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_RECT 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 fragment_pipeline
*fragment_pipe
;
101 struct wine_rb_tree ffp_fragment_shaders
;
102 BOOL ffp_proj_control
;
105 struct glsl_vs_program
107 struct list shader_entry
;
109 GLenum vertex_color_clamp
;
110 GLint
*uniform_f_locations
;
111 GLint uniform_i_locations
[MAX_CONST_I
];
112 GLint pos_fixup_location
;
115 struct glsl_gs_program
117 struct list shader_entry
;
121 struct glsl_ps_program
123 struct list shader_entry
;
125 GLint
*uniform_f_locations
;
126 GLint uniform_i_locations
[MAX_CONST_I
];
127 GLint bumpenv_mat_location
[MAX_TEXTURES
];
128 GLint bumpenv_lum_scale_location
[MAX_TEXTURES
];
129 GLint bumpenv_lum_offset_location
[MAX_TEXTURES
];
130 GLint tex_factor_location
;
131 GLint specular_enable_location
;
132 GLint ycorrection_location
;
133 GLint np2_fixup_location
;
134 const struct ps_np2fixup_info
*np2_fixup_info
;
137 /* Struct to maintain data about a linked GLSL program */
138 struct glsl_shader_prog_link
140 struct wine_rb_entry program_lookup_entry
;
141 struct glsl_vs_program vs
;
142 struct glsl_gs_program gs
;
143 struct glsl_ps_program ps
;
144 GLhandleARB programId
;
145 UINT constant_version
;
148 struct glsl_program_key
155 struct shader_glsl_ctx_priv
{
156 const struct vs_compile_args
*cur_vs_args
;
157 const struct ps_compile_args
*cur_ps_args
;
158 struct ps_np2fixup_info
*cur_np2fixup_info
;
161 struct glsl_ps_compiled_shader
163 struct ps_compile_args args
;
164 struct ps_np2fixup_info np2fixup
;
168 struct glsl_vs_compiled_shader
170 struct vs_compile_args args
;
174 struct glsl_gs_compiled_shader
179 struct glsl_shader_private
183 struct glsl_vs_compiled_shader
*vs
;
184 struct glsl_gs_compiled_shader
*gs
;
185 struct glsl_ps_compiled_shader
*ps
;
187 UINT num_gl_shaders
, shader_array_size
;
190 struct glsl_ffp_fragment_shader
192 struct ffp_frag_desc entry
;
194 struct list linked_programs
;
197 static const char *debug_gl_shader_type(GLenum type
)
201 #define WINED3D_TO_STR(u) case u: return #u
202 WINED3D_TO_STR(GL_VERTEX_SHADER_ARB
);
203 WINED3D_TO_STR(GL_GEOMETRY_SHADER_ARB
);
204 WINED3D_TO_STR(GL_FRAGMENT_SHADER_ARB
);
205 #undef WINED3D_TO_STR
207 return wine_dbg_sprintf("UNKNOWN(%#x)", type
);
211 static const char *shader_glsl_get_prefix(enum wined3d_shader_type type
)
215 case WINED3D_SHADER_TYPE_VERTEX
:
218 case WINED3D_SHADER_TYPE_GEOMETRY
:
221 case WINED3D_SHADER_TYPE_PIXEL
:
225 FIXME("Unhandled shader type %#x.\n", type
);
230 /* Extract a line from the info log.
231 * Note that this modifies the source string. */
232 static char *get_info_log_line(char **ptr
)
237 if (!(q
= strstr(p
, "\n")))
239 if (!*p
) return NULL
;
249 /* Context activation is done by the caller. */
250 static void print_glsl_info_log(const struct wined3d_gl_info
*gl_info
, GLhandleARB obj
)
252 int infologLength
= 0;
255 if (!WARN_ON(d3d_shader
) && !FIXME_ON(d3d_shader
))
258 GL_EXTCALL(glGetObjectParameterivARB(obj
,
259 GL_OBJECT_INFO_LOG_LENGTH_ARB
,
262 /* A size of 1 is just a null-terminated string, so the log should be bigger than
263 * that if there are errors. */
264 if (infologLength
> 1)
268 infoLog
= HeapAlloc(GetProcessHeap(), 0, infologLength
);
269 /* The info log is supposed to be zero-terminated, but at least some
270 * versions of fglrx don't terminate the string properly. The reported
271 * length does include the terminator, so explicitly set it to zero
273 infoLog
[infologLength
- 1] = 0;
274 GL_EXTCALL(glGetInfoLogARB(obj
, infologLength
, NULL
, infoLog
));
277 if (gl_info
->quirks
& WINED3D_QUIRK_INFO_LOG_SPAM
)
279 WARN("Info log received from GLSL shader #%u:\n", obj
);
280 while ((line
= get_info_log_line(&ptr
))) WARN(" %s\n", line
);
284 FIXME("Info log received from GLSL shader #%u:\n", obj
);
285 while ((line
= get_info_log_line(&ptr
))) FIXME(" %s\n", line
);
287 HeapFree(GetProcessHeap(), 0, infoLog
);
291 /* Context activation is done by the caller. */
292 static void shader_glsl_compile(const struct wined3d_gl_info
*gl_info
, GLhandleARB shader
, const char *src
)
294 TRACE("Compiling shader object %u.\n", shader
);
295 GL_EXTCALL(glShaderSourceARB(shader
, 1, &src
, NULL
));
296 checkGLcall("glShaderSourceARB");
297 GL_EXTCALL(glCompileShaderARB(shader
));
298 checkGLcall("glCompileShaderARB");
299 print_glsl_info_log(gl_info
, shader
);
302 /* Context activation is done by the caller. */
303 static void shader_glsl_dump_program_source(const struct wined3d_gl_info
*gl_info
, GLhandleARB program
)
305 GLint i
, object_count
, source_size
= -1;
306 GLhandleARB
*objects
;
309 GL_EXTCALL(glGetObjectParameterivARB(program
, GL_OBJECT_ATTACHED_OBJECTS_ARB
, &object_count
));
310 objects
= HeapAlloc(GetProcessHeap(), 0, object_count
* sizeof(*objects
));
313 ERR("Failed to allocate object array memory.\n");
317 GL_EXTCALL(glGetAttachedObjectsARB(program
, object_count
, NULL
, objects
));
318 for (i
= 0; i
< object_count
; ++i
)
323 GL_EXTCALL(glGetObjectParameterivARB(objects
[i
], GL_OBJECT_SHADER_SOURCE_LENGTH_ARB
, &tmp
));
325 if (source_size
< tmp
)
327 HeapFree(GetProcessHeap(), 0, source
);
329 source
= HeapAlloc(GetProcessHeap(), HEAP_ZERO_MEMORY
, tmp
);
332 ERR("Failed to allocate %d bytes for shader source.\n", tmp
);
333 HeapFree(GetProcessHeap(), 0, objects
);
339 FIXME("Object %u:\n", objects
[i
]);
340 GL_EXTCALL(glGetObjectParameterivARB(objects
[i
], GL_OBJECT_SUBTYPE_ARB
, &tmp
));
341 FIXME(" GL_OBJECT_SUBTYPE_ARB: %s.\n", debug_gl_shader_type(tmp
));
342 GL_EXTCALL(glGetObjectParameterivARB(objects
[i
], GL_OBJECT_COMPILE_STATUS_ARB
, &tmp
));
343 FIXME(" GL_OBJECT_COMPILE_STATUS_ARB: %d.\n", tmp
);
347 GL_EXTCALL(glGetShaderSourceARB(objects
[i
], source_size
, NULL
, source
));
348 while ((line
= get_info_log_line(&ptr
))) FIXME(" %s\n", line
);
352 HeapFree(GetProcessHeap(), 0, source
);
353 HeapFree(GetProcessHeap(), 0, objects
);
356 /* Context activation is done by the caller. */
357 static void shader_glsl_validate_link(const struct wined3d_gl_info
*gl_info
, GLhandleARB program
)
361 if (!TRACE_ON(d3d_shader
) && !FIXME_ON(d3d_shader
)) return;
363 GL_EXTCALL(glGetObjectParameterivARB(program
, GL_OBJECT_TYPE_ARB
, &tmp
));
364 if (tmp
== GL_PROGRAM_OBJECT_ARB
)
366 GL_EXTCALL(glGetObjectParameterivARB(program
, GL_OBJECT_LINK_STATUS_ARB
, &tmp
));
369 FIXME("Program %u link status invalid.\n", program
);
370 shader_glsl_dump_program_source(gl_info
, program
);
374 print_glsl_info_log(gl_info
, program
);
377 /* Context activation is done by the caller. */
378 static void shader_glsl_load_psamplers(const struct wined3d_gl_info
*gl_info
,
379 const DWORD
*tex_unit_map
, GLhandleARB programId
)
382 char sampler_name
[20];
385 for (i
= 0; i
< MAX_FRAGMENT_SAMPLERS
; ++i
)
387 snprintf(sampler_name
, sizeof(sampler_name
), "ps_sampler%u", i
);
388 name_loc
= GL_EXTCALL(glGetUniformLocationARB(programId
, sampler_name
));
389 if (name_loc
!= -1) {
390 DWORD mapped_unit
= tex_unit_map
[i
];
391 if (mapped_unit
!= WINED3D_UNMAPPED_STAGE
&& mapped_unit
< gl_info
->limits
.fragment_samplers
)
393 TRACE("Loading %s for texture %d\n", sampler_name
, mapped_unit
);
394 GL_EXTCALL(glUniform1iARB(name_loc
, mapped_unit
));
395 checkGLcall("glUniform1iARB");
397 ERR("Trying to load sampler %s on unsupported unit %d\n", sampler_name
, mapped_unit
);
403 /* Context activation is done by the caller. */
404 static void shader_glsl_load_vsamplers(const struct wined3d_gl_info
*gl_info
,
405 const DWORD
*tex_unit_map
, GLhandleARB programId
)
408 char sampler_name
[20];
411 for (i
= 0; i
< MAX_VERTEX_SAMPLERS
; ++i
)
413 snprintf(sampler_name
, sizeof(sampler_name
), "vs_sampler%u", i
);
414 name_loc
= GL_EXTCALL(glGetUniformLocationARB(programId
, sampler_name
));
415 if (name_loc
!= -1) {
416 DWORD mapped_unit
= tex_unit_map
[MAX_FRAGMENT_SAMPLERS
+ i
];
417 if (mapped_unit
!= WINED3D_UNMAPPED_STAGE
&& mapped_unit
< gl_info
->limits
.combined_samplers
)
419 TRACE("Loading %s for texture %d\n", sampler_name
, mapped_unit
);
420 GL_EXTCALL(glUniform1iARB(name_loc
, mapped_unit
));
421 checkGLcall("glUniform1iARB");
423 ERR("Trying to load sampler %s on unsupported unit %d\n", sampler_name
, mapped_unit
);
429 /* Context activation is done by the caller. */
430 static inline void walk_constant_heap(const struct wined3d_gl_info
*gl_info
, const float *constants
,
431 const GLint
*constant_locations
, const struct constant_heap
*heap
, unsigned char *stack
, DWORD version
)
434 unsigned int heap_idx
= 1;
437 if (heap
->entries
[heap_idx
].version
<= version
) return;
439 idx
= heap
->entries
[heap_idx
].idx
;
440 if (constant_locations
[idx
] != -1) GL_EXTCALL(glUniform4fvARB(constant_locations
[idx
], 1, &constants
[idx
* 4]));
441 stack
[stack_idx
] = HEAP_NODE_TRAVERSE_LEFT
;
443 while (stack_idx
>= 0)
445 /* Note that we fall through to the next case statement. */
446 switch(stack
[stack_idx
])
448 case HEAP_NODE_TRAVERSE_LEFT
:
450 unsigned int left_idx
= heap_idx
<< 1;
451 if (left_idx
< heap
->size
&& heap
->entries
[left_idx
].version
> version
)
454 idx
= heap
->entries
[heap_idx
].idx
;
455 if (constant_locations
[idx
] != -1)
456 GL_EXTCALL(glUniform4fvARB(constant_locations
[idx
], 1, &constants
[idx
* 4]));
458 stack
[stack_idx
++] = HEAP_NODE_TRAVERSE_RIGHT
;
459 stack
[stack_idx
] = HEAP_NODE_TRAVERSE_LEFT
;
464 case HEAP_NODE_TRAVERSE_RIGHT
:
466 unsigned int right_idx
= (heap_idx
<< 1) + 1;
467 if (right_idx
< heap
->size
&& heap
->entries
[right_idx
].version
> version
)
469 heap_idx
= right_idx
;
470 idx
= heap
->entries
[heap_idx
].idx
;
471 if (constant_locations
[idx
] != -1)
472 GL_EXTCALL(glUniform4fvARB(constant_locations
[idx
], 1, &constants
[idx
* 4]));
474 stack
[stack_idx
++] = HEAP_NODE_POP
;
475 stack
[stack_idx
] = HEAP_NODE_TRAVERSE_LEFT
;
486 checkGLcall("walk_constant_heap()");
489 /* Context activation is done by the caller. */
490 static inline void apply_clamped_constant(const struct wined3d_gl_info
*gl_info
, GLint location
, const GLfloat
*data
)
492 GLfloat clamped_constant
[4];
494 if (location
== -1) return;
496 clamped_constant
[0] = data
[0] < -1.0f
? -1.0f
: data
[0] > 1.0f
? 1.0f
: data
[0];
497 clamped_constant
[1] = data
[1] < -1.0f
? -1.0f
: data
[1] > 1.0f
? 1.0f
: data
[1];
498 clamped_constant
[2] = data
[2] < -1.0f
? -1.0f
: data
[2] > 1.0f
? 1.0f
: data
[2];
499 clamped_constant
[3] = data
[3] < -1.0f
? -1.0f
: data
[3] > 1.0f
? 1.0f
: data
[3];
501 GL_EXTCALL(glUniform4fvARB(location
, 1, clamped_constant
));
504 /* Context activation is done by the caller. */
505 static inline void walk_constant_heap_clamped(const struct wined3d_gl_info
*gl_info
, const float *constants
,
506 const GLint
*constant_locations
, const struct constant_heap
*heap
, unsigned char *stack
, DWORD version
)
509 unsigned int heap_idx
= 1;
512 if (heap
->entries
[heap_idx
].version
<= version
) return;
514 idx
= heap
->entries
[heap_idx
].idx
;
515 apply_clamped_constant(gl_info
, constant_locations
[idx
], &constants
[idx
* 4]);
516 stack
[stack_idx
] = HEAP_NODE_TRAVERSE_LEFT
;
518 while (stack_idx
>= 0)
520 /* Note that we fall through to the next case statement. */
521 switch(stack
[stack_idx
])
523 case HEAP_NODE_TRAVERSE_LEFT
:
525 unsigned int left_idx
= heap_idx
<< 1;
526 if (left_idx
< heap
->size
&& heap
->entries
[left_idx
].version
> version
)
529 idx
= heap
->entries
[heap_idx
].idx
;
530 apply_clamped_constant(gl_info
, constant_locations
[idx
], &constants
[idx
* 4]);
532 stack
[stack_idx
++] = HEAP_NODE_TRAVERSE_RIGHT
;
533 stack
[stack_idx
] = HEAP_NODE_TRAVERSE_LEFT
;
538 case HEAP_NODE_TRAVERSE_RIGHT
:
540 unsigned int right_idx
= (heap_idx
<< 1) + 1;
541 if (right_idx
< heap
->size
&& heap
->entries
[right_idx
].version
> version
)
543 heap_idx
= right_idx
;
544 idx
= heap
->entries
[heap_idx
].idx
;
545 apply_clamped_constant(gl_info
, constant_locations
[idx
], &constants
[idx
* 4]);
547 stack
[stack_idx
++] = HEAP_NODE_POP
;
548 stack
[stack_idx
] = HEAP_NODE_TRAVERSE_LEFT
;
559 checkGLcall("walk_constant_heap_clamped()");
562 /* Context activation is done by the caller. */
563 static void shader_glsl_load_constantsF(const struct wined3d_shader
*shader
, const struct wined3d_gl_info
*gl_info
,
564 const float *constants
, const GLint
*constant_locations
, const struct constant_heap
*heap
,
565 unsigned char *stack
, UINT version
)
567 const struct wined3d_shader_lconst
*lconst
;
569 /* 1.X pshaders have the constants clamped to [-1;1] implicitly. */
570 if (shader
->reg_maps
.shader_version
.major
== 1
571 && shader
->reg_maps
.shader_version
.type
== WINED3D_SHADER_TYPE_PIXEL
)
572 walk_constant_heap_clamped(gl_info
, constants
, constant_locations
, heap
, stack
, version
);
574 walk_constant_heap(gl_info
, constants
, constant_locations
, heap
, stack
, version
);
576 if (!shader
->load_local_constsF
)
578 TRACE("No need to load local float constants for this shader\n");
582 /* Immediate constants are clamped to [-1;1] at shader creation time if needed */
583 LIST_FOR_EACH_ENTRY(lconst
, &shader
->constantsF
, struct wined3d_shader_lconst
, entry
)
585 GLint location
= constant_locations
[lconst
->idx
];
586 /* We found this uniform name in the program - go ahead and send the data */
587 if (location
!= -1) GL_EXTCALL(glUniform4fvARB(location
, 1, (const GLfloat
*)lconst
->value
));
589 checkGLcall("glUniform4fvARB()");
592 /* Context activation is done by the caller. */
593 static void shader_glsl_load_constantsI(const struct wined3d_shader
*shader
, const struct wined3d_gl_info
*gl_info
,
594 const GLint locations
[MAX_CONST_I
], const int *constants
, WORD constants_set
)
599 for (i
= 0; constants_set
; constants_set
>>= 1, ++i
)
601 if (!(constants_set
& 1)) continue;
603 TRACE_(d3d_constants
)("Loading constants %u: %i, %i, %i, %i\n",
604 i
, constants
[i
*4], constants
[i
*4+1], constants
[i
*4+2], constants
[i
*4+3]);
606 /* We found this uniform name in the program - go ahead and send the data */
607 GL_EXTCALL(glUniform4ivARB(locations
[i
], 1, &constants
[i
*4]));
608 checkGLcall("glUniform4ivARB");
611 /* Load immediate constants */
612 ptr
= list_head(&shader
->constantsI
);
615 const struct wined3d_shader_lconst
*lconst
= LIST_ENTRY(ptr
, const struct wined3d_shader_lconst
, entry
);
616 unsigned int idx
= lconst
->idx
;
617 const GLint
*values
= (const GLint
*)lconst
->value
;
619 TRACE_(d3d_constants
)("Loading local constants %i: %i, %i, %i, %i\n", idx
,
620 values
[0], values
[1], values
[2], values
[3]);
622 /* We found this uniform name in the program - go ahead and send the data */
623 GL_EXTCALL(glUniform4ivARB(locations
[idx
], 1, values
));
624 checkGLcall("glUniform4ivARB");
625 ptr
= list_next(&shader
->constantsI
, ptr
);
629 /* Context activation is done by the caller. */
630 static void shader_glsl_load_constantsB(const struct wined3d_shader
*shader
, const struct wined3d_gl_info
*gl_info
,
631 GLhandleARB programId
, const BOOL
*constants
, WORD constants_set
)
639 prefix
= shader_glsl_get_prefix(shader
->reg_maps
.shader_version
.type
);
641 /* TODO: Benchmark and see if it would be beneficial to store the
642 * locations of the constants to avoid looking up each time */
643 for (i
= 0; constants_set
; constants_set
>>= 1, ++i
)
645 if (!(constants_set
& 1)) continue;
647 TRACE_(d3d_constants
)("Loading constants %i: %i;\n", i
, constants
[i
]);
649 /* TODO: Benchmark and see if it would be beneficial to store the
650 * locations of the constants to avoid looking up each time */
651 snprintf(tmp_name
, sizeof(tmp_name
), "%s_b[%i]", prefix
, i
);
652 tmp_loc
= GL_EXTCALL(glGetUniformLocationARB(programId
, tmp_name
));
655 /* We found this uniform name in the program - go ahead and send the data */
656 GL_EXTCALL(glUniform1ivARB(tmp_loc
, 1, &constants
[i
]));
657 checkGLcall("glUniform1ivARB");
661 /* Load immediate constants */
662 ptr
= list_head(&shader
->constantsB
);
665 const struct wined3d_shader_lconst
*lconst
= LIST_ENTRY(ptr
, const struct wined3d_shader_lconst
, entry
);
666 unsigned int idx
= lconst
->idx
;
667 const GLint
*values
= (const GLint
*)lconst
->value
;
669 TRACE_(d3d_constants
)("Loading local constants %i: %i\n", idx
, values
[0]);
671 snprintf(tmp_name
, sizeof(tmp_name
), "%s_b[%i]", prefix
, idx
);
672 tmp_loc
= GL_EXTCALL(glGetUniformLocationARB(programId
, tmp_name
));
674 /* We found this uniform name in the program - go ahead and send the data */
675 GL_EXTCALL(glUniform1ivARB(tmp_loc
, 1, values
));
676 checkGLcall("glUniform1ivARB");
678 ptr
= list_next(&shader
->constantsB
, ptr
);
682 static void reset_program_constant_version(struct wine_rb_entry
*entry
, void *context
)
684 WINE_RB_ENTRY_VALUE(entry
, struct glsl_shader_prog_link
, program_lookup_entry
)->constant_version
= 0;
687 /* Context activation is done by the caller (state handler). */
688 static void shader_glsl_load_np2fixup_constants(void *shader_priv
,
689 const struct wined3d_gl_info
*gl_info
, const struct wined3d_state
*state
)
691 struct shader_glsl_priv
*glsl_priv
= shader_priv
;
692 const struct glsl_shader_prog_link
*prog
= glsl_priv
->glsl_program
;
694 /* No GLSL program set - nothing to do. */
697 /* NP2 texcoord fixup is (currently) only done for pixelshaders. */
698 if (!use_ps(state
)) return;
700 if (prog
->ps
.np2_fixup_info
&& prog
->ps
.np2_fixup_location
!= -1)
703 UINT fixup
= prog
->ps
.np2_fixup_info
->active
;
704 GLfloat np2fixup_constants
[4 * MAX_FRAGMENT_SAMPLERS
];
706 for (i
= 0; fixup
; fixup
>>= 1, ++i
)
708 const struct wined3d_texture
*tex
= state
->textures
[i
];
709 const unsigned char idx
= prog
->ps
.np2_fixup_info
->idx
[i
];
710 GLfloat
*tex_dim
= &np2fixup_constants
[(idx
>> 1) * 4];
714 ERR("Nonexistent texture is flagged for NP2 texcoord fixup.\n");
720 tex_dim
[2] = tex
->pow2_matrix
[0];
721 tex_dim
[3] = tex
->pow2_matrix
[5];
725 tex_dim
[0] = tex
->pow2_matrix
[0];
726 tex_dim
[1] = tex
->pow2_matrix
[5];
730 GL_EXTCALL(glUniform4fvARB(prog
->ps
.np2_fixup_location
,
731 prog
->ps
.np2_fixup_info
->num_consts
, np2fixup_constants
));
735 /* Context activation is done by the caller (state handler). */
736 static void shader_glsl_load_constants(const struct wined3d_context
*context
,
737 BOOL usePixelShader
, BOOL useVertexShader
)
739 const struct wined3d_gl_info
*gl_info
= context
->gl_info
;
740 struct wined3d_device
*device
= context
->swapchain
->device
;
741 struct wined3d_stateblock
*stateBlock
= device
->stateBlock
;
742 const struct wined3d_state
*state
= &stateBlock
->state
;
743 struct shader_glsl_priv
*priv
= device
->shader_priv
;
744 float position_fixup
[4];
746 GLhandleARB programId
;
747 struct glsl_shader_prog_link
*prog
= priv
->glsl_program
;
748 UINT constant_version
;
752 /* No GLSL program set - nothing to do. */
755 programId
= prog
->programId
;
756 constant_version
= prog
->constant_version
;
760 const struct wined3d_shader
*vshader
= state
->vertex_shader
;
762 /* Load DirectX 9 float constants/uniforms for vertex shader */
763 shader_glsl_load_constantsF(vshader
, gl_info
, state
->vs_consts_f
,
764 prog
->vs
.uniform_f_locations
, &priv
->vconst_heap
, priv
->stack
, constant_version
);
766 /* Load DirectX 9 integer constants/uniforms for vertex shader */
767 shader_glsl_load_constantsI(vshader
, gl_info
, prog
->vs
.uniform_i_locations
, state
->vs_consts_i
,
768 stateBlock
->changed
.vertexShaderConstantsI
& vshader
->reg_maps
.integer_constants
);
770 /* Load DirectX 9 boolean constants/uniforms for vertex shader */
771 shader_glsl_load_constantsB(vshader
, gl_info
, programId
, state
->vs_consts_b
,
772 stateBlock
->changed
.vertexShaderConstantsB
& vshader
->reg_maps
.boolean_constants
);
774 /* Upload the position fixup params */
775 shader_get_position_fixup(context
, state
, position_fixup
);
776 GL_EXTCALL(glUniform4fvARB(prog
->vs
.pos_fixup_location
, 1, position_fixup
));
777 checkGLcall("glUniform4fvARB");
782 const struct wined3d_shader
*pshader
= state
->pixel_shader
;
784 /* Load DirectX 9 float constants/uniforms for pixel shader */
785 shader_glsl_load_constantsF(pshader
, gl_info
, state
->ps_consts_f
,
786 prog
->ps
.uniform_f_locations
, &priv
->pconst_heap
, priv
->stack
, constant_version
);
788 /* Load DirectX 9 integer constants/uniforms for pixel shader */
789 shader_glsl_load_constantsI(pshader
, gl_info
, prog
->ps
.uniform_i_locations
, state
->ps_consts_i
,
790 stateBlock
->changed
.pixelShaderConstantsI
& pshader
->reg_maps
.integer_constants
);
792 /* Load DirectX 9 boolean constants/uniforms for pixel shader */
793 shader_glsl_load_constantsB(pshader
, gl_info
, programId
, state
->ps_consts_b
,
794 stateBlock
->changed
.pixelShaderConstantsB
& pshader
->reg_maps
.boolean_constants
);
796 /* Upload the environment bump map matrix if needed. The needsbumpmat
797 * member specifies the texture stage to load the matrix from. It
798 * can't be 0 for a valid texbem instruction. */
799 for (i
= 0; i
< MAX_TEXTURES
; ++i
)
803 if (prog
->ps
.bumpenv_mat_location
[i
] == -1)
806 data
= (const float *)&state
->texture_states
[i
][WINED3D_TSS_BUMPENV_MAT00
];
807 GL_EXTCALL(glUniformMatrix2fvARB(prog
->ps
.bumpenv_mat_location
[i
], 1, 0, data
));
808 checkGLcall("glUniformMatrix2fvARB");
810 /* texbeml needs the luminance scale and offset too. If texbeml
811 * is used, needsbumpmat is set too, so we can check that in the
812 * needsbumpmat check. */
813 if (prog
->ps
.bumpenv_lum_scale_location
[i
] != -1)
815 const GLfloat
*scale
= (const GLfloat
*)&state
->texture_states
[i
][WINED3D_TSS_BUMPENV_LSCALE
];
816 const GLfloat
*offset
= (const GLfloat
*)&state
->texture_states
[i
][WINED3D_TSS_BUMPENV_LOFFSET
];
818 GL_EXTCALL(glUniform1fvARB(prog
->ps
.bumpenv_lum_scale_location
[i
], 1, scale
));
819 checkGLcall("glUniform1fvARB");
820 GL_EXTCALL(glUniform1fvARB(prog
->ps
.bumpenv_lum_offset_location
[i
], 1, offset
));
821 checkGLcall("glUniform1fvARB");
825 if (prog
->ps
.ycorrection_location
!= -1)
827 float correction_params
[4];
829 if (context
->render_offscreen
)
831 correction_params
[0] = 0.0f
;
832 correction_params
[1] = 1.0f
;
834 /* position is window relative, not viewport relative */
835 correction_params
[0] = (float) context
->current_rt
->resource
.height
;
836 correction_params
[1] = -1.0f
;
838 GL_EXTCALL(glUniform4fvARB(prog
->ps
.ycorrection_location
, 1, correction_params
));
841 else if (priv
->fragment_pipe
== &glsl_fragment_pipe
)
845 for (i
= 0; i
< MAX_TEXTURES
; ++i
)
847 GL_EXTCALL(glUniformMatrix2fvARB(prog
->ps
.bumpenv_mat_location
[i
], 1, 0,
848 (const float *)&state
->texture_states
[i
][WINED3D_TSS_BUMPENV_MAT00
]));
849 GL_EXTCALL(glUniform1fARB(prog
->ps
.bumpenv_lum_scale_location
[i
],
850 *(const float *)&state
->texture_states
[i
][WINED3D_TSS_BUMPENV_LSCALE
]));
851 GL_EXTCALL(glUniform1fARB(prog
->ps
.bumpenv_lum_offset_location
[i
],
852 *(const float *)&state
->texture_states
[i
][WINED3D_TSS_BUMPENV_LOFFSET
]));
855 D3DCOLORTOGLFLOAT4(state
->render_states
[WINED3D_RS_TEXTUREFACTOR
], col
);
856 GL_EXTCALL(glUniform4fARB(prog
->ps
.tex_factor_location
, col
[0], col
[1], col
[2], col
[3]));
858 if (state
->render_states
[WINED3D_RS_SPECULARENABLE
])
859 GL_EXTCALL(glUniform4fARB(prog
->ps
.specular_enable_location
, 1.0f
, 1.0f
, 1.0f
, 0.0f
));
861 GL_EXTCALL(glUniform4fARB(prog
->ps
.specular_enable_location
, 0.0f
, 0.0f
, 0.0f
, 0.0f
));
863 checkGLcall("fixed function uniforms");
866 if (priv
->next_constant_version
== UINT_MAX
)
868 TRACE("Max constant version reached, resetting to 0.\n");
869 wine_rb_for_each_entry(&priv
->program_lookup
, reset_program_constant_version
, NULL
);
870 priv
->next_constant_version
= 1;
874 prog
->constant_version
= priv
->next_constant_version
++;
878 static void update_heap_entry(const struct constant_heap
*heap
, unsigned int idx
,
879 unsigned int heap_idx
, DWORD new_version
)
881 struct constant_entry
*entries
= heap
->entries
;
882 unsigned int *positions
= heap
->positions
;
883 unsigned int parent_idx
;
887 parent_idx
= heap_idx
>> 1;
889 if (new_version
<= entries
[parent_idx
].version
) break;
891 entries
[heap_idx
] = entries
[parent_idx
];
892 positions
[entries
[parent_idx
].idx
] = heap_idx
;
893 heap_idx
= parent_idx
;
896 entries
[heap_idx
].version
= new_version
;
897 entries
[heap_idx
].idx
= idx
;
898 positions
[idx
] = heap_idx
;
901 static void shader_glsl_update_float_vertex_constants(struct wined3d_device
*device
, UINT start
, UINT count
)
903 struct shader_glsl_priv
*priv
= device
->shader_priv
;
904 struct constant_heap
*heap
= &priv
->vconst_heap
;
907 for (i
= start
; i
< count
+ start
; ++i
)
909 if (!device
->stateBlock
->changed
.vertexShaderConstantsF
[i
])
910 update_heap_entry(heap
, i
, heap
->size
++, priv
->next_constant_version
);
912 update_heap_entry(heap
, i
, heap
->positions
[i
], priv
->next_constant_version
);
916 static void shader_glsl_update_float_pixel_constants(struct wined3d_device
*device
, UINT start
, UINT count
)
918 struct shader_glsl_priv
*priv
= device
->shader_priv
;
919 struct constant_heap
*heap
= &priv
->pconst_heap
;
922 for (i
= start
; i
< count
+ start
; ++i
)
924 if (!device
->stateBlock
->changed
.pixelShaderConstantsF
[i
])
925 update_heap_entry(heap
, i
, heap
->size
++, priv
->next_constant_version
);
927 update_heap_entry(heap
, i
, heap
->positions
[i
], priv
->next_constant_version
);
931 static unsigned int vec4_varyings(DWORD shader_major
, const struct wined3d_gl_info
*gl_info
)
933 unsigned int ret
= gl_info
->limits
.glsl_varyings
/ 4;
934 /* 4.0 shaders do not write clip coords because d3d10 does not support user clipplanes */
935 if(shader_major
> 3) return ret
;
937 /* 3.0 shaders may need an extra varying for the clip coord on some cards(mostly dx10 ones) */
938 if (gl_info
->quirks
& WINED3D_QUIRK_GLSL_CLIP_VARYING
) ret
-= 1;
942 /** Generate the variable & register declarations for the GLSL output target */
943 static void shader_generate_glsl_declarations(const struct wined3d_context
*context
,
944 struct wined3d_shader_buffer
*buffer
, const struct wined3d_shader
*shader
,
945 const struct wined3d_shader_reg_maps
*reg_maps
, const struct shader_glsl_ctx_priv
*ctx_priv
)
947 const struct wined3d_shader_version
*version
= ®_maps
->shader_version
;
948 const struct wined3d_state
*state
= &shader
->device
->stateBlock
->state
;
949 const struct ps_compile_args
*ps_args
= ctx_priv
->cur_ps_args
;
950 const struct wined3d_gl_info
*gl_info
= context
->gl_info
;
951 const struct wined3d_fb_state
*fb
= &shader
->device
->fb
;
952 unsigned int i
, extra_constants_needed
= 0;
953 const struct wined3d_shader_lconst
*lconst
;
957 prefix
= shader_glsl_get_prefix(version
->type
);
959 /* Prototype the subroutines */
960 for (i
= 0, map
= reg_maps
->labels
; map
; map
>>= 1, ++i
)
962 if (map
& 1) shader_addline(buffer
, "void subroutine%u();\n", i
);
965 /* Declare the constants (aka uniforms) */
966 if (shader
->limits
.constant_float
> 0)
968 unsigned max_constantsF
;
970 /* Unless the shader uses indirect addressing, always declare the
971 * maximum array size and ignore that we need some uniforms privately.
972 * E.g. if GL supports 256 uniforms, and we need 2 for the pos fixup
973 * and immediate values, still declare VC[256]. If the shader needs
974 * more uniforms than we have it won't work in any case. If it uses
975 * less, the compiler will figure out which uniforms are really used
976 * and strip them out. This allows a shader to use c255 on a dx9 card,
977 * as long as it doesn't also use all the other constants.
979 * If the shader uses indirect addressing the compiler must assume
980 * that all declared uniforms are used. In this case, declare only the
981 * amount that we're assured to have.
983 * Thus we run into problems in these two cases:
984 * 1) The shader really uses more uniforms than supported.
985 * 2) The shader uses indirect addressing, less constants than
986 * supported, but uses a constant index > #supported consts. */
987 if (version
->type
== WINED3D_SHADER_TYPE_PIXEL
)
989 /* No indirect addressing here. */
990 max_constantsF
= gl_info
->limits
.glsl_ps_float_constants
;
994 if (reg_maps
->usesrelconstF
)
996 /* Subtract the other potential uniforms from the max
997 * available (bools, ints, and 1 row of projection matrix).
998 * Subtract another uniform for immediate values, which have
999 * to be loaded via uniform by the driver as well. The shader
1000 * code only uses 0.5, 2.0, 1.0, 128 and -128 in vertex
1001 * shader code, so one vec4 should be enough. (Unfortunately
1002 * the Nvidia driver doesn't store 128 and -128 in one float).
1004 * Writing gl_ClipVertex requires one uniform for each
1005 * clipplane as well. */
1006 max_constantsF
= gl_info
->limits
.glsl_vs_float_constants
- 3;
1007 if(ctx_priv
->cur_vs_args
->clip_enabled
)
1009 max_constantsF
-= gl_info
->limits
.clipplanes
;
1011 max_constantsF
-= count_bits(reg_maps
->integer_constants
);
1012 /* Strictly speaking a bool only uses one scalar, but the nvidia(Linux) compiler doesn't pack them properly,
1013 * so each scalar requires a full vec4. We could work around this by packing the booleans ourselves, but
1014 * for now take this into account when calculating the number of available constants
1016 max_constantsF
-= count_bits(reg_maps
->boolean_constants
);
1017 /* Set by driver quirks in directx.c */
1018 max_constantsF
-= gl_info
->reserved_glsl_constants
;
1020 if (max_constantsF
< shader
->limits
.constant_float
)
1022 static unsigned int once
;
1025 ERR_(winediag
)("The hardware does not support enough uniform components to run this shader,"
1026 " it may not render correctly.\n");
1028 WARN("The hardware does not support enough uniform components to run this shader.\n");
1033 max_constantsF
= gl_info
->limits
.glsl_vs_float_constants
;
1036 max_constantsF
= min(shader
->limits
.constant_float
, max_constantsF
);
1037 shader_addline(buffer
, "uniform vec4 %s_c[%u];\n", prefix
, max_constantsF
);
1040 /* Always declare the full set of constants, the compiler can remove the
1041 * unused ones because d3d doesn't (yet) support indirect int and bool
1042 * constant addressing. This avoids problems if the app uses e.g. i0 and i9. */
1043 if (shader
->limits
.constant_int
> 0 && reg_maps
->integer_constants
)
1044 shader_addline(buffer
, "uniform ivec4 %s_i[%u];\n", prefix
, shader
->limits
.constant_int
);
1046 if (shader
->limits
.constant_bool
> 0 && reg_maps
->boolean_constants
)
1047 shader_addline(buffer
, "uniform bool %s_b[%u];\n", prefix
, shader
->limits
.constant_bool
);
1049 for (i
= 0; i
< WINED3D_MAX_CBS
; ++i
)
1051 if (reg_maps
->cb_sizes
[i
])
1052 shader_addline(buffer
, "uniform vec4 %s_cb%u[%u];\n", prefix
, i
, reg_maps
->cb_sizes
[i
]);
1055 /* Declare texture samplers */
1056 for (i
= 0; i
< shader
->limits
.sampler
; ++i
)
1058 if (reg_maps
->sampler_type
[i
])
1060 BOOL shadow_sampler
= version
->type
== WINED3D_SHADER_TYPE_PIXEL
&& (ps_args
->shadow
& (1 << i
));
1061 const struct wined3d_texture
*texture
;
1063 switch (reg_maps
->sampler_type
[i
])
1067 shader_addline(buffer
, "uniform sampler1DShadow %s_sampler%u;\n", prefix
, i
);
1069 shader_addline(buffer
, "uniform sampler1D %s_sampler%u;\n", prefix
, i
);
1072 texture
= state
->textures
[i
];
1075 if (texture
&& texture
->target
== GL_TEXTURE_RECTANGLE_ARB
)
1076 shader_addline(buffer
, "uniform sampler2DRectShadow %s_sampler%u;\n", prefix
, i
);
1078 shader_addline(buffer
, "uniform sampler2DShadow %s_sampler%u;\n", prefix
, i
);
1082 if (texture
&& texture
->target
== GL_TEXTURE_RECTANGLE_ARB
)
1083 shader_addline(buffer
, "uniform sampler2DRect %s_sampler%u;\n", prefix
, i
);
1085 shader_addline(buffer
, "uniform sampler2D %s_sampler%u;\n", prefix
, i
);
1088 case WINED3DSTT_CUBE
:
1090 FIXME("Unsupported Cube shadow sampler.\n");
1091 shader_addline(buffer
, "uniform samplerCube %s_sampler%u;\n", prefix
, i
);
1093 case WINED3DSTT_VOLUME
:
1095 FIXME("Unsupported 3D shadow sampler.\n");
1096 shader_addline(buffer
, "uniform sampler3D %s_sampler%u;\n", prefix
, i
);
1099 shader_addline(buffer
, "uniform unsupported_sampler %s_sampler%u;\n", prefix
, i
);
1100 FIXME("Unrecognized sampler type: %#x\n", reg_maps
->sampler_type
[i
]);
1106 /* Declare uniforms for NP2 texcoord fixup:
1107 * This is NOT done inside the loop that declares the texture samplers
1108 * since the NP2 fixup code is currently only used for the GeforceFX
1109 * series and when forcing the ARB_npot extension off. Modern cards just
1110 * skip the code anyway, so put it inside a separate loop. */
1111 if (version
->type
== WINED3D_SHADER_TYPE_PIXEL
&& ps_args
->np2_fixup
)
1113 struct ps_np2fixup_info
*fixup
= ctx_priv
->cur_np2fixup_info
;
1116 /* NP2/RECT textures in OpenGL use texcoords in the range [0,width]x[0,height]
1117 * while D3D has them in the (normalized) [0,1]x[0,1] range.
1118 * samplerNP2Fixup stores texture dimensions and is updated through
1119 * shader_glsl_load_np2fixup_constants when the sampler changes. */
1121 for (i
= 0; i
< shader
->limits
.sampler
; ++i
)
1123 if (reg_maps
->sampler_type
[i
])
1125 if (!(ps_args
->np2_fixup
& (1 << i
))) continue;
1127 if (WINED3DSTT_2D
!= reg_maps
->sampler_type
[i
]) {
1128 FIXME("Non-2D texture is flagged for NP2 texcoord fixup.\n");
1132 fixup
->idx
[i
] = cur
++;
1136 fixup
->num_consts
= (cur
+ 1) >> 1;
1137 fixup
->active
= ps_args
->np2_fixup
;
1138 shader_addline(buffer
, "uniform vec4 %s_samplerNP2Fixup[%u];\n", prefix
, fixup
->num_consts
);
1141 /* Declare address variables */
1142 for (i
= 0, map
= reg_maps
->address
; map
; map
>>= 1, ++i
)
1144 if (map
& 1) shader_addline(buffer
, "ivec4 A%u;\n", i
);
1147 /* Declare texture coordinate temporaries and initialize them */
1148 for (i
= 0, map
= reg_maps
->texcoord
; map
; map
>>= 1, ++i
)
1150 if (map
& 1) shader_addline(buffer
, "vec4 T%u = gl_TexCoord[%u];\n", i
, i
);
1153 if (version
->type
== WINED3D_SHADER_TYPE_VERTEX
)
1155 /* Declare attributes. */
1156 for (i
= 0, map
= reg_maps
->input_registers
; map
; map
>>= 1, ++i
)
1159 shader_addline(buffer
, "attribute vec4 %s_in%u;\n", prefix
, i
);
1162 shader_addline(buffer
, "uniform vec4 posFixup;\n");
1163 shader_addline(buffer
, "void order_ps_input(in vec4[%u]);\n", shader
->limits
.packed_output
);
1165 else if (version
->type
== WINED3D_SHADER_TYPE_GEOMETRY
)
1167 shader_addline(buffer
, "varying in vec4 gs_in[][%u];\n", shader
->limits
.packed_input
);
1169 else if (version
->type
== WINED3D_SHADER_TYPE_PIXEL
)
1171 if (version
->major
>= 3)
1173 UINT in_count
= min(vec4_varyings(version
->major
, gl_info
), shader
->limits
.packed_input
);
1176 shader_addline(buffer
, "varying vec4 %s_in[%u];\n", prefix
, in_count
);
1178 /* TODO: Write a replacement shader for the fixed function
1179 * vertex pipeline, so this isn't needed. For fixed function
1180 * vertex processing + 3.0 pixel shader we need a separate
1181 * function in the pixel shader that reads the fixed function
1182 * color into the packed input registers. */
1183 shader_addline(buffer
, "vec4 %s_in[%u];\n", prefix
, in_count
);
1186 for (i
= 0, map
= reg_maps
->bumpmat
; map
; map
>>= 1, ++i
)
1191 shader_addline(buffer
, "uniform mat2 bumpenv_mat%u;\n", i
);
1193 if (reg_maps
->luminanceparams
& (1 << i
))
1195 shader_addline(buffer
, "uniform float bumpenv_lum_scale%u;\n", i
);
1196 shader_addline(buffer
, "uniform float bumpenv_lum_offset%u;\n", i
);
1197 extra_constants_needed
++;
1200 extra_constants_needed
++;
1203 if (ps_args
->srgb_correction
)
1205 shader_addline(buffer
, "const vec4 srgb_const0 = vec4(%.8e, %.8e, %.8e, %.8e);\n",
1206 srgb_pow
, srgb_mul_high
, srgb_sub_high
, srgb_mul_low
);
1207 shader_addline(buffer
, "const vec4 srgb_const1 = vec4(%.8e, 0.0, 0.0, 0.0);\n",
1210 if (reg_maps
->vpos
|| reg_maps
->usesdsy
)
1212 if (shader
->limits
.constant_float
+ extra_constants_needed
1213 + 1 < gl_info
->limits
.glsl_ps_float_constants
)
1215 shader_addline(buffer
, "uniform vec4 ycorrection;\n");
1216 extra_constants_needed
++;
1220 /* This happens because we do not have proper tracking of the constant registers that are
1221 * actually used, only the max limit of the shader version
1223 FIXME("Cannot find a free uniform for vpos correction params\n");
1224 shader_addline(buffer
, "const vec4 ycorrection = vec4(%f, %f, 0.0, 0.0);\n",
1225 context
->render_offscreen
? 0.0f
: fb
->render_targets
[0]->resource
.height
,
1226 context
->render_offscreen
? 1.0f
: -1.0f
);
1228 shader_addline(buffer
, "vec4 vpos;\n");
1232 /* Declare output register temporaries */
1233 if (shader
->limits
.packed_output
)
1234 shader_addline(buffer
, "vec4 %s_out[%u];\n", prefix
, shader
->limits
.packed_output
);
1236 /* Declare temporary variables */
1237 for (i
= 0, map
= reg_maps
->temporary
; map
; map
>>= 1, ++i
)
1239 if (map
& 1) shader_addline(buffer
, "vec4 R%u;\n", i
);
1242 /* Declare loop registers aLx */
1243 if (version
->major
< 4)
1245 for (i
= 0; i
< reg_maps
->loop_depth
; ++i
)
1247 shader_addline(buffer
, "int aL%u;\n", i
);
1248 shader_addline(buffer
, "int tmpInt%u;\n", i
);
1252 /* Temporary variables for matrix operations */
1253 shader_addline(buffer
, "vec4 tmp0;\n");
1254 shader_addline(buffer
, "vec4 tmp1;\n");
1256 /* Local constants use a different name so they can be loaded once at shader link time
1257 * They can't be hardcoded into the shader text via LC = {x, y, z, w}; because the
1258 * float -> string conversion can cause precision loss.
1260 if (!shader
->load_local_constsF
)
1262 LIST_FOR_EACH_ENTRY(lconst
, &shader
->constantsF
, struct wined3d_shader_lconst
, entry
)
1264 shader_addline(buffer
, "uniform vec4 %s_lc%u;\n", prefix
, lconst
->idx
);
1268 /* Start the main program. */
1269 shader_addline(buffer
, "void main()\n{\n");
1271 /* Direct3D applications expect integer vPos values, while OpenGL drivers
1272 * add approximately 0.5. This causes off-by-one problems as spotted by
1273 * the vPos d3d9 visual test. Unfortunately ATI cards do not add exactly
1274 * 0.5, but rather something like 0.49999999 or 0.50000001, which still
1275 * causes precision troubles when we just subtract 0.5.
1277 * To deal with that, just floor() the position. This will eliminate the
1278 * fraction on all cards.
1280 * TODO: Test how this behaves with multisampling.
1282 * An advantage of floor is that it works even if the driver doesn't add
1283 * 0.5. It is somewhat questionable if 1.5, 2.5, ... are the proper values
1284 * to return in gl_FragCoord, even though coordinates specify the pixel
1285 * centers instead of the pixel corners. This code will behave correctly
1286 * on drivers that returns integer values. */
1287 if (version
->type
== WINED3D_SHADER_TYPE_PIXEL
&& reg_maps
->vpos
)
1288 shader_addline(buffer
,
1289 "vpos = floor(vec4(0, ycorrection[0], 0, 0) + gl_FragCoord * vec4(1, ycorrection[1], 1, 1));\n");
1292 /*****************************************************************************
1293 * Functions to generate GLSL strings from DirectX Shader bytecode begin here.
1295 * For more information, see http://wiki.winehq.org/DirectX-Shaders
1296 ****************************************************************************/
1299 static void shader_glsl_add_src_param(const struct wined3d_shader_instruction
*ins
,
1300 const struct wined3d_shader_src_param
*wined3d_src
, DWORD mask
, struct glsl_src_param
*glsl_src
);
1302 /** Used for opcode modifiers - They multiply the result by the specified amount */
1303 static const char * const shift_glsl_tab
[] = {
1305 "2.0 * ", /* 1 (x2) */
1306 "4.0 * ", /* 2 (x4) */
1307 "8.0 * ", /* 3 (x8) */
1308 "16.0 * ", /* 4 (x16) */
1309 "32.0 * ", /* 5 (x32) */
1316 "0.0625 * ", /* 12 (d16) */
1317 "0.125 * ", /* 13 (d8) */
1318 "0.25 * ", /* 14 (d4) */
1319 "0.5 * " /* 15 (d2) */
1322 /* Generate a GLSL parameter that does the input modifier computation and return the input register/mask to use */
1323 static void shader_glsl_gen_modifier(enum wined3d_shader_src_modifier src_modifier
,
1324 const char *in_reg
, const char *in_regswizzle
, char *out_str
)
1328 switch (src_modifier
)
1330 case WINED3DSPSM_DZ
: /* Need to handle this in the instructions itself (texld & texcrd). */
1331 case WINED3DSPSM_DW
:
1332 case WINED3DSPSM_NONE
:
1333 sprintf(out_str
, "%s%s", in_reg
, in_regswizzle
);
1335 case WINED3DSPSM_NEG
:
1336 sprintf(out_str
, "-%s%s", in_reg
, in_regswizzle
);
1338 case WINED3DSPSM_NOT
:
1339 sprintf(out_str
, "!%s%s", in_reg
, in_regswizzle
);
1341 case WINED3DSPSM_BIAS
:
1342 sprintf(out_str
, "(%s%s - vec4(0.5)%s)", in_reg
, in_regswizzle
, in_regswizzle
);
1344 case WINED3DSPSM_BIASNEG
:
1345 sprintf(out_str
, "-(%s%s - vec4(0.5)%s)", in_reg
, in_regswizzle
, in_regswizzle
);
1347 case WINED3DSPSM_SIGN
:
1348 sprintf(out_str
, "(2.0 * (%s%s - 0.5))", in_reg
, in_regswizzle
);
1350 case WINED3DSPSM_SIGNNEG
:
1351 sprintf(out_str
, "-(2.0 * (%s%s - 0.5))", in_reg
, in_regswizzle
);
1353 case WINED3DSPSM_COMP
:
1354 sprintf(out_str
, "(1.0 - %s%s)", in_reg
, in_regswizzle
);
1356 case WINED3DSPSM_X2
:
1357 sprintf(out_str
, "(2.0 * %s%s)", in_reg
, in_regswizzle
);
1359 case WINED3DSPSM_X2NEG
:
1360 sprintf(out_str
, "-(2.0 * %s%s)", in_reg
, in_regswizzle
);
1362 case WINED3DSPSM_ABS
:
1363 sprintf(out_str
, "abs(%s%s)", in_reg
, in_regswizzle
);
1365 case WINED3DSPSM_ABSNEG
:
1366 sprintf(out_str
, "-abs(%s%s)", in_reg
, in_regswizzle
);
1369 FIXME("Unhandled modifier %u\n", src_modifier
);
1370 sprintf(out_str
, "%s%s", in_reg
, in_regswizzle
);
1374 /** Writes the GLSL variable name that corresponds to the register that the
1375 * DX opcode parameter is trying to access */
1376 static void shader_glsl_get_register_name(const struct wined3d_shader_register
*reg
,
1377 char *register_name
, BOOL
*is_color
, const struct wined3d_shader_instruction
*ins
)
1379 /* oPos, oFog and oPts in D3D */
1380 static const char * const hwrastout_reg_names
[] = {"vs_out[10]", "vs_out[11].x", "vs_out[11].y"};
1382 const struct wined3d_shader
*shader
= ins
->ctx
->shader
;
1383 const struct wined3d_shader_reg_maps
*reg_maps
= ins
->ctx
->reg_maps
;
1384 const struct wined3d_shader_version
*version
= ®_maps
->shader_version
;
1385 const struct wined3d_gl_info
*gl_info
= ins
->ctx
->gl_info
;
1386 const char *prefix
= shader_glsl_get_prefix(version
->type
);
1387 struct glsl_src_param rel_param0
, rel_param1
;
1389 if (reg
->idx
[0].offset
!= ~0U && reg
->idx
[0].rel_addr
)
1390 shader_glsl_add_src_param(ins
, reg
->idx
[0].rel_addr
, WINED3DSP_WRITEMASK_0
, &rel_param0
);
1391 if (reg
->idx
[1].offset
!= ~0U && reg
->idx
[1].rel_addr
)
1392 shader_glsl_add_src_param(ins
, reg
->idx
[1].rel_addr
, WINED3DSP_WRITEMASK_0
, &rel_param1
);
1397 case WINED3DSPR_TEMP
:
1398 sprintf(register_name
, "R%u", reg
->idx
[0].offset
);
1401 case WINED3DSPR_INPUT
:
1402 /* vertex shaders */
1403 if (version
->type
== WINED3D_SHADER_TYPE_VERTEX
)
1405 struct shader_glsl_ctx_priv
*priv
= ins
->ctx
->backend_data
;
1406 if (priv
->cur_vs_args
->swizzle_map
& (1 << reg
->idx
[0].offset
))
1408 sprintf(register_name
, "%s_in%u", prefix
, reg
->idx
[0].offset
);
1412 if (version
->type
== WINED3D_SHADER_TYPE_GEOMETRY
)
1414 if (reg
->idx
[0].rel_addr
)
1416 if (reg
->idx
[1].rel_addr
)
1417 sprintf(register_name
, "gs_in[%s + %u][%s + %u]",
1418 rel_param0
.param_str
, reg
->idx
[0].offset
, rel_param1
.param_str
, reg
->idx
[1].offset
);
1420 sprintf(register_name
, "gs_in[%s + %u][%u]",
1421 rel_param0
.param_str
, reg
->idx
[0].offset
, reg
->idx
[1].offset
);
1423 else if (reg
->idx
[1].rel_addr
)
1424 sprintf(register_name
, "gs_in[%u][%s + %u]",
1425 reg
->idx
[0].offset
, rel_param1
.param_str
, reg
->idx
[1].offset
);
1427 sprintf(register_name
, "gs_in[%u][%u]", reg
->idx
[0].offset
, reg
->idx
[1].offset
);
1431 /* pixel shaders >= 3.0 */
1432 if (version
->major
>= 3)
1434 DWORD idx
= shader
->u
.ps
.input_reg_map
[reg
->idx
[0].offset
];
1435 unsigned int in_count
= vec4_varyings(version
->major
, gl_info
);
1437 if (reg
->idx
[0].rel_addr
)
1439 /* Removing a + 0 would be an obvious optimization, but
1440 * OS X doesn't see the NOP operation there. */
1443 if (shader
->u
.ps
.declared_in_count
> in_count
)
1445 sprintf(register_name
,
1446 "((%s + %u) > %u ? (%s + %u) > %u ? gl_SecondaryColor : gl_Color : %s_in[%s + %u])",
1447 rel_param0
.param_str
, idx
, in_count
- 1, rel_param0
.param_str
, idx
, in_count
,
1448 prefix
, rel_param0
.param_str
, idx
);
1452 sprintf(register_name
, "%s_in[%s + %u]", prefix
, rel_param0
.param_str
, idx
);
1457 if (shader
->u
.ps
.declared_in_count
> in_count
)
1459 sprintf(register_name
, "((%s) > %u ? (%s) > %u ? gl_SecondaryColor : gl_Color : %s_in[%s])",
1460 rel_param0
.param_str
, in_count
- 1, rel_param0
.param_str
, in_count
,
1461 prefix
, rel_param0
.param_str
);
1465 sprintf(register_name
, "%s_in[%s]", prefix
, rel_param0
.param_str
);
1471 if (idx
== in_count
) sprintf(register_name
, "gl_Color");
1472 else if (idx
== in_count
+ 1) sprintf(register_name
, "gl_SecondaryColor");
1473 else sprintf(register_name
, "%s_in[%u]", prefix
, idx
);
1478 if (!reg
->idx
[0].offset
)
1479 strcpy(register_name
, "gl_Color");
1481 strcpy(register_name
, "gl_SecondaryColor");
1486 case WINED3DSPR_CONST
:
1488 /* Relative addressing */
1489 if (reg
->idx
[0].rel_addr
)
1491 if (reg
->idx
[0].offset
)
1492 sprintf(register_name
, "%s_c[%s + %u]", prefix
, rel_param0
.param_str
, reg
->idx
[0].offset
);
1494 sprintf(register_name
, "%s_c[%s]", prefix
, rel_param0
.param_str
);
1498 if (shader_constant_is_local(shader
, reg
->idx
[0].offset
))
1499 sprintf(register_name
, "%s_lc%u", prefix
, reg
->idx
[0].offset
);
1501 sprintf(register_name
, "%s_c[%u]", prefix
, reg
->idx
[0].offset
);
1506 case WINED3DSPR_CONSTINT
:
1507 sprintf(register_name
, "%s_i[%u]", prefix
, reg
->idx
[0].offset
);
1510 case WINED3DSPR_CONSTBOOL
:
1511 sprintf(register_name
, "%s_b[%u]", prefix
, reg
->idx
[0].offset
);
1514 case WINED3DSPR_TEXTURE
: /* case WINED3DSPR_ADDR: */
1515 if (version
->type
== WINED3D_SHADER_TYPE_PIXEL
)
1516 sprintf(register_name
, "T%u", reg
->idx
[0].offset
);
1518 sprintf(register_name
, "A%u", reg
->idx
[0].offset
);
1521 case WINED3DSPR_LOOP
:
1522 sprintf(register_name
, "aL%u", ins
->ctx
->loop_state
->current_reg
- 1);
1525 case WINED3DSPR_SAMPLER
:
1526 sprintf(register_name
, "%s_sampler%u", prefix
, reg
->idx
[0].offset
);
1529 case WINED3DSPR_COLOROUT
:
1530 if (reg
->idx
[0].offset
>= gl_info
->limits
.buffers
)
1531 WARN("Write to render target %u, only %d supported.\n",
1532 reg
->idx
[0].offset
, gl_info
->limits
.buffers
);
1534 sprintf(register_name
, "gl_FragData[%u]", reg
->idx
[0].offset
);
1537 case WINED3DSPR_RASTOUT
:
1538 sprintf(register_name
, "%s", hwrastout_reg_names
[reg
->idx
[0].offset
]);
1541 case WINED3DSPR_DEPTHOUT
:
1542 sprintf(register_name
, "gl_FragDepth");
1545 case WINED3DSPR_ATTROUT
:
1546 if (!reg
->idx
[0].offset
)
1547 sprintf(register_name
, "%s_out[8]", prefix
);
1549 sprintf(register_name
, "%s_out[9]", prefix
);
1552 case WINED3DSPR_TEXCRDOUT
:
1553 /* Vertex shaders >= 3.0: WINED3DSPR_OUTPUT */
1554 sprintf(register_name
, "%s_out[%u]", prefix
, reg
->idx
[0].offset
);
1557 case WINED3DSPR_MISCTYPE
:
1558 if (!reg
->idx
[0].offset
)
1561 sprintf(register_name
, "vpos");
1563 else if (reg
->idx
[0].offset
== 1)
1565 /* Note that gl_FrontFacing is a bool, while vFace is
1566 * a float for which the sign determines front/back */
1567 sprintf(register_name
, "(gl_FrontFacing ? 1.0 : -1.0)");
1571 FIXME("Unhandled misctype register %u.\n", reg
->idx
[0].offset
);
1572 sprintf(register_name
, "unrecognized_register");
1576 case WINED3DSPR_IMMCONST
:
1577 switch (reg
->immconst_type
)
1579 case WINED3D_IMMCONST_SCALAR
:
1580 switch (reg
->data_type
)
1582 case WINED3D_DATA_FLOAT
:
1583 sprintf(register_name
, "%.8e", *(const float *)reg
->immconst_data
);
1585 case WINED3D_DATA_INT
:
1586 sprintf(register_name
, "%#x", reg
->immconst_data
[0]);
1588 case WINED3D_DATA_RESOURCE
:
1589 case WINED3D_DATA_SAMPLER
:
1590 case WINED3D_DATA_UINT
:
1591 sprintf(register_name
, "%#xu", reg
->immconst_data
[0]);
1594 sprintf(register_name
, "<unhandled data type %#x>", reg
->data_type
);
1599 case WINED3D_IMMCONST_VEC4
:
1600 switch (reg
->data_type
)
1602 case WINED3D_DATA_FLOAT
:
1603 sprintf(register_name
, "vec4(%.8e, %.8e, %.8e, %.8e)",
1604 *(const float *)®
->immconst_data
[0], *(const float *)®
->immconst_data
[1],
1605 *(const float *)®
->immconst_data
[2], *(const float *)®
->immconst_data
[3]);
1607 case WINED3D_DATA_INT
:
1608 sprintf(register_name
, "ivec4(%#x, %#x, %#x, %#x)",
1609 reg
->immconst_data
[0], reg
->immconst_data
[1],
1610 reg
->immconst_data
[2], reg
->immconst_data
[3]);
1612 case WINED3D_DATA_RESOURCE
:
1613 case WINED3D_DATA_SAMPLER
:
1614 case WINED3D_DATA_UINT
:
1615 sprintf(register_name
, "uvec4(%#xu, %#xu, %#xu, %#xu)",
1616 reg
->immconst_data
[0], reg
->immconst_data
[1],
1617 reg
->immconst_data
[2], reg
->immconst_data
[3]);
1620 sprintf(register_name
, "<unhandled data type %#x>", reg
->data_type
);
1626 FIXME("Unhandled immconst type %#x\n", reg
->immconst_type
);
1627 sprintf(register_name
, "<unhandled_immconst_type %#x>", reg
->immconst_type
);
1631 case WINED3DSPR_CONSTBUFFER
:
1632 if (reg
->idx
[1].rel_addr
)
1633 sprintf(register_name
, "%s_cb%u[%s + %u]",
1634 prefix
, reg
->idx
[0].offset
, rel_param1
.param_str
, reg
->idx
[1].offset
);
1636 sprintf(register_name
, "%s_cb%u[%u]", prefix
, reg
->idx
[0].offset
, reg
->idx
[1].offset
);
1639 case WINED3DSPR_PRIMID
:
1640 sprintf(register_name
, "uint(gl_PrimitiveIDIn)");
1644 FIXME("Unhandled register type %#x.\n", reg
->type
);
1645 sprintf(register_name
, "unrecognized_register");
1650 static void shader_glsl_write_mask_to_str(DWORD write_mask
, char *str
)
1653 if (write_mask
& WINED3DSP_WRITEMASK_0
) *str
++ = 'x';
1654 if (write_mask
& WINED3DSP_WRITEMASK_1
) *str
++ = 'y';
1655 if (write_mask
& WINED3DSP_WRITEMASK_2
) *str
++ = 'z';
1656 if (write_mask
& WINED3DSP_WRITEMASK_3
) *str
++ = 'w';
1660 /* Get the GLSL write mask for the destination register */
1661 static DWORD
shader_glsl_get_write_mask(const struct wined3d_shader_dst_param
*param
, char *write_mask
)
1663 DWORD mask
= param
->write_mask
;
1665 if (shader_is_scalar(¶m
->reg
))
1667 mask
= WINED3DSP_WRITEMASK_0
;
1672 shader_glsl_write_mask_to_str(mask
, write_mask
);
1678 static unsigned int shader_glsl_get_write_mask_size(DWORD write_mask
) {
1679 unsigned int size
= 0;
1681 if (write_mask
& WINED3DSP_WRITEMASK_0
) ++size
;
1682 if (write_mask
& WINED3DSP_WRITEMASK_1
) ++size
;
1683 if (write_mask
& WINED3DSP_WRITEMASK_2
) ++size
;
1684 if (write_mask
& WINED3DSP_WRITEMASK_3
) ++size
;
1689 static void shader_glsl_swizzle_to_str(const DWORD swizzle
, BOOL fixup
, DWORD mask
, char *str
)
1691 /* For registers of type WINED3DDECLTYPE_D3DCOLOR, data is stored as "bgra",
1692 * but addressed as "rgba". To fix this we need to swap the register's x
1693 * and z components. */
1694 const char *swizzle_chars
= fixup
? "zyxw" : "xyzw";
1697 /* swizzle bits fields: wwzzyyxx */
1698 if (mask
& WINED3DSP_WRITEMASK_0
) *str
++ = swizzle_chars
[swizzle
& 0x03];
1699 if (mask
& WINED3DSP_WRITEMASK_1
) *str
++ = swizzle_chars
[(swizzle
>> 2) & 0x03];
1700 if (mask
& WINED3DSP_WRITEMASK_2
) *str
++ = swizzle_chars
[(swizzle
>> 4) & 0x03];
1701 if (mask
& WINED3DSP_WRITEMASK_3
) *str
++ = swizzle_chars
[(swizzle
>> 6) & 0x03];
1705 static void shader_glsl_get_swizzle(const struct wined3d_shader_src_param
*param
,
1706 BOOL fixup
, DWORD mask
, char *swizzle_str
)
1708 if (shader_is_scalar(¶m
->reg
))
1709 *swizzle_str
= '\0';
1711 shader_glsl_swizzle_to_str(param
->swizzle
, fixup
, mask
, swizzle_str
);
1714 /* From a given parameter token, generate the corresponding GLSL string.
1715 * Also, return the actual register name and swizzle in case the
1716 * caller needs this information as well. */
1717 static void shader_glsl_add_src_param(const struct wined3d_shader_instruction
*ins
,
1718 const struct wined3d_shader_src_param
*wined3d_src
, DWORD mask
, struct glsl_src_param
*glsl_src
)
1720 BOOL is_color
= FALSE
;
1721 char swizzle_str
[6];
1723 glsl_src
->reg_name
[0] = '\0';
1724 glsl_src
->param_str
[0] = '\0';
1725 swizzle_str
[0] = '\0';
1727 shader_glsl_get_register_name(&wined3d_src
->reg
, glsl_src
->reg_name
, &is_color
, ins
);
1728 shader_glsl_get_swizzle(wined3d_src
, is_color
, mask
, swizzle_str
);
1730 if (wined3d_src
->reg
.type
== WINED3DSPR_IMMCONST
|| wined3d_src
->reg
.type
== WINED3DSPR_PRIMID
)
1732 shader_glsl_gen_modifier(wined3d_src
->modifiers
, glsl_src
->reg_name
, swizzle_str
, glsl_src
->param_str
);
1736 char param_str
[200];
1738 shader_glsl_gen_modifier(wined3d_src
->modifiers
, glsl_src
->reg_name
, swizzle_str
, param_str
);
1740 switch (wined3d_src
->reg
.data_type
)
1742 case WINED3D_DATA_FLOAT
:
1743 sprintf(glsl_src
->param_str
, "%s", param_str
);
1745 case WINED3D_DATA_INT
:
1746 sprintf(glsl_src
->param_str
, "floatBitsToInt(%s)", param_str
);
1748 case WINED3D_DATA_RESOURCE
:
1749 case WINED3D_DATA_SAMPLER
:
1750 case WINED3D_DATA_UINT
:
1751 sprintf(glsl_src
->param_str
, "floatBitsToUint(%s)", param_str
);
1754 FIXME("Unhandled data type %#x.\n", wined3d_src
->reg
.data_type
);
1755 sprintf(glsl_src
->param_str
, "%s", param_str
);
1761 /* From a given parameter token, generate the corresponding GLSL string.
1762 * Also, return the actual register name and swizzle in case the
1763 * caller needs this information as well. */
1764 static DWORD
shader_glsl_add_dst_param(const struct wined3d_shader_instruction
*ins
,
1765 const struct wined3d_shader_dst_param
*wined3d_dst
, struct glsl_dst_param
*glsl_dst
)
1767 BOOL is_color
= FALSE
;
1769 glsl_dst
->mask_str
[0] = '\0';
1770 glsl_dst
->reg_name
[0] = '\0';
1772 shader_glsl_get_register_name(&wined3d_dst
->reg
, glsl_dst
->reg_name
, &is_color
, ins
);
1773 return shader_glsl_get_write_mask(wined3d_dst
, glsl_dst
->mask_str
);
1776 /* Append the destination part of the instruction to the buffer, return the effective write mask */
1777 static DWORD
shader_glsl_append_dst_ext(struct wined3d_shader_buffer
*buffer
,
1778 const struct wined3d_shader_instruction
*ins
, const struct wined3d_shader_dst_param
*dst
)
1780 struct glsl_dst_param glsl_dst
;
1783 if ((mask
= shader_glsl_add_dst_param(ins
, dst
, &glsl_dst
)))
1785 switch (dst
->reg
.data_type
)
1787 case WINED3D_DATA_FLOAT
:
1788 shader_addline(buffer
, "%s%s = %s(",
1789 glsl_dst
.reg_name
, glsl_dst
.mask_str
, shift_glsl_tab
[dst
->shift
]);
1791 case WINED3D_DATA_INT
:
1792 shader_addline(buffer
, "%s%s = %sintBitsToFloat(",
1793 glsl_dst
.reg_name
, glsl_dst
.mask_str
, shift_glsl_tab
[dst
->shift
]);
1795 case WINED3D_DATA_RESOURCE
:
1796 case WINED3D_DATA_SAMPLER
:
1797 case WINED3D_DATA_UINT
:
1798 shader_addline(buffer
, "%s%s = %suintBitsToFloat(",
1799 glsl_dst
.reg_name
, glsl_dst
.mask_str
, shift_glsl_tab
[dst
->shift
]);
1802 FIXME("Unhandled data type %#x.\n", dst
->reg
.data_type
);
1803 shader_addline(buffer
, "%s%s = %s(",
1804 glsl_dst
.reg_name
, glsl_dst
.mask_str
, shift_glsl_tab
[dst
->shift
]);
1812 /* Append the destination part of the instruction to the buffer, return the effective write mask */
1813 static DWORD
shader_glsl_append_dst(struct wined3d_shader_buffer
*buffer
, const struct wined3d_shader_instruction
*ins
)
1815 return shader_glsl_append_dst_ext(buffer
, ins
, &ins
->dst
[0]);
1818 /** Process GLSL instruction modifiers */
1819 static void shader_glsl_add_instruction_modifiers(const struct wined3d_shader_instruction
*ins
)
1821 struct glsl_dst_param dst_param
;
1824 if (!ins
->dst_count
) return;
1826 modifiers
= ins
->dst
[0].modifiers
;
1827 if (!modifiers
) return;
1829 shader_glsl_add_dst_param(ins
, &ins
->dst
[0], &dst_param
);
1831 if (modifiers
& WINED3DSPDM_SATURATE
)
1833 /* _SAT means to clamp the value of the register to between 0 and 1 */
1834 shader_addline(ins
->ctx
->buffer
, "%s%s = clamp(%s%s, 0.0, 1.0);\n", dst_param
.reg_name
,
1835 dst_param
.mask_str
, dst_param
.reg_name
, dst_param
.mask_str
);
1838 if (modifiers
& WINED3DSPDM_MSAMPCENTROID
)
1840 FIXME("_centroid modifier not handled\n");
1843 if (modifiers
& WINED3DSPDM_PARTIALPRECISION
)
1845 /* MSDN says this modifier can be safely ignored, so that's what we'll do. */
1849 static const char *shader_glsl_get_rel_op(enum wined3d_shader_rel_op op
)
1853 case WINED3D_SHADER_REL_OP_GT
: return ">";
1854 case WINED3D_SHADER_REL_OP_EQ
: return "==";
1855 case WINED3D_SHADER_REL_OP_GE
: return ">=";
1856 case WINED3D_SHADER_REL_OP_LT
: return "<";
1857 case WINED3D_SHADER_REL_OP_NE
: return "!=";
1858 case WINED3D_SHADER_REL_OP_LE
: return "<=";
1860 FIXME("Unrecognized operator %#x.\n", op
);
1865 static void shader_glsl_get_sample_function(const struct wined3d_shader_context
*ctx
,
1866 DWORD sampler_idx
, DWORD flags
, struct glsl_sample_function
*sample_function
)
1868 enum wined3d_sampler_texture_type sampler_type
= ctx
->reg_maps
->sampler_type
[sampler_idx
];
1869 const struct wined3d_gl_info
*gl_info
= ctx
->gl_info
;
1870 BOOL shadow
= ctx
->reg_maps
->shader_version
.type
== WINED3D_SHADER_TYPE_PIXEL
1871 && (((const struct shader_glsl_ctx_priv
*)ctx
->backend_data
)->cur_ps_args
->shadow
& (1 << sampler_idx
));
1872 BOOL projected
= flags
& WINED3D_GLSL_SAMPLE_PROJECTED
;
1873 BOOL texrect
= flags
& WINED3D_GLSL_SAMPLE_RECT
;
1874 BOOL lod
= flags
& WINED3D_GLSL_SAMPLE_LOD
;
1875 BOOL grad
= flags
& WINED3D_GLSL_SAMPLE_GRAD
;
1877 /* Note that there's no such thing as a projected cube texture. */
1878 switch(sampler_type
) {
1884 sample_function
->name
= projected
? "shadow1DProjLod" : "shadow1DLod";
1888 if (gl_info
->supported
[EXT_GPU_SHADER4
])
1889 sample_function
->name
= projected
? "shadow1DProjGrad" : "shadow1DGrad";
1890 else if (gl_info
->supported
[ARB_SHADER_TEXTURE_LOD
])
1891 sample_function
->name
= projected
? "shadow1DProjGradARB" : "shadow1DGradARB";
1894 FIXME("Unsupported 1D shadow grad function.\n");
1895 sample_function
->name
= "unsupported1DGrad";
1900 sample_function
->name
= projected
? "shadow1DProj" : "shadow1D";
1902 sample_function
->coord_mask
= WINED3DSP_WRITEMASK_0
| WINED3DSP_WRITEMASK_1
;
1908 sample_function
->name
= projected
? "texture1DProjLod" : "texture1DLod";
1912 if (gl_info
->supported
[EXT_GPU_SHADER4
])
1913 sample_function
->name
= projected
? "texture1DProjGrad" : "texture1DGrad";
1914 else if (gl_info
->supported
[ARB_SHADER_TEXTURE_LOD
])
1915 sample_function
->name
= projected
? "texture1DProjGradARB" : "texture1DGradARB";
1918 FIXME("Unsupported 1D grad function.\n");
1919 sample_function
->name
= "unsupported1DGrad";
1924 sample_function
->name
= projected
? "texture1DProj" : "texture1D";
1926 sample_function
->coord_mask
= WINED3DSP_WRITEMASK_0
;
1937 sample_function
->name
= projected
? "shadow2DRectProjLod" : "shadow2DRectLod";
1941 if (gl_info
->supported
[EXT_GPU_SHADER4
])
1942 sample_function
->name
= projected
? "shadow2DRectProjGrad" : "shadow2DRectGrad";
1943 else if (gl_info
->supported
[ARB_SHADER_TEXTURE_LOD
])
1944 sample_function
->name
= projected
? "shadow2DRectProjGradARB" : "shadow2DRectGradARB";
1947 FIXME("Unsupported RECT shadow grad function.\n");
1948 sample_function
->name
= "unsupported2DRectGrad";
1953 sample_function
->name
= projected
? "shadow2DRectProj" : "shadow2DRect";
1960 sample_function
->name
= projected
? "shadow2DProjLod" : "shadow2DLod";
1964 if (gl_info
->supported
[EXT_GPU_SHADER4
])
1965 sample_function
->name
= projected
? "shadow2DProjGrad" : "shadow2DGrad";
1966 else if (gl_info
->supported
[ARB_SHADER_TEXTURE_LOD
])
1967 sample_function
->name
= projected
? "shadow2DProjGradARB" : "shadow2DGradARB";
1970 FIXME("Unsupported 2D shadow grad function.\n");
1971 sample_function
->name
= "unsupported2DGrad";
1976 sample_function
->name
= projected
? "shadow2DProj" : "shadow2D";
1979 sample_function
->coord_mask
= WINED3DSP_WRITEMASK_0
| WINED3DSP_WRITEMASK_1
| WINED3DSP_WRITEMASK_2
;
1987 sample_function
->name
= projected
? "texture2DRectProjLod" : "texture2DRectLod";
1991 if (gl_info
->supported
[EXT_GPU_SHADER4
])
1992 sample_function
->name
= projected
? "texture2DRectProjGrad" : "texture2DRectGrad";
1993 else if (gl_info
->supported
[ARB_SHADER_TEXTURE_LOD
])
1994 sample_function
->name
= projected
? "texture2DRectProjGradARB" : "texture2DRectGradARB";
1997 FIXME("Unsupported RECT grad function.\n");
1998 sample_function
->name
= "unsupported2DRectGrad";
2003 sample_function
->name
= projected
? "texture2DRectProj" : "texture2DRect";
2010 sample_function
->name
= projected
? "texture2DProjLod" : "texture2DLod";
2014 if (gl_info
->supported
[EXT_GPU_SHADER4
])
2015 sample_function
->name
= projected
? "texture2DProjGrad" : "texture2DGrad";
2016 else if (gl_info
->supported
[ARB_SHADER_TEXTURE_LOD
])
2017 sample_function
->name
= projected
? "texture2DProjGradARB" : "texture2DGradARB";
2020 FIXME("Unsupported 2D grad function.\n");
2021 sample_function
->name
= "unsupported2DGrad";
2026 sample_function
->name
= projected
? "texture2DProj" : "texture2D";
2029 sample_function
->coord_mask
= WINED3DSP_WRITEMASK_0
| WINED3DSP_WRITEMASK_1
;
2033 case WINED3DSTT_CUBE
:
2036 FIXME("Unsupported Cube shadow function.\n");
2037 sample_function
->name
= "unsupportedCubeShadow";
2038 sample_function
->coord_mask
= 0;
2044 sample_function
->name
= "textureCubeLod";
2048 if (gl_info
->supported
[EXT_GPU_SHADER4
])
2049 sample_function
->name
= "textureCubeGrad";
2050 else if (gl_info
->supported
[ARB_SHADER_TEXTURE_LOD
])
2051 sample_function
->name
= "textureCubeGradARB";
2054 FIXME("Unsupported Cube grad function.\n");
2055 sample_function
->name
= "unsupportedCubeGrad";
2060 sample_function
->name
= "textureCube";
2062 sample_function
->coord_mask
= WINED3DSP_WRITEMASK_0
| WINED3DSP_WRITEMASK_1
| WINED3DSP_WRITEMASK_2
;
2066 case WINED3DSTT_VOLUME
:
2069 FIXME("Unsupported 3D shadow function.\n");
2070 sample_function
->name
= "unsupported3DShadow";
2071 sample_function
->coord_mask
= 0;
2077 sample_function
->name
= projected
? "texture3DProjLod" : "texture3DLod";
2081 if (gl_info
->supported
[EXT_GPU_SHADER4
])
2082 sample_function
->name
= projected
? "texture3DProjGrad" : "texture3DGrad";
2083 else if (gl_info
->supported
[ARB_SHADER_TEXTURE_LOD
])
2084 sample_function
->name
= projected
? "texture3DProjGradARB" : "texture3DGradARB";
2087 FIXME("Unsupported 3D grad function.\n");
2088 sample_function
->name
= "unsupported3DGrad";
2093 sample_function
->name
= projected
? "texture3DProj" : "texture3D";
2095 sample_function
->coord_mask
= WINED3DSP_WRITEMASK_0
| WINED3DSP_WRITEMASK_1
| WINED3DSP_WRITEMASK_2
;
2100 sample_function
->name
= "";
2101 sample_function
->coord_mask
= 0;
2102 FIXME("Unrecognized sampler type: %#x;\n", sampler_type
);
2107 static void shader_glsl_append_fixup_arg(char *arguments
, const char *reg_name
,
2108 BOOL sign_fixup
, enum fixup_channel_source channel_source
)
2110 switch(channel_source
)
2112 case CHANNEL_SOURCE_ZERO
:
2113 strcat(arguments
, "0.0");
2116 case CHANNEL_SOURCE_ONE
:
2117 strcat(arguments
, "1.0");
2120 case CHANNEL_SOURCE_X
:
2121 strcat(arguments
, reg_name
);
2122 strcat(arguments
, ".x");
2125 case CHANNEL_SOURCE_Y
:
2126 strcat(arguments
, reg_name
);
2127 strcat(arguments
, ".y");
2130 case CHANNEL_SOURCE_Z
:
2131 strcat(arguments
, reg_name
);
2132 strcat(arguments
, ".z");
2135 case CHANNEL_SOURCE_W
:
2136 strcat(arguments
, reg_name
);
2137 strcat(arguments
, ".w");
2141 FIXME("Unhandled channel source %#x\n", channel_source
);
2142 strcat(arguments
, "undefined");
2146 if (sign_fixup
) strcat(arguments
, " * 2.0 - 1.0");
2149 static void shader_glsl_color_correction_ext(struct wined3d_shader_buffer
*buffer
,
2150 const char *reg_name
, DWORD mask
, struct color_fixup_desc fixup
)
2152 unsigned int mask_size
, remaining
;
2153 DWORD fixup_mask
= 0;
2154 char arguments
[256];
2157 if (fixup
.x_sign_fixup
|| fixup
.x_source
!= CHANNEL_SOURCE_X
) fixup_mask
|= WINED3DSP_WRITEMASK_0
;
2158 if (fixup
.y_sign_fixup
|| fixup
.y_source
!= CHANNEL_SOURCE_Y
) fixup_mask
|= WINED3DSP_WRITEMASK_1
;
2159 if (fixup
.z_sign_fixup
|| fixup
.z_source
!= CHANNEL_SOURCE_Z
) fixup_mask
|= WINED3DSP_WRITEMASK_2
;
2160 if (fixup
.w_sign_fixup
|| fixup
.w_source
!= CHANNEL_SOURCE_W
) fixup_mask
|= WINED3DSP_WRITEMASK_3
;
2161 if (!(mask
&= fixup_mask
))
2164 if (is_complex_fixup(fixup
))
2166 enum complex_fixup complex_fixup
= get_complex_fixup(fixup
);
2167 FIXME("Complex fixup (%#x) not supported\n",complex_fixup
);
2171 shader_glsl_write_mask_to_str(mask
, mask_str
);
2172 mask_size
= shader_glsl_get_write_mask_size(mask
);
2174 arguments
[0] = '\0';
2175 remaining
= mask_size
;
2176 if (mask
& WINED3DSP_WRITEMASK_0
)
2178 shader_glsl_append_fixup_arg(arguments
, reg_name
, fixup
.x_sign_fixup
, fixup
.x_source
);
2179 if (--remaining
) strcat(arguments
, ", ");
2181 if (mask
& WINED3DSP_WRITEMASK_1
)
2183 shader_glsl_append_fixup_arg(arguments
, reg_name
, fixup
.y_sign_fixup
, fixup
.y_source
);
2184 if (--remaining
) strcat(arguments
, ", ");
2186 if (mask
& WINED3DSP_WRITEMASK_2
)
2188 shader_glsl_append_fixup_arg(arguments
, reg_name
, fixup
.z_sign_fixup
, fixup
.z_source
);
2189 if (--remaining
) strcat(arguments
, ", ");
2191 if (mask
& WINED3DSP_WRITEMASK_3
)
2193 shader_glsl_append_fixup_arg(arguments
, reg_name
, fixup
.w_sign_fixup
, fixup
.w_source
);
2194 if (--remaining
) strcat(arguments
, ", ");
2198 shader_addline(buffer
, "%s%s = vec%u(%s);\n", reg_name
, mask_str
, mask_size
, arguments
);
2200 shader_addline(buffer
, "%s%s = %s;\n", reg_name
, mask_str
, arguments
);
2203 static void shader_glsl_color_correction(const struct wined3d_shader_instruction
*ins
, struct color_fixup_desc fixup
)
2208 shader_glsl_get_register_name(&ins
->dst
[0].reg
, reg_name
, &is_color
, ins
);
2209 shader_glsl_color_correction_ext(ins
->ctx
->buffer
, reg_name
, ins
->dst
[0].write_mask
, fixup
);
2212 static void PRINTF_ATTR(8, 9) shader_glsl_gen_sample_code(const struct wined3d_shader_instruction
*ins
,
2213 DWORD sampler
, const struct glsl_sample_function
*sample_function
, DWORD swizzle
,
2214 const char *dx
, const char *dy
, const char *bias
, const char *coord_reg_fmt
, ...)
2216 const struct wined3d_shader_version
*version
= &ins
->ctx
->reg_maps
->shader_version
;
2217 char dst_swizzle
[6];
2218 struct color_fixup_desc fixup
;
2219 BOOL np2_fixup
= FALSE
;
2222 shader_glsl_swizzle_to_str(swizzle
, FALSE
, ins
->dst
[0].write_mask
, dst_swizzle
);
2224 if (version
->type
== WINED3D_SHADER_TYPE_PIXEL
)
2226 const struct shader_glsl_ctx_priv
*priv
= ins
->ctx
->backend_data
;
2227 fixup
= priv
->cur_ps_args
->color_fixup
[sampler
];
2229 if(priv
->cur_ps_args
->np2_fixup
& (1 << sampler
)) {
2231 FIXME("Biased sampling from NP2 textures is unsupported\n");
2239 fixup
= COLOR_FIXUP_IDENTITY
; /* FIXME: Vshader color fixup */
2242 shader_glsl_append_dst(ins
->ctx
->buffer
, ins
);
2244 shader_addline(ins
->ctx
->buffer
, "%s(%s_sampler%u, ",
2245 sample_function
->name
, shader_glsl_get_prefix(version
->type
), sampler
);
2247 va_start(args
, coord_reg_fmt
);
2248 shader_vaddline(ins
->ctx
->buffer
, coord_reg_fmt
, args
);
2252 shader_addline(ins
->ctx
->buffer
, ", %s)%s);\n", bias
, dst_swizzle
);
2255 const struct shader_glsl_ctx_priv
*priv
= ins
->ctx
->backend_data
;
2256 const unsigned char idx
= priv
->cur_np2fixup_info
->idx
[sampler
];
2258 shader_addline(ins
->ctx
->buffer
, " * ps_samplerNP2Fixup[%u].%s)%s);\n", idx
>> 1,
2259 (idx
% 2) ? "zw" : "xy", dst_swizzle
);
2260 } else if(dx
&& dy
) {
2261 shader_addline(ins
->ctx
->buffer
, ", %s, %s)%s);\n", dx
, dy
, dst_swizzle
);
2263 shader_addline(ins
->ctx
->buffer
, ")%s);\n", dst_swizzle
);
2267 if(!is_identity_fixup(fixup
)) {
2268 shader_glsl_color_correction(ins
, fixup
);
2272 /*****************************************************************************
2273 * Begin processing individual instruction opcodes
2274 ****************************************************************************/
2276 static void shader_glsl_binop(const struct wined3d_shader_instruction
*ins
)
2278 struct wined3d_shader_buffer
*buffer
= ins
->ctx
->buffer
;
2279 struct glsl_src_param src0_param
;
2280 struct glsl_src_param src1_param
;
2284 /* Determine the GLSL operator to use based on the opcode */
2285 switch (ins
->handler_idx
)
2287 case WINED3DSIH_ADD
: op
= "+"; break;
2288 case WINED3DSIH_AND
: op
= "&"; break;
2289 case WINED3DSIH_DIV
: op
= "/"; break;
2290 case WINED3DSIH_IADD
: op
= "+"; break;
2291 case WINED3DSIH_MUL
: op
= "*"; break;
2292 case WINED3DSIH_SUB
: op
= "-"; break;
2293 case WINED3DSIH_USHR
: op
= ">>"; break;
2294 case WINED3DSIH_XOR
: op
= "^"; break;
2296 op
= "<unhandled operator>";
2297 FIXME("Opcode %#x not yet handled in GLSL\n", ins
->handler_idx
);
2301 write_mask
= shader_glsl_append_dst(buffer
, ins
);
2302 shader_glsl_add_src_param(ins
, &ins
->src
[0], write_mask
, &src0_param
);
2303 shader_glsl_add_src_param(ins
, &ins
->src
[1], write_mask
, &src1_param
);
2304 shader_addline(buffer
, "%s %s %s);\n", src0_param
.param_str
, op
, src1_param
.param_str
);
2307 static void shader_glsl_relop(const struct wined3d_shader_instruction
*ins
)
2309 struct wined3d_shader_buffer
*buffer
= ins
->ctx
->buffer
;
2310 struct glsl_src_param src0_param
;
2311 struct glsl_src_param src1_param
;
2312 unsigned int mask_size
;
2316 write_mask
= shader_glsl_append_dst(buffer
, ins
);
2317 mask_size
= shader_glsl_get_write_mask_size(write_mask
);
2318 shader_glsl_add_src_param(ins
, &ins
->src
[0], write_mask
, &src0_param
);
2319 shader_glsl_add_src_param(ins
, &ins
->src
[1], write_mask
, &src1_param
);
2323 switch (ins
->handler_idx
)
2325 case WINED3DSIH_EQ
: op
= "equal"; break;
2326 case WINED3DSIH_GE
: op
= "greaterThanEqual"; break;
2327 case WINED3DSIH_IGE
: op
= "greaterThanEqual"; break;
2328 case WINED3DSIH_LT
: op
= "lessThan"; break;
2330 op
= "<unhandled operator>";
2331 ERR("Unhandled opcode %#x.\n", ins
->handler_idx
);
2335 shader_addline(buffer
, "uvec%u(%s(%s, %s)) * 0xffffffffu);\n",
2336 mask_size
, op
, src0_param
.param_str
, src1_param
.param_str
);
2340 switch (ins
->handler_idx
)
2342 case WINED3DSIH_EQ
: op
= "=="; break;
2343 case WINED3DSIH_GE
: op
= ">="; break;
2344 case WINED3DSIH_IGE
: op
= ">="; break;
2345 case WINED3DSIH_LT
: op
= "<"; break;
2347 op
= "<unhandled operator>";
2348 ERR("Unhandled opcode %#x.\n", ins
->handler_idx
);
2352 shader_addline(buffer
, "%s %s %s ? 0xffffffffu : 0u);\n",
2353 src0_param
.param_str
, op
, src1_param
.param_str
);
2357 static void shader_glsl_imul(const struct wined3d_shader_instruction
*ins
)
2359 struct wined3d_shader_buffer
*buffer
= ins
->ctx
->buffer
;
2360 struct glsl_src_param src0_param
;
2361 struct glsl_src_param src1_param
;
2364 /* If we have ARB_gpu_shader5 or GLSL 4.0, we can use imulExtended(). If
2365 * not, we can emulate it. */
2366 if (ins
->dst
[0].reg
.type
!= WINED3DSPR_NULL
)
2367 FIXME("64-bit integer multiplies not implemented.\n");
2369 if (ins
->dst
[1].reg
.type
!= WINED3DSPR_NULL
)
2371 write_mask
= shader_glsl_append_dst_ext(buffer
, ins
, &ins
->dst
[1]);
2372 shader_glsl_add_src_param(ins
, &ins
->src
[0], write_mask
, &src0_param
);
2373 shader_glsl_add_src_param(ins
, &ins
->src
[1], write_mask
, &src1_param
);
2375 shader_addline(ins
->ctx
->buffer
, "%s * %s);\n",
2376 src0_param
.param_str
, src1_param
.param_str
);
2380 static void shader_glsl_udiv(const struct wined3d_shader_instruction
*ins
)
2382 struct wined3d_shader_buffer
*buffer
= ins
->ctx
->buffer
;
2383 struct glsl_src_param src0_param
, src1_param
;
2386 if (ins
->dst
[0].reg
.type
!= WINED3DSPR_NULL
)
2389 if (ins
->dst
[1].reg
.type
!= WINED3DSPR_NULL
)
2393 write_mask
= shader_glsl_get_write_mask(&ins
->dst
[0], dst_mask
);
2394 shader_glsl_add_src_param(ins
, &ins
->src
[0], write_mask
, &src0_param
);
2395 shader_glsl_add_src_param(ins
, &ins
->src
[1], write_mask
, &src1_param
);
2396 shader_addline(buffer
, "tmp0%s = %s / %s;\n",
2397 dst_mask
, src0_param
.param_str
, src1_param
.param_str
);
2399 write_mask
= shader_glsl_append_dst_ext(buffer
, ins
, &ins
->dst
[1]);
2400 shader_glsl_add_src_param(ins
, &ins
->src
[0], write_mask
, &src0_param
);
2401 shader_glsl_add_src_param(ins
, &ins
->src
[1], write_mask
, &src1_param
);
2402 shader_addline(buffer
, "%s %% %s));\n", src0_param
.param_str
, src1_param
.param_str
);
2404 shader_glsl_append_dst_ext(buffer
, ins
, &ins
->dst
[0]);
2405 shader_addline(buffer
, "tmp0%s);\n", dst_mask
);
2409 write_mask
= shader_glsl_append_dst_ext(buffer
, ins
, &ins
->dst
[0]);
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
);
2415 else if (ins
->dst
[1].reg
.type
!= WINED3DSPR_NULL
)
2417 write_mask
= shader_glsl_append_dst_ext(buffer
, ins
, &ins
->dst
[1]);
2418 shader_glsl_add_src_param(ins
, &ins
->src
[0], write_mask
, &src0_param
);
2419 shader_glsl_add_src_param(ins
, &ins
->src
[1], write_mask
, &src1_param
);
2420 shader_addline(buffer
, "%s %% %s);\n", src0_param
.param_str
, src1_param
.param_str
);
2424 /* Process the WINED3DSIO_MOV opcode using GLSL (dst = src) */
2425 static void shader_glsl_mov(const struct wined3d_shader_instruction
*ins
)
2427 const struct wined3d_gl_info
*gl_info
= ins
->ctx
->gl_info
;
2428 struct wined3d_shader_buffer
*buffer
= ins
->ctx
->buffer
;
2429 struct glsl_src_param src0_param
;
2432 write_mask
= shader_glsl_append_dst(buffer
, ins
);
2433 shader_glsl_add_src_param(ins
, &ins
->src
[0], write_mask
, &src0_param
);
2435 /* In vs_1_1 WINED3DSIO_MOV can write to the address register. In later
2436 * shader versions WINED3DSIO_MOVA is used for this. */
2437 if (ins
->ctx
->reg_maps
->shader_version
.major
== 1
2438 && ins
->ctx
->reg_maps
->shader_version
.type
== WINED3D_SHADER_TYPE_VERTEX
2439 && ins
->dst
[0].reg
.type
== WINED3DSPR_ADDR
)
2441 /* This is a simple floor() */
2442 unsigned int mask_size
= shader_glsl_get_write_mask_size(write_mask
);
2443 if (mask_size
> 1) {
2444 shader_addline(buffer
, "ivec%d(floor(%s)));\n", mask_size
, src0_param
.param_str
);
2446 shader_addline(buffer
, "int(floor(%s)));\n", src0_param
.param_str
);
2449 else if(ins
->handler_idx
== WINED3DSIH_MOVA
)
2451 /* We need to *round* to the nearest int here. */
2452 unsigned int mask_size
= shader_glsl_get_write_mask_size(write_mask
);
2454 if (gl_info
->supported
[EXT_GPU_SHADER4
])
2457 shader_addline(buffer
, "ivec%d(round(%s)));\n", mask_size
, src0_param
.param_str
);
2459 shader_addline(buffer
, "int(round(%s)));\n", src0_param
.param_str
);
2464 shader_addline(buffer
, "ivec%d(floor(abs(%s) + vec%d(0.5)) * sign(%s)));\n",
2465 mask_size
, src0_param
.param_str
, mask_size
, src0_param
.param_str
);
2467 shader_addline(buffer
, "int(floor(abs(%s) + 0.5) * sign(%s)));\n",
2468 src0_param
.param_str
, src0_param
.param_str
);
2473 shader_addline(buffer
, "%s);\n", src0_param
.param_str
);
2477 /* Process the dot product operators DP3 and DP4 in GLSL (dst = dot(src0, src1)) */
2478 static void shader_glsl_dot(const struct wined3d_shader_instruction
*ins
)
2480 struct wined3d_shader_buffer
*buffer
= ins
->ctx
->buffer
;
2481 struct glsl_src_param src0_param
;
2482 struct glsl_src_param src1_param
;
2483 DWORD dst_write_mask
, src_write_mask
;
2484 unsigned int dst_size
= 0;
2486 dst_write_mask
= shader_glsl_append_dst(buffer
, ins
);
2487 dst_size
= shader_glsl_get_write_mask_size(dst_write_mask
);
2489 /* dp3 works on vec3, dp4 on vec4 */
2490 if (ins
->handler_idx
== WINED3DSIH_DP4
)
2492 src_write_mask
= WINED3DSP_WRITEMASK_ALL
;
2494 src_write_mask
= WINED3DSP_WRITEMASK_0
| WINED3DSP_WRITEMASK_1
| WINED3DSP_WRITEMASK_2
;
2497 shader_glsl_add_src_param(ins
, &ins
->src
[0], src_write_mask
, &src0_param
);
2498 shader_glsl_add_src_param(ins
, &ins
->src
[1], src_write_mask
, &src1_param
);
2501 shader_addline(buffer
, "vec%d(dot(%s, %s)));\n", dst_size
, src0_param
.param_str
, src1_param
.param_str
);
2503 shader_addline(buffer
, "dot(%s, %s));\n", src0_param
.param_str
, src1_param
.param_str
);
2507 /* Note that this instruction has some restrictions. The destination write mask
2508 * can't contain the w component, and the source swizzles have to be .xyzw */
2509 static void shader_glsl_cross(const struct wined3d_shader_instruction
*ins
)
2511 DWORD src_mask
= WINED3DSP_WRITEMASK_0
| WINED3DSP_WRITEMASK_1
| WINED3DSP_WRITEMASK_2
;
2512 struct glsl_src_param src0_param
;
2513 struct glsl_src_param src1_param
;
2516 shader_glsl_get_write_mask(&ins
->dst
[0], dst_mask
);
2517 shader_glsl_append_dst(ins
->ctx
->buffer
, ins
);
2518 shader_glsl_add_src_param(ins
, &ins
->src
[0], src_mask
, &src0_param
);
2519 shader_glsl_add_src_param(ins
, &ins
->src
[1], src_mask
, &src1_param
);
2520 shader_addline(ins
->ctx
->buffer
, "cross(%s, %s)%s);\n", src0_param
.param_str
, src1_param
.param_str
, dst_mask
);
2523 static void shader_glsl_cut(const struct wined3d_shader_instruction
*ins
)
2525 shader_addline(ins
->ctx
->buffer
, "EndPrimitive();\n");
2528 /* Process the WINED3DSIO_POW instruction in GLSL (dst = |src0|^src1)
2529 * Src0 and src1 are scalars. Note that D3D uses the absolute of src0, while
2530 * GLSL uses the value as-is. */
2531 static void shader_glsl_pow(const struct wined3d_shader_instruction
*ins
)
2533 struct wined3d_shader_buffer
*buffer
= ins
->ctx
->buffer
;
2534 struct glsl_src_param src0_param
;
2535 struct glsl_src_param src1_param
;
2536 DWORD dst_write_mask
;
2537 unsigned int dst_size
;
2539 dst_write_mask
= shader_glsl_append_dst(buffer
, ins
);
2540 dst_size
= shader_glsl_get_write_mask_size(dst_write_mask
);
2542 shader_glsl_add_src_param(ins
, &ins
->src
[0], WINED3DSP_WRITEMASK_0
, &src0_param
);
2543 shader_glsl_add_src_param(ins
, &ins
->src
[1], WINED3DSP_WRITEMASK_0
, &src1_param
);
2547 shader_addline(buffer
, "vec%u(%s == 0.0 ? 1.0 : pow(abs(%s), %s)));\n",
2548 dst_size
, src1_param
.param_str
, src0_param
.param_str
, src1_param
.param_str
);
2552 shader_addline(buffer
, "%s == 0.0 ? 1.0 : pow(abs(%s), %s));\n",
2553 src1_param
.param_str
, src0_param
.param_str
, src1_param
.param_str
);
2557 /* Process the WINED3DSIO_LOG instruction in GLSL (dst = log2(|src0|))
2558 * Src0 is a scalar. Note that D3D uses the absolute of src0, while
2559 * GLSL uses the value as-is. */
2560 static void shader_glsl_log(const struct wined3d_shader_instruction
*ins
)
2562 struct wined3d_shader_buffer
*buffer
= ins
->ctx
->buffer
;
2563 struct glsl_src_param src0_param
;
2564 DWORD dst_write_mask
;
2565 unsigned int dst_size
;
2567 dst_write_mask
= shader_glsl_append_dst(buffer
, ins
);
2568 dst_size
= shader_glsl_get_write_mask_size(dst_write_mask
);
2570 shader_glsl_add_src_param(ins
, &ins
->src
[0], WINED3DSP_WRITEMASK_0
, &src0_param
);
2574 shader_addline(buffer
, "vec%u(log2(abs(%s))));\n",
2575 dst_size
, src0_param
.param_str
);
2579 shader_addline(buffer
, "log2(abs(%s)));\n",
2580 src0_param
.param_str
);
2584 /* Map the opcode 1-to-1 to the GL code (arg->dst = instruction(src0, src1, ...) */
2585 static void shader_glsl_map2gl(const struct wined3d_shader_instruction
*ins
)
2587 struct wined3d_shader_buffer
*buffer
= ins
->ctx
->buffer
;
2588 struct glsl_src_param src_param
;
2589 const char *instruction
;
2593 /* Determine the GLSL function to use based on the opcode */
2594 /* TODO: Possibly make this a table for faster lookups */
2595 switch (ins
->handler_idx
)
2597 case WINED3DSIH_MIN
: instruction
= "min"; break;
2598 case WINED3DSIH_MAX
: instruction
= "max"; break;
2599 case WINED3DSIH_ABS
: instruction
= "abs"; break;
2600 case WINED3DSIH_FRC
: instruction
= "fract"; break;
2601 case WINED3DSIH_EXP
: instruction
= "exp2"; break;
2602 case WINED3DSIH_DSX
: instruction
= "dFdx"; break;
2603 case WINED3DSIH_DSY
: instruction
= "ycorrection.y * dFdy"; break;
2604 case WINED3DSIH_ROUND_NI
: instruction
= "floor"; break;
2605 default: instruction
= "";
2606 FIXME("Opcode %#x not yet handled in GLSL\n", ins
->handler_idx
);
2610 write_mask
= shader_glsl_append_dst(buffer
, ins
);
2612 shader_addline(buffer
, "%s(", instruction
);
2616 shader_glsl_add_src_param(ins
, &ins
->src
[0], write_mask
, &src_param
);
2617 shader_addline(buffer
, "%s", src_param
.param_str
);
2618 for (i
= 1; i
< ins
->src_count
; ++i
)
2620 shader_glsl_add_src_param(ins
, &ins
->src
[i
], write_mask
, &src_param
);
2621 shader_addline(buffer
, ", %s", src_param
.param_str
);
2625 shader_addline(buffer
, "));\n");
2628 static void shader_glsl_nop(const struct wined3d_shader_instruction
*ins
) {}
2630 static void shader_glsl_nrm(const struct wined3d_shader_instruction
*ins
)
2632 struct wined3d_shader_buffer
*buffer
= ins
->ctx
->buffer
;
2633 struct glsl_src_param src_param
;
2634 unsigned int mask_size
;
2638 write_mask
= shader_glsl_get_write_mask(ins
->dst
, dst_mask
);
2639 mask_size
= shader_glsl_get_write_mask_size(write_mask
);
2640 shader_glsl_add_src_param(ins
, &ins
->src
[0], write_mask
, &src_param
);
2642 shader_addline(buffer
, "tmp0.x = dot(%s, %s);\n",
2643 src_param
.param_str
, src_param
.param_str
);
2644 shader_glsl_append_dst(buffer
, ins
);
2648 shader_addline(buffer
, "tmp0.x == 0.0 ? vec%u(0.0) : (%s * inversesqrt(tmp0.x)));\n",
2649 mask_size
, src_param
.param_str
);
2653 shader_addline(buffer
, "tmp0.x == 0.0 ? 0.0 : (%s * inversesqrt(tmp0.x)));\n",
2654 src_param
.param_str
);
2658 /** Process the WINED3DSIO_EXPP instruction in GLSL:
2659 * For shader model 1.x, do the following (and honor the writemask, so use a temporary variable):
2660 * dst.x = 2^(floor(src))
2661 * dst.y = src - floor(src)
2662 * dst.z = 2^src (partial precision is allowed, but optional)
2664 * For 2.0 shaders, just do this (honoring writemask and swizzle):
2665 * dst = 2^src; (partial precision is allowed, but optional)
2667 static void shader_glsl_expp(const struct wined3d_shader_instruction
*ins
)
2669 struct glsl_src_param src_param
;
2671 shader_glsl_add_src_param(ins
, &ins
->src
[0], WINED3DSP_WRITEMASK_0
, &src_param
);
2673 if (ins
->ctx
->reg_maps
->shader_version
.major
< 2)
2677 shader_addline(ins
->ctx
->buffer
, "tmp0.x = exp2(floor(%s));\n", src_param
.param_str
);
2678 shader_addline(ins
->ctx
->buffer
, "tmp0.y = %s - floor(%s);\n", src_param
.param_str
, src_param
.param_str
);
2679 shader_addline(ins
->ctx
->buffer
, "tmp0.z = exp2(%s);\n", src_param
.param_str
);
2680 shader_addline(ins
->ctx
->buffer
, "tmp0.w = 1.0;\n");
2682 shader_glsl_append_dst(ins
->ctx
->buffer
, ins
);
2683 shader_glsl_get_write_mask(&ins
->dst
[0], dst_mask
);
2684 shader_addline(ins
->ctx
->buffer
, "tmp0%s);\n", dst_mask
);
2687 unsigned int mask_size
;
2689 write_mask
= shader_glsl_append_dst(ins
->ctx
->buffer
, ins
);
2690 mask_size
= shader_glsl_get_write_mask_size(write_mask
);
2692 if (mask_size
> 1) {
2693 shader_addline(ins
->ctx
->buffer
, "vec%d(exp2(%s)));\n", mask_size
, src_param
.param_str
);
2695 shader_addline(ins
->ctx
->buffer
, "exp2(%s));\n", src_param
.param_str
);
2700 static void shader_glsl_to_int(const struct wined3d_shader_instruction
*ins
)
2702 struct wined3d_shader_buffer
*buffer
= ins
->ctx
->buffer
;
2703 struct glsl_src_param src_param
;
2704 unsigned int mask_size
;
2707 write_mask
= shader_glsl_append_dst(buffer
, ins
);
2708 mask_size
= shader_glsl_get_write_mask_size(write_mask
);
2709 shader_glsl_add_src_param(ins
, &ins
->src
[0], write_mask
, &src_param
);
2712 shader_addline(buffer
, "ivec%u(%s));\n", mask_size
, src_param
.param_str
);
2714 shader_addline(buffer
, "int(%s));\n", src_param
.param_str
);
2717 static void shader_glsl_to_float(const struct wined3d_shader_instruction
*ins
)
2719 struct wined3d_shader_buffer
*buffer
= ins
->ctx
->buffer
;
2720 struct glsl_src_param src_param
;
2721 unsigned int mask_size
;
2724 write_mask
= shader_glsl_append_dst(buffer
, ins
);
2725 mask_size
= shader_glsl_get_write_mask_size(write_mask
);
2726 shader_glsl_add_src_param(ins
, &ins
->src
[0], write_mask
, &src_param
);
2729 shader_addline(buffer
, "vec%u(%s));\n", mask_size
, src_param
.param_str
);
2731 shader_addline(buffer
, "float(%s));\n", src_param
.param_str
);
2734 /** Process the RCP (reciprocal or inverse) opcode in GLSL (dst = 1 / src) */
2735 static void shader_glsl_rcp(const struct wined3d_shader_instruction
*ins
)
2737 struct glsl_src_param src_param
;
2739 unsigned int mask_size
;
2741 write_mask
= shader_glsl_append_dst(ins
->ctx
->buffer
, ins
);
2742 mask_size
= shader_glsl_get_write_mask_size(write_mask
);
2743 shader_glsl_add_src_param(ins
, &ins
->src
[0], WINED3DSP_WRITEMASK_3
, &src_param
);
2747 shader_addline(ins
->ctx
->buffer
, "vec%u(1.0 / %s));\n",
2748 mask_size
, src_param
.param_str
);
2752 shader_addline(ins
->ctx
->buffer
, "1.0 / %s);\n",
2753 src_param
.param_str
);
2757 static void shader_glsl_rsq(const struct wined3d_shader_instruction
*ins
)
2759 struct wined3d_shader_buffer
*buffer
= ins
->ctx
->buffer
;
2760 struct glsl_src_param src_param
;
2762 unsigned int mask_size
;
2764 write_mask
= shader_glsl_append_dst(buffer
, ins
);
2765 mask_size
= shader_glsl_get_write_mask_size(write_mask
);
2767 shader_glsl_add_src_param(ins
, &ins
->src
[0], WINED3DSP_WRITEMASK_3
, &src_param
);
2771 shader_addline(buffer
, "vec%u(inversesqrt(abs(%s))));\n",
2772 mask_size
, src_param
.param_str
);
2776 shader_addline(buffer
, "inversesqrt(abs(%s)));\n",
2777 src_param
.param_str
);
2781 /** Process signed comparison opcodes in GLSL. */
2782 static void shader_glsl_compare(const struct wined3d_shader_instruction
*ins
)
2784 struct glsl_src_param src0_param
;
2785 struct glsl_src_param src1_param
;
2787 unsigned int mask_size
;
2789 write_mask
= shader_glsl_append_dst(ins
->ctx
->buffer
, ins
);
2790 mask_size
= shader_glsl_get_write_mask_size(write_mask
);
2791 shader_glsl_add_src_param(ins
, &ins
->src
[0], write_mask
, &src0_param
);
2792 shader_glsl_add_src_param(ins
, &ins
->src
[1], write_mask
, &src1_param
);
2794 if (mask_size
> 1) {
2795 const char *compare
;
2797 switch(ins
->handler_idx
)
2799 case WINED3DSIH_SLT
: compare
= "lessThan"; break;
2800 case WINED3DSIH_SGE
: compare
= "greaterThanEqual"; break;
2801 default: compare
= "";
2802 FIXME("Can't handle opcode %#x\n", ins
->handler_idx
);
2805 shader_addline(ins
->ctx
->buffer
, "vec%d(%s(%s, %s)));\n", mask_size
, compare
,
2806 src0_param
.param_str
, src1_param
.param_str
);
2808 switch(ins
->handler_idx
)
2810 case WINED3DSIH_SLT
:
2811 /* Step(src0, src1) is not suitable here because if src0 == src1 SLT is supposed,
2812 * to return 0.0 but step returns 1.0 because step is not < x
2813 * An alternative is a bvec compare padded with an unused second component.
2814 * step(src1 * -1.0, src0 * -1.0) is not an option because it suffers from the same
2815 * issue. Playing with not() is not possible either because not() does not accept
2818 shader_addline(ins
->ctx
->buffer
, "(%s < %s) ? 1.0 : 0.0);\n",
2819 src0_param
.param_str
, src1_param
.param_str
);
2821 case WINED3DSIH_SGE
:
2822 /* Here we can use the step() function and safe a conditional */
2823 shader_addline(ins
->ctx
->buffer
, "step(%s, %s));\n", src1_param
.param_str
, src0_param
.param_str
);
2826 FIXME("Can't handle opcode %#x\n", ins
->handler_idx
);
2832 static void shader_glsl_conditional_move(const struct wined3d_shader_instruction
*ins
)
2834 const char *condition_prefix
, *condition_suffix
;
2835 struct wined3d_shader_dst_param dst
;
2836 struct glsl_src_param src0_param
;
2837 struct glsl_src_param src1_param
;
2838 struct glsl_src_param src2_param
;
2839 BOOL temp_destination
= FALSE
;
2840 DWORD cmp_channel
= 0;
2845 switch (ins
->handler_idx
)
2847 case WINED3DSIH_CMP
:
2848 condition_prefix
= "";
2849 condition_suffix
= " >= 0.0";
2852 case WINED3DSIH_CND
:
2853 condition_prefix
= "";
2854 condition_suffix
= " > 0.5";
2857 case WINED3DSIH_MOVC
:
2858 condition_prefix
= "bool(";
2859 condition_suffix
= ")";
2863 FIXME("Unhandled instruction %#x.\n", ins
->handler_idx
);
2864 condition_prefix
= "<unhandled prefix>";
2865 condition_suffix
= "<unhandled suffix>";
2869 if (shader_is_scalar(&ins
->dst
[0].reg
) || shader_is_scalar(&ins
->src
[0].reg
))
2871 write_mask
= shader_glsl_append_dst(ins
->ctx
->buffer
, ins
);
2872 shader_glsl_add_src_param(ins
, &ins
->src
[0], write_mask
, &src0_param
);
2873 shader_glsl_add_src_param(ins
, &ins
->src
[1], write_mask
, &src1_param
);
2874 shader_glsl_add_src_param(ins
, &ins
->src
[2], write_mask
, &src2_param
);
2876 shader_addline(ins
->ctx
->buffer
, "%s%s%s ? %s : %s);\n",
2877 condition_prefix
, src0_param
.param_str
, condition_suffix
,
2878 src1_param
.param_str
, src2_param
.param_str
);
2884 /* Splitting the instruction up in multiple lines imposes a problem:
2885 * The first lines may overwrite source parameters of the following lines.
2886 * Deal with that by using a temporary destination register if needed. */
2887 if ((ins
->src
[0].reg
.idx
[0].offset
== dst
.reg
.idx
[0].offset
2888 && ins
->src
[0].reg
.type
== dst
.reg
.type
)
2889 || (ins
->src
[1].reg
.idx
[0].offset
== dst
.reg
.idx
[0].offset
2890 && ins
->src
[1].reg
.type
== dst
.reg
.type
)
2891 || (ins
->src
[2].reg
.idx
[0].offset
== dst
.reg
.idx
[0].offset
2892 && ins
->src
[2].reg
.type
== dst
.reg
.type
))
2893 temp_destination
= TRUE
;
2895 /* Cycle through all source0 channels. */
2896 for (i
= 0; i
< 4; ++i
)
2899 /* Find the destination channels which use the current source0 channel. */
2900 for (j
= 0; j
< 4; ++j
)
2902 if (((ins
->src
[0].swizzle
>> (2 * j
)) & 0x3) == i
)
2904 write_mask
|= WINED3DSP_WRITEMASK_0
<< j
;
2905 cmp_channel
= WINED3DSP_WRITEMASK_0
<< j
;
2908 dst
.write_mask
= ins
->dst
[0].write_mask
& write_mask
;
2910 if (temp_destination
)
2912 if (!(write_mask
= shader_glsl_get_write_mask(&dst
, mask_char
)))
2914 shader_addline(ins
->ctx
->buffer
, "tmp0%s = (", mask_char
);
2916 else if (!(write_mask
= shader_glsl_append_dst_ext(ins
->ctx
->buffer
, ins
, &dst
)))
2919 shader_glsl_add_src_param(ins
, &ins
->src
[0], cmp_channel
, &src0_param
);
2920 shader_glsl_add_src_param(ins
, &ins
->src
[1], write_mask
, &src1_param
);
2921 shader_glsl_add_src_param(ins
, &ins
->src
[2], write_mask
, &src2_param
);
2923 shader_addline(ins
->ctx
->buffer
, "%s%s%s ? %s : %s);\n",
2924 condition_prefix
, src0_param
.param_str
, condition_suffix
,
2925 src1_param
.param_str
, src2_param
.param_str
);
2928 if (temp_destination
)
2930 shader_glsl_get_write_mask(&ins
->dst
[0], mask_char
);
2931 shader_glsl_append_dst(ins
->ctx
->buffer
, ins
);
2932 shader_addline(ins
->ctx
->buffer
, "tmp0%s);\n", mask_char
);
2936 /** Process the CND opcode in GLSL (dst = (src0 > 0.5) ? src1 : src2) */
2937 /* For ps 1.1-1.3, only a single component of src0 is used. For ps 1.4
2938 * the compare is done per component of src0. */
2939 static void shader_glsl_cnd(const struct wined3d_shader_instruction
*ins
)
2941 struct glsl_src_param src0_param
;
2942 struct glsl_src_param src1_param
;
2943 struct glsl_src_param src2_param
;
2945 DWORD shader_version
= WINED3D_SHADER_VERSION(ins
->ctx
->reg_maps
->shader_version
.major
,
2946 ins
->ctx
->reg_maps
->shader_version
.minor
);
2948 if (shader_version
< WINED3D_SHADER_VERSION(1, 4))
2950 write_mask
= shader_glsl_append_dst(ins
->ctx
->buffer
, ins
);
2951 shader_glsl_add_src_param(ins
, &ins
->src
[0], WINED3DSP_WRITEMASK_0
, &src0_param
);
2952 shader_glsl_add_src_param(ins
, &ins
->src
[1], write_mask
, &src1_param
);
2953 shader_glsl_add_src_param(ins
, &ins
->src
[2], write_mask
, &src2_param
);
2955 /* Fun: The D3DSI_COISSUE flag changes the semantic of the cnd instruction for < 1.4 shaders */
2958 shader_addline(ins
->ctx
->buffer
, "%s /* COISSUE! */);\n", src1_param
.param_str
);
2960 shader_addline(ins
->ctx
->buffer
, "%s > 0.5 ? %s : %s);\n",
2961 src0_param
.param_str
, src1_param
.param_str
, src2_param
.param_str
);
2966 shader_glsl_conditional_move(ins
);
2969 /** GLSL code generation for WINED3DSIO_MAD: Multiply the first 2 opcodes, then add the last */
2970 static void shader_glsl_mad(const struct wined3d_shader_instruction
*ins
)
2972 struct glsl_src_param src0_param
;
2973 struct glsl_src_param src1_param
;
2974 struct glsl_src_param src2_param
;
2977 write_mask
= shader_glsl_append_dst(ins
->ctx
->buffer
, ins
);
2978 shader_glsl_add_src_param(ins
, &ins
->src
[0], write_mask
, &src0_param
);
2979 shader_glsl_add_src_param(ins
, &ins
->src
[1], write_mask
, &src1_param
);
2980 shader_glsl_add_src_param(ins
, &ins
->src
[2], write_mask
, &src2_param
);
2981 shader_addline(ins
->ctx
->buffer
, "(%s * %s) + %s);\n",
2982 src0_param
.param_str
, src1_param
.param_str
, src2_param
.param_str
);
2985 /* Handles transforming all WINED3DSIO_M?x? opcodes for
2986 Vertex shaders to GLSL codes */
2987 static void shader_glsl_mnxn(const struct wined3d_shader_instruction
*ins
)
2990 int nComponents
= 0;
2991 struct wined3d_shader_dst_param tmp_dst
= {{0}};
2992 struct wined3d_shader_src_param tmp_src
[2] = {{{0}}};
2993 struct wined3d_shader_instruction tmp_ins
;
2995 memset(&tmp_ins
, 0, sizeof(tmp_ins
));
2997 /* Set constants for the temporary argument */
2998 tmp_ins
.ctx
= ins
->ctx
;
2999 tmp_ins
.dst_count
= 1;
3000 tmp_ins
.dst
= &tmp_dst
;
3001 tmp_ins
.src_count
= 2;
3002 tmp_ins
.src
= tmp_src
;
3004 switch(ins
->handler_idx
)
3006 case WINED3DSIH_M4x4
:
3008 tmp_ins
.handler_idx
= WINED3DSIH_DP4
;
3010 case WINED3DSIH_M4x3
:
3012 tmp_ins
.handler_idx
= WINED3DSIH_DP4
;
3014 case WINED3DSIH_M3x4
:
3016 tmp_ins
.handler_idx
= WINED3DSIH_DP3
;
3018 case WINED3DSIH_M3x3
:
3020 tmp_ins
.handler_idx
= WINED3DSIH_DP3
;
3022 case WINED3DSIH_M3x2
:
3024 tmp_ins
.handler_idx
= WINED3DSIH_DP3
;
3030 tmp_dst
= ins
->dst
[0];
3031 tmp_src
[0] = ins
->src
[0];
3032 tmp_src
[1] = ins
->src
[1];
3033 for (i
= 0; i
< nComponents
; ++i
)
3035 tmp_dst
.write_mask
= WINED3DSP_WRITEMASK_0
<< i
;
3036 shader_glsl_dot(&tmp_ins
);
3037 ++tmp_src
[1].reg
.idx
[0].offset
;
3042 The LRP instruction performs a component-wise linear interpolation
3043 between the second and third operands using the first operand as the
3044 blend factor. Equation: (dst = src2 + src0 * (src1 - src2))
3045 This is equivalent to mix(src2, src1, src0);
3047 static void shader_glsl_lrp(const struct wined3d_shader_instruction
*ins
)
3049 struct glsl_src_param src0_param
;
3050 struct glsl_src_param src1_param
;
3051 struct glsl_src_param src2_param
;
3054 write_mask
= shader_glsl_append_dst(ins
->ctx
->buffer
, ins
);
3056 shader_glsl_add_src_param(ins
, &ins
->src
[0], write_mask
, &src0_param
);
3057 shader_glsl_add_src_param(ins
, &ins
->src
[1], write_mask
, &src1_param
);
3058 shader_glsl_add_src_param(ins
, &ins
->src
[2], write_mask
, &src2_param
);
3060 shader_addline(ins
->ctx
->buffer
, "mix(%s, %s, %s));\n",
3061 src2_param
.param_str
, src1_param
.param_str
, src0_param
.param_str
);
3064 /** Process the WINED3DSIO_LIT instruction in GLSL:
3065 * dst.x = dst.w = 1.0
3066 * dst.y = (src0.x > 0) ? src0.x
3067 * dst.z = (src0.x > 0) ? ((src0.y > 0) ? pow(src0.y, src.w) : 0) : 0
3068 * where src.w is clamped at +- 128
3070 static void shader_glsl_lit(const struct wined3d_shader_instruction
*ins
)
3072 struct glsl_src_param src0_param
;
3073 struct glsl_src_param src1_param
;
3074 struct glsl_src_param src3_param
;
3077 shader_glsl_append_dst(ins
->ctx
->buffer
, ins
);
3078 shader_glsl_get_write_mask(&ins
->dst
[0], dst_mask
);
3080 shader_glsl_add_src_param(ins
, &ins
->src
[0], WINED3DSP_WRITEMASK_0
, &src0_param
);
3081 shader_glsl_add_src_param(ins
, &ins
->src
[0], WINED3DSP_WRITEMASK_1
, &src1_param
);
3082 shader_glsl_add_src_param(ins
, &ins
->src
[0], WINED3DSP_WRITEMASK_3
, &src3_param
);
3084 /* The sdk specifies the instruction like this
3086 * if(src.x > 0.0) dst.y = src.x
3088 * if(src.x > 0.0 && src.y > 0.0) dst.z = pow(src.y, power);
3091 * (where power = src.w clamped between -128 and 128)
3093 * Obviously that has quite a few conditionals in it which we don't like. So the first step is this:
3094 * dst.x = 1.0 ... No further explanation needed
3095 * dst.y = max(src.y, 0.0); ... If x < 0.0, use 0.0, otherwise x. Same as the conditional
3096 * dst.z = x > 0.0 ? pow(max(y, 0.0), p) : 0; ... 0 ^ power is 0, and otherwise we use y anyway
3097 * dst.w = 1.0. ... Nothing fancy.
3099 * So we still have one conditional in there. So do this:
3100 * dst.z = pow(max(0.0, src.y) * step(0.0, src.x), power);
3102 * 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),
3103 * 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.
3104 * if both x and y are > 0, we get pow(y * 1.0, power), as it is supposed to.
3106 * Unfortunately pow(0.0 ^ 0.0) returns NaN on most GPUs, but lit with src.y = 0 and src.w = 0 returns
3107 * a non-NaN value in dst.z. What we return doesn't matter, as long as it is not NaN. Return 0, which is
3108 * what all Windows HW drivers and GL_ARB_vertex_program's LIT do.
3110 shader_addline(ins
->ctx
->buffer
,
3111 "vec4(1.0, max(%s, 0.0), %s == 0.0 ? 0.0 : "
3112 "pow(max(0.0, %s) * step(0.0, %s), clamp(%s, -128.0, 128.0)), 1.0)%s);\n",
3113 src0_param
.param_str
, src3_param
.param_str
, src1_param
.param_str
,
3114 src0_param
.param_str
, src3_param
.param_str
, dst_mask
);
3117 /** Process the WINED3DSIO_DST instruction in GLSL:
3119 * dst.y = src0.x * src0.y
3123 static void shader_glsl_dst(const struct wined3d_shader_instruction
*ins
)
3125 struct glsl_src_param src0y_param
;
3126 struct glsl_src_param src0z_param
;
3127 struct glsl_src_param src1y_param
;
3128 struct glsl_src_param src1w_param
;
3131 shader_glsl_append_dst(ins
->ctx
->buffer
, ins
);
3132 shader_glsl_get_write_mask(&ins
->dst
[0], dst_mask
);
3134 shader_glsl_add_src_param(ins
, &ins
->src
[0], WINED3DSP_WRITEMASK_1
, &src0y_param
);
3135 shader_glsl_add_src_param(ins
, &ins
->src
[0], WINED3DSP_WRITEMASK_2
, &src0z_param
);
3136 shader_glsl_add_src_param(ins
, &ins
->src
[1], WINED3DSP_WRITEMASK_1
, &src1y_param
);
3137 shader_glsl_add_src_param(ins
, &ins
->src
[1], WINED3DSP_WRITEMASK_3
, &src1w_param
);
3139 shader_addline(ins
->ctx
->buffer
, "vec4(1.0, %s * %s, %s, %s))%s;\n",
3140 src0y_param
.param_str
, src1y_param
.param_str
, src0z_param
.param_str
, src1w_param
.param_str
, dst_mask
);
3143 /** Process the WINED3DSIO_SINCOS instruction in GLSL:
3144 * VS 2.0 requires that specific cosine and sine constants be passed to this instruction so the hardware
3145 * can handle it. But, these functions are built-in for GLSL, so we can just ignore the last 2 params.
3147 * dst.x = cos(src0.?)
3148 * dst.y = sin(src0.?)
3152 static void shader_glsl_sincos(const struct wined3d_shader_instruction
*ins
)
3154 struct wined3d_shader_buffer
*buffer
= ins
->ctx
->buffer
;
3155 struct glsl_src_param src0_param
;
3158 if (ins
->ctx
->reg_maps
->shader_version
.major
< 4)
3160 shader_glsl_add_src_param(ins
, &ins
->src
[0], WINED3DSP_WRITEMASK_0
, &src0_param
);
3162 write_mask
= shader_glsl_append_dst(buffer
, ins
);
3165 case WINED3DSP_WRITEMASK_0
:
3166 shader_addline(buffer
, "cos(%s));\n", src0_param
.param_str
);
3169 case WINED3DSP_WRITEMASK_1
:
3170 shader_addline(buffer
, "sin(%s));\n", src0_param
.param_str
);
3173 case (WINED3DSP_WRITEMASK_0
| WINED3DSP_WRITEMASK_1
):
3174 shader_addline(buffer
, "vec2(cos(%s), sin(%s)));\n",
3175 src0_param
.param_str
, src0_param
.param_str
);
3179 ERR("Write mask should be .x, .y or .xy\n");
3186 if (ins
->dst
[0].reg
.type
!= WINED3DSPR_NULL
)
3189 if (ins
->dst
[1].reg
.type
!= WINED3DSPR_NULL
)
3193 write_mask
= shader_glsl_get_write_mask(&ins
->dst
[0], dst_mask
);
3194 shader_glsl_add_src_param(ins
, &ins
->src
[0], write_mask
, &src0_param
);
3195 shader_addline(buffer
, "tmp0%s = sin(%s);\n", dst_mask
, src0_param
.param_str
);
3197 write_mask
= shader_glsl_append_dst_ext(buffer
, ins
, &ins
->dst
[1]);
3198 shader_glsl_add_src_param(ins
, &ins
->src
[0], write_mask
, &src0_param
);
3199 shader_addline(buffer
, "cos(%s));\n", src0_param
.param_str
);
3201 shader_glsl_append_dst_ext(buffer
, ins
, &ins
->dst
[0]);
3202 shader_addline(buffer
, "tmp0%s);\n", dst_mask
);
3206 write_mask
= shader_glsl_append_dst_ext(buffer
, ins
, &ins
->dst
[0]);
3207 shader_glsl_add_src_param(ins
, &ins
->src
[0], write_mask
, &src0_param
);
3208 shader_addline(buffer
, "sin(%s));\n", src0_param
.param_str
);
3211 else if (ins
->dst
[1].reg
.type
!= WINED3DSPR_NULL
)
3213 write_mask
= shader_glsl_append_dst_ext(buffer
, ins
, &ins
->dst
[1]);
3214 shader_glsl_add_src_param(ins
, &ins
->src
[0], write_mask
, &src0_param
);
3215 shader_addline(buffer
, "cos(%s));\n", src0_param
.param_str
);
3219 /* sgn in vs_2_0 has 2 extra parameters(registers for temporary storage) which we don't use
3220 * here. But those extra parameters require a dedicated function for sgn, since map2gl would
3221 * generate invalid code
3223 static void shader_glsl_sgn(const struct wined3d_shader_instruction
*ins
)
3225 struct glsl_src_param src0_param
;
3228 write_mask
= shader_glsl_append_dst(ins
->ctx
->buffer
, ins
);
3229 shader_glsl_add_src_param(ins
, &ins
->src
[0], write_mask
, &src0_param
);
3231 shader_addline(ins
->ctx
->buffer
, "sign(%s));\n", src0_param
.param_str
);
3234 /** Process the WINED3DSIO_LOOP instruction in GLSL:
3235 * Start a for() loop where src1.y is the initial value of aL,
3236 * increment aL by src1.z for a total of src1.x iterations.
3237 * Need to use a temporary variable for this operation.
3239 /* FIXME: I don't think nested loops will work correctly this way. */
3240 static void shader_glsl_loop(const struct wined3d_shader_instruction
*ins
)
3242 struct wined3d_shader_loop_state
*loop_state
= ins
->ctx
->loop_state
;
3243 struct wined3d_shader_buffer
*buffer
= ins
->ctx
->buffer
;
3244 const struct wined3d_shader
*shader
= ins
->ctx
->shader
;
3245 const struct wined3d_shader_lconst
*constant
;
3246 struct glsl_src_param src1_param
;
3247 const DWORD
*control_values
= NULL
;
3249 if (ins
->ctx
->reg_maps
->shader_version
.major
< 4)
3251 shader_glsl_add_src_param(ins
, &ins
->src
[1], WINED3DSP_WRITEMASK_ALL
, &src1_param
);
3253 /* Try to hardcode the loop control parameters if possible. Direct3D 9
3254 * class hardware doesn't support real varying indexing, but Microsoft
3255 * designed this feature for Shader model 2.x+. If the loop control is
3256 * known at compile time, the GLSL compiler can unroll the loop, and
3257 * replace indirect addressing with direct addressing. */
3258 if (ins
->src
[1].reg
.type
== WINED3DSPR_CONSTINT
)
3260 LIST_FOR_EACH_ENTRY(constant
, &shader
->constantsI
, struct wined3d_shader_lconst
, entry
)
3262 if (constant
->idx
== ins
->src
[1].reg
.idx
[0].offset
)
3264 control_values
= constant
->value
;
3272 struct wined3d_shader_loop_control loop_control
;
3273 loop_control
.count
= control_values
[0];
3274 loop_control
.start
= control_values
[1];
3275 loop_control
.step
= (int)control_values
[2];
3277 if (loop_control
.step
> 0)
3279 shader_addline(buffer
, "for (aL%u = %u; aL%u < (%u * %d + %u); aL%u += %d)\n{\n",
3280 loop_state
->current_depth
, loop_control
.start
,
3281 loop_state
->current_depth
, loop_control
.count
, loop_control
.step
, loop_control
.start
,
3282 loop_state
->current_depth
, loop_control
.step
);
3284 else if (loop_control
.step
< 0)
3286 shader_addline(buffer
, "for (aL%u = %u; aL%u > (%u * %d + %u); aL%u += %d)\n{\n",
3287 loop_state
->current_depth
, loop_control
.start
,
3288 loop_state
->current_depth
, loop_control
.count
, loop_control
.step
, loop_control
.start
,
3289 loop_state
->current_depth
, loop_control
.step
);
3293 shader_addline(buffer
, "for (aL%u = %u, tmpInt%u = 0; tmpInt%u < %u; tmpInt%u++)\n{\n",
3294 loop_state
->current_depth
, loop_control
.start
, loop_state
->current_depth
,
3295 loop_state
->current_depth
, loop_control
.count
,
3296 loop_state
->current_depth
);
3301 shader_addline(buffer
, "for (tmpInt%u = 0, aL%u = %s.y; tmpInt%u < %s.x; tmpInt%u++, aL%u += %s.z)\n{\n",
3302 loop_state
->current_depth
, loop_state
->current_reg
,
3303 src1_param
.reg_name
, loop_state
->current_depth
, src1_param
.reg_name
,
3304 loop_state
->current_depth
, loop_state
->current_reg
, src1_param
.reg_name
);
3307 ++loop_state
->current_reg
;
3311 shader_addline(buffer
, "for (;;)\n{\n");
3314 ++loop_state
->current_depth
;
3317 static void shader_glsl_end(const struct wined3d_shader_instruction
*ins
)
3319 struct wined3d_shader_loop_state
*loop_state
= ins
->ctx
->loop_state
;
3321 shader_addline(ins
->ctx
->buffer
, "}\n");
3323 if (ins
->handler_idx
== WINED3DSIH_ENDLOOP
)
3325 --loop_state
->current_depth
;
3326 --loop_state
->current_reg
;
3329 if (ins
->handler_idx
== WINED3DSIH_ENDREP
)
3331 --loop_state
->current_depth
;
3335 static void shader_glsl_rep(const struct wined3d_shader_instruction
*ins
)
3337 const struct wined3d_shader
*shader
= ins
->ctx
->shader
;
3338 struct wined3d_shader_loop_state
*loop_state
= ins
->ctx
->loop_state
;
3339 const struct wined3d_shader_lconst
*constant
;
3340 struct glsl_src_param src0_param
;
3341 const DWORD
*control_values
= NULL
;
3343 /* Try to hardcode local values to help the GLSL compiler to unroll and optimize the loop */
3344 if (ins
->src
[0].reg
.type
== WINED3DSPR_CONSTINT
)
3346 LIST_FOR_EACH_ENTRY(constant
, &shader
->constantsI
, struct wined3d_shader_lconst
, entry
)
3348 if (constant
->idx
== ins
->src
[0].reg
.idx
[0].offset
)
3350 control_values
= constant
->value
;
3358 shader_addline(ins
->ctx
->buffer
, "for (tmpInt%d = 0; tmpInt%d < %d; tmpInt%d++) {\n",
3359 loop_state
->current_depth
, loop_state
->current_depth
,
3360 control_values
[0], loop_state
->current_depth
);
3364 shader_glsl_add_src_param(ins
, &ins
->src
[0], WINED3DSP_WRITEMASK_0
, &src0_param
);
3365 shader_addline(ins
->ctx
->buffer
, "for (tmpInt%d = 0; tmpInt%d < %s; tmpInt%d++) {\n",
3366 loop_state
->current_depth
, loop_state
->current_depth
,
3367 src0_param
.param_str
, loop_state
->current_depth
);
3370 ++loop_state
->current_depth
;
3373 static void shader_glsl_if(const struct wined3d_shader_instruction
*ins
)
3375 struct glsl_src_param src0_param
;
3377 shader_glsl_add_src_param(ins
, &ins
->src
[0], WINED3DSP_WRITEMASK_0
, &src0_param
);
3378 shader_addline(ins
->ctx
->buffer
, "if (%s) {\n", src0_param
.param_str
);
3381 static void shader_glsl_ifc(const struct wined3d_shader_instruction
*ins
)
3383 struct glsl_src_param src0_param
;
3384 struct glsl_src_param src1_param
;
3386 shader_glsl_add_src_param(ins
, &ins
->src
[0], WINED3DSP_WRITEMASK_0
, &src0_param
);
3387 shader_glsl_add_src_param(ins
, &ins
->src
[1], WINED3DSP_WRITEMASK_0
, &src1_param
);
3389 shader_addline(ins
->ctx
->buffer
, "if (%s %s %s) {\n",
3390 src0_param
.param_str
, shader_glsl_get_rel_op(ins
->flags
), src1_param
.param_str
);
3393 static void shader_glsl_else(const struct wined3d_shader_instruction
*ins
)
3395 shader_addline(ins
->ctx
->buffer
, "} else {\n");
3398 static void shader_glsl_emit(const struct wined3d_shader_instruction
*ins
)
3400 shader_addline(ins
->ctx
->buffer
, "EmitVertex();\n");
3403 static void shader_glsl_break(const struct wined3d_shader_instruction
*ins
)
3405 shader_addline(ins
->ctx
->buffer
, "break;\n");
3408 /* FIXME: According to MSDN the compare is done per component. */
3409 static void shader_glsl_breakc(const struct wined3d_shader_instruction
*ins
)
3411 struct glsl_src_param src0_param
;
3412 struct glsl_src_param src1_param
;
3414 shader_glsl_add_src_param(ins
, &ins
->src
[0], WINED3DSP_WRITEMASK_0
, &src0_param
);
3415 shader_glsl_add_src_param(ins
, &ins
->src
[1], WINED3DSP_WRITEMASK_0
, &src1_param
);
3417 shader_addline(ins
->ctx
->buffer
, "if (%s %s %s) break;\n",
3418 src0_param
.param_str
, shader_glsl_get_rel_op(ins
->flags
), src1_param
.param_str
);
3421 static void shader_glsl_breakp(const struct wined3d_shader_instruction
*ins
)
3423 struct glsl_src_param src_param
;
3425 shader_glsl_add_src_param(ins
, &ins
->src
[0], WINED3DSP_WRITEMASK_0
, &src_param
);
3426 shader_addline(ins
->ctx
->buffer
, "if (bool(%s)) break;\n", src_param
.param_str
);
3429 static void shader_glsl_label(const struct wined3d_shader_instruction
*ins
)
3431 shader_addline(ins
->ctx
->buffer
, "}\n");
3432 shader_addline(ins
->ctx
->buffer
, "void subroutine%u()\n{\n", ins
->src
[0].reg
.idx
[0].offset
);
3435 static void shader_glsl_call(const struct wined3d_shader_instruction
*ins
)
3437 shader_addline(ins
->ctx
->buffer
, "subroutine%u();\n", ins
->src
[0].reg
.idx
[0].offset
);
3440 static void shader_glsl_callnz(const struct wined3d_shader_instruction
*ins
)
3442 struct glsl_src_param src1_param
;
3444 shader_glsl_add_src_param(ins
, &ins
->src
[1], WINED3DSP_WRITEMASK_0
, &src1_param
);
3445 shader_addline(ins
->ctx
->buffer
, "if (%s) subroutine%u();\n",
3446 src1_param
.param_str
, ins
->src
[0].reg
.idx
[0].offset
);
3449 static void shader_glsl_ret(const struct wined3d_shader_instruction
*ins
)
3451 /* No-op. The closing } is written when a new function is started, and at the end of the shader. This
3452 * function only suppresses the unhandled instruction warning
3456 /*********************************************
3457 * Pixel Shader Specific Code begins here
3458 ********************************************/
3459 static void shader_glsl_tex(const struct wined3d_shader_instruction
*ins
)
3461 const struct wined3d_shader
*shader
= ins
->ctx
->shader
;
3462 struct wined3d_device
*device
= shader
->device
;
3463 DWORD shader_version
= WINED3D_SHADER_VERSION(ins
->ctx
->reg_maps
->shader_version
.major
,
3464 ins
->ctx
->reg_maps
->shader_version
.minor
);
3465 struct glsl_sample_function sample_function
;
3466 const struct wined3d_texture
*texture
;
3467 DWORD sample_flags
= 0;
3469 DWORD mask
= 0, swizzle
;
3471 /* 1.0-1.4: Use destination register as sampler source.
3472 * 2.0+: Use provided sampler source. */
3473 if (shader_version
< WINED3D_SHADER_VERSION(2,0))
3474 sampler_idx
= ins
->dst
[0].reg
.idx
[0].offset
;
3476 sampler_idx
= ins
->src
[1].reg
.idx
[0].offset
;
3477 texture
= device
->stateBlock
->state
.textures
[sampler_idx
];
3479 if (shader_version
< WINED3D_SHADER_VERSION(1,4))
3481 const struct shader_glsl_ctx_priv
*priv
= ins
->ctx
->backend_data
;
3482 DWORD flags
= (priv
->cur_ps_args
->tex_transform
>> sampler_idx
* WINED3D_PSARGS_TEXTRANSFORM_SHIFT
)
3483 & WINED3D_PSARGS_TEXTRANSFORM_MASK
;
3484 enum wined3d_sampler_texture_type sampler_type
= ins
->ctx
->reg_maps
->sampler_type
[sampler_idx
];
3486 /* Projected cube textures don't make a lot of sense, the resulting coordinates stay the same. */
3487 if (flags
& WINED3D_PSARGS_PROJECTED
&& sampler_type
!= WINED3DSTT_CUBE
)
3489 sample_flags
|= WINED3D_GLSL_SAMPLE_PROJECTED
;
3490 switch (flags
& ~WINED3D_PSARGS_PROJECTED
)
3492 case WINED3D_TTFF_COUNT1
:
3493 FIXME("WINED3D_TTFF_PROJECTED with WINED3D_TTFF_COUNT1?\n");
3495 case WINED3D_TTFF_COUNT2
:
3496 mask
= WINED3DSP_WRITEMASK_1
;
3498 case WINED3D_TTFF_COUNT3
:
3499 mask
= WINED3DSP_WRITEMASK_2
;
3501 case WINED3D_TTFF_COUNT4
:
3502 case WINED3D_TTFF_DISABLE
:
3503 mask
= WINED3DSP_WRITEMASK_3
;
3508 else if (shader_version
< WINED3D_SHADER_VERSION(2,0))
3510 enum wined3d_shader_src_modifier src_mod
= ins
->src
[0].modifiers
;
3512 if (src_mod
== WINED3DSPSM_DZ
) {
3513 sample_flags
|= WINED3D_GLSL_SAMPLE_PROJECTED
;
3514 mask
= WINED3DSP_WRITEMASK_2
;
3515 } else if (src_mod
== WINED3DSPSM_DW
) {
3516 sample_flags
|= WINED3D_GLSL_SAMPLE_PROJECTED
;
3517 mask
= WINED3DSP_WRITEMASK_3
;
3520 if (ins
->flags
& WINED3DSI_TEXLD_PROJECT
)
3522 /* ps 2.0 texldp instruction always divides by the fourth component. */
3523 sample_flags
|= WINED3D_GLSL_SAMPLE_PROJECTED
;
3524 mask
= WINED3DSP_WRITEMASK_3
;
3528 if (texture
&& texture
->target
== GL_TEXTURE_RECTANGLE_ARB
)
3529 sample_flags
|= WINED3D_GLSL_SAMPLE_RECT
;
3531 shader_glsl_get_sample_function(ins
->ctx
, sampler_idx
, sample_flags
, &sample_function
);
3532 mask
|= sample_function
.coord_mask
;
3534 if (shader_version
< WINED3D_SHADER_VERSION(2,0)) swizzle
= WINED3DSP_NOSWIZZLE
;
3535 else swizzle
= ins
->src
[1].swizzle
;
3537 /* 1.0-1.3: Use destination register as coordinate source.
3538 1.4+: Use provided coordinate source register. */
3539 if (shader_version
< WINED3D_SHADER_VERSION(1,4))
3542 shader_glsl_write_mask_to_str(mask
, coord_mask
);
3543 shader_glsl_gen_sample_code(ins
, sampler_idx
, &sample_function
, swizzle
, NULL
, NULL
, NULL
,
3544 "T%u%s", sampler_idx
, coord_mask
);
3548 struct glsl_src_param coord_param
;
3549 shader_glsl_add_src_param(ins
, &ins
->src
[0], mask
, &coord_param
);
3550 if (ins
->flags
& WINED3DSI_TEXLD_BIAS
)
3552 struct glsl_src_param bias
;
3553 shader_glsl_add_src_param(ins
, &ins
->src
[0], WINED3DSP_WRITEMASK_3
, &bias
);
3554 shader_glsl_gen_sample_code(ins
, sampler_idx
, &sample_function
, swizzle
, NULL
, NULL
, bias
.param_str
,
3555 "%s", coord_param
.param_str
);
3557 shader_glsl_gen_sample_code(ins
, sampler_idx
, &sample_function
, swizzle
, NULL
, NULL
, NULL
,
3558 "%s", coord_param
.param_str
);
3563 static void shader_glsl_texldd(const struct wined3d_shader_instruction
*ins
)
3565 const struct wined3d_shader
*shader
= ins
->ctx
->shader
;
3566 struct wined3d_device
*device
= shader
->device
;
3567 const struct wined3d_gl_info
*gl_info
= ins
->ctx
->gl_info
;
3568 struct glsl_src_param coord_param
, dx_param
, dy_param
;
3569 DWORD sample_flags
= WINED3D_GLSL_SAMPLE_GRAD
;
3570 struct glsl_sample_function sample_function
;
3572 DWORD swizzle
= ins
->src
[1].swizzle
;
3573 const struct wined3d_texture
*texture
;
3575 if (!gl_info
->supported
[ARB_SHADER_TEXTURE_LOD
] && !gl_info
->supported
[EXT_GPU_SHADER4
])
3577 FIXME("texldd used, but not supported by hardware. Falling back to regular tex\n");
3578 shader_glsl_tex(ins
);
3582 sampler_idx
= ins
->src
[1].reg
.idx
[0].offset
;
3583 texture
= device
->stateBlock
->state
.textures
[sampler_idx
];
3584 if (texture
&& texture
->target
== GL_TEXTURE_RECTANGLE_ARB
)
3585 sample_flags
|= WINED3D_GLSL_SAMPLE_RECT
;
3587 shader_glsl_get_sample_function(ins
->ctx
, sampler_idx
, sample_flags
, &sample_function
);
3588 shader_glsl_add_src_param(ins
, &ins
->src
[0], sample_function
.coord_mask
, &coord_param
);
3589 shader_glsl_add_src_param(ins
, &ins
->src
[2], sample_function
.coord_mask
, &dx_param
);
3590 shader_glsl_add_src_param(ins
, &ins
->src
[3], sample_function
.coord_mask
, &dy_param
);
3592 shader_glsl_gen_sample_code(ins
, sampler_idx
, &sample_function
, swizzle
, dx_param
.param_str
, dy_param
.param_str
, NULL
,
3593 "%s", coord_param
.param_str
);
3596 static void shader_glsl_texldl(const struct wined3d_shader_instruction
*ins
)
3598 const struct wined3d_shader
*shader
= ins
->ctx
->shader
;
3599 struct wined3d_device
*device
= shader
->device
;
3600 const struct wined3d_gl_info
*gl_info
= ins
->ctx
->gl_info
;
3601 struct glsl_src_param coord_param
, lod_param
;
3602 DWORD sample_flags
= WINED3D_GLSL_SAMPLE_LOD
;
3603 struct glsl_sample_function sample_function
;
3605 DWORD swizzle
= ins
->src
[1].swizzle
;
3606 const struct wined3d_texture
*texture
;
3608 sampler_idx
= ins
->src
[1].reg
.idx
[0].offset
;
3609 texture
= device
->stateBlock
->state
.textures
[sampler_idx
];
3610 if (texture
&& texture
->target
== GL_TEXTURE_RECTANGLE_ARB
)
3611 sample_flags
|= WINED3D_GLSL_SAMPLE_RECT
;
3613 shader_glsl_get_sample_function(ins
->ctx
, sampler_idx
, sample_flags
, &sample_function
);
3614 shader_glsl_add_src_param(ins
, &ins
->src
[0], sample_function
.coord_mask
, &coord_param
);
3616 shader_glsl_add_src_param(ins
, &ins
->src
[0], WINED3DSP_WRITEMASK_3
, &lod_param
);
3618 if (!gl_info
->supported
[ARB_SHADER_TEXTURE_LOD
] && !gl_info
->supported
[EXT_GPU_SHADER4
]
3619 && ins
->ctx
->reg_maps
->shader_version
.type
== WINED3D_SHADER_TYPE_PIXEL
)
3621 /* Plain GLSL only supports Lod sampling functions in vertex shaders.
3622 * However, the NVIDIA drivers allow them in fragment shaders as well,
3623 * even without the appropriate extension. */
3624 WARN("Using %s in fragment shader.\n", sample_function
.name
);
3626 shader_glsl_gen_sample_code(ins
, sampler_idx
, &sample_function
, swizzle
, NULL
, NULL
, lod_param
.param_str
,
3627 "%s", coord_param
.param_str
);
3630 static void shader_glsl_texcoord(const struct wined3d_shader_instruction
*ins
)
3632 /* FIXME: Make this work for more than just 2D textures */
3633 struct wined3d_shader_buffer
*buffer
= ins
->ctx
->buffer
;
3634 DWORD write_mask
= shader_glsl_append_dst(ins
->ctx
->buffer
, ins
);
3636 if (!(ins
->ctx
->reg_maps
->shader_version
.major
== 1 && ins
->ctx
->reg_maps
->shader_version
.minor
== 4))
3640 shader_glsl_get_write_mask(&ins
->dst
[0], dst_mask
);
3641 shader_addline(buffer
, "clamp(gl_TexCoord[%u], 0.0, 1.0)%s);\n",
3642 ins
->dst
[0].reg
.idx
[0].offset
, dst_mask
);
3646 enum wined3d_shader_src_modifier src_mod
= ins
->src
[0].modifiers
;
3647 DWORD reg
= ins
->src
[0].reg
.idx
[0].offset
;
3648 char dst_swizzle
[6];
3650 shader_glsl_get_swizzle(&ins
->src
[0], FALSE
, write_mask
, dst_swizzle
);
3652 if (src_mod
== WINED3DSPSM_DZ
)
3654 unsigned int mask_size
= shader_glsl_get_write_mask_size(write_mask
);
3655 struct glsl_src_param div_param
;
3657 shader_glsl_add_src_param(ins
, &ins
->src
[0], WINED3DSP_WRITEMASK_2
, &div_param
);
3659 if (mask_size
> 1) {
3660 shader_addline(buffer
, "gl_TexCoord[%u]%s / vec%d(%s));\n", reg
, dst_swizzle
, mask_size
, div_param
.param_str
);
3662 shader_addline(buffer
, "gl_TexCoord[%u]%s / %s);\n", reg
, dst_swizzle
, div_param
.param_str
);
3665 else if (src_mod
== WINED3DSPSM_DW
)
3667 unsigned int mask_size
= shader_glsl_get_write_mask_size(write_mask
);
3668 struct glsl_src_param div_param
;
3670 shader_glsl_add_src_param(ins
, &ins
->src
[0], WINED3DSP_WRITEMASK_3
, &div_param
);
3672 if (mask_size
> 1) {
3673 shader_addline(buffer
, "gl_TexCoord[%u]%s / vec%d(%s));\n", reg
, dst_swizzle
, mask_size
, div_param
.param_str
);
3675 shader_addline(buffer
, "gl_TexCoord[%u]%s / %s);\n", reg
, dst_swizzle
, div_param
.param_str
);
3678 shader_addline(buffer
, "gl_TexCoord[%u]%s);\n", reg
, dst_swizzle
);
3683 /** Process the WINED3DSIO_TEXDP3TEX instruction in GLSL:
3684 * Take a 3-component dot product of the TexCoord[dstreg] and src,
3685 * then perform a 1D texture lookup from stage dstregnum, place into dst. */
3686 static void shader_glsl_texdp3tex(const struct wined3d_shader_instruction
*ins
)
3688 DWORD src_mask
= WINED3DSP_WRITEMASK_0
| WINED3DSP_WRITEMASK_1
| WINED3DSP_WRITEMASK_2
;
3689 DWORD sampler_idx
= ins
->dst
[0].reg
.idx
[0].offset
;
3690 struct glsl_sample_function sample_function
;
3691 struct glsl_src_param src0_param
;
3694 shader_glsl_add_src_param(ins
, &ins
->src
[0], src_mask
, &src0_param
);
3696 /* Do I have to take care about the projected bit? I don't think so, since the dp3 returns only one
3697 * scalar, and projected sampling would require 4.
3699 * It is a dependent read - not valid with conditional NP2 textures
3701 shader_glsl_get_sample_function(ins
->ctx
, sampler_idx
, 0, &sample_function
);
3702 mask_size
= shader_glsl_get_write_mask_size(sample_function
.coord_mask
);
3707 shader_glsl_gen_sample_code(ins
, sampler_idx
, &sample_function
, WINED3DSP_NOSWIZZLE
, NULL
, NULL
, NULL
,
3708 "dot(gl_TexCoord[%u].xyz, %s)", sampler_idx
, src0_param
.param_str
);
3712 shader_glsl_gen_sample_code(ins
, sampler_idx
, &sample_function
, WINED3DSP_NOSWIZZLE
, NULL
, NULL
, NULL
,
3713 "vec2(dot(gl_TexCoord[%u].xyz, %s), 0.0)", sampler_idx
, src0_param
.param_str
);
3717 shader_glsl_gen_sample_code(ins
, sampler_idx
, &sample_function
, WINED3DSP_NOSWIZZLE
, NULL
, NULL
, NULL
,
3718 "vec3(dot(gl_TexCoord[%u].xyz, %s), 0.0, 0.0)", sampler_idx
, src0_param
.param_str
);
3722 FIXME("Unexpected mask size %u\n", mask_size
);
3727 /** Process the WINED3DSIO_TEXDP3 instruction in GLSL:
3728 * Take a 3-component dot product of the TexCoord[dstreg] and src. */
3729 static void shader_glsl_texdp3(const struct wined3d_shader_instruction
*ins
)
3731 DWORD src_mask
= WINED3DSP_WRITEMASK_0
| WINED3DSP_WRITEMASK_1
| WINED3DSP_WRITEMASK_2
;
3732 DWORD dstreg
= ins
->dst
[0].reg
.idx
[0].offset
;
3733 struct glsl_src_param src0_param
;
3735 unsigned int mask_size
;
3737 dst_mask
= shader_glsl_append_dst(ins
->ctx
->buffer
, ins
);
3738 mask_size
= shader_glsl_get_write_mask_size(dst_mask
);
3739 shader_glsl_add_src_param(ins
, &ins
->src
[0], src_mask
, &src0_param
);
3741 if (mask_size
> 1) {
3742 shader_addline(ins
->ctx
->buffer
, "vec%d(dot(T%u.xyz, %s)));\n", mask_size
, dstreg
, src0_param
.param_str
);
3744 shader_addline(ins
->ctx
->buffer
, "dot(T%u.xyz, %s));\n", dstreg
, src0_param
.param_str
);
3748 /** Process the WINED3DSIO_TEXDEPTH instruction in GLSL:
3749 * Calculate the depth as dst.x / dst.y */
3750 static void shader_glsl_texdepth(const struct wined3d_shader_instruction
*ins
)
3752 struct glsl_dst_param dst_param
;
3754 shader_glsl_add_dst_param(ins
, &ins
->dst
[0], &dst_param
);
3756 /* Tests show that texdepth never returns anything below 0.0, and that r5.y is clamped to 1.0.
3757 * Negative input is accepted, -0.25 / -0.5 returns 0.5. GL should clamp gl_FragDepth to [0;1], but
3758 * this doesn't always work, so clamp the results manually. Whether or not the x value is clamped at 1
3759 * too is irrelevant, since if x = 0, any y value < 1.0 (and > 1.0 is not allowed) results in a result
3762 shader_addline(ins
->ctx
->buffer
, "gl_FragDepth = clamp((%s.x / min(%s.y, 1.0)), 0.0, 1.0);\n",
3763 dst_param
.reg_name
, dst_param
.reg_name
);
3766 /** Process the WINED3DSIO_TEXM3X2DEPTH instruction in GLSL:
3767 * Last row of a 3x2 matrix multiply, use the result to calculate the depth:
3768 * Calculate tmp0.y = TexCoord[dstreg] . src.xyz; (tmp0.x has already been calculated)
3769 * depth = (tmp0.y == 0.0) ? 1.0 : tmp0.x / tmp0.y
3771 static void shader_glsl_texm3x2depth(const struct wined3d_shader_instruction
*ins
)
3773 DWORD src_mask
= WINED3DSP_WRITEMASK_0
| WINED3DSP_WRITEMASK_1
| WINED3DSP_WRITEMASK_2
;
3774 DWORD dstreg
= ins
->dst
[0].reg
.idx
[0].offset
;
3775 struct glsl_src_param src0_param
;
3777 shader_glsl_add_src_param(ins
, &ins
->src
[0], src_mask
, &src0_param
);
3779 shader_addline(ins
->ctx
->buffer
, "tmp0.y = dot(T%u.xyz, %s);\n", dstreg
, src0_param
.param_str
);
3780 shader_addline(ins
->ctx
->buffer
, "gl_FragDepth = (tmp0.y == 0.0) ? 1.0 : clamp(tmp0.x / tmp0.y, 0.0, 1.0);\n");
3783 /** Process the WINED3DSIO_TEXM3X2PAD instruction in GLSL
3784 * Calculate the 1st of a 2-row matrix multiplication. */
3785 static void shader_glsl_texm3x2pad(const struct wined3d_shader_instruction
*ins
)
3787 DWORD src_mask
= WINED3DSP_WRITEMASK_0
| WINED3DSP_WRITEMASK_1
| WINED3DSP_WRITEMASK_2
;
3788 DWORD reg
= ins
->dst
[0].reg
.idx
[0].offset
;
3789 struct wined3d_shader_buffer
*buffer
= ins
->ctx
->buffer
;
3790 struct glsl_src_param src0_param
;
3792 shader_glsl_add_src_param(ins
, &ins
->src
[0], src_mask
, &src0_param
);
3793 shader_addline(buffer
, "tmp0.x = dot(T%u.xyz, %s);\n", reg
, src0_param
.param_str
);
3796 /** Process the WINED3DSIO_TEXM3X3PAD instruction in GLSL
3797 * Calculate the 1st or 2nd row of a 3-row matrix multiplication. */
3798 static void shader_glsl_texm3x3pad(const struct wined3d_shader_instruction
*ins
)
3800 DWORD src_mask
= WINED3DSP_WRITEMASK_0
| WINED3DSP_WRITEMASK_1
| WINED3DSP_WRITEMASK_2
;
3801 struct wined3d_shader_buffer
*buffer
= ins
->ctx
->buffer
;
3802 struct wined3d_shader_tex_mx
*tex_mx
= ins
->ctx
->tex_mx
;
3803 DWORD reg
= ins
->dst
[0].reg
.idx
[0].offset
;
3804 struct glsl_src_param src0_param
;
3806 shader_glsl_add_src_param(ins
, &ins
->src
[0], src_mask
, &src0_param
);
3807 shader_addline(buffer
, "tmp0.%c = dot(T%u.xyz, %s);\n", 'x' + tex_mx
->current_row
, reg
, src0_param
.param_str
);
3808 tex_mx
->texcoord_w
[tex_mx
->current_row
++] = reg
;
3811 static void shader_glsl_texm3x2tex(const struct wined3d_shader_instruction
*ins
)
3813 DWORD src_mask
= WINED3DSP_WRITEMASK_0
| WINED3DSP_WRITEMASK_1
| WINED3DSP_WRITEMASK_2
;
3814 struct wined3d_shader_buffer
*buffer
= ins
->ctx
->buffer
;
3815 struct glsl_sample_function sample_function
;
3816 DWORD reg
= ins
->dst
[0].reg
.idx
[0].offset
;
3817 struct glsl_src_param src0_param
;
3819 shader_glsl_add_src_param(ins
, &ins
->src
[0], src_mask
, &src0_param
);
3820 shader_addline(buffer
, "tmp0.y = dot(T%u.xyz, %s);\n", reg
, src0_param
.param_str
);
3822 shader_glsl_get_sample_function(ins
->ctx
, reg
, 0, &sample_function
);
3824 /* Sample the texture using the calculated coordinates */
3825 shader_glsl_gen_sample_code(ins
, reg
, &sample_function
, WINED3DSP_NOSWIZZLE
, NULL
, NULL
, NULL
, "tmp0.xy");
3828 /** Process the WINED3DSIO_TEXM3X3TEX instruction in GLSL
3829 * Perform the 3rd row of a 3x3 matrix multiply, then sample the texture using the calculated coordinates */
3830 static void shader_glsl_texm3x3tex(const struct wined3d_shader_instruction
*ins
)
3832 DWORD src_mask
= WINED3DSP_WRITEMASK_0
| WINED3DSP_WRITEMASK_1
| WINED3DSP_WRITEMASK_2
;
3833 struct wined3d_shader_tex_mx
*tex_mx
= ins
->ctx
->tex_mx
;
3834 struct glsl_sample_function sample_function
;
3835 DWORD reg
= ins
->dst
[0].reg
.idx
[0].offset
;
3836 struct glsl_src_param src0_param
;
3838 shader_glsl_add_src_param(ins
, &ins
->src
[0], src_mask
, &src0_param
);
3839 shader_addline(ins
->ctx
->buffer
, "tmp0.z = dot(T%u.xyz, %s);\n", reg
, src0_param
.param_str
);
3841 /* Dependent read, not valid with conditional NP2 */
3842 shader_glsl_get_sample_function(ins
->ctx
, reg
, 0, &sample_function
);
3844 /* Sample the texture using the calculated coordinates */
3845 shader_glsl_gen_sample_code(ins
, reg
, &sample_function
, WINED3DSP_NOSWIZZLE
, NULL
, NULL
, NULL
, "tmp0.xyz");
3847 tex_mx
->current_row
= 0;
3850 /** Process the WINED3DSIO_TEXM3X3 instruction in GLSL
3851 * Perform the 3rd row of a 3x3 matrix multiply */
3852 static void shader_glsl_texm3x3(const struct wined3d_shader_instruction
*ins
)
3854 DWORD src_mask
= WINED3DSP_WRITEMASK_0
| WINED3DSP_WRITEMASK_1
| WINED3DSP_WRITEMASK_2
;
3855 struct wined3d_shader_tex_mx
*tex_mx
= ins
->ctx
->tex_mx
;
3856 DWORD reg
= ins
->dst
[0].reg
.idx
[0].offset
;
3857 struct glsl_src_param src0_param
;
3860 shader_glsl_add_src_param(ins
, &ins
->src
[0], src_mask
, &src0_param
);
3862 shader_glsl_append_dst(ins
->ctx
->buffer
, ins
);
3863 shader_glsl_get_write_mask(&ins
->dst
[0], dst_mask
);
3864 shader_addline(ins
->ctx
->buffer
, "vec4(tmp0.xy, dot(T%u.xyz, %s), 1.0)%s);\n", reg
, src0_param
.param_str
, dst_mask
);
3866 tex_mx
->current_row
= 0;
3869 /* Process the WINED3DSIO_TEXM3X3SPEC instruction in GLSL
3870 * Perform the final texture lookup based on the previous 2 3x3 matrix multiplies */
3871 static void shader_glsl_texm3x3spec(const struct wined3d_shader_instruction
*ins
)
3873 struct glsl_src_param src0_param
;
3874 struct glsl_src_param src1_param
;
3875 struct wined3d_shader_buffer
*buffer
= ins
->ctx
->buffer
;
3876 struct wined3d_shader_tex_mx
*tex_mx
= ins
->ctx
->tex_mx
;
3877 DWORD src_mask
= WINED3DSP_WRITEMASK_0
| WINED3DSP_WRITEMASK_1
| WINED3DSP_WRITEMASK_2
;
3878 struct glsl_sample_function sample_function
;
3879 DWORD reg
= ins
->dst
[0].reg
.idx
[0].offset
;
3882 shader_glsl_add_src_param(ins
, &ins
->src
[0], src_mask
, &src0_param
);
3883 shader_glsl_add_src_param(ins
, &ins
->src
[1], src_mask
, &src1_param
);
3885 /* Perform the last matrix multiply operation */
3886 shader_addline(buffer
, "tmp0.z = dot(T%u.xyz, %s);\n", reg
, src0_param
.param_str
);
3887 /* Reflection calculation */
3888 shader_addline(buffer
, "tmp0.xyz = -reflect((%s), normalize(tmp0.xyz));\n", src1_param
.param_str
);
3890 /* Dependent read, not valid with conditional NP2 */
3891 shader_glsl_get_sample_function(ins
->ctx
, reg
, 0, &sample_function
);
3892 shader_glsl_write_mask_to_str(sample_function
.coord_mask
, coord_mask
);
3894 /* Sample the texture */
3895 shader_glsl_gen_sample_code(ins
, reg
, &sample_function
, WINED3DSP_NOSWIZZLE
,
3896 NULL
, NULL
, NULL
, "tmp0%s", coord_mask
);
3898 tex_mx
->current_row
= 0;
3901 /* Process the WINED3DSIO_TEXM3X3VSPEC instruction in GLSL
3902 * Perform the final texture lookup based on the previous 2 3x3 matrix multiplies */
3903 static void shader_glsl_texm3x3vspec(const struct wined3d_shader_instruction
*ins
)
3905 struct wined3d_shader_buffer
*buffer
= ins
->ctx
->buffer
;
3906 struct wined3d_shader_tex_mx
*tex_mx
= ins
->ctx
->tex_mx
;
3907 DWORD src_mask
= WINED3DSP_WRITEMASK_0
| WINED3DSP_WRITEMASK_1
| WINED3DSP_WRITEMASK_2
;
3908 struct glsl_sample_function sample_function
;
3909 DWORD reg
= ins
->dst
[0].reg
.idx
[0].offset
;
3910 struct glsl_src_param src0_param
;
3913 shader_glsl_add_src_param(ins
, &ins
->src
[0], src_mask
, &src0_param
);
3915 /* Perform the last matrix multiply operation */
3916 shader_addline(buffer
, "tmp0.z = dot(vec3(T%u), vec3(%s));\n", reg
, src0_param
.param_str
);
3918 /* Construct the eye-ray vector from w coordinates */
3919 shader_addline(buffer
, "tmp1.xyz = normalize(vec3(gl_TexCoord[%u].w, gl_TexCoord[%u].w, gl_TexCoord[%u].w));\n",
3920 tex_mx
->texcoord_w
[0], tex_mx
->texcoord_w
[1], reg
);
3921 shader_addline(buffer
, "tmp0.xyz = -reflect(tmp1.xyz, normalize(tmp0.xyz));\n");
3923 /* Dependent read, not valid with conditional NP2 */
3924 shader_glsl_get_sample_function(ins
->ctx
, reg
, 0, &sample_function
);
3925 shader_glsl_write_mask_to_str(sample_function
.coord_mask
, coord_mask
);
3927 /* Sample the texture using the calculated coordinates */
3928 shader_glsl_gen_sample_code(ins
, reg
, &sample_function
, WINED3DSP_NOSWIZZLE
,
3929 NULL
, NULL
, NULL
, "tmp0%s", coord_mask
);
3931 tex_mx
->current_row
= 0;
3934 /** Process the WINED3DSIO_TEXBEM instruction in GLSL.
3935 * Apply a fake bump map transform.
3936 * texbem is pshader <= 1.3 only, this saves a few version checks
3938 static void shader_glsl_texbem(const struct wined3d_shader_instruction
*ins
)
3940 const struct shader_glsl_ctx_priv
*priv
= ins
->ctx
->backend_data
;
3941 struct glsl_sample_function sample_function
;
3942 struct glsl_src_param coord_param
;
3948 sampler_idx
= ins
->dst
[0].reg
.idx
[0].offset
;
3949 flags
= (priv
->cur_ps_args
->tex_transform
>> sampler_idx
* WINED3D_PSARGS_TEXTRANSFORM_SHIFT
)
3950 & WINED3D_PSARGS_TEXTRANSFORM_MASK
;
3952 /* Dependent read, not valid with conditional NP2 */
3953 shader_glsl_get_sample_function(ins
->ctx
, sampler_idx
, 0, &sample_function
);
3954 mask
= sample_function
.coord_mask
;
3956 shader_glsl_write_mask_to_str(mask
, coord_mask
);
3958 /* With projected textures, texbem only divides the static texture coord,
3959 * not the displacement, so we can't let GL handle this. */
3960 if (flags
& WINED3D_PSARGS_PROJECTED
)
3963 char coord_div_mask
[3];
3964 switch (flags
& ~WINED3D_PSARGS_PROJECTED
)
3966 case WINED3D_TTFF_COUNT1
:
3967 FIXME("WINED3D_TTFF_PROJECTED with WINED3D_TTFF_COUNT1?\n");
3969 case WINED3D_TTFF_COUNT2
:
3970 div_mask
= WINED3DSP_WRITEMASK_1
;
3972 case WINED3D_TTFF_COUNT3
:
3973 div_mask
= WINED3DSP_WRITEMASK_2
;
3975 case WINED3D_TTFF_COUNT4
:
3976 case WINED3D_TTFF_DISABLE
:
3977 div_mask
= WINED3DSP_WRITEMASK_3
;
3980 shader_glsl_write_mask_to_str(div_mask
, coord_div_mask
);
3981 shader_addline(ins
->ctx
->buffer
, "T%u%s /= T%u%s;\n", sampler_idx
, coord_mask
, sampler_idx
, coord_div_mask
);
3984 shader_glsl_add_src_param(ins
, &ins
->src
[0], WINED3DSP_WRITEMASK_0
| WINED3DSP_WRITEMASK_1
, &coord_param
);
3986 shader_glsl_gen_sample_code(ins
, sampler_idx
, &sample_function
, WINED3DSP_NOSWIZZLE
, NULL
, NULL
, NULL
,
3987 "T%u%s + vec4(bumpenv_mat%u * %s, 0.0, 0.0)%s", sampler_idx
, coord_mask
, sampler_idx
,
3988 coord_param
.param_str
, coord_mask
);
3990 if (ins
->handler_idx
== WINED3DSIH_TEXBEML
)
3992 struct glsl_src_param luminance_param
;
3993 struct glsl_dst_param dst_param
;
3995 shader_glsl_add_src_param(ins
, &ins
->src
[0], WINED3DSP_WRITEMASK_2
, &luminance_param
);
3996 shader_glsl_add_dst_param(ins
, &ins
->dst
[0], &dst_param
);
3998 shader_addline(ins
->ctx
->buffer
, "%s%s *= (%s * bumpenv_lum_scale%u + bumpenv_lum_offset%u);\n",
3999 dst_param
.reg_name
, dst_param
.mask_str
,
4000 luminance_param
.param_str
, sampler_idx
, sampler_idx
);
4004 static void shader_glsl_bem(const struct wined3d_shader_instruction
*ins
)
4006 DWORD sampler_idx
= ins
->dst
[0].reg
.idx
[0].offset
;
4007 struct glsl_src_param src0_param
, src1_param
;
4009 shader_glsl_add_src_param(ins
, &ins
->src
[0], WINED3DSP_WRITEMASK_0
| WINED3DSP_WRITEMASK_1
, &src0_param
);
4010 shader_glsl_add_src_param(ins
, &ins
->src
[1], WINED3DSP_WRITEMASK_0
| WINED3DSP_WRITEMASK_1
, &src1_param
);
4012 shader_glsl_append_dst(ins
->ctx
->buffer
, ins
);
4013 shader_addline(ins
->ctx
->buffer
, "%s + bumpenv_mat%u * %s);\n",
4014 src0_param
.param_str
, sampler_idx
, src1_param
.param_str
);
4017 /** Process the WINED3DSIO_TEXREG2AR instruction in GLSL
4018 * Sample 2D texture at dst using the alpha & red (wx) components of src as texture coordinates */
4019 static void shader_glsl_texreg2ar(const struct wined3d_shader_instruction
*ins
)
4021 DWORD sampler_idx
= ins
->dst
[0].reg
.idx
[0].offset
;
4022 struct glsl_sample_function sample_function
;
4023 struct glsl_src_param src0_param
;
4025 shader_glsl_add_src_param(ins
, &ins
->src
[0], WINED3DSP_WRITEMASK_ALL
, &src0_param
);
4027 shader_glsl_get_sample_function(ins
->ctx
, sampler_idx
, 0, &sample_function
);
4028 shader_glsl_gen_sample_code(ins
, sampler_idx
, &sample_function
, WINED3DSP_NOSWIZZLE
, NULL
, NULL
, NULL
,
4029 "%s.wx", src0_param
.reg_name
);
4032 /** Process the WINED3DSIO_TEXREG2GB instruction in GLSL
4033 * Sample 2D texture at dst using the green & blue (yz) components of src as texture coordinates */
4034 static void shader_glsl_texreg2gb(const struct wined3d_shader_instruction
*ins
)
4036 DWORD sampler_idx
= ins
->dst
[0].reg
.idx
[0].offset
;
4037 struct glsl_sample_function sample_function
;
4038 struct glsl_src_param src0_param
;
4040 shader_glsl_add_src_param(ins
, &ins
->src
[0], WINED3DSP_WRITEMASK_ALL
, &src0_param
);
4042 shader_glsl_get_sample_function(ins
->ctx
, sampler_idx
, 0, &sample_function
);
4043 shader_glsl_gen_sample_code(ins
, sampler_idx
, &sample_function
, WINED3DSP_NOSWIZZLE
, NULL
, NULL
, NULL
,
4044 "%s.yz", src0_param
.reg_name
);
4047 /** Process the WINED3DSIO_TEXREG2RGB instruction in GLSL
4048 * Sample texture at dst using the rgb (xyz) components of src as texture coordinates */
4049 static void shader_glsl_texreg2rgb(const struct wined3d_shader_instruction
*ins
)
4051 DWORD sampler_idx
= ins
->dst
[0].reg
.idx
[0].offset
;
4052 struct glsl_sample_function sample_function
;
4053 struct glsl_src_param src0_param
;
4055 /* Dependent read, not valid with conditional NP2 */
4056 shader_glsl_get_sample_function(ins
->ctx
, sampler_idx
, 0, &sample_function
);
4057 shader_glsl_add_src_param(ins
, &ins
->src
[0], sample_function
.coord_mask
, &src0_param
);
4059 shader_glsl_gen_sample_code(ins
, sampler_idx
, &sample_function
, WINED3DSP_NOSWIZZLE
, NULL
, NULL
, NULL
,
4060 "%s", src0_param
.param_str
);
4063 /** Process the WINED3DSIO_TEXKILL instruction in GLSL.
4064 * If any of the first 3 components are < 0, discard this pixel */
4065 static void shader_glsl_texkill(const struct wined3d_shader_instruction
*ins
)
4067 struct glsl_dst_param dst_param
;
4069 /* The argument is a destination parameter, and no writemasks are allowed */
4070 shader_glsl_add_dst_param(ins
, &ins
->dst
[0], &dst_param
);
4071 if (ins
->ctx
->reg_maps
->shader_version
.major
>= 2)
4073 /* 2.0 shaders compare all 4 components in texkill */
4074 shader_addline(ins
->ctx
->buffer
, "if (any(lessThan(%s.xyzw, vec4(0.0)))) discard;\n", dst_param
.reg_name
);
4076 /* 1.X shaders only compare the first 3 components, probably due to the nature of the texkill
4077 * instruction as a tex* instruction, and phase, which kills all a / w components. Even if all
4078 * 4 components are defined, only the first 3 are used
4080 shader_addline(ins
->ctx
->buffer
, "if (any(lessThan(%s.xyz, vec3(0.0)))) discard;\n", dst_param
.reg_name
);
4084 /** Process the WINED3DSIO_DP2ADD instruction in GLSL.
4085 * dst = dot2(src0, src1) + src2 */
4086 static void shader_glsl_dp2add(const struct wined3d_shader_instruction
*ins
)
4088 struct glsl_src_param src0_param
;
4089 struct glsl_src_param src1_param
;
4090 struct glsl_src_param src2_param
;
4092 unsigned int mask_size
;
4094 write_mask
= shader_glsl_append_dst(ins
->ctx
->buffer
, ins
);
4095 mask_size
= shader_glsl_get_write_mask_size(write_mask
);
4097 shader_glsl_add_src_param(ins
, &ins
->src
[0], WINED3DSP_WRITEMASK_0
| WINED3DSP_WRITEMASK_1
, &src0_param
);
4098 shader_glsl_add_src_param(ins
, &ins
->src
[1], WINED3DSP_WRITEMASK_0
| WINED3DSP_WRITEMASK_1
, &src1_param
);
4099 shader_glsl_add_src_param(ins
, &ins
->src
[2], WINED3DSP_WRITEMASK_0
, &src2_param
);
4101 if (mask_size
> 1) {
4102 shader_addline(ins
->ctx
->buffer
, "vec%d(dot(%s, %s) + %s));\n",
4103 mask_size
, src0_param
.param_str
, src1_param
.param_str
, src2_param
.param_str
);
4105 shader_addline(ins
->ctx
->buffer
, "dot(%s, %s) + %s);\n",
4106 src0_param
.param_str
, src1_param
.param_str
, src2_param
.param_str
);
4110 static void shader_glsl_input_pack(const struct wined3d_shader
*shader
, struct wined3d_shader_buffer
*buffer
,
4111 const struct wined3d_shader_signature_element
*input_signature
,
4112 const struct wined3d_shader_reg_maps
*reg_maps
,
4113 enum vertexprocessing_mode vertexprocessing
)
4115 WORD map
= reg_maps
->input_registers
;
4118 for (i
= 0; map
; map
>>= 1, ++i
)
4120 const char *semantic_name
;
4125 if (!(map
& 1)) continue;
4127 semantic_name
= input_signature
[i
].semantic_name
;
4128 semantic_idx
= input_signature
[i
].semantic_idx
;
4129 shader_glsl_write_mask_to_str(input_signature
[i
].mask
, reg_mask
);
4131 if (shader_match_semantic(semantic_name
, WINED3D_DECL_USAGE_TEXCOORD
))
4133 if (semantic_idx
< 8 && vertexprocessing
== pretransformed
)
4134 shader_addline(buffer
, "ps_in[%u]%s = gl_TexCoord[%u]%s;\n",
4135 shader
->u
.ps
.input_reg_map
[i
], reg_mask
, semantic_idx
, 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
);
4140 else if (shader_match_semantic(semantic_name
, WINED3D_DECL_USAGE_COLOR
))
4143 shader_addline(buffer
, "ps_in[%u]%s = vec4(gl_Color)%s;\n",
4144 shader
->u
.ps
.input_reg_map
[i
], reg_mask
, reg_mask
);
4145 else if (semantic_idx
== 1)
4146 shader_addline(buffer
, "ps_in[%u]%s = vec4(gl_SecondaryColor)%s;\n",
4147 shader
->u
.ps
.input_reg_map
[i
], reg_mask
, reg_mask
);
4149 shader_addline(buffer
, "ps_in[%u]%s = vec4(0.0, 0.0, 0.0, 0.0)%s;\n",
4150 shader
->u
.ps
.input_reg_map
[i
], reg_mask
, reg_mask
);
4154 shader_addline(buffer
, "ps_in[%u]%s = vec4(0.0, 0.0, 0.0, 0.0)%s;\n",
4155 shader
->u
.ps
.input_reg_map
[i
], reg_mask
, reg_mask
);
4160 /*********************************************
4161 * Vertex Shader Specific Code begins here
4162 ********************************************/
4164 static void add_glsl_program_entry(struct shader_glsl_priv
*priv
, struct glsl_shader_prog_link
*entry
)
4166 struct glsl_program_key key
;
4168 key
.vs_id
= entry
->vs
.id
;
4169 key
.gs_id
= entry
->gs
.id
;
4170 key
.ps_id
= entry
->ps
.id
;
4172 if (wine_rb_put(&priv
->program_lookup
, &key
, &entry
->program_lookup_entry
) == -1)
4174 ERR("Failed to insert program entry.\n");
4178 static struct glsl_shader_prog_link
*get_glsl_program_entry(const struct shader_glsl_priv
*priv
,
4179 GLhandleARB vs_id
, GLhandleARB gs_id
, GLhandleARB ps_id
)
4181 struct wine_rb_entry
*entry
;
4182 struct glsl_program_key key
;
4188 entry
= wine_rb_get(&priv
->program_lookup
, &key
);
4189 return entry
? WINE_RB_ENTRY_VALUE(entry
, struct glsl_shader_prog_link
, program_lookup_entry
) : NULL
;
4192 /* Context activation is done by the caller. */
4193 static void delete_glsl_program_entry(struct shader_glsl_priv
*priv
, const struct wined3d_gl_info
*gl_info
,
4194 struct glsl_shader_prog_link
*entry
)
4196 struct glsl_program_key key
;
4198 key
.vs_id
= entry
->vs
.id
;
4199 key
.gs_id
= entry
->gs
.id
;
4200 key
.ps_id
= entry
->ps
.id
;
4201 wine_rb_remove(&priv
->program_lookup
, &key
);
4203 GL_EXTCALL(glDeleteObjectARB(entry
->programId
));
4205 list_remove(&entry
->vs
.shader_entry
);
4207 list_remove(&entry
->gs
.shader_entry
);
4209 list_remove(&entry
->ps
.shader_entry
);
4210 HeapFree(GetProcessHeap(), 0, entry
->vs
.uniform_f_locations
);
4211 HeapFree(GetProcessHeap(), 0, entry
->ps
.uniform_f_locations
);
4212 HeapFree(GetProcessHeap(), 0, entry
);
4215 static void handle_ps3_input(struct wined3d_shader_buffer
*buffer
,
4216 const struct wined3d_gl_info
*gl_info
, const DWORD
*map
,
4217 const struct wined3d_shader_signature_element
*input_signature
,
4218 const struct wined3d_shader_reg_maps
*reg_maps_in
,
4219 const struct wined3d_shader_signature_element
*output_signature
,
4220 const struct wined3d_shader_reg_maps
*reg_maps_out
)
4223 const char *semantic_name_in
;
4224 UINT semantic_idx_in
;
4227 unsigned int in_count
= vec4_varyings(3, gl_info
);
4229 char destination
[50];
4230 WORD input_map
, output_map
;
4232 set
= HeapAlloc(GetProcessHeap(), HEAP_ZERO_MEMORY
, sizeof(*set
) * (in_count
+ 2));
4234 input_map
= reg_maps_in
->input_registers
;
4235 for (i
= 0; input_map
; input_map
>>= 1, ++i
)
4237 if (!(input_map
& 1)) continue;
4240 /* Declared, but not read register */
4241 if (in_idx
== ~0U) continue;
4242 if (in_idx
>= (in_count
+ 2))
4244 FIXME("More input varyings declared than supported, expect issues.\n");
4248 if (in_idx
== in_count
)
4249 sprintf(destination
, "gl_FrontColor");
4250 else if (in_idx
== in_count
+ 1)
4251 sprintf(destination
, "gl_FrontSecondaryColor");
4253 sprintf(destination
, "ps_in[%u]", in_idx
);
4255 semantic_name_in
= input_signature
[i
].semantic_name
;
4256 semantic_idx_in
= input_signature
[i
].semantic_idx
;
4259 output_map
= reg_maps_out
->output_registers
;
4260 for (j
= 0; output_map
; output_map
>>= 1, ++j
)
4264 if (!(output_map
& 1)
4265 || semantic_idx_in
!= output_signature
[j
].semantic_idx
4266 || strcmp(semantic_name_in
, output_signature
[j
].semantic_name
)
4267 || !(mask
= input_signature
[i
].mask
& output_signature
[j
].mask
))
4271 shader_glsl_write_mask_to_str(mask
, reg_mask
);
4273 shader_addline(buffer
, "%s%s = vs_out[%u]%s;\n",
4274 destination
, reg_mask
, j
, reg_mask
);
4278 for (i
= 0; i
< in_count
+ 2; ++i
)
4282 if (!set
[i
] || set
[i
] == WINED3DSP_WRITEMASK_ALL
)
4285 if (set
[i
] == ~0U) set
[i
] = 0;
4288 if (!(set
[i
] & WINED3DSP_WRITEMASK_0
)) reg_mask
[size
++] = 'x';
4289 if (!(set
[i
] & WINED3DSP_WRITEMASK_1
)) reg_mask
[size
++] = 'y';
4290 if (!(set
[i
] & WINED3DSP_WRITEMASK_2
)) reg_mask
[size
++] = 'z';
4291 if (!(set
[i
] & WINED3DSP_WRITEMASK_3
)) reg_mask
[size
++] = 'w';
4292 reg_mask
[size
] = '\0';
4295 sprintf(destination
, "gl_FrontColor");
4296 else if (i
== in_count
+ 1)
4297 sprintf(destination
, "gl_FrontSecondaryColor");
4299 sprintf(destination
, "ps_in[%u]", i
);
4301 if (size
== 1) shader_addline(buffer
, "%s.%s = 0.0;\n", destination
, reg_mask
);
4302 else shader_addline(buffer
, "%s.%s = vec%u(0.0);\n", destination
, reg_mask
, size
);
4305 HeapFree(GetProcessHeap(), 0, set
);
4308 /* Context activation is done by the caller. */
4309 static GLhandleARB
generate_param_reorder_function(struct wined3d_shader_buffer
*buffer
,
4310 const struct wined3d_shader
*vs
, const struct wined3d_shader
*ps
,
4311 const struct wined3d_gl_info
*gl_info
)
4313 GLhandleARB ret
= 0;
4314 DWORD ps_major
= ps
? ps
->reg_maps
.shader_version
.major
: 0;
4316 const char *semantic_name
;
4319 const struct wined3d_shader_signature_element
*output_signature
= vs
->output_signature
;
4320 WORD map
= vs
->reg_maps
.output_registers
;
4322 shader_buffer_clear(buffer
);
4324 shader_addline(buffer
, "#version 120\n");
4328 shader_addline(buffer
, "void order_ps_input(in vec4 vs_out[%u])\n{\n", vs
->limits
.packed_output
);
4330 for (i
= 0; map
; map
>>= 1, ++i
)
4334 if (!(map
& 1)) continue;
4336 semantic_name
= output_signature
[i
].semantic_name
;
4337 semantic_idx
= output_signature
[i
].semantic_idx
;
4338 write_mask
= output_signature
[i
].mask
;
4339 shader_glsl_write_mask_to_str(write_mask
, reg_mask
);
4341 if (shader_match_semantic(semantic_name
, WINED3D_DECL_USAGE_COLOR
))
4344 shader_addline(buffer
, "gl_FrontColor%s = vs_out[%u]%s;\n",
4345 reg_mask
, i
, reg_mask
);
4346 else if (semantic_idx
== 1)
4347 shader_addline(buffer
, "gl_FrontSecondaryColor%s = vs_out[%u]%s;\n",
4348 reg_mask
, i
, reg_mask
);
4350 else if (shader_match_semantic(semantic_name
, WINED3D_DECL_USAGE_POSITION
))
4352 shader_addline(buffer
, "gl_Position%s = vs_out[%u]%s;\n",
4353 reg_mask
, i
, reg_mask
);
4355 else if (shader_match_semantic(semantic_name
, WINED3D_DECL_USAGE_TEXCOORD
))
4357 if (semantic_idx
< 8)
4359 if (!(gl_info
->quirks
& WINED3D_QUIRK_SET_TEXCOORD_W
) || ps_major
> 0)
4360 write_mask
|= WINED3DSP_WRITEMASK_3
;
4362 shader_addline(buffer
, "gl_TexCoord[%u]%s = vs_out[%u]%s;\n",
4363 semantic_idx
, reg_mask
, i
, reg_mask
);
4364 if (!(write_mask
& WINED3DSP_WRITEMASK_3
))
4365 shader_addline(buffer
, "gl_TexCoord[%u].w = 1.0;\n", semantic_idx
);
4368 else if (shader_match_semantic(semantic_name
, WINED3D_DECL_USAGE_PSIZE
))
4370 shader_addline(buffer
, "gl_PointSize = vs_out[%u].%c;\n", i
, reg_mask
[1]);
4372 else if (shader_match_semantic(semantic_name
, WINED3D_DECL_USAGE_FOG
))
4374 shader_addline(buffer
, "gl_FogFragCoord = clamp(vs_out[%u].%c, 0.0, 1.0);\n", i
, reg_mask
[1]);
4377 shader_addline(buffer
, "}\n");
4381 UINT in_count
= min(vec4_varyings(ps_major
, gl_info
), ps
->limits
.packed_input
);
4382 /* This one is tricky: a 3.0 pixel shader reads from a 3.0 vertex shader */
4383 shader_addline(buffer
, "varying vec4 ps_in[%u];\n", in_count
);
4384 shader_addline(buffer
, "void order_ps_input(in vec4 vs_out[%u])\n{\n", vs
->limits
.packed_output
);
4386 /* First, sort out position and point size. Those are not passed to the pixel shader */
4387 for (i
= 0; map
; map
>>= 1, ++i
)
4389 if (!(map
& 1)) continue;
4391 semantic_name
= output_signature
[i
].semantic_name
;
4392 shader_glsl_write_mask_to_str(output_signature
[i
].mask
, reg_mask
);
4394 if (shader_match_semantic(semantic_name
, WINED3D_DECL_USAGE_POSITION
))
4396 shader_addline(buffer
, "gl_Position%s = vs_out[%u]%s;\n",
4397 reg_mask
, i
, reg_mask
);
4399 else if (shader_match_semantic(semantic_name
, WINED3D_DECL_USAGE_PSIZE
))
4401 shader_addline(buffer
, "gl_PointSize = vs_out[%u].%c;\n", i
, reg_mask
[1]);
4405 /* Then, fix the pixel shader input */
4406 handle_ps3_input(buffer
, gl_info
, ps
->u
.ps
.input_reg_map
, ps
->input_signature
,
4407 &ps
->reg_maps
, output_signature
, &vs
->reg_maps
);
4409 shader_addline(buffer
, "}\n");
4412 ret
= GL_EXTCALL(glCreateShaderObjectARB(GL_VERTEX_SHADER_ARB
));
4413 checkGLcall("glCreateShaderObjectARB(GL_VERTEX_SHADER_ARB)");
4414 shader_glsl_compile(gl_info
, ret
, buffer
->buffer
);
4419 static void shader_glsl_generate_srgb_write_correction(struct wined3d_shader_buffer
*buffer
)
4421 shader_addline(buffer
, "tmp0.xyz = pow(gl_FragData[0].xyz, vec3(srgb_const0.x));\n");
4422 shader_addline(buffer
, "tmp0.xyz = tmp0.xyz * vec3(srgb_const0.y) - vec3(srgb_const0.z);\n");
4423 shader_addline(buffer
, "tmp1.xyz = gl_FragData[0].xyz * vec3(srgb_const0.w);\n");
4424 shader_addline(buffer
, "bvec3 srgb_compare = lessThan(gl_FragData[0].xyz, vec3(srgb_const1.x));\n");
4425 shader_addline(buffer
, "gl_FragData[0].xyz = mix(tmp0.xyz, tmp1.xyz, vec3(srgb_compare));\n");
4426 shader_addline(buffer
, "gl_FragData[0] = clamp(gl_FragData[0], 0.0, 1.0);\n");
4429 static void shader_glsl_generate_fog_code(struct wined3d_shader_buffer
*buffer
, enum fogmode mode
)
4437 /* Fog = (gl_Fog.end - gl_FogFragCoord) / (gl_Fog.end - gl_Fog.start) */
4438 shader_addline(buffer
, "float Fog = (gl_Fog.end - gl_FogFragCoord) / (gl_Fog.end - gl_Fog.start);\n");
4442 /* Fog = e^-(gl_Fog.density * gl_FogFragCoord) */
4443 shader_addline(buffer
, "float Fog = exp(-gl_Fog.density * gl_FogFragCoord);\n");
4447 /* Fog = e^-((gl_Fog.density * gl_FogFragCoord)^2) */
4448 shader_addline(buffer
, "float Fog = exp(-gl_Fog.density * gl_Fog.density * gl_FogFragCoord * gl_FogFragCoord);\n");
4452 ERR("Invalid fog mode %#x.\n", mode
);
4456 shader_addline(buffer
, "gl_FragData[0].xyz = mix(gl_Fog.color.xyz, gl_FragData[0].xyz, clamp(Fog, 0.0, 1.0));\n");
4459 /* Context activation is done by the caller. */
4460 static void hardcode_local_constants(const struct wined3d_shader
*shader
,
4461 const struct wined3d_gl_info
*gl_info
, GLhandleARB programId
, const char *prefix
)
4463 const struct wined3d_shader_lconst
*lconst
;
4468 LIST_FOR_EACH_ENTRY(lconst
, &shader
->constantsF
, struct wined3d_shader_lconst
, entry
)
4470 value
= (const float *)lconst
->value
;
4471 snprintf(glsl_name
, sizeof(glsl_name
), "%s_lc%u", prefix
, lconst
->idx
);
4472 tmp_loc
= GL_EXTCALL(glGetUniformLocationARB(programId
, glsl_name
));
4473 GL_EXTCALL(glUniform4fvARB(tmp_loc
, 1, value
));
4475 checkGLcall("Hardcoding local constants");
4478 /* Context activation is done by the caller. */
4479 static GLuint
shader_glsl_generate_pshader(const struct wined3d_context
*context
,
4480 struct wined3d_shader_buffer
*buffer
, const struct wined3d_shader
*shader
,
4481 const struct ps_compile_args
*args
, struct ps_np2fixup_info
*np2fixup_info
)
4483 const struct wined3d_shader_reg_maps
*reg_maps
= &shader
->reg_maps
;
4484 const struct wined3d_gl_info
*gl_info
= context
->gl_info
;
4485 const DWORD
*function
= shader
->function
;
4486 struct shader_glsl_ctx_priv priv_ctx
;
4488 /* Create the hw GLSL shader object and assign it as the shader->prgId */
4489 GLhandleARB shader_obj
= GL_EXTCALL(glCreateShaderObjectARB(GL_FRAGMENT_SHADER_ARB
));
4491 memset(&priv_ctx
, 0, sizeof(priv_ctx
));
4492 priv_ctx
.cur_ps_args
= args
;
4493 priv_ctx
.cur_np2fixup_info
= np2fixup_info
;
4495 shader_addline(buffer
, "#version 120\n");
4497 if (gl_info
->supported
[ARB_SHADER_BIT_ENCODING
])
4498 shader_addline(buffer
, "#extension GL_ARB_shader_bit_encoding : enable\n");
4499 if (gl_info
->supported
[ARB_SHADER_TEXTURE_LOD
])
4500 shader_addline(buffer
, "#extension GL_ARB_shader_texture_lod : enable\n");
4501 /* The spec says that it doesn't have to be explicitly enabled, but the
4502 * nvidia drivers write a warning if we don't do so. */
4503 if (gl_info
->supported
[ARB_TEXTURE_RECTANGLE
])
4504 shader_addline(buffer
, "#extension GL_ARB_texture_rectangle : enable\n");
4505 if (gl_info
->supported
[EXT_GPU_SHADER4
])
4506 shader_addline(buffer
, "#extension GL_EXT_gpu_shader4 : enable\n");
4508 /* Base Declarations */
4509 shader_generate_glsl_declarations(context
, buffer
, shader
, reg_maps
, &priv_ctx
);
4511 /* Pack 3.0 inputs */
4512 if (reg_maps
->shader_version
.major
>= 3 && args
->vp_mode
!= vertexshader
)
4513 shader_glsl_input_pack(shader
, buffer
, shader
->input_signature
, reg_maps
, args
->vp_mode
);
4515 /* Base Shader Body */
4516 shader_generate_main(shader
, buffer
, reg_maps
, function
, &priv_ctx
);
4518 /* Pixel shaders < 2.0 place the resulting color in R0 implicitly */
4519 if (reg_maps
->shader_version
.major
< 2)
4521 /* Some older cards like GeforceFX ones don't support multiple buffers, so also not gl_FragData */
4522 shader_addline(buffer
, "gl_FragData[0] = R0;\n");
4525 if (args
->srgb_correction
)
4526 shader_glsl_generate_srgb_write_correction(buffer
);
4528 /* SM < 3 does not replace the fog stage. */
4529 if (reg_maps
->shader_version
.major
< 3)
4530 shader_glsl_generate_fog_code(buffer
, args
->fog
);
4532 shader_addline(buffer
, "}\n");
4534 TRACE("Compiling shader object %u\n", shader_obj
);
4535 shader_glsl_compile(gl_info
, shader_obj
, buffer
->buffer
);
4537 /* Store the shader object */
4541 /* Context activation is done by the caller. */
4542 static GLuint
shader_glsl_generate_vshader(const struct wined3d_context
*context
,
4543 struct wined3d_shader_buffer
*buffer
, const struct wined3d_shader
*shader
,
4544 const struct vs_compile_args
*args
)
4546 const struct wined3d_shader_reg_maps
*reg_maps
= &shader
->reg_maps
;
4547 const struct wined3d_gl_info
*gl_info
= context
->gl_info
;
4548 const DWORD
*function
= shader
->function
;
4549 struct shader_glsl_ctx_priv priv_ctx
;
4551 /* Create the hw GLSL shader program and assign it as the shader->prgId */
4552 GLhandleARB shader_obj
= GL_EXTCALL(glCreateShaderObjectARB(GL_VERTEX_SHADER_ARB
));
4554 shader_addline(buffer
, "#version 120\n");
4556 if (gl_info
->supported
[ARB_SHADER_BIT_ENCODING
])
4557 shader_addline(buffer
, "#extension GL_ARB_shader_bit_encoding : enable\n");
4558 if (gl_info
->supported
[EXT_GPU_SHADER4
])
4559 shader_addline(buffer
, "#extension GL_EXT_gpu_shader4 : enable\n");
4561 memset(&priv_ctx
, 0, sizeof(priv_ctx
));
4562 priv_ctx
.cur_vs_args
= args
;
4564 /* Base Declarations */
4565 shader_generate_glsl_declarations(context
, buffer
, shader
, reg_maps
, &priv_ctx
);
4567 /* Base Shader Body */
4568 shader_generate_main(shader
, buffer
, reg_maps
, function
, &priv_ctx
);
4570 /* Unpack outputs */
4571 shader_addline(buffer
, "order_ps_input(vs_out);\n");
4573 /* The D3DRS_FOGTABLEMODE render state defines if the shader-generated fog coord is used
4574 * or if the fragment depth is used. If the fragment depth is used(FOGTABLEMODE != NONE),
4575 * the fog frag coord is thrown away. If the fog frag coord is used, but not written by
4576 * the shader, it is set to 0.0(fully fogged, since start = 1.0, end = 0.0)
4578 if (args
->fog_src
== VS_FOG_Z
)
4579 shader_addline(buffer
, "gl_FogFragCoord = gl_Position.z;\n");
4580 else if (!reg_maps
->fog
)
4581 shader_addline(buffer
, "gl_FogFragCoord = 0.0;\n");
4583 /* We always store the clipplanes without y inversion */
4584 if (args
->clip_enabled
)
4585 shader_addline(buffer
, "gl_ClipVertex = gl_Position;\n");
4587 /* Write the final position.
4589 * OpenGL coordinates specify the center of the pixel while d3d coords specify
4590 * the corner. The offsets are stored in z and w in posFixup. posFixup.y contains
4591 * 1.0 or -1.0 to turn the rendering upside down for offscreen rendering. PosFixup.x
4592 * contains 1.0 to allow a mad.
4594 shader_addline(buffer
, "gl_Position.y = gl_Position.y * posFixup.y;\n");
4595 shader_addline(buffer
, "gl_Position.xy += posFixup.zw * gl_Position.ww;\n");
4597 /* Z coord [0;1]->[-1;1] mapping, see comment in transform_projection in state.c
4599 * Basically we want (in homogeneous coordinates) z = z * 2 - 1. However, shaders are run
4600 * before the homogeneous divide, so we have to take the w into account: z = ((z / w) * 2 - 1) * w,
4601 * which is the same as z = z * 2 - w.
4603 shader_addline(buffer
, "gl_Position.z = gl_Position.z * 2.0 - gl_Position.w;\n");
4605 shader_addline(buffer
, "}\n");
4607 TRACE("Compiling shader object %u\n", shader_obj
);
4608 shader_glsl_compile(gl_info
, shader_obj
, buffer
->buffer
);
4613 /* Context activation is done by the caller. */
4614 static GLhandleARB
shader_glsl_generate_geometry_shader(const struct wined3d_context
*context
,
4615 struct wined3d_shader_buffer
*buffer
, const struct wined3d_shader
*shader
)
4617 const struct wined3d_shader_reg_maps
*reg_maps
= &shader
->reg_maps
;
4618 const struct wined3d_gl_info
*gl_info
= context
->gl_info
;
4619 const DWORD
*function
= shader
->function
;
4620 struct shader_glsl_ctx_priv priv_ctx
;
4621 GLhandleARB shader_id
;
4623 shader_id
= GL_EXTCALL(glCreateShaderObjectARB(GL_GEOMETRY_SHADER_ARB
));
4625 shader_addline(buffer
, "#version 120\n");
4627 if (gl_info
->supported
[ARB_GEOMETRY_SHADER4
])
4628 shader_addline(buffer
, "#extension GL_ARB_geometry_shader4 : enable\n");
4629 if (gl_info
->supported
[ARB_SHADER_BIT_ENCODING
])
4630 shader_addline(buffer
, "#extension GL_ARB_shader_bit_encoding : enable\n");
4631 if (gl_info
->supported
[EXT_GPU_SHADER4
])
4632 shader_addline(buffer
, "#extension GL_EXT_gpu_shader4 : enable\n");
4634 memset(&priv_ctx
, 0, sizeof(priv_ctx
));
4635 shader_generate_glsl_declarations(context
, buffer
, shader
, reg_maps
, &priv_ctx
);
4636 shader_generate_main(shader
, buffer
, reg_maps
, function
, &priv_ctx
);
4637 shader_addline(buffer
, "}\n");
4639 TRACE("Compiling shader object %u.\n", shader_id
);
4640 shader_glsl_compile(gl_info
, shader_id
, buffer
->buffer
);
4645 static GLhandleARB
find_glsl_pshader(const struct wined3d_context
*context
,
4646 struct wined3d_shader_buffer
*buffer
, struct wined3d_shader
*shader
,
4647 const struct ps_compile_args
*args
, const struct ps_np2fixup_info
**np2fixup_info
)
4649 struct wined3d_state
*state
= &shader
->device
->stateBlock
->state
;
4650 struct glsl_ps_compiled_shader
*gl_shaders
, *new_array
;
4651 struct glsl_shader_private
*shader_data
;
4652 struct ps_np2fixup_info
*np2fixup
;
4657 if (!shader
->backend_data
)
4659 shader
->backend_data
= HeapAlloc(GetProcessHeap(), HEAP_ZERO_MEMORY
, sizeof(*shader_data
));
4660 if (!shader
->backend_data
)
4662 ERR("Failed to allocate backend data.\n");
4666 shader_data
= shader
->backend_data
;
4667 gl_shaders
= shader_data
->gl_shaders
.ps
;
4669 /* Usually we have very few GL shaders for each d3d shader(just 1 or maybe 2),
4670 * so a linear search is more performant than a hashmap or a binary search
4671 * (cache coherency etc)
4673 for (i
= 0; i
< shader_data
->num_gl_shaders
; ++i
)
4675 if (!memcmp(&gl_shaders
[i
].args
, args
, sizeof(*args
)))
4677 if (args
->np2_fixup
)
4678 *np2fixup_info
= &gl_shaders
[i
].np2fixup
;
4679 return gl_shaders
[i
].prgId
;
4683 TRACE("No matching GL shader found for shader %p, compiling a new shader.\n", shader
);
4684 if(shader_data
->shader_array_size
== shader_data
->num_gl_shaders
) {
4685 if (shader_data
->num_gl_shaders
)
4687 new_size
= shader_data
->shader_array_size
+ max(1, shader_data
->shader_array_size
/ 2);
4688 new_array
= HeapReAlloc(GetProcessHeap(), 0, shader_data
->gl_shaders
.ps
,
4689 new_size
* sizeof(*gl_shaders
));
4693 new_array
= HeapAlloc(GetProcessHeap(), 0, sizeof(*gl_shaders
));
4698 ERR("Out of memory\n");
4701 shader_data
->gl_shaders
.ps
= new_array
;
4702 shader_data
->shader_array_size
= new_size
;
4703 gl_shaders
= new_array
;
4706 gl_shaders
[shader_data
->num_gl_shaders
].args
= *args
;
4708 np2fixup
= &gl_shaders
[shader_data
->num_gl_shaders
].np2fixup
;
4709 memset(np2fixup
, 0, sizeof(*np2fixup
));
4710 *np2fixup_info
= args
->np2_fixup
? np2fixup
: NULL
;
4712 pixelshader_update_samplers(&shader
->reg_maps
, state
->textures
);
4714 shader_buffer_clear(buffer
);
4715 ret
= shader_glsl_generate_pshader(context
, buffer
, shader
, args
, np2fixup
);
4716 gl_shaders
[shader_data
->num_gl_shaders
++].prgId
= ret
;
4721 static inline BOOL
vs_args_equal(const struct vs_compile_args
*stored
, const struct vs_compile_args
*new,
4722 const DWORD use_map
) {
4723 if((stored
->swizzle_map
& use_map
) != new->swizzle_map
) return FALSE
;
4724 if((stored
->clip_enabled
) != new->clip_enabled
) return FALSE
;
4725 return stored
->fog_src
== new->fog_src
;
4728 static GLhandleARB
find_glsl_vshader(const struct wined3d_context
*context
,
4729 struct wined3d_shader_buffer
*buffer
, struct wined3d_shader
*shader
,
4730 const struct vs_compile_args
*args
)
4734 DWORD use_map
= shader
->device
->strided_streams
.use_map
;
4735 struct glsl_vs_compiled_shader
*gl_shaders
, *new_array
;
4736 struct glsl_shader_private
*shader_data
;
4739 if (!shader
->backend_data
)
4741 shader
->backend_data
= HeapAlloc(GetProcessHeap(), HEAP_ZERO_MEMORY
, sizeof(*shader_data
));
4742 if (!shader
->backend_data
)
4744 ERR("Failed to allocate backend data.\n");
4748 shader_data
= shader
->backend_data
;
4749 gl_shaders
= shader_data
->gl_shaders
.vs
;
4751 /* Usually we have very few GL shaders for each d3d shader(just 1 or maybe 2),
4752 * so a linear search is more performant than a hashmap or a binary search
4753 * (cache coherency etc)
4755 for (i
= 0; i
< shader_data
->num_gl_shaders
; ++i
)
4757 if (vs_args_equal(&gl_shaders
[i
].args
, args
, use_map
))
4758 return gl_shaders
[i
].prgId
;
4761 TRACE("No matching GL shader found for shader %p, compiling a new shader.\n", shader
);
4763 if(shader_data
->shader_array_size
== shader_data
->num_gl_shaders
) {
4764 if (shader_data
->num_gl_shaders
)
4766 new_size
= shader_data
->shader_array_size
+ max(1, shader_data
->shader_array_size
/ 2);
4767 new_array
= HeapReAlloc(GetProcessHeap(), 0, shader_data
->gl_shaders
.vs
,
4768 new_size
* sizeof(*gl_shaders
));
4772 new_array
= HeapAlloc(GetProcessHeap(), 0, sizeof(*gl_shaders
));
4777 ERR("Out of memory\n");
4780 shader_data
->gl_shaders
.vs
= new_array
;
4781 shader_data
->shader_array_size
= new_size
;
4782 gl_shaders
= new_array
;
4785 gl_shaders
[shader_data
->num_gl_shaders
].args
= *args
;
4787 shader_buffer_clear(buffer
);
4788 ret
= shader_glsl_generate_vshader(context
, buffer
, shader
, args
);
4789 gl_shaders
[shader_data
->num_gl_shaders
++].prgId
= ret
;
4794 static GLhandleARB
find_glsl_geometry_shader(const struct wined3d_context
*context
,
4795 struct wined3d_shader_buffer
*buffer
, struct wined3d_shader
*shader
)
4797 struct glsl_gs_compiled_shader
*gl_shaders
;
4798 struct glsl_shader_private
*shader_data
;
4801 if (!shader
->backend_data
)
4803 if (!(shader
->backend_data
= HeapAlloc(GetProcessHeap(), HEAP_ZERO_MEMORY
, sizeof(*shader_data
))))
4805 ERR("Failed to allocate backend data.\n");
4809 shader_data
= shader
->backend_data
;
4810 gl_shaders
= shader_data
->gl_shaders
.gs
;
4812 if (shader_data
->num_gl_shaders
)
4813 return gl_shaders
[0].id
;
4815 TRACE("No matching GL shader found for shader %p, compiling a new shader.\n", shader
);
4817 if (!(shader_data
->gl_shaders
.gs
= HeapAlloc(GetProcessHeap(), 0, sizeof(*gl_shaders
))))
4819 ERR("Failed to allocate GL shader array.\n");
4822 shader_data
->shader_array_size
= 1;
4823 gl_shaders
= shader_data
->gl_shaders
.gs
;
4825 shader_buffer_clear(buffer
);
4826 ret
= shader_glsl_generate_geometry_shader(context
, buffer
, shader
);
4827 gl_shaders
[shader_data
->num_gl_shaders
++].id
= ret
;
4832 static const char *shader_glsl_get_ffp_fragment_op_arg(struct wined3d_shader_buffer
*buffer
,
4833 DWORD argnum
, unsigned int stage
, DWORD arg
)
4837 if (arg
== ARG_UNUSED
)
4838 return "<unused arg>";
4840 switch (arg
& WINED3DTA_SELECTMASK
)
4842 case WINED3DTA_DIFFUSE
:
4846 case WINED3DTA_CURRENT
:
4853 case WINED3DTA_TEXTURE
:
4856 case 0: ret
= "tex0"; break;
4857 case 1: ret
= "tex1"; break;
4858 case 2: ret
= "tex2"; break;
4859 case 3: ret
= "tex3"; break;
4860 case 4: ret
= "tex4"; break;
4861 case 5: ret
= "tex5"; break;
4862 case 6: ret
= "tex6"; break;
4863 case 7: ret
= "tex7"; break;
4865 ret
= "<invalid texture>";
4870 case WINED3DTA_TFACTOR
:
4874 case WINED3DTA_SPECULAR
:
4875 ret
= "gl_SecondaryColor";
4878 case WINED3DTA_TEMP
:
4882 case WINED3DTA_CONSTANT
:
4883 FIXME("Per-stage constants not implemented.\n");
4886 case 0: ret
= "const0"; break;
4887 case 1: ret
= "const1"; break;
4888 case 2: ret
= "const2"; break;
4889 case 3: ret
= "const3"; break;
4890 case 4: ret
= "const4"; break;
4891 case 5: ret
= "const5"; break;
4892 case 6: ret
= "const6"; break;
4893 case 7: ret
= "const7"; break;
4895 ret
= "<invalid constant>";
4901 return "<unhandled arg>";
4904 if (arg
& WINED3DTA_COMPLEMENT
)
4906 shader_addline(buffer
, "arg%u = vec4(1.0) - %s;\n", argnum
, ret
);
4909 else if (argnum
== 1)
4911 else if (argnum
== 2)
4915 if (arg
& WINED3DTA_ALPHAREPLICATE
)
4917 shader_addline(buffer
, "arg%u = vec4(%s.w);\n", argnum
, ret
);
4920 else if (argnum
== 1)
4922 else if (argnum
== 2)
4929 static void shader_glsl_ffp_fragment_op(struct wined3d_shader_buffer
*buffer
, unsigned int stage
, BOOL color
,
4930 BOOL alpha
, DWORD dst
, DWORD op
, DWORD dw_arg0
, DWORD dw_arg1
, DWORD dw_arg2
)
4932 const char *dstmask
, *dstreg
, *arg0
, *arg1
, *arg2
;
4942 dstreg
= "temp_reg";
4946 arg0
= shader_glsl_get_ffp_fragment_op_arg(buffer
, 0, stage
, dw_arg0
);
4947 arg1
= shader_glsl_get_ffp_fragment_op_arg(buffer
, 1, stage
, dw_arg1
);
4948 arg2
= shader_glsl_get_ffp_fragment_op_arg(buffer
, 2, stage
, dw_arg2
);
4952 case WINED3D_TOP_DISABLE
:
4954 shader_addline(buffer
, "%s%s = gl_Color%s;\n", dstreg
, dstmask
, dstmask
);
4957 case WINED3D_TOP_SELECT_ARG1
:
4958 shader_addline(buffer
, "%s%s = %s%s;\n", dstreg
, dstmask
, arg1
, dstmask
);
4961 case WINED3D_TOP_SELECT_ARG2
:
4962 shader_addline(buffer
, "%s%s = %s%s;\n", dstreg
, dstmask
, arg2
, dstmask
);
4965 case WINED3D_TOP_MODULATE
:
4966 shader_addline(buffer
, "%s%s = %s%s * %s%s;\n", dstreg
, dstmask
, arg1
, dstmask
, arg2
, dstmask
);
4969 case WINED3D_TOP_MODULATE_4X
:
4970 shader_addline(buffer
, "%s%s = clamp(%s%s * %s%s * 4.0, 0.0, 1.0);\n",
4971 dstreg
, dstmask
, arg1
, dstmask
, arg2
, dstmask
);
4974 case WINED3D_TOP_MODULATE_2X
:
4975 shader_addline(buffer
, "%s%s = clamp(%s%s * %s%s * 2.0, 0.0, 1.0);\n",
4976 dstreg
, dstmask
, arg1
, dstmask
, arg2
, dstmask
);
4979 case WINED3D_TOP_ADD
:
4980 shader_addline(buffer
, "%s%s = clamp(%s%s + %s%s, 0.0, 1.0);\n",
4981 dstreg
, dstmask
, arg1
, dstmask
, arg2
, dstmask
);
4984 case WINED3D_TOP_ADD_SIGNED
:
4985 shader_addline(buffer
, "%s%s = clamp(%s%s + (%s - vec4(0.5))%s, 0.0, 1.0);\n",
4986 dstreg
, dstmask
, arg1
, dstmask
, arg2
, dstmask
);
4989 case WINED3D_TOP_ADD_SIGNED_2X
:
4990 shader_addline(buffer
, "%s%s = clamp((%s%s + (%s - vec4(0.5))%s) * 2.0, 0.0, 1.0);\n",
4991 dstreg
, dstmask
, arg1
, dstmask
, arg2
, dstmask
);
4994 case WINED3D_TOP_SUBTRACT
:
4995 shader_addline(buffer
, "%s%s = clamp(%s%s - %s%s, 0.0, 1.0);\n",
4996 dstreg
, dstmask
, arg1
, dstmask
, arg2
, dstmask
);
4999 case WINED3D_TOP_ADD_SMOOTH
:
5000 shader_addline(buffer
, "%s%s = clamp((vec4(1.0) - %s)%s * %s%s + %s%s, 0.0, 1.0);\n",
5001 dstreg
, dstmask
, arg1
, dstmask
, arg2
, dstmask
, arg1
, dstmask
);
5004 case WINED3D_TOP_BLEND_DIFFUSE_ALPHA
:
5005 arg0
= shader_glsl_get_ffp_fragment_op_arg(buffer
, 0, stage
, WINED3DTA_DIFFUSE
);
5006 shader_addline(buffer
, "%s%s = mix(%s%s, %s%s, %s.w);\n",
5007 dstreg
, dstmask
, arg2
, dstmask
, arg1
, dstmask
, arg0
);
5010 case WINED3D_TOP_BLEND_TEXTURE_ALPHA
:
5011 arg0
= shader_glsl_get_ffp_fragment_op_arg(buffer
, 0, stage
, WINED3DTA_TEXTURE
);
5012 shader_addline(buffer
, "%s%s = mix(%s%s, %s%s, %s.w);\n",
5013 dstreg
, dstmask
, arg2
, dstmask
, arg1
, dstmask
, arg0
);
5016 case WINED3D_TOP_BLEND_FACTOR_ALPHA
:
5017 arg0
= shader_glsl_get_ffp_fragment_op_arg(buffer
, 0, stage
, WINED3DTA_TFACTOR
);
5018 shader_addline(buffer
, "%s%s = mix(%s%s, %s%s, %s.w);\n",
5019 dstreg
, dstmask
, arg2
, dstmask
, arg1
, dstmask
, arg0
);
5022 case WINED3D_TOP_BLEND_TEXTURE_ALPHA_PM
:
5023 arg0
= shader_glsl_get_ffp_fragment_op_arg(buffer
, 0, stage
, WINED3DTA_TEXTURE
);
5024 shader_addline(buffer
, "%s%s = clamp(%s%s * (1.0 - %s.w) + %s%s, 0.0, 1.0);\n",
5025 dstreg
, dstmask
, arg2
, dstmask
, arg0
, arg1
, dstmask
);
5028 case WINED3D_TOP_BLEND_CURRENT_ALPHA
:
5029 arg0
= shader_glsl_get_ffp_fragment_op_arg(buffer
, 0, stage
, WINED3DTA_CURRENT
);
5030 shader_addline(buffer
, "%s%s = mix(%s%s, %s%s, %s.w);\n",
5031 dstreg
, dstmask
, arg2
, dstmask
, arg1
, dstmask
, arg0
);
5034 case WINED3D_TOP_MODULATE_ALPHA_ADD_COLOR
:
5035 shader_addline(buffer
, "%s%s = clamp(%s%s * %s.w + %s%s, 0.0, 1.0);\n",
5036 dstreg
, dstmask
, arg2
, dstmask
, arg1
, arg1
, dstmask
);
5039 case WINED3D_TOP_MODULATE_COLOR_ADD_ALPHA
:
5040 shader_addline(buffer
, "%s%s = clamp(%s%s * %s%s + %s.w, 0.0, 1.0);\n",
5041 dstreg
, dstmask
, arg1
, dstmask
, arg2
, dstmask
, arg1
);
5044 case WINED3D_TOP_MODULATE_INVALPHA_ADD_COLOR
:
5045 shader_addline(buffer
, "%s%s = clamp(%s%s * (1.0 - %s.w) + %s%s, 0.0, 1.0);\n",
5046 dstreg
, dstmask
, arg2
, dstmask
, arg1
, arg1
, dstmask
);
5048 case WINED3D_TOP_MODULATE_INVCOLOR_ADD_ALPHA
:
5049 shader_addline(buffer
, "%s%s = clamp((vec4(1.0) - %s)%s * %s%s + %s.w, 0.0, 1.0);\n",
5050 dstreg
, dstmask
, arg1
, dstmask
, arg2
, dstmask
, arg1
);
5053 case WINED3D_TOP_BUMPENVMAP
:
5054 case WINED3D_TOP_BUMPENVMAP_LUMINANCE
:
5055 /* These are handled in the first pass, nothing to do. */
5058 case WINED3D_TOP_DOTPRODUCT3
:
5059 shader_addline(buffer
, "%s%s = vec4(clamp(dot(%s.xyz - 0.5, %s.xyz - 0.5) * 4.0, 0.0, 1.0))%s;\n",
5060 dstreg
, dstmask
, arg1
, arg2
, dstmask
);
5063 case WINED3D_TOP_MULTIPLY_ADD
:
5064 shader_addline(buffer
, "%s%s = clamp(%s%s * %s%s + %s%s, 0.0, 1.0);\n",
5065 dstreg
, dstmask
, arg1
, dstmask
, arg2
, dstmask
, arg0
, dstmask
);
5068 case WINED3D_TOP_LERP
:
5069 /* MSDN isn't quite right here. */
5070 shader_addline(buffer
, "%s%s = mix(%s%s, %s%s, %s%s);\n",
5071 dstreg
, dstmask
, arg2
, dstmask
, arg1
, dstmask
, arg0
, dstmask
);
5075 FIXME("Unhandled operation %#x.\n", op
);
5080 /* Context activation is done by the caller. */
5081 static GLuint
shader_glsl_generate_ffp_fragment_shader(struct wined3d_shader_buffer
*buffer
,
5082 const struct ffp_frag_settings
*settings
, const struct wined3d_gl_info
*gl_info
)
5084 BOOL tempreg_used
= FALSE
, tfactor_used
= FALSE
;
5085 BYTE lum_map
= 0, bump_map
= 0, tex_map
= 0;
5086 const char *final_combiner_src
= "ret";
5087 UINT lowest_disabled_stage
;
5088 GLhandleARB shader_obj
;
5089 DWORD arg0
, arg1
, arg2
;
5092 shader_buffer_clear(buffer
);
5094 /* Find out which textures are read */
5095 for (stage
= 0; stage
< MAX_TEXTURES
; ++stage
)
5097 if (settings
->op
[stage
].cop
== WINED3D_TOP_DISABLE
)
5100 arg0
= settings
->op
[stage
].carg0
& WINED3DTA_SELECTMASK
;
5101 arg1
= settings
->op
[stage
].carg1
& WINED3DTA_SELECTMASK
;
5102 arg2
= settings
->op
[stage
].carg2
& 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
;
5110 if (settings
->op
[stage
].dst
== tempreg
)
5111 tempreg_used
= TRUE
;
5113 switch (settings
->op
[stage
].cop
)
5115 case WINED3D_TOP_BUMPENVMAP_LUMINANCE
:
5116 lum_map
|= 1 << stage
;
5118 case WINED3D_TOP_BUMPENVMAP
:
5119 bump_map
|= 1 << stage
;
5121 case WINED3D_TOP_BLEND_TEXTURE_ALPHA
:
5122 case WINED3D_TOP_BLEND_TEXTURE_ALPHA_PM
:
5123 tex_map
|= 1 << stage
;
5126 case WINED3D_TOP_BLEND_FACTOR_ALPHA
:
5127 tfactor_used
= TRUE
;
5134 if (settings
->op
[stage
].aop
== WINED3D_TOP_DISABLE
)
5137 arg0
= settings
->op
[stage
].aarg0
& WINED3DTA_SELECTMASK
;
5138 arg1
= settings
->op
[stage
].aarg1
& WINED3DTA_SELECTMASK
;
5139 arg2
= settings
->op
[stage
].aarg2
& WINED3DTA_SELECTMASK
;
5141 if (arg0
== WINED3DTA_TEXTURE
|| arg1
== WINED3DTA_TEXTURE
|| arg2
== WINED3DTA_TEXTURE
)
5142 tex_map
|= 1 << stage
;
5143 if (arg0
== WINED3DTA_TFACTOR
|| arg1
== WINED3DTA_TFACTOR
|| arg2
== WINED3DTA_TFACTOR
)
5144 tfactor_used
= TRUE
;
5145 if (arg0
== WINED3DTA_TEMP
|| arg1
== WINED3DTA_TEMP
|| arg2
== WINED3DTA_TEMP
)
5146 tempreg_used
= TRUE
;
5148 lowest_disabled_stage
= stage
;
5150 shader_addline(buffer
, "#version 120\n");
5152 shader_addline(buffer
, "vec4 tmp0, tmp1;\n");
5153 shader_addline(buffer
, "vec4 ret;\n");
5154 if (tempreg_used
|| settings
->sRGB_write
)
5155 shader_addline(buffer
, "vec4 temp_reg;\n");
5156 shader_addline(buffer
, "vec4 arg0, arg1, arg2;\n");
5158 for (stage
= 0; stage
< MAX_TEXTURES
; ++stage
)
5160 if (!(tex_map
& (1 << stage
)))
5163 switch (settings
->op
[stage
].tex_type
)
5166 shader_addline(buffer
, "uniform sampler1D ps_sampler%u;\n", stage
);
5169 shader_addline(buffer
, "uniform sampler2D ps_sampler%u;\n", stage
);
5172 shader_addline(buffer
, "uniform sampler3D ps_sampler%u;\n", stage
);
5175 shader_addline(buffer
, "uniform samplerCube ps_sampler%u;\n", stage
);
5178 shader_addline(buffer
, "uniform sampler2DRect ps_sampler%u;\n", stage
);
5181 FIXME("Unhandled sampler type %#x.\n", settings
->op
[stage
].tex_type
);
5185 shader_addline(buffer
, "vec4 tex%u;\n", stage
);
5187 if (!(bump_map
& (1 << stage
)))
5189 shader_addline(buffer
, "uniform mat2 bumpenv_mat%u;\n", stage
);
5191 if (!(lum_map
& (1 << stage
)))
5193 shader_addline(buffer
, "uniform float bumpenv_lum_scale%u;\n", stage
);
5194 shader_addline(buffer
, "uniform float bumpenv_lum_offset%u;\n", stage
);
5197 shader_addline(buffer
, "uniform vec4 tex_factor;\n");
5198 shader_addline(buffer
, "uniform vec4 specular_enable;\n");
5200 if (settings
->sRGB_write
)
5202 shader_addline(buffer
, "const vec4 srgb_const0 = vec4(%.8e, %.8e, %.8e, %.8e);\n",
5203 srgb_pow
, srgb_mul_high
, srgb_sub_high
, srgb_mul_low
);
5204 shader_addline(buffer
, "const vec4 srgb_const1 = vec4(%.8e, 0.0, 0.0, 0.0);\n",
5208 shader_addline(buffer
, "void main()\n{\n");
5210 if (lowest_disabled_stage
< 7 && settings
->emul_clipplanes
)
5211 shader_addline(buffer
, "if (any(lessThan(gl_texCoord[7], vec4(0.0)))) discard;\n");
5213 /* Generate texture sampling instructions) */
5214 for (stage
= 0; stage
< MAX_TEXTURES
&& settings
->op
[stage
].cop
!= WINED3D_TOP_DISABLE
; ++stage
)
5216 const char *texture_function
, *coord_mask
;
5217 char tex_reg_name
[8];
5220 if (!(tex_map
& (1 << stage
)))
5223 if (settings
->op
[stage
].projected
== proj_none
)
5227 else if (settings
->op
[stage
].projected
== proj_count4
5228 || settings
->op
[stage
].projected
== proj_count3
)
5234 FIXME("Unexpected projection mode %d\n", settings
->op
[stage
].projected
);
5238 if (settings
->op
[stage
].cop
== WINED3D_TOP_BUMPENVMAP
5239 || settings
->op
[stage
].cop
== WINED3D_TOP_BUMPENVMAP_LUMINANCE
)
5244 switch (settings
->op
[stage
].tex_type
)
5249 texture_function
= "texture1DProj";
5254 texture_function
= "texture1D";
5261 texture_function
= "texture2DProj";
5266 texture_function
= "texture2D";
5273 texture_function
= "texture3DProj";
5274 coord_mask
= "xyzw";
5278 texture_function
= "texture3D";
5283 texture_function
= "textureCube";
5289 texture_function
= "texture2DRectProj";
5294 texture_function
= "texture2DRect";
5299 FIXME("Unhandled texture type %#x.\n", settings
->op
[stage
].tex_type
);
5300 texture_function
= "";
5301 coord_mask
= "xyzw";
5306 && (settings
->op
[stage
- 1].cop
== WINED3D_TOP_BUMPENVMAP
5307 || settings
->op
[stage
- 1].cop
== WINED3D_TOP_BUMPENVMAP_LUMINANCE
))
5309 shader_addline(buffer
, "ret.xy = bumpenv_mat%u * tex%u.xy;\n", stage
- 1, stage
- 1);
5311 /* With projective textures, texbem only divides the static
5312 * texture coord, not the displacement, so multiply the
5313 * displacement with the dividing parameter before passing it to
5315 if (settings
->op
[stage
].projected
!= proj_none
)
5317 if (settings
->op
[stage
].projected
== proj_count4
)
5319 shader_addline(buffer
, "ret.xy = (ret.xy * gl_TexCoord[%u].w) + gl_TexCoord[%u].xy;\n",
5321 shader_addline(buffer
, "ret.zw = gl_TexCoord[%u].ww;\n", stage
);
5325 shader_addline(buffer
, "ret.xy = (ret.xy * gl_TexCoord[%u].z) + gl_TexCoord[%u].xy;\n",
5327 shader_addline(buffer
, "ret.zw = gl_TexCoord[%u].zz;\n", stage
);
5332 shader_addline(buffer
, "ret = gl_TexCoord[%u] + ret.xyxy;\n", stage
);
5336 shader_addline(buffer
, "tex%u = clamp(%s(ps_sampler%u, ret.%s), 0.0, 1.0);\n",
5337 stage
, texture_function
, stage
, coord_mask
);
5339 shader_addline(buffer
, "tex%u = %s(ps_sampler%u, ret.%s);\n",
5340 stage
, texture_function
, stage
, coord_mask
);
5342 if (settings
->op
[stage
- 1].cop
== WINED3D_TOP_BUMPENVMAP_LUMINANCE
)
5343 shader_addline(buffer
, "tex%u *= clamp(tex%u.z * bumpenv_lum_scale%u + bumpenv_lum_offset%u, 0.0, 1.0);\n",
5344 stage
, stage
- 1, stage
- 1, stage
- 1);
5346 else if (settings
->op
[stage
].projected
== proj_count3
)
5349 shader_addline(buffer
, "tex%u = clamp(%s(ps_sampler%u, gl_TexCoord[%u].xyz), 0.0, 1.0);\n",
5350 stage
, texture_function
, stage
, stage
);
5352 shader_addline(buffer
, "tex%u = %s(ps_sampler%u, gl_TexCoord[%u].xyz);\n",
5353 stage
, texture_function
, stage
, stage
);
5358 shader_addline(buffer
, "tex%u = clamp(%s(ps_sampler%u, gl_TexCoord[%u].%s), 0.0, 1.0);\n",
5359 stage
, texture_function
, stage
, stage
, coord_mask
);
5361 shader_addline(buffer
, "tex%u = %s(ps_sampler%u, gl_TexCoord[%u].%s);\n",
5362 stage
, texture_function
, stage
, stage
, coord_mask
);
5365 sprintf(tex_reg_name
, "tex%u", stage
);
5366 shader_glsl_color_correction_ext(buffer
, tex_reg_name
, WINED3DSP_WRITEMASK_ALL
,
5367 settings
->op
[stage
].color_fixup
);
5370 /* Generate the main shader */
5371 for (stage
= 0; stage
< MAX_TEXTURES
; ++stage
)
5375 if (settings
->op
[stage
].cop
== WINED3D_TOP_DISABLE
)
5378 final_combiner_src
= "gl_Color";
5382 if (settings
->op
[stage
].cop
== WINED3D_TOP_SELECT_ARG1
5383 && settings
->op
[stage
].aop
== WINED3D_TOP_SELECT_ARG1
)
5384 op_equal
= settings
->op
[stage
].carg1
== settings
->op
[stage
].aarg1
;
5385 else if (settings
->op
[stage
].cop
== WINED3D_TOP_SELECT_ARG1
5386 && settings
->op
[stage
].aop
== WINED3D_TOP_SELECT_ARG2
)
5387 op_equal
= settings
->op
[stage
].carg1
== settings
->op
[stage
].aarg2
;
5388 else if (settings
->op
[stage
].cop
== WINED3D_TOP_SELECT_ARG2
5389 && settings
->op
[stage
].aop
== WINED3D_TOP_SELECT_ARG1
)
5390 op_equal
= settings
->op
[stage
].carg2
== settings
->op
[stage
].aarg1
;
5391 else if (settings
->op
[stage
].cop
== WINED3D_TOP_SELECT_ARG2
5392 && settings
->op
[stage
].aop
== WINED3D_TOP_SELECT_ARG2
)
5393 op_equal
= settings
->op
[stage
].carg2
== settings
->op
[stage
].aarg2
;
5395 op_equal
= settings
->op
[stage
].aop
== settings
->op
[stage
].cop
5396 && settings
->op
[stage
].carg0
== settings
->op
[stage
].aarg0
5397 && settings
->op
[stage
].carg1
== settings
->op
[stage
].aarg1
5398 && settings
->op
[stage
].carg2
== settings
->op
[stage
].aarg2
;
5400 if (settings
->op
[stage
].aop
== WINED3D_TOP_DISABLE
)
5402 shader_glsl_ffp_fragment_op(buffer
, stage
, TRUE
, FALSE
, settings
->op
[stage
].dst
,
5403 settings
->op
[stage
].cop
, settings
->op
[stage
].carg0
,
5404 settings
->op
[stage
].carg1
, settings
->op
[stage
].carg2
);
5406 shader_addline(buffer
, "ret.w = gl_Color.w;\n");
5410 shader_glsl_ffp_fragment_op(buffer
, stage
, TRUE
, TRUE
, settings
->op
[stage
].dst
,
5411 settings
->op
[stage
].cop
, settings
->op
[stage
].carg0
,
5412 settings
->op
[stage
].carg1
, settings
->op
[stage
].carg2
);
5416 shader_glsl_ffp_fragment_op(buffer
, stage
, TRUE
, FALSE
, settings
->op
[stage
].dst
,
5417 settings
->op
[stage
].cop
, settings
->op
[stage
].carg0
,
5418 settings
->op
[stage
].carg1
, settings
->op
[stage
].carg2
);
5419 shader_glsl_ffp_fragment_op(buffer
, stage
, FALSE
, TRUE
, settings
->op
[stage
].dst
,
5420 settings
->op
[stage
].aop
, settings
->op
[stage
].aarg0
,
5421 settings
->op
[stage
].aarg1
, settings
->op
[stage
].aarg2
);
5425 shader_addline(buffer
, "gl_FragData[0] = gl_SecondaryColor * specular_enable + %s;\n", final_combiner_src
);
5427 if (settings
->sRGB_write
)
5428 shader_glsl_generate_srgb_write_correction(buffer
);
5430 shader_glsl_generate_fog_code(buffer
, settings
->fog
);
5432 shader_addline(buffer
, "}\n");
5434 shader_obj
= GL_EXTCALL(glCreateShaderObjectARB(GL_FRAGMENT_SHADER_ARB
));
5435 shader_glsl_compile(gl_info
, shader_obj
, buffer
->buffer
);
5439 static struct glsl_ffp_fragment_shader
*shader_glsl_find_ffp_fragment_shader(struct shader_glsl_priv
*priv
,
5440 const struct wined3d_gl_info
*gl_info
, const struct ffp_frag_settings
*args
)
5442 struct glsl_ffp_fragment_shader
*glsl_desc
;
5443 const struct ffp_frag_desc
*desc
;
5445 if ((desc
= find_ffp_frag_shader(&priv
->ffp_fragment_shaders
, args
)))
5446 return CONTAINING_RECORD(desc
, struct glsl_ffp_fragment_shader
, entry
);
5448 if (!(glsl_desc
= HeapAlloc(GetProcessHeap(), 0, sizeof(*glsl_desc
))))
5451 glsl_desc
->entry
.settings
= *args
;
5452 glsl_desc
->id
= shader_glsl_generate_ffp_fragment_shader(&priv
->shader_buffer
, args
, gl_info
);
5453 list_init(&glsl_desc
->linked_programs
);
5454 add_ffp_frag_shader(&priv
->ffp_fragment_shaders
, &glsl_desc
->entry
);
5460 static void shader_glsl_init_vs_uniform_locations(const struct wined3d_gl_info
*gl_info
,
5461 GLhandleARB program_id
, struct glsl_vs_program
*vs
)
5466 vs
->uniform_f_locations
= HeapAlloc(GetProcessHeap(), 0,
5467 sizeof(GLhandleARB
) * gl_info
->limits
.glsl_vs_float_constants
);
5468 for (i
= 0; i
< gl_info
->limits
.glsl_vs_float_constants
; ++i
)
5470 snprintf(name
, sizeof(name
), "vs_c[%u]", i
);
5471 vs
->uniform_f_locations
[i
] = GL_EXTCALL(glGetUniformLocationARB(program_id
, name
));
5474 for (i
= 0; i
< MAX_CONST_I
; ++i
)
5476 snprintf(name
, sizeof(name
), "vs_i[%u]", i
);
5477 vs
->uniform_i_locations
[i
] = GL_EXTCALL(glGetUniformLocationARB(program_id
, name
));
5480 vs
->pos_fixup_location
= GL_EXTCALL(glGetUniformLocationARB(program_id
, "posFixup"));
5483 static void shader_glsl_init_ps_uniform_locations(const struct wined3d_gl_info
*gl_info
,
5484 GLhandleARB program_id
, struct glsl_ps_program
*ps
)
5489 ps
->uniform_f_locations
= HeapAlloc(GetProcessHeap(), 0,
5490 sizeof(GLhandleARB
) * gl_info
->limits
.glsl_ps_float_constants
);
5491 for (i
= 0; i
< gl_info
->limits
.glsl_ps_float_constants
; ++i
)
5493 snprintf(name
, sizeof(name
), "ps_c[%u]", i
);
5494 ps
->uniform_f_locations
[i
] = GL_EXTCALL(glGetUniformLocationARB(program_id
, name
));
5497 for (i
= 0; i
< MAX_CONST_I
; ++i
)
5499 snprintf(name
, sizeof(name
), "ps_i[%u]", i
);
5500 ps
->uniform_i_locations
[i
] = GL_EXTCALL(glGetUniformLocationARB(program_id
, name
));
5503 for (i
= 0; i
< MAX_TEXTURES
; ++i
)
5505 snprintf(name
, sizeof(name
), "bumpenv_mat%u", i
);
5506 ps
->bumpenv_mat_location
[i
] = GL_EXTCALL(glGetUniformLocationARB(program_id
, name
));
5507 snprintf(name
, sizeof(name
), "bumpenv_lum_scale%u", i
);
5508 ps
->bumpenv_lum_scale_location
[i
] = GL_EXTCALL(glGetUniformLocationARB(program_id
, name
));
5509 snprintf(name
, sizeof(name
), "bumpenv_lum_offset%u", i
);
5510 ps
->bumpenv_lum_offset_location
[i
] = GL_EXTCALL(glGetUniformLocationARB(program_id
, name
));
5513 ps
->tex_factor_location
= GL_EXTCALL(glGetUniformLocationARB(program_id
, "tex_factor"));
5514 ps
->specular_enable_location
= GL_EXTCALL(glGetUniformLocationARB(program_id
, "specular_enable"));
5515 ps
->np2_fixup_location
= GL_EXTCALL(glGetUniformLocationARB(program_id
, "ps_samplerNP2Fixup"));
5516 ps
->ycorrection_location
= GL_EXTCALL(glGetUniformLocationARB(program_id
, "ycorrection"));
5519 /* Context activation is done by the caller. */
5520 static void set_glsl_shader_program(const struct wined3d_context
*context
, struct wined3d_device
*device
,
5521 enum wined3d_shader_mode vertex_mode
, enum wined3d_shader_mode fragment_mode
)
5523 const struct wined3d_state
*state
= &device
->stateBlock
->state
;
5524 const struct wined3d_gl_info
*gl_info
= context
->gl_info
;
5525 const struct ps_np2fixup_info
*np2fixup_info
= NULL
;
5526 struct shader_glsl_priv
*priv
= device
->shader_priv
;
5527 struct glsl_shader_prog_link
*entry
= NULL
;
5528 struct wined3d_shader
*vshader
= NULL
;
5529 struct wined3d_shader
*gshader
= NULL
;
5530 struct wined3d_shader
*pshader
= NULL
;
5531 GLhandleARB programId
= 0;
5532 GLhandleARB reorder_shader_id
= 0;
5534 struct ps_compile_args ps_compile_args
;
5535 struct vs_compile_args vs_compile_args
;
5536 GLhandleARB vs_id
, gs_id
, ps_id
;
5537 struct list
*ps_list
;
5539 if (vertex_mode
== WINED3D_SHADER_MODE_SHADER
)
5541 vshader
= state
->vertex_shader
;
5542 find_vs_compile_args(state
, vshader
, &vs_compile_args
);
5543 vs_id
= find_glsl_vshader(context
, &priv
->shader_buffer
, vshader
, &vs_compile_args
);
5545 if ((gshader
= state
->geometry_shader
))
5546 gs_id
= find_glsl_geometry_shader(context
, &priv
->shader_buffer
, gshader
);
5556 if (fragment_mode
== WINED3D_SHADER_MODE_SHADER
)
5558 pshader
= state
->pixel_shader
;
5559 find_ps_compile_args(state
, pshader
, &ps_compile_args
);
5560 ps_id
= find_glsl_pshader(context
, &priv
->shader_buffer
,
5561 pshader
, &ps_compile_args
, &np2fixup_info
);
5562 ps_list
= &pshader
->linked_programs
;
5564 else if (fragment_mode
== WINED3D_SHADER_MODE_FFP
&& priv
->fragment_pipe
== &glsl_fragment_pipe
)
5566 struct glsl_ffp_fragment_shader
*ffp_shader
;
5567 struct ffp_frag_settings settings
;
5569 gen_ffp_frag_op(device
, state
, &settings
, FALSE
);
5570 ffp_shader
= shader_glsl_find_ffp_fragment_shader(priv
, gl_info
, &settings
);
5571 ps_id
= ffp_shader
->id
;
5572 ps_list
= &ffp_shader
->linked_programs
;
5579 if ((!vs_id
&& !gs_id
&& !ps_id
) || (entry
= get_glsl_program_entry(priv
, vs_id
, gs_id
, ps_id
)))
5581 priv
->glsl_program
= entry
;
5585 /* If we get to this point, then no matching program exists, so we create one */
5586 programId
= GL_EXTCALL(glCreateProgramObjectARB());
5587 TRACE("Created new GLSL shader program %u\n", programId
);
5589 /* Create the entry */
5590 entry
= HeapAlloc(GetProcessHeap(), 0, sizeof(struct glsl_shader_prog_link
));
5591 entry
->programId
= programId
;
5592 entry
->vs
.id
= vs_id
;
5593 entry
->gs
.id
= gs_id
;
5594 entry
->ps
.id
= ps_id
;
5595 entry
->constant_version
= 0;
5596 entry
->ps
.np2_fixup_info
= np2fixup_info
;
5597 /* Add the hash table entry */
5598 add_glsl_program_entry(priv
, entry
);
5600 /* Set the current program */
5601 priv
->glsl_program
= entry
;
5603 /* Attach GLSL vshader */
5606 WORD map
= vshader
->reg_maps
.input_registers
;
5609 reorder_shader_id
= generate_param_reorder_function(&priv
->shader_buffer
, vshader
, pshader
, gl_info
);
5610 TRACE("Attaching GLSL shader object %u to program %u\n", reorder_shader_id
, programId
);
5611 GL_EXTCALL(glAttachObjectARB(programId
, reorder_shader_id
));
5612 checkGLcall("glAttachObjectARB");
5613 /* Flag the reorder function for deletion, then it will be freed automatically when the program
5616 GL_EXTCALL(glDeleteObjectARB(reorder_shader_id
));
5618 TRACE("Attaching GLSL shader object %u to program %u.\n", vs_id
, programId
);
5619 GL_EXTCALL(glAttachObjectARB(programId
, vs_id
));
5620 checkGLcall("glAttachObjectARB");
5622 /* Bind vertex attributes to a corresponding index number to match
5623 * the same index numbers as ARB_vertex_programs (makes loading
5624 * vertex attributes simpler). With this method, we can use the
5625 * exact same code to load the attributes later for both ARB and
5628 * We have to do this here because we need to know the Program ID
5629 * in order to make the bindings work, and it has to be done prior
5630 * to linking the GLSL program. */
5631 for (i
= 0; map
; map
>>= 1, ++i
)
5633 if (!(map
& 1)) continue;
5635 snprintf(tmp_name
, sizeof(tmp_name
), "vs_in%u", i
);
5636 GL_EXTCALL(glBindAttribLocationARB(programId
, i
, tmp_name
));
5638 checkGLcall("glBindAttribLocationARB");
5640 list_add_head(&vshader
->linked_programs
, &entry
->vs
.shader_entry
);
5645 TRACE("Attaching GLSL geometry shader object %u to program %u.\n", gs_id
, programId
);
5646 GL_EXTCALL(glAttachObjectARB(programId
, gs_id
));
5647 checkGLcall("glAttachObjectARB");
5649 TRACE("input type %s, output type %s, vertices out %u.\n",
5650 debug_d3dprimitivetype(gshader
->u
.gs
.input_type
),
5651 debug_d3dprimitivetype(gshader
->u
.gs
.output_type
),
5652 gshader
->u
.gs
.vertices_out
);
5653 GL_EXTCALL(glProgramParameteriARB(programId
, GL_GEOMETRY_INPUT_TYPE_ARB
,
5654 gl_primitive_type_from_d3d(gshader
->u
.gs
.input_type
)));
5655 GL_EXTCALL(glProgramParameteriARB(programId
, GL_GEOMETRY_OUTPUT_TYPE_ARB
,
5656 gl_primitive_type_from_d3d(gshader
->u
.gs
.output_type
)));
5657 GL_EXTCALL(glProgramParameteriARB(programId
, GL_GEOMETRY_VERTICES_OUT_ARB
,
5658 gshader
->u
.gs
.vertices_out
));
5659 checkGLcall("glProgramParameteriARB");
5661 list_add_head(&gshader
->linked_programs
, &entry
->gs
.shader_entry
);
5664 /* Attach GLSL pshader */
5667 TRACE("Attaching GLSL shader object %u to program %u.\n", ps_id
, programId
);
5668 GL_EXTCALL(glAttachObjectARB(programId
, ps_id
));
5669 checkGLcall("glAttachObjectARB");
5671 list_add_head(ps_list
, &entry
->ps
.shader_entry
);
5674 /* Link the program */
5675 TRACE("Linking GLSL shader program %u\n", programId
);
5676 GL_EXTCALL(glLinkProgramARB(programId
));
5677 shader_glsl_validate_link(gl_info
, programId
);
5679 shader_glsl_init_vs_uniform_locations(gl_info
, programId
, &entry
->vs
);
5680 shader_glsl_init_ps_uniform_locations(gl_info
, programId
, &entry
->ps
);
5681 checkGLcall("Find glsl program uniform locations");
5683 if (pshader
&& pshader
->reg_maps
.shader_version
.major
>= 3
5684 && pshader
->u
.ps
.declared_in_count
> vec4_varyings(3, gl_info
))
5686 TRACE("Shader %d needs vertex color clamping disabled\n", programId
);
5687 entry
->vs
.vertex_color_clamp
= GL_FALSE
;
5691 entry
->vs
.vertex_color_clamp
= GL_FIXED_ONLY_ARB
;
5694 /* Set the shader to allow uniform loading on it */
5695 GL_EXTCALL(glUseProgramObjectARB(programId
));
5696 checkGLcall("glUseProgramObjectARB(programId)");
5698 /* Load the vertex and pixel samplers now. The function that finds the mappings makes sure
5699 * that it stays the same for each vertexshader-pixelshader pair(=linked glsl program). If
5700 * a pshader with fixed function pipeline is used there are no vertex samplers, and if a
5701 * vertex shader with fixed function pixel processing is used we make sure that the card
5702 * supports enough samplers to allow the max number of vertex samplers with all possible
5703 * fixed function fragment processing setups. So once the program is linked these samplers
5706 shader_glsl_load_vsamplers(gl_info
, device
->texUnitMap
, programId
);
5707 shader_glsl_load_psamplers(gl_info
, device
->texUnitMap
, programId
);
5709 /* If the local constants do not have to be loaded with the environment constants,
5710 * load them now to have them hardcoded in the GLSL program. This saves some CPU cycles
5713 if (pshader
&& !pshader
->load_local_constsF
)
5714 hardcode_local_constants(pshader
, gl_info
, programId
, "ps");
5715 if (vshader
&& !vshader
->load_local_constsF
)
5716 hardcode_local_constants(vshader
, gl_info
, programId
, "vs");
5719 /* Context activation is done by the caller. */
5720 static GLhandleARB
create_glsl_blt_shader(const struct wined3d_gl_info
*gl_info
, enum tex_types tex_type
, BOOL masked
)
5722 GLhandleARB program_id
;
5723 GLhandleARB vshader_id
, pshader_id
;
5724 const char *blt_pshader
;
5726 static const char *blt_vshader
=
5730 " gl_Position = gl_Vertex;\n"
5731 " gl_FrontColor = vec4(1.0);\n"
5732 " gl_TexCoord[0] = gl_MultiTexCoord0;\n"
5735 static const char * const blt_pshaders_full
[tex_type_count
] =
5741 "uniform sampler2D sampler;\n"
5744 " gl_FragDepth = texture2D(sampler, gl_TexCoord[0].xy).x;\n"
5750 "uniform samplerCube sampler;\n"
5753 " gl_FragDepth = textureCube(sampler, gl_TexCoord[0].xyz).x;\n"
5757 "#extension GL_ARB_texture_rectangle : enable\n"
5758 "uniform sampler2DRect sampler;\n"
5761 " gl_FragDepth = texture2DRect(sampler, gl_TexCoord[0].xy).x;\n"
5765 static const char * const blt_pshaders_masked
[tex_type_count
] =
5771 "uniform sampler2D sampler;\n"
5772 "uniform vec4 mask;\n"
5775 " if (all(lessThan(gl_FragCoord.xy, mask.zw))) discard;\n"
5776 " gl_FragDepth = texture2D(sampler, gl_TexCoord[0].xy).x;\n"
5782 "uniform samplerCube sampler;\n"
5783 "uniform vec4 mask;\n"
5786 " if (all(lessThan(gl_FragCoord.xy, mask.zw))) discard;\n"
5787 " gl_FragDepth = textureCube(sampler, gl_TexCoord[0].xyz).x;\n"
5791 "#extension GL_ARB_texture_rectangle : enable\n"
5792 "uniform sampler2DRect sampler;\n"
5793 "uniform vec4 mask;\n"
5796 " if (all(lessThan(gl_FragCoord.xy, mask.zw))) discard;\n"
5797 " gl_FragDepth = texture2DRect(sampler, gl_TexCoord[0].xy).x;\n"
5801 blt_pshader
= masked
? blt_pshaders_masked
[tex_type
] : blt_pshaders_full
[tex_type
];
5804 FIXME("tex_type %#x not supported\n", tex_type
);
5808 vshader_id
= GL_EXTCALL(glCreateShaderObjectARB(GL_VERTEX_SHADER_ARB
));
5809 shader_glsl_compile(gl_info
, vshader_id
, blt_vshader
);
5811 pshader_id
= GL_EXTCALL(glCreateShaderObjectARB(GL_FRAGMENT_SHADER_ARB
));
5812 shader_glsl_compile(gl_info
, pshader_id
, blt_pshader
);
5814 program_id
= GL_EXTCALL(glCreateProgramObjectARB());
5815 GL_EXTCALL(glAttachObjectARB(program_id
, vshader_id
));
5816 GL_EXTCALL(glAttachObjectARB(program_id
, pshader_id
));
5817 GL_EXTCALL(glLinkProgramARB(program_id
));
5819 shader_glsl_validate_link(gl_info
, program_id
);
5821 /* Once linked we can mark the shaders for deletion. They will be deleted once the program
5824 GL_EXTCALL(glDeleteObjectARB(vshader_id
));
5825 GL_EXTCALL(glDeleteObjectARB(pshader_id
));
5829 /* Context activation is done by the caller. */
5830 static void shader_glsl_select(const struct wined3d_context
*context
, enum wined3d_shader_mode vertex_mode
,
5831 enum wined3d_shader_mode fragment_mode
)
5833 const struct wined3d_gl_info
*gl_info
= context
->gl_info
;
5834 struct wined3d_device
*device
= context
->swapchain
->device
;
5835 struct shader_glsl_priv
*priv
= device
->shader_priv
;
5836 GLhandleARB program_id
= 0;
5837 GLenum old_vertex_color_clamp
, current_vertex_color_clamp
;
5839 old_vertex_color_clamp
= priv
->glsl_program
? priv
->glsl_program
->vs
.vertex_color_clamp
: GL_FIXED_ONLY_ARB
;
5840 set_glsl_shader_program(context
, device
, vertex_mode
, fragment_mode
);
5841 current_vertex_color_clamp
= priv
->glsl_program
? priv
->glsl_program
->vs
.vertex_color_clamp
: GL_FIXED_ONLY_ARB
;
5842 if (old_vertex_color_clamp
!= current_vertex_color_clamp
)
5844 if (gl_info
->supported
[ARB_COLOR_BUFFER_FLOAT
])
5846 GL_EXTCALL(glClampColorARB(GL_CLAMP_VERTEX_COLOR_ARB
, current_vertex_color_clamp
));
5847 checkGLcall("glClampColorARB");
5851 FIXME("vertex color clamp needs to be changed, but extension not supported.\n");
5855 program_id
= priv
->glsl_program
? priv
->glsl_program
->programId
: 0;
5856 if (program_id
) TRACE("Using GLSL program %u\n", program_id
);
5857 GL_EXTCALL(glUseProgramObjectARB(program_id
));
5858 checkGLcall("glUseProgramObjectARB");
5860 /* In case that NP2 texcoord fixup data is found for the selected program, trigger a reload of the
5861 * constants. This has to be done because it can't be guaranteed that sampler() (from state.c) is
5862 * called between selecting the shader and using it, which results in wrong fixup for some frames. */
5863 if (priv
->glsl_program
&& priv
->glsl_program
->ps
.np2_fixup_info
)
5865 shader_glsl_load_np2fixup_constants(priv
, gl_info
, &device
->stateBlock
->state
);
5869 /* Context activation is done by the caller. */
5870 static void shader_glsl_select_depth_blt(void *shader_priv
, const struct wined3d_gl_info
*gl_info
,
5871 enum tex_types tex_type
, const SIZE
*ds_mask_size
)
5873 BOOL masked
= ds_mask_size
->cx
&& ds_mask_size
->cy
;
5874 struct shader_glsl_priv
*priv
= shader_priv
;
5875 GLhandleARB
*blt_program
;
5878 blt_program
= masked
? &priv
->depth_blt_program_masked
[tex_type
] : &priv
->depth_blt_program_full
[tex_type
];
5881 *blt_program
= create_glsl_blt_shader(gl_info
, tex_type
, masked
);
5882 loc
= GL_EXTCALL(glGetUniformLocationARB(*blt_program
, "sampler"));
5883 GL_EXTCALL(glUseProgramObjectARB(*blt_program
));
5884 GL_EXTCALL(glUniform1iARB(loc
, 0));
5888 GL_EXTCALL(glUseProgramObjectARB(*blt_program
));
5893 loc
= GL_EXTCALL(glGetUniformLocationARB(*blt_program
, "mask"));
5894 GL_EXTCALL(glUniform4fARB(loc
, 0.0f
, 0.0f
, (float)ds_mask_size
->cx
, (float)ds_mask_size
->cy
));
5898 /* Context activation is done by the caller. */
5899 static void shader_glsl_deselect_depth_blt(void *shader_priv
, const struct wined3d_gl_info
*gl_info
)
5901 struct shader_glsl_priv
*priv
= shader_priv
;
5902 GLhandleARB program_id
;
5904 program_id
= priv
->glsl_program
? priv
->glsl_program
->programId
: 0;
5905 if (program_id
) TRACE("Using GLSL program %u\n", program_id
);
5907 GL_EXTCALL(glUseProgramObjectARB(program_id
));
5908 checkGLcall("glUseProgramObjectARB");
5911 static void shader_glsl_destroy(struct wined3d_shader
*shader
)
5913 struct glsl_shader_private
*shader_data
= shader
->backend_data
;
5914 struct wined3d_device
*device
= shader
->device
;
5915 struct shader_glsl_priv
*priv
= device
->shader_priv
;
5916 const struct wined3d_gl_info
*gl_info
;
5917 const struct list
*linked_programs
;
5918 struct wined3d_context
*context
;
5920 if (!shader_data
|| !shader_data
->num_gl_shaders
)
5922 HeapFree(GetProcessHeap(), 0, shader_data
);
5923 shader
->backend_data
= NULL
;
5927 context
= context_acquire(device
, NULL
);
5928 gl_info
= context
->gl_info
;
5930 TRACE("Deleting linked programs.\n");
5931 linked_programs
= &shader
->linked_programs
;
5932 if (linked_programs
->next
)
5934 struct glsl_shader_prog_link
*entry
, *entry2
;
5937 switch (shader
->reg_maps
.shader_version
.type
)
5939 case WINED3D_SHADER_TYPE_PIXEL
:
5941 struct glsl_ps_compiled_shader
*gl_shaders
= shader_data
->gl_shaders
.ps
;
5943 LIST_FOR_EACH_ENTRY_SAFE(entry
, entry2
, linked_programs
,
5944 struct glsl_shader_prog_link
, ps
.shader_entry
)
5946 delete_glsl_program_entry(priv
, gl_info
, entry
);
5949 for (i
= 0; i
< shader_data
->num_gl_shaders
; ++i
)
5951 TRACE("Deleting pixel shader %u.\n", gl_shaders
[i
].prgId
);
5952 if (priv
->glsl_program
&& priv
->glsl_program
->ps
.id
== gl_shaders
[i
].prgId
)
5953 shader_glsl_select(context
, WINED3D_SHADER_MODE_NONE
, WINED3D_SHADER_MODE_NONE
);
5954 GL_EXTCALL(glDeleteObjectARB(gl_shaders
[i
].prgId
));
5955 checkGLcall("glDeleteObjectARB");
5957 HeapFree(GetProcessHeap(), 0, shader_data
->gl_shaders
.ps
);
5962 case WINED3D_SHADER_TYPE_VERTEX
:
5964 struct glsl_vs_compiled_shader
*gl_shaders
= shader_data
->gl_shaders
.vs
;
5966 LIST_FOR_EACH_ENTRY_SAFE(entry
, entry2
, linked_programs
,
5967 struct glsl_shader_prog_link
, vs
.shader_entry
)
5969 delete_glsl_program_entry(priv
, gl_info
, entry
);
5972 for (i
= 0; i
< shader_data
->num_gl_shaders
; ++i
)
5974 TRACE("Deleting vertex shader %u.\n", gl_shaders
[i
].prgId
);
5975 if (priv
->glsl_program
&& priv
->glsl_program
->vs
.id
== gl_shaders
[i
].prgId
)
5976 shader_glsl_select(context
, WINED3D_SHADER_MODE_NONE
, WINED3D_SHADER_MODE_NONE
);
5977 GL_EXTCALL(glDeleteObjectARB(gl_shaders
[i
].prgId
));
5978 checkGLcall("glDeleteObjectARB");
5980 HeapFree(GetProcessHeap(), 0, shader_data
->gl_shaders
.vs
);
5985 case WINED3D_SHADER_TYPE_GEOMETRY
:
5987 struct glsl_gs_compiled_shader
*gl_shaders
= shader_data
->gl_shaders
.gs
;
5989 LIST_FOR_EACH_ENTRY_SAFE(entry
, entry2
, linked_programs
,
5990 struct glsl_shader_prog_link
, gs
.shader_entry
)
5992 delete_glsl_program_entry(priv
, gl_info
, entry
);
5995 for (i
= 0; i
< shader_data
->num_gl_shaders
; ++i
)
5997 TRACE("Deleting geometry shader %u.\n", gl_shaders
[i
].id
);
5998 if (priv
->glsl_program
&& priv
->glsl_program
->gs
.id
== gl_shaders
[i
].id
)
5999 shader_glsl_select(context
, WINED3D_SHADER_MODE_NONE
, WINED3D_SHADER_MODE_NONE
);
6000 GL_EXTCALL(glDeleteObjectARB(gl_shaders
[i
].id
));
6001 checkGLcall("glDeleteObjectARB");
6003 HeapFree(GetProcessHeap(), 0, shader_data
->gl_shaders
.gs
);
6009 ERR("Unhandled shader type %#x.\n", shader
->reg_maps
.shader_version
.type
);
6014 HeapFree(GetProcessHeap(), 0, shader
->backend_data
);
6015 shader
->backend_data
= NULL
;
6017 context_release(context
);
6020 static int glsl_program_key_compare(const void *key
, const struct wine_rb_entry
*entry
)
6022 const struct glsl_program_key
*k
= key
;
6023 const struct glsl_shader_prog_link
*prog
= WINE_RB_ENTRY_VALUE(entry
,
6024 const struct glsl_shader_prog_link
, program_lookup_entry
);
6026 if (k
->vs_id
> prog
->vs
.id
) return 1;
6027 else if (k
->vs_id
< prog
->vs
.id
) return -1;
6029 if (k
->gs_id
> prog
->gs
.id
) return 1;
6030 else if (k
->gs_id
< prog
->gs
.id
) return -1;
6032 if (k
->ps_id
> prog
->ps
.id
) return 1;
6033 else if (k
->ps_id
< prog
->ps
.id
) return -1;
6038 static BOOL
constant_heap_init(struct constant_heap
*heap
, unsigned int constant_count
)
6040 SIZE_T size
= (constant_count
+ 1) * sizeof(*heap
->entries
) + constant_count
* sizeof(*heap
->positions
);
6041 void *mem
= HeapAlloc(GetProcessHeap(), 0, size
);
6045 ERR("Failed to allocate memory\n");
6049 heap
->entries
= mem
;
6050 heap
->entries
[1].version
= 0;
6051 heap
->positions
= (unsigned int *)(heap
->entries
+ constant_count
+ 1);
6057 static void constant_heap_free(struct constant_heap
*heap
)
6059 HeapFree(GetProcessHeap(), 0, heap
->entries
);
6062 static const struct wine_rb_functions wined3d_glsl_program_rb_functions
=
6067 glsl_program_key_compare
,
6070 static HRESULT
shader_glsl_alloc(struct wined3d_device
*device
, const struct fragment_pipeline
*fragment_pipe
)
6072 const struct wined3d_gl_info
*gl_info
= &device
->adapter
->gl_info
;
6073 struct shader_glsl_priv
*priv
= HeapAlloc(GetProcessHeap(), HEAP_ZERO_MEMORY
, sizeof(struct shader_glsl_priv
));
6074 SIZE_T stack_size
= wined3d_log2i(max(gl_info
->limits
.glsl_vs_float_constants
,
6075 gl_info
->limits
.glsl_ps_float_constants
)) + 1;
6076 struct fragment_caps fragment_caps
;
6077 void *fragment_priv
;
6079 if (!(fragment_priv
= fragment_pipe
->alloc_private(&glsl_shader_backend
, priv
)))
6081 ERR("Failed to initialize fragment pipe.\n");
6082 HeapFree(GetProcessHeap(), 0, priv
);
6086 if (!shader_buffer_init(&priv
->shader_buffer
))
6088 ERR("Failed to initialize shader buffer.\n");
6092 priv
->stack
= HeapAlloc(GetProcessHeap(), 0, stack_size
* sizeof(*priv
->stack
));
6095 ERR("Failed to allocate memory.\n");
6099 if (!constant_heap_init(&priv
->vconst_heap
, gl_info
->limits
.glsl_vs_float_constants
))
6101 ERR("Failed to initialize vertex shader constant heap\n");
6105 if (!constant_heap_init(&priv
->pconst_heap
, gl_info
->limits
.glsl_ps_float_constants
))
6107 ERR("Failed to initialize pixel shader constant heap\n");
6111 if (wine_rb_init(&priv
->program_lookup
, &wined3d_glsl_program_rb_functions
) == -1)
6113 ERR("Failed to initialize rbtree.\n");
6117 priv
->next_constant_version
= 1;
6118 fragment_pipe
->get_caps(gl_info
, &fragment_caps
);
6119 priv
->ffp_proj_control
= fragment_caps
.wined3d_caps
& WINED3D_FRAGMENT_CAP_PROJ_CONTROL
;
6120 device
->fragment_priv
= fragment_priv
;
6121 priv
->fragment_pipe
= fragment_pipe
;
6123 device
->shader_priv
= priv
;
6127 constant_heap_free(&priv
->pconst_heap
);
6128 constant_heap_free(&priv
->vconst_heap
);
6129 HeapFree(GetProcessHeap(), 0, priv
->stack
);
6130 shader_buffer_free(&priv
->shader_buffer
);
6131 fragment_pipe
->free_private(device
);
6132 HeapFree(GetProcessHeap(), 0, priv
);
6133 return E_OUTOFMEMORY
;
6136 /* Context activation is done by the caller. */
6137 static void shader_glsl_free(struct wined3d_device
*device
)
6139 const struct wined3d_gl_info
*gl_info
= &device
->adapter
->gl_info
;
6140 struct shader_glsl_priv
*priv
= device
->shader_priv
;
6143 for (i
= 0; i
< tex_type_count
; ++i
)
6145 if (priv
->depth_blt_program_full
[i
])
6147 GL_EXTCALL(glDeleteObjectARB(priv
->depth_blt_program_full
[i
]));
6149 if (priv
->depth_blt_program_masked
[i
])
6151 GL_EXTCALL(glDeleteObjectARB(priv
->depth_blt_program_masked
[i
]));
6155 wine_rb_destroy(&priv
->program_lookup
, NULL
, NULL
);
6156 constant_heap_free(&priv
->pconst_heap
);
6157 constant_heap_free(&priv
->vconst_heap
);
6158 HeapFree(GetProcessHeap(), 0, priv
->stack
);
6159 shader_buffer_free(&priv
->shader_buffer
);
6160 priv
->fragment_pipe
->free_private(device
);
6162 HeapFree(GetProcessHeap(), 0, device
->shader_priv
);
6163 device
->shader_priv
= NULL
;
6166 static void shader_glsl_context_destroyed(void *shader_priv
, const struct wined3d_context
*context
) {}
6168 static void shader_glsl_get_caps(const struct wined3d_gl_info
*gl_info
, struct shader_caps
*caps
)
6172 if (gl_info
->supported
[EXT_GPU_SHADER4
] && gl_info
->supported
[ARB_SHADER_BIT_ENCODING
]
6173 && gl_info
->supported
[ARB_GEOMETRY_SHADER4
] && gl_info
->glsl_version
>= MAKEDWORD_VERSION(1, 50)
6174 && gl_info
->supported
[ARB_DRAW_ELEMENTS_BASE_VERTEX
] && gl_info
->supported
[ARB_DRAW_INSTANCED
])
6176 /* ARB_shader_texture_lod or EXT_gpu_shader4 is required for the SM3
6177 * texldd and texldl instructions. */
6178 else if (gl_info
->supported
[ARB_SHADER_TEXTURE_LOD
] || gl_info
->supported
[EXT_GPU_SHADER4
])
6182 TRACE("Shader model %u.\n", shader_model
);
6184 caps
->vs_version
= min(wined3d_settings
.max_sm_vs
, shader_model
);
6185 caps
->gs_version
= min(wined3d_settings
.max_sm_gs
, shader_model
);
6186 caps
->ps_version
= min(wined3d_settings
.max_sm_ps
, shader_model
);
6188 caps
->vs_uniform_count
= gl_info
->limits
.glsl_vs_float_constants
;
6189 caps
->ps_uniform_count
= gl_info
->limits
.glsl_ps_float_constants
;
6191 /* FIXME: The following line is card dependent. -8.0 to 8.0 is the
6192 * Direct3D minimum requirement.
6194 * Both GL_ARB_fragment_program and GLSL require a "maximum representable magnitude"
6195 * of colors to be 2^10, and 2^32 for other floats. Should we use 1024 here?
6197 * The problem is that the refrast clamps temporary results in the shader to
6198 * [-MaxValue;+MaxValue]. If the card's max value is bigger than the one we advertize here,
6199 * then applications may miss the clamping behavior. On the other hand, if it is smaller,
6200 * the shader will generate incorrect results too. Unfortunately, GL deliberately doesn't
6201 * offer a way to query this.
6203 caps
->ps_1x_max_value
= 8.0;
6205 /* Ideally we'd only set caps like sRGB writes here if supported by both
6206 * the shader backend and the fragment pipe, but we can get called before
6207 * shader_glsl_alloc(). */
6208 caps
->wined3d_caps
= WINED3D_SHADER_CAP_VS_CLIPPING
6209 | WINED3D_SHADER_CAP_SRGB_WRITE
;
6212 static BOOL
shader_glsl_color_fixup_supported(struct color_fixup_desc fixup
)
6214 if (TRACE_ON(d3d_shader
) && TRACE_ON(d3d
))
6216 TRACE("Checking support for fixup:\n");
6217 dump_color_fixup_desc(fixup
);
6220 /* We support everything except YUV conversions. */
6221 if (!is_complex_fixup(fixup
))
6227 TRACE("[FAILED]\n");
6231 static const SHADER_HANDLER shader_glsl_instruction_handler_table
[WINED3DSIH_TABLE_SIZE
] =
6233 /* WINED3DSIH_ABS */ shader_glsl_map2gl
,
6234 /* WINED3DSIH_ADD */ shader_glsl_binop
,
6235 /* WINED3DSIH_AND */ shader_glsl_binop
,
6236 /* WINED3DSIH_BEM */ shader_glsl_bem
,
6237 /* WINED3DSIH_BREAK */ shader_glsl_break
,
6238 /* WINED3DSIH_BREAKC */ shader_glsl_breakc
,
6239 /* WINED3DSIH_BREAKP */ shader_glsl_breakp
,
6240 /* WINED3DSIH_CALL */ shader_glsl_call
,
6241 /* WINED3DSIH_CALLNZ */ shader_glsl_callnz
,
6242 /* WINED3DSIH_CMP */ shader_glsl_conditional_move
,
6243 /* WINED3DSIH_CND */ shader_glsl_cnd
,
6244 /* WINED3DSIH_CRS */ shader_glsl_cross
,
6245 /* WINED3DSIH_CUT */ shader_glsl_cut
,
6246 /* WINED3DSIH_DCL */ shader_glsl_nop
,
6247 /* WINED3DSIH_DCL_CONSTANT_BUFFER */ shader_glsl_nop
,
6248 /* WINED3DSIH_DCL_INPUT_PRIMITIVE */ shader_glsl_nop
,
6249 /* WINED3DSIH_DCL_OUTPUT_TOPOLOGY */ shader_glsl_nop
,
6250 /* WINED3DSIH_DCL_VERTICES_OUT */ shader_glsl_nop
,
6251 /* WINED3DSIH_DEF */ shader_glsl_nop
,
6252 /* WINED3DSIH_DEFB */ shader_glsl_nop
,
6253 /* WINED3DSIH_DEFI */ shader_glsl_nop
,
6254 /* WINED3DSIH_DIV */ shader_glsl_binop
,
6255 /* WINED3DSIH_DP2ADD */ shader_glsl_dp2add
,
6256 /* WINED3DSIH_DP3 */ shader_glsl_dot
,
6257 /* WINED3DSIH_DP4 */ shader_glsl_dot
,
6258 /* WINED3DSIH_DST */ shader_glsl_dst
,
6259 /* WINED3DSIH_DSX */ shader_glsl_map2gl
,
6260 /* WINED3DSIH_DSY */ shader_glsl_map2gl
,
6261 /* WINED3DSIH_ELSE */ shader_glsl_else
,
6262 /* WINED3DSIH_EMIT */ shader_glsl_emit
,
6263 /* WINED3DSIH_ENDIF */ shader_glsl_end
,
6264 /* WINED3DSIH_ENDLOOP */ shader_glsl_end
,
6265 /* WINED3DSIH_ENDREP */ shader_glsl_end
,
6266 /* WINED3DSIH_EQ */ shader_glsl_relop
,
6267 /* WINED3DSIH_EXP */ shader_glsl_map2gl
,
6268 /* WINED3DSIH_EXPP */ shader_glsl_expp
,
6269 /* WINED3DSIH_FRC */ shader_glsl_map2gl
,
6270 /* WINED3DSIH_FTOI */ shader_glsl_to_int
,
6271 /* WINED3DSIH_GE */ shader_glsl_relop
,
6272 /* WINED3DSIH_IADD */ shader_glsl_binop
,
6273 /* WINED3DSIH_IEQ */ NULL
,
6274 /* WINED3DSIH_IF */ shader_glsl_if
,
6275 /* WINED3DSIH_IFC */ shader_glsl_ifc
,
6276 /* WINED3DSIH_IGE */ shader_glsl_relop
,
6277 /* WINED3DSIH_IMUL */ shader_glsl_imul
,
6278 /* WINED3DSIH_ITOF */ shader_glsl_to_float
,
6279 /* WINED3DSIH_LABEL */ shader_glsl_label
,
6280 /* WINED3DSIH_LD */ NULL
,
6281 /* WINED3DSIH_LIT */ shader_glsl_lit
,
6282 /* WINED3DSIH_LOG */ shader_glsl_log
,
6283 /* WINED3DSIH_LOGP */ shader_glsl_log
,
6284 /* WINED3DSIH_LOOP */ shader_glsl_loop
,
6285 /* WINED3DSIH_LRP */ shader_glsl_lrp
,
6286 /* WINED3DSIH_LT */ shader_glsl_relop
,
6287 /* WINED3DSIH_M3x2 */ shader_glsl_mnxn
,
6288 /* WINED3DSIH_M3x3 */ shader_glsl_mnxn
,
6289 /* WINED3DSIH_M3x4 */ shader_glsl_mnxn
,
6290 /* WINED3DSIH_M4x3 */ shader_glsl_mnxn
,
6291 /* WINED3DSIH_M4x4 */ shader_glsl_mnxn
,
6292 /* WINED3DSIH_MAD */ shader_glsl_mad
,
6293 /* WINED3DSIH_MAX */ shader_glsl_map2gl
,
6294 /* WINED3DSIH_MIN */ shader_glsl_map2gl
,
6295 /* WINED3DSIH_MOV */ shader_glsl_mov
,
6296 /* WINED3DSIH_MOVA */ shader_glsl_mov
,
6297 /* WINED3DSIH_MOVC */ shader_glsl_conditional_move
,
6298 /* WINED3DSIH_MUL */ shader_glsl_binop
,
6299 /* WINED3DSIH_NOP */ shader_glsl_nop
,
6300 /* WINED3DSIH_NRM */ shader_glsl_nrm
,
6301 /* WINED3DSIH_PHASE */ shader_glsl_nop
,
6302 /* WINED3DSIH_POW */ shader_glsl_pow
,
6303 /* WINED3DSIH_RCP */ shader_glsl_rcp
,
6304 /* WINED3DSIH_REP */ shader_glsl_rep
,
6305 /* WINED3DSIH_RET */ shader_glsl_ret
,
6306 /* WINED3DSIH_ROUND_NI */ shader_glsl_map2gl
,
6307 /* WINED3DSIH_RSQ */ shader_glsl_rsq
,
6308 /* WINED3DSIH_SAMPLE */ NULL
,
6309 /* WINED3DSIH_SAMPLE_GRAD */ NULL
,
6310 /* WINED3DSIH_SAMPLE_LOD */ NULL
,
6311 /* WINED3DSIH_SETP */ NULL
,
6312 /* WINED3DSIH_SGE */ shader_glsl_compare
,
6313 /* WINED3DSIH_SGN */ shader_glsl_sgn
,
6314 /* WINED3DSIH_SINCOS */ shader_glsl_sincos
,
6315 /* WINED3DSIH_SLT */ shader_glsl_compare
,
6316 /* WINED3DSIH_SQRT */ NULL
,
6317 /* WINED3DSIH_SUB */ shader_glsl_binop
,
6318 /* WINED3DSIH_TEX */ shader_glsl_tex
,
6319 /* WINED3DSIH_TEXBEM */ shader_glsl_texbem
,
6320 /* WINED3DSIH_TEXBEML */ shader_glsl_texbem
,
6321 /* WINED3DSIH_TEXCOORD */ shader_glsl_texcoord
,
6322 /* WINED3DSIH_TEXDEPTH */ shader_glsl_texdepth
,
6323 /* WINED3DSIH_TEXDP3 */ shader_glsl_texdp3
,
6324 /* WINED3DSIH_TEXDP3TEX */ shader_glsl_texdp3tex
,
6325 /* WINED3DSIH_TEXKILL */ shader_glsl_texkill
,
6326 /* WINED3DSIH_TEXLDD */ shader_glsl_texldd
,
6327 /* WINED3DSIH_TEXLDL */ shader_glsl_texldl
,
6328 /* WINED3DSIH_TEXM3x2DEPTH */ shader_glsl_texm3x2depth
,
6329 /* WINED3DSIH_TEXM3x2PAD */ shader_glsl_texm3x2pad
,
6330 /* WINED3DSIH_TEXM3x2TEX */ shader_glsl_texm3x2tex
,
6331 /* WINED3DSIH_TEXM3x3 */ shader_glsl_texm3x3
,
6332 /* WINED3DSIH_TEXM3x3DIFF */ NULL
,
6333 /* WINED3DSIH_TEXM3x3PAD */ shader_glsl_texm3x3pad
,
6334 /* WINED3DSIH_TEXM3x3SPEC */ shader_glsl_texm3x3spec
,
6335 /* WINED3DSIH_TEXM3x3TEX */ shader_glsl_texm3x3tex
,
6336 /* WINED3DSIH_TEXM3x3VSPEC */ shader_glsl_texm3x3vspec
,
6337 /* WINED3DSIH_TEXREG2AR */ shader_glsl_texreg2ar
,
6338 /* WINED3DSIH_TEXREG2GB */ shader_glsl_texreg2gb
,
6339 /* WINED3DSIH_TEXREG2RGB */ shader_glsl_texreg2rgb
,
6340 /* WINED3DSIH_UDIV */ shader_glsl_udiv
,
6341 /* WINED3DSIH_USHR */ shader_glsl_binop
,
6342 /* WINED3DSIH_UTOF */ shader_glsl_to_float
,
6343 /* WINED3DSIH_XOR */ shader_glsl_binop
,
6346 static void shader_glsl_handle_instruction(const struct wined3d_shader_instruction
*ins
) {
6347 SHADER_HANDLER hw_fct
;
6349 /* Select handler */
6350 hw_fct
= shader_glsl_instruction_handler_table
[ins
->handler_idx
];
6352 /* Unhandled opcode */
6355 FIXME("Backend can't handle opcode %#x\n", ins
->handler_idx
);
6360 shader_glsl_add_instruction_modifiers(ins
);
6363 static BOOL
shader_glsl_has_ffp_proj_control(void *shader_priv
)
6365 struct shader_glsl_priv
*priv
= shader_priv
;
6367 return priv
->ffp_proj_control
;
6370 const struct wined3d_shader_backend_ops glsl_shader_backend
=
6372 shader_glsl_handle_instruction
,
6374 shader_glsl_select_depth_blt
,
6375 shader_glsl_deselect_depth_blt
,
6376 shader_glsl_update_float_vertex_constants
,
6377 shader_glsl_update_float_pixel_constants
,
6378 shader_glsl_load_constants
,
6379 shader_glsl_load_np2fixup_constants
,
6380 shader_glsl_destroy
,
6383 shader_glsl_context_destroyed
,
6384 shader_glsl_get_caps
,
6385 shader_glsl_color_fixup_supported
,
6386 shader_glsl_has_ffp_proj_control
,
6389 static void glsl_fragment_pipe_enable(const struct wined3d_gl_info
*gl_info
, BOOL enable
)
6391 /* Nothing to do. */
6394 static void glsl_fragment_pipe_get_caps(const struct wined3d_gl_info
*gl_info
, struct fragment_caps
*caps
)
6396 caps
->wined3d_caps
= WINED3D_FRAGMENT_CAP_PROJ_CONTROL
6397 | WINED3D_FRAGMENT_CAP_SRGB_WRITE
;
6398 caps
->PrimitiveMiscCaps
= WINED3DPMISCCAPS_TSSARGTEMP
;
6399 caps
->TextureOpCaps
= WINED3DTEXOPCAPS_DISABLE
6400 | WINED3DTEXOPCAPS_SELECTARG1
6401 | WINED3DTEXOPCAPS_SELECTARG2
6402 | WINED3DTEXOPCAPS_MODULATE4X
6403 | WINED3DTEXOPCAPS_MODULATE2X
6404 | WINED3DTEXOPCAPS_MODULATE
6405 | WINED3DTEXOPCAPS_ADDSIGNED2X
6406 | WINED3DTEXOPCAPS_ADDSIGNED
6407 | WINED3DTEXOPCAPS_ADD
6408 | WINED3DTEXOPCAPS_SUBTRACT
6409 | WINED3DTEXOPCAPS_ADDSMOOTH
6410 | WINED3DTEXOPCAPS_BLENDCURRENTALPHA
6411 | WINED3DTEXOPCAPS_BLENDFACTORALPHA
6412 | WINED3DTEXOPCAPS_BLENDTEXTUREALPHA
6413 | WINED3DTEXOPCAPS_BLENDDIFFUSEALPHA
6414 | WINED3DTEXOPCAPS_BLENDTEXTUREALPHAPM
6415 | WINED3DTEXOPCAPS_MODULATEALPHA_ADDCOLOR
6416 | WINED3DTEXOPCAPS_MODULATECOLOR_ADDALPHA
6417 | WINED3DTEXOPCAPS_MODULATEINVCOLOR_ADDALPHA
6418 | WINED3DTEXOPCAPS_MODULATEINVALPHA_ADDCOLOR
6419 | WINED3DTEXOPCAPS_DOTPRODUCT3
6420 | WINED3DTEXOPCAPS_MULTIPLYADD
6421 | WINED3DTEXOPCAPS_LERP
6422 | WINED3DTEXOPCAPS_BUMPENVMAP
6423 | WINED3DTEXOPCAPS_BUMPENVMAPLUMINANCE
;
6424 caps
->MaxTextureBlendStages
= 8;
6425 caps
->MaxSimultaneousTextures
= min(gl_info
->limits
.fragment_samplers
, 8);
6428 static void *glsl_fragment_pipe_alloc(const struct wined3d_shader_backend_ops
*shader_backend
, void *shader_priv
)
6430 struct shader_glsl_priv
*priv
;
6432 if (shader_backend
== &glsl_shader_backend
)
6436 if (wine_rb_init(&priv
->ffp_fragment_shaders
, &wined3d_ffp_frag_program_rb_functions
) == -1)
6438 ERR("Failed to initialize rbtree.\n");
6445 FIXME("GLSL fragment pipe without GLSL shader backend not implemented.\n");
6450 struct glsl_ffp_destroy_ctx
6452 struct shader_glsl_priv
*priv
;
6453 const struct wined3d_gl_info
*gl_info
;
6456 static void shader_glsl_free_ffp_fragment_shader(struct wine_rb_entry
*entry
, void *context
)
6458 struct glsl_ffp_fragment_shader
*shader
= WINE_RB_ENTRY_VALUE(entry
,
6459 struct glsl_ffp_fragment_shader
, entry
.entry
);
6460 struct glsl_shader_prog_link
*program
, *program2
;
6461 struct glsl_ffp_destroy_ctx
*ctx
= context
;
6463 LIST_FOR_EACH_ENTRY_SAFE(program
, program2
, &shader
->linked_programs
,
6464 struct glsl_shader_prog_link
, ps
.shader_entry
)
6466 delete_glsl_program_entry(ctx
->priv
, ctx
->gl_info
, program
);
6468 ctx
->gl_info
->gl_ops
.ext
.p_glDeleteObjectARB(shader
->id
);
6469 HeapFree(GetProcessHeap(), 0, shader
);
6472 /* Context activation is done by the caller. */
6473 static void glsl_fragment_pipe_free(struct wined3d_device
*device
)
6475 struct shader_glsl_priv
*priv
= device
->fragment_priv
;
6476 struct glsl_ffp_destroy_ctx ctx
;
6479 ctx
.gl_info
= &device
->adapter
->gl_info
;
6480 wine_rb_destroy(&priv
->ffp_fragment_shaders
, shader_glsl_free_ffp_fragment_shader
, &ctx
);
6483 static void glsl_fragment_pipe_shader(struct wined3d_context
*context
,
6484 const struct wined3d_state
*state
, DWORD state_id
)
6486 context
->last_was_pshader
= use_ps(state
);
6488 context
->select_shader
= 1;
6489 context
->load_constants
= 1;
6492 static void glsl_fragment_pipe_fog(struct wined3d_context
*context
,
6493 const struct wined3d_state
*state
, DWORD state_id
)
6495 BOOL use_vshader
= use_vs(state
);
6496 enum fogsource new_source
;
6498 context
->select_shader
= 1;
6499 context
->load_constants
= 1;
6501 if (!state
->render_states
[WINED3D_RS_FOGENABLE
])
6504 if (state
->render_states
[WINED3D_RS_FOGTABLEMODE
] == WINED3D_FOG_NONE
)
6507 new_source
= FOGSOURCE_VS
;
6508 else if (state
->render_states
[WINED3D_RS_FOGVERTEXMODE
] == WINED3D_FOG_NONE
|| context
->last_was_rhw
)
6509 new_source
= FOGSOURCE_COORD
;
6511 new_source
= FOGSOURCE_FFP
;
6515 new_source
= FOGSOURCE_FFP
;
6518 if (new_source
!= context
->fog_source
)
6520 context
->fog_source
= new_source
;
6521 state_fogstartend(context
, state
, STATE_RENDER(WINED3D_RS_FOGSTART
));
6525 static void glsl_fragment_pipe_tex_transform(struct wined3d_context
*context
,
6526 const struct wined3d_state
*state
, DWORD state_id
)
6528 context
->select_shader
= 1;
6529 context
->load_constants
= 1;
6532 static void glsl_fragment_pipe_invalidate_constants(struct wined3d_context
*context
,
6533 const struct wined3d_state
*state
, DWORD state_id
)
6535 context
->load_constants
= 1;
6538 static const struct StateEntryTemplate glsl_fragment_pipe_state_template
[] =
6540 {STATE_RENDER(WINED3D_RS_TEXTUREFACTOR
), {STATE_RENDER(WINED3D_RS_TEXTUREFACTOR
), glsl_fragment_pipe_invalidate_constants
}, WINED3D_GL_EXT_NONE
},
6541 {STATE_TEXTURESTAGE(0, WINED3D_TSS_COLOR_OP
), {STATE_PIXELSHADER
, NULL
}, WINED3D_GL_EXT_NONE
},
6542 {STATE_TEXTURESTAGE(0, WINED3D_TSS_COLOR_ARG1
), {STATE_PIXELSHADER
, NULL
}, WINED3D_GL_EXT_NONE
},
6543 {STATE_TEXTURESTAGE(0, WINED3D_TSS_COLOR_ARG2
), {STATE_PIXELSHADER
, NULL
}, WINED3D_GL_EXT_NONE
},
6544 {STATE_TEXTURESTAGE(0, WINED3D_TSS_COLOR_ARG0
), {STATE_PIXELSHADER
, NULL
}, WINED3D_GL_EXT_NONE
},
6545 {STATE_TEXTURESTAGE(0, WINED3D_TSS_ALPHA_OP
), {STATE_PIXELSHADER
, NULL
}, WINED3D_GL_EXT_NONE
},
6546 {STATE_TEXTURESTAGE(0, WINED3D_TSS_ALPHA_ARG1
), {STATE_PIXELSHADER
, NULL
}, WINED3D_GL_EXT_NONE
},
6547 {STATE_TEXTURESTAGE(0, WINED3D_TSS_ALPHA_ARG2
), {STATE_PIXELSHADER
, NULL
}, WINED3D_GL_EXT_NONE
},
6548 {STATE_TEXTURESTAGE(0, WINED3D_TSS_ALPHA_ARG0
), {STATE_PIXELSHADER
, NULL
}, WINED3D_GL_EXT_NONE
},
6549 {STATE_TEXTURESTAGE(0, WINED3D_TSS_RESULT_ARG
), {STATE_PIXELSHADER
, NULL
}, WINED3D_GL_EXT_NONE
},
6550 {STATE_TEXTURESTAGE(0, WINED3D_TSS_BUMPENV_MAT00
), {STATE_TEXTURESTAGE(0, WINED3D_TSS_BUMPENV_MAT00
), glsl_fragment_pipe_invalidate_constants
}, WINED3D_GL_EXT_NONE
},
6551 {STATE_TEXTURESTAGE(0, WINED3D_TSS_BUMPENV_MAT01
), {STATE_TEXTURESTAGE(0, WINED3D_TSS_BUMPENV_MAT00
), NULL
}, WINED3D_GL_EXT_NONE
},
6552 {STATE_TEXTURESTAGE(0, WINED3D_TSS_BUMPENV_MAT10
), {STATE_TEXTURESTAGE(0, WINED3D_TSS_BUMPENV_MAT00
), NULL
}, WINED3D_GL_EXT_NONE
},
6553 {STATE_TEXTURESTAGE(0, WINED3D_TSS_BUMPENV_MAT11
), {STATE_TEXTURESTAGE(0, WINED3D_TSS_BUMPENV_MAT00
), NULL
}, WINED3D_GL_EXT_NONE
},
6554 {STATE_TEXTURESTAGE(0, WINED3D_TSS_BUMPENV_LSCALE
), {STATE_TEXTURESTAGE(0, WINED3D_TSS_BUMPENV_LSCALE
), glsl_fragment_pipe_invalidate_constants
}, WINED3D_GL_EXT_NONE
},
6555 {STATE_TEXTURESTAGE(0, WINED3D_TSS_BUMPENV_LOFFSET
), {STATE_TEXTURESTAGE(0, WINED3D_TSS_BUMPENV_LSCALE
), NULL
}, WINED3D_GL_EXT_NONE
},
6556 {STATE_TEXTURESTAGE(1, WINED3D_TSS_COLOR_OP
), {STATE_PIXELSHADER
, NULL
}, WINED3D_GL_EXT_NONE
},
6557 {STATE_TEXTURESTAGE(1, WINED3D_TSS_COLOR_ARG1
), {STATE_PIXELSHADER
, NULL
}, WINED3D_GL_EXT_NONE
},
6558 {STATE_TEXTURESTAGE(1, WINED3D_TSS_COLOR_ARG2
), {STATE_PIXELSHADER
, NULL
}, WINED3D_GL_EXT_NONE
},
6559 {STATE_TEXTURESTAGE(1, WINED3D_TSS_COLOR_ARG0
), {STATE_PIXELSHADER
, NULL
}, WINED3D_GL_EXT_NONE
},
6560 {STATE_TEXTURESTAGE(1, WINED3D_TSS_ALPHA_OP
), {STATE_PIXELSHADER
, NULL
}, WINED3D_GL_EXT_NONE
},
6561 {STATE_TEXTURESTAGE(1, WINED3D_TSS_ALPHA_ARG1
), {STATE_PIXELSHADER
, NULL
}, WINED3D_GL_EXT_NONE
},
6562 {STATE_TEXTURESTAGE(1, WINED3D_TSS_ALPHA_ARG2
), {STATE_PIXELSHADER
, NULL
}, WINED3D_GL_EXT_NONE
},
6563 {STATE_TEXTURESTAGE(1, WINED3D_TSS_ALPHA_ARG0
), {STATE_PIXELSHADER
, NULL
}, WINED3D_GL_EXT_NONE
},
6564 {STATE_TEXTURESTAGE(1, WINED3D_TSS_RESULT_ARG
), {STATE_PIXELSHADER
, NULL
}, WINED3D_GL_EXT_NONE
},
6565 {STATE_TEXTURESTAGE(1, WINED3D_TSS_BUMPENV_MAT00
), {STATE_TEXTURESTAGE(1, WINED3D_TSS_BUMPENV_MAT00
), glsl_fragment_pipe_invalidate_constants
}, WINED3D_GL_EXT_NONE
},
6566 {STATE_TEXTURESTAGE(1, WINED3D_TSS_BUMPENV_MAT01
), {STATE_TEXTURESTAGE(1, WINED3D_TSS_BUMPENV_MAT00
), NULL
}, WINED3D_GL_EXT_NONE
},
6567 {STATE_TEXTURESTAGE(1, WINED3D_TSS_BUMPENV_MAT10
), {STATE_TEXTURESTAGE(1, WINED3D_TSS_BUMPENV_MAT00
), NULL
}, WINED3D_GL_EXT_NONE
},
6568 {STATE_TEXTURESTAGE(1, WINED3D_TSS_BUMPENV_MAT11
), {STATE_TEXTURESTAGE(1, WINED3D_TSS_BUMPENV_MAT00
), NULL
}, WINED3D_GL_EXT_NONE
},
6569 {STATE_TEXTURESTAGE(1, WINED3D_TSS_BUMPENV_LSCALE
), {STATE_TEXTURESTAGE(1, WINED3D_TSS_BUMPENV_LSCALE
), glsl_fragment_pipe_invalidate_constants
}, WINED3D_GL_EXT_NONE
},
6570 {STATE_TEXTURESTAGE(1, WINED3D_TSS_BUMPENV_LOFFSET
), {STATE_TEXTURESTAGE(1, WINED3D_TSS_BUMPENV_LSCALE
), NULL
}, WINED3D_GL_EXT_NONE
},
6571 {STATE_TEXTURESTAGE(2, WINED3D_TSS_COLOR_OP
), {STATE_PIXELSHADER
, NULL
}, WINED3D_GL_EXT_NONE
},
6572 {STATE_TEXTURESTAGE(2, WINED3D_TSS_COLOR_ARG1
), {STATE_PIXELSHADER
, NULL
}, WINED3D_GL_EXT_NONE
},
6573 {STATE_TEXTURESTAGE(2, WINED3D_TSS_COLOR_ARG2
), {STATE_PIXELSHADER
, NULL
}, WINED3D_GL_EXT_NONE
},
6574 {STATE_TEXTURESTAGE(2, WINED3D_TSS_COLOR_ARG0
), {STATE_PIXELSHADER
, NULL
}, WINED3D_GL_EXT_NONE
},
6575 {STATE_TEXTURESTAGE(2, WINED3D_TSS_ALPHA_OP
), {STATE_PIXELSHADER
, NULL
}, WINED3D_GL_EXT_NONE
},
6576 {STATE_TEXTURESTAGE(2, WINED3D_TSS_ALPHA_ARG1
), {STATE_PIXELSHADER
, NULL
}, WINED3D_GL_EXT_NONE
},
6577 {STATE_TEXTURESTAGE(2, WINED3D_TSS_ALPHA_ARG2
), {STATE_PIXELSHADER
, NULL
}, WINED3D_GL_EXT_NONE
},
6578 {STATE_TEXTURESTAGE(2, WINED3D_TSS_ALPHA_ARG0
), {STATE_PIXELSHADER
, NULL
}, WINED3D_GL_EXT_NONE
},
6579 {STATE_TEXTURESTAGE(2, WINED3D_TSS_RESULT_ARG
), {STATE_PIXELSHADER
, NULL
}, WINED3D_GL_EXT_NONE
},
6580 {STATE_TEXTURESTAGE(2, WINED3D_TSS_BUMPENV_MAT00
), {STATE_TEXTURESTAGE(2, WINED3D_TSS_BUMPENV_MAT00
), glsl_fragment_pipe_invalidate_constants
}, WINED3D_GL_EXT_NONE
},
6581 {STATE_TEXTURESTAGE(2, WINED3D_TSS_BUMPENV_MAT01
), {STATE_TEXTURESTAGE(2, WINED3D_TSS_BUMPENV_MAT00
), NULL
}, WINED3D_GL_EXT_NONE
},
6582 {STATE_TEXTURESTAGE(2, WINED3D_TSS_BUMPENV_MAT10
), {STATE_TEXTURESTAGE(2, WINED3D_TSS_BUMPENV_MAT00
), NULL
}, WINED3D_GL_EXT_NONE
},
6583 {STATE_TEXTURESTAGE(2, WINED3D_TSS_BUMPENV_MAT11
), {STATE_TEXTURESTAGE(2, WINED3D_TSS_BUMPENV_MAT00
), NULL
}, WINED3D_GL_EXT_NONE
},
6584 {STATE_TEXTURESTAGE(2, WINED3D_TSS_BUMPENV_LSCALE
), {STATE_TEXTURESTAGE(2, WINED3D_TSS_BUMPENV_LSCALE
), glsl_fragment_pipe_invalidate_constants
}, WINED3D_GL_EXT_NONE
},
6585 {STATE_TEXTURESTAGE(2, WINED3D_TSS_BUMPENV_LOFFSET
), {STATE_TEXTURESTAGE(2, WINED3D_TSS_BUMPENV_LSCALE
), NULL
}, WINED3D_GL_EXT_NONE
},
6586 {STATE_TEXTURESTAGE(3, WINED3D_TSS_COLOR_OP
), {STATE_PIXELSHADER
, NULL
}, WINED3D_GL_EXT_NONE
},
6587 {STATE_TEXTURESTAGE(3, WINED3D_TSS_COLOR_ARG1
), {STATE_PIXELSHADER
, NULL
}, WINED3D_GL_EXT_NONE
},
6588 {STATE_TEXTURESTAGE(3, WINED3D_TSS_COLOR_ARG2
), {STATE_PIXELSHADER
, NULL
}, WINED3D_GL_EXT_NONE
},
6589 {STATE_TEXTURESTAGE(3, WINED3D_TSS_COLOR_ARG0
), {STATE_PIXELSHADER
, NULL
}, WINED3D_GL_EXT_NONE
},
6590 {STATE_TEXTURESTAGE(3, WINED3D_TSS_ALPHA_OP
), {STATE_PIXELSHADER
, NULL
}, WINED3D_GL_EXT_NONE
},
6591 {STATE_TEXTURESTAGE(3, WINED3D_TSS_ALPHA_ARG1
), {STATE_PIXELSHADER
, NULL
}, WINED3D_GL_EXT_NONE
},
6592 {STATE_TEXTURESTAGE(3, WINED3D_TSS_ALPHA_ARG2
), {STATE_PIXELSHADER
, NULL
}, WINED3D_GL_EXT_NONE
},
6593 {STATE_TEXTURESTAGE(3, WINED3D_TSS_ALPHA_ARG0
), {STATE_PIXELSHADER
, NULL
}, WINED3D_GL_EXT_NONE
},
6594 {STATE_TEXTURESTAGE(3, WINED3D_TSS_RESULT_ARG
), {STATE_PIXELSHADER
, NULL
}, WINED3D_GL_EXT_NONE
},
6595 {STATE_TEXTURESTAGE(3, WINED3D_TSS_BUMPENV_MAT00
), {STATE_TEXTURESTAGE(3, WINED3D_TSS_BUMPENV_MAT00
), glsl_fragment_pipe_invalidate_constants
}, WINED3D_GL_EXT_NONE
},
6596 {STATE_TEXTURESTAGE(3, WINED3D_TSS_BUMPENV_MAT01
), {STATE_TEXTURESTAGE(3, WINED3D_TSS_BUMPENV_MAT00
), NULL
}, WINED3D_GL_EXT_NONE
},
6597 {STATE_TEXTURESTAGE(3, WINED3D_TSS_BUMPENV_MAT10
), {STATE_TEXTURESTAGE(3, WINED3D_TSS_BUMPENV_MAT00
), NULL
}, WINED3D_GL_EXT_NONE
},
6598 {STATE_TEXTURESTAGE(3, WINED3D_TSS_BUMPENV_MAT11
), {STATE_TEXTURESTAGE(3, WINED3D_TSS_BUMPENV_MAT00
), NULL
}, WINED3D_GL_EXT_NONE
},
6599 {STATE_TEXTURESTAGE(3, WINED3D_TSS_BUMPENV_LSCALE
), {STATE_TEXTURESTAGE(3, WINED3D_TSS_BUMPENV_LSCALE
), glsl_fragment_pipe_invalidate_constants
}, WINED3D_GL_EXT_NONE
},
6600 {STATE_TEXTURESTAGE(3, WINED3D_TSS_BUMPENV_LOFFSET
), {STATE_TEXTURESTAGE(3, WINED3D_TSS_BUMPENV_LSCALE
), NULL
}, WINED3D_GL_EXT_NONE
},
6601 {STATE_TEXTURESTAGE(4, WINED3D_TSS_COLOR_OP
), {STATE_PIXELSHADER
, NULL
}, WINED3D_GL_EXT_NONE
},
6602 {STATE_TEXTURESTAGE(4, WINED3D_TSS_COLOR_ARG1
), {STATE_PIXELSHADER
, NULL
}, WINED3D_GL_EXT_NONE
},
6603 {STATE_TEXTURESTAGE(4, WINED3D_TSS_COLOR_ARG2
), {STATE_PIXELSHADER
, NULL
}, WINED3D_GL_EXT_NONE
},
6604 {STATE_TEXTURESTAGE(4, WINED3D_TSS_COLOR_ARG0
), {STATE_PIXELSHADER
, NULL
}, WINED3D_GL_EXT_NONE
},
6605 {STATE_TEXTURESTAGE(4, WINED3D_TSS_ALPHA_OP
), {STATE_PIXELSHADER
, NULL
}, WINED3D_GL_EXT_NONE
},
6606 {STATE_TEXTURESTAGE(4, WINED3D_TSS_ALPHA_ARG1
), {STATE_PIXELSHADER
, NULL
}, WINED3D_GL_EXT_NONE
},
6607 {STATE_TEXTURESTAGE(4, WINED3D_TSS_ALPHA_ARG2
), {STATE_PIXELSHADER
, NULL
}, WINED3D_GL_EXT_NONE
},
6608 {STATE_TEXTURESTAGE(4, WINED3D_TSS_ALPHA_ARG0
), {STATE_PIXELSHADER
, NULL
}, WINED3D_GL_EXT_NONE
},
6609 {STATE_TEXTURESTAGE(4, WINED3D_TSS_RESULT_ARG
), {STATE_PIXELSHADER
, NULL
}, WINED3D_GL_EXT_NONE
},
6610 {STATE_TEXTURESTAGE(4, WINED3D_TSS_BUMPENV_MAT00
), {STATE_TEXTURESTAGE(4, WINED3D_TSS_BUMPENV_MAT00
), glsl_fragment_pipe_invalidate_constants
}, WINED3D_GL_EXT_NONE
},
6611 {STATE_TEXTURESTAGE(4, WINED3D_TSS_BUMPENV_MAT01
), {STATE_TEXTURESTAGE(4, WINED3D_TSS_BUMPENV_MAT00
), NULL
}, WINED3D_GL_EXT_NONE
},
6612 {STATE_TEXTURESTAGE(4, WINED3D_TSS_BUMPENV_MAT10
), {STATE_TEXTURESTAGE(4, WINED3D_TSS_BUMPENV_MAT00
), NULL
}, WINED3D_GL_EXT_NONE
},
6613 {STATE_TEXTURESTAGE(4, WINED3D_TSS_BUMPENV_MAT11
), {STATE_TEXTURESTAGE(4, WINED3D_TSS_BUMPENV_MAT00
), NULL
}, WINED3D_GL_EXT_NONE
},
6614 {STATE_TEXTURESTAGE(4, WINED3D_TSS_BUMPENV_LSCALE
), {STATE_TEXTURESTAGE(4, WINED3D_TSS_BUMPENV_LSCALE
), glsl_fragment_pipe_invalidate_constants
}, WINED3D_GL_EXT_NONE
},
6615 {STATE_TEXTURESTAGE(4, WINED3D_TSS_BUMPENV_LOFFSET
), {STATE_TEXTURESTAGE(4, WINED3D_TSS_BUMPENV_LSCALE
), NULL
}, WINED3D_GL_EXT_NONE
},
6616 {STATE_TEXTURESTAGE(5, WINED3D_TSS_COLOR_OP
), {STATE_PIXELSHADER
, NULL
}, WINED3D_GL_EXT_NONE
},
6617 {STATE_TEXTURESTAGE(5, WINED3D_TSS_COLOR_ARG1
), {STATE_PIXELSHADER
, NULL
}, WINED3D_GL_EXT_NONE
},
6618 {STATE_TEXTURESTAGE(5, WINED3D_TSS_COLOR_ARG2
), {STATE_PIXELSHADER
, NULL
}, WINED3D_GL_EXT_NONE
},
6619 {STATE_TEXTURESTAGE(5, WINED3D_TSS_COLOR_ARG0
), {STATE_PIXELSHADER
, NULL
}, WINED3D_GL_EXT_NONE
},
6620 {STATE_TEXTURESTAGE(5, WINED3D_TSS_ALPHA_OP
), {STATE_PIXELSHADER
, NULL
}, WINED3D_GL_EXT_NONE
},
6621 {STATE_TEXTURESTAGE(5, WINED3D_TSS_ALPHA_ARG1
), {STATE_PIXELSHADER
, NULL
}, WINED3D_GL_EXT_NONE
},
6622 {STATE_TEXTURESTAGE(5, WINED3D_TSS_ALPHA_ARG2
), {STATE_PIXELSHADER
, NULL
}, WINED3D_GL_EXT_NONE
},
6623 {STATE_TEXTURESTAGE(5, WINED3D_TSS_ALPHA_ARG0
), {STATE_PIXELSHADER
, NULL
}, WINED3D_GL_EXT_NONE
},
6624 {STATE_TEXTURESTAGE(5, WINED3D_TSS_RESULT_ARG
), {STATE_PIXELSHADER
, NULL
}, WINED3D_GL_EXT_NONE
},
6625 {STATE_TEXTURESTAGE(5, WINED3D_TSS_BUMPENV_MAT00
), {STATE_TEXTURESTAGE(5, WINED3D_TSS_BUMPENV_MAT00
), glsl_fragment_pipe_invalidate_constants
}, WINED3D_GL_EXT_NONE
},
6626 {STATE_TEXTURESTAGE(5, WINED3D_TSS_BUMPENV_MAT01
), {STATE_TEXTURESTAGE(5, WINED3D_TSS_BUMPENV_MAT00
), NULL
}, WINED3D_GL_EXT_NONE
},
6627 {STATE_TEXTURESTAGE(5, WINED3D_TSS_BUMPENV_MAT10
), {STATE_TEXTURESTAGE(5, WINED3D_TSS_BUMPENV_MAT00
), NULL
}, WINED3D_GL_EXT_NONE
},
6628 {STATE_TEXTURESTAGE(5, WINED3D_TSS_BUMPENV_MAT11
), {STATE_TEXTURESTAGE(5, WINED3D_TSS_BUMPENV_MAT00
), NULL
}, WINED3D_GL_EXT_NONE
},
6629 {STATE_TEXTURESTAGE(5, WINED3D_TSS_BUMPENV_LSCALE
), {STATE_TEXTURESTAGE(5, WINED3D_TSS_BUMPENV_LSCALE
), glsl_fragment_pipe_invalidate_constants
}, WINED3D_GL_EXT_NONE
},
6630 {STATE_TEXTURESTAGE(5, WINED3D_TSS_BUMPENV_LOFFSET
), {STATE_TEXTURESTAGE(5, WINED3D_TSS_BUMPENV_LSCALE
), NULL
}, WINED3D_GL_EXT_NONE
},
6631 {STATE_TEXTURESTAGE(6, WINED3D_TSS_COLOR_OP
), {STATE_PIXELSHADER
, NULL
}, WINED3D_GL_EXT_NONE
},
6632 {STATE_TEXTURESTAGE(6, WINED3D_TSS_COLOR_ARG1
), {STATE_PIXELSHADER
, NULL
}, WINED3D_GL_EXT_NONE
},
6633 {STATE_TEXTURESTAGE(6, WINED3D_TSS_COLOR_ARG2
), {STATE_PIXELSHADER
, NULL
}, WINED3D_GL_EXT_NONE
},
6634 {STATE_TEXTURESTAGE(6, WINED3D_TSS_COLOR_ARG0
), {STATE_PIXELSHADER
, NULL
}, WINED3D_GL_EXT_NONE
},
6635 {STATE_TEXTURESTAGE(6, WINED3D_TSS_ALPHA_OP
), {STATE_PIXELSHADER
, NULL
}, WINED3D_GL_EXT_NONE
},
6636 {STATE_TEXTURESTAGE(6, WINED3D_TSS_ALPHA_ARG1
), {STATE_PIXELSHADER
, NULL
}, WINED3D_GL_EXT_NONE
},
6637 {STATE_TEXTURESTAGE(6, WINED3D_TSS_ALPHA_ARG2
), {STATE_PIXELSHADER
, NULL
}, WINED3D_GL_EXT_NONE
},
6638 {STATE_TEXTURESTAGE(6, WINED3D_TSS_ALPHA_ARG0
), {STATE_PIXELSHADER
, NULL
}, WINED3D_GL_EXT_NONE
},
6639 {STATE_TEXTURESTAGE(6, WINED3D_TSS_RESULT_ARG
), {STATE_PIXELSHADER
, NULL
}, WINED3D_GL_EXT_NONE
},
6640 {STATE_TEXTURESTAGE(6, WINED3D_TSS_BUMPENV_MAT00
), {STATE_TEXTURESTAGE(6, WINED3D_TSS_BUMPENV_MAT00
), glsl_fragment_pipe_invalidate_constants
}, WINED3D_GL_EXT_NONE
},
6641 {STATE_TEXTURESTAGE(6, WINED3D_TSS_BUMPENV_MAT01
), {STATE_TEXTURESTAGE(6, WINED3D_TSS_BUMPENV_MAT00
), NULL
}, WINED3D_GL_EXT_NONE
},
6642 {STATE_TEXTURESTAGE(6, WINED3D_TSS_BUMPENV_MAT10
), {STATE_TEXTURESTAGE(6, WINED3D_TSS_BUMPENV_MAT00
), NULL
}, WINED3D_GL_EXT_NONE
},
6643 {STATE_TEXTURESTAGE(6, WINED3D_TSS_BUMPENV_MAT11
), {STATE_TEXTURESTAGE(6, WINED3D_TSS_BUMPENV_MAT00
), NULL
}, WINED3D_GL_EXT_NONE
},
6644 {STATE_TEXTURESTAGE(6, WINED3D_TSS_BUMPENV_LSCALE
), {STATE_TEXTURESTAGE(6, WINED3D_TSS_BUMPENV_LSCALE
), glsl_fragment_pipe_invalidate_constants
}, WINED3D_GL_EXT_NONE
},
6645 {STATE_TEXTURESTAGE(6, WINED3D_TSS_BUMPENV_LOFFSET
), {STATE_TEXTURESTAGE(6, WINED3D_TSS_BUMPENV_LSCALE
), NULL
}, WINED3D_GL_EXT_NONE
},
6646 {STATE_TEXTURESTAGE(7, WINED3D_TSS_COLOR_OP
), {STATE_PIXELSHADER
, NULL
}, WINED3D_GL_EXT_NONE
},
6647 {STATE_TEXTURESTAGE(7, WINED3D_TSS_COLOR_ARG1
), {STATE_PIXELSHADER
, NULL
}, WINED3D_GL_EXT_NONE
},
6648 {STATE_TEXTURESTAGE(7, WINED3D_TSS_COLOR_ARG2
), {STATE_PIXELSHADER
, NULL
}, WINED3D_GL_EXT_NONE
},
6649 {STATE_TEXTURESTAGE(7, WINED3D_TSS_COLOR_ARG0
), {STATE_PIXELSHADER
, NULL
}, WINED3D_GL_EXT_NONE
},
6650 {STATE_TEXTURESTAGE(7, WINED3D_TSS_ALPHA_OP
), {STATE_PIXELSHADER
, NULL
}, WINED3D_GL_EXT_NONE
},
6651 {STATE_TEXTURESTAGE(7, WINED3D_TSS_ALPHA_ARG1
), {STATE_PIXELSHADER
, NULL
}, WINED3D_GL_EXT_NONE
},
6652 {STATE_TEXTURESTAGE(7, WINED3D_TSS_ALPHA_ARG2
), {STATE_PIXELSHADER
, NULL
}, WINED3D_GL_EXT_NONE
},
6653 {STATE_TEXTURESTAGE(7, WINED3D_TSS_ALPHA_ARG0
), {STATE_PIXELSHADER
, NULL
}, WINED3D_GL_EXT_NONE
},
6654 {STATE_TEXTURESTAGE(7, WINED3D_TSS_RESULT_ARG
), {STATE_PIXELSHADER
, NULL
}, WINED3D_GL_EXT_NONE
},
6655 {STATE_TEXTURESTAGE(7, WINED3D_TSS_BUMPENV_MAT00
), {STATE_TEXTURESTAGE(7, WINED3D_TSS_BUMPENV_MAT00
), glsl_fragment_pipe_invalidate_constants
}, WINED3D_GL_EXT_NONE
},
6656 {STATE_TEXTURESTAGE(7, WINED3D_TSS_BUMPENV_MAT01
), {STATE_TEXTURESTAGE(7, WINED3D_TSS_BUMPENV_MAT00
), NULL
}, WINED3D_GL_EXT_NONE
},
6657 {STATE_TEXTURESTAGE(7, WINED3D_TSS_BUMPENV_MAT10
), {STATE_TEXTURESTAGE(7, WINED3D_TSS_BUMPENV_MAT00
), NULL
}, WINED3D_GL_EXT_NONE
},
6658 {STATE_TEXTURESTAGE(7, WINED3D_TSS_BUMPENV_MAT11
), {STATE_TEXTURESTAGE(7, WINED3D_TSS_BUMPENV_MAT00
), NULL
}, WINED3D_GL_EXT_NONE
},
6659 {STATE_TEXTURESTAGE(7, WINED3D_TSS_BUMPENV_LSCALE
), {STATE_TEXTURESTAGE(7, WINED3D_TSS_BUMPENV_LSCALE
), glsl_fragment_pipe_invalidate_constants
}, WINED3D_GL_EXT_NONE
},
6660 {STATE_TEXTURESTAGE(7, WINED3D_TSS_BUMPENV_LOFFSET
), {STATE_TEXTURESTAGE(7, WINED3D_TSS_BUMPENV_LSCALE
), NULL
}, WINED3D_GL_EXT_NONE
},
6661 {STATE_PIXELSHADER
, {STATE_PIXELSHADER
, glsl_fragment_pipe_shader
}, WINED3D_GL_EXT_NONE
},
6662 {STATE_RENDER(WINED3D_RS_FOGENABLE
), {STATE_RENDER(WINED3D_RS_FOGENABLE
), glsl_fragment_pipe_fog
}, WINED3D_GL_EXT_NONE
},
6663 {STATE_RENDER(WINED3D_RS_FOGTABLEMODE
), {STATE_RENDER(WINED3D_RS_FOGENABLE
), NULL
}, WINED3D_GL_EXT_NONE
},
6664 {STATE_RENDER(WINED3D_RS_FOGVERTEXMODE
), {STATE_RENDER(WINED3D_RS_FOGENABLE
), NULL
}, WINED3D_GL_EXT_NONE
},
6665 {STATE_RENDER(WINED3D_RS_FOGSTART
), {STATE_RENDER(WINED3D_RS_FOGSTART
), state_fogstartend
}, WINED3D_GL_EXT_NONE
},
6666 {STATE_RENDER(WINED3D_RS_FOGEND
), {STATE_RENDER(WINED3D_RS_FOGSTART
), NULL
}, WINED3D_GL_EXT_NONE
},
6667 {STATE_RENDER(WINED3D_RS_SRGBWRITEENABLE
), {STATE_RENDER(WINED3D_RS_SRGBWRITEENABLE
), state_srgbwrite
}, ARB_FRAMEBUFFER_SRGB
},
6668 {STATE_RENDER(WINED3D_RS_SRGBWRITEENABLE
), {STATE_PIXELSHADER
, NULL
}, WINED3D_GL_EXT_NONE
},
6669 {STATE_RENDER(WINED3D_RS_FOGCOLOR
), {STATE_RENDER(WINED3D_RS_FOGCOLOR
), state_fogcolor
}, WINED3D_GL_EXT_NONE
},
6670 {STATE_RENDER(WINED3D_RS_FOGDENSITY
), {STATE_RENDER(WINED3D_RS_FOGDENSITY
), state_fogdensity
}, WINED3D_GL_EXT_NONE
},
6671 {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
},
6672 {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
},
6673 {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
},
6674 {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
},
6675 {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
},
6676 {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
},
6677 {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
},
6678 {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
},
6679 {STATE_RENDER(WINED3D_RS_SPECULARENABLE
), {STATE_RENDER(WINED3D_RS_SPECULARENABLE
), glsl_fragment_pipe_invalidate_constants
}, WINED3D_GL_EXT_NONE
},
6680 {0 /* Terminate */, {0, 0 }, WINED3D_GL_EXT_NONE
},
6683 const struct fragment_pipeline glsl_fragment_pipe
=
6685 glsl_fragment_pipe_enable
,
6686 glsl_fragment_pipe_get_caps
,
6687 glsl_fragment_pipe_alloc
,
6688 glsl_fragment_pipe_free
,
6689 shader_glsl_color_fixup_supported
,
6690 glsl_fragment_pipe_state_template
,