachObjectARB(programId
, ps_id
));
5635 checkGLcall("glAttachObjectARB");
5637 list_add_head(ps_list
, &entry
->ps
.shader_entry
);
5640 /* Link the program */
5641 TRACE("Linking GLSL shader program %u\n", programId
);
5642 GL_EXTCALL(glLinkProgramARB(programId
));
5643 shader_glsl_validate_link(gl_info
, programId
);
5645 shader_glsl_init_vs_uniform_locations(gl_info
, programId
, &entry
->vs
);
5646 shader_glsl_init_ps_uniform_locations(gl_info
, programId
, &entry
->ps
);
5647 checkGLcall("Find glsl program uniform locations");
5649 if (pshader
&& pshader
->reg_maps
.shader_version
.major
>= 3
5650 && pshader
->u
.ps
.declared_in_count
> vec4_varyings(3, gl_info
))
5652 TRACE("Shader %d needs vertex color clamping disabled\n", programId
);
5653 entry
->vs
.vertex_color_clamp
= GL_FALSE
;
5657 entry
->vs
.vertex_color_clamp
= GL_FIXED_ONLY_ARB
;
5660 /* Set the shader to allow uniform loading on it */
5661 GL_EXTCALL(glUseProgramObjectARB(programId
));
5662 checkGLcall("glUseProgramObjectARB(programId)");
5664 /* Load the vertex and pixel samplers now. The function that finds the mappings makes sure
5665 * that it stays the same for each vertexshader-pixelshader pair(=linked glsl program). If
5666 * a pshader with fixed function pipeline is used there are no vertex samplers, and if a
5667 * vertex shader with fixed function pixel processing is used we make sure that the card
5668 * supports enough samplers to allow the max number of vertex samplers with all possible
5669 * fixed function fragment processing setups. So once the program is linked these samplers
5672 shader_glsl_load_vsamplers(gl_info
, device
->texUnitMap
, programId
);
5673 shader_glsl_load_psamplers(gl_info
, device
->texUnitMap
, programId
);
5676 /* Context activation is done by the caller. */
5677 static GLhandleARB
create_glsl_blt_shader(const struct wined3d_gl_info
*gl_info
, enum tex_types tex_type
, BOOL masked
)
5679 GLhandleARB program_id
;
5680 GLhandleARB vshader_id
, pshader_id
;
5681 const char *blt_pshader
;
5683 static const char *blt_vshader
=
5687 " gl_Position = gl_Vertex;\n"
5688 " gl_FrontColor = vec4(1.0);\n"
5689 " gl_TexCoord[0] = gl_MultiTexCoord0;\n"
5692 static const char * const blt_pshaders_full
[tex_type_count
] =
5698 "uniform sampler2D sampler;\n"
5701 " gl_FragDepth = texture2D(sampler, gl_TexCoord[0].xy).x;\n"
5707 "uniform samplerCube sampler;\n"
5710 " gl_FragDepth = textureCube(sampler, gl_TexCoord[0].xyz).x;\n"
5714 "#extension GL_ARB_texture_rectangle : enable\n"
5715 "uniform sampler2DRect sampler;\n"
5718 " gl_FragDepth = texture2DRect(sampler, gl_TexCoord[0].xy).x;\n"
5722 static const char * const blt_pshaders_masked
[tex_type_count
] =
5728 "uniform sampler2D sampler;\n"
5729 "uniform vec4 mask;\n"
5732 " if (all(lessThan(gl_FragCoord.xy, mask.zw))) discard;\n"
5733 " gl_FragDepth = texture2D(sampler, gl_TexCoord[0].xy).x;\n"
5739 "uniform samplerCube sampler;\n"
5740 "uniform vec4 mask;\n"
5743 " if (all(lessThan(gl_FragCoord.xy, mask.zw))) discard;\n"
5744 " gl_FragDepth = textureCube(sampler, gl_TexCoord[0].xyz).x;\n"
5748 "#extension GL_ARB_texture_rectangle : enable\n"
5749 "uniform sampler2DRect sampler;\n"
5750 "uniform vec4 mask;\n"
5753 " if (all(lessThan(gl_FragCoord.xy, mask.zw))) discard;\n"
5754 " gl_FragDepth = texture2DRect(sampler, gl_TexCoord[0].xy).x;\n"
5758 blt_pshader
= masked
? blt_pshaders_masked
[tex_type
] : blt_pshaders_full
[tex_type
];
5761 FIXME("tex_type %#x not supported\n", tex_type
);
5765 vshader_id
= GL_EXTCALL(glCreateShaderObjectARB(GL_VERTEX_SHADER_ARB
));
5766 shader_glsl_compile(gl_info
, vshader_id
, blt_vshader
);
5768 pshader_id
= GL_EXTCALL(glCreateShaderObjectARB(GL_FRAGMENT_SHADER_ARB
));
5769 shader_glsl_compile(gl_info
, pshader_id
, blt_pshader
);
5771 program_id
= GL_EXTCALL(glCreateProgramObjectARB());
5772 GL_EXTCALL(glAttachObjectARB(program_id
, vshader_id
));
5773 GL_EXTCALL(glAttachObjectARB(program_id
, pshader_id
));
5774 GL_EXTCALL(glLinkProgramARB(program_id
));
5776 shader_glsl_validate_link(gl_info
, program_id
);
5778 /* Once linked we can mark the shaders for deletion. They will be deleted once the program
5781 GL_EXTCALL(glDeleteObjectARB(vshader_id
));
5782 GL_EXTCALL(glDeleteObjectARB(pshader_id
));
5786 /* Context activation is done by the caller. */
5787 static void shader_glsl_select(const struct wined3d_context
*context
, enum wined3d_shader_mode vertex_mode
,
5788 enum wined3d_shader_mode fragment_mode
)
5790 const struct wined3d_gl_info
*gl_info
= context
->gl_info
;
5791 struct wined3d_device
*device
= context
->swapchain
->device
;
5792 struct shader_glsl_priv
*priv
= device
->shader_priv
;
5793 GLhandleARB program_id
= 0;
5794 GLenum old_vertex_color_clamp
, current_vertex_color_clamp
;
5796 priv
->vertex_pipe
->vp_enable(gl_info
, vertex_mode
== WINED3D_SHADER_MODE_FFP
);
5797 priv
->fragment_pipe
->enable_extension(gl_info
, fragment_mode
== WINED3D_SHADER_MODE_FFP
);
5799 old_vertex_color_clamp
= priv
->glsl_program
? priv
->glsl_program
->vs
.vertex_color_clamp
: GL_FIXED_ONLY_ARB
;
5800 set_glsl_shader_program(context
, device
, vertex_mode
, fragment_mode
);
5801 current_vertex_color_clamp
= priv
->glsl_program
? priv
->glsl_program
->vs
.vertex_color_clamp
: GL_FIXED_ONLY_ARB
;
5802 if (old_vertex_color_clamp
!= current_vertex_color_clamp
)
5804 if (gl_info
->supported
[ARB_COLOR_BUFFER_FLOAT
])
5806 GL_EXTCALL(glClampColorARB(GL_CLAMP_VERTEX_COLOR_ARB
, current_vertex_color_clamp
));
5807 checkGLcall("glClampColorARB");
5811 FIXME("vertex color clamp needs to be changed, but extension not supported.\n");
5815 program_id
= priv
->glsl_program
? priv
->glsl_program
->programId
: 0;
5816 if (program_id
) TRACE("Using GLSL program %u\n", program_id
);
5817 GL_EXTCALL(glUseProgramObjectARB(program_id
));
5818 checkGLcall("glUseProgramObjectARB");
5820 /* In case that NP2 texcoord fixup data is found for the selected program, trigger a reload of the
5821 * constants. This has to be done because it can't be guaranteed that sampler() (from state.c) is
5822 * called between selecting the shader and using it, which results in wrong fixup for some frames. */
5823 if (priv
->glsl_program
&& priv
->glsl_program
->ps
.np2_fixup_info
)
5825 shader_glsl_load_np2fixup_constants(priv
, gl_info
, &device
->stateBlock
->state
);
5829 /* Context activation is done by the caller. */
5830 static void shader_glsl_select_depth_blt(void *shader_priv
, const struct wined3d_gl_info
*gl_info
,
5831 enum tex_types tex_type
, const SIZE
*ds_mask_size
)
5833 BOOL masked
= ds_mask_size
->cx
&& ds_mask_size
->cy
;
5834 struct shader_glsl_priv
*priv
= shader_priv
;
5835 GLhandleARB
*blt_program
;
5838 blt_program
= masked
? &priv
->depth_blt_program_masked
[tex_type
] : &priv
->depth_blt_program_full
[tex_type
];
5841 *blt_program
= create_glsl_blt_shader(gl_info
, tex_type
, masked
);
5842 loc
= GL_EXTCALL(glGetUniformLocationARB(*blt_program
, "sampler"));
5843 GL_EXTCALL(glUseProgramObjectARB(*blt_program
));
5844 GL_EXTCALL(glUniform1iARB(loc
, 0));
5848 GL_EXTCALL(glUseProgramObjectARB(*blt_program
));
5853 loc
= GL_EXTCALL(glGetUniformLocationARB(*blt_program
, "mask"));
5854 GL_EXTCALL(glUniform4fARB(loc
, 0.0f
, 0.0f
, (float)ds_mask_size
->cx
, (float)ds_mask_size
->cy
));
5858 /* Context activation is done by the caller. */
5859 static void shader_glsl_deselect_depth_blt(void *shader_priv
, const struct wined3d_gl_info
*gl_info
)
5861 struct shader_glsl_priv
*priv
= shader_priv
;
5862 GLhandleARB program_id
;
5864 program_id
= priv
->glsl_program
? priv
->glsl_program
->programId
: 0;
5865 if (program_id
) TRACE("Using GLSL program %u\n", program_id
);
5867 GL_EXTCALL(glUseProgramObjectARB(program_id
));
5868 checkGLcall("glUseProgramObjectARB");
5871 static void shader_glsl_destroy(struct wined3d_shader
*shader
)
5873 struct glsl_shader_private
*shader_data
= shader
->backend_data
;
5874 struct wined3d_device
*device
= shader
->device
;
5875 struct shader_glsl_priv
*priv
= device
->shader_priv
;
5876 const struct wined3d_gl_info
*gl_info
;
5877 const struct list
*linked_programs
;
5878 struct wined3d_context
*context
;
5880 if (!shader_data
|| !shader_data
->num_gl_shaders
)
5882 HeapFree(GetProcessHeap(), 0, shader_data
);
5883 shader
->backend_data
= NULL
;
5887 context
= context_acquire(device
, NULL
);
5888 gl_info
= context
->gl_info
;
5890 TRACE("Deleting linked programs.\n");
5891 linked_programs
= &shader
->linked_programs
;
5892 if (linked_programs
->next
)
5894 struct glsl_shader_prog_link
*entry
, *entry2
;
5897 switch (shader
->reg_maps
.shader_version
.type
)
5899 case WINED3D_SHADER_TYPE_PIXEL
:
5901 struct glsl_ps_compiled_shader
*gl_shaders
= shader_data
->gl_shaders
.ps
;
5903 LIST_FOR_EACH_ENTRY_SAFE(entry
, entry2
, linked_programs
,
5904 struct glsl_shader_prog_link
, ps
.shader_entry
)
5906 delete_glsl_program_entry(priv
, gl_info
, entry
);
5909 for (i
= 0; i
< shader_data
->num_gl_shaders
; ++i
)
5911 TRACE("Deleting pixel shader %u.\n", gl_shaders
[i
].prgId
);
5912 if (priv
->glsl_program
&& priv
->glsl_program
->ps
.id
== gl_shaders
[i
].prgId
)
5913 shader_glsl_select(context
, WINED3D_SHADER_MODE_NONE
, WINED3D_SHADER_MODE_NONE
);
5914 GL_EXTCALL(glDeleteObjectARB(gl_shaders
[i
].prgId
));
5915 checkGLcall("glDeleteObjectARB");
5917 HeapFree(GetProcessHeap(), 0, shader_data
->gl_shaders
.ps
);
5922 case WINED3D_SHADER_TYPE_VERTEX
:
5924 struct glsl_vs_compiled_shader
*gl_shaders
= shader_data
->gl_shaders
.vs
;
5926 LIST_FOR_EACH_ENTRY_SAFE(entry
, entry2
, linked_programs
,
5927 struct glsl_shader_prog_link
, vs
.shader_entry
)
5929 delete_glsl_program_entry(priv
, gl_info
, entry
);
5932 for (i
= 0; i
< shader_data
->num_gl_shaders
; ++i
)
5934 TRACE("Deleting vertex shader %u.\n", gl_shaders
[i
].prgId
);
5935 if (priv
->glsl_program
&& priv
->glsl_program
->vs
.id
== gl_shaders
[i
].prgId
)
5936 shader_glsl_select(context
, WINED3D_SHADER_MODE_NONE
, WINED3D_SHADER_MODE_NONE
);
5937 GL_EXTCALL(glDeleteObjectARB(gl_shaders
[i
].prgId
));
5938 checkGLcall("glDeleteObjectARB");
5940 HeapFree(GetProcessHeap(), 0, shader_data
->gl_shaders
.vs
);
5945 case WINED3D_SHADER_TYPE_GEOMETRY
:
5947 struct glsl_gs_compiled_shader
*gl_shaders
= shader_data
->gl_shaders
.gs
;
5949 LIST_FOR_EACH_ENTRY_SAFE(entry
, entry2
, linked_programs
,
5950 struct glsl_shader_prog_link
, gs
.shader_entry
)
5952 delete_glsl_program_entry(priv
, gl_info
, entry
);
5955 for (i
= 0; i
< shader_data
->num_gl_shaders
; ++i
)
5957 TRACE("Deleting geometry shader %u.\n", gl_shaders
[i
].id
);
5958 if (priv
->glsl_program
&& priv
->glsl_program
->gs
.id
== gl_shaders
[i
].id
)
5959 shader_glsl_select(context
, WINED3D_SHADER_MODE_NONE
, WINED3D_SHADER_MODE_NONE
);
5960 GL_EXTCALL(glDeleteObjectARB(gl_shaders
[i
].id
));
5961 checkGLcall("glDeleteObjectARB");
5963 HeapFree(GetProcessHeap(), 0, shader_data
->gl_shaders
.gs
);
5969 ERR("Unhandled shader type %#x.\n", shader
->reg_maps
.shader_version
.type
);
5974 HeapFree(GetProcessHeap(), 0, shader
->backend_data
);
5975 shader
->backend_data
= NULL
;
5977 context_release(context
);
5980 static int glsl_program_key_compare(const void *key
, const struct wine_rb_entry
*entry
)
5982 const struct glsl_program_key
*k
= key
;
5983 const struct glsl_shader_prog_link
*prog
= WINE_RB_ENTRY_VALUE(entry
,
5984 const struct glsl_shader_prog_link
, program_lookup_entry
);
5986 if (k
->vs_id
> prog
->vs
.id
) return 1;
5987 else if (k
->vs_id
< prog
->vs
.id
) return -1;
5989 if (k
->gs_id
> prog
->gs
.id
) return 1;
5990 else if (k
->gs_id
< prog
->gs
.id
) return -1;
5992 if (k
->ps_id
> prog
->ps
.id
) return 1;
5993 else if (k
->ps_id
< prog
->ps
.id
) return -1;
5998 static BOOL
constant_heap_init(struct constant_heap
*heap
, unsigned int constant_count
)
6000 SIZE_T size
= (constant_count
+ 1) * sizeof(*heap
->entries
) + constant_count
* sizeof(*heap
->positions
);
6001 void *mem
= HeapAlloc(GetProcessHeap(), 0, size
);
6005 ERR("Failed to allocate memory\n");
6009 heap
->entries
= mem
;
6010 heap
->entries
[1].version
= 0;
6011 heap
->positions
= (unsigned int *)(heap
->entries
+ constant_count
+ 1);
6017 static void constant_heap_free(struct constant_heap
*heap
)
6019 HeapFree(GetProcessHeap(), 0, heap
->entries
);
6022 static const struct wine_rb_functions wined3d_glsl_program_rb_functions
=
6027 glsl_program_key_compare
,
6030 static HRESULT
shader_glsl_alloc(struct wined3d_device
*device
, const struct wined3d_vertex_pipe_ops
*vertex_pipe
,
6031 const struct fragment_pipeline
*fragment_pipe
)
6033 const struct wined3d_gl_info
*gl_info
= &device
->adapter
->gl_info
;
6034 struct shader_glsl_priv
*priv
= HeapAlloc(GetProcessHeap(), HEAP_ZERO_MEMORY
, sizeof(struct shader_glsl_priv
));
6035 SIZE_T stack_size
= wined3d_log2i(max(gl_info
->limits
.glsl_vs_float_constants
,
6036 gl_info
->limits
.glsl_ps_float_constants
)) + 1;
6037 struct fragment_caps fragment_caps
;
6038 void *vertex_priv
, *fragment_priv
;
6040 if (!(vertex_priv
= vertex_pipe
->vp_alloc(&glsl_shader_backend
, priv
)))
6042 ERR("Failed to initialize vertex pipe.\n");
6043 HeapFree(GetProcessHeap(), 0, priv
);
6047 if (!(fragment_priv
= fragment_pipe
->alloc_private(&glsl_shader_backend
, priv
)))
6049 ERR("Failed to initialize fragment pipe.\n");
6050 vertex_pipe
->vp_free(device
);
6051 HeapFree(GetProcessHeap(), 0, priv
);
6055 if (!shader_buffer_init(&priv
->shader_buffer
))
6057 ERR("Failed to initialize shader buffer.\n");
6061 priv
->stack
= HeapAlloc(GetProcessHeap(), 0, stack_size
* sizeof(*priv
->stack
));
6064 ERR("Failed to allocate memory.\n");
6068 if (!constant_heap_init(&priv
->vconst_heap
, gl_info
->limits
.glsl_vs_float_constants
))
6070 ERR("Failed to initialize vertex shader constant heap\n");
6074 if (!constant_heap_init(&priv
->pconst_heap
, gl_info
->limits
.glsl_ps_float_constants
))
6076 ERR("Failed to initialize pixel shader constant heap\n");
6080 if (wine_rb_init(&priv
->program_lookup
, &wined3d_glsl_program_rb_functions
) == -1)
6082 ERR("Failed to initialize rbtree.\n");
6086 priv
->next_constant_version
= 1;
6087 priv
->vertex_pipe
= vertex_pipe
;
6088 priv
->fragment_pipe
= fragment_pipe
;
6089 fragment_pipe
->get_caps(gl_info
, &fragment_caps
);
6090 priv
->ffp_proj_control
= fragment_caps
.wined3d_caps
& WINED3D_FRAGMENT_CAP_PROJ_CONTROL
;
6092 device
->vertex_priv
= vertex_priv
;
6093 device
->fragment_priv
= fragment_priv
;
6094 device
->shader_priv
= priv
;
6099 constant_heap_free(&priv
->pconst_heap
);
6100 constant_heap_free(&priv
->vconst_heap
);
6101 HeapFree(GetProcessHeap(), 0, priv
->stack
);
6102 shader_buffer_free(&priv
->shader_buffer
);
6103 fragment_pipe
->free_private(device
);
6104 vertex_pipe
->vp_free(device
);
6105 HeapFree(GetProcessHeap(), 0, priv
);
6106 return E_OUTOFMEMORY
;
6109 /* Context activation is done by the caller. */
6110 static void shader_glsl_free(struct wined3d_device
*device
)
6112 const struct wined3d_gl_info
*gl_info
= &device
->adapter
->gl_info
;
6113 struct shader_glsl_priv
*priv
= device
->shader_priv
;
6116 for (i
= 0; i
< tex_type_count
; ++i
)
6118 if (priv
->depth_blt_program_full
[i
])
6120 GL_EXTCALL(glDeleteObjectARB(priv
->depth_blt_program_full
[i
]));
6122 if (priv
->depth_blt_program_masked
[i
])
6124 GL_EXTCALL(glDeleteObjectARB(priv
->depth_blt_program_masked
[i
]));
6128 wine_rb_destroy(&priv
->program_lookup
, NULL
, NULL
);
6129 constant_heap_free(&priv
->pconst_heap
);
6130 constant_heap_free(&priv
->vconst_heap
);
6131 HeapFree(GetProcessHeap(), 0, priv
->stack
);
6132 shader_buffer_free(&priv
->shader_buffer
);
6133 priv
->fragment_pipe
->free_private(device
);
6134 priv
->vertex_pipe
->vp_free(device
);
6136 HeapFree(GetProcessHeap(), 0, device
->shader_priv
);
6137 device
->shader_priv
= NULL
;
6140 static void shader_glsl_context_destroyed(void *shader_priv
, const struct wined3d_context
*context
) {}
6142 static void shader_glsl_get_caps(const struct wined3d_gl_info
*gl_info
, struct shader_caps
*caps
)
6146 if (gl_info
->supported
[EXT_GPU_SHADER4
] && gl_info
->supported
[ARB_SHADER_BIT_ENCODING
]
6147 && gl_info
->supported
[ARB_GEOMETRY_SHADER4
] && gl_info
->glsl_version
>= MAKEDWORD_VERSION(1, 50)
6148 && gl_info
->supported
[ARB_DRAW_ELEMENTS_BASE_VERTEX
] && gl_info
->supported
[ARB_DRAW_INSTANCED
])
6150 /* ARB_shader_texture_lod or EXT_gpu_shader4 is required for the SM3
6151 * texldd and texldl instructions. */
6152 else if (gl_info
->supported
[ARB_SHADER_TEXTURE_LOD
] || gl_info
->supported
[EXT_GPU_SHADER4
])
6156 TRACE("Shader model %u.\n", shader_model
);
6158 caps
->vs_version
= min(wined3d_settings
.max_sm_vs
, shader_model
);
6159 caps
->gs_version
= min(wined3d_settings
.max_sm_gs
, shader_model
);
6160 caps
->ps_version
= min(wined3d_settings
.max_sm_ps
, shader_model
);
6162 caps
->vs_uniform_count
= gl_info
->limits
.glsl_vs_float_constants
;
6163 caps
->ps_uniform_count
= gl_info
->limits
.glsl_ps_float_constants
;
6165 /* FIXME: The following line is card dependent. -8.0 to 8.0 is the
6166 * Direct3D minimum requirement.
6168 * Both GL_ARB_fragment_program and GLSL require a "maximum representable magnitude"
6169 * of colors to be 2^10, and 2^32 for other floats. Should we use 1024 here?
6171 * The problem is that the refrast clamps temporary results in the shader to
6172 * [-MaxValue;+MaxValue]. If the card's max value is bigger than the one we advertize here,
6173 * then applications may miss the clamping behavior. On the other hand, if it is smaller,
6174 * the shader will generate incorrect results too. Unfortunately, GL deliberately doesn't
6175 * offer a way to query this.
6177 caps
->ps_1x_max_value
= 8.0;
6179 /* Ideally we'd only set caps like sRGB writes here if supported by both
6180 * the shader backend and the fragment pipe, but we can get called before
6181 * shader_glsl_alloc(). */
6182 caps
->wined3d_caps
= WINED3D_SHADER_CAP_VS_CLIPPING
6183 | WINED3D_SHADER_CAP_SRGB_WRITE
;
6186 static BOOL
shader_glsl_color_fixup_supported(struct color_fixup_desc fixup
)
6188 if (TRACE_ON(d3d_shader
) && TRACE_ON(d3d
))
6190 TRACE("Checking support for fixup:\n");
6191 dump_color_fixup_desc(fixup
);
6194 /* We support everything except YUV conversions. */
6195 if (!is_complex_fixup(fixup
))
6201 TRACE("[FAILED]\n");
6205 static const SHADER_HANDLER shader_glsl_instruction_handler_table
[WINED3DSIH_TABLE_SIZE
] =
6207 /* WINED3DSIH_ABS */ shader_glsl_map2gl
,
6208 /* WINED3DSIH_ADD */ shader_glsl_binop
,
6209 /* WINED3DSIH_AND */ shader_glsl_binop
,
6210 /* WINED3DSIH_BEM */ shader_glsl_bem
,
6211 /* WINED3DSIH_BREAK */ shader_glsl_break
,
6212 /* WINED3DSIH_BREAKC */ shader_glsl_breakc
,
6213 /* WINED3DSIH_BREAKP */ shader_glsl_breakp
,
6214 /* WINED3DSIH_CALL */ shader_glsl_call
,
6215 /* WINED3DSIH_CALLNZ */ shader_glsl_callnz
,
6216 /* WINED3DSIH_CMP */ shader_glsl_conditional_move
,
6217 /* WINED3DSIH_CND */ shader_glsl_cnd
,
6218 /* WINED3DSIH_CRS */ shader_glsl_cross
,
6219 /* WINED3DSIH_CUT */ shader_glsl_cut
,
6220 /* WINED3DSIH_DCL */ shader_glsl_nop
,
6221 /* WINED3DSIH_DCL_CONSTANT_BUFFER */ shader_glsl_nop
,
6222 /* WINED3DSIH_DCL_INPUT_PRIMITIVE */ shader_glsl_nop
,
6223 /* WINED3DSIH_DCL_OUTPUT_TOPOLOGY */ shader_glsl_nop
,
6224 /* WINED3DSIH_DCL_VERTICES_OUT */ shader_glsl_nop
,
6225 /* WINED3DSIH_DEF */ shader_glsl_nop
,
6226 /* WINED3DSIH_DEFB */ shader_glsl_nop
,
6227 /* WINED3DSIH_DEFI */ shader_glsl_nop
,
6228 /* WINED3DSIH_DIV */ shader_glsl_binop
,
6229 /* WINED3DSIH_DP2ADD */ shader_glsl_dp2add
,
6230 /* WINED3DSIH_DP3 */ shader_glsl_dot
,
6231 /* WINED3DSIH_DP4 */ shader_glsl_dot
,
6232 /* WINED3DSIH_DST */ shader_glsl_dst
,
6233 /* WINED3DSIH_DSX */ shader_glsl_map2gl
,
6234 /* WINED3DSIH_DSY */ shader_glsl_map2gl
,
6235 /* WINED3DSIH_ELSE */ shader_glsl_else
,
6236 /* WINED3DSIH_EMIT */ shader_glsl_emit
,
6237 /* WINED3DSIH_ENDIF */ shader_glsl_end
,
6238 /* WINED3DSIH_ENDLOOP */ shader_glsl_end
,
6239 /* WINED3DSIH_ENDREP */ shader_glsl_end
,
6240 /* WINED3DSIH_EQ */ shader_glsl_relop
,
6241 /* WINED3DSIH_EXP */ shader_glsl_map2gl
,
6242 /* WINED3DSIH_EXPP */ shader_glsl_expp
,
6243 /* WINED3DSIH_FRC */ shader_glsl_map2gl
,
6244 /* WINED3DSIH_FTOI */ shader_glsl_to_int
,
6245 /* WINED3DSIH_GE */ shader_glsl_relop
,
6246 /* WINED3DSIH_IADD */ shader_glsl_binop
,
6247 /* WINED3DSIH_IEQ */ NULL
,
6248 /* WINED3DSIH_IF */ shader_glsl_if
,
6249 /* WINED3DSIH_IFC */ shader_glsl_ifc
,
6250 /* WINED3DSIH_IGE */ shader_glsl_relop
,
6251 /* WINED3DSIH_IMUL */ shader_glsl_imul
,
6252 /* WINED3DSIH_ITOF */ shader_glsl_to_float
,
6253 /* WINED3DSIH_LABEL */ shader_glsl_label
,
6254 /* WINED3DSIH_LD */ NULL
,
6255 /* WINED3DSIH_LIT */ shader_glsl_lit
,
6256 /* WINED3DSIH_LOG */ shader_glsl_log
,
6257 /* WINED3DSIH_LOGP */ shader_glsl_log
,
6258 /* WINED3DSIH_LOOP */ shader_glsl_loop
,
6259 /* WINED3DSIH_LRP */ shader_glsl_lrp
,
6260 /* WINED3DSIH_LT */ shader_glsl_relop
,
6261 /* WINED3DSIH_M3x2 */ shader_glsl_mnxn
,
6262 /* WINED3DSIH_M3x3 */ shader_glsl_mnxn
,
6263 /* WINED3DSIH_M3x4 */ shader_glsl_mnxn
,
6264 /* WINED3DSIH_M4x3 */ shader_glsl_mnxn
,
6265 /* WINED3DSIH_M4x4 */ shader_glsl_mnxn
,
6266 /* WINED3DSIH_MAD */ shader_glsl_mad
,
6267 /* WINED3DSIH_MAX */ shader_glsl_map2gl
,
6268 /* WINED3DSIH_MIN */ shader_glsl_map2gl
,
6269 /* WINED3DSIH_MOV */ shader_glsl_mov
,
6270 /* WINED3DSIH_MOVA */ shader_glsl_mov
,
6271 /* WINED3DSIH_MOVC */ shader_glsl_conditional_move
,
6272 /* WINED3DSIH_MUL */ shader_glsl_binop
,
6273 /* WINED3DSIH_NOP */ shader_glsl_nop
,
6274 /* WINED3DSIH_NRM */ shader_glsl_nrm
,
6275 /* WINED3DSIH_PHASE */ shader_glsl_nop
,
6276 /* WINED3DSIH_POW */ shader_glsl_pow
,
6277 /* WINED3DSIH_RCP */ shader_glsl_rcp
,
6278 /* WINED3DSIH_REP */ shader_glsl_rep
,
6279 /* WINED3DSIH_RET */ shader_glsl_ret
,
6280 /* WINED3DSIH_ROUND_NI */ shader_glsl_map2gl
,
6281 /* WINED3DSIH_RSQ */ shader_glsl_rsq
,
6282 /* WINED3DSIH_SAMPLE */ NULL
,
6283 /* WINED3DSIH_SAMPLE_GRAD */ NULL
,
6284 /* WINED3DSIH_SAMPLE_LOD */ NULL
,
6285 /* WINED3DSIH_SETP */ NULL
,
6286 /* WINED3DSIH_SGE */ shader_glsl_compare
,
6287 /* WINED3DSIH_SGN */ shader_glsl_sgn
,
6288 /* WINED3DSIH_SINCOS */ shader_glsl_sincos
,
6289 /* WINED3DSIH_SLT */ shader_glsl_compare
,
6290 /* WINED3DSIH_SQRT */ NULL
,
6291 /* WINED3DSIH_SUB */ shader_glsl_binop
,
6292 /* WINED3DSIH_TEX */ shader_glsl_tex
,
6293 /* WINED3DSIH_TEXBEM */ shader_glsl_texbem
,
6294 /* WINED3DSIH_TEXBEML */ shader_glsl_texbem
,
6295 /* WINED3DSIH_TEXCOORD */ shader_glsl_texcoord
,
6296 /* WINED3DSIH_TEXDEPTH */ shader_glsl_texdepth
,
6297 /* WINED3DSIH_TEXDP3 */ shader_glsl_texdp3
,
6298 /* WINED3DSIH_TEXDP3TEX */ shader_glsl_texdp3tex
,
6299 /* WINED3DSIH_TEXKILL */ shader_glsl_texkill
,
6300 /* WINED3DSIH_TEXLDD */ shader_glsl_texldd
,
6301 /* WINED3DSIH_TEXLDL */ shader_glsl_texldl
,
6302 /* WINED3DSIH_TEXM3x2DEPTH */ shader_glsl_texm3x2depth
,
6303 /* WINED3DSIH_TEXM3x2PAD */ shader_glsl_texm3x2pad
,
6304 /* WINED3DSIH_TEXM3x2TEX */ shader_glsl_texm3x2tex
,
6305 /* WINED3DSIH_TEXM3x3 */ shader_glsl_texm3x3
,
6306 /* WINED3DSIH_TEXM3x3DIFF */ NULL
,
6307 /* WINED3DSIH_TEXM3x3PAD */ shader_glsl_texm3x3pad
,
6308 /* WINED3DSIH_TEXM3x3SPEC */ shader_glsl_texm3x3spec
,
6309 /* WINED3DSIH_TEXM3x3TEX */ shader_glsl_texm3x3tex
,
6310 /* WINED3DSIH_TEXM3x3VSPEC */ shader_glsl_texm3x3vspec
,
6311 /* WINED3DSIH_TEXREG2AR */ shader_glsl_texreg2ar
,
6312 /* WINED3DSIH_TEXREG2GB */ shader_glsl_texreg2gb
,
6313 /* WINED3DSIH_TEXREG2RGB */ shader_glsl_texreg2rgb
,
6314 /* WINED3DSIH_UDIV */ shader_glsl_udiv
,
6315 /* WINED3DSIH_USHR */ shader_glsl_binop
,
6316 /* WINED3DSIH_UTOF */ shader_glsl_to_float
,
6317 /* WINED3DSIH_XOR */ shader_glsl_binop
,
6320 static void shader_glsl_handle_instruction(const struct wined3d_shader_instruction
*ins
) {
6321 SHADER_HANDLER hw_fct
;
6323 /* Select handler */
6324 hw_fct
= shader_glsl_instruction_handler_table
[ins
->handler_idx
];
6326 /* Unhandled opcode */
6329 FIXME("Backend can't handle opcode %#x\n", ins
->handler_idx
);
6334 shader_glsl_add_instruction_modifiers(ins
);
6337 static BOOL
shader_glsl_has_ffp_proj_control(void *shader_priv
)
6339 struct shader_glsl_priv
*priv
= shader_priv
;
6341 return priv
->ffp_proj_control
;
6344 const struct wined3d_shader_backend_ops glsl_shader_backend
=
6346 shader_glsl_handle_instruction
,
6348 shader_glsl_select_depth_blt
,
6349 shader_glsl_deselect_depth_blt
,
6350 shader_glsl_update_float_vertex_constants
,
6351 shader_glsl_update_float_pixel_constants
,
6352 shader_glsl_load_constants
,
6353 shader_glsl_load_np2fixup_constants
,
6354 shader_glsl_destroy
,
6357 shader_glsl_context_destroyed
,
6358 shader_glsl_get_caps
,
6359 shader_glsl_color_fixup_supported
,
6360 shader_glsl_has_ffp_proj_control
,
6363 static void glsl_fragment_pipe_enable(const struct wined3d_gl_info
*gl_info
, BOOL enable
)
6365 /* Nothing to do. */
6368 static void glsl_fragment_pipe_get_caps(const struct wined3d_gl_info
*gl_info
, struct fragment_caps
*caps
)
6370 caps
->wined3d_caps
= WINED3D_FRAGMENT_CAP_PROJ_CONTROL
6371 | WINED3D_FRAGMENT_CAP_SRGB_WRITE
;
6372 caps
->PrimitiveMiscCaps
= WINED3DPMISCCAPS_TSSARGTEMP
;
6373 caps
->TextureOpCaps
= WINED3DTEXOPCAPS_DISABLE
6374 | WINED3DTEXOPCAPS_SELECTARG1
6375 | WINED3DTEXOPCAPS_SELECTARG2
6376 | WINED3DTEXOPCAPS_MODULATE4X
6377 | WINED3DTEXOPCAPS_MODULATE2X
6378 | WINED3DTEXOPCAPS_MODULATE
6379 | WINED3DTEXOPCAPS_ADDSIGNED2X
6380 | WINED3DTEXOPCAPS_ADDSIGNED
6381 | WINED3DTEXOPCAPS_ADD
6382 | WINED3DTEXOPCAPS_SUBTRACT
6383 | WINED3DTEXOPCAPS_ADDSMOOTH
6384 | WINED3DTEXOPCAPS_BLENDCURRENTALPHA
6385 | WINED3DTEXOPCAPS_BLENDFACTORALPHA
6386 | WINED3DTEXOPCAPS_BLENDTEXTUREALPHA
6387 | WINED3DTEXOPCAPS_BLENDDIFFUSEALPHA
6388 | WINED3DTEXOPCAPS_BLENDTEXTUREALPHAPM
6389 | WINED3DTEXOPCAPS_MODULATEALPHA_ADDCOLOR
6390 | WINED3DTEXOPCAPS_MODULATECOLOR_ADDALPHA
6391 | WINED3DTEXOPCAPS_MODULATEINVCOLOR_ADDALPHA
6392 | WINED3DTEXOPCAPS_MODULATEINVALPHA_ADDCOLOR
6393 | WINED3DTEXOPCAPS_DOTPRODUCT3
6394 | WINED3DTEXOPCAPS_MULTIPLYADD
6395 | WINED3DTEXOPCAPS_LERP
6396 | WINED3DTEXOPCAPS_BUMPENVMAP
6397 | WINED3DTEXOPCAPS_BUMPENVMAPLUMINANCE
;
6398 caps
->MaxTextureBlendStages
= 8;
6399 caps
->MaxSimultaneousTextures
= min(gl_info
->limits
.fragment_samplers
, 8);
6402 static void *glsl_fragment_pipe_alloc(const struct wined3d_shader_backend_ops
*shader_backend
, void *shader_priv
)
6404 struct shader_glsl_priv
*priv
;
6406 if (shader_backend
== &glsl_shader_backend
)
6410 if (wine_rb_init(&priv
->ffp_fragment_shaders
, &wined3d_ffp_frag_program_rb_functions
) == -1)
6412 ERR("Failed to initialize rbtree.\n");
6419 FIXME("GLSL fragment pipe without GLSL shader backend not implemented.\n");
6424 struct glsl_ffp_destroy_ctx
6426 struct shader_glsl_priv
*priv
;
6427 const struct wined3d_gl_info
*gl_info
;
6430 static void shader_glsl_free_ffp_fragment_shader(struct wine_rb_entry
*entry
, void *context
)
6432 struct glsl_ffp_fragment_shader
*shader
= WINE_RB_ENTRY_VALUE(entry
,
6433 struct glsl_ffp_fragment_shader
, entry
.entry
);
6434 struct glsl_shader_prog_link
*program
, *program2
;
6435 struct glsl_ffp_destroy_ctx
*ctx
= context
;
6437 LIST_FOR_EACH_ENTRY_SAFE(program
, program2
, &shader
->linked_programs
,
6438 struct glsl_shader_prog_link
, ps
.shader_entry
)
6440 delete_glsl_program_entry(ctx
->priv
, ctx
->gl_info
, program
);
6442 ctx
->gl_info
->gl_ops
.ext
.p_glDeleteObjectARB(shader
->id
);
6443 HeapFree(GetProcessHeap(), 0, shader
);
6446 /* Context activation is done by the caller. */
6447 static void glsl_fragment_pipe_free(struct wined3d_device
*device
)
6449 struct shader_glsl_priv
*priv
= device
->fragment_priv
;
6450 struct glsl_ffp_destroy_ctx ctx
;
6453 ctx
.gl_info
= &device
->adapter
->gl_info
;
6454 wine_rb_destroy(&priv
->ffp_fragment_shaders
, shader_glsl_free_ffp_fragment_shader
, &ctx
);
6457 static void glsl_fragment_pipe_shader(struct wined3d_context
*context
,
6458 const struct wined3d_state
*state
, DWORD state_id
)
6460 context
->last_was_pshader
= use_ps(state
);
6462 context
->select_shader
= 1;
6463 context
->load_constants
= 1;
6466 static void glsl_fragment_pipe_fog(struct wined3d_context
*context
,
6467 const struct wined3d_state
*state
, DWORD state_id
)
6469 BOOL use_vshader
= use_vs(state
);
6470 enum fogsource new_source
;
6472 context
->select_shader
= 1;
6473 context
->load_constants
= 1;
6475 if (!state
->render_states
[WINED3D_RS_FOGENABLE
])
6478 if (state
->render_states
[WINED3D_RS_FOGTABLEMODE
] == WINED3D_FOG_NONE
)
6481 new_source
= FOGSOURCE_VS
;
6482 else if (state
->render_states
[WINED3D_RS_FOGVERTEXMODE
] == WINED3D_FOG_NONE
|| context
->last_was_rhw
)
6483 new_source
= FOGSOURCE_COORD
;
6485 new_source
= FOGSOURCE_FFP
;
6489 new_source
= FOGSOURCE_FFP
;
6492 if (new_source
!= context
->fog_source
)
6494 context
->fog_source
= new_source
;
6495 state_fogstartend(context
, state
, STATE_RENDER(WINED3D_RS_FOGSTART
));
6499 static void glsl_fragment_pipe_tex_transform(struct wined3d_context
*context
,
6500 const struct wined3d_state
*state
, DWORD state_id
)
6502 context
->select_shader
= 1;
6503 context
->load_constants
= 1;
6506 static void glsl_fragment_pipe_invalidate_constants(struct wined3d_context
*context
,
6507 const struct wined3d_state
*state
, DWORD state_id
)
6509 context
->load_constants
= 1;
6512 static const struct StateEntryTemplate glsl_fragment_pipe_state_template
[] =
6514 {STATE_RENDER(WINED3D_RS_TEXTUREFACTOR
), {STATE_RENDER(WINED3D_RS_TEXTUREFACTOR
), glsl_fragment_pipe_invalidate_constants
}, WINED3D_GL_EXT_NONE
},
6515 {STATE_TEXTURESTAGE(0, WINED3D_TSS_COLOR_OP
), {STATE_PIXELSHADER
, NULL
}, WINED3D_GL_EXT_NONE
},
6516 {STATE_TEXTURESTAGE(0, WINED3D_TSS_COLOR_ARG1
), {STATE_PIXELSHADER
, NULL
}, WINED3D_GL_EXT_NONE
},
6517 {STATE_TEXTURESTAGE(0, WINED3D_TSS_COLOR_ARG2
), {STATE_PIXELSHADER
, NULL
}, WINED3D_GL_EXT_NONE
},
6518 {STATE_TEXTURESTAGE(0, WINED3D_TSS_COLOR_ARG0
), {STATE_PIXELSHADER
, NULL
}, WINED3D_GL_EXT_NONE
},
6519 {STATE_TEXTURESTAGE(0, WINED3D_TSS_ALPHA_OP
), {STATE_PIXELSHADER
, NULL
}, WINED3D_GL_EXT_NONE
},
6520 {STATE_TEXTURESTAGE(0, WINED3D_TSS_ALPHA_ARG1
), {STATE_PIXELSHADER
, NULL
}, WINED3D_GL_EXT_NONE
},
6521 {STATE_TEXTURESTAGE(0, WINED3D_TSS_ALPHA_ARG2
), {STATE_PIXELSHADER
, NULL
}, WINED3D_GL_EXT_NONE
},
6522 {STATE_TEXTURESTAGE(0, WINED3D_TSS_ALPHA_ARG0
), {STATE_PIXELSHADER
, NULL
}, WINED3D_GL_EXT_NONE
},
6523 {STATE_TEXTURESTAGE(0, WINED3D_TSS_RESULT_ARG
), {STATE_PIXELSHADER
, NULL
}, WINED3D_GL_EXT_NONE
},
6524 {STATE_TEXTURESTAGE(0, WINED3D_TSS_BUMPENV_MAT00
), {STATE_TEXTURESTAGE(0, WINED3D_TSS_BUMPENV_MAT00
), glsl_fragment_pipe_invalidate_constants
}, WINED3D_GL_EXT_NONE
},
6525 {STATE_TEXTURESTAGE(0, WINED3D_TSS_BUMPENV_MAT01
), {STATE_TEXTURESTAGE(0, WINED3D_TSS_BUMPENV_MAT00
), NULL
}, WINED3D_GL_EXT_NONE
},
6526 {STATE_TEXTURESTAGE(0, WINED3D_TSS_BUMPENV_MAT10
), {STATE_TEXTURESTAGE(0, WINED3D_TSS_BUMPENV_MAT00
), NULL
}, WINED3D_GL_EXT_NONE
},
6527 {STATE_TEXTURESTAGE(0, WINED3D_TSS_BUMPENV_MAT11
), {STATE_TEXTURESTAGE(0, WINED3D_TSS_BUMPENV_MAT00
), NULL
}, WINED3D_GL_EXT_NONE
},
6528 {STATE_TEXTURESTAGE(0, WINED3D_TSS_BUMPENV_LSCALE
), {STATE_TEXTURESTAGE(0, WINED3D_TSS_BUMPENV_LSCALE
), glsl_fragment_pipe_invalidate_constants
}, WINED3D_GL_EXT_NONE
},
6529 {STATE_TEXTURESTAGE(0, WINED3D_TSS_BUMPENV_LOFFSET
), {STATE_TEXTURESTAGE(0, WINED3D_TSS_BUMPENV_LSCALE
), NULL
}, WINED3D_GL_EXT_NONE
},
6530 {STATE_TEXTURESTAGE(1, WINED3D_TSS_COLOR_OP
), {STATE_PIXELSHADER
, NULL
}, WINED3D_GL_EXT_NONE
},
6531 {STATE_TEXTURESTAGE(1, WINED3D_TSS_COLOR_ARG1
), {STATE_PIXELSHADER
, NULL
}, WINED3D_GL_EXT_NONE
},
6532 {STATE_TEXTURESTAGE(1, WINED3D_TSS_COLOR_ARG2
), {STATE_PIXELSHADER
, NULL
}, WINED3D_GL_EXT_NONE
},
6533 {STATE_TEXTURESTAGE(1, WINED3D_TSS_COLOR_ARG0
), {STATE_PIXELSHADER
, NULL
}, WINED3D_GL_EXT_NONE
},
6534 {STATE_TEXTURESTAGE(1, WINED3D_TSS_ALPHA_OP
), {STATE_PIXELSHADER
, NULL
}, WINED3D_GL_EXT_NONE
},
6535 {STATE_TEXTURESTAGE(1, WINED3D_TSS_ALPHA_ARG1
), {STATE_PIXELSHADER
, NULL
}, WINED3D_GL_EXT_NONE
},
6536 {STATE_TEXTURESTAGE(1, WINED3D_TSS_ALPHA_ARG2
), {STATE_PIXELSHADER
, NULL
}, WINED3D_GL_EXT_NONE
},
6537 {STATE_TEXTURESTAGE(1, WINED3D_TSS_ALPHA_ARG0
), {STATE_PIXELSHADER
, NULL
}, WINED3D_GL_EXT_NONE
},
6538 {STATE_TEXTURESTAGE(1, WINED3D_TSS_RESULT_ARG
), {STATE_PIXELSHADER
, NULL
}, WINED3D_GL_EXT_NONE
},
6539 {STATE_TEXTURESTAGE(1, WINED3D_TSS_BUMPENV_MAT00
), {STATE_TEXTURESTAGE(1, WINED3D_TSS_BUMPENV_MAT00
), glsl_fragment_pipe_invalidate_constants
}, WINED3D_GL_EXT_NONE
},
6540 {STATE_TEXTURESTAGE(1, WINED3D_TSS_BUMPENV_MAT01
), {STATE_TEXTURESTAGE(1, WINED3D_TSS_BUMPENV_MAT00
), NULL
}, WINED3D_GL_EXT_NONE
},
6541 {STATE_TEXTURESTAGE(1, WINED3D_TSS_BUMPENV_MAT10
), {STATE_TEXTURESTAGE(1, WINED3D_TSS_BUMPENV_MAT00
), NULL
}, WINED3D_GL_EXT_NONE
},
6542 {STATE_TEXTURESTAGE(1, WINED3D_TSS_BUMPENV_MAT11
), {STATE_TEXTURESTAGE(1, WINED3D_TSS_BUMPENV_MAT00
), NULL
}, WINED3D_GL_EXT_NONE
},
6543 {STATE_TEXTURESTAGE(1, WINED3D_TSS_BUMPENV_LSCALE
), {STATE_TEXTURESTAGE(1, WINED3D_TSS_BUMPENV_LSCALE
), glsl_fragment_pipe_invalidate_constants
}, WINED3D_GL_EXT_NONE
},
6544 {STATE_TEXTURESTAGE(1, WINED3D_TSS_BUMPENV_LOFFSET
), {STATE_TEXTURESTAGE(1, WINED3D_TSS_BUMPENV_LSCALE
), NULL
}, WINED3D_GL_EXT_NONE
},
6545 {STATE_TEXTURESTAGE(2, WINED3D_TSS_COLOR_OP
), {STATE_PIXELSHADER
, NULL
}, WINED3D_GL_EXT_NONE
},
6546 {STATE_TEXTURESTAGE(2, WINED3D_TSS_COLOR_ARG1
), {STATE_PIXELSHADER
, NULL
}, WINED3D_GL_EXT_NONE
},
6547 {STATE_TEXTURESTAGE(2, WINED3D_TSS_COLOR_ARG2
), {STATE_PIXELSHADER
, NULL
}, WINED3D_GL_EXT_NONE
},
6548 {STATE_TEXTURESTAGE(2, WINED3D_TSS_COLOR_ARG0
), {STATE_PIXELSHADER
, NULL
}, WINED3D_GL_EXT_NONE
},
6549 {STATE_TEXTURESTAGE(2, WINED3D_TSS_ALPHA_OP
), {STATE_PIXELSHADER
, NULL
}, WINED3D_GL_EXT_NONE
},
6550 {STATE_TEXTURESTAGE(2, WINED3D_TSS_ALPHA_ARG1
), {STATE_PIXELSHADER
, NULL
}, WINED3D_GL_EXT_NONE
},
6551 {STATE_TEXTURESTAGE(2, WINED3D_TSS_ALPHA_ARG2
), {STATE_PIXELSHADER
, NULL
}, WINED3D_GL_EXT_NONE
},
6552 {STATE_TEXTURESTAGE(2, WINED3D_TSS_ALPHA_ARG0
), {STATE_PIXELSHADER
, NULL
}, WINED3D_GL_EXT_NONE
},
6553 {STATE_TEXTURESTAGE(2, WINED3D_TSS_RESULT_ARG
), {STATE_PIXELSHADER
, NULL
}, WINED3D_GL_EXT_NONE
},
6554 {STATE_TEXTURESTAGE(2, WINED3D_TSS_BUMPENV_MAT00
), {STATE_TEXTURESTAGE(2, WINED3D_TSS_BUMPENV_MAT00
), glsl_fragment_pipe_invalidate_constants
}, WINED3D_GL_EXT_NONE
},
6555 {STATE_TEXTURESTAGE(2, WINED3D_TSS_BUMPENV_MAT01
), {STATE_TEXTURESTAGE(2, WINED3D_TSS_BUMPENV_MAT00
), NULL
}, WINED3D_GL_EXT_NONE
},
6556 {STATE_TEXTURESTAGE(2, WINED3D_TSS_BUMPENV_MAT10
), {STATE_TEXTURESTAGE(2, WINED3D_TSS_BUMPENV_MAT00
), NULL
}, WINED3D_GL_EXT_NONE
},
6557 {STATE_TEXTURESTAGE(2, WINED3D_TSS_BUMPENV_MAT11
), {STATE_TEXTURESTAGE(2, WINED3D_TSS_BUMPENV_MAT00
), NULL
}, WINED3D_GL_EXT_NONE
},
6558 {STATE_TEXTURESTAGE(2, WINED3D_TSS_BUMPENV_LSCALE
), {STATE_TEXTURESTAGE(2, WINED3D_TSS_BUMPENV_LSCALE
), glsl_fragment_pipe_invalidate_constants
}, WINED3D_GL_EXT_NONE
},
6559 {STATE_TEXTURESTAGE(2, WINED3D_TSS_BUMPENV_LOFFSET
), {STATE_TEXTURESTAGE(2, WINED3D_TSS_BUMPENV_LSCALE
), NULL
}, WINED3D_GL_EXT_NONE
},
6560 {STATE_TEXTURESTAGE(3, WINED3D_TSS_COLOR_OP
), {STATE_PIXELSHADER
, NULL
}, WINED3D_GL_EXT_NONE
},
6561 {STATE_TEXTURESTAGE(3, WINED3D_TSS_COLOR_ARG1
), {STATE_PIXELSHADER
, NULL
}, WINED3D_GL_EXT_NONE
},
6562 {STATE_TEXTURESTAGE(3, WINED3D_TSS_COLOR_ARG2
), {STATE_PIXELSHADER
, NULL
}, WINED3D_GL_EXT_NONE
},
6563 {STATE_TEXTURESTAGE(3, WINED3D_TSS_COLOR_ARG0
), {STATE_PIXELSHADER
, NULL
}, WINED3D_GL_EXT_NONE
},
6564 {STATE_TEXTURESTAGE(3, WINED3D_TSS_ALPHA_OP
), {STATE_PIXELSHADER
, NULL
}, WINED3D_GL_EXT_NONE
},
6565 {STATE_TEXTURESTAGE(3, WINED3D_TSS_ALPHA_ARG1
), {STATE_PIXELSHADER
, NULL
}, WINED3D_GL_EXT_NONE
},
6566 {STATE_TEXTURESTAGE(3, WINED3D_TSS_ALPHA_ARG2
), {STATE_PIXELSHADER
, NULL
}, WINED3D_GL_EXT_NONE
},
6567 {STATE_TEXTURESTAGE(3, WINED3D_TSS_ALPHA_ARG0
), {STATE_PIXELSHADER
, NULL
}, WINED3D_GL_EXT_NONE
},
6568 {STATE_TEXTURESTAGE(3, WINED3D_TSS_RESULT_ARG
), {STATE_PIXELSHADER
, NULL
}, WINED3D_GL_EXT_NONE
},
6569 {STATE_TEXTURESTAGE(3, WINED3D_TSS_BUMPENV_MAT00
), {STATE_TEXTURESTAGE(3, WINED3D_TSS_BUMPENV_MAT00
), glsl_fragment_pipe_invalidate_constants
}, WINED3D_GL_EXT_NONE
},
6570 {STATE_TEXTURESTAGE(3, WINED3D_TSS_BUMPENV_MAT01
), {STATE_TEXTURESTAGE(3, WINED3D_TSS_BUMPENV_MAT00
), NULL
}, WINED3D_GL_EXT_NONE
},
6571 {STATE_TEXTURESTAGE(3, WINED3D_TSS_BUMPENV_MAT10
), {STATE_TEXTURESTAGE(3, WINED3D_TSS_BUMPENV_MAT00
), NULL
}, WINED3D_GL_EXT_NONE
},
6572 {STATE_TEXTURESTAGE(3, WINED3D_TSS_BUMPENV_MAT11
), {STATE_TEXTURESTAGE(3, WINED3D_TSS_BUMPENV_MAT00
), NULL
}, WINED3D_GL_EXT_NONE
},
6573 {STATE_TEXTURESTAGE(3, WINED3D_TSS_BUMPENV_LSCALE
), {STATE_TEXTURESTAGE(3, WINED3D_TSS_BUMPENV_LSCALE
), glsl_fragment_pipe_invalidate_constants
}, WINED3D_GL_EXT_NONE
},
6574 {STATE_TEXTURESTAGE(3, WINED3D_TSS_BUMPENV_LOFFSET
), {STATE_TEXTURESTAGE(3, WINED3D_TSS_BUMPENV_LSCALE
), NULL
}, WINED3D_GL_EXT_NONE
},
6575 {STATE_TEXTURESTAGE(4, WINED3D_TSS_COLOR_OP
), {STATE_PIXELSHADER
, NULL
}, WINED3D_GL_EXT_NONE
},
6576 {STATE_TEXTURESTAGE(4, WINED3D_TSS_COLOR_ARG1
), {STATE_PIXELSHADER
, NULL
}, WINED3D_GL_EXT_NONE
},
6577 {STATE_TEXTURESTAGE(4, WINED3D_TSS_COLOR_ARG2
), {STATE_PIXELSHADER
, NULL
}, WINED3D_GL_EXT_NONE
},
6578 {STATE_TEXTURESTAGE(4, WINED3D_TSS_COLOR_ARG0
), {STATE_PIXELSHADER
, NULL
}, WINED3D_GL_EXT_NONE
},
6579 {STATE_TEXTURESTAGE(4, WINED3D_TSS_ALPHA_OP
), {STATE_PIXELSHADER
, NULL
}, WINED3D_GL_EXT_NONE
},
6580 {STATE_TEXTURESTAGE(4, WINED3D_TSS_ALPHA_ARG1
), {STATE_PIXELSHADER
, NULL
}, WINED3D_GL_EXT_NONE
},
6581 {STATE_TEXTURESTAGE(4, WINED3D_TSS_ALPHA_ARG2
), {STATE_PIXELSHADER
, NULL
}, WINED3D_GL_EXT_NONE
},
6582 {STATE_TEXTURESTAGE(4, WINED3D_TSS_ALPHA_ARG0
), {STATE_PIXELSHADER
, NULL
}, WINED3D_GL_EXT_NONE
},
6583 {STATE_TEXTURESTAGE(4, WINED3D_TSS_RESULT_ARG
), {STATE_PIXELSHADER
, NULL
}, WINED3D_GL_EXT_NONE
},
6584 {STATE_TEXTURESTAGE(4, WINED3D_TSS_BUMPENV_MAT00
), {STATE_TEXTURESTAGE(4, WINED3D_TSS_BUMPENV_MAT00
), glsl_fragment_pipe_invalidate_constants
}, WINED3D_GL_EXT_NONE
},
6585 {STATE_TEXTURESTAGE(4, WINED3D_TSS_BUMPENV_MAT01
), {STATE_TEXTURESTAGE(4, WINED3D_TSS_BUMPENV_MAT00
), NULL
}, WINED3D_GL_EXT_NONE
},
6586 {STATE_TEXTURESTAGE(4, WINED3D_TSS_BUMPENV_MAT10
), {STATE_TEXTURESTAGE(4, WINED3D_TSS_BUMPENV_MAT00
), NULL
}, WINED3D_GL_EXT_NONE
},
6587 {STATE_TEXTURESTAGE(4, WINED3D_TSS_BUMPENV_MAT11
), {STATE_TEXTURESTAGE(4, WINED3D_TSS_BUMPENV_MAT00
), NULL
}, WINED3D_GL_EXT_NONE
},
6588 {STATE_TEXTURESTAGE(4, WINED3D_TSS_BUMPENV_LSCALE
), {STATE_TEXTURESTAGE(4, WINED3D_TSS_BUMPENV_LSCALE
), glsl_fragment_pipe_invalidate_constants
}, WINED3D_GL_EXT_NONE
},
6589 {STATE_TEXTURESTAGE(4, WINED3D_TSS_BUMPENV_LOFFSET
), {STATE_TEXTURESTAGE(4, WINED3D_TSS_BUMPENV_LSCALE
), NULL
}, WINED3D_GL_EXT_NONE
},
6590 {STATE_TEXTURESTAGE(5, WINED3D_TSS_COLOR_OP
), {STATE_PIXELSHADER
, NULL
}, WINED3D_GL_EXT_NONE
},
6591 {STATE_TEXTURESTAGE(5, WINED3D_TSS_COLOR_ARG1
), {STATE_PIXELSHADER
, NULL
}, WINED3D_GL_EXT_NONE
},
6592 {STATE_TEXTURESTAGE(5, WINED3D_TSS_COLOR_ARG2
), {STATE_PIXELSHADER
, NULL
}, WINED3D_GL_EXT_NONE
},
6593 {STATE_TEXTURESTAGE(5, WINED3D_TSS_COLOR_ARG0
), {STATE_PIXELSHADER
, NULL
}, WINED3D_GL_EXT_NONE
},
6594 {STATE_TEXTURESTAGE(5, WINED3D_TSS_ALPHA_OP
), {STATE_PIXELSHADER
, NULL
}, WINED3D_GL_EXT_NONE
},
6595 {STATE_TEXTURESTAGE(5, WINED3D_TSS_ALPHA_ARG1
), {STATE_PIXELSHADER
, NULL
}, WINED3D_GL_EXT_NONE
},
6596 {STATE_TEXTURESTAGE(5, WINED3D_TSS_ALPHA_ARG2
), {STATE_PIXELSHADER
, NULL
}, WINED3D_GL_EXT_NONE
},
6597 {STATE_TEXTURESTAGE(5, WINED3D_TSS_ALPHA_ARG0
), {STATE_PIXELSHADER
, NULL
}, WINED3D_GL_EXT_NONE
},
6598 {STATE_TEXTURESTAGE(5, WINED3D_TSS_RESULT_ARG
), {STATE_PIXELSHADER
, NULL
}, WINED3D_GL_EXT_NONE
},
6599 {STATE_TEXTURESTAGE(5, WINED3D_TSS_BUMPENV_MAT00
), {STATE_TEXTURESTAGE(5, WINED3D_TSS_BUMPENV_MAT00
), glsl_fragment_pipe_invalidate_constants
}, WINED3D_GL_EXT_NONE
},
6600 {STATE_TEXTURESTAGE(5, WINED3D_TSS_BUMPENV_MAT01
), {STATE_TEXTURESTAGE(5, WINED3D_TSS_BUMPENV_MAT00
), NULL
}, WINED3D_GL_EXT_NONE
},
6601 {STATE_TEXTURESTAGE(5, WINED3D_TSS_BUMPENV_MAT10
), {STATE_TEXTURESTAGE(5, WINED3D_TSS_BUMPENV_MAT00
), NULL
}, WINED3D_GL_EXT_NONE
},
6602 {STATE_TEXTURESTAGE(5, WINED3D_TSS_BUMPENV_MAT11
), {STATE_TEXTURESTAGE(5, WINED3D_TSS_BUMPENV_MAT00
), NULL
}, WINED3D_GL_EXT_NONE
},
6603 {STATE_TEXTURESTAGE(5, WINED3D_TSS_BUMPENV_LSCALE
), {STATE_TEXTURESTAGE(5, WINED3D_TSS_BUMPENV_LSCALE
), glsl_fragment_pipe_invalidate_constants
}, WINED3D_GL_EXT_NONE
},
6604 {STATE_TEXTURESTAGE(5, WINED3D_TSS_BUMPENV_LOFFSET
), {STATE_TEXTURESTAGE(5, WINED3D_TSS_BUMPENV_LSCALE
), NULL
}, WINED3D_GL_EXT_NONE
},
6605 {STATE_TEXTURESTAGE(6, WINED3D_TSS_COLOR_OP
), {STATE_PIXELSHADER
, NULL
}, WINED3D_GL_EXT_NONE
},
6606 {STATE_TEXTURESTAGE(6, WINED3D_TSS_COLOR_ARG1
), {STATE_PIXELSHADER
, NULL
}, WINED3D_GL_EXT_NONE
},
6607 {STATE_TEXTURESTAGE(6, WINED3D_TSS_COLOR_ARG2
), {STATE_PIXELSHADER
, NULL
}, WINED3D_GL_EXT_NONE
},
6608 {STATE_TEXTURESTAGE(6, WINED3D_TSS_COLOR_ARG0
), {STATE_PIXELSHADER
, NULL
}, WINED3D_GL_EXT_NONE
},
6609 {STATE_TEXTURESTAGE(6, WINED3D_TSS_ALPHA_OP
), {STATE_PIXELSHADER
, NULL
}, WINED3D_GL_EXT_NONE
},
6610 {STATE_TEXTURESTAGE(6, WINED3D_TSS_ALPHA_ARG1
), {STATE_PIXELSHADER
, NULL
}, WINED3D_GL_EXT_NONE
},
6611 {STATE_TEXTURESTAGE(6, WINED3D_TSS_ALPHA_ARG2
), {STATE_PIXELSHADER
, NULL
}, WINED3D_GL_EXT_NONE
},
6612 {STATE_TEXTURESTAGE(6, WINED3D_TSS_ALPHA_ARG0
), {STATE_PIXELSHADER
, NULL
}, WINED3D_GL_EXT_NONE
},
6613 {STATE_TEXTURESTAGE(6, WINED3D_TSS_RESULT_ARG
), {STATE_PIXELSHADER
, NULL
}, WINED3D_GL_EXT_NONE
},
6614 {STATE_TEXTURESTAGE(6, WINED3D_TSS_BUMPENV_MAT00
), {STATE_TEXTURESTAGE(6, WINED3D_TSS_BUMPENV_MAT00
), glsl_fragment_pipe_invalidate_constants
}, WINED3D_GL_EXT_NONE
},
6615 {STATE_TEXTURESTAGE(6, WINED3D_TSS_BUMPENV_MAT01
), {STATE_TEXTURESTAGE(6, WINED3D_TSS_BUMPENV_MAT00
), NULL
}, WINED3D_GL_EXT_NONE
},
6616 {STATE_TEXTURESTAGE(6, WINED3D_TSS_BUMPENV_MAT10
), {STATE_TEXTURESTAGE(6, WINED3D_TSS_BUMPENV_MAT00
), NULL
}, WINED3D_GL_EXT_NONE
},
6617 {STATE_TEXTURESTAGE(6, WINED3D_TSS_BUMPENV_MAT11
), {STATE_TEXTURESTAGE(6, WINED3D_TSS_BUMPENV_MAT00
), NULL
}, WINED3D_GL_EXT_NONE
},
6618 {STATE_TEXTURESTAGE(6, WINED3D_TSS_BUMPENV_LSCALE
), {STATE_TEXTURESTAGE(6, WINED3D_TSS_BUMPENV_LSCALE
), glsl_fragment_pipe_invalidate_constants
}, WINED3D_GL_EXT_NONE
},
6619 {STATE_TEXTURESTAGE(6, WINED3D_TSS_BUMPENV_LOFFSET
), {STATE_TEXTURESTAGE(6, WINED3D_TSS_BUMPENV_LSCALE
), NULL
}, WINED3D_GL_EXT_NONE
},
6620 {STATE_TEXTURESTAGE(7, WINED3D_TSS_COLOR_OP
), {STATE_PIXELSHADER
, NULL
}, WINED3D_GL_EXT_NONE
},
6621 {STATE_TEXTURESTAGE(7, WINED3D_TSS_COLOR_ARG1
), {STATE_PIXELSHADER
, NULL
}, WINED3D_GL_EXT_NONE
},
6622 {STATE_TEXTURESTAGE(7, WINED3D_TSS_COLOR_ARG2
), {STATE_PIXELSHADER
, NULL
}, WINED3D_GL_EXT_NONE
},
6623 {STATE_TEXTURESTAGE(7, WINED3D_TSS_COLOR_ARG0
), {STATE_PIXELSHADER
, NULL
}, WINED3D_GL_EXT_NONE
},
6624 {STATE_TEXTURESTAGE(7, WINED3D_TSS_ALPHA_OP
), {STATE_PIXELSHADER
, NULL
}, WINED3D_GL_EXT_NONE
},
6625 {STATE_TEXTURESTAGE(7, WINED3D_TSS_ALPHA_ARG1
), {STATE_PIXELSHADER
, NULL
}, WINED3D_GL_EXT_NONE
},
6626 {STATE_TEXTURESTAGE(7, WINED3D_TSS_ALPHA_ARG2
), {STATE_PIXELSHADER
, NULL
}, WINED3D_GL_EXT_NONE
},
6627 {STATE_TEXTURESTAGE(7, WINED3D_TSS_ALPHA_ARG0
), {STATE_PIXELSHADER
, NULL
}, WINED3D_GL_EXT_NONE
},
6628 {STATE_TEXTURESTAGE(7, WINED3D_TSS_RESULT_ARG
), {STATE_PIXELSHADER
, NULL
}, WINED3D_GL_EXT_NONE
},
6629 {STATE_TEXTURESTAGE(7, WINED3D_TSS_BUMPENV_MAT00
), {STATE_TEXTURESTAGE(7, WINED3D_TSS_BUMPENV_MAT00
), glsl_fragment_pipe_invalidate_constants
}, WINED3D_GL_EXT_NONE
},
6630 {STATE_TEXTURESTAGE(7, WINED3D_TSS_BUMPENV_MAT01
), {STATE_TEXTURESTAGE(7, WINED3D_TSS_BUMPENV_MAT00
), NULL
}, WINED3D_GL_EXT_NONE
},
6631 {STATE_TEXTURESTAGE(7, WINED3D_TSS_BUMPENV_MAT10
), {STATE_TEXTURESTAGE(7, WINED3D_TSS_BUMPENV_MAT00
), NULL
}, WINED3D_GL_EXT_NONE
},
6632 {STATE_TEXTURESTAGE(7, WINED3D_TSS_BUMPENV_MAT11
), {STATE_TEXTURESTAGE(7, WINED3D_TSS_BUMPENV_MAT00
), NULL
}, WINED3D_GL_EXT_NONE
},
6633 {STATE_TEXTURESTAGE(7, WINED3D_TSS_BUMPENV_LSCALE
), {STATE_TEXTURESTAGE(7, WINED3D_TSS_BUMPENV_LSCALE
), glsl_fragment_pipe_invalidate_constants
}, WINED3D_GL_EXT_NONE
},
6634 {STATE_TEXTURESTAGE(7, WINED3D_TSS_BUMPENV_LOFFSET
), {STATE_TEXTURESTAGE(7, WINED3D_TSS_BUMPENV_LSCALE
), NULL
}, WINED3D_GL_EXT_NONE
},
6635 {STATE_PIXELSHADER
, {STATE_PIXELSHADER
, glsl_fragment_pipe_shader
}, WINED3D_GL_EXT_NONE
},
6636 {STATE_RENDER(WINED3D_RS_FOGENABLE
), {STATE_RENDER(WINED3D_RS_FOGENABLE
), glsl_fragment_pipe_fog
}, WINED3D_GL_EXT_NONE
},
6637 {STATE_RENDER(WINED3D_RS_FOGTABLEMODE
), {STATE_RENDER(WINED3D_RS_FOGENABLE
), NULL
}, WINED3D_GL_EXT_NONE
},
6638 {STATE_RENDER(WINED3D_RS_FOGVERTEXMODE
), {STATE_RENDER(WINED3D_RS_FOGENABLE
), NULL
}, WINED3D_GL_EXT_NONE
},
6639 {STATE_RENDER(WINED3D_RS_FOGSTART
), {STATE_RENDER(WINED3D_RS_FOGSTART
), state_fogstartend
}, WINED3D_GL_EXT_NONE
},
6640 {STATE_RENDER(WINED3D_RS_FOGEND
), {STATE_RENDER(WINED3D_RS_FOGSTART
), NULL
}, WINED3D_GL_EXT_NONE
},
6641 {STATE_RENDER(WINED3D_RS_SRGBWRITEENABLE
), {STATE_RENDER(WINED3D_RS_SRGBWRITEENABLE
), state_srgbwrite
}, ARB_FRAMEBUFFER_SRGB
},
6642 {STATE_RENDER(WINED3D_RS_SRGBWRITEENABLE
), {STATE_PIXELSHADER
, NULL
}, WINED3D_GL_EXT_NONE
},
6643 {STATE_RENDER(WINED3D_RS_FOGCOLOR
), {STATE_RENDER(WINED3D_RS_FOGCOLOR
), state_fogcolor
}, WINED3D_GL_EXT_NONE
},
6644 {STATE_RENDER(WINED3D_RS_FOGDENSITY
), {STATE_RENDER(WINED3D_RS_FOGDENSITY
), state_fogdensity
}, WINED3D_GL_EXT_NONE
},
6645 {STATE_TEXTURESTAGE(0,WINED3D_TSS_TEXTURE_TRANSFORM_FLAGS
), {STATE_TEXTURESTAGE(0, WINED3D_TSS_TEXTURE_TRANSFORM_FLAGS
), glsl_fragment_pipe_tex_transform
}, WINED3D_GL_EXT_NONE
},
6646 {STATE_TEXTURESTAGE(1,WINED3D_TSS_TEXTURE_TRANSFORM_FLAGS
), {STATE_TEXTURESTAGE(1, WINED3D_TSS_TEXTURE_TRANSFORM_FLAGS
), glsl_fragment_pipe_tex_transform
}, WINED3D_GL_EXT_NONE
},
6647 {STATE_TEXTURESTAGE(2,WINED3D_TSS_TEXTURE_TRANSFORM_FLAGS
), {STATE_TEXTURESTAGE(2, WINED3D_TSS_TEXTURE_TRANSFORM_FLAGS
), glsl_fragment_pipe_tex_transform
}, WINED3D_GL_EXT_NONE
},
6648 {STATE_TEXTURESTAGE(3,WINED3D_TSS_TEXTURE_TRANSFORM_FLAGS
), {STATE_TEXTURESTAGE(3, WINED3D_TSS_TEXTURE_TRANSFORM_FLAGS
), glsl_fragment_pipe_tex_transform
}, WINED3D_GL_EXT_NONE
},
6649 {STATE_TEXTURESTAGE(4,WINED3D_TSS_TEXTURE_TRANSFORM_FLAGS
), {STATE_TEXTURESTAGE(4, WINED3D_TSS_TEXTURE_TRANSFORM_FLAGS
), glsl_fragment_pipe_tex_transform
}, WINED3D_GL_EXT_NONE
},
6650 {STATE_TEXTURESTAGE(5,WINED3D_TSS_TEXTURE_TRANSFORM_FLAGS
), {STATE_TEXTURESTAGE(5, WINED3D_TSS_TEXTURE_TRANSFORM_FLAGS
), glsl_fragment_pipe_tex_transform
}, WINED3D_GL_EXT_NONE
},
6651 {STATE_TEXTURESTAGE(6,WINED3D_TSS_TEXTURE_TRANSFORM_FLAGS
), {STATE_TEXTURESTAGE(6, WINED3D_TSS_TEXTURE_TRANSFORM_FLAGS
), glsl_fragment_pipe_tex_transform
}, WINED3D_GL_EXT_NONE
},
6652 {STATE_TEXTURESTAGE(7,WINED3D_TSS_TEXTURE_TRANSFORM_FLAGS
), {STATE_TEXTURESTAGE(7, WINED3D_TSS_TEXTURE_TRANSFORM_FLAGS
), glsl_fragment_pipe_tex_transform
}, WINED3D_GL_EXT_NONE
},
6653 {STATE_RENDER(WINED3D_RS_SPECULARENABLE
), {STATE_RENDER(WINED3D_RS_SPECULARENABLE
), glsl_fragment_pipe_invalidate_constants
}, WINED3D_GL_EXT_NONE
},
6654 {0 /* Terminate */, {0, 0 }, WINED3D_GL_EXT_NONE
},
6657 const struct fragment_pipeline glsl_fragment_pipe
=
6659 glsl_fragment_pipe_enable
,
6660 glsl_fragment_pipe_get_caps
,
6661 glsl_fragment_pipe_alloc
,
6662 glsl_fragment_pipe_free
,
6663 shader_glsl_color_fixup_supported
,
6664 glsl_fragment_pipe_state_template
,