2 * GLSL pixel and vertex shader implementation
4 * Copyright 2006 Jason Green
5 * Copyright 2006-2007 Henri Verbeet
6 * Copyright 2007-2009, 2013 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"
41 #include "wined3d_private.h"
43 WINE_DEFAULT_DEBUG_CHANNEL(d3d_shader
);
44 WINE_DECLARE_DEBUG_CHANNEL(d3d
);
45 WINE_DECLARE_DEBUG_CHANNEL(winediag
);
47 #define WINED3D_GLSL_SAMPLE_PROJECTED 0x1
48 #define WINED3D_GLSL_SAMPLE_NPOT 0x2
49 #define WINED3D_GLSL_SAMPLE_LOD 0x4
50 #define WINED3D_GLSL_SAMPLE_GRAD 0x8
64 struct glsl_sample_function
68 enum wined3d_data_type data_type
;
73 HEAP_NODE_TRAVERSE_LEFT
,
74 HEAP_NODE_TRAVERSE_RIGHT
,
86 struct constant_entry
*entries
;
88 unsigned int *positions
;
92 /* GLSL shader private data */
93 struct shader_glsl_priv
{
94 struct wined3d_shader_buffer shader_buffer
;
95 struct wine_rb_tree program_lookup
;
96 struct constant_heap vconst_heap
;
97 struct constant_heap pconst_heap
;
99 GLuint depth_blt_program_full
[WINED3D_GL_RES_TYPE_COUNT
];
100 GLuint depth_blt_program_masked
[WINED3D_GL_RES_TYPE_COUNT
];
101 UINT next_constant_version
;
103 const struct wined3d_vertex_pipe_ops
*vertex_pipe
;
104 const struct fragment_pipeline
*fragment_pipe
;
105 struct wine_rb_tree ffp_vertex_shaders
;
106 struct wine_rb_tree ffp_fragment_shaders
;
107 BOOL ffp_proj_control
;
110 struct glsl_vs_program
112 struct list shader_entry
;
114 GLenum vertex_color_clamp
;
115 GLint
*uniform_f_locations
;
116 GLint uniform_i_locations
[MAX_CONST_I
];
117 GLint uniform_b_locations
[MAX_CONST_B
];
118 GLint pos_fixup_location
;
120 GLint modelview_matrix_location
;
121 GLint projection_matrix_location
;
122 GLint normal_matrix_location
;
125 struct glsl_gs_program
127 struct list shader_entry
;
131 struct glsl_ps_program
133 struct list shader_entry
;
135 GLint
*uniform_f_locations
;
136 GLint uniform_i_locations
[MAX_CONST_I
];
137 GLint uniform_b_locations
[MAX_CONST_B
];
138 GLint bumpenv_mat_location
[MAX_TEXTURES
];
139 GLint bumpenv_lum_scale_location
[MAX_TEXTURES
];
140 GLint bumpenv_lum_offset_location
[MAX_TEXTURES
];
141 GLint tss_constant_location
[MAX_TEXTURES
];
142 GLint tex_factor_location
;
143 GLint specular_enable_location
;
144 GLint ycorrection_location
;
145 GLint np2_fixup_location
;
146 GLint color_key_location
;
147 const struct ps_np2fixup_info
*np2_fixup_info
;
150 /* Struct to maintain data about a linked GLSL program */
151 struct glsl_shader_prog_link
153 struct wine_rb_entry program_lookup_entry
;
154 struct glsl_vs_program vs
;
155 struct glsl_gs_program gs
;
156 struct glsl_ps_program ps
;
158 DWORD constant_update_mask
;
159 UINT constant_version
;
162 struct glsl_program_key
169 struct shader_glsl_ctx_priv
{
170 const struct vs_compile_args
*cur_vs_args
;
171 const struct ps_compile_args
*cur_ps_args
;
172 struct ps_np2fixup_info
*cur_np2fixup_info
;
175 struct glsl_context_data
177 struct glsl_shader_prog_link
*glsl_program
;
180 struct glsl_ps_compiled_shader
182 struct ps_compile_args args
;
183 struct ps_np2fixup_info np2fixup
;
187 struct glsl_vs_compiled_shader
189 struct vs_compile_args args
;
193 struct glsl_gs_compiled_shader
198 struct glsl_shader_private
202 struct glsl_vs_compiled_shader
*vs
;
203 struct glsl_gs_compiled_shader
*gs
;
204 struct glsl_ps_compiled_shader
*ps
;
206 UINT num_gl_shaders
, shader_array_size
;
209 struct glsl_ffp_vertex_shader
211 struct wined3d_ffp_vs_desc desc
;
213 struct list linked_programs
;
216 struct glsl_ffp_fragment_shader
218 struct ffp_frag_desc entry
;
220 struct list linked_programs
;
223 struct glsl_ffp_destroy_ctx
225 struct shader_glsl_priv
*priv
;
226 const struct wined3d_gl_info
*gl_info
;
229 static const char *debug_gl_shader_type(GLenum type
)
233 #define WINED3D_TO_STR(u) case u: return #u
234 WINED3D_TO_STR(GL_VERTEX_SHADER
);
235 WINED3D_TO_STR(GL_GEOMETRY_SHADER
);
236 WINED3D_TO_STR(GL_FRAGMENT_SHADER
);
237 #undef WINED3D_TO_STR
239 return wine_dbg_sprintf("UNKNOWN(%#x)", type
);
243 static const char *shader_glsl_get_prefix(enum wined3d_shader_type type
)
247 case WINED3D_SHADER_TYPE_VERTEX
:
250 case WINED3D_SHADER_TYPE_GEOMETRY
:
253 case WINED3D_SHADER_TYPE_PIXEL
:
257 FIXME("Unhandled shader type %#x.\n", type
);
262 static const char *shader_glsl_get_version(const struct wined3d_gl_info
*gl_info
,
263 const struct wined3d_shader_version
*version
)
265 if (gl_info
->glsl_version
>= MAKEDWORD_VERSION(1, 30) && version
->major
>= 4)
266 return "#version 130";
268 return "#version 120";
271 static void shader_glsl_append_imm_vec4(struct wined3d_shader_buffer
*buffer
, const float *values
)
275 wined3d_ftoa(values
[0], str
[0]);
276 wined3d_ftoa(values
[1], str
[1]);
277 wined3d_ftoa(values
[2], str
[2]);
278 wined3d_ftoa(values
[3], str
[3]);
279 shader_addline(buffer
, "vec4(%s, %s, %s, %s)", str
[0], str
[1], str
[2], str
[3]);
282 /* Extract a line from the info log.
283 * Note that this modifies the source string. */
284 static char *get_info_log_line(char **ptr
)
289 if (!(q
= strstr(p
, "\n")))
291 if (!*p
) return NULL
;
301 /* Context activation is done by the caller. */
302 static void print_glsl_info_log(const struct wined3d_gl_info
*gl_info
, GLuint id
, BOOL program
)
307 if (!WARN_ON(d3d_shader
) && !FIXME_ON(d3d_shader
))
311 GL_EXTCALL(glGetProgramiv(id
, GL_INFO_LOG_LENGTH
, &length
));
313 GL_EXTCALL(glGetShaderiv(id
, GL_INFO_LOG_LENGTH
, &length
));
315 /* A size of 1 is just a null-terminated string, so the log should be bigger than
316 * that if there are errors. */
321 log
= HeapAlloc(GetProcessHeap(), 0, length
);
322 /* The info log is supposed to be zero-terminated, but at least some
323 * versions of fglrx don't terminate the string properly. The reported
324 * length does include the terminator, so explicitly set it to zero
328 GL_EXTCALL(glGetProgramInfoLog(id
, length
, NULL
, log
));
330 GL_EXTCALL(glGetShaderInfoLog(id
, length
, NULL
, log
));
333 if (gl_info
->quirks
& WINED3D_QUIRK_INFO_LOG_SPAM
)
335 WARN("Info log received from GLSL shader #%u:\n", id
);
336 while ((line
= get_info_log_line(&ptr
))) WARN(" %s\n", line
);
340 FIXME("Info log received from GLSL shader #%u:\n", id
);
341 while ((line
= get_info_log_line(&ptr
))) FIXME(" %s\n", line
);
343 HeapFree(GetProcessHeap(), 0, log
);
347 /* Context activation is done by the caller. */
348 static void shader_glsl_compile(const struct wined3d_gl_info
*gl_info
, GLuint shader
, const char *src
)
350 TRACE("Compiling shader object %u.\n", shader
);
351 GL_EXTCALL(glShaderSource(shader
, 1, &src
, NULL
));
352 checkGLcall("glShaderSource");
353 GL_EXTCALL(glCompileShader(shader
));
354 checkGLcall("glCompileShader");
355 print_glsl_info_log(gl_info
, shader
, FALSE
);
358 /* Context activation is done by the caller. */
359 static void shader_glsl_dump_program_source(const struct wined3d_gl_info
*gl_info
, GLuint program
)
361 GLint i
, shader_count
, source_size
= -1;
365 GL_EXTCALL(glGetProgramiv(program
, GL_ATTACHED_SHADERS
, &shader_count
));
366 shaders
= HeapAlloc(GetProcessHeap(), 0, shader_count
* sizeof(*shaders
));
369 ERR("Failed to allocate shader array memory.\n");
373 GL_EXTCALL(glGetAttachedShaders(program
, shader_count
, NULL
, shaders
));
374 for (i
= 0; i
< shader_count
; ++i
)
379 GL_EXTCALL(glGetShaderiv(shaders
[i
], GL_SHADER_SOURCE_LENGTH
, &tmp
));
381 if (source_size
< tmp
)
383 HeapFree(GetProcessHeap(), 0, source
);
385 source
= HeapAlloc(GetProcessHeap(), HEAP_ZERO_MEMORY
, tmp
);
388 ERR("Failed to allocate %d bytes for shader source.\n", tmp
);
389 HeapFree(GetProcessHeap(), 0, shaders
);
395 FIXME("Shader %u:\n", shaders
[i
]);
396 GL_EXTCALL(glGetShaderiv(shaders
[i
], GL_SHADER_TYPE
, &tmp
));
397 FIXME(" GL_SHADER_TYPE: %s.\n", debug_gl_shader_type(tmp
));
398 GL_EXTCALL(glGetShaderiv(shaders
[i
], GL_COMPILE_STATUS
, &tmp
));
399 FIXME(" GL_COMPILE_STATUS: %d.\n", tmp
);
403 GL_EXTCALL(glGetShaderSource(shaders
[i
], source_size
, NULL
, source
));
404 while ((line
= get_info_log_line(&ptr
))) FIXME(" %s\n", line
);
408 HeapFree(GetProcessHeap(), 0, source
);
409 HeapFree(GetProcessHeap(), 0, shaders
);
412 /* Context activation is done by the caller. */
413 static void shader_glsl_validate_link(const struct wined3d_gl_info
*gl_info
, GLuint program
)
417 if (!TRACE_ON(d3d_shader
) && !FIXME_ON(d3d_shader
))
420 GL_EXTCALL(glGetProgramiv(program
, GL_LINK_STATUS
, &tmp
));
423 FIXME("Program %u link status invalid.\n", program
);
424 shader_glsl_dump_program_source(gl_info
, program
);
427 print_glsl_info_log(gl_info
, program
, TRUE
);
430 /* Context activation is done by the caller. */
431 static void shader_glsl_load_samplers(const struct wined3d_gl_info
*gl_info
,
432 const DWORD
*tex_unit_map
, GLuint program_id
)
434 unsigned int mapped_unit
;
435 char sampler_name
[20];
442 enum wined3d_shader_type type
;
443 unsigned int base_idx
;
448 {WINED3D_SHADER_TYPE_PIXEL
, 0, MAX_FRAGMENT_SAMPLERS
},
449 {WINED3D_SHADER_TYPE_VERTEX
, MAX_FRAGMENT_SAMPLERS
, MAX_VERTEX_SAMPLERS
},
452 for (i
= 0; i
< ARRAY_SIZE(sampler_info
); ++i
)
454 prefix
= shader_glsl_get_prefix(sampler_info
[i
].type
);
456 for (j
= 0; j
< sampler_info
[i
].count
; ++j
)
458 snprintf(sampler_name
, sizeof(sampler_name
), "%s_sampler%u", prefix
, j
);
459 name_loc
= GL_EXTCALL(glGetUniformLocation(program_id
, sampler_name
));
463 mapped_unit
= tex_unit_map
[sampler_info
[i
].base_idx
+ j
];
464 if (mapped_unit
== WINED3D_UNMAPPED_STAGE
|| mapped_unit
>= gl_info
->limits
.combined_samplers
)
466 ERR("Trying to load sampler %s on unsupported unit %u.\n", sampler_name
, mapped_unit
);
470 TRACE("Loading sampler %s on unit %u.\n", sampler_name
, mapped_unit
);
471 GL_EXTCALL(glUniform1i(name_loc
, mapped_unit
));
474 checkGLcall("glUniform1i");
477 /* Context activation is done by the caller. */
478 static inline void walk_constant_heap(const struct wined3d_gl_info
*gl_info
, const float *constants
,
479 const GLint
*constant_locations
, const struct constant_heap
*heap
, unsigned char *stack
, DWORD version
)
481 unsigned int start
= ~0U, end
= 0;
483 unsigned int heap_idx
= 1;
486 if (heap
->entries
[heap_idx
].version
<= version
) return;
488 idx
= heap
->entries
[heap_idx
].idx
;
489 if (constant_locations
[idx
] != -1)
491 stack
[stack_idx
] = HEAP_NODE_TRAVERSE_LEFT
;
493 while (stack_idx
>= 0)
495 /* Note that we fall through to the next case statement. */
496 switch(stack
[stack_idx
])
498 case HEAP_NODE_TRAVERSE_LEFT
:
500 unsigned int left_idx
= heap_idx
<< 1;
501 if (left_idx
< heap
->size
&& heap
->entries
[left_idx
].version
> version
)
504 idx
= heap
->entries
[heap_idx
].idx
;
505 if (constant_locations
[idx
] != -1)
513 stack
[stack_idx
++] = HEAP_NODE_TRAVERSE_RIGHT
;
514 stack
[stack_idx
] = HEAP_NODE_TRAVERSE_LEFT
;
519 case HEAP_NODE_TRAVERSE_RIGHT
:
521 unsigned int right_idx
= (heap_idx
<< 1) + 1;
522 if (right_idx
< heap
->size
&& heap
->entries
[right_idx
].version
> version
)
524 heap_idx
= right_idx
;
525 idx
= heap
->entries
[heap_idx
].idx
;
526 if (constant_locations
[idx
] != -1)
534 stack
[stack_idx
++] = HEAP_NODE_POP
;
535 stack
[stack_idx
] = HEAP_NODE_TRAVERSE_LEFT
;
547 GL_EXTCALL(glUniform4fv(constant_locations
[start
], end
- start
+ 1, &constants
[start
* 4]));
548 checkGLcall("walk_constant_heap()");
551 /* Context activation is done by the caller. */
552 static inline void apply_clamped_constant(const struct wined3d_gl_info
*gl_info
, GLint location
, const GLfloat
*data
)
554 GLfloat clamped_constant
[4];
556 if (location
== -1) return;
558 clamped_constant
[0] = data
[0] < -1.0f
? -1.0f
: data
[0] > 1.0f
? 1.0f
: data
[0];
559 clamped_constant
[1] = data
[1] < -1.0f
? -1.0f
: data
[1] > 1.0f
? 1.0f
: data
[1];
560 clamped_constant
[2] = data
[2] < -1.0f
? -1.0f
: data
[2] > 1.0f
? 1.0f
: data
[2];
561 clamped_constant
[3] = data
[3] < -1.0f
? -1.0f
: data
[3] > 1.0f
? 1.0f
: data
[3];
563 GL_EXTCALL(glUniform4fv(location
, 1, clamped_constant
));
566 /* Context activation is done by the caller. */
567 static inline void walk_constant_heap_clamped(const struct wined3d_gl_info
*gl_info
, const float *constants
,
568 const GLint
*constant_locations
, const struct constant_heap
*heap
, unsigned char *stack
, DWORD version
)
571 unsigned int heap_idx
= 1;
574 if (heap
->entries
[heap_idx
].version
<= version
) return;
576 idx
= heap
->entries
[heap_idx
].idx
;
577 apply_clamped_constant(gl_info
, constant_locations
[idx
], &constants
[idx
* 4]);
578 stack
[stack_idx
] = HEAP_NODE_TRAVERSE_LEFT
;
580 while (stack_idx
>= 0)
582 /* Note that we fall through to the next case statement. */
583 switch(stack
[stack_idx
])
585 case HEAP_NODE_TRAVERSE_LEFT
:
587 unsigned int left_idx
= heap_idx
<< 1;
588 if (left_idx
< heap
->size
&& heap
->entries
[left_idx
].version
> version
)
591 idx
= heap
->entries
[heap_idx
].idx
;
592 apply_clamped_constant(gl_info
, constant_locations
[idx
], &constants
[idx
* 4]);
594 stack
[stack_idx
++] = HEAP_NODE_TRAVERSE_RIGHT
;
595 stack
[stack_idx
] = HEAP_NODE_TRAVERSE_LEFT
;
600 case HEAP_NODE_TRAVERSE_RIGHT
:
602 unsigned int right_idx
= (heap_idx
<< 1) + 1;
603 if (right_idx
< heap
->size
&& heap
->entries
[right_idx
].version
> version
)
605 heap_idx
= right_idx
;
606 idx
= heap
->entries
[heap_idx
].idx
;
607 apply_clamped_constant(gl_info
, constant_locations
[idx
], &constants
[idx
* 4]);
609 stack
[stack_idx
++] = HEAP_NODE_POP
;
610 stack
[stack_idx
] = HEAP_NODE_TRAVERSE_LEFT
;
621 checkGLcall("walk_constant_heap_clamped()");
624 /* Context activation is done by the caller. */
625 static void shader_glsl_load_constantsF(const struct wined3d_shader
*shader
, const struct wined3d_gl_info
*gl_info
,
626 const float *constants
, const GLint
*constant_locations
, const struct constant_heap
*heap
,
627 unsigned char *stack
, UINT version
)
629 const struct wined3d_shader_lconst
*lconst
;
631 /* 1.X pshaders have the constants clamped to [-1;1] implicitly. */
632 if (shader
->reg_maps
.shader_version
.major
== 1
633 && shader
->reg_maps
.shader_version
.type
== WINED3D_SHADER_TYPE_PIXEL
)
634 walk_constant_heap_clamped(gl_info
, constants
, constant_locations
, heap
, stack
, version
);
636 walk_constant_heap(gl_info
, constants
, constant_locations
, heap
, stack
, version
);
638 if (!shader
->load_local_constsF
)
640 TRACE("No need to load local float constants for this shader\n");
644 /* Immediate constants are clamped to [-1;1] at shader creation time if needed */
645 LIST_FOR_EACH_ENTRY(lconst
, &shader
->constantsF
, struct wined3d_shader_lconst
, entry
)
647 GL_EXTCALL(glUniform4fv(constant_locations
[lconst
->idx
], 1, (const GLfloat
*)lconst
->value
));
649 checkGLcall("glUniform4fv()");
652 /* Context activation is done by the caller. */
653 static void shader_glsl_load_constantsI(const struct wined3d_shader
*shader
, const struct wined3d_gl_info
*gl_info
,
654 const GLint locations
[MAX_CONST_I
], const int *constants
, WORD constants_set
)
659 for (i
= 0; constants_set
; constants_set
>>= 1, ++i
)
661 if (!(constants_set
& 1)) continue;
663 /* We found this uniform name in the program - go ahead and send the data */
664 GL_EXTCALL(glUniform4iv(locations
[i
], 1, &constants
[i
* 4]));
667 /* Load immediate constants */
668 ptr
= list_head(&shader
->constantsI
);
671 const struct wined3d_shader_lconst
*lconst
= LIST_ENTRY(ptr
, const struct wined3d_shader_lconst
, entry
);
672 unsigned int idx
= lconst
->idx
;
673 const GLint
*values
= (const GLint
*)lconst
->value
;
675 /* We found this uniform name in the program - go ahead and send the data */
676 GL_EXTCALL(glUniform4iv(locations
[idx
], 1, values
));
677 ptr
= list_next(&shader
->constantsI
, ptr
);
679 checkGLcall("glUniform4iv()");
682 /* Context activation is done by the caller. */
683 static void shader_glsl_load_constantsB(const struct wined3d_shader
*shader
, const struct wined3d_gl_info
*gl_info
,
684 const GLint locations
[MAX_CONST_B
], const BOOL
*constants
, WORD constants_set
)
689 for (i
= 0; constants_set
; constants_set
>>= 1, ++i
)
691 if (!(constants_set
& 1)) continue;
693 GL_EXTCALL(glUniform1iv(locations
[i
], 1, &constants
[i
]));
696 /* Load immediate constants */
697 ptr
= list_head(&shader
->constantsB
);
700 const struct wined3d_shader_lconst
*lconst
= LIST_ENTRY(ptr
, const struct wined3d_shader_lconst
, entry
);
701 unsigned int idx
= lconst
->idx
;
702 const GLint
*values
= (const GLint
*)lconst
->value
;
704 GL_EXTCALL(glUniform1iv(locations
[idx
], 1, values
));
705 ptr
= list_next(&shader
->constantsB
, ptr
);
707 checkGLcall("glUniform1iv()");
710 static void reset_program_constant_version(struct wine_rb_entry
*entry
, void *context
)
712 WINE_RB_ENTRY_VALUE(entry
, struct glsl_shader_prog_link
, program_lookup_entry
)->constant_version
= 0;
715 /* Context activation is done by the caller (state handler). */
716 static void shader_glsl_load_np2fixup_constants(const struct glsl_ps_program
*ps
,
717 const struct wined3d_gl_info
*gl_info
, const struct wined3d_state
*state
)
719 GLfloat np2fixup_constants
[4 * MAX_FRAGMENT_SAMPLERS
];
720 UINT fixup
= ps
->np2_fixup_info
->active
;
723 for (i
= 0; fixup
; fixup
>>= 1, ++i
)
725 const struct wined3d_texture
*tex
= state
->textures
[i
];
726 unsigned char idx
= ps
->np2_fixup_info
->idx
[i
];
727 GLfloat
*tex_dim
= &np2fixup_constants
[(idx
>> 1) * 4];
731 ERR("Nonexistent texture is flagged for NP2 texcoord fixup.\n");
737 tex_dim
[2] = tex
->pow2_matrix
[0];
738 tex_dim
[3] = tex
->pow2_matrix
[5];
742 tex_dim
[0] = tex
->pow2_matrix
[0];
743 tex_dim
[1] = tex
->pow2_matrix
[5];
747 GL_EXTCALL(glUniform4fv(ps
->np2_fixup_location
, ps
->np2_fixup_info
->num_consts
, np2fixup_constants
));
750 /* Taken and adapted from Mesa. */
751 static BOOL
invert_matrix_3d(struct wined3d_matrix
*out
, const struct wined3d_matrix
*in
)
753 float pos
, neg
, t
, det
;
754 struct wined3d_matrix temp
;
756 /* Calculate the determinant of upper left 3x3 submatrix and
757 * determine if the matrix is singular. */
759 t
= in
->_11
* in
->_22
* in
->_33
;
765 t
= in
->_21
* in
->_32
* in
->_13
;
770 t
= in
->_31
* in
->_12
* in
->_23
;
776 t
= -in
->_31
* in
->_22
* in
->_13
;
781 t
= -in
->_21
* in
->_12
* in
->_33
;
787 t
= -in
->_11
* in
->_32
* in
->_23
;
795 if (fabsf(det
) < 1e-25f
)
799 temp
._11
= (in
->_22
* in
->_33
- in
->_32
* in
->_23
) * det
;
800 temp
._12
= -(in
->_12
* in
->_33
- in
->_32
* in
->_13
) * det
;
801 temp
._13
= (in
->_12
* in
->_23
- in
->_22
* in
->_13
) * det
;
802 temp
._21
= -(in
->_21
* in
->_33
- in
->_31
* in
->_23
) * det
;
803 temp
._22
= (in
->_11
* in
->_33
- in
->_31
* in
->_13
) * det
;
804 temp
._23
= -(in
->_11
* in
->_23
- in
->_21
* in
->_13
) * det
;
805 temp
._31
= (in
->_21
* in
->_32
- in
->_31
* in
->_22
) * det
;
806 temp
._32
= -(in
->_11
* in
->_32
- in
->_31
* in
->_12
) * det
;
807 temp
._33
= (in
->_11
* in
->_22
- in
->_21
* in
->_12
) * det
;
813 static void shader_glsl_ffp_vertex_normalmatrix_uniform(const struct wined3d_context
*context
,
814 const struct wined3d_state
*state
, struct glsl_shader_prog_link
*prog
)
816 const struct wined3d_gl_info
*gl_info
= context
->gl_info
;
818 struct wined3d_matrix mv
;
821 /* gl_NormalMatrix is defined in the OpenGL spec as "transpose of the
822 * inverse of the upper leftmost 3x3 of gl_ModelViewMatrix" and that
823 * seems to be correct for D3D too. */
824 get_modelview_matrix(context
, state
, &mv
);
825 invert_matrix_3d(&mv
, &mv
);
826 /* Tests show that singular modelview matrices are used unchanged as normal
827 * matrices on D3D3 and older. There seems to be no clearly consistent
828 * behavior on newer D3D versions so always follow older ddraw behavior. */
829 for (i
= 0; i
< 3; ++i
)
830 for (j
= 0; j
< 3; ++j
)
831 mat
[i
* 3 + j
] = (&mv
._11
)[j
* 4 + i
];
833 GL_EXTCALL(glUniformMatrix3fv(prog
->vs
.normal_matrix_location
, 1, FALSE
, mat
));
834 checkGLcall("glUniformMatrix3fv");
837 /* Context activation is done by the caller (state handler). */
838 static void shader_glsl_load_color_key_constant(const struct glsl_ps_program
*ps
,
839 const struct wined3d_gl_info
*gl_info
, const struct wined3d_state
*state
)
841 struct wined3d_color float_key
;
842 const struct wined3d_texture
*texture
= state
->textures
[0];
844 wined3d_format_convert_color_to_float(texture
->resource
.format
, NULL
,
845 texture
->async
.src_blt_color_key
.color_space_high_value
, &float_key
);
846 GL_EXTCALL(glUniform4fv(ps
->color_key_location
, 1, &float_key
.r
));
849 /* Context activation is done by the caller (state handler). */
850 static void shader_glsl_load_constants(void *shader_priv
, struct wined3d_context
*context
,
851 const struct wined3d_state
*state
)
853 const struct glsl_context_data
*ctx_data
= context
->shader_backend_data
;
854 const struct wined3d_shader
*vshader
= state
->shader
[WINED3D_SHADER_TYPE_VERTEX
];
855 const struct wined3d_shader
*pshader
= state
->shader
[WINED3D_SHADER_TYPE_PIXEL
];
856 const struct wined3d_gl_info
*gl_info
= context
->gl_info
;
857 struct shader_glsl_priv
*priv
= shader_priv
;
858 float position_fixup
[4];
859 DWORD update_mask
= 0;
861 struct glsl_shader_prog_link
*prog
= ctx_data
->glsl_program
;
862 UINT constant_version
;
866 /* No GLSL program set - nothing to do. */
869 constant_version
= prog
->constant_version
;
870 update_mask
= context
->constant_update_mask
& prog
->constant_update_mask
;
872 if (update_mask
& WINED3D_SHADER_CONST_VS_F
)
873 shader_glsl_load_constantsF(vshader
, gl_info
, state
->vs_consts_f
,
874 prog
->vs
.uniform_f_locations
, &priv
->vconst_heap
, priv
->stack
, constant_version
);
876 if (update_mask
& WINED3D_SHADER_CONST_VS_I
)
877 shader_glsl_load_constantsI(vshader
, gl_info
, prog
->vs
.uniform_i_locations
, state
->vs_consts_i
,
878 vshader
->reg_maps
.integer_constants
);
880 if (update_mask
& WINED3D_SHADER_CONST_VS_B
)
881 shader_glsl_load_constantsB(vshader
, gl_info
, prog
->vs
.uniform_b_locations
, state
->vs_consts_b
,
882 vshader
->reg_maps
.boolean_constants
);
884 if (update_mask
& WINED3D_SHADER_CONST_VS_POS_FIXUP
)
886 shader_get_position_fixup(context
, state
, position_fixup
);
887 GL_EXTCALL(glUniform4fv(prog
->vs
.pos_fixup_location
, 1, position_fixup
));
888 checkGLcall("glUniform4fv");
891 if (update_mask
& WINED3D_SHADER_CONST_FFP_MODELVIEW
)
893 struct wined3d_matrix mat
;
895 get_modelview_matrix(context
, state
, &mat
);
896 GL_EXTCALL(glUniformMatrix4fv(prog
->vs
.modelview_matrix_location
, 1, FALSE
, &mat
._11
));
897 checkGLcall("glUniformMatrix4fv");
899 shader_glsl_ffp_vertex_normalmatrix_uniform(context
, state
, prog
);
902 if (update_mask
& WINED3D_SHADER_CONST_FFP_PROJ
)
904 struct wined3d_matrix projection
;
906 get_projection_matrix(context
, state
, &projection
);
907 GL_EXTCALL(glUniformMatrix4fv(prog
->vs
.projection_matrix_location
, 1, FALSE
, &projection
._11
));
908 checkGLcall("glUniformMatrix4fv");
911 if (update_mask
& WINED3D_SHADER_CONST_PS_F
)
912 shader_glsl_load_constantsF(pshader
, gl_info
, state
->ps_consts_f
,
913 prog
->ps
.uniform_f_locations
, &priv
->pconst_heap
, priv
->stack
, constant_version
);
915 if (update_mask
& WINED3D_SHADER_CONST_PS_I
)
916 shader_glsl_load_constantsI(pshader
, gl_info
, prog
->ps
.uniform_i_locations
, state
->ps_consts_i
,
917 pshader
->reg_maps
.integer_constants
);
919 if (update_mask
& WINED3D_SHADER_CONST_PS_B
)
920 shader_glsl_load_constantsB(pshader
, gl_info
, prog
->ps
.uniform_b_locations
, state
->ps_consts_b
,
921 pshader
->reg_maps
.boolean_constants
);
923 if (update_mask
& WINED3D_SHADER_CONST_PS_BUMP_ENV
)
925 for (i
= 0; i
< MAX_TEXTURES
; ++i
)
927 if (prog
->ps
.bumpenv_mat_location
[i
] == -1)
930 GL_EXTCALL(glUniformMatrix2fv(prog
->ps
.bumpenv_mat_location
[i
], 1, 0,
931 (const GLfloat
*)&state
->texture_states
[i
][WINED3D_TSS_BUMPENV_MAT00
]));
933 if (prog
->ps
.bumpenv_lum_scale_location
[i
] != -1)
935 GL_EXTCALL(glUniform1fv(prog
->ps
.bumpenv_lum_scale_location
[i
], 1,
936 (const GLfloat
*)&state
->texture_states
[i
][WINED3D_TSS_BUMPENV_LSCALE
]));
937 GL_EXTCALL(glUniform1fv(prog
->ps
.bumpenv_lum_offset_location
[i
], 1,
938 (const GLfloat
*)&state
->texture_states
[i
][WINED3D_TSS_BUMPENV_LOFFSET
]));
942 checkGLcall("bump env uniforms");
945 if (update_mask
& WINED3D_SHADER_CONST_PS_Y_CORR
)
947 float correction_params
[4];
949 if (context
->render_offscreen
)
951 correction_params
[0] = 0.0f
;
952 correction_params
[1] = 1.0f
;
954 /* position is window relative, not viewport relative */
955 correction_params
[0] = (float) context
->current_rt
->resource
.height
;
956 correction_params
[1] = -1.0f
;
958 GL_EXTCALL(glUniform4fv(prog
->ps
.ycorrection_location
, 1, correction_params
));
961 if (update_mask
& WINED3D_SHADER_CONST_PS_NP2_FIXUP
)
962 shader_glsl_load_np2fixup_constants(&prog
->ps
, gl_info
, state
);
963 if (update_mask
& WINED3D_SHADER_CONST_FFP_COLOR_KEY
)
964 shader_glsl_load_color_key_constant(&prog
->ps
, gl_info
, state
);
966 if (update_mask
& WINED3D_SHADER_CONST_FFP_PS
)
970 if (prog
->ps
.tex_factor_location
!= -1)
972 D3DCOLORTOGLFLOAT4(state
->render_states
[WINED3D_RS_TEXTUREFACTOR
], col
);
973 GL_EXTCALL(glUniform4fv(prog
->ps
.tex_factor_location
, 1, col
));
976 if (state
->render_states
[WINED3D_RS_SPECULARENABLE
])
977 GL_EXTCALL(glUniform4f(prog
->ps
.specular_enable_location
, 1.0f
, 1.0f
, 1.0f
, 0.0f
));
979 GL_EXTCALL(glUniform4f(prog
->ps
.specular_enable_location
, 0.0f
, 0.0f
, 0.0f
, 0.0f
));
981 for (i
= 0; i
< MAX_TEXTURES
; ++i
)
983 if (prog
->ps
.tss_constant_location
[i
] == -1)
986 D3DCOLORTOGLFLOAT4(state
->texture_states
[i
][WINED3D_TSS_CONSTANT
], col
);
987 GL_EXTCALL(glUniform4fv(prog
->ps
.tss_constant_location
[i
], 1, col
));
990 checkGLcall("fixed function uniforms");
993 if (priv
->next_constant_version
== UINT_MAX
)
995 TRACE("Max constant version reached, resetting to 0.\n");
996 wine_rb_for_each_entry(&priv
->program_lookup
, reset_program_constant_version
, NULL
);
997 priv
->next_constant_version
= 1;
1001 prog
->constant_version
= priv
->next_constant_version
++;
1005 static void update_heap_entry(struct constant_heap
*heap
, unsigned int idx
, DWORD new_version
)
1007 struct constant_entry
*entries
= heap
->entries
;
1008 unsigned int *positions
= heap
->positions
;
1009 unsigned int heap_idx
, parent_idx
;
1011 if (!heap
->contained
[idx
])
1013 heap_idx
= heap
->size
++;
1014 heap
->contained
[idx
] = TRUE
;
1018 heap_idx
= positions
[idx
];
1021 while (heap_idx
> 1)
1023 parent_idx
= heap_idx
>> 1;
1025 if (new_version
<= entries
[parent_idx
].version
) break;
1027 entries
[heap_idx
] = entries
[parent_idx
];
1028 positions
[entries
[parent_idx
].idx
] = heap_idx
;
1029 heap_idx
= parent_idx
;
1032 entries
[heap_idx
].version
= new_version
;
1033 entries
[heap_idx
].idx
= idx
;
1034 positions
[idx
] = heap_idx
;
1037 static void shader_glsl_update_float_vertex_constants(struct wined3d_device
*device
, UINT start
, UINT count
)
1039 struct shader_glsl_priv
*priv
= device
->shader_priv
;
1040 struct constant_heap
*heap
= &priv
->vconst_heap
;
1043 for (i
= start
; i
< count
+ start
; ++i
)
1045 update_heap_entry(heap
, i
, priv
->next_constant_version
);
1048 for (i
= 0; i
< device
->context_count
; ++i
)
1050 device
->contexts
[i
]->constant_update_mask
|= WINED3D_SHADER_CONST_VS_F
;
1054 static void shader_glsl_update_float_pixel_constants(struct wined3d_device
*device
, UINT start
, UINT count
)
1056 struct shader_glsl_priv
*priv
= device
->shader_priv
;
1057 struct constant_heap
*heap
= &priv
->pconst_heap
;
1060 for (i
= start
; i
< count
+ start
; ++i
)
1062 update_heap_entry(heap
, i
, priv
->next_constant_version
);
1065 for (i
= 0; i
< device
->context_count
; ++i
)
1067 device
->contexts
[i
]->constant_update_mask
|= WINED3D_SHADER_CONST_PS_F
;
1071 static unsigned int vec4_varyings(DWORD shader_major
, const struct wined3d_gl_info
*gl_info
)
1073 unsigned int ret
= gl_info
->limits
.glsl_varyings
/ 4;
1074 /* 4.0 shaders do not write clip coords because d3d10 does not support user clipplanes */
1075 if(shader_major
> 3) return ret
;
1077 /* 3.0 shaders may need an extra varying for the clip coord on some cards(mostly dx10 ones) */
1078 if (gl_info
->quirks
& WINED3D_QUIRK_GLSL_CLIP_VARYING
) ret
-= 1;
1082 /** Generate the variable & register declarations for the GLSL output target */
1083 static void shader_generate_glsl_declarations(const struct wined3d_context
*context
,
1084 struct wined3d_shader_buffer
*buffer
, const struct wined3d_shader
*shader
,
1085 const struct wined3d_shader_reg_maps
*reg_maps
, const struct shader_glsl_ctx_priv
*ctx_priv
)
1087 const struct wined3d_shader_version
*version
= ®_maps
->shader_version
;
1088 const struct wined3d_state
*state
= &shader
->device
->state
;
1089 const struct ps_compile_args
*ps_args
= ctx_priv
->cur_ps_args
;
1090 const struct wined3d_gl_info
*gl_info
= context
->gl_info
;
1091 const struct wined3d_fb_state
*fb
= &shader
->device
->fb
;
1092 unsigned int i
, extra_constants_needed
= 0;
1093 const struct wined3d_shader_lconst
*lconst
;
1097 prefix
= shader_glsl_get_prefix(version
->type
);
1099 /* Prototype the subroutines */
1100 for (i
= 0, map
= reg_maps
->labels
; map
; map
>>= 1, ++i
)
1102 if (map
& 1) shader_addline(buffer
, "void subroutine%u();\n", i
);
1105 /* Declare the constants (aka uniforms) */
1106 if (shader
->limits
->constant_float
> 0)
1108 unsigned max_constantsF
;
1110 /* Unless the shader uses indirect addressing, always declare the
1111 * maximum array size and ignore that we need some uniforms privately.
1112 * E.g. if GL supports 256 uniforms, and we need 2 for the pos fixup
1113 * and immediate values, still declare VC[256]. If the shader needs
1114 * more uniforms than we have it won't work in any case. If it uses
1115 * less, the compiler will figure out which uniforms are really used
1116 * and strip them out. This allows a shader to use c255 on a dx9 card,
1117 * as long as it doesn't also use all the other constants.
1119 * If the shader uses indirect addressing the compiler must assume
1120 * that all declared uniforms are used. In this case, declare only the
1121 * amount that we're assured to have.
1123 * Thus we run into problems in these two cases:
1124 * 1) The shader really uses more uniforms than supported.
1125 * 2) The shader uses indirect addressing, less constants than
1126 * supported, but uses a constant index > #supported consts. */
1127 if (version
->type
== WINED3D_SHADER_TYPE_PIXEL
)
1129 /* No indirect addressing here. */
1130 max_constantsF
= gl_info
->limits
.glsl_ps_float_constants
;
1134 if (reg_maps
->usesrelconstF
)
1136 /* Subtract the other potential uniforms from the max
1137 * available (bools, ints, and 1 row of projection matrix).
1138 * Subtract another uniform for immediate values, which have
1139 * to be loaded via uniform by the driver as well. The shader
1140 * code only uses 0.5, 2.0, 1.0, 128 and -128 in vertex
1141 * shader code, so one vec4 should be enough. (Unfortunately
1142 * the Nvidia driver doesn't store 128 and -128 in one float).
1144 * Writing gl_ClipVertex requires one uniform for each
1145 * clipplane as well. */
1146 max_constantsF
= gl_info
->limits
.glsl_vs_float_constants
- 3;
1147 if(ctx_priv
->cur_vs_args
->clip_enabled
)
1149 max_constantsF
-= gl_info
->limits
.clipplanes
;
1151 max_constantsF
-= count_bits(reg_maps
->integer_constants
);
1152 /* Strictly speaking a bool only uses one scalar, but the nvidia(Linux) compiler doesn't pack them properly,
1153 * so each scalar requires a full vec4. We could work around this by packing the booleans ourselves, but
1154 * for now take this into account when calculating the number of available constants
1156 max_constantsF
-= count_bits(reg_maps
->boolean_constants
);
1157 /* Set by driver quirks in directx.c */
1158 max_constantsF
-= gl_info
->reserved_glsl_constants
;
1160 if (max_constantsF
< shader
->limits
->constant_float
)
1162 static unsigned int once
;
1165 ERR_(winediag
)("The hardware does not support enough uniform components to run this shader,"
1166 " it may not render correctly.\n");
1168 WARN("The hardware does not support enough uniform components to run this shader.\n");
1173 max_constantsF
= gl_info
->limits
.glsl_vs_float_constants
;
1176 max_constantsF
= min(shader
->limits
->constant_float
, max_constantsF
);
1177 shader_addline(buffer
, "uniform vec4 %s_c[%u];\n", prefix
, max_constantsF
);
1180 /* Always declare the full set of constants, the compiler can remove the
1181 * unused ones because d3d doesn't (yet) support indirect int and bool
1182 * constant addressing. This avoids problems if the app uses e.g. i0 and i9. */
1183 if (shader
->limits
->constant_int
> 0 && reg_maps
->integer_constants
)
1184 shader_addline(buffer
, "uniform ivec4 %s_i[%u];\n", prefix
, shader
->limits
->constant_int
);
1186 if (shader
->limits
->constant_bool
> 0 && reg_maps
->boolean_constants
)
1187 shader_addline(buffer
, "uniform bool %s_b[%u];\n", prefix
, shader
->limits
->constant_bool
);
1189 for (i
= 0; i
< WINED3D_MAX_CBS
; ++i
)
1191 if (reg_maps
->cb_sizes
[i
])
1192 shader_addline(buffer
, "layout(std140) uniform block_%s_cb%u { vec4 %s_cb%u[%u]; };\n",
1193 prefix
, i
, prefix
, i
, reg_maps
->cb_sizes
[i
]);
1196 /* Declare texture samplers */
1197 for (i
= 0; i
< reg_maps
->sampler_map
.count
; ++i
)
1199 struct wined3d_shader_sampler_map_entry
*entry
;
1200 BOOL shadow_sampler
, tex_rect
;
1201 const char *sampler_type
;
1203 entry
= ®_maps
->sampler_map
.entries
[i
];
1205 if (entry
->resource_idx
>= ARRAY_SIZE(reg_maps
->resource_info
))
1207 ERR("Invalid resource index %u.\n", entry
->resource_idx
);
1211 shadow_sampler
= version
->type
== WINED3D_SHADER_TYPE_PIXEL
&& (ps_args
->shadow
& (1 << entry
->sampler_idx
));
1212 switch (reg_maps
->resource_info
[entry
->resource_idx
].type
)
1214 case WINED3D_SHADER_RESOURCE_TEXTURE_1D
:
1216 sampler_type
= "sampler1DShadow";
1218 sampler_type
= "sampler1D";
1221 case WINED3D_SHADER_RESOURCE_TEXTURE_2D
:
1222 tex_rect
= version
->type
== WINED3D_SHADER_TYPE_PIXEL
1223 && (ps_args
->np2_fixup
& (1 << entry
->resource_idx
))
1224 && gl_info
->supported
[ARB_TEXTURE_RECTANGLE
];
1228 sampler_type
= "sampler2DRectShadow";
1230 sampler_type
= "sampler2DShadow";
1235 sampler_type
= "sampler2DRect";
1237 sampler_type
= "sampler2D";
1241 case WINED3D_SHADER_RESOURCE_TEXTURE_3D
:
1243 FIXME("Unsupported 3D shadow sampler.\n");
1244 sampler_type
= "sampler3D";
1247 case WINED3D_SHADER_RESOURCE_TEXTURE_CUBE
:
1249 FIXME("Unsupported Cube shadow sampler.\n");
1250 sampler_type
= "samplerCube";
1254 sampler_type
= "unsupported_sampler";
1255 FIXME("Unhandled resource type %#x.\n", reg_maps
->resource_info
[i
].type
);
1258 shader_addline(buffer
, "uniform %s %s_sampler%u;\n", sampler_type
, prefix
, entry
->bind_idx
);
1261 /* Declare uniforms for NP2 texcoord fixup:
1262 * This is NOT done inside the loop that declares the texture samplers
1263 * since the NP2 fixup code is currently only used for the GeforceFX
1264 * series and when forcing the ARB_npot extension off. Modern cards just
1265 * skip the code anyway, so put it inside a separate loop. */
1266 if (version
->type
== WINED3D_SHADER_TYPE_PIXEL
&& ps_args
->np2_fixup
)
1268 struct ps_np2fixup_info
*fixup
= ctx_priv
->cur_np2fixup_info
;
1271 /* NP2/RECT textures in OpenGL use texcoords in the range [0,width]x[0,height]
1272 * while D3D has them in the (normalized) [0,1]x[0,1] range.
1273 * samplerNP2Fixup stores texture dimensions and is updated through
1274 * shader_glsl_load_np2fixup_constants when the sampler changes. */
1276 for (i
= 0; i
< shader
->limits
->sampler
; ++i
)
1278 if (!reg_maps
->resource_info
[i
].type
|| !(ps_args
->np2_fixup
& (1 << i
)))
1281 if (reg_maps
->resource_info
[i
].type
!= WINED3D_SHADER_RESOURCE_TEXTURE_2D
)
1283 FIXME("Non-2D texture is flagged for NP2 texcoord fixup.\n");
1287 fixup
->idx
[i
] = cur
++;
1290 fixup
->num_consts
= (cur
+ 1) >> 1;
1291 fixup
->active
= ps_args
->np2_fixup
;
1292 shader_addline(buffer
, "uniform vec4 %s_samplerNP2Fixup[%u];\n", prefix
, fixup
->num_consts
);
1295 /* Declare address variables */
1296 for (i
= 0, map
= reg_maps
->address
; map
; map
>>= 1, ++i
)
1298 if (map
& 1) shader_addline(buffer
, "ivec4 A%u;\n", i
);
1301 /* Declare texture coordinate temporaries and initialize them */
1302 for (i
= 0, map
= reg_maps
->texcoord
; map
; map
>>= 1, ++i
)
1304 if (map
& 1) shader_addline(buffer
, "vec4 T%u = gl_TexCoord[%u];\n", i
, i
);
1307 if (version
->type
== WINED3D_SHADER_TYPE_VERTEX
)
1309 for (i
= 0; i
< shader
->input_signature
.element_count
; ++i
)
1311 const struct wined3d_shader_signature_element
*e
= &shader
->input_signature
.elements
[i
];
1312 if (e
->sysval_semantic
== WINED3D_SV_INSTANCEID
)
1313 shader_addline(buffer
, "vec4 %s_in%u = vec4(intBitsToFloat(gl_InstanceID), 0.0, 0.0, 0.0);\n",
1314 prefix
, e
->register_idx
);
1316 shader_addline(buffer
, "attribute vec4 %s_in%u;\n", prefix
, e
->register_idx
);
1319 shader_addline(buffer
, "uniform vec4 posFixup;\n");
1320 shader_addline(buffer
, "void order_ps_input(in vec4[%u]);\n", shader
->limits
->packed_output
);
1322 else if (version
->type
== WINED3D_SHADER_TYPE_GEOMETRY
)
1324 shader_addline(buffer
, "varying in vec4 gs_in[][%u];\n", shader
->limits
->packed_input
);
1326 else if (version
->type
== WINED3D_SHADER_TYPE_PIXEL
)
1328 if (version
->major
>= 3)
1330 UINT in_count
= min(vec4_varyings(version
->major
, gl_info
), shader
->limits
->packed_input
);
1333 shader_addline(buffer
, "varying vec4 %s_link[%u];\n", prefix
, in_count
);
1334 shader_addline(buffer
, "vec4 %s_in[%u];\n", prefix
, in_count
);
1337 for (i
= 0, map
= reg_maps
->bumpmat
; map
; map
>>= 1, ++i
)
1342 shader_addline(buffer
, "uniform mat2 bumpenv_mat%u;\n", i
);
1344 if (reg_maps
->luminanceparams
& (1 << i
))
1346 shader_addline(buffer
, "uniform float bumpenv_lum_scale%u;\n", i
);
1347 shader_addline(buffer
, "uniform float bumpenv_lum_offset%u;\n", i
);
1348 extra_constants_needed
++;
1351 extra_constants_needed
++;
1354 if (ps_args
->srgb_correction
)
1356 shader_addline(buffer
, "const vec4 srgb_const0 = ");
1357 shader_glsl_append_imm_vec4(buffer
, wined3d_srgb_const0
);
1358 shader_addline(buffer
, ";\n");
1359 shader_addline(buffer
, "const vec4 srgb_const1 = ");
1360 shader_glsl_append_imm_vec4(buffer
, wined3d_srgb_const1
);
1361 shader_addline(buffer
, ";\n");
1363 if (reg_maps
->vpos
|| reg_maps
->usesdsy
)
1365 if (shader
->limits
->constant_float
+ extra_constants_needed
1366 + 1 < gl_info
->limits
.glsl_ps_float_constants
)
1368 shader_addline(buffer
, "uniform vec4 ycorrection;\n");
1369 extra_constants_needed
++;
1373 float ycorrection
[] =
1375 context
->render_offscreen
? 0.0f
: fb
->render_targets
[0]->height
,
1376 context
->render_offscreen
? 1.0f
: -1.0f
,
1381 /* This happens because we do not have proper tracking of the
1382 * constant registers that are actually used, only the max
1383 * limit of the shader version. */
1384 FIXME("Cannot find a free uniform for vpos correction params\n");
1385 shader_addline(buffer
, "const vec4 ycorrection = ");
1386 shader_glsl_append_imm_vec4(buffer
, ycorrection
);
1387 shader_addline(buffer
, ";\n");
1389 shader_addline(buffer
, "vec4 vpos;\n");
1393 /* Declare output register temporaries */
1394 if (shader
->limits
->packed_output
)
1395 shader_addline(buffer
, "vec4 %s_out[%u];\n", prefix
, shader
->limits
->packed_output
);
1397 /* Declare temporary variables */
1398 for (i
= 0, map
= reg_maps
->temporary
; map
; map
>>= 1, ++i
)
1400 if (map
& 1) shader_addline(buffer
, "vec4 R%u;\n", i
);
1403 /* Declare loop registers aLx */
1404 if (version
->major
< 4)
1406 for (i
= 0; i
< reg_maps
->loop_depth
; ++i
)
1408 shader_addline(buffer
, "int aL%u;\n", i
);
1409 shader_addline(buffer
, "int tmpInt%u;\n", i
);
1413 /* Temporary variables for matrix operations */
1414 shader_addline(buffer
, "vec4 tmp0;\n");
1415 shader_addline(buffer
, "vec4 tmp1;\n");
1417 if (!shader
->load_local_constsF
)
1419 LIST_FOR_EACH_ENTRY(lconst
, &shader
->constantsF
, struct wined3d_shader_lconst
, entry
)
1421 shader_addline(buffer
, "const vec4 %s_lc%u = ", prefix
, lconst
->idx
);
1422 shader_glsl_append_imm_vec4(buffer
, (const float *)lconst
->value
);
1423 shader_addline(buffer
, ";\n");
1427 /* Start the main program. */
1428 shader_addline(buffer
, "void main()\n{\n");
1430 /* Direct3D applications expect integer vPos values, while OpenGL drivers
1431 * add approximately 0.5. This causes off-by-one problems as spotted by
1432 * the vPos d3d9 visual test. Unfortunately ATI cards do not add exactly
1433 * 0.5, but rather something like 0.49999999 or 0.50000001, which still
1434 * causes precision troubles when we just subtract 0.5.
1436 * To deal with that, just floor() the position. This will eliminate the
1437 * fraction on all cards.
1439 * TODO: Test how this behaves with multisampling.
1441 * An advantage of floor is that it works even if the driver doesn't add
1442 * 0.5. It is somewhat questionable if 1.5, 2.5, ... are the proper values
1443 * to return in gl_FragCoord, even though coordinates specify the pixel
1444 * centers instead of the pixel corners. This code will behave correctly
1445 * on drivers that returns integer values. */
1446 if (version
->type
== WINED3D_SHADER_TYPE_PIXEL
&& reg_maps
->vpos
)
1448 if (shader
->device
->wined3d
->flags
& WINED3D_PIXEL_CENTER_INTEGER
)
1449 shader_addline(buffer
,
1450 "vpos = floor(vec4(0, ycorrection[0], 0, 0) + gl_FragCoord * vec4(1, ycorrection[1], 1, 1));\n");
1452 shader_addline(buffer
,
1453 "vpos = vec4(0, ycorrection[0], 0, 0) + gl_FragCoord * vec4(1, ycorrection[1], 1, 1);\n");
1457 /*****************************************************************************
1458 * Functions to generate GLSL strings from DirectX Shader bytecode begin here.
1460 * For more information, see http://wiki.winehq.org/DirectX-Shaders
1461 ****************************************************************************/
1464 static void shader_glsl_add_src_param(const struct wined3d_shader_instruction
*ins
,
1465 const struct wined3d_shader_src_param
*wined3d_src
, DWORD mask
, struct glsl_src_param
*glsl_src
);
1467 /** Used for opcode modifiers - They multiply the result by the specified amount */
1468 static const char * const shift_glsl_tab
[] = {
1470 "2.0 * ", /* 1 (x2) */
1471 "4.0 * ", /* 2 (x4) */
1472 "8.0 * ", /* 3 (x8) */
1473 "16.0 * ", /* 4 (x16) */
1474 "32.0 * ", /* 5 (x32) */
1481 "0.0625 * ", /* 12 (d16) */
1482 "0.125 * ", /* 13 (d8) */
1483 "0.25 * ", /* 14 (d4) */
1484 "0.5 * " /* 15 (d2) */
1487 /* Generate a GLSL parameter that does the input modifier computation and return the input register/mask to use */
1488 static void shader_glsl_gen_modifier(enum wined3d_shader_src_modifier src_modifier
,
1489 const char *in_reg
, const char *in_regswizzle
, char *out_str
)
1493 switch (src_modifier
)
1495 case WINED3DSPSM_DZ
: /* Need to handle this in the instructions itself (texld & texcrd). */
1496 case WINED3DSPSM_DW
:
1497 case WINED3DSPSM_NONE
:
1498 sprintf(out_str
, "%s%s", in_reg
, in_regswizzle
);
1500 case WINED3DSPSM_NEG
:
1501 sprintf(out_str
, "-%s%s", in_reg
, in_regswizzle
);
1503 case WINED3DSPSM_NOT
:
1504 sprintf(out_str
, "!%s%s", in_reg
, in_regswizzle
);
1506 case WINED3DSPSM_BIAS
:
1507 sprintf(out_str
, "(%s%s - vec4(0.5)%s)", in_reg
, in_regswizzle
, in_regswizzle
);
1509 case WINED3DSPSM_BIASNEG
:
1510 sprintf(out_str
, "-(%s%s - vec4(0.5)%s)", in_reg
, in_regswizzle
, in_regswizzle
);
1512 case WINED3DSPSM_SIGN
:
1513 sprintf(out_str
, "(2.0 * (%s%s - 0.5))", in_reg
, in_regswizzle
);
1515 case WINED3DSPSM_SIGNNEG
:
1516 sprintf(out_str
, "-(2.0 * (%s%s - 0.5))", in_reg
, in_regswizzle
);
1518 case WINED3DSPSM_COMP
:
1519 sprintf(out_str
, "(1.0 - %s%s)", in_reg
, in_regswizzle
);
1521 case WINED3DSPSM_X2
:
1522 sprintf(out_str
, "(2.0 * %s%s)", in_reg
, in_regswizzle
);
1524 case WINED3DSPSM_X2NEG
:
1525 sprintf(out_str
, "-(2.0 * %s%s)", in_reg
, in_regswizzle
);
1527 case WINED3DSPSM_ABS
:
1528 sprintf(out_str
, "abs(%s%s)", in_reg
, in_regswizzle
);
1530 case WINED3DSPSM_ABSNEG
:
1531 sprintf(out_str
, "-abs(%s%s)", in_reg
, in_regswizzle
);
1534 FIXME("Unhandled modifier %u\n", src_modifier
);
1535 sprintf(out_str
, "%s%s", in_reg
, in_regswizzle
);
1539 /** Writes the GLSL variable name that corresponds to the register that the
1540 * DX opcode parameter is trying to access */
1541 static void shader_glsl_get_register_name(const struct wined3d_shader_register
*reg
,
1542 char *register_name
, BOOL
*is_color
, const struct wined3d_shader_instruction
*ins
)
1544 /* oPos, oFog and oPts in D3D */
1545 static const char * const hwrastout_reg_names
[] = {"vs_out[10]", "vs_out[11].x", "vs_out[11].y"};
1547 const struct wined3d_shader
*shader
= ins
->ctx
->shader
;
1548 const struct wined3d_shader_reg_maps
*reg_maps
= ins
->ctx
->reg_maps
;
1549 const struct wined3d_shader_version
*version
= ®_maps
->shader_version
;
1550 const struct wined3d_gl_info
*gl_info
= ins
->ctx
->gl_info
;
1551 const char *prefix
= shader_glsl_get_prefix(version
->type
);
1552 struct glsl_src_param rel_param0
, rel_param1
;
1553 char imm_str
[4][17];
1555 if (reg
->idx
[0].offset
!= ~0U && reg
->idx
[0].rel_addr
)
1556 shader_glsl_add_src_param(ins
, reg
->idx
[0].rel_addr
, WINED3DSP_WRITEMASK_0
, &rel_param0
);
1557 if (reg
->idx
[1].offset
!= ~0U && reg
->idx
[1].rel_addr
)
1558 shader_glsl_add_src_param(ins
, reg
->idx
[1].rel_addr
, WINED3DSP_WRITEMASK_0
, &rel_param1
);
1563 case WINED3DSPR_TEMP
:
1564 sprintf(register_name
, "R%u", reg
->idx
[0].offset
);
1567 case WINED3DSPR_INPUT
:
1568 /* vertex shaders */
1569 if (version
->type
== WINED3D_SHADER_TYPE_VERTEX
)
1571 struct shader_glsl_ctx_priv
*priv
= ins
->ctx
->backend_data
;
1572 if (priv
->cur_vs_args
->swizzle_map
& (1 << reg
->idx
[0].offset
))
1574 sprintf(register_name
, "%s_in%u", prefix
, reg
->idx
[0].offset
);
1578 if (version
->type
== WINED3D_SHADER_TYPE_GEOMETRY
)
1580 if (reg
->idx
[0].rel_addr
)
1582 if (reg
->idx
[1].rel_addr
)
1583 sprintf(register_name
, "gs_in[%s + %u][%s + %u]",
1584 rel_param0
.param_str
, reg
->idx
[0].offset
, rel_param1
.param_str
, reg
->idx
[1].offset
);
1586 sprintf(register_name
, "gs_in[%s + %u][%u]",
1587 rel_param0
.param_str
, reg
->idx
[0].offset
, reg
->idx
[1].offset
);
1589 else if (reg
->idx
[1].rel_addr
)
1590 sprintf(register_name
, "gs_in[%u][%s + %u]",
1591 reg
->idx
[0].offset
, rel_param1
.param_str
, reg
->idx
[1].offset
);
1593 sprintf(register_name
, "gs_in[%u][%u]", reg
->idx
[0].offset
, reg
->idx
[1].offset
);
1597 /* pixel shaders >= 3.0 */
1598 if (version
->major
>= 3)
1600 DWORD idx
= shader
->u
.ps
.input_reg_map
[reg
->idx
[0].offset
];
1601 unsigned int in_count
= vec4_varyings(version
->major
, gl_info
);
1603 if (reg
->idx
[0].rel_addr
)
1605 /* Removing a + 0 would be an obvious optimization, but
1606 * OS X doesn't see the NOP operation there. */
1609 if (shader
->u
.ps
.declared_in_count
> in_count
)
1611 sprintf(register_name
,
1612 "((%s + %u) > %u ? (%s + %u) > %u ? gl_SecondaryColor : gl_Color : %s_in[%s + %u])",
1613 rel_param0
.param_str
, idx
, in_count
- 1, rel_param0
.param_str
, idx
, in_count
,
1614 prefix
, rel_param0
.param_str
, idx
);
1618 sprintf(register_name
, "%s_in[%s + %u]", prefix
, rel_param0
.param_str
, idx
);
1623 if (shader
->u
.ps
.declared_in_count
> in_count
)
1625 sprintf(register_name
, "((%s) > %u ? (%s) > %u ? gl_SecondaryColor : gl_Color : %s_in[%s])",
1626 rel_param0
.param_str
, in_count
- 1, rel_param0
.param_str
, in_count
,
1627 prefix
, rel_param0
.param_str
);
1631 sprintf(register_name
, "%s_in[%s]", prefix
, rel_param0
.param_str
);
1637 if (idx
== in_count
) sprintf(register_name
, "gl_Color");
1638 else if (idx
== in_count
+ 1) sprintf(register_name
, "gl_SecondaryColor");
1639 else sprintf(register_name
, "%s_in[%u]", prefix
, idx
);
1644 if (!reg
->idx
[0].offset
)
1645 strcpy(register_name
, "gl_Color");
1647 strcpy(register_name
, "gl_SecondaryColor");
1652 case WINED3DSPR_CONST
:
1654 /* Relative addressing */
1655 if (reg
->idx
[0].rel_addr
)
1657 if (reg
->idx
[0].offset
)
1658 sprintf(register_name
, "%s_c[%s + %u]", prefix
, rel_param0
.param_str
, reg
->idx
[0].offset
);
1660 sprintf(register_name
, "%s_c[%s]", prefix
, rel_param0
.param_str
);
1664 if (shader_constant_is_local(shader
, reg
->idx
[0].offset
))
1665 sprintf(register_name
, "%s_lc%u", prefix
, reg
->idx
[0].offset
);
1667 sprintf(register_name
, "%s_c[%u]", prefix
, reg
->idx
[0].offset
);
1672 case WINED3DSPR_CONSTINT
:
1673 sprintf(register_name
, "%s_i[%u]", prefix
, reg
->idx
[0].offset
);
1676 case WINED3DSPR_CONSTBOOL
:
1677 sprintf(register_name
, "%s_b[%u]", prefix
, reg
->idx
[0].offset
);
1680 case WINED3DSPR_TEXTURE
: /* case WINED3DSPR_ADDR: */
1681 if (version
->type
== WINED3D_SHADER_TYPE_PIXEL
)
1682 sprintf(register_name
, "T%u", reg
->idx
[0].offset
);
1684 sprintf(register_name
, "A%u", reg
->idx
[0].offset
);
1687 case WINED3DSPR_LOOP
:
1688 sprintf(register_name
, "aL%u", ins
->ctx
->loop_state
->current_reg
- 1);
1691 case WINED3DSPR_SAMPLER
:
1692 sprintf(register_name
, "%s_sampler%u", prefix
, reg
->idx
[0].offset
);
1695 case WINED3DSPR_COLOROUT
:
1696 if (reg
->idx
[0].offset
>= gl_info
->limits
.buffers
)
1697 WARN("Write to render target %u, only %d supported.\n",
1698 reg
->idx
[0].offset
, gl_info
->limits
.buffers
);
1700 sprintf(register_name
, "gl_FragData[%u]", reg
->idx
[0].offset
);
1703 case WINED3DSPR_RASTOUT
:
1704 sprintf(register_name
, "%s", hwrastout_reg_names
[reg
->idx
[0].offset
]);
1707 case WINED3DSPR_DEPTHOUT
:
1708 sprintf(register_name
, "gl_FragDepth");
1711 case WINED3DSPR_ATTROUT
:
1712 if (!reg
->idx
[0].offset
)
1713 sprintf(register_name
, "%s_out[8]", prefix
);
1715 sprintf(register_name
, "%s_out[9]", prefix
);
1718 case WINED3DSPR_TEXCRDOUT
:
1719 /* Vertex shaders >= 3.0: WINED3DSPR_OUTPUT */
1720 sprintf(register_name
, "%s_out[%u]", prefix
, reg
->idx
[0].offset
);
1723 case WINED3DSPR_MISCTYPE
:
1724 if (!reg
->idx
[0].offset
)
1727 sprintf(register_name
, "vpos");
1729 else if (reg
->idx
[0].offset
== 1)
1731 /* Note that gl_FrontFacing is a bool, while vFace is
1732 * a float for which the sign determines front/back */
1733 sprintf(register_name
, "(gl_FrontFacing ? 1.0 : -1.0)");
1737 FIXME("Unhandled misctype register %u.\n", reg
->idx
[0].offset
);
1738 sprintf(register_name
, "unrecognized_register");
1742 case WINED3DSPR_IMMCONST
:
1743 switch (reg
->immconst_type
)
1745 case WINED3D_IMMCONST_SCALAR
:
1746 switch (reg
->data_type
)
1748 case WINED3D_DATA_FLOAT
:
1749 wined3d_ftoa(*(const float *)reg
->immconst_data
, register_name
);
1751 case WINED3D_DATA_INT
:
1752 sprintf(register_name
, "%#x", reg
->immconst_data
[0]);
1754 case WINED3D_DATA_RESOURCE
:
1755 case WINED3D_DATA_SAMPLER
:
1756 case WINED3D_DATA_UINT
:
1757 sprintf(register_name
, "%#xu", reg
->immconst_data
[0]);
1760 sprintf(register_name
, "<unhandled data type %#x>", reg
->data_type
);
1765 case WINED3D_IMMCONST_VEC4
:
1766 switch (reg
->data_type
)
1768 case WINED3D_DATA_FLOAT
:
1769 wined3d_ftoa(*(const float *)®
->immconst_data
[0], imm_str
[0]);
1770 wined3d_ftoa(*(const float *)®
->immconst_data
[1], imm_str
[1]);
1771 wined3d_ftoa(*(const float *)®
->immconst_data
[2], imm_str
[2]);
1772 wined3d_ftoa(*(const float *)®
->immconst_data
[3], imm_str
[3]);
1773 sprintf(register_name
, "vec4(%s, %s, %s, %s)",
1774 imm_str
[0], imm_str
[1], imm_str
[2], imm_str
[3]);
1776 case WINED3D_DATA_INT
:
1777 sprintf(register_name
, "ivec4(%#x, %#x, %#x, %#x)",
1778 reg
->immconst_data
[0], reg
->immconst_data
[1],
1779 reg
->immconst_data
[2], reg
->immconst_data
[3]);
1781 case WINED3D_DATA_RESOURCE
:
1782 case WINED3D_DATA_SAMPLER
:
1783 case WINED3D_DATA_UINT
:
1784 sprintf(register_name
, "uvec4(%#xu, %#xu, %#xu, %#xu)",
1785 reg
->immconst_data
[0], reg
->immconst_data
[1],
1786 reg
->immconst_data
[2], reg
->immconst_data
[3]);
1789 sprintf(register_name
, "<unhandled data type %#x>", reg
->data_type
);
1795 FIXME("Unhandled immconst type %#x\n", reg
->immconst_type
);
1796 sprintf(register_name
, "<unhandled_immconst_type %#x>", reg
->immconst_type
);
1800 case WINED3DSPR_CONSTBUFFER
:
1801 if (reg
->idx
[1].rel_addr
)
1802 sprintf(register_name
, "%s_cb%u[%s + %u]",
1803 prefix
, reg
->idx
[0].offset
, rel_param1
.param_str
, reg
->idx
[1].offset
);
1805 sprintf(register_name
, "%s_cb%u[%u]", prefix
, reg
->idx
[0].offset
, reg
->idx
[1].offset
);
1808 case WINED3DSPR_PRIMID
:
1809 sprintf(register_name
, "uint(gl_PrimitiveIDIn)");
1813 FIXME("Unhandled register type %#x.\n", reg
->type
);
1814 sprintf(register_name
, "unrecognized_register");
1819 static void shader_glsl_write_mask_to_str(DWORD write_mask
, char *str
)
1822 if (write_mask
& WINED3DSP_WRITEMASK_0
) *str
++ = 'x';
1823 if (write_mask
& WINED3DSP_WRITEMASK_1
) *str
++ = 'y';
1824 if (write_mask
& WINED3DSP_WRITEMASK_2
) *str
++ = 'z';
1825 if (write_mask
& WINED3DSP_WRITEMASK_3
) *str
++ = 'w';
1829 /* Get the GLSL write mask for the destination register */
1830 static DWORD
shader_glsl_get_write_mask(const struct wined3d_shader_dst_param
*param
, char *write_mask
)
1832 DWORD mask
= param
->write_mask
;
1834 if (shader_is_scalar(¶m
->reg
))
1836 mask
= WINED3DSP_WRITEMASK_0
;
1841 shader_glsl_write_mask_to_str(mask
, write_mask
);
1847 static unsigned int shader_glsl_get_write_mask_size(DWORD write_mask
) {
1848 unsigned int size
= 0;
1850 if (write_mask
& WINED3DSP_WRITEMASK_0
) ++size
;
1851 if (write_mask
& WINED3DSP_WRITEMASK_1
) ++size
;
1852 if (write_mask
& WINED3DSP_WRITEMASK_2
) ++size
;
1853 if (write_mask
& WINED3DSP_WRITEMASK_3
) ++size
;
1858 static void shader_glsl_swizzle_to_str(const DWORD swizzle
, BOOL fixup
, DWORD mask
, char *str
)
1860 /* For registers of type WINED3DDECLTYPE_D3DCOLOR, data is stored as "bgra",
1861 * but addressed as "rgba". To fix this we need to swap the register's x
1862 * and z components. */
1863 const char *swizzle_chars
= fixup
? "zyxw" : "xyzw";
1866 /* swizzle bits fields: wwzzyyxx */
1867 if (mask
& WINED3DSP_WRITEMASK_0
) *str
++ = swizzle_chars
[swizzle
& 0x03];
1868 if (mask
& WINED3DSP_WRITEMASK_1
) *str
++ = swizzle_chars
[(swizzle
>> 2) & 0x03];
1869 if (mask
& WINED3DSP_WRITEMASK_2
) *str
++ = swizzle_chars
[(swizzle
>> 4) & 0x03];
1870 if (mask
& WINED3DSP_WRITEMASK_3
) *str
++ = swizzle_chars
[(swizzle
>> 6) & 0x03];
1874 static void shader_glsl_get_swizzle(const struct wined3d_shader_src_param
*param
,
1875 BOOL fixup
, DWORD mask
, char *swizzle_str
)
1877 if (shader_is_scalar(¶m
->reg
))
1878 *swizzle_str
= '\0';
1880 shader_glsl_swizzle_to_str(param
->swizzle
, fixup
, mask
, swizzle_str
);
1883 /* From a given parameter token, generate the corresponding GLSL string.
1884 * Also, return the actual register name and swizzle in case the
1885 * caller needs this information as well. */
1886 static void shader_glsl_add_src_param(const struct wined3d_shader_instruction
*ins
,
1887 const struct wined3d_shader_src_param
*wined3d_src
, DWORD mask
, struct glsl_src_param
*glsl_src
)
1889 BOOL is_color
= FALSE
;
1890 char swizzle_str
[6];
1892 glsl_src
->reg_name
[0] = '\0';
1893 glsl_src
->param_str
[0] = '\0';
1894 swizzle_str
[0] = '\0';
1896 shader_glsl_get_register_name(&wined3d_src
->reg
, glsl_src
->reg_name
, &is_color
, ins
);
1897 shader_glsl_get_swizzle(wined3d_src
, is_color
, mask
, swizzle_str
);
1899 if (wined3d_src
->reg
.type
== WINED3DSPR_IMMCONST
|| wined3d_src
->reg
.type
== WINED3DSPR_PRIMID
)
1901 shader_glsl_gen_modifier(wined3d_src
->modifiers
, glsl_src
->reg_name
, swizzle_str
, glsl_src
->param_str
);
1907 switch (wined3d_src
->reg
.data_type
)
1909 case WINED3D_DATA_FLOAT
:
1910 sprintf(reg_name
, "%s", glsl_src
->reg_name
);
1912 case WINED3D_DATA_INT
:
1913 sprintf(reg_name
, "floatBitsToInt(%s)", glsl_src
->reg_name
);
1915 case WINED3D_DATA_RESOURCE
:
1916 case WINED3D_DATA_SAMPLER
:
1917 case WINED3D_DATA_UINT
:
1918 sprintf(reg_name
, "floatBitsToUint(%s)", glsl_src
->reg_name
);
1921 FIXME("Unhandled data type %#x.\n", wined3d_src
->reg
.data_type
);
1922 sprintf(reg_name
, "%s", glsl_src
->reg_name
);
1926 shader_glsl_gen_modifier(wined3d_src
->modifiers
, reg_name
, swizzle_str
, glsl_src
->param_str
);
1930 /* From a given parameter token, generate the corresponding GLSL string.
1931 * Also, return the actual register name and swizzle in case the
1932 * caller needs this information as well. */
1933 static DWORD
shader_glsl_add_dst_param(const struct wined3d_shader_instruction
*ins
,
1934 const struct wined3d_shader_dst_param
*wined3d_dst
, struct glsl_dst_param
*glsl_dst
)
1936 BOOL is_color
= FALSE
;
1938 glsl_dst
->mask_str
[0] = '\0';
1939 glsl_dst
->reg_name
[0] = '\0';
1941 shader_glsl_get_register_name(&wined3d_dst
->reg
, glsl_dst
->reg_name
, &is_color
, ins
);
1942 return shader_glsl_get_write_mask(wined3d_dst
, glsl_dst
->mask_str
);
1945 /* Append the destination part of the instruction to the buffer, return the effective write mask */
1946 static DWORD
shader_glsl_append_dst_ext(struct wined3d_shader_buffer
*buffer
,
1947 const struct wined3d_shader_instruction
*ins
, const struct wined3d_shader_dst_param
*dst
,
1948 enum wined3d_data_type data_type
)
1950 struct glsl_dst_param glsl_dst
;
1953 if ((mask
= shader_glsl_add_dst_param(ins
, dst
, &glsl_dst
)))
1957 case WINED3D_DATA_FLOAT
:
1958 shader_addline(buffer
, "%s%s = %s(",
1959 glsl_dst
.reg_name
, glsl_dst
.mask_str
, shift_glsl_tab
[dst
->shift
]);
1961 case WINED3D_DATA_INT
:
1962 shader_addline(buffer
, "%s%s = %sintBitsToFloat(",
1963 glsl_dst
.reg_name
, glsl_dst
.mask_str
, shift_glsl_tab
[dst
->shift
]);
1965 case WINED3D_DATA_RESOURCE
:
1966 case WINED3D_DATA_SAMPLER
:
1967 case WINED3D_DATA_UINT
:
1968 shader_addline(buffer
, "%s%s = %suintBitsToFloat(",
1969 glsl_dst
.reg_name
, glsl_dst
.mask_str
, shift_glsl_tab
[dst
->shift
]);
1972 FIXME("Unhandled data type %#x.\n", data_type
);
1973 shader_addline(buffer
, "%s%s = %s(",
1974 glsl_dst
.reg_name
, glsl_dst
.mask_str
, shift_glsl_tab
[dst
->shift
]);
1982 /* Append the destination part of the instruction to the buffer, return the effective write mask */
1983 static DWORD
shader_glsl_append_dst(struct wined3d_shader_buffer
*buffer
, const struct wined3d_shader_instruction
*ins
)
1985 return shader_glsl_append_dst_ext(buffer
, ins
, &ins
->dst
[0], ins
->dst
[0].reg
.data_type
);
1988 /** Process GLSL instruction modifiers */
1989 static void shader_glsl_add_instruction_modifiers(const struct wined3d_shader_instruction
*ins
)
1991 struct glsl_dst_param dst_param
;
1994 if (!ins
->dst_count
) return;
1996 modifiers
= ins
->dst
[0].modifiers
;
1997 if (!modifiers
) return;
1999 shader_glsl_add_dst_param(ins
, &ins
->dst
[0], &dst_param
);
2001 if (modifiers
& WINED3DSPDM_SATURATE
)
2003 /* _SAT means to clamp the value of the register to between 0 and 1 */
2004 shader_addline(ins
->ctx
->buffer
, "%s%s = clamp(%s%s, 0.0, 1.0);\n", dst_param
.reg_name
,
2005 dst_param
.mask_str
, dst_param
.reg_name
, dst_param
.mask_str
);
2008 if (modifiers
& WINED3DSPDM_MSAMPCENTROID
)
2010 FIXME("_centroid modifier not handled\n");
2013 if (modifiers
& WINED3DSPDM_PARTIALPRECISION
)
2015 /* MSDN says this modifier can be safely ignored, so that's what we'll do. */
2019 static const char *shader_glsl_get_rel_op(enum wined3d_shader_rel_op op
)
2023 case WINED3D_SHADER_REL_OP_GT
: return ">";
2024 case WINED3D_SHADER_REL_OP_EQ
: return "==";
2025 case WINED3D_SHADER_REL_OP_GE
: return ">=";
2026 case WINED3D_SHADER_REL_OP_LT
: return "<";
2027 case WINED3D_SHADER_REL_OP_NE
: return "!=";
2028 case WINED3D_SHADER_REL_OP_LE
: return "<=";
2030 FIXME("Unrecognized operator %#x.\n", op
);
2035 static void shader_glsl_get_sample_function(const struct wined3d_shader_context
*ctx
,
2036 DWORD resource_idx
, DWORD flags
, struct glsl_sample_function
*sample_function
)
2038 enum wined3d_shader_resource_type resource_type
= ctx
->reg_maps
->resource_info
[resource_idx
].type
;
2039 const struct wined3d_gl_info
*gl_info
= ctx
->gl_info
;
2040 BOOL shadow
= ctx
->reg_maps
->shader_version
.type
== WINED3D_SHADER_TYPE_PIXEL
2041 && (((const struct shader_glsl_ctx_priv
*)ctx
->backend_data
)->cur_ps_args
->shadow
& (1 << resource_idx
));
2042 BOOL projected
= flags
& WINED3D_GLSL_SAMPLE_PROJECTED
;
2043 BOOL texrect
= flags
& WINED3D_GLSL_SAMPLE_NPOT
&& gl_info
->supported
[ARB_TEXTURE_RECTANGLE
];
2044 BOOL lod
= flags
& WINED3D_GLSL_SAMPLE_LOD
;
2045 BOOL grad
= flags
& WINED3D_GLSL_SAMPLE_GRAD
;
2047 sample_function
->data_type
= ctx
->reg_maps
->resource_info
[resource_idx
].data_type
;
2049 /* Note that there's no such thing as a projected cube texture. */
2050 switch (resource_type
)
2052 case WINED3D_SHADER_RESOURCE_TEXTURE_1D
:
2057 sample_function
->name
= projected
? "shadow1DProjLod" : "shadow1DLod";
2061 if (gl_info
->supported
[EXT_GPU_SHADER4
])
2062 sample_function
->name
= projected
? "shadow1DProjGrad" : "shadow1DGrad";
2063 else if (gl_info
->supported
[ARB_SHADER_TEXTURE_LOD
])
2064 sample_function
->name
= projected
? "shadow1DProjGradARB" : "shadow1DGradARB";
2067 FIXME("Unsupported 1D shadow grad function.\n");
2068 sample_function
->name
= "unsupported1DGrad";
2073 sample_function
->name
= projected
? "shadow1DProj" : "shadow1D";
2075 sample_function
->coord_mask
= WINED3DSP_WRITEMASK_0
| WINED3DSP_WRITEMASK_1
;
2081 sample_function
->name
= projected
? "texture1DProjLod" : "texture1DLod";
2085 if (gl_info
->supported
[EXT_GPU_SHADER4
])
2086 sample_function
->name
= projected
? "texture1DProjGrad" : "texture1DGrad";
2087 else if (gl_info
->supported
[ARB_SHADER_TEXTURE_LOD
])
2088 sample_function
->name
= projected
? "texture1DProjGradARB" : "texture1DGradARB";
2091 FIXME("Unsupported 1D grad function.\n");
2092 sample_function
->name
= "unsupported1DGrad";
2097 sample_function
->name
= projected
? "texture1DProj" : "texture1D";
2099 sample_function
->coord_mask
= WINED3DSP_WRITEMASK_0
;
2103 case WINED3D_SHADER_RESOURCE_TEXTURE_2D
:
2110 sample_function
->name
= projected
? "shadow2DRectProjLod" : "shadow2DRectLod";
2114 if (gl_info
->supported
[EXT_GPU_SHADER4
])
2115 sample_function
->name
= projected
? "shadow2DRectProjGrad" : "shadow2DRectGrad";
2116 else if (gl_info
->supported
[ARB_SHADER_TEXTURE_LOD
])
2117 sample_function
->name
= projected
? "shadow2DRectProjGradARB" : "shadow2DRectGradARB";
2120 FIXME("Unsupported RECT shadow grad function.\n");
2121 sample_function
->name
= "unsupported2DRectGrad";
2126 sample_function
->name
= projected
? "shadow2DRectProj" : "shadow2DRect";
2133 sample_function
->name
= projected
? "shadow2DProjLod" : "shadow2DLod";
2137 if (gl_info
->supported
[EXT_GPU_SHADER4
])
2138 sample_function
->name
= projected
? "shadow2DProjGrad" : "shadow2DGrad";
2139 else if (gl_info
->supported
[ARB_SHADER_TEXTURE_LOD
])
2140 sample_function
->name
= projected
? "shadow2DProjGradARB" : "shadow2DGradARB";
2143 FIXME("Unsupported 2D shadow grad function.\n");
2144 sample_function
->name
= "unsupported2DGrad";
2149 sample_function
->name
= projected
? "shadow2DProj" : "shadow2D";
2152 sample_function
->coord_mask
= WINED3DSP_WRITEMASK_0
| WINED3DSP_WRITEMASK_1
| WINED3DSP_WRITEMASK_2
;
2160 sample_function
->name
= projected
? "texture2DRectProjLod" : "texture2DRectLod";
2164 if (gl_info
->supported
[EXT_GPU_SHADER4
])
2165 sample_function
->name
= projected
? "texture2DRectProjGrad" : "texture2DRectGrad";
2166 else if (gl_info
->supported
[ARB_SHADER_TEXTURE_LOD
])
2167 sample_function
->name
= projected
? "texture2DRectProjGradARB" : "texture2DRectGradARB";
2170 FIXME("Unsupported RECT grad function.\n");
2171 sample_function
->name
= "unsupported2DRectGrad";
2176 sample_function
->name
= projected
? "texture2DRectProj" : "texture2DRect";
2183 sample_function
->name
= projected
? "texture2DProjLod" : "texture2DLod";
2187 if (gl_info
->supported
[EXT_GPU_SHADER4
])
2188 sample_function
->name
= projected
? "texture2DProjGrad" : "texture2DGrad";
2189 else if (gl_info
->supported
[ARB_SHADER_TEXTURE_LOD
])
2190 sample_function
->name
= projected
? "texture2DProjGradARB" : "texture2DGradARB";
2193 FIXME("Unsupported 2D grad function.\n");
2194 sample_function
->name
= "unsupported2DGrad";
2199 sample_function
->name
= projected
? "texture2DProj" : "texture2D";
2202 sample_function
->coord_mask
= WINED3DSP_WRITEMASK_0
| WINED3DSP_WRITEMASK_1
;
2206 case WINED3D_SHADER_RESOURCE_TEXTURE_3D
:
2209 FIXME("Unsupported 3D shadow function.\n");
2210 sample_function
->name
= "unsupported3DShadow";
2211 sample_function
->coord_mask
= 0;
2217 sample_function
->name
= projected
? "texture3DProjLod" : "texture3DLod";
2221 if (gl_info
->supported
[EXT_GPU_SHADER4
])
2222 sample_function
->name
= projected
? "texture3DProjGrad" : "texture3DGrad";
2223 else if (gl_info
->supported
[ARB_SHADER_TEXTURE_LOD
])
2224 sample_function
->name
= projected
? "texture3DProjGradARB" : "texture3DGradARB";
2227 FIXME("Unsupported 3D grad function.\n");
2228 sample_function
->name
= "unsupported3DGrad";
2233 sample_function
->name
= projected
? "texture3DProj" : "texture3D";
2235 sample_function
->coord_mask
= WINED3DSP_WRITEMASK_0
| WINED3DSP_WRITEMASK_1
| WINED3DSP_WRITEMASK_2
;
2239 case WINED3D_SHADER_RESOURCE_TEXTURE_CUBE
:
2242 FIXME("Unsupported Cube shadow function.\n");
2243 sample_function
->name
= "unsupportedCubeShadow";
2244 sample_function
->coord_mask
= 0;
2250 sample_function
->name
= "textureCubeLod";
2254 if (gl_info
->supported
[EXT_GPU_SHADER4
])
2255 sample_function
->name
= "textureCubeGrad";
2256 else if (gl_info
->supported
[ARB_SHADER_TEXTURE_LOD
])
2257 sample_function
->name
= "textureCubeGradARB";
2260 FIXME("Unsupported Cube grad function.\n");
2261 sample_function
->name
= "unsupportedCubeGrad";
2266 sample_function
->name
= "textureCube";
2268 sample_function
->coord_mask
= WINED3DSP_WRITEMASK_0
| WINED3DSP_WRITEMASK_1
| WINED3DSP_WRITEMASK_2
;
2273 sample_function
->name
= "";
2274 sample_function
->coord_mask
= 0;
2275 FIXME("Unhandled resource type %#x.\n", resource_type
);
2280 static void shader_glsl_append_fixup_arg(char *arguments
, const char *reg_name
,
2281 BOOL sign_fixup
, enum fixup_channel_source channel_source
)
2283 switch(channel_source
)
2285 case CHANNEL_SOURCE_ZERO
:
2286 strcat(arguments
, "0.0");
2289 case CHANNEL_SOURCE_ONE
:
2290 strcat(arguments
, "1.0");
2293 case CHANNEL_SOURCE_X
:
2294 strcat(arguments
, reg_name
);
2295 strcat(arguments
, ".x");
2298 case CHANNEL_SOURCE_Y
:
2299 strcat(arguments
, reg_name
);
2300 strcat(arguments
, ".y");
2303 case CHANNEL_SOURCE_Z
:
2304 strcat(arguments
, reg_name
);
2305 strcat(arguments
, ".z");
2308 case CHANNEL_SOURCE_W
:
2309 strcat(arguments
, reg_name
);
2310 strcat(arguments
, ".w");
2314 FIXME("Unhandled channel source %#x\n", channel_source
);
2315 strcat(arguments
, "undefined");
2319 if (sign_fixup
) strcat(arguments
, " * 2.0 - 1.0");
2322 static void shader_glsl_color_correction_ext(struct wined3d_shader_buffer
*buffer
,
2323 const char *reg_name
, DWORD mask
, struct color_fixup_desc fixup
)
2325 unsigned int mask_size
, remaining
;
2326 DWORD fixup_mask
= 0;
2327 char arguments
[256];
2330 if (fixup
.x_sign_fixup
|| fixup
.x_source
!= CHANNEL_SOURCE_X
) fixup_mask
|= WINED3DSP_WRITEMASK_0
;
2331 if (fixup
.y_sign_fixup
|| fixup
.y_source
!= CHANNEL_SOURCE_Y
) fixup_mask
|= WINED3DSP_WRITEMASK_1
;
2332 if (fixup
.z_sign_fixup
|| fixup
.z_source
!= CHANNEL_SOURCE_Z
) fixup_mask
|= WINED3DSP_WRITEMASK_2
;
2333 if (fixup
.w_sign_fixup
|| fixup
.w_source
!= CHANNEL_SOURCE_W
) fixup_mask
|= WINED3DSP_WRITEMASK_3
;
2334 if (!(mask
&= fixup_mask
))
2337 if (is_complex_fixup(fixup
))
2339 enum complex_fixup complex_fixup
= get_complex_fixup(fixup
);
2340 FIXME("Complex fixup (%#x) not supported\n",complex_fixup
);
2344 shader_glsl_write_mask_to_str(mask
, mask_str
);
2345 mask_size
= shader_glsl_get_write_mask_size(mask
);
2347 arguments
[0] = '\0';
2348 remaining
= mask_size
;
2349 if (mask
& WINED3DSP_WRITEMASK_0
)
2351 shader_glsl_append_fixup_arg(arguments
, reg_name
, fixup
.x_sign_fixup
, fixup
.x_source
);
2352 if (--remaining
) strcat(arguments
, ", ");
2354 if (mask
& WINED3DSP_WRITEMASK_1
)
2356 shader_glsl_append_fixup_arg(arguments
, reg_name
, fixup
.y_sign_fixup
, fixup
.y_source
);
2357 if (--remaining
) strcat(arguments
, ", ");
2359 if (mask
& WINED3DSP_WRITEMASK_2
)
2361 shader_glsl_append_fixup_arg(arguments
, reg_name
, fixup
.z_sign_fixup
, fixup
.z_source
);
2362 if (--remaining
) strcat(arguments
, ", ");
2364 if (mask
& WINED3DSP_WRITEMASK_3
)
2366 shader_glsl_append_fixup_arg(arguments
, reg_name
, fixup
.w_sign_fixup
, fixup
.w_source
);
2367 if (--remaining
) strcat(arguments
, ", ");
2371 shader_addline(buffer
, "%s%s = vec%u(%s);\n", reg_name
, mask_str
, mask_size
, arguments
);
2373 shader_addline(buffer
, "%s%s = %s;\n", reg_name
, mask_str
, arguments
);
2376 static void shader_glsl_color_correction(const struct wined3d_shader_instruction
*ins
, struct color_fixup_desc fixup
)
2381 shader_glsl_get_register_name(&ins
->dst
[0].reg
, reg_name
, &is_color
, ins
);
2382 shader_glsl_color_correction_ext(ins
->ctx
->buffer
, reg_name
, ins
->dst
[0].write_mask
, fixup
);
2385 static void PRINTF_ATTR(8, 9) shader_glsl_gen_sample_code(const struct wined3d_shader_instruction
*ins
,
2386 DWORD sampler
, const struct glsl_sample_function
*sample_function
, DWORD swizzle
,
2387 const char *dx
, const char *dy
, const char *bias
, const char *coord_reg_fmt
, ...)
2389 const struct wined3d_shader_version
*version
= &ins
->ctx
->reg_maps
->shader_version
;
2390 char dst_swizzle
[6];
2391 struct color_fixup_desc fixup
;
2392 BOOL np2_fixup
= FALSE
;
2395 shader_glsl_swizzle_to_str(swizzle
, FALSE
, ins
->dst
[0].write_mask
, dst_swizzle
);
2397 if (version
->type
== WINED3D_SHADER_TYPE_PIXEL
)
2399 const struct shader_glsl_ctx_priv
*priv
= ins
->ctx
->backend_data
;
2400 fixup
= priv
->cur_ps_args
->color_fixup
[sampler
];
2402 if(priv
->cur_ps_args
->np2_fixup
& (1 << sampler
)) {
2404 FIXME("Biased sampling from NP2 textures is unsupported\n");
2412 fixup
= COLOR_FIXUP_IDENTITY
; /* FIXME: Vshader color fixup */
2415 shader_glsl_append_dst_ext(ins
->ctx
->buffer
, ins
, &ins
->dst
[0], sample_function
->data_type
);
2417 shader_addline(ins
->ctx
->buffer
, "%s(%s_sampler%u, ",
2418 sample_function
->name
, shader_glsl_get_prefix(version
->type
), sampler
);
2420 va_start(args
, coord_reg_fmt
);
2421 shader_vaddline(ins
->ctx
->buffer
, coord_reg_fmt
, args
);
2425 shader_addline(ins
->ctx
->buffer
, ", %s)%s);\n", bias
, dst_swizzle
);
2428 const struct shader_glsl_ctx_priv
*priv
= ins
->ctx
->backend_data
;
2429 const unsigned char idx
= priv
->cur_np2fixup_info
->idx
[sampler
];
2431 shader_addline(ins
->ctx
->buffer
, " * ps_samplerNP2Fixup[%u].%s)%s);\n", idx
>> 1,
2432 (idx
% 2) ? "zw" : "xy", dst_swizzle
);
2433 } else if(dx
&& dy
) {
2434 shader_addline(ins
->ctx
->buffer
, ", %s, %s)%s);\n", dx
, dy
, dst_swizzle
);
2436 shader_addline(ins
->ctx
->buffer
, ")%s);\n", dst_swizzle
);
2440 if(!is_identity_fixup(fixup
)) {
2441 shader_glsl_color_correction(ins
, fixup
);
2445 /*****************************************************************************
2446 * Begin processing individual instruction opcodes
2447 ****************************************************************************/
2449 static void shader_glsl_binop(const struct wined3d_shader_instruction
*ins
)
2451 struct wined3d_shader_buffer
*buffer
= ins
->ctx
->buffer
;
2452 struct glsl_src_param src0_param
;
2453 struct glsl_src_param src1_param
;
2457 /* Determine the GLSL operator to use based on the opcode */
2458 switch (ins
->handler_idx
)
2460 case WINED3DSIH_ADD
: op
= "+"; break;
2461 case WINED3DSIH_AND
: op
= "&"; break;
2462 case WINED3DSIH_DIV
: op
= "/"; break;
2463 case WINED3DSIH_IADD
: op
= "+"; break;
2464 case WINED3DSIH_ISHL
: op
= "<<"; break;
2465 case WINED3DSIH_MUL
: op
= "*"; break;
2466 case WINED3DSIH_OR
: op
= "|"; break;
2467 case WINED3DSIH_SUB
: op
= "-"; break;
2468 case WINED3DSIH_USHR
: op
= ">>"; break;
2469 case WINED3DSIH_XOR
: op
= "^"; break;
2471 op
= "<unhandled operator>";
2472 FIXME("Opcode %#x not yet handled in GLSL\n", ins
->handler_idx
);
2476 write_mask
= shader_glsl_append_dst(buffer
, ins
);
2477 shader_glsl_add_src_param(ins
, &ins
->src
[0], write_mask
, &src0_param
);
2478 shader_glsl_add_src_param(ins
, &ins
->src
[1], write_mask
, &src1_param
);
2479 shader_addline(buffer
, "%s %s %s);\n", src0_param
.param_str
, op
, src1_param
.param_str
);
2482 static void shader_glsl_relop(const struct wined3d_shader_instruction
*ins
)
2484 struct wined3d_shader_buffer
*buffer
= ins
->ctx
->buffer
;
2485 struct glsl_src_param src0_param
;
2486 struct glsl_src_param src1_param
;
2487 unsigned int mask_size
;
2491 write_mask
= shader_glsl_append_dst(buffer
, ins
);
2492 mask_size
= shader_glsl_get_write_mask_size(write_mask
);
2493 shader_glsl_add_src_param(ins
, &ins
->src
[0], write_mask
, &src0_param
);
2494 shader_glsl_add_src_param(ins
, &ins
->src
[1], write_mask
, &src1_param
);
2498 switch (ins
->handler_idx
)
2500 case WINED3DSIH_EQ
: op
= "equal"; break;
2501 case WINED3DSIH_GE
: op
= "greaterThanEqual"; break;
2502 case WINED3DSIH_IGE
: op
= "greaterThanEqual"; break;
2503 case WINED3DSIH_UGE
: op
= "greaterThanEqual"; break;
2504 case WINED3DSIH_LT
: op
= "lessThan"; break;
2505 case WINED3DSIH_NE
: op
= "notEqual"; break;
2507 op
= "<unhandled operator>";
2508 ERR("Unhandled opcode %#x.\n", ins
->handler_idx
);
2512 shader_addline(buffer
, "uvec%u(%s(%s, %s)) * 0xffffffffu);\n",
2513 mask_size
, op
, src0_param
.param_str
, src1_param
.param_str
);
2517 switch (ins
->handler_idx
)
2519 case WINED3DSIH_EQ
: op
= "=="; break;
2520 case WINED3DSIH_GE
: op
= ">="; break;
2521 case WINED3DSIH_IGE
: op
= ">="; break;
2522 case WINED3DSIH_UGE
: op
= ">="; break;
2523 case WINED3DSIH_LT
: op
= "<"; break;
2524 case WINED3DSIH_NE
: op
= "!="; break;
2526 op
= "<unhandled operator>";
2527 ERR("Unhandled opcode %#x.\n", ins
->handler_idx
);
2531 shader_addline(buffer
, "%s %s %s ? 0xffffffffu : 0u);\n",
2532 src0_param
.param_str
, op
, src1_param
.param_str
);
2536 static void shader_glsl_imul(const struct wined3d_shader_instruction
*ins
)
2538 struct wined3d_shader_buffer
*buffer
= ins
->ctx
->buffer
;
2539 struct glsl_src_param src0_param
;
2540 struct glsl_src_param src1_param
;
2543 /* If we have ARB_gpu_shader5 or GLSL 4.0, we can use imulExtended(). If
2544 * not, we can emulate it. */
2545 if (ins
->dst
[0].reg
.type
!= WINED3DSPR_NULL
)
2546 FIXME("64-bit integer multiplies not implemented.\n");
2548 if (ins
->dst
[1].reg
.type
!= WINED3DSPR_NULL
)
2550 write_mask
= shader_glsl_append_dst_ext(buffer
, ins
, &ins
->dst
[1], ins
->dst
[1].reg
.data_type
);
2551 shader_glsl_add_src_param(ins
, &ins
->src
[0], write_mask
, &src0_param
);
2552 shader_glsl_add_src_param(ins
, &ins
->src
[1], write_mask
, &src1_param
);
2554 shader_addline(ins
->ctx
->buffer
, "%s * %s);\n",
2555 src0_param
.param_str
, src1_param
.param_str
);
2559 static void shader_glsl_udiv(const struct wined3d_shader_instruction
*ins
)
2561 struct wined3d_shader_buffer
*buffer
= ins
->ctx
->buffer
;
2562 struct glsl_src_param src0_param
, src1_param
;
2565 if (ins
->dst
[0].reg
.type
!= WINED3DSPR_NULL
)
2568 if (ins
->dst
[1].reg
.type
!= WINED3DSPR_NULL
)
2572 write_mask
= shader_glsl_get_write_mask(&ins
->dst
[0], dst_mask
);
2573 shader_glsl_add_src_param(ins
, &ins
->src
[0], write_mask
, &src0_param
);
2574 shader_glsl_add_src_param(ins
, &ins
->src
[1], write_mask
, &src1_param
);
2575 shader_addline(buffer
, "tmp0%s = %s / %s;\n",
2576 dst_mask
, src0_param
.param_str
, src1_param
.param_str
);
2578 write_mask
= shader_glsl_append_dst_ext(buffer
, ins
, &ins
->dst
[1], ins
->dst
[1].reg
.data_type
);
2579 shader_glsl_add_src_param(ins
, &ins
->src
[0], write_mask
, &src0_param
);
2580 shader_glsl_add_src_param(ins
, &ins
->src
[1], write_mask
, &src1_param
);
2581 shader_addline(buffer
, "%s %% %s));\n", src0_param
.param_str
, src1_param
.param_str
);
2583 shader_glsl_append_dst_ext(buffer
, ins
, &ins
->dst
[0], ins
->dst
[0].reg
.data_type
);
2584 shader_addline(buffer
, "tmp0%s);\n", dst_mask
);
2588 write_mask
= shader_glsl_append_dst_ext(buffer
, ins
, &ins
->dst
[0], ins
->dst
[0].reg
.data_type
);
2589 shader_glsl_add_src_param(ins
, &ins
->src
[0], write_mask
, &src0_param
);
2590 shader_glsl_add_src_param(ins
, &ins
->src
[1], write_mask
, &src1_param
);
2591 shader_addline(buffer
, "%s / %s);\n", src0_param
.param_str
, src1_param
.param_str
);
2594 else if (ins
->dst
[1].reg
.type
!= WINED3DSPR_NULL
)
2596 write_mask
= shader_glsl_append_dst_ext(buffer
, ins
, &ins
->dst
[1], ins
->dst
[1].reg
.data_type
);
2597 shader_glsl_add_src_param(ins
, &ins
->src
[0], write_mask
, &src0_param
);
2598 shader_glsl_add_src_param(ins
, &ins
->src
[1], write_mask
, &src1_param
);
2599 shader_addline(buffer
, "%s %% %s);\n", src0_param
.param_str
, src1_param
.param_str
);
2603 /* Process the WINED3DSIO_MOV opcode using GLSL (dst = src) */
2604 static void shader_glsl_mov(const struct wined3d_shader_instruction
*ins
)
2606 const struct wined3d_gl_info
*gl_info
= ins
->ctx
->gl_info
;
2607 struct wined3d_shader_buffer
*buffer
= ins
->ctx
->buffer
;
2608 struct glsl_src_param src0_param
;
2611 write_mask
= shader_glsl_append_dst(buffer
, ins
);
2612 shader_glsl_add_src_param(ins
, &ins
->src
[0], write_mask
, &src0_param
);
2614 /* In vs_1_1 WINED3DSIO_MOV can write to the address register. In later
2615 * shader versions WINED3DSIO_MOVA is used for this. */
2616 if (ins
->ctx
->reg_maps
->shader_version
.major
== 1
2617 && ins
->ctx
->reg_maps
->shader_version
.type
== WINED3D_SHADER_TYPE_VERTEX
2618 && ins
->dst
[0].reg
.type
== WINED3DSPR_ADDR
)
2620 /* This is a simple floor() */
2621 unsigned int mask_size
= shader_glsl_get_write_mask_size(write_mask
);
2622 if (mask_size
> 1) {
2623 shader_addline(buffer
, "ivec%d(floor(%s)));\n", mask_size
, src0_param
.param_str
);
2625 shader_addline(buffer
, "int(floor(%s)));\n", src0_param
.param_str
);
2628 else if(ins
->handler_idx
== WINED3DSIH_MOVA
)
2630 /* We need to *round* to the nearest int here. */
2631 unsigned int mask_size
= shader_glsl_get_write_mask_size(write_mask
);
2633 if (gl_info
->supported
[EXT_GPU_SHADER4
])
2636 shader_addline(buffer
, "ivec%d(round(%s)));\n", mask_size
, src0_param
.param_str
);
2638 shader_addline(buffer
, "int(round(%s)));\n", src0_param
.param_str
);
2643 shader_addline(buffer
, "ivec%d(floor(abs(%s) + vec%d(0.5)) * sign(%s)));\n",
2644 mask_size
, src0_param
.param_str
, mask_size
, src0_param
.param_str
);
2646 shader_addline(buffer
, "int(floor(abs(%s) + 0.5) * sign(%s)));\n",
2647 src0_param
.param_str
, src0_param
.param_str
);
2652 shader_addline(buffer
, "%s);\n", src0_param
.param_str
);
2656 /* Process the dot product operators DP3 and DP4 in GLSL (dst = dot(src0, src1)) */
2657 static void shader_glsl_dot(const struct wined3d_shader_instruction
*ins
)
2659 struct wined3d_shader_buffer
*buffer
= ins
->ctx
->buffer
;
2660 struct glsl_src_param src0_param
;
2661 struct glsl_src_param src1_param
;
2662 DWORD dst_write_mask
, src_write_mask
;
2663 unsigned int dst_size
= 0;
2665 dst_write_mask
= shader_glsl_append_dst(buffer
, ins
);
2666 dst_size
= shader_glsl_get_write_mask_size(dst_write_mask
);
2668 /* dp3 works on vec3, dp4 on vec4 */
2669 if (ins
->handler_idx
== WINED3DSIH_DP4
)
2671 src_write_mask
= WINED3DSP_WRITEMASK_ALL
;
2673 src_write_mask
= WINED3DSP_WRITEMASK_0
| WINED3DSP_WRITEMASK_1
| WINED3DSP_WRITEMASK_2
;
2676 shader_glsl_add_src_param(ins
, &ins
->src
[0], src_write_mask
, &src0_param
);
2677 shader_glsl_add_src_param(ins
, &ins
->src
[1], src_write_mask
, &src1_param
);
2680 shader_addline(buffer
, "vec%d(dot(%s, %s)));\n", dst_size
, src0_param
.param_str
, src1_param
.param_str
);
2682 shader_addline(buffer
, "dot(%s, %s));\n", src0_param
.param_str
, src1_param
.param_str
);
2686 /* Note that this instruction has some restrictions. The destination write mask
2687 * can't contain the w component, and the source swizzles have to be .xyzw */
2688 static void shader_glsl_cross(const struct wined3d_shader_instruction
*ins
)
2690 DWORD src_mask
= WINED3DSP_WRITEMASK_0
| WINED3DSP_WRITEMASK_1
| WINED3DSP_WRITEMASK_2
;
2691 struct glsl_src_param src0_param
;
2692 struct glsl_src_param src1_param
;
2695 shader_glsl_get_write_mask(&ins
->dst
[0], dst_mask
);
2696 shader_glsl_append_dst(ins
->ctx
->buffer
, ins
);
2697 shader_glsl_add_src_param(ins
, &ins
->src
[0], src_mask
, &src0_param
);
2698 shader_glsl_add_src_param(ins
, &ins
->src
[1], src_mask
, &src1_param
);
2699 shader_addline(ins
->ctx
->buffer
, "cross(%s, %s)%s);\n", src0_param
.param_str
, src1_param
.param_str
, dst_mask
);
2702 static void shader_glsl_cut(const struct wined3d_shader_instruction
*ins
)
2704 shader_addline(ins
->ctx
->buffer
, "EndPrimitive();\n");
2707 /* Process the WINED3DSIO_POW instruction in GLSL (dst = |src0|^src1)
2708 * Src0 and src1 are scalars. Note that D3D uses the absolute of src0, while
2709 * GLSL uses the value as-is. */
2710 static void shader_glsl_pow(const struct wined3d_shader_instruction
*ins
)
2712 struct wined3d_shader_buffer
*buffer
= ins
->ctx
->buffer
;
2713 struct glsl_src_param src0_param
;
2714 struct glsl_src_param src1_param
;
2715 DWORD dst_write_mask
;
2716 unsigned int dst_size
;
2718 dst_write_mask
= shader_glsl_append_dst(buffer
, ins
);
2719 dst_size
= shader_glsl_get_write_mask_size(dst_write_mask
);
2721 shader_glsl_add_src_param(ins
, &ins
->src
[0], WINED3DSP_WRITEMASK_0
, &src0_param
);
2722 shader_glsl_add_src_param(ins
, &ins
->src
[1], WINED3DSP_WRITEMASK_0
, &src1_param
);
2726 shader_addline(buffer
, "vec%u(%s == 0.0 ? 1.0 : pow(abs(%s), %s)));\n",
2727 dst_size
, src1_param
.param_str
, src0_param
.param_str
, src1_param
.param_str
);
2731 shader_addline(buffer
, "%s == 0.0 ? 1.0 : pow(abs(%s), %s));\n",
2732 src1_param
.param_str
, src0_param
.param_str
, src1_param
.param_str
);
2736 /* Map the opcode 1-to-1 to the GL code (arg->dst = instruction(src0, src1, ...) */
2737 static void shader_glsl_map2gl(const struct wined3d_shader_instruction
*ins
)
2739 struct wined3d_shader_buffer
*buffer
= ins
->ctx
->buffer
;
2740 struct glsl_src_param src_param
;
2741 const char *instruction
;
2745 /* Determine the GLSL function to use based on the opcode */
2746 /* TODO: Possibly make this a table for faster lookups */
2747 switch (ins
->handler_idx
)
2749 case WINED3DSIH_MIN
: instruction
= "min"; break;
2750 case WINED3DSIH_MAX
: instruction
= "max"; break;
2751 case WINED3DSIH_ABS
: instruction
= "abs"; break;
2752 case WINED3DSIH_FRC
: instruction
= "fract"; break;
2753 case WINED3DSIH_DSX
: instruction
= "dFdx"; break;
2754 case WINED3DSIH_DSY
: instruction
= "ycorrection.y * dFdy"; break;
2755 case WINED3DSIH_ROUND_NI
: instruction
= "floor"; break;
2756 case WINED3DSIH_SQRT
: instruction
= "sqrt"; break;
2757 default: instruction
= "";
2758 FIXME("Opcode %#x not yet handled in GLSL\n", ins
->handler_idx
);
2762 write_mask
= shader_glsl_append_dst(buffer
, ins
);
2764 shader_addline(buffer
, "%s(", instruction
);
2768 shader_glsl_add_src_param(ins
, &ins
->src
[0], write_mask
, &src_param
);
2769 shader_addline(buffer
, "%s", src_param
.param_str
);
2770 for (i
= 1; i
< ins
->src_count
; ++i
)
2772 shader_glsl_add_src_param(ins
, &ins
->src
[i
], write_mask
, &src_param
);
2773 shader_addline(buffer
, ", %s", src_param
.param_str
);
2777 shader_addline(buffer
, "));\n");
2780 static void shader_glsl_nop(const struct wined3d_shader_instruction
*ins
) {}
2782 static void shader_glsl_nrm(const struct wined3d_shader_instruction
*ins
)
2784 struct wined3d_shader_buffer
*buffer
= ins
->ctx
->buffer
;
2785 struct glsl_src_param src_param
;
2786 unsigned int mask_size
;
2790 write_mask
= shader_glsl_get_write_mask(ins
->dst
, dst_mask
);
2791 mask_size
= shader_glsl_get_write_mask_size(write_mask
);
2792 shader_glsl_add_src_param(ins
, &ins
->src
[0], write_mask
, &src_param
);
2794 shader_addline(buffer
, "tmp0.x = dot(%s, %s);\n",
2795 src_param
.param_str
, src_param
.param_str
);
2796 shader_glsl_append_dst(buffer
, ins
);
2800 shader_addline(buffer
, "tmp0.x == 0.0 ? vec%u(0.0) : (%s * inversesqrt(tmp0.x)));\n",
2801 mask_size
, src_param
.param_str
);
2805 shader_addline(buffer
, "tmp0.x == 0.0 ? 0.0 : (%s * inversesqrt(tmp0.x)));\n",
2806 src_param
.param_str
);
2810 static void shader_glsl_scalar_op(const struct wined3d_shader_instruction
*ins
)
2812 struct wined3d_shader_buffer
*buffer
= ins
->ctx
->buffer
;
2813 struct glsl_src_param src0_param
;
2814 const char *prefix
, *suffix
;
2815 unsigned int dst_size
;
2816 DWORD dst_write_mask
;
2818 dst_write_mask
= shader_glsl_append_dst(buffer
, ins
);
2819 dst_size
= shader_glsl_get_write_mask_size(dst_write_mask
);
2821 shader_glsl_add_src_param(ins
, &ins
->src
[0], WINED3DSP_WRITEMASK_3
, &src0_param
);
2823 switch (ins
->handler_idx
)
2825 case WINED3DSIH_EXP
:
2826 case WINED3DSIH_EXPP
:
2831 case WINED3DSIH_LOG
:
2832 case WINED3DSIH_LOGP
:
2833 prefix
= "log2(abs(";
2837 case WINED3DSIH_RCP
:
2842 case WINED3DSIH_RSQ
:
2843 prefix
= "inversesqrt(abs(";
2850 FIXME("Unhandled instruction %#x.\n", ins
->handler_idx
);
2855 shader_addline(buffer
, "vec%u(%s%s%s));\n", dst_size
, prefix
, src0_param
.param_str
, suffix
);
2857 shader_addline(buffer
, "%s%s%s);\n", prefix
, src0_param
.param_str
, suffix
);
2860 /** Process the WINED3DSIO_EXPP instruction in GLSL:
2861 * For shader model 1.x, do the following (and honor the writemask, so use a temporary variable):
2862 * dst.x = 2^(floor(src))
2863 * dst.y = src - floor(src)
2864 * dst.z = 2^src (partial precision is allowed, but optional)
2866 * For 2.0 shaders, just do this (honoring writemask and swizzle):
2867 * dst = 2^src; (partial precision is allowed, but optional)
2869 static void shader_glsl_expp(const struct wined3d_shader_instruction
*ins
)
2871 if (ins
->ctx
->reg_maps
->shader_version
.major
< 2)
2873 struct glsl_src_param src_param
;
2876 shader_glsl_add_src_param(ins
, &ins
->src
[0], WINED3DSP_WRITEMASK_3
, &src_param
);
2878 shader_addline(ins
->ctx
->buffer
, "tmp0.x = exp2(floor(%s));\n", src_param
.param_str
);
2879 shader_addline(ins
->ctx
->buffer
, "tmp0.y = %s - floor(%s);\n", src_param
.param_str
, src_param
.param_str
);
2880 shader_addline(ins
->ctx
->buffer
, "tmp0.z = exp2(%s);\n", src_param
.param_str
);
2881 shader_addline(ins
->ctx
->buffer
, "tmp0.w = 1.0;\n");
2883 shader_glsl_append_dst(ins
->ctx
->buffer
, ins
);
2884 shader_glsl_get_write_mask(&ins
->dst
[0], dst_mask
);
2885 shader_addline(ins
->ctx
->buffer
, "tmp0%s);\n", dst_mask
);
2889 shader_glsl_scalar_op(ins
);
2892 static void shader_glsl_to_int(const struct wined3d_shader_instruction
*ins
)
2894 struct wined3d_shader_buffer
*buffer
= ins
->ctx
->buffer
;
2895 struct glsl_src_param src_param
;
2896 unsigned int mask_size
;
2899 write_mask
= shader_glsl_append_dst(buffer
, ins
);
2900 mask_size
= shader_glsl_get_write_mask_size(write_mask
);
2901 shader_glsl_add_src_param(ins
, &ins
->src
[0], write_mask
, &src_param
);
2904 shader_addline(buffer
, "ivec%u(%s));\n", mask_size
, src_param
.param_str
);
2906 shader_addline(buffer
, "int(%s));\n", src_param
.param_str
);
2909 static void shader_glsl_to_float(const struct wined3d_shader_instruction
*ins
)
2911 struct wined3d_shader_buffer
*buffer
= ins
->ctx
->buffer
;
2912 struct glsl_src_param src_param
;
2913 unsigned int mask_size
;
2916 write_mask
= shader_glsl_append_dst(buffer
, ins
);
2917 mask_size
= shader_glsl_get_write_mask_size(write_mask
);
2918 shader_glsl_add_src_param(ins
, &ins
->src
[0], write_mask
, &src_param
);
2921 shader_addline(buffer
, "vec%u(%s));\n", mask_size
, src_param
.param_str
);
2923 shader_addline(buffer
, "float(%s));\n", src_param
.param_str
);
2926 /** Process signed comparison opcodes in GLSL. */
2927 static void shader_glsl_compare(const struct wined3d_shader_instruction
*ins
)
2929 struct glsl_src_param src0_param
;
2930 struct glsl_src_param src1_param
;
2932 unsigned int mask_size
;
2934 write_mask
= shader_glsl_append_dst(ins
->ctx
->buffer
, ins
);
2935 mask_size
= shader_glsl_get_write_mask_size(write_mask
);
2936 shader_glsl_add_src_param(ins
, &ins
->src
[0], write_mask
, &src0_param
);
2937 shader_glsl_add_src_param(ins
, &ins
->src
[1], write_mask
, &src1_param
);
2939 if (mask_size
> 1) {
2940 const char *compare
;
2942 switch(ins
->handler_idx
)
2944 case WINED3DSIH_SLT
: compare
= "lessThan"; break;
2945 case WINED3DSIH_SGE
: compare
= "greaterThanEqual"; break;
2946 default: compare
= "";
2947 FIXME("Can't handle opcode %#x\n", ins
->handler_idx
);
2950 shader_addline(ins
->ctx
->buffer
, "vec%d(%s(%s, %s)));\n", mask_size
, compare
,
2951 src0_param
.param_str
, src1_param
.param_str
);
2953 switch(ins
->handler_idx
)
2955 case WINED3DSIH_SLT
:
2956 /* Step(src0, src1) is not suitable here because if src0 == src1 SLT is supposed,
2957 * to return 0.0 but step returns 1.0 because step is not < x
2958 * An alternative is a bvec compare padded with an unused second component.
2959 * step(src1 * -1.0, src0 * -1.0) is not an option because it suffers from the same
2960 * issue. Playing with not() is not possible either because not() does not accept
2963 shader_addline(ins
->ctx
->buffer
, "(%s < %s) ? 1.0 : 0.0);\n",
2964 src0_param
.param_str
, src1_param
.param_str
);
2966 case WINED3DSIH_SGE
:
2967 /* Here we can use the step() function and safe a conditional */
2968 shader_addline(ins
->ctx
->buffer
, "step(%s, %s));\n", src1_param
.param_str
, src0_param
.param_str
);
2971 FIXME("Can't handle opcode %#x\n", ins
->handler_idx
);
2977 static void shader_glsl_conditional_move(const struct wined3d_shader_instruction
*ins
)
2979 const char *condition_prefix
, *condition_suffix
;
2980 struct wined3d_shader_dst_param dst
;
2981 struct glsl_src_param src0_param
;
2982 struct glsl_src_param src1_param
;
2983 struct glsl_src_param src2_param
;
2984 BOOL temp_destination
= FALSE
;
2985 DWORD cmp_channel
= 0;
2990 switch (ins
->handler_idx
)
2992 case WINED3DSIH_CMP
:
2993 condition_prefix
= "";
2994 condition_suffix
= " >= 0.0";
2997 case WINED3DSIH_CND
:
2998 condition_prefix
= "";
2999 condition_suffix
= " > 0.5";
3002 case WINED3DSIH_MOVC
:
3003 condition_prefix
= "bool(";
3004 condition_suffix
= ")";
3008 FIXME("Unhandled instruction %#x.\n", ins
->handler_idx
);
3009 condition_prefix
= "<unhandled prefix>";
3010 condition_suffix
= "<unhandled suffix>";
3014 if (shader_is_scalar(&ins
->dst
[0].reg
) || shader_is_scalar(&ins
->src
[0].reg
))
3016 write_mask
= shader_glsl_append_dst(ins
->ctx
->buffer
, ins
);
3017 shader_glsl_add_src_param(ins
, &ins
->src
[0], write_mask
, &src0_param
);
3018 shader_glsl_add_src_param(ins
, &ins
->src
[1], write_mask
, &src1_param
);
3019 shader_glsl_add_src_param(ins
, &ins
->src
[2], write_mask
, &src2_param
);
3021 shader_addline(ins
->ctx
->buffer
, "%s%s%s ? %s : %s);\n",
3022 condition_prefix
, src0_param
.param_str
, condition_suffix
,
3023 src1_param
.param_str
, src2_param
.param_str
);
3029 /* Splitting the instruction up in multiple lines imposes a problem:
3030 * The first lines may overwrite source parameters of the following lines.
3031 * Deal with that by using a temporary destination register if needed. */
3032 if ((ins
->src
[0].reg
.idx
[0].offset
== dst
.reg
.idx
[0].offset
3033 && ins
->src
[0].reg
.type
== dst
.reg
.type
)
3034 || (ins
->src
[1].reg
.idx
[0].offset
== dst
.reg
.idx
[0].offset
3035 && ins
->src
[1].reg
.type
== dst
.reg
.type
)
3036 || (ins
->src
[2].reg
.idx
[0].offset
== dst
.reg
.idx
[0].offset
3037 && ins
->src
[2].reg
.type
== dst
.reg
.type
))
3038 temp_destination
= TRUE
;
3040 /* Cycle through all source0 channels. */
3041 for (i
= 0; i
< 4; ++i
)
3044 /* Find the destination channels which use the current source0 channel. */
3045 for (j
= 0; j
< 4; ++j
)
3047 if (((ins
->src
[0].swizzle
>> (2 * j
)) & 0x3) == i
)
3049 write_mask
|= WINED3DSP_WRITEMASK_0
<< j
;
3050 cmp_channel
= WINED3DSP_WRITEMASK_0
<< j
;
3053 dst
.write_mask
= ins
->dst
[0].write_mask
& write_mask
;
3055 if (temp_destination
)
3057 if (!(write_mask
= shader_glsl_get_write_mask(&dst
, mask_char
)))
3059 shader_addline(ins
->ctx
->buffer
, "tmp0%s = (", mask_char
);
3061 else if (!(write_mask
= shader_glsl_append_dst_ext(ins
->ctx
->buffer
, ins
, &dst
, dst
.reg
.data_type
)))
3064 shader_glsl_add_src_param(ins
, &ins
->src
[0], cmp_channel
, &src0_param
);
3065 shader_glsl_add_src_param(ins
, &ins
->src
[1], write_mask
, &src1_param
);
3066 shader_glsl_add_src_param(ins
, &ins
->src
[2], write_mask
, &src2_param
);
3068 shader_addline(ins
->ctx
->buffer
, "%s%s%s ? %s : %s);\n",
3069 condition_prefix
, src0_param
.param_str
, condition_suffix
,
3070 src1_param
.param_str
, src2_param
.param_str
);
3073 if (temp_destination
)
3075 shader_glsl_get_write_mask(&ins
->dst
[0], mask_char
);
3076 shader_glsl_append_dst(ins
->ctx
->buffer
, ins
);
3077 shader_addline(ins
->ctx
->buffer
, "tmp0%s);\n", mask_char
);
3081 /** Process the CND opcode in GLSL (dst = (src0 > 0.5) ? src1 : src2) */
3082 /* For ps 1.1-1.3, only a single component of src0 is used. For ps 1.4
3083 * the compare is done per component of src0. */
3084 static void shader_glsl_cnd(const struct wined3d_shader_instruction
*ins
)
3086 struct glsl_src_param src0_param
;
3087 struct glsl_src_param src1_param
;
3088 struct glsl_src_param src2_param
;
3090 DWORD shader_version
= WINED3D_SHADER_VERSION(ins
->ctx
->reg_maps
->shader_version
.major
,
3091 ins
->ctx
->reg_maps
->shader_version
.minor
);
3093 if (shader_version
< WINED3D_SHADER_VERSION(1, 4))
3095 write_mask
= shader_glsl_append_dst(ins
->ctx
->buffer
, ins
);
3096 shader_glsl_add_src_param(ins
, &ins
->src
[0], WINED3DSP_WRITEMASK_0
, &src0_param
);
3097 shader_glsl_add_src_param(ins
, &ins
->src
[1], write_mask
, &src1_param
);
3098 shader_glsl_add_src_param(ins
, &ins
->src
[2], write_mask
, &src2_param
);
3100 if (ins
->coissue
&& ins
->dst
->write_mask
!= WINED3DSP_WRITEMASK_3
)
3101 shader_addline(ins
->ctx
->buffer
, "%s /* COISSUE! */);\n", src1_param
.param_str
);
3103 shader_addline(ins
->ctx
->buffer
, "%s > 0.5 ? %s : %s);\n",
3104 src0_param
.param_str
, src1_param
.param_str
, src2_param
.param_str
);
3108 shader_glsl_conditional_move(ins
);
3111 /** GLSL code generation for WINED3DSIO_MAD: Multiply the first 2 opcodes, then add the last */
3112 static void shader_glsl_mad(const struct wined3d_shader_instruction
*ins
)
3114 struct glsl_src_param src0_param
;
3115 struct glsl_src_param src1_param
;
3116 struct glsl_src_param src2_param
;
3119 write_mask
= shader_glsl_append_dst(ins
->ctx
->buffer
, ins
);
3120 shader_glsl_add_src_param(ins
, &ins
->src
[0], write_mask
, &src0_param
);
3121 shader_glsl_add_src_param(ins
, &ins
->src
[1], write_mask
, &src1_param
);
3122 shader_glsl_add_src_param(ins
, &ins
->src
[2], write_mask
, &src2_param
);
3123 shader_addline(ins
->ctx
->buffer
, "(%s * %s) + %s);\n",
3124 src0_param
.param_str
, src1_param
.param_str
, src2_param
.param_str
);
3127 /* Handles transforming all WINED3DSIO_M?x? opcodes for
3128 Vertex shaders to GLSL codes */
3129 static void shader_glsl_mnxn(const struct wined3d_shader_instruction
*ins
)
3132 int nComponents
= 0;
3133 struct wined3d_shader_dst_param tmp_dst
= {{0}};
3134 struct wined3d_shader_src_param tmp_src
[2] = {{{0}}};
3135 struct wined3d_shader_instruction tmp_ins
;
3137 memset(&tmp_ins
, 0, sizeof(tmp_ins
));
3139 /* Set constants for the temporary argument */
3140 tmp_ins
.ctx
= ins
->ctx
;
3141 tmp_ins
.dst_count
= 1;
3142 tmp_ins
.dst
= &tmp_dst
;
3143 tmp_ins
.src_count
= 2;
3144 tmp_ins
.src
= tmp_src
;
3146 switch(ins
->handler_idx
)
3148 case WINED3DSIH_M4x4
:
3150 tmp_ins
.handler_idx
= WINED3DSIH_DP4
;
3152 case WINED3DSIH_M4x3
:
3154 tmp_ins
.handler_idx
= WINED3DSIH_DP4
;
3156 case WINED3DSIH_M3x4
:
3158 tmp_ins
.handler_idx
= WINED3DSIH_DP3
;
3160 case WINED3DSIH_M3x3
:
3162 tmp_ins
.handler_idx
= WINED3DSIH_DP3
;
3164 case WINED3DSIH_M3x2
:
3166 tmp_ins
.handler_idx
= WINED3DSIH_DP3
;
3172 tmp_dst
= ins
->dst
[0];
3173 tmp_src
[0] = ins
->src
[0];
3174 tmp_src
[1] = ins
->src
[1];
3175 for (i
= 0; i
< nComponents
; ++i
)
3177 tmp_dst
.write_mask
= WINED3DSP_WRITEMASK_0
<< i
;
3178 shader_glsl_dot(&tmp_ins
);
3179 ++tmp_src
[1].reg
.idx
[0].offset
;
3184 The LRP instruction performs a component-wise linear interpolation
3185 between the second and third operands using the first operand as the
3186 blend factor. Equation: (dst = src2 + src0 * (src1 - src2))
3187 This is equivalent to mix(src2, src1, src0);
3189 static void shader_glsl_lrp(const struct wined3d_shader_instruction
*ins
)
3191 struct glsl_src_param src0_param
;
3192 struct glsl_src_param src1_param
;
3193 struct glsl_src_param src2_param
;
3196 write_mask
= shader_glsl_append_dst(ins
->ctx
->buffer
, ins
);
3198 shader_glsl_add_src_param(ins
, &ins
->src
[0], write_mask
, &src0_param
);
3199 shader_glsl_add_src_param(ins
, &ins
->src
[1], write_mask
, &src1_param
);
3200 shader_glsl_add_src_param(ins
, &ins
->src
[2], write_mask
, &src2_param
);
3202 shader_addline(ins
->ctx
->buffer
, "mix(%s, %s, %s));\n",
3203 src2_param
.param_str
, src1_param
.param_str
, src0_param
.param_str
);
3206 /** Process the WINED3DSIO_LIT instruction in GLSL:
3207 * dst.x = dst.w = 1.0
3208 * dst.y = (src0.x > 0) ? src0.x
3209 * dst.z = (src0.x > 0) ? ((src0.y > 0) ? pow(src0.y, src.w) : 0) : 0
3210 * where src.w is clamped at +- 128
3212 static void shader_glsl_lit(const struct wined3d_shader_instruction
*ins
)
3214 struct glsl_src_param src0_param
;
3215 struct glsl_src_param src1_param
;
3216 struct glsl_src_param src3_param
;
3219 shader_glsl_append_dst(ins
->ctx
->buffer
, ins
);
3220 shader_glsl_get_write_mask(&ins
->dst
[0], dst_mask
);
3222 shader_glsl_add_src_param(ins
, &ins
->src
[0], WINED3DSP_WRITEMASK_0
, &src0_param
);
3223 shader_glsl_add_src_param(ins
, &ins
->src
[0], WINED3DSP_WRITEMASK_1
, &src1_param
);
3224 shader_glsl_add_src_param(ins
, &ins
->src
[0], WINED3DSP_WRITEMASK_3
, &src3_param
);
3226 /* The sdk specifies the instruction like this
3228 * if(src.x > 0.0) dst.y = src.x
3230 * if(src.x > 0.0 && src.y > 0.0) dst.z = pow(src.y, power);
3233 * (where power = src.w clamped between -128 and 128)
3235 * Obviously that has quite a few conditionals in it which we don't like. So the first step is this:
3236 * dst.x = 1.0 ... No further explanation needed
3237 * dst.y = max(src.y, 0.0); ... If x < 0.0, use 0.0, otherwise x. Same as the conditional
3238 * dst.z = x > 0.0 ? pow(max(y, 0.0), p) : 0; ... 0 ^ power is 0, and otherwise we use y anyway
3239 * dst.w = 1.0. ... Nothing fancy.
3241 * So we still have one conditional in there. So do this:
3242 * dst.z = pow(max(0.0, src.y) * step(0.0, src.x), power);
3244 * 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),
3245 * 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.
3246 * if both x and y are > 0, we get pow(y * 1.0, power), as it is supposed to.
3248 * Unfortunately pow(0.0 ^ 0.0) returns NaN on most GPUs, but lit with src.y = 0 and src.w = 0 returns
3249 * a non-NaN value in dst.z. What we return doesn't matter, as long as it is not NaN. Return 0, which is
3250 * what all Windows HW drivers and GL_ARB_vertex_program's LIT do.
3252 shader_addline(ins
->ctx
->buffer
,
3253 "vec4(1.0, max(%s, 0.0), %s == 0.0 ? 0.0 : "
3254 "pow(max(0.0, %s) * step(0.0, %s), clamp(%s, -128.0, 128.0)), 1.0)%s);\n",
3255 src0_param
.param_str
, src3_param
.param_str
, src1_param
.param_str
,
3256 src0_param
.param_str
, src3_param
.param_str
, dst_mask
);
3259 /** Process the WINED3DSIO_DST instruction in GLSL:
3261 * dst.y = src0.x * src0.y
3265 static void shader_glsl_dst(const struct wined3d_shader_instruction
*ins
)
3267 struct glsl_src_param src0y_param
;
3268 struct glsl_src_param src0z_param
;
3269 struct glsl_src_param src1y_param
;
3270 struct glsl_src_param src1w_param
;
3273 shader_glsl_append_dst(ins
->ctx
->buffer
, ins
);
3274 shader_glsl_get_write_mask(&ins
->dst
[0], dst_mask
);
3276 shader_glsl_add_src_param(ins
, &ins
->src
[0], WINED3DSP_WRITEMASK_1
, &src0y_param
);
3277 shader_glsl_add_src_param(ins
, &ins
->src
[0], WINED3DSP_WRITEMASK_2
, &src0z_param
);
3278 shader_glsl_add_src_param(ins
, &ins
->src
[1], WINED3DSP_WRITEMASK_1
, &src1y_param
);
3279 shader_glsl_add_src_param(ins
, &ins
->src
[1], WINED3DSP_WRITEMASK_3
, &src1w_param
);
3281 shader_addline(ins
->ctx
->buffer
, "vec4(1.0, %s * %s, %s, %s))%s;\n",
3282 src0y_param
.param_str
, src1y_param
.param_str
, src0z_param
.param_str
, src1w_param
.param_str
, dst_mask
);
3285 /** Process the WINED3DSIO_SINCOS instruction in GLSL:
3286 * VS 2.0 requires that specific cosine and sine constants be passed to this instruction so the hardware
3287 * can handle it. But, these functions are built-in for GLSL, so we can just ignore the last 2 params.
3289 * dst.x = cos(src0.?)
3290 * dst.y = sin(src0.?)
3294 static void shader_glsl_sincos(const struct wined3d_shader_instruction
*ins
)
3296 struct wined3d_shader_buffer
*buffer
= ins
->ctx
->buffer
;
3297 struct glsl_src_param src0_param
;
3300 if (ins
->ctx
->reg_maps
->shader_version
.major
< 4)
3302 shader_glsl_add_src_param(ins
, &ins
->src
[0], WINED3DSP_WRITEMASK_0
, &src0_param
);
3304 write_mask
= shader_glsl_append_dst(buffer
, ins
);
3307 case WINED3DSP_WRITEMASK_0
:
3308 shader_addline(buffer
, "cos(%s));\n", src0_param
.param_str
);
3311 case WINED3DSP_WRITEMASK_1
:
3312 shader_addline(buffer
, "sin(%s));\n", src0_param
.param_str
);
3315 case (WINED3DSP_WRITEMASK_0
| WINED3DSP_WRITEMASK_1
):
3316 shader_addline(buffer
, "vec2(cos(%s), sin(%s)));\n",
3317 src0_param
.param_str
, src0_param
.param_str
);
3321 ERR("Write mask should be .x, .y or .xy\n");
3328 if (ins
->dst
[0].reg
.type
!= WINED3DSPR_NULL
)
3331 if (ins
->dst
[1].reg
.type
!= WINED3DSPR_NULL
)
3335 write_mask
= shader_glsl_get_write_mask(&ins
->dst
[0], dst_mask
);
3336 shader_glsl_add_src_param(ins
, &ins
->src
[0], write_mask
, &src0_param
);
3337 shader_addline(buffer
, "tmp0%s = sin(%s);\n", dst_mask
, src0_param
.param_str
);
3339 write_mask
= shader_glsl_append_dst_ext(buffer
, ins
, &ins
->dst
[1], ins
->dst
[1].reg
.data_type
);
3340 shader_glsl_add_src_param(ins
, &ins
->src
[0], write_mask
, &src0_param
);
3341 shader_addline(buffer
, "cos(%s));\n", src0_param
.param_str
);
3343 shader_glsl_append_dst_ext(buffer
, ins
, &ins
->dst
[0], ins
->dst
[0].reg
.data_type
);
3344 shader_addline(buffer
, "tmp0%s);\n", dst_mask
);
3348 write_mask
= shader_glsl_append_dst_ext(buffer
, ins
, &ins
->dst
[0], ins
->dst
[0].reg
.data_type
);
3349 shader_glsl_add_src_param(ins
, &ins
->src
[0], write_mask
, &src0_param
);
3350 shader_addline(buffer
, "sin(%s));\n", src0_param
.param_str
);
3353 else if (ins
->dst
[1].reg
.type
!= WINED3DSPR_NULL
)
3355 write_mask
= shader_glsl_append_dst_ext(buffer
, ins
, &ins
->dst
[1], ins
->dst
[1].reg
.data_type
);
3356 shader_glsl_add_src_param(ins
, &ins
->src
[0], write_mask
, &src0_param
);
3357 shader_addline(buffer
, "cos(%s));\n", src0_param
.param_str
);
3361 /* sgn in vs_2_0 has 2 extra parameters(registers for temporary storage) which we don't use
3362 * here. But those extra parameters require a dedicated function for sgn, since map2gl would
3363 * generate invalid code
3365 static void shader_glsl_sgn(const struct wined3d_shader_instruction
*ins
)
3367 struct glsl_src_param src0_param
;
3370 write_mask
= shader_glsl_append_dst(ins
->ctx
->buffer
, ins
);
3371 shader_glsl_add_src_param(ins
, &ins
->src
[0], write_mask
, &src0_param
);
3373 shader_addline(ins
->ctx
->buffer
, "sign(%s));\n", src0_param
.param_str
);
3376 /** Process the WINED3DSIO_LOOP instruction in GLSL:
3377 * Start a for() loop where src1.y is the initial value of aL,
3378 * increment aL by src1.z for a total of src1.x iterations.
3379 * Need to use a temporary variable for this operation.
3381 /* FIXME: I don't think nested loops will work correctly this way. */
3382 static void shader_glsl_loop(const struct wined3d_shader_instruction
*ins
)
3384 struct wined3d_shader_loop_state
*loop_state
= ins
->ctx
->loop_state
;
3385 struct wined3d_shader_buffer
*buffer
= ins
->ctx
->buffer
;
3386 const struct wined3d_shader
*shader
= ins
->ctx
->shader
;
3387 const struct wined3d_shader_lconst
*constant
;
3388 struct glsl_src_param src1_param
;
3389 const DWORD
*control_values
= NULL
;
3391 if (ins
->ctx
->reg_maps
->shader_version
.major
< 4)
3393 shader_glsl_add_src_param(ins
, &ins
->src
[1], WINED3DSP_WRITEMASK_ALL
, &src1_param
);
3395 /* Try to hardcode the loop control parameters if possible. Direct3D 9
3396 * class hardware doesn't support real varying indexing, but Microsoft
3397 * designed this feature for Shader model 2.x+. If the loop control is
3398 * known at compile time, the GLSL compiler can unroll the loop, and
3399 * replace indirect addressing with direct addressing. */
3400 if (ins
->src
[1].reg
.type
== WINED3DSPR_CONSTINT
)
3402 LIST_FOR_EACH_ENTRY(constant
, &shader
->constantsI
, struct wined3d_shader_lconst
, entry
)
3404 if (constant
->idx
== ins
->src
[1].reg
.idx
[0].offset
)
3406 control_values
= constant
->value
;
3414 struct wined3d_shader_loop_control loop_control
;
3415 loop_control
.count
= control_values
[0];
3416 loop_control
.start
= control_values
[1];
3417 loop_control
.step
= (int)control_values
[2];
3419 if (loop_control
.step
> 0)
3421 shader_addline(buffer
, "for (aL%u = %u; aL%u < (%u * %d + %u); aL%u += %d)\n{\n",
3422 loop_state
->current_depth
, loop_control
.start
,
3423 loop_state
->current_depth
, loop_control
.count
, loop_control
.step
, loop_control
.start
,
3424 loop_state
->current_depth
, loop_control
.step
);
3426 else if (loop_control
.step
< 0)
3428 shader_addline(buffer
, "for (aL%u = %u; aL%u > (%u * %d + %u); aL%u += %d)\n{\n",
3429 loop_state
->current_depth
, loop_control
.start
,
3430 loop_state
->current_depth
, loop_control
.count
, loop_control
.step
, loop_control
.start
,
3431 loop_state
->current_depth
, loop_control
.step
);
3435 shader_addline(buffer
, "for (aL%u = %u, tmpInt%u = 0; tmpInt%u < %u; tmpInt%u++)\n{\n",
3436 loop_state
->current_depth
, loop_control
.start
, loop_state
->current_depth
,
3437 loop_state
->current_depth
, loop_control
.count
,
3438 loop_state
->current_depth
);
3443 shader_addline(buffer
, "for (tmpInt%u = 0, aL%u = %s.y; tmpInt%u < %s.x; tmpInt%u++, aL%u += %s.z)\n{\n",
3444 loop_state
->current_depth
, loop_state
->current_reg
,
3445 src1_param
.reg_name
, loop_state
->current_depth
, src1_param
.reg_name
,
3446 loop_state
->current_depth
, loop_state
->current_reg
, src1_param
.reg_name
);
3449 ++loop_state
->current_reg
;
3453 shader_addline(buffer
, "for (;;)\n{\n");
3456 ++loop_state
->current_depth
;
3459 static void shader_glsl_end(const struct wined3d_shader_instruction
*ins
)
3461 struct wined3d_shader_loop_state
*loop_state
= ins
->ctx
->loop_state
;
3463 shader_addline(ins
->ctx
->buffer
, "}\n");
3465 if (ins
->handler_idx
== WINED3DSIH_ENDLOOP
)
3467 --loop_state
->current_depth
;
3468 --loop_state
->current_reg
;
3471 if (ins
->handler_idx
== WINED3DSIH_ENDREP
)
3473 --loop_state
->current_depth
;
3477 static void shader_glsl_rep(const struct wined3d_shader_instruction
*ins
)
3479 const struct wined3d_shader
*shader
= ins
->ctx
->shader
;
3480 struct wined3d_shader_loop_state
*loop_state
= ins
->ctx
->loop_state
;
3481 const struct wined3d_shader_lconst
*constant
;
3482 struct glsl_src_param src0_param
;
3483 const DWORD
*control_values
= NULL
;
3485 /* Try to hardcode local values to help the GLSL compiler to unroll and optimize the loop */
3486 if (ins
->src
[0].reg
.type
== WINED3DSPR_CONSTINT
)
3488 LIST_FOR_EACH_ENTRY(constant
, &shader
->constantsI
, struct wined3d_shader_lconst
, entry
)
3490 if (constant
->idx
== ins
->src
[0].reg
.idx
[0].offset
)
3492 control_values
= constant
->value
;
3500 shader_addline(ins
->ctx
->buffer
, "for (tmpInt%d = 0; tmpInt%d < %d; tmpInt%d++) {\n",
3501 loop_state
->current_depth
, loop_state
->current_depth
,
3502 control_values
[0], loop_state
->current_depth
);
3506 shader_glsl_add_src_param(ins
, &ins
->src
[0], WINED3DSP_WRITEMASK_0
, &src0_param
);
3507 shader_addline(ins
->ctx
->buffer
, "for (tmpInt%d = 0; tmpInt%d < %s; tmpInt%d++) {\n",
3508 loop_state
->current_depth
, loop_state
->current_depth
,
3509 src0_param
.param_str
, loop_state
->current_depth
);
3512 ++loop_state
->current_depth
;
3515 static void shader_glsl_if(const struct wined3d_shader_instruction
*ins
)
3517 struct glsl_src_param src0_param
;
3519 shader_glsl_add_src_param(ins
, &ins
->src
[0], WINED3DSP_WRITEMASK_0
, &src0_param
);
3520 shader_addline(ins
->ctx
->buffer
, "if (bool(%s)) {\n", src0_param
.param_str
);
3523 static void shader_glsl_ifc(const struct wined3d_shader_instruction
*ins
)
3525 struct glsl_src_param src0_param
;
3526 struct glsl_src_param src1_param
;
3528 shader_glsl_add_src_param(ins
, &ins
->src
[0], WINED3DSP_WRITEMASK_0
, &src0_param
);
3529 shader_glsl_add_src_param(ins
, &ins
->src
[1], WINED3DSP_WRITEMASK_0
, &src1_param
);
3531 shader_addline(ins
->ctx
->buffer
, "if (%s %s %s) {\n",
3532 src0_param
.param_str
, shader_glsl_get_rel_op(ins
->flags
), src1_param
.param_str
);
3535 static void shader_glsl_else(const struct wined3d_shader_instruction
*ins
)
3537 shader_addline(ins
->ctx
->buffer
, "} else {\n");
3540 static void shader_glsl_emit(const struct wined3d_shader_instruction
*ins
)
3542 shader_addline(ins
->ctx
->buffer
, "EmitVertex();\n");
3545 static void shader_glsl_break(const struct wined3d_shader_instruction
*ins
)
3547 shader_addline(ins
->ctx
->buffer
, "break;\n");
3550 /* FIXME: According to MSDN the compare is done per component. */
3551 static void shader_glsl_breakc(const struct wined3d_shader_instruction
*ins
)
3553 struct glsl_src_param src0_param
;
3554 struct glsl_src_param src1_param
;
3556 shader_glsl_add_src_param(ins
, &ins
->src
[0], WINED3DSP_WRITEMASK_0
, &src0_param
);
3557 shader_glsl_add_src_param(ins
, &ins
->src
[1], WINED3DSP_WRITEMASK_0
, &src1_param
);
3559 shader_addline(ins
->ctx
->buffer
, "if (%s %s %s) break;\n",
3560 src0_param
.param_str
, shader_glsl_get_rel_op(ins
->flags
), src1_param
.param_str
);
3563 static void shader_glsl_breakp(const struct wined3d_shader_instruction
*ins
)
3565 struct glsl_src_param src_param
;
3567 shader_glsl_add_src_param(ins
, &ins
->src
[0], WINED3DSP_WRITEMASK_0
, &src_param
);
3568 shader_addline(ins
->ctx
->buffer
, "if (bool(%s)) break;\n", src_param
.param_str
);
3571 static void shader_glsl_label(const struct wined3d_shader_instruction
*ins
)
3573 shader_addline(ins
->ctx
->buffer
, "}\n");
3574 shader_addline(ins
->ctx
->buffer
, "void subroutine%u()\n{\n", ins
->src
[0].reg
.idx
[0].offset
);
3577 static void shader_glsl_call(const struct wined3d_shader_instruction
*ins
)
3579 shader_addline(ins
->ctx
->buffer
, "subroutine%u();\n", ins
->src
[0].reg
.idx
[0].offset
);
3582 static void shader_glsl_callnz(const struct wined3d_shader_instruction
*ins
)
3584 struct glsl_src_param src1_param
;
3586 shader_glsl_add_src_param(ins
, &ins
->src
[1], WINED3DSP_WRITEMASK_0
, &src1_param
);
3587 shader_addline(ins
->ctx
->buffer
, "if (%s) subroutine%u();\n",
3588 src1_param
.param_str
, ins
->src
[0].reg
.idx
[0].offset
);
3591 static void shader_glsl_ret(const struct wined3d_shader_instruction
*ins
)
3593 /* No-op. The closing } is written when a new function is started, and at the end of the shader. This
3594 * function only suppresses the unhandled instruction warning
3598 /*********************************************
3599 * Pixel Shader Specific Code begins here
3600 ********************************************/
3601 static void shader_glsl_tex(const struct wined3d_shader_instruction
*ins
)
3603 DWORD shader_version
= WINED3D_SHADER_VERSION(ins
->ctx
->reg_maps
->shader_version
.major
,
3604 ins
->ctx
->reg_maps
->shader_version
.minor
);
3605 struct glsl_sample_function sample_function
;
3606 DWORD sample_flags
= 0;
3608 DWORD mask
= 0, swizzle
;
3609 const struct shader_glsl_ctx_priv
*priv
= ins
->ctx
->backend_data
;
3611 /* 1.0-1.4: Use destination register as sampler source.
3612 * 2.0+: Use provided sampler source. */
3613 if (shader_version
< WINED3D_SHADER_VERSION(2,0))
3614 resource_idx
= ins
->dst
[0].reg
.idx
[0].offset
;
3616 resource_idx
= ins
->src
[1].reg
.idx
[0].offset
;
3618 if (shader_version
< WINED3D_SHADER_VERSION(1,4))
3620 DWORD flags
= (priv
->cur_ps_args
->tex_transform
>> resource_idx
* WINED3D_PSARGS_TEXTRANSFORM_SHIFT
)
3621 & WINED3D_PSARGS_TEXTRANSFORM_MASK
;
3622 enum wined3d_shader_resource_type resource_type
= ins
->ctx
->reg_maps
->resource_info
[resource_idx
].type
;
3624 /* Projected cube textures don't make a lot of sense, the resulting coordinates stay the same. */
3625 if (flags
& WINED3D_PSARGS_PROJECTED
&& resource_type
!= WINED3D_SHADER_RESOURCE_TEXTURE_CUBE
)
3627 sample_flags
|= WINED3D_GLSL_SAMPLE_PROJECTED
;
3628 switch (flags
& ~WINED3D_PSARGS_PROJECTED
)
3630 case WINED3D_TTFF_COUNT1
:
3631 FIXME("WINED3D_TTFF_PROJECTED with WINED3D_TTFF_COUNT1?\n");
3633 case WINED3D_TTFF_COUNT2
:
3634 mask
= WINED3DSP_WRITEMASK_1
;
3636 case WINED3D_TTFF_COUNT3
:
3637 mask
= WINED3DSP_WRITEMASK_2
;
3639 case WINED3D_TTFF_COUNT4
:
3640 case WINED3D_TTFF_DISABLE
:
3641 mask
= WINED3DSP_WRITEMASK_3
;
3646 else if (shader_version
< WINED3D_SHADER_VERSION(2,0))
3648 enum wined3d_shader_src_modifier src_mod
= ins
->src
[0].modifiers
;
3650 if (src_mod
== WINED3DSPSM_DZ
) {
3651 sample_flags
|= WINED3D_GLSL_SAMPLE_PROJECTED
;
3652 mask
= WINED3DSP_WRITEMASK_2
;
3653 } else if (src_mod
== WINED3DSPSM_DW
) {
3654 sample_flags
|= WINED3D_GLSL_SAMPLE_PROJECTED
;
3655 mask
= WINED3DSP_WRITEMASK_3
;
3660 if ((ins
->flags
& WINED3DSI_TEXLD_PROJECT
)
3661 && ins
->ctx
->reg_maps
->resource_info
[resource_idx
].type
!= WINED3D_SHADER_RESOURCE_TEXTURE_CUBE
)
3663 /* ps 2.0 texldp instruction always divides by the fourth component. */
3664 sample_flags
|= WINED3D_GLSL_SAMPLE_PROJECTED
;
3665 mask
= WINED3DSP_WRITEMASK_3
;
3669 if (priv
->cur_ps_args
->np2_fixup
& (1 << resource_idx
))
3670 sample_flags
|= WINED3D_GLSL_SAMPLE_NPOT
;
3672 shader_glsl_get_sample_function(ins
->ctx
, resource_idx
, sample_flags
, &sample_function
);
3673 mask
|= sample_function
.coord_mask
;
3675 if (shader_version
< WINED3D_SHADER_VERSION(2,0)) swizzle
= WINED3DSP_NOSWIZZLE
;
3676 else swizzle
= ins
->src
[1].swizzle
;
3678 /* 1.0-1.3: Use destination register as coordinate source.
3679 1.4+: Use provided coordinate source register. */
3680 if (shader_version
< WINED3D_SHADER_VERSION(1,4))
3683 shader_glsl_write_mask_to_str(mask
, coord_mask
);
3684 shader_glsl_gen_sample_code(ins
, resource_idx
, &sample_function
, swizzle
, NULL
, NULL
, NULL
,
3685 "T%u%s", resource_idx
, coord_mask
);
3689 struct glsl_src_param coord_param
;
3690 shader_glsl_add_src_param(ins
, &ins
->src
[0], mask
, &coord_param
);
3691 if (ins
->flags
& WINED3DSI_TEXLD_BIAS
)
3693 struct glsl_src_param bias
;
3694 shader_glsl_add_src_param(ins
, &ins
->src
[0], WINED3DSP_WRITEMASK_3
, &bias
);
3695 shader_glsl_gen_sample_code(ins
, resource_idx
, &sample_function
, swizzle
, NULL
, NULL
, bias
.param_str
,
3696 "%s", coord_param
.param_str
);
3698 shader_glsl_gen_sample_code(ins
, resource_idx
, &sample_function
, swizzle
, NULL
, NULL
, NULL
,
3699 "%s", coord_param
.param_str
);
3704 static void shader_glsl_texldd(const struct wined3d_shader_instruction
*ins
)
3706 const struct wined3d_gl_info
*gl_info
= ins
->ctx
->gl_info
;
3707 struct glsl_src_param coord_param
, dx_param
, dy_param
;
3708 DWORD sample_flags
= WINED3D_GLSL_SAMPLE_GRAD
;
3709 struct glsl_sample_function sample_function
;
3711 DWORD swizzle
= ins
->src
[1].swizzle
;
3712 const struct shader_glsl_ctx_priv
*priv
= ins
->ctx
->backend_data
;
3714 if (!gl_info
->supported
[ARB_SHADER_TEXTURE_LOD
] && !gl_info
->supported
[EXT_GPU_SHADER4
])
3716 FIXME("texldd used, but not supported by hardware. Falling back to regular tex\n");
3717 shader_glsl_tex(ins
);
3721 sampler_idx
= ins
->src
[1].reg
.idx
[0].offset
;
3722 if (priv
->cur_ps_args
->np2_fixup
& (1 << sampler_idx
))
3723 sample_flags
|= WINED3D_GLSL_SAMPLE_NPOT
;
3725 shader_glsl_get_sample_function(ins
->ctx
, sampler_idx
, sample_flags
, &sample_function
);
3726 shader_glsl_add_src_param(ins
, &ins
->src
[0], sample_function
.coord_mask
, &coord_param
);
3727 shader_glsl_add_src_param(ins
, &ins
->src
[2], sample_function
.coord_mask
, &dx_param
);
3728 shader_glsl_add_src_param(ins
, &ins
->src
[3], sample_function
.coord_mask
, &dy_param
);
3730 shader_glsl_gen_sample_code(ins
, sampler_idx
, &sample_function
, swizzle
, dx_param
.param_str
, dy_param
.param_str
, NULL
,
3731 "%s", coord_param
.param_str
);
3734 static void shader_glsl_texldl(const struct wined3d_shader_instruction
*ins
)
3736 const struct wined3d_gl_info
*gl_info
= ins
->ctx
->gl_info
;
3737 struct glsl_src_param coord_param
, lod_param
;
3738 DWORD sample_flags
= WINED3D_GLSL_SAMPLE_LOD
;
3739 struct glsl_sample_function sample_function
;
3741 DWORD swizzle
= ins
->src
[1].swizzle
;
3742 const struct shader_glsl_ctx_priv
*priv
= ins
->ctx
->backend_data
;
3744 sampler_idx
= ins
->src
[1].reg
.idx
[0].offset
;
3745 if (ins
->ctx
->reg_maps
->shader_version
.type
== WINED3D_SHADER_TYPE_PIXEL
3746 && priv
->cur_ps_args
->np2_fixup
& (1 << sampler_idx
))
3747 sample_flags
|= WINED3D_GLSL_SAMPLE_NPOT
;
3749 shader_glsl_get_sample_function(ins
->ctx
, sampler_idx
, sample_flags
, &sample_function
);
3750 shader_glsl_add_src_param(ins
, &ins
->src
[0], sample_function
.coord_mask
, &coord_param
);
3752 shader_glsl_add_src_param(ins
, &ins
->src
[0], WINED3DSP_WRITEMASK_3
, &lod_param
);
3754 if (!gl_info
->supported
[ARB_SHADER_TEXTURE_LOD
] && !gl_info
->supported
[EXT_GPU_SHADER4
]
3755 && ins
->ctx
->reg_maps
->shader_version
.type
== WINED3D_SHADER_TYPE_PIXEL
)
3757 /* Plain GLSL only supports Lod sampling functions in vertex shaders.
3758 * However, the NVIDIA drivers allow them in fragment shaders as well,
3759 * even without the appropriate extension. */
3760 WARN("Using %s in fragment shader.\n", sample_function
.name
);
3762 shader_glsl_gen_sample_code(ins
, sampler_idx
, &sample_function
, swizzle
, NULL
, NULL
, lod_param
.param_str
,
3763 "%s", coord_param
.param_str
);
3766 static unsigned int shader_glsl_find_sampler(const struct wined3d_shader_sampler_map
*sampler_map
,
3767 unsigned int resource_idx
, unsigned int sampler_idx
)
3769 struct wined3d_shader_sampler_map_entry
*entries
= sampler_map
->entries
;
3772 for (i
= 0; i
< sampler_map
->count
; ++i
)
3774 if (entries
[i
].resource_idx
== resource_idx
&& entries
[i
].sampler_idx
== sampler_idx
)
3775 return entries
[i
].bind_idx
;
3778 ERR("No GLSL sampler found for resource %u / sampler %u.\n", resource_idx
, sampler_idx
);
3783 static void shader_glsl_sample(const struct wined3d_shader_instruction
*ins
)
3785 struct glsl_sample_function sample_function
;
3786 struct glsl_src_param coord_param
;
3787 unsigned int sampler_idx
;
3789 shader_glsl_get_sample_function(ins
->ctx
, ins
->src
[1].reg
.idx
[0].offset
, 0, &sample_function
);
3790 shader_glsl_add_src_param(ins
, &ins
->src
[0], sample_function
.coord_mask
, &coord_param
);
3791 sampler_idx
= shader_glsl_find_sampler(&ins
->ctx
->reg_maps
->sampler_map
,
3792 ins
->src
[1].reg
.idx
[0].offset
, ins
->src
[2].reg
.idx
[0].offset
);
3793 shader_glsl_gen_sample_code(ins
, sampler_idx
, &sample_function
, WINED3DSP_NOSWIZZLE
,
3794 NULL
, NULL
, NULL
, "%s", coord_param
.param_str
);
3797 static void shader_glsl_texcoord(const struct wined3d_shader_instruction
*ins
)
3799 /* FIXME: Make this work for more than just 2D textures */
3800 struct wined3d_shader_buffer
*buffer
= ins
->ctx
->buffer
;
3801 DWORD write_mask
= shader_glsl_append_dst(ins
->ctx
->buffer
, ins
);
3803 if (!(ins
->ctx
->reg_maps
->shader_version
.major
== 1 && ins
->ctx
->reg_maps
->shader_version
.minor
== 4))
3807 shader_glsl_get_write_mask(&ins
->dst
[0], dst_mask
);
3808 shader_addline(buffer
, "clamp(gl_TexCoord[%u], 0.0, 1.0)%s);\n",
3809 ins
->dst
[0].reg
.idx
[0].offset
, dst_mask
);
3813 enum wined3d_shader_src_modifier src_mod
= ins
->src
[0].modifiers
;
3814 DWORD reg
= ins
->src
[0].reg
.idx
[0].offset
;
3815 char dst_swizzle
[6];
3817 shader_glsl_get_swizzle(&ins
->src
[0], FALSE
, write_mask
, dst_swizzle
);
3819 if (src_mod
== WINED3DSPSM_DZ
)
3821 unsigned int mask_size
= shader_glsl_get_write_mask_size(write_mask
);
3822 struct glsl_src_param div_param
;
3824 shader_glsl_add_src_param(ins
, &ins
->src
[0], WINED3DSP_WRITEMASK_2
, &div_param
);
3826 if (mask_size
> 1) {
3827 shader_addline(buffer
, "gl_TexCoord[%u]%s / vec%d(%s));\n", reg
, dst_swizzle
, mask_size
, div_param
.param_str
);
3829 shader_addline(buffer
, "gl_TexCoord[%u]%s / %s);\n", reg
, dst_swizzle
, div_param
.param_str
);
3832 else if (src_mod
== WINED3DSPSM_DW
)
3834 unsigned int mask_size
= shader_glsl_get_write_mask_size(write_mask
);
3835 struct glsl_src_param div_param
;
3837 shader_glsl_add_src_param(ins
, &ins
->src
[0], WINED3DSP_WRITEMASK_3
, &div_param
);
3839 if (mask_size
> 1) {
3840 shader_addline(buffer
, "gl_TexCoord[%u]%s / vec%d(%s));\n", reg
, dst_swizzle
, mask_size
, div_param
.param_str
);
3842 shader_addline(buffer
, "gl_TexCoord[%u]%s / %s);\n", reg
, dst_swizzle
, div_param
.param_str
);
3845 shader_addline(buffer
, "gl_TexCoord[%u]%s);\n", reg
, dst_swizzle
);
3850 /** Process the WINED3DSIO_TEXDP3TEX instruction in GLSL:
3851 * Take a 3-component dot product of the TexCoord[dstreg] and src,
3852 * then perform a 1D texture lookup from stage dstregnum, place into dst. */
3853 static void shader_glsl_texdp3tex(const struct wined3d_shader_instruction
*ins
)
3855 DWORD src_mask
= WINED3DSP_WRITEMASK_0
| WINED3DSP_WRITEMASK_1
| WINED3DSP_WRITEMASK_2
;
3856 DWORD sampler_idx
= ins
->dst
[0].reg
.idx
[0].offset
;
3857 struct glsl_sample_function sample_function
;
3858 struct glsl_src_param src0_param
;
3861 shader_glsl_add_src_param(ins
, &ins
->src
[0], src_mask
, &src0_param
);
3863 /* Do I have to take care about the projected bit? I don't think so, since the dp3 returns only one
3864 * scalar, and projected sampling would require 4.
3866 * It is a dependent read - not valid with conditional NP2 textures
3868 shader_glsl_get_sample_function(ins
->ctx
, sampler_idx
, 0, &sample_function
);
3869 mask_size
= shader_glsl_get_write_mask_size(sample_function
.coord_mask
);
3874 shader_glsl_gen_sample_code(ins
, sampler_idx
, &sample_function
, WINED3DSP_NOSWIZZLE
, NULL
, NULL
, NULL
,
3875 "dot(gl_TexCoord[%u].xyz, %s)", sampler_idx
, src0_param
.param_str
);
3879 shader_glsl_gen_sample_code(ins
, sampler_idx
, &sample_function
, WINED3DSP_NOSWIZZLE
, NULL
, NULL
, NULL
,
3880 "vec2(dot(gl_TexCoord[%u].xyz, %s), 0.0)", sampler_idx
, src0_param
.param_str
);
3884 shader_glsl_gen_sample_code(ins
, sampler_idx
, &sample_function
, WINED3DSP_NOSWIZZLE
, NULL
, NULL
, NULL
,
3885 "vec3(dot(gl_TexCoord[%u].xyz, %s), 0.0, 0.0)", sampler_idx
, src0_param
.param_str
);
3889 FIXME("Unexpected mask size %u\n", mask_size
);
3894 /** Process the WINED3DSIO_TEXDP3 instruction in GLSL:
3895 * Take a 3-component dot product of the TexCoord[dstreg] and src. */
3896 static void shader_glsl_texdp3(const struct wined3d_shader_instruction
*ins
)
3898 DWORD src_mask
= WINED3DSP_WRITEMASK_0
| WINED3DSP_WRITEMASK_1
| WINED3DSP_WRITEMASK_2
;
3899 DWORD dstreg
= ins
->dst
[0].reg
.idx
[0].offset
;
3900 struct glsl_src_param src0_param
;
3902 unsigned int mask_size
;
3904 dst_mask
= shader_glsl_append_dst(ins
->ctx
->buffer
, ins
);
3905 mask_size
= shader_glsl_get_write_mask_size(dst_mask
);
3906 shader_glsl_add_src_param(ins
, &ins
->src
[0], src_mask
, &src0_param
);
3908 if (mask_size
> 1) {
3909 shader_addline(ins
->ctx
->buffer
, "vec%d(dot(T%u.xyz, %s)));\n", mask_size
, dstreg
, src0_param
.param_str
);
3911 shader_addline(ins
->ctx
->buffer
, "dot(T%u.xyz, %s));\n", dstreg
, src0_param
.param_str
);
3915 /** Process the WINED3DSIO_TEXDEPTH instruction in GLSL:
3916 * Calculate the depth as dst.x / dst.y */
3917 static void shader_glsl_texdepth(const struct wined3d_shader_instruction
*ins
)
3919 struct glsl_dst_param dst_param
;
3921 shader_glsl_add_dst_param(ins
, &ins
->dst
[0], &dst_param
);
3923 /* Tests show that texdepth never returns anything below 0.0, and that r5.y is clamped to 1.0.
3924 * Negative input is accepted, -0.25 / -0.5 returns 0.5. GL should clamp gl_FragDepth to [0;1], but
3925 * this doesn't always work, so clamp the results manually. Whether or not the x value is clamped at 1
3926 * too is irrelevant, since if x = 0, any y value < 1.0 (and > 1.0 is not allowed) results in a result
3929 shader_addline(ins
->ctx
->buffer
, "gl_FragDepth = clamp((%s.x / min(%s.y, 1.0)), 0.0, 1.0);\n",
3930 dst_param
.reg_name
, dst_param
.reg_name
);
3933 /** Process the WINED3DSIO_TEXM3X2DEPTH instruction in GLSL:
3934 * Last row of a 3x2 matrix multiply, use the result to calculate the depth:
3935 * Calculate tmp0.y = TexCoord[dstreg] . src.xyz; (tmp0.x has already been calculated)
3936 * depth = (tmp0.y == 0.0) ? 1.0 : tmp0.x / tmp0.y
3938 static void shader_glsl_texm3x2depth(const struct wined3d_shader_instruction
*ins
)
3940 DWORD src_mask
= WINED3DSP_WRITEMASK_0
| WINED3DSP_WRITEMASK_1
| WINED3DSP_WRITEMASK_2
;
3941 DWORD dstreg
= ins
->dst
[0].reg
.idx
[0].offset
;
3942 struct glsl_src_param src0_param
;
3944 shader_glsl_add_src_param(ins
, &ins
->src
[0], src_mask
, &src0_param
);
3946 shader_addline(ins
->ctx
->buffer
, "tmp0.y = dot(T%u.xyz, %s);\n", dstreg
, src0_param
.param_str
);
3947 shader_addline(ins
->ctx
->buffer
, "gl_FragDepth = (tmp0.y == 0.0) ? 1.0 : clamp(tmp0.x / tmp0.y, 0.0, 1.0);\n");
3950 /** Process the WINED3DSIO_TEXM3X2PAD instruction in GLSL
3951 * Calculate the 1st of a 2-row matrix multiplication. */
3952 static void shader_glsl_texm3x2pad(const struct wined3d_shader_instruction
*ins
)
3954 DWORD src_mask
= WINED3DSP_WRITEMASK_0
| WINED3DSP_WRITEMASK_1
| WINED3DSP_WRITEMASK_2
;
3955 DWORD reg
= ins
->dst
[0].reg
.idx
[0].offset
;
3956 struct wined3d_shader_buffer
*buffer
= ins
->ctx
->buffer
;
3957 struct glsl_src_param src0_param
;
3959 shader_glsl_add_src_param(ins
, &ins
->src
[0], src_mask
, &src0_param
);
3960 shader_addline(buffer
, "tmp0.x = dot(T%u.xyz, %s);\n", reg
, src0_param
.param_str
);
3963 /** Process the WINED3DSIO_TEXM3X3PAD instruction in GLSL
3964 * Calculate the 1st or 2nd row of a 3-row matrix multiplication. */
3965 static void shader_glsl_texm3x3pad(const struct wined3d_shader_instruction
*ins
)
3967 DWORD src_mask
= WINED3DSP_WRITEMASK_0
| WINED3DSP_WRITEMASK_1
| WINED3DSP_WRITEMASK_2
;
3968 struct wined3d_shader_buffer
*buffer
= ins
->ctx
->buffer
;
3969 struct wined3d_shader_tex_mx
*tex_mx
= ins
->ctx
->tex_mx
;
3970 DWORD reg
= ins
->dst
[0].reg
.idx
[0].offset
;
3971 struct glsl_src_param src0_param
;
3973 shader_glsl_add_src_param(ins
, &ins
->src
[0], src_mask
, &src0_param
);
3974 shader_addline(buffer
, "tmp0.%c = dot(T%u.xyz, %s);\n", 'x' + tex_mx
->current_row
, reg
, src0_param
.param_str
);
3975 tex_mx
->texcoord_w
[tex_mx
->current_row
++] = reg
;
3978 static void shader_glsl_texm3x2tex(const struct wined3d_shader_instruction
*ins
)
3980 DWORD src_mask
= WINED3DSP_WRITEMASK_0
| WINED3DSP_WRITEMASK_1
| WINED3DSP_WRITEMASK_2
;
3981 struct wined3d_shader_buffer
*buffer
= ins
->ctx
->buffer
;
3982 struct glsl_sample_function sample_function
;
3983 DWORD reg
= ins
->dst
[0].reg
.idx
[0].offset
;
3984 struct glsl_src_param src0_param
;
3986 shader_glsl_add_src_param(ins
, &ins
->src
[0], src_mask
, &src0_param
);
3987 shader_addline(buffer
, "tmp0.y = dot(T%u.xyz, %s);\n", reg
, src0_param
.param_str
);
3989 shader_glsl_get_sample_function(ins
->ctx
, reg
, 0, &sample_function
);
3991 /* Sample the texture using the calculated coordinates */
3992 shader_glsl_gen_sample_code(ins
, reg
, &sample_function
, WINED3DSP_NOSWIZZLE
, NULL
, NULL
, NULL
, "tmp0.xy");
3995 /** Process the WINED3DSIO_TEXM3X3TEX instruction in GLSL
3996 * Perform the 3rd row of a 3x3 matrix multiply, then sample the texture using the calculated coordinates */
3997 static void shader_glsl_texm3x3tex(const struct wined3d_shader_instruction
*ins
)
3999 DWORD src_mask
= WINED3DSP_WRITEMASK_0
| WINED3DSP_WRITEMASK_1
| WINED3DSP_WRITEMASK_2
;
4000 struct wined3d_shader_tex_mx
*tex_mx
= ins
->ctx
->tex_mx
;
4001 struct glsl_sample_function sample_function
;
4002 DWORD reg
= ins
->dst
[0].reg
.idx
[0].offset
;
4003 struct glsl_src_param src0_param
;
4005 shader_glsl_add_src_param(ins
, &ins
->src
[0], src_mask
, &src0_param
);
4006 shader_addline(ins
->ctx
->buffer
, "tmp0.z = dot(T%u.xyz, %s);\n", reg
, src0_param
.param_str
);
4008 /* Dependent read, not valid with conditional NP2 */
4009 shader_glsl_get_sample_function(ins
->ctx
, reg
, 0, &sample_function
);
4011 /* Sample the texture using the calculated coordinates */
4012 shader_glsl_gen_sample_code(ins
, reg
, &sample_function
, WINED3DSP_NOSWIZZLE
, NULL
, NULL
, NULL
, "tmp0.xyz");
4014 tex_mx
->current_row
= 0;
4017 /** Process the WINED3DSIO_TEXM3X3 instruction in GLSL
4018 * Perform the 3rd row of a 3x3 matrix multiply */
4019 static void shader_glsl_texm3x3(const struct wined3d_shader_instruction
*ins
)
4021 DWORD src_mask
= WINED3DSP_WRITEMASK_0
| WINED3DSP_WRITEMASK_1
| WINED3DSP_WRITEMASK_2
;
4022 struct wined3d_shader_tex_mx
*tex_mx
= ins
->ctx
->tex_mx
;
4023 DWORD reg
= ins
->dst
[0].reg
.idx
[0].offset
;
4024 struct glsl_src_param src0_param
;
4027 shader_glsl_add_src_param(ins
, &ins
->src
[0], src_mask
, &src0_param
);
4029 shader_glsl_append_dst(ins
->ctx
->buffer
, ins
);
4030 shader_glsl_get_write_mask(&ins
->dst
[0], dst_mask
);
4031 shader_addline(ins
->ctx
->buffer
, "vec4(tmp0.xy, dot(T%u.xyz, %s), 1.0)%s);\n", reg
, src0_param
.param_str
, dst_mask
);
4033 tex_mx
->current_row
= 0;
4036 /* Process the WINED3DSIO_TEXM3X3SPEC instruction in GLSL
4037 * Perform the final texture lookup based on the previous 2 3x3 matrix multiplies */
4038 static void shader_glsl_texm3x3spec(const struct wined3d_shader_instruction
*ins
)
4040 struct glsl_src_param src0_param
;
4041 struct glsl_src_param src1_param
;
4042 struct wined3d_shader_buffer
*buffer
= ins
->ctx
->buffer
;
4043 struct wined3d_shader_tex_mx
*tex_mx
= ins
->ctx
->tex_mx
;
4044 DWORD src_mask
= WINED3DSP_WRITEMASK_0
| WINED3DSP_WRITEMASK_1
| WINED3DSP_WRITEMASK_2
;
4045 struct glsl_sample_function sample_function
;
4046 DWORD reg
= ins
->dst
[0].reg
.idx
[0].offset
;
4049 shader_glsl_add_src_param(ins
, &ins
->src
[0], src_mask
, &src0_param
);
4050 shader_glsl_add_src_param(ins
, &ins
->src
[1], src_mask
, &src1_param
);
4052 /* Perform the last matrix multiply operation */
4053 shader_addline(buffer
, "tmp0.z = dot(T%u.xyz, %s);\n", reg
, src0_param
.param_str
);
4054 /* Reflection calculation */
4055 shader_addline(buffer
, "tmp0.xyz = -reflect((%s), normalize(tmp0.xyz));\n", src1_param
.param_str
);
4057 /* Dependent read, not valid with conditional NP2 */
4058 shader_glsl_get_sample_function(ins
->ctx
, reg
, 0, &sample_function
);
4059 shader_glsl_write_mask_to_str(sample_function
.coord_mask
, coord_mask
);
4061 /* Sample the texture */
4062 shader_glsl_gen_sample_code(ins
, reg
, &sample_function
, WINED3DSP_NOSWIZZLE
,
4063 NULL
, NULL
, NULL
, "tmp0%s", coord_mask
);
4065 tex_mx
->current_row
= 0;
4068 /* Process the WINED3DSIO_TEXM3X3VSPEC instruction in GLSL
4069 * Perform the final texture lookup based on the previous 2 3x3 matrix multiplies */
4070 static void shader_glsl_texm3x3vspec(const struct wined3d_shader_instruction
*ins
)
4072 struct wined3d_shader_buffer
*buffer
= ins
->ctx
->buffer
;
4073 struct wined3d_shader_tex_mx
*tex_mx
= ins
->ctx
->tex_mx
;
4074 DWORD src_mask
= WINED3DSP_WRITEMASK_0
| WINED3DSP_WRITEMASK_1
| WINED3DSP_WRITEMASK_2
;
4075 struct glsl_sample_function sample_function
;
4076 DWORD reg
= ins
->dst
[0].reg
.idx
[0].offset
;
4077 struct glsl_src_param src0_param
;
4080 shader_glsl_add_src_param(ins
, &ins
->src
[0], src_mask
, &src0_param
);
4082 /* Perform the last matrix multiply operation */
4083 shader_addline(buffer
, "tmp0.z = dot(vec3(T%u), vec3(%s));\n", reg
, src0_param
.param_str
);
4085 /* Construct the eye-ray vector from w coordinates */
4086 shader_addline(buffer
, "tmp1.xyz = normalize(vec3(gl_TexCoord[%u].w, gl_TexCoord[%u].w, gl_TexCoord[%u].w));\n",
4087 tex_mx
->texcoord_w
[0], tex_mx
->texcoord_w
[1], reg
);
4088 shader_addline(buffer
, "tmp0.xyz = -reflect(tmp1.xyz, normalize(tmp0.xyz));\n");
4090 /* Dependent read, not valid with conditional NP2 */
4091 shader_glsl_get_sample_function(ins
->ctx
, reg
, 0, &sample_function
);
4092 shader_glsl_write_mask_to_str(sample_function
.coord_mask
, coord_mask
);
4094 /* Sample the texture using the calculated coordinates */
4095 shader_glsl_gen_sample_code(ins
, reg
, &sample_function
, WINED3DSP_NOSWIZZLE
,
4096 NULL
, NULL
, NULL
, "tmp0%s", coord_mask
);
4098 tex_mx
->current_row
= 0;
4101 /** Process the WINED3DSIO_TEXBEM instruction in GLSL.
4102 * Apply a fake bump map transform.
4103 * texbem is pshader <= 1.3 only, this saves a few version checks
4105 static void shader_glsl_texbem(const struct wined3d_shader_instruction
*ins
)
4107 const struct shader_glsl_ctx_priv
*priv
= ins
->ctx
->backend_data
;
4108 struct glsl_sample_function sample_function
;
4109 struct glsl_src_param coord_param
;
4115 sampler_idx
= ins
->dst
[0].reg
.idx
[0].offset
;
4116 flags
= (priv
->cur_ps_args
->tex_transform
>> sampler_idx
* WINED3D_PSARGS_TEXTRANSFORM_SHIFT
)
4117 & WINED3D_PSARGS_TEXTRANSFORM_MASK
;
4119 /* Dependent read, not valid with conditional NP2 */
4120 shader_glsl_get_sample_function(ins
->ctx
, sampler_idx
, 0, &sample_function
);
4121 mask
= sample_function
.coord_mask
;
4123 shader_glsl_write_mask_to_str(mask
, coord_mask
);
4125 /* With projected textures, texbem only divides the static texture coord,
4126 * not the displacement, so we can't let GL handle this. */
4127 if (flags
& WINED3D_PSARGS_PROJECTED
)
4130 char coord_div_mask
[3];
4131 switch (flags
& ~WINED3D_PSARGS_PROJECTED
)
4133 case WINED3D_TTFF_COUNT1
:
4134 FIXME("WINED3D_TTFF_PROJECTED with WINED3D_TTFF_COUNT1?\n");
4136 case WINED3D_TTFF_COUNT2
:
4137 div_mask
= WINED3DSP_WRITEMASK_1
;
4139 case WINED3D_TTFF_COUNT3
:
4140 div_mask
= WINED3DSP_WRITEMASK_2
;
4142 case WINED3D_TTFF_COUNT4
:
4143 case WINED3D_TTFF_DISABLE
:
4144 div_mask
= WINED3DSP_WRITEMASK_3
;
4147 shader_glsl_write_mask_to_str(div_mask
, coord_div_mask
);
4148 shader_addline(ins
->ctx
->buffer
, "T%u%s /= T%u%s;\n", sampler_idx
, coord_mask
, sampler_idx
, coord_div_mask
);
4151 shader_glsl_add_src_param(ins
, &ins
->src
[0], WINED3DSP_WRITEMASK_0
| WINED3DSP_WRITEMASK_1
, &coord_param
);
4153 shader_glsl_gen_sample_code(ins
, sampler_idx
, &sample_function
, WINED3DSP_NOSWIZZLE
, NULL
, NULL
, NULL
,
4154 "T%u%s + vec4(bumpenv_mat%u * %s, 0.0, 0.0)%s", sampler_idx
, coord_mask
, sampler_idx
,
4155 coord_param
.param_str
, coord_mask
);
4157 if (ins
->handler_idx
== WINED3DSIH_TEXBEML
)
4159 struct glsl_src_param luminance_param
;
4160 struct glsl_dst_param dst_param
;
4162 shader_glsl_add_src_param(ins
, &ins
->src
[0], WINED3DSP_WRITEMASK_2
, &luminance_param
);
4163 shader_glsl_add_dst_param(ins
, &ins
->dst
[0], &dst_param
);
4165 shader_addline(ins
->ctx
->buffer
, "%s%s *= (%s * bumpenv_lum_scale%u + bumpenv_lum_offset%u);\n",
4166 dst_param
.reg_name
, dst_param
.mask_str
,
4167 luminance_param
.param_str
, sampler_idx
, sampler_idx
);
4171 static void shader_glsl_bem(const struct wined3d_shader_instruction
*ins
)
4173 DWORD sampler_idx
= ins
->dst
[0].reg
.idx
[0].offset
;
4174 struct glsl_src_param src0_param
, src1_param
;
4176 shader_glsl_add_src_param(ins
, &ins
->src
[0], WINED3DSP_WRITEMASK_0
| WINED3DSP_WRITEMASK_1
, &src0_param
);
4177 shader_glsl_add_src_param(ins
, &ins
->src
[1], WINED3DSP_WRITEMASK_0
| WINED3DSP_WRITEMASK_1
, &src1_param
);
4179 shader_glsl_append_dst(ins
->ctx
->buffer
, ins
);
4180 shader_addline(ins
->ctx
->buffer
, "%s + bumpenv_mat%u * %s);\n",
4181 src0_param
.param_str
, sampler_idx
, src1_param
.param_str
);
4184 /** Process the WINED3DSIO_TEXREG2AR instruction in GLSL
4185 * Sample 2D texture at dst using the alpha & red (wx) components of src as texture coordinates */
4186 static void shader_glsl_texreg2ar(const struct wined3d_shader_instruction
*ins
)
4188 DWORD sampler_idx
= ins
->dst
[0].reg
.idx
[0].offset
;
4189 struct glsl_sample_function sample_function
;
4190 struct glsl_src_param src0_param
;
4192 shader_glsl_add_src_param(ins
, &ins
->src
[0], WINED3DSP_WRITEMASK_ALL
, &src0_param
);
4194 shader_glsl_get_sample_function(ins
->ctx
, sampler_idx
, 0, &sample_function
);
4195 shader_glsl_gen_sample_code(ins
, sampler_idx
, &sample_function
, WINED3DSP_NOSWIZZLE
, NULL
, NULL
, NULL
,
4196 "%s.wx", src0_param
.reg_name
);
4199 /** Process the WINED3DSIO_TEXREG2GB instruction in GLSL
4200 * Sample 2D texture at dst using the green & blue (yz) components of src as texture coordinates */
4201 static void shader_glsl_texreg2gb(const struct wined3d_shader_instruction
*ins
)
4203 DWORD sampler_idx
= ins
->dst
[0].reg
.idx
[0].offset
;
4204 struct glsl_sample_function sample_function
;
4205 struct glsl_src_param src0_param
;
4207 shader_glsl_add_src_param(ins
, &ins
->src
[0], WINED3DSP_WRITEMASK_ALL
, &src0_param
);
4209 shader_glsl_get_sample_function(ins
->ctx
, sampler_idx
, 0, &sample_function
);
4210 shader_glsl_gen_sample_code(ins
, sampler_idx
, &sample_function
, WINED3DSP_NOSWIZZLE
, NULL
, NULL
, NULL
,
4211 "%s.yz", src0_param
.reg_name
);
4214 /** Process the WINED3DSIO_TEXREG2RGB instruction in GLSL
4215 * Sample texture at dst using the rgb (xyz) components of src as texture coordinates */
4216 static void shader_glsl_texreg2rgb(const struct wined3d_shader_instruction
*ins
)
4218 DWORD sampler_idx
= ins
->dst
[0].reg
.idx
[0].offset
;
4219 struct glsl_sample_function sample_function
;
4220 struct glsl_src_param src0_param
;
4222 /* Dependent read, not valid with conditional NP2 */
4223 shader_glsl_get_sample_function(ins
->ctx
, sampler_idx
, 0, &sample_function
);
4224 shader_glsl_add_src_param(ins
, &ins
->src
[0], sample_function
.coord_mask
, &src0_param
);
4226 shader_glsl_gen_sample_code(ins
, sampler_idx
, &sample_function
, WINED3DSP_NOSWIZZLE
, NULL
, NULL
, NULL
,
4227 "%s", src0_param
.param_str
);
4230 /** Process the WINED3DSIO_TEXKILL instruction in GLSL.
4231 * If any of the first 3 components are < 0, discard this pixel */
4232 static void shader_glsl_texkill(const struct wined3d_shader_instruction
*ins
)
4234 struct glsl_dst_param dst_param
;
4236 /* The argument is a destination parameter, and no writemasks are allowed */
4237 shader_glsl_add_dst_param(ins
, &ins
->dst
[0], &dst_param
);
4238 if (ins
->ctx
->reg_maps
->shader_version
.major
>= 2)
4240 if (ins
->ctx
->reg_maps
->shader_version
.major
>= 4)
4241 FIXME("SM4 discard not implemented.\n");
4242 /* 2.0 shaders compare all 4 components in texkill */
4243 shader_addline(ins
->ctx
->buffer
, "if (any(lessThan(%s.xyzw, vec4(0.0)))) discard;\n", dst_param
.reg_name
);
4245 /* 1.X shaders only compare the first 3 components, probably due to the nature of the texkill
4246 * instruction as a tex* instruction, and phase, which kills all a / w components. Even if all
4247 * 4 components are defined, only the first 3 are used
4249 shader_addline(ins
->ctx
->buffer
, "if (any(lessThan(%s.xyz, vec3(0.0)))) discard;\n", dst_param
.reg_name
);
4253 /** Process the WINED3DSIO_DP2ADD instruction in GLSL.
4254 * dst = dot2(src0, src1) + src2 */
4255 static void shader_glsl_dp2add(const struct wined3d_shader_instruction
*ins
)
4257 struct glsl_src_param src0_param
;
4258 struct glsl_src_param src1_param
;
4259 struct glsl_src_param src2_param
;
4261 unsigned int mask_size
;
4263 write_mask
= shader_glsl_append_dst(ins
->ctx
->buffer
, ins
);
4264 mask_size
= shader_glsl_get_write_mask_size(write_mask
);
4266 shader_glsl_add_src_param(ins
, &ins
->src
[0], WINED3DSP_WRITEMASK_0
| WINED3DSP_WRITEMASK_1
, &src0_param
);
4267 shader_glsl_add_src_param(ins
, &ins
->src
[1], WINED3DSP_WRITEMASK_0
| WINED3DSP_WRITEMASK_1
, &src1_param
);
4268 shader_glsl_add_src_param(ins
, &ins
->src
[2], WINED3DSP_WRITEMASK_0
, &src2_param
);
4270 if (mask_size
> 1) {
4271 shader_addline(ins
->ctx
->buffer
, "vec%d(dot(%s, %s) + %s));\n",
4272 mask_size
, src0_param
.param_str
, src1_param
.param_str
, src2_param
.param_str
);
4274 shader_addline(ins
->ctx
->buffer
, "dot(%s, %s) + %s);\n",
4275 src0_param
.param_str
, src1_param
.param_str
, src2_param
.param_str
);
4279 static void shader_glsl_input_pack(const struct wined3d_shader
*shader
, struct wined3d_shader_buffer
*buffer
,
4280 const struct wined3d_shader_signature
*input_signature
,
4281 const struct wined3d_shader_reg_maps
*reg_maps
,
4282 enum vertexprocessing_mode vertexprocessing
)
4286 for (i
= 0; i
< input_signature
->element_count
; ++i
)
4288 const struct wined3d_shader_signature_element
*input
= &input_signature
->elements
[i
];
4289 const char *semantic_name
;
4294 if (!(reg_maps
->input_registers
& (1 << input
->register_idx
)))
4297 semantic_name
= input
->semantic_name
;
4298 semantic_idx
= input
->semantic_idx
;
4299 shader_glsl_write_mask_to_str(input
->mask
, reg_mask
);
4301 if (vertexprocessing
== vertexshader
)
4303 if (input
->sysval_semantic
== WINED3D_SV_POSITION
)
4304 shader_addline(buffer
, "ps_in[%u]%s = vpos%s;\n",
4305 shader
->u
.ps
.input_reg_map
[input
->register_idx
], reg_mask
, reg_mask
);
4307 shader_addline(buffer
, "ps_in[%u]%s = ps_link[%u]%s;\n",
4308 shader
->u
.ps
.input_reg_map
[input
->register_idx
], reg_mask
,
4309 shader
->u
.ps
.input_reg_map
[input
->register_idx
], reg_mask
);
4311 else if (shader_match_semantic(semantic_name
, WINED3D_DECL_USAGE_TEXCOORD
))
4313 if (semantic_idx
< 8 && vertexprocessing
== pretransformed
)
4314 shader_addline(buffer
, "ps_in[%u]%s = gl_TexCoord[%u]%s;\n",
4315 shader
->u
.ps
.input_reg_map
[input
->register_idx
], reg_mask
, semantic_idx
, reg_mask
);
4317 shader_addline(buffer
, "ps_in[%u]%s = vec4(0.0, 0.0, 0.0, 0.0)%s;\n",
4318 shader
->u
.ps
.input_reg_map
[input
->register_idx
], reg_mask
, reg_mask
);
4320 else if (shader_match_semantic(semantic_name
, WINED3D_DECL_USAGE_COLOR
))
4323 shader_addline(buffer
, "ps_in[%u]%s = vec4(gl_Color)%s;\n",
4324 shader
->u
.ps
.input_reg_map
[input
->register_idx
], reg_mask
, reg_mask
);
4325 else if (semantic_idx
== 1)
4326 shader_addline(buffer
, "ps_in[%u]%s = vec4(gl_SecondaryColor)%s;\n",
4327 shader
->u
.ps
.input_reg_map
[input
->register_idx
], reg_mask
, reg_mask
);
4329 shader_addline(buffer
, "ps_in[%u]%s = vec4(0.0, 0.0, 0.0, 0.0)%s;\n",
4330 shader
->u
.ps
.input_reg_map
[input
->register_idx
], reg_mask
, reg_mask
);
4334 shader_addline(buffer
, "ps_in[%u]%s = vec4(0.0, 0.0, 0.0, 0.0)%s;\n",
4335 shader
->u
.ps
.input_reg_map
[input
->register_idx
], reg_mask
, reg_mask
);
4340 /*********************************************
4341 * Vertex Shader Specific Code begins here
4342 ********************************************/
4344 static void add_glsl_program_entry(struct shader_glsl_priv
*priv
, struct glsl_shader_prog_link
*entry
)
4346 struct glsl_program_key key
;
4348 key
.vs_id
= entry
->vs
.id
;
4349 key
.gs_id
= entry
->gs
.id
;
4350 key
.ps_id
= entry
->ps
.id
;
4352 if (wine_rb_put(&priv
->program_lookup
, &key
, &entry
->program_lookup_entry
) == -1)
4354 ERR("Failed to insert program entry.\n");
4358 static struct glsl_shader_prog_link
*get_glsl_program_entry(const struct shader_glsl_priv
*priv
,
4359 GLuint vs_id
, GLuint gs_id
, GLuint ps_id
)
4361 struct wine_rb_entry
*entry
;
4362 struct glsl_program_key key
;
4368 entry
= wine_rb_get(&priv
->program_lookup
, &key
);
4369 return entry
? WINE_RB_ENTRY_VALUE(entry
, struct glsl_shader_prog_link
, program_lookup_entry
) : NULL
;
4372 /* Context activation is done by the caller. */
4373 static void delete_glsl_program_entry(struct shader_glsl_priv
*priv
, const struct wined3d_gl_info
*gl_info
,
4374 struct glsl_shader_prog_link
*entry
)
4376 struct glsl_program_key key
;
4378 key
.vs_id
= entry
->vs
.id
;
4379 key
.gs_id
= entry
->gs
.id
;
4380 key
.ps_id
= entry
->ps
.id
;
4381 wine_rb_remove(&priv
->program_lookup
, &key
);
4383 GL_EXTCALL(glDeleteProgram(entry
->id
));
4385 list_remove(&entry
->vs
.shader_entry
);
4387 list_remove(&entry
->gs
.shader_entry
);
4389 list_remove(&entry
->ps
.shader_entry
);
4390 HeapFree(GetProcessHeap(), 0, entry
->vs
.uniform_f_locations
);
4391 HeapFree(GetProcessHeap(), 0, entry
->ps
.uniform_f_locations
);
4392 HeapFree(GetProcessHeap(), 0, entry
);
4395 static void handle_ps3_input(struct wined3d_shader_buffer
*buffer
,
4396 const struct wined3d_gl_info
*gl_info
, const DWORD
*map
,
4397 const struct wined3d_shader_signature
*input_signature
,
4398 const struct wined3d_shader_reg_maps
*reg_maps_in
,
4399 const struct wined3d_shader_signature
*output_signature
,
4400 const struct wined3d_shader_reg_maps
*reg_maps_out
)
4405 unsigned int in_count
= vec4_varyings(3, gl_info
);
4407 char destination
[50];
4409 set
= HeapAlloc(GetProcessHeap(), HEAP_ZERO_MEMORY
, sizeof(*set
) * (in_count
+ 2));
4411 for (i
= 0; i
< input_signature
->element_count
; ++i
)
4413 const struct wined3d_shader_signature_element
*input
= &input_signature
->elements
[i
];
4415 if (!(reg_maps_in
->input_registers
& (1 << input
->register_idx
)))
4418 in_idx
= map
[input
->register_idx
];
4419 /* Declared, but not read register */
4422 if (in_idx
>= (in_count
+ 2))
4424 FIXME("More input varyings declared than supported, expect issues.\n");
4428 if (in_idx
== in_count
)
4429 sprintf(destination
, "gl_FrontColor");
4430 else if (in_idx
== in_count
+ 1)
4431 sprintf(destination
, "gl_FrontSecondaryColor");
4433 sprintf(destination
, "ps_link[%u]", in_idx
);
4438 for (j
= 0; j
< output_signature
->element_count
; ++j
)
4440 const struct wined3d_shader_signature_element
*output
= &output_signature
->elements
[j
];
4443 if (!(reg_maps_out
->output_registers
& (1 << output
->register_idx
))
4444 || input
->semantic_idx
!= output
->semantic_idx
4445 || strcmp(input
->semantic_name
, output
->semantic_name
)
4446 || !(mask
= input
->mask
& output
->mask
))
4449 if (set
[in_idx
] == ~0u)
4452 set
[in_idx
] |= mask
;
4453 shader_glsl_write_mask_to_str(mask
, reg_mask
);
4455 shader_addline(buffer
, "%s%s = vs_out[%u]%s;\n",
4456 destination
, reg_mask
, output
->register_idx
, reg_mask
);
4460 for (i
= 0; i
< in_count
+ 2; ++i
)
4464 if (!set
[i
] || set
[i
] == WINED3DSP_WRITEMASK_ALL
)
4467 if (set
[i
] == ~0U) set
[i
] = 0;
4470 if (!(set
[i
] & WINED3DSP_WRITEMASK_0
)) reg_mask
[size
++] = 'x';
4471 if (!(set
[i
] & WINED3DSP_WRITEMASK_1
)) reg_mask
[size
++] = 'y';
4472 if (!(set
[i
] & WINED3DSP_WRITEMASK_2
)) reg_mask
[size
++] = 'z';
4473 if (!(set
[i
] & WINED3DSP_WRITEMASK_3
)) reg_mask
[size
++] = 'w';
4474 reg_mask
[size
] = '\0';
4477 sprintf(destination
, "gl_FrontColor");
4478 else if (i
== in_count
+ 1)
4479 sprintf(destination
, "gl_FrontSecondaryColor");
4481 sprintf(destination
, "ps_link[%u]", i
);
4483 if (size
== 1) shader_addline(buffer
, "%s.%s = 0.0;\n", destination
, reg_mask
);
4484 else shader_addline(buffer
, "%s.%s = vec%u(0.0);\n", destination
, reg_mask
, size
);
4487 HeapFree(GetProcessHeap(), 0, set
);
4490 /* Context activation is done by the caller. */
4491 static GLuint
generate_param_reorder_function(struct wined3d_shader_buffer
*buffer
,
4492 const struct wined3d_shader
*vs
, const struct wined3d_shader
*ps
,
4493 const struct wined3d_gl_info
*gl_info
)
4496 DWORD ps_major
= ps
? ps
->reg_maps
.shader_version
.major
: 0;
4498 const char *semantic_name
;
4502 shader_buffer_clear(buffer
);
4504 shader_addline(buffer
, "%s\n", shader_glsl_get_version(gl_info
, &vs
->reg_maps
.shader_version
));
4508 shader_addline(buffer
, "void order_ps_input(in vec4 vs_out[%u])\n{\n", vs
->limits
->packed_output
);
4510 for (i
= 0; i
< vs
->output_signature
.element_count
; ++i
)
4512 const struct wined3d_shader_signature_element
*output
= &vs
->output_signature
.elements
[i
];
4515 if (!(vs
->reg_maps
.output_registers
& (1 << output
->register_idx
)))
4518 semantic_name
= output
->semantic_name
;
4519 semantic_idx
= output
->semantic_idx
;
4520 write_mask
= output
->mask
;
4521 shader_glsl_write_mask_to_str(write_mask
, reg_mask
);
4523 if (shader_match_semantic(semantic_name
, WINED3D_DECL_USAGE_COLOR
))
4526 shader_addline(buffer
, "gl_FrontColor%s = vs_out[%u]%s;\n",
4527 reg_mask
, output
->register_idx
, reg_mask
);
4528 else if (semantic_idx
== 1)
4529 shader_addline(buffer
, "gl_FrontSecondaryColor%s = vs_out[%u]%s;\n",
4530 reg_mask
, output
->register_idx
, reg_mask
);
4532 else if (shader_match_semantic(semantic_name
, WINED3D_DECL_USAGE_POSITION
) && !semantic_idx
)
4534 shader_addline(buffer
, "gl_Position%s = vs_out[%u]%s;\n",
4535 reg_mask
, output
->register_idx
, reg_mask
);
4537 else if (shader_match_semantic(semantic_name
, WINED3D_DECL_USAGE_TEXCOORD
))
4539 if (semantic_idx
< 8)
4541 if (!(gl_info
->quirks
& WINED3D_QUIRK_SET_TEXCOORD_W
) || ps_major
> 0)
4542 write_mask
|= WINED3DSP_WRITEMASK_3
;
4544 shader_addline(buffer
, "gl_TexCoord[%u]%s = vs_out[%u]%s;\n",
4545 semantic_idx
, reg_mask
, output
->register_idx
, reg_mask
);
4546 if (!(write_mask
& WINED3DSP_WRITEMASK_3
))
4547 shader_addline(buffer
, "gl_TexCoord[%u].w = 1.0;\n", semantic_idx
);
4550 else if (shader_match_semantic(semantic_name
, WINED3D_DECL_USAGE_PSIZE
))
4552 shader_addline(buffer
, "gl_PointSize = vs_out[%u].%c;\n", output
->register_idx
, reg_mask
[1]);
4554 else if (shader_match_semantic(semantic_name
, WINED3D_DECL_USAGE_FOG
))
4556 shader_addline(buffer
, "gl_FogFragCoord = clamp(vs_out[%u].%c, 0.0, 1.0);\n",
4557 output
->register_idx
, reg_mask
[1]);
4560 shader_addline(buffer
, "}\n");
4564 UINT in_count
= min(vec4_varyings(ps_major
, gl_info
), ps
->limits
->packed_input
);
4565 /* This one is tricky: a 3.0 pixel shader reads from a 3.0 vertex shader */
4566 shader_addline(buffer
, "varying vec4 ps_link[%u];\n", in_count
);
4567 shader_addline(buffer
, "void order_ps_input(in vec4 vs_out[%u])\n{\n", vs
->limits
->packed_output
);
4569 /* First, sort out position and point size. Those are not passed to the pixel shader */
4570 for (i
= 0; i
< vs
->output_signature
.element_count
; ++i
)
4572 const struct wined3d_shader_signature_element
*output
= &vs
->output_signature
.elements
[i
];
4574 if (!(vs
->reg_maps
.output_registers
& (1 << output
->register_idx
)))
4577 semantic_name
= output
->semantic_name
;
4578 semantic_idx
= output
->semantic_idx
;
4579 shader_glsl_write_mask_to_str(output
->mask
, reg_mask
);
4581 if (shader_match_semantic(semantic_name
, WINED3D_DECL_USAGE_POSITION
) && !semantic_idx
)
4583 shader_addline(buffer
, "gl_Position%s = vs_out[%u]%s;\n",
4584 reg_mask
, output
->register_idx
, reg_mask
);
4586 else if (shader_match_semantic(semantic_name
, WINED3D_DECL_USAGE_PSIZE
))
4588 shader_addline(buffer
, "gl_PointSize = vs_out[%u].%c;\n", output
->register_idx
, reg_mask
[1]);
4592 /* Then, fix the pixel shader input */
4593 handle_ps3_input(buffer
, gl_info
, ps
->u
.ps
.input_reg_map
, &ps
->input_signature
,
4594 &ps
->reg_maps
, &vs
->output_signature
, &vs
->reg_maps
);
4596 shader_addline(buffer
, "}\n");
4599 ret
= GL_EXTCALL(glCreateShader(GL_VERTEX_SHADER
));
4600 checkGLcall("glCreateShader(GL_VERTEX_SHADER)");
4601 shader_glsl_compile(gl_info
, ret
, buffer
->buffer
);
4606 static void shader_glsl_generate_srgb_write_correction(struct wined3d_shader_buffer
*buffer
)
4608 shader_addline(buffer
, "tmp0.xyz = pow(gl_FragData[0].xyz, vec3(srgb_const0.x));\n");
4609 shader_addline(buffer
, "tmp0.xyz = tmp0.xyz * vec3(srgb_const0.y) - vec3(srgb_const0.z);\n");
4610 shader_addline(buffer
, "tmp1.xyz = gl_FragData[0].xyz * vec3(srgb_const0.w);\n");
4611 shader_addline(buffer
, "bvec3 srgb_compare = lessThan(gl_FragData[0].xyz, vec3(srgb_const1.x));\n");
4612 shader_addline(buffer
, "gl_FragData[0].xyz = mix(tmp0.xyz, tmp1.xyz, vec3(srgb_compare));\n");
4613 shader_addline(buffer
, "gl_FragData[0] = clamp(gl_FragData[0], 0.0, 1.0);\n");
4616 static void shader_glsl_generate_fog_code(struct wined3d_shader_buffer
*buffer
, enum wined3d_ffp_ps_fog_mode mode
)
4620 case WINED3D_FFP_PS_FOG_OFF
:
4623 case WINED3D_FFP_PS_FOG_LINEAR
:
4624 shader_addline(buffer
, "float Fog = (gl_Fog.end - gl_FogFragCoord) * gl_Fog.scale;\n");
4627 case WINED3D_FFP_PS_FOG_EXP
:
4628 /* Fog = e^-(gl_Fog.density * gl_FogFragCoord) */
4629 shader_addline(buffer
, "float Fog = exp(-gl_Fog.density * gl_FogFragCoord);\n");
4632 case WINED3D_FFP_PS_FOG_EXP2
:
4633 /* Fog = e^-((gl_Fog.density * gl_FogFragCoord)^2) */
4634 shader_addline(buffer
, "float Fog = exp(-gl_Fog.density * gl_Fog.density * gl_FogFragCoord * gl_FogFragCoord);\n");
4638 ERR("Invalid fog mode %#x.\n", mode
);
4642 shader_addline(buffer
, "gl_FragData[0].xyz = mix(gl_Fog.color.xyz, gl_FragData[0].xyz, clamp(Fog, 0.0, 1.0));\n");
4645 /* Context activation is done by the caller. */
4646 static GLuint
shader_glsl_generate_pshader(const struct wined3d_context
*context
,
4647 struct wined3d_shader_buffer
*buffer
, const struct wined3d_shader
*shader
,
4648 const struct ps_compile_args
*args
, struct ps_np2fixup_info
*np2fixup_info
)
4650 const struct wined3d_shader_reg_maps
*reg_maps
= &shader
->reg_maps
;
4651 const struct wined3d_gl_info
*gl_info
= context
->gl_info
;
4652 const DWORD
*function
= shader
->function
;
4653 struct shader_glsl_ctx_priv priv_ctx
;
4655 /* Create the hw GLSL shader object and assign it as the shader->prgId */
4656 GLuint shader_id
= GL_EXTCALL(glCreateShader(GL_FRAGMENT_SHADER
));
4658 memset(&priv_ctx
, 0, sizeof(priv_ctx
));
4659 priv_ctx
.cur_ps_args
= args
;
4660 priv_ctx
.cur_np2fixup_info
= np2fixup_info
;
4662 shader_addline(buffer
, "%s\n", shader_glsl_get_version(gl_info
, ®_maps
->shader_version
));
4664 if (gl_info
->supported
[ARB_SHADER_BIT_ENCODING
])
4665 shader_addline(buffer
, "#extension GL_ARB_shader_bit_encoding : enable\n");
4666 if (gl_info
->supported
[ARB_SHADER_TEXTURE_LOD
])
4667 shader_addline(buffer
, "#extension GL_ARB_shader_texture_lod : enable\n");
4668 /* The spec says that it doesn't have to be explicitly enabled, but the
4669 * nvidia drivers write a warning if we don't do so. */
4670 if (gl_info
->supported
[ARB_TEXTURE_RECTANGLE
])
4671 shader_addline(buffer
, "#extension GL_ARB_texture_rectangle : enable\n");
4672 if (gl_info
->supported
[ARB_UNIFORM_BUFFER_OBJECT
])
4673 shader_addline(buffer
, "#extension GL_ARB_uniform_buffer_object : enable\n");
4674 if (gl_info
->supported
[EXT_GPU_SHADER4
])
4675 shader_addline(buffer
, "#extension GL_EXT_gpu_shader4 : enable\n");
4677 /* Base Declarations */
4678 shader_generate_glsl_declarations(context
, buffer
, shader
, reg_maps
, &priv_ctx
);
4680 /* Pack 3.0 inputs */
4681 if (reg_maps
->shader_version
.major
>= 3)
4682 shader_glsl_input_pack(shader
, buffer
, &shader
->input_signature
, reg_maps
, args
->vp_mode
);
4684 /* Base Shader Body */
4685 shader_generate_main(shader
, buffer
, reg_maps
, function
, &priv_ctx
);
4687 /* Pixel shaders < 2.0 place the resulting color in R0 implicitly */
4688 if (reg_maps
->shader_version
.major
< 2)
4690 /* Some older cards like GeforceFX ones don't support multiple buffers, so also not gl_FragData */
4691 shader_addline(buffer
, "gl_FragData[0] = R0;\n");
4694 if (args
->srgb_correction
)
4695 shader_glsl_generate_srgb_write_correction(buffer
);
4697 /* SM < 3 does not replace the fog stage. */
4698 if (reg_maps
->shader_version
.major
< 3)
4699 shader_glsl_generate_fog_code(buffer
, args
->fog
);
4701 shader_addline(buffer
, "}\n");
4703 TRACE("Compiling shader object %u.\n", shader_id
);
4704 shader_glsl_compile(gl_info
, shader_id
, buffer
->buffer
);
4709 /* Context activation is done by the caller. */
4710 static GLuint
shader_glsl_generate_vshader(const struct wined3d_context
*context
,
4711 struct wined3d_shader_buffer
*buffer
, const struct wined3d_shader
*shader
,
4712 const struct vs_compile_args
*args
)
4714 const struct wined3d_shader_reg_maps
*reg_maps
= &shader
->reg_maps
;
4715 const struct wined3d_gl_info
*gl_info
= context
->gl_info
;
4716 const DWORD
*function
= shader
->function
;
4717 struct shader_glsl_ctx_priv priv_ctx
;
4719 /* Create the hw GLSL shader program and assign it as the shader->prgId */
4720 GLuint shader_id
= GL_EXTCALL(glCreateShader(GL_VERTEX_SHADER
));
4722 shader_addline(buffer
, "%s\n", shader_glsl_get_version(gl_info
, ®_maps
->shader_version
));
4724 if (gl_info
->supported
[ARB_DRAW_INSTANCED
])
4725 shader_addline(buffer
, "#extension GL_ARB_draw_instanced : enable\n");
4726 if (gl_info
->supported
[ARB_SHADER_BIT_ENCODING
])
4727 shader_addline(buffer
, "#extension GL_ARB_shader_bit_encoding : enable\n");
4728 if (gl_info
->supported
[ARB_UNIFORM_BUFFER_OBJECT
])
4729 shader_addline(buffer
, "#extension GL_ARB_uniform_buffer_object : enable\n");
4730 if (gl_info
->supported
[EXT_GPU_SHADER4
])
4731 shader_addline(buffer
, "#extension GL_EXT_gpu_shader4 : enable\n");
4733 memset(&priv_ctx
, 0, sizeof(priv_ctx
));
4734 priv_ctx
.cur_vs_args
= args
;
4736 /* Base Declarations */
4737 shader_generate_glsl_declarations(context
, buffer
, shader
, reg_maps
, &priv_ctx
);
4739 /* Base Shader Body */
4740 shader_generate_main(shader
, buffer
, reg_maps
, function
, &priv_ctx
);
4742 /* Unpack outputs */
4743 shader_addline(buffer
, "order_ps_input(vs_out);\n");
4745 /* The D3DRS_FOGTABLEMODE render state defines if the shader-generated fog coord is used
4746 * or if the fragment depth is used. If the fragment depth is used(FOGTABLEMODE != NONE),
4747 * the fog frag coord is thrown away. If the fog frag coord is used, but not written by
4748 * the shader, it is set to 0.0(fully fogged, since start = 1.0, end = 0.0)
4750 if (args
->fog_src
== VS_FOG_Z
)
4751 shader_addline(buffer
, "gl_FogFragCoord = gl_Position.z;\n");
4752 else if (!reg_maps
->fog
)
4753 shader_addline(buffer
, "gl_FogFragCoord = 0.0;\n");
4755 /* We always store the clipplanes without y inversion */
4756 if (args
->clip_enabled
)
4757 shader_addline(buffer
, "gl_ClipVertex = gl_Position;\n");
4759 /* Write the final position.
4761 * OpenGL coordinates specify the center of the pixel while d3d coords specify
4762 * the corner. The offsets are stored in z and w in posFixup. posFixup.y contains
4763 * 1.0 or -1.0 to turn the rendering upside down for offscreen rendering. PosFixup.x
4764 * contains 1.0 to allow a mad.
4766 shader_addline(buffer
, "gl_Position.y = gl_Position.y * posFixup.y;\n");
4767 shader_addline(buffer
, "gl_Position.xy += posFixup.zw * gl_Position.ww;\n");
4769 /* Z coord [0;1]->[-1;1] mapping, see comment in transform_projection in state.c
4771 * Basically we want (in homogeneous coordinates) z = z * 2 - 1. However, shaders are run
4772 * before the homogeneous divide, so we have to take the w into account: z = ((z / w) * 2 - 1) * w,
4773 * which is the same as z = z * 2 - w.
4775 shader_addline(buffer
, "gl_Position.z = gl_Position.z * 2.0 - gl_Position.w;\n");
4777 shader_addline(buffer
, "}\n");
4779 TRACE("Compiling shader object %u.\n", shader_id
);
4780 shader_glsl_compile(gl_info
, shader_id
, buffer
->buffer
);
4785 /* Context activation is done by the caller. */
4786 static GLuint
shader_glsl_generate_geometry_shader(const struct wined3d_context
*context
,
4787 struct wined3d_shader_buffer
*buffer
, const struct wined3d_shader
*shader
)
4789 const struct wined3d_shader_reg_maps
*reg_maps
= &shader
->reg_maps
;
4790 const struct wined3d_gl_info
*gl_info
= context
->gl_info
;
4791 const DWORD
*function
= shader
->function
;
4792 struct shader_glsl_ctx_priv priv_ctx
;
4795 shader_id
= GL_EXTCALL(glCreateShader(GL_GEOMETRY_SHADER
));
4797 shader_addline(buffer
, "%s\n", shader_glsl_get_version(gl_info
, ®_maps
->shader_version
));
4799 if (gl_info
->supported
[ARB_GEOMETRY_SHADER4
])
4800 shader_addline(buffer
, "#extension GL_ARB_geometry_shader4 : enable\n");
4801 if (gl_info
->supported
[ARB_SHADER_BIT_ENCODING
])
4802 shader_addline(buffer
, "#extension GL_ARB_shader_bit_encoding : enable\n");
4803 if (gl_info
->supported
[ARB_UNIFORM_BUFFER_OBJECT
])
4804 shader_addline(buffer
, "#extension GL_ARB_uniform_buffer_object : enable\n");
4805 if (gl_info
->supported
[EXT_GPU_SHADER4
])
4806 shader_addline(buffer
, "#extension GL_EXT_gpu_shader4 : enable\n");
4808 memset(&priv_ctx
, 0, sizeof(priv_ctx
));
4809 shader_generate_glsl_declarations(context
, buffer
, shader
, reg_maps
, &priv_ctx
);
4810 shader_generate_main(shader
, buffer
, reg_maps
, function
, &priv_ctx
);
4811 shader_addline(buffer
, "}\n");
4813 TRACE("Compiling shader object %u.\n", shader_id
);
4814 shader_glsl_compile(gl_info
, shader_id
, buffer
->buffer
);
4819 static GLuint
find_glsl_pshader(const struct wined3d_context
*context
,
4820 struct wined3d_shader_buffer
*buffer
, struct wined3d_shader
*shader
,
4821 const struct ps_compile_args
*args
, const struct ps_np2fixup_info
**np2fixup_info
)
4823 struct glsl_ps_compiled_shader
*gl_shaders
, *new_array
;
4824 struct glsl_shader_private
*shader_data
;
4825 struct ps_np2fixup_info
*np2fixup
;
4830 if (!shader
->backend_data
)
4832 shader
->backend_data
= HeapAlloc(GetProcessHeap(), HEAP_ZERO_MEMORY
, sizeof(*shader_data
));
4833 if (!shader
->backend_data
)
4835 ERR("Failed to allocate backend data.\n");
4839 shader_data
= shader
->backend_data
;
4840 gl_shaders
= shader_data
->gl_shaders
.ps
;
4842 /* Usually we have very few GL shaders for each d3d shader(just 1 or maybe 2),
4843 * so a linear search is more performant than a hashmap or a binary search
4844 * (cache coherency etc)
4846 for (i
= 0; i
< shader_data
->num_gl_shaders
; ++i
)
4848 if (!memcmp(&gl_shaders
[i
].args
, args
, sizeof(*args
)))
4850 if (args
->np2_fixup
)
4851 *np2fixup_info
= &gl_shaders
[i
].np2fixup
;
4852 return gl_shaders
[i
].id
;
4856 TRACE("No matching GL shader found for shader %p, compiling a new shader.\n", shader
);
4857 if(shader_data
->shader_array_size
== shader_data
->num_gl_shaders
) {
4858 if (shader_data
->num_gl_shaders
)
4860 new_size
= shader_data
->shader_array_size
+ max(1, shader_data
->shader_array_size
/ 2);
4861 new_array
= HeapReAlloc(GetProcessHeap(), 0, shader_data
->gl_shaders
.ps
,
4862 new_size
* sizeof(*gl_shaders
));
4866 new_array
= HeapAlloc(GetProcessHeap(), 0, sizeof(*gl_shaders
));
4871 ERR("Out of memory\n");
4874 shader_data
->gl_shaders
.ps
= new_array
;
4875 shader_data
->shader_array_size
= new_size
;
4876 gl_shaders
= new_array
;
4879 gl_shaders
[shader_data
->num_gl_shaders
].args
= *args
;
4881 np2fixup
= &gl_shaders
[shader_data
->num_gl_shaders
].np2fixup
;
4882 memset(np2fixup
, 0, sizeof(*np2fixup
));
4883 *np2fixup_info
= args
->np2_fixup
? np2fixup
: NULL
;
4885 pixelshader_update_resource_types(shader
, args
->tex_types
);
4887 shader_buffer_clear(buffer
);
4888 ret
= shader_glsl_generate_pshader(context
, buffer
, shader
, args
, np2fixup
);
4889 gl_shaders
[shader_data
->num_gl_shaders
++].id
= ret
;
4894 static inline BOOL
vs_args_equal(const struct vs_compile_args
*stored
, const struct vs_compile_args
*new,
4895 const DWORD use_map
) {
4896 if((stored
->swizzle_map
& use_map
) != new->swizzle_map
) return FALSE
;
4897 if((stored
->clip_enabled
) != new->clip_enabled
) return FALSE
;
4898 return stored
->fog_src
== new->fog_src
;
4901 static GLuint
find_glsl_vshader(const struct wined3d_context
*context
,
4902 struct wined3d_shader_buffer
*buffer
, struct wined3d_shader
*shader
,
4903 const struct vs_compile_args
*args
)
4907 DWORD use_map
= context
->stream_info
.use_map
;
4908 struct glsl_vs_compiled_shader
*gl_shaders
, *new_array
;
4909 struct glsl_shader_private
*shader_data
;
4912 if (!shader
->backend_data
)
4914 shader
->backend_data
= HeapAlloc(GetProcessHeap(), HEAP_ZERO_MEMORY
, sizeof(*shader_data
));
4915 if (!shader
->backend_data
)
4917 ERR("Failed to allocate backend data.\n");
4921 shader_data
= shader
->backend_data
;
4922 gl_shaders
= shader_data
->gl_shaders
.vs
;
4924 /* Usually we have very few GL shaders for each d3d shader(just 1 or maybe 2),
4925 * so a linear search is more performant than a hashmap or a binary search
4926 * (cache coherency etc)
4928 for (i
= 0; i
< shader_data
->num_gl_shaders
; ++i
)
4930 if (vs_args_equal(&gl_shaders
[i
].args
, args
, use_map
))
4931 return gl_shaders
[i
].id
;
4934 TRACE("No matching GL shader found for shader %p, compiling a new shader.\n", shader
);
4936 if(shader_data
->shader_array_size
== shader_data
->num_gl_shaders
) {
4937 if (shader_data
->num_gl_shaders
)
4939 new_size
= shader_data
->shader_array_size
+ max(1, shader_data
->shader_array_size
/ 2);
4940 new_array
= HeapReAlloc(GetProcessHeap(), 0, shader_data
->gl_shaders
.vs
,
4941 new_size
* sizeof(*gl_shaders
));
4945 new_array
= HeapAlloc(GetProcessHeap(), 0, sizeof(*gl_shaders
));
4950 ERR("Out of memory\n");
4953 shader_data
->gl_shaders
.vs
= new_array
;
4954 shader_data
->shader_array_size
= new_size
;
4955 gl_shaders
= new_array
;
4958 gl_shaders
[shader_data
->num_gl_shaders
].args
= *args
;
4960 shader_buffer_clear(buffer
);
4961 ret
= shader_glsl_generate_vshader(context
, buffer
, shader
, args
);
4962 gl_shaders
[shader_data
->num_gl_shaders
++].id
= ret
;
4967 static GLuint
find_glsl_geometry_shader(const struct wined3d_context
*context
,
4968 struct wined3d_shader_buffer
*buffer
, struct wined3d_shader
*shader
)
4970 struct glsl_gs_compiled_shader
*gl_shaders
;
4971 struct glsl_shader_private
*shader_data
;
4974 if (!shader
->backend_data
)
4976 if (!(shader
->backend_data
= HeapAlloc(GetProcessHeap(), HEAP_ZERO_MEMORY
, sizeof(*shader_data
))))
4978 ERR("Failed to allocate backend data.\n");
4982 shader_data
= shader
->backend_data
;
4983 gl_shaders
= shader_data
->gl_shaders
.gs
;
4985 if (shader_data
->num_gl_shaders
)
4986 return gl_shaders
[0].id
;
4988 TRACE("No matching GL shader found for shader %p, compiling a new shader.\n", shader
);
4990 if (!(shader_data
->gl_shaders
.gs
= HeapAlloc(GetProcessHeap(), 0, sizeof(*gl_shaders
))))
4992 ERR("Failed to allocate GL shader array.\n");
4995 shader_data
->shader_array_size
= 1;
4996 gl_shaders
= shader_data
->gl_shaders
.gs
;
4998 shader_buffer_clear(buffer
);
4999 ret
= shader_glsl_generate_geometry_shader(context
, buffer
, shader
);
5000 gl_shaders
[shader_data
->num_gl_shaders
++].id
= ret
;
5005 static const char *shader_glsl_ffp_mcs(enum wined3d_material_color_source mcs
, const char *material
)
5009 case WINED3D_MCS_MATERIAL
:
5011 case WINED3D_MCS_COLOR1
:
5013 case WINED3D_MCS_COLOR2
:
5014 return "gl_SecondaryColor";
5016 ERR("Invalid material color source %#x.\n", mcs
);
5021 static void shader_glsl_ffp_vertex_lighting(struct wined3d_shader_buffer
*buffer
,
5022 const struct wined3d_ffp_vs_settings
*settings
, const struct wined3d_gl_info
*gl_info
)
5024 const char *diffuse
, *specular
, *emission
, *ambient
;
5025 enum wined3d_light_type light_type
;
5028 if (!settings
->lighting
)
5030 shader_addline(buffer
, "gl_FrontColor = gl_Color;\n");
5031 shader_addline(buffer
, "gl_FrontSecondaryColor = gl_SecondaryColor;\n");
5035 shader_addline(buffer
, "vec3 ambient = gl_LightModel.ambient.xyz;\n");
5036 shader_addline(buffer
, "vec3 diffuse = vec3(0.0);\n");
5037 shader_addline(buffer
, "vec4 specular = vec4(0.0);\n");
5038 shader_addline(buffer
, "vec3 dir, dst;\n");
5039 shader_addline(buffer
, "float att, t;\n");
5041 ambient
= shader_glsl_ffp_mcs(settings
->ambient_source
, "gl_FrontMaterial.ambient");
5042 diffuse
= shader_glsl_ffp_mcs(settings
->diffuse_source
, "gl_FrontMaterial.diffuse");
5043 specular
= shader_glsl_ffp_mcs(settings
->specular_source
, "gl_FrontMaterial.specular");
5044 emission
= shader_glsl_ffp_mcs(settings
->emission_source
, "gl_FrontMaterial.emission");
5046 for (i
= 0; i
< MAX_ACTIVE_LIGHTS
; ++i
)
5048 light_type
= (settings
->light_type
>> WINED3D_FFP_LIGHT_TYPE_SHIFT(i
)) & WINED3D_FFP_LIGHT_TYPE_MASK
;
5051 case WINED3D_LIGHT_POINT
:
5052 shader_addline(buffer
, "dir = gl_LightSource[%u].position.xyz - ec_pos.xyz;\n", i
);
5053 shader_addline(buffer
, "dst.z = dot(dir, dir);\n");
5054 shader_addline(buffer
, "dst.y = sqrt(dst.z);\n");
5055 shader_addline(buffer
, "dst.x = 1.0;\n");
5056 shader_addline(buffer
, "att = dot(dst.xyz, vec3(gl_LightSource[%u].constantAttenuation,"
5057 " gl_LightSource[%u].linearAttenuation, gl_LightSource[%u].quadraticAttenuation));\n", i
, i
, i
);
5058 shader_addline(buffer
, "ambient += gl_LightSource[%u].ambient.xyz / att;\n", i
);
5059 if (!settings
->normal
)
5061 shader_addline(buffer
, "dir = normalize(dir);\n");
5062 shader_addline(buffer
, "diffuse += (clamp(dot(dir, normal), 0.0, 1.0)"
5063 " * gl_LightSource[%u].diffuse.xyz) / att;\n", i
);
5064 if (settings
->localviewer
)
5065 shader_addline(buffer
, "t = dot(normal, normalize(dir - normalize(ec_pos.xyz)));\n");
5067 shader_addline(buffer
, "t = dot(normal, normalize(dir + vec3(0.0, 0.0, 1.0)));\n");
5068 shader_addline(buffer
, "if (t > 0.0) specular += (pow(t, gl_FrontMaterial.shininess)"
5069 " * gl_LightSource[%u].specular) / att;\n", i
);
5072 case WINED3D_LIGHT_SPOT
:
5073 shader_addline(buffer
, "dir = gl_LightSource[%u].position.xyz - ec_pos.xyz;\n", i
);
5074 shader_addline(buffer
, "dst.z = dot(dir, dir);\n");
5075 shader_addline(buffer
, "dst.y = sqrt(dst.z);\n");
5076 shader_addline(buffer
, "dst.x = 1.0;\n");
5077 shader_addline(buffer
, "dir = normalize(dir);\n");
5078 shader_addline(buffer
, "t = dot(-dir, normalize(gl_LightSource[%u].spotDirection));\n", i
);
5079 shader_addline(buffer
, "if (t < gl_LightSource[%u].spotCosCutoff) att = 0.0;\n", i
);
5080 shader_addline(buffer
, "else att = pow(t, gl_LightSource[%u].spotExponent)"
5081 " / dot(dst.xyz, vec3(gl_LightSource[%u].constantAttenuation,"
5082 " gl_LightSource[%u].linearAttenuation, gl_LightSource[%u].quadraticAttenuation));\n",
5084 shader_addline(buffer
, "ambient += gl_LightSource[%u].ambient.xyz * att;\n", i
);
5085 if (!settings
->normal
)
5087 shader_addline(buffer
, "diffuse += (clamp(dot(dir, normal), 0.0, 1.0)"
5088 " * gl_LightSource[%u].diffuse.xyz) * att;\n", i
);
5089 if (settings
->localviewer
)
5090 shader_addline(buffer
, "t = dot(normal, normalize(dir - normalize(ec_pos.xyz)));\n");
5092 shader_addline(buffer
, "t = dot(normal, normalize(dir + vec3(0.0, 0.0, 1.0)));\n");
5093 shader_addline(buffer
, "if (t > 0.0) specular += (pow(t, gl_FrontMaterial.shininess)"
5094 " * gl_LightSource[%u].specular) * att;\n", i
);
5097 case WINED3D_LIGHT_DIRECTIONAL
:
5098 shader_addline(buffer
, "ambient += gl_LightSource[%u].ambient.xyz;\n", i
);
5099 if (!settings
->normal
)
5101 shader_addline(buffer
, "dir = normalize(gl_LightSource[%u].position.xyz);\n", i
);
5102 shader_addline(buffer
, "diffuse += clamp(dot(dir, normal), 0.0, 1.0)"
5103 " * gl_LightSource[%u].diffuse.xyz;\n", i
);
5104 shader_addline(buffer
, "t = dot(normal, gl_LightSource[%u].halfVector.xyz);\n", i
);
5105 shader_addline(buffer
, "if (t > 0.0) specular += pow(t, gl_FrontMaterial.shininess)"
5106 " * gl_LightSource[%u].specular;\n", i
);
5111 FIXME("Unhandled light type %#x.\n", light_type
);
5116 shader_addline(buffer
, "gl_FrontColor.xyz = %s.xyz * ambient + %s.xyz * diffuse + %s.xyz;\n",
5117 ambient
, diffuse
, emission
);
5118 shader_addline(buffer
, "gl_FrontColor.w = %s.w;\n", diffuse
);
5119 shader_addline(buffer
, "gl_FrontSecondaryColor = %s * specular;\n", specular
);
5122 /* Context activation is done by the caller. */
5123 static GLuint
shader_glsl_generate_ffp_vertex_shader(struct wined3d_shader_buffer
*buffer
,
5124 const struct wined3d_ffp_vs_settings
*settings
, const struct wined3d_gl_info
*gl_info
)
5129 shader_buffer_clear(buffer
);
5131 shader_addline(buffer
, "#version 120\n");
5132 shader_addline(buffer
, "\n");
5134 shader_addline(buffer
, "uniform mat4 ffp_modelview_matrix;\n");
5135 shader_addline(buffer
, "uniform mat4 ffp_projection_matrix;\n");
5136 shader_addline(buffer
, "uniform mat3 ffp_normal_matrix;\n");
5138 shader_addline(buffer
, "\nvoid main()\n{\n");
5139 shader_addline(buffer
, "float m;\n");
5140 shader_addline(buffer
, "vec3 r;\n");
5142 if (settings
->transformed
)
5144 shader_addline(buffer
, "vec4 ec_pos = vec4(gl_Vertex.xyz, 1.0);\n");
5145 shader_addline(buffer
, "gl_Position = ffp_projection_matrix * ec_pos;\n");
5146 shader_addline(buffer
, "if (gl_Vertex.w != 0.0) gl_Position /= gl_Vertex.w;\n");
5150 shader_addline(buffer
, "vec4 ec_pos = ffp_modelview_matrix * gl_Vertex;\n");
5151 shader_addline(buffer
, "gl_Position = ffp_projection_matrix * ec_pos;\n");
5152 if (settings
->clipping
)
5153 shader_addline(buffer
, "gl_ClipVertex = ec_pos;\n");
5154 shader_addline(buffer
, "ec_pos /= ec_pos.w;\n");
5157 if (!settings
->normal
)
5158 shader_addline(buffer
, "vec3 normal = vec3(0.0);\n");
5159 else if (settings
->normalize
)
5160 shader_addline(buffer
, "vec3 normal = normalize(ffp_normal_matrix * gl_Normal);\n");
5162 shader_addline(buffer
, "vec3 normal = ffp_normal_matrix * gl_Normal;\n");
5164 shader_glsl_ffp_vertex_lighting(buffer
, settings
, gl_info
);
5166 for (i
= 0; i
< MAX_TEXTURES
; ++i
)
5168 switch (settings
->texgen
[i
] << WINED3D_FFP_TCI_SHIFT
)
5170 case WINED3DTSS_TCI_PASSTHRU
:
5171 if (settings
->texcoords
& (1 << i
))
5172 shader_addline(buffer
, "gl_TexCoord[%u] = gl_TextureMatrix[%u] * gl_MultiTexCoord%d;\n",
5176 case WINED3DTSS_TCI_CAMERASPACENORMAL
:
5177 shader_addline(buffer
, "gl_TexCoord[%u] = gl_TextureMatrix[%u] * vec4(normal, 1.0);\n", i
, i
);
5180 case WINED3DTSS_TCI_CAMERASPACEPOSITION
:
5181 shader_addline(buffer
, "gl_TexCoord[%u] = gl_TextureMatrix[%u] * ec_pos;\n", i
, i
);
5184 case WINED3DTSS_TCI_CAMERASPACEREFLECTIONVECTOR
:
5185 shader_addline(buffer
, "gl_TexCoord[%u] = gl_TextureMatrix[%u]"
5186 " * vec4(reflect(normalize(ec_pos.xyz), normal), 1.0);\n", i
, i
);
5189 case WINED3DTSS_TCI_SPHEREMAP
:
5190 shader_addline(buffer
, "r = reflect(normalize(ec_pos.xyz), normal);\n");
5191 shader_addline(buffer
, "m = 2.0 * length(vec3(r.x, r.y, r.z + 1.0));\n");
5192 shader_addline(buffer
, "gl_TexCoord[%u] = gl_TextureMatrix[%u]"
5193 " * vec4(r.x / m + 0.5, r.y / m + 0.5, 0.0, 1.0);\n", i
, i
);
5197 ERR("Unhandled texgen %#x.\n", settings
->texgen
[i
]);
5202 switch (settings
->fog_mode
)
5204 case WINED3D_FFP_VS_FOG_OFF
:
5207 case WINED3D_FFP_VS_FOG_FOGCOORD
:
5208 shader_addline(buffer
, "gl_FogFragCoord = gl_SecondaryColor.w * 255.0;\n");
5211 case WINED3D_FFP_VS_FOG_RANGE
:
5212 shader_addline(buffer
, "gl_FogFragCoord = length(ec_pos.xyz);\n");
5215 case WINED3D_FFP_VS_FOG_DEPTH
:
5216 if (settings
->ortho_fog
)
5217 /* Need to undo the [0.0 - 1.0] -> [-1.0 - 1.0] transformation from D3D to GL coordinates. */
5218 shader_addline(buffer
, "gl_FogFragCoord = gl_Position.z * 0.5 + 0.5;\n");
5219 else if (settings
->transformed
)
5220 shader_addline(buffer
, "gl_FogFragCoord = ec_pos.z;\n");
5222 shader_addline(buffer
, "gl_FogFragCoord = abs(ec_pos.z);\n");
5226 ERR("Unhandled fog mode %#x.\n", settings
->fog_mode
);
5230 if (settings
->point_size
)
5232 shader_addline(buffer
, "gl_PointSize = gl_Point.size / sqrt(gl_Point.distanceConstantAttenuation"
5233 " + gl_Point.distanceLinearAttenuation * length(ec_pos.xyz)"
5234 " + gl_Point.distanceQuadraticAttenuation * dot(ec_pos.xyz, ec_pos.xyz));\n");
5235 shader_addline(buffer
, "gl_PointSize = clamp(gl_PointSize, gl_Point.sizeMin, gl_Point.sizeMax);\n");
5238 shader_addline(buffer
, "}\n");
5240 shader_obj
= GL_EXTCALL(glCreateShader(GL_VERTEX_SHADER
));
5241 shader_glsl_compile(gl_info
, shader_obj
, buffer
->buffer
);
5246 static const char *shader_glsl_get_ffp_fragment_op_arg(struct wined3d_shader_buffer
*buffer
,
5247 DWORD argnum
, unsigned int stage
, DWORD arg
)
5251 if (arg
== ARG_UNUSED
)
5252 return "<unused arg>";
5254 switch (arg
& WINED3DTA_SELECTMASK
)
5256 case WINED3DTA_DIFFUSE
:
5260 case WINED3DTA_CURRENT
:
5267 case WINED3DTA_TEXTURE
:
5270 case 0: ret
= "tex0"; break;
5271 case 1: ret
= "tex1"; break;
5272 case 2: ret
= "tex2"; break;
5273 case 3: ret
= "tex3"; break;
5274 case 4: ret
= "tex4"; break;
5275 case 5: ret
= "tex5"; break;
5276 case 6: ret
= "tex6"; break;
5277 case 7: ret
= "tex7"; break;
5279 ret
= "<invalid texture>";
5284 case WINED3DTA_TFACTOR
:
5288 case WINED3DTA_SPECULAR
:
5289 ret
= "gl_SecondaryColor";
5292 case WINED3DTA_TEMP
:
5296 case WINED3DTA_CONSTANT
:
5299 case 0: ret
= "tss_const0"; break;
5300 case 1: ret
= "tss_const1"; break;
5301 case 2: ret
= "tss_const2"; break;
5302 case 3: ret
= "tss_const3"; break;
5303 case 4: ret
= "tss_const4"; break;
5304 case 5: ret
= "tss_const5"; break;
5305 case 6: ret
= "tss_const6"; break;
5306 case 7: ret
= "tss_const7"; break;
5308 ret
= "<invalid constant>";
5314 return "<unhandled arg>";
5317 if (arg
& WINED3DTA_COMPLEMENT
)
5319 shader_addline(buffer
, "arg%u = vec4(1.0) - %s;\n", argnum
, ret
);
5322 else if (argnum
== 1)
5324 else if (argnum
== 2)
5328 if (arg
& WINED3DTA_ALPHAREPLICATE
)
5330 shader_addline(buffer
, "arg%u = vec4(%s.w);\n", argnum
, ret
);
5333 else if (argnum
== 1)
5335 else if (argnum
== 2)
5342 static void shader_glsl_ffp_fragment_op(struct wined3d_shader_buffer
*buffer
, unsigned int stage
, BOOL color
,
5343 BOOL alpha
, DWORD dst
, DWORD op
, DWORD dw_arg0
, DWORD dw_arg1
, DWORD dw_arg2
)
5345 const char *dstmask
, *dstreg
, *arg0
, *arg1
, *arg2
;
5355 dstreg
= "temp_reg";
5359 arg0
= shader_glsl_get_ffp_fragment_op_arg(buffer
, 0, stage
, dw_arg0
);
5360 arg1
= shader_glsl_get_ffp_fragment_op_arg(buffer
, 1, stage
, dw_arg1
);
5361 arg2
= shader_glsl_get_ffp_fragment_op_arg(buffer
, 2, stage
, dw_arg2
);
5365 case WINED3D_TOP_DISABLE
:
5367 shader_addline(buffer
, "%s%s = gl_Color%s;\n", dstreg
, dstmask
, dstmask
);
5370 case WINED3D_TOP_SELECT_ARG1
:
5371 shader_addline(buffer
, "%s%s = %s%s;\n", dstreg
, dstmask
, arg1
, dstmask
);
5374 case WINED3D_TOP_SELECT_ARG2
:
5375 shader_addline(buffer
, "%s%s = %s%s;\n", dstreg
, dstmask
, arg2
, dstmask
);
5378 case WINED3D_TOP_MODULATE
:
5379 shader_addline(buffer
, "%s%s = %s%s * %s%s;\n", dstreg
, dstmask
, arg1
, dstmask
, arg2
, dstmask
);
5382 case WINED3D_TOP_MODULATE_4X
:
5383 shader_addline(buffer
, "%s%s = clamp(%s%s * %s%s * 4.0, 0.0, 1.0);\n",
5384 dstreg
, dstmask
, arg1
, dstmask
, arg2
, dstmask
);
5387 case WINED3D_TOP_MODULATE_2X
:
5388 shader_addline(buffer
, "%s%s = clamp(%s%s * %s%s * 2.0, 0.0, 1.0);\n",
5389 dstreg
, dstmask
, arg1
, dstmask
, arg2
, dstmask
);
5392 case WINED3D_TOP_ADD
:
5393 shader_addline(buffer
, "%s%s = clamp(%s%s + %s%s, 0.0, 1.0);\n",
5394 dstreg
, dstmask
, arg1
, dstmask
, arg2
, dstmask
);
5397 case WINED3D_TOP_ADD_SIGNED
:
5398 shader_addline(buffer
, "%s%s = clamp(%s%s + (%s - vec4(0.5))%s, 0.0, 1.0);\n",
5399 dstreg
, dstmask
, arg1
, dstmask
, arg2
, dstmask
);
5402 case WINED3D_TOP_ADD_SIGNED_2X
:
5403 shader_addline(buffer
, "%s%s = clamp((%s%s + (%s - vec4(0.5))%s) * 2.0, 0.0, 1.0);\n",
5404 dstreg
, dstmask
, arg1
, dstmask
, arg2
, dstmask
);
5407 case WINED3D_TOP_SUBTRACT
:
5408 shader_addline(buffer
, "%s%s = clamp(%s%s - %s%s, 0.0, 1.0);\n",
5409 dstreg
, dstmask
, arg1
, dstmask
, arg2
, dstmask
);
5412 case WINED3D_TOP_ADD_SMOOTH
:
5413 shader_addline(buffer
, "%s%s = clamp((vec4(1.0) - %s)%s * %s%s + %s%s, 0.0, 1.0);\n",
5414 dstreg
, dstmask
, arg1
, dstmask
, arg2
, dstmask
, arg1
, dstmask
);
5417 case WINED3D_TOP_BLEND_DIFFUSE_ALPHA
:
5418 arg0
= shader_glsl_get_ffp_fragment_op_arg(buffer
, 0, stage
, WINED3DTA_DIFFUSE
);
5419 shader_addline(buffer
, "%s%s = mix(%s%s, %s%s, %s.w);\n",
5420 dstreg
, dstmask
, arg2
, dstmask
, arg1
, dstmask
, arg0
);
5423 case WINED3D_TOP_BLEND_TEXTURE_ALPHA
:
5424 arg0
= shader_glsl_get_ffp_fragment_op_arg(buffer
, 0, stage
, WINED3DTA_TEXTURE
);
5425 shader_addline(buffer
, "%s%s = mix(%s%s, %s%s, %s.w);\n",
5426 dstreg
, dstmask
, arg2
, dstmask
, arg1
, dstmask
, arg0
);
5429 case WINED3D_TOP_BLEND_FACTOR_ALPHA
:
5430 arg0
= shader_glsl_get_ffp_fragment_op_arg(buffer
, 0, stage
, WINED3DTA_TFACTOR
);
5431 shader_addline(buffer
, "%s%s = mix(%s%s, %s%s, %s.w);\n",
5432 dstreg
, dstmask
, arg2
, dstmask
, arg1
, dstmask
, arg0
);
5435 case WINED3D_TOP_BLEND_TEXTURE_ALPHA_PM
:
5436 arg0
= shader_glsl_get_ffp_fragment_op_arg(buffer
, 0, stage
, WINED3DTA_TEXTURE
);
5437 shader_addline(buffer
, "%s%s = clamp(%s%s * (1.0 - %s.w) + %s%s, 0.0, 1.0);\n",
5438 dstreg
, dstmask
, arg2
, dstmask
, arg0
, arg1
, dstmask
);
5441 case WINED3D_TOP_BLEND_CURRENT_ALPHA
:
5442 arg0
= shader_glsl_get_ffp_fragment_op_arg(buffer
, 0, stage
, WINED3DTA_CURRENT
);
5443 shader_addline(buffer
, "%s%s = mix(%s%s, %s%s, %s.w);\n",
5444 dstreg
, dstmask
, arg2
, dstmask
, arg1
, dstmask
, arg0
);
5447 case WINED3D_TOP_MODULATE_ALPHA_ADD_COLOR
:
5448 shader_addline(buffer
, "%s%s = clamp(%s%s * %s.w + %s%s, 0.0, 1.0);\n",
5449 dstreg
, dstmask
, arg2
, dstmask
, arg1
, arg1
, dstmask
);
5452 case WINED3D_TOP_MODULATE_COLOR_ADD_ALPHA
:
5453 shader_addline(buffer
, "%s%s = clamp(%s%s * %s%s + %s.w, 0.0, 1.0);\n",
5454 dstreg
, dstmask
, arg1
, dstmask
, arg2
, dstmask
, arg1
);
5457 case WINED3D_TOP_MODULATE_INVALPHA_ADD_COLOR
:
5458 shader_addline(buffer
, "%s%s = clamp(%s%s * (1.0 - %s.w) + %s%s, 0.0, 1.0);\n",
5459 dstreg
, dstmask
, arg2
, dstmask
, arg1
, arg1
, dstmask
);
5461 case WINED3D_TOP_MODULATE_INVCOLOR_ADD_ALPHA
:
5462 shader_addline(buffer
, "%s%s = clamp((vec4(1.0) - %s)%s * %s%s + %s.w, 0.0, 1.0);\n",
5463 dstreg
, dstmask
, arg1
, dstmask
, arg2
, dstmask
, arg1
);
5466 case WINED3D_TOP_BUMPENVMAP
:
5467 case WINED3D_TOP_BUMPENVMAP_LUMINANCE
:
5468 /* These are handled in the first pass, nothing to do. */
5471 case WINED3D_TOP_DOTPRODUCT3
:
5472 shader_addline(buffer
, "%s%s = vec4(clamp(dot(%s.xyz - 0.5, %s.xyz - 0.5) * 4.0, 0.0, 1.0))%s;\n",
5473 dstreg
, dstmask
, arg1
, arg2
, dstmask
);
5476 case WINED3D_TOP_MULTIPLY_ADD
:
5477 shader_addline(buffer
, "%s%s = clamp(%s%s * %s%s + %s%s, 0.0, 1.0);\n",
5478 dstreg
, dstmask
, arg1
, dstmask
, arg2
, dstmask
, arg0
, dstmask
);
5481 case WINED3D_TOP_LERP
:
5482 /* MSDN isn't quite right here. */
5483 shader_addline(buffer
, "%s%s = mix(%s%s, %s%s, %s%s);\n",
5484 dstreg
, dstmask
, arg2
, dstmask
, arg1
, dstmask
, arg0
, dstmask
);
5488 FIXME("Unhandled operation %#x.\n", op
);
5493 /* Context activation is done by the caller. */
5494 static GLuint
shader_glsl_generate_ffp_fragment_shader(struct wined3d_shader_buffer
*buffer
,
5495 const struct ffp_frag_settings
*settings
, const struct wined3d_gl_info
*gl_info
)
5497 BYTE lum_map
= 0, bump_map
= 0, tex_map
= 0, tss_const_map
= 0;
5498 BOOL tempreg_used
= FALSE
, tfactor_used
= FALSE
;
5499 const char *final_combiner_src
= "ret";
5500 UINT lowest_disabled_stage
;
5502 DWORD arg0
, arg1
, arg2
;
5505 shader_buffer_clear(buffer
);
5507 /* Find out which textures are read */
5508 for (stage
= 0; stage
< MAX_TEXTURES
; ++stage
)
5510 if (settings
->op
[stage
].cop
== WINED3D_TOP_DISABLE
)
5513 arg0
= settings
->op
[stage
].carg0
& WINED3DTA_SELECTMASK
;
5514 arg1
= settings
->op
[stage
].carg1
& WINED3DTA_SELECTMASK
;
5515 arg2
= settings
->op
[stage
].carg2
& WINED3DTA_SELECTMASK
;
5517 if (arg0
== WINED3DTA_TEXTURE
|| arg1
== WINED3DTA_TEXTURE
|| arg2
== WINED3DTA_TEXTURE
5518 || (stage
== 0 && settings
->color_key_enabled
))
5519 tex_map
|= 1 << stage
;
5520 if (arg0
== WINED3DTA_TFACTOR
|| arg1
== WINED3DTA_TFACTOR
|| arg2
== WINED3DTA_TFACTOR
)
5521 tfactor_used
= TRUE
;
5522 if (arg0
== WINED3DTA_TEMP
|| arg1
== WINED3DTA_TEMP
|| arg2
== WINED3DTA_TEMP
)
5523 tempreg_used
= TRUE
;
5524 if (settings
->op
[stage
].dst
== tempreg
)
5525 tempreg_used
= TRUE
;
5526 if (arg0
== WINED3DTA_CONSTANT
|| arg1
== WINED3DTA_CONSTANT
|| arg2
== WINED3DTA_CONSTANT
)
5527 tss_const_map
|= 1 << stage
;
5529 switch (settings
->op
[stage
].cop
)
5531 case WINED3D_TOP_BUMPENVMAP_LUMINANCE
:
5532 lum_map
|= 1 << stage
;
5534 case WINED3D_TOP_BUMPENVMAP
:
5535 bump_map
|= 1 << stage
;
5537 case WINED3D_TOP_BLEND_TEXTURE_ALPHA
:
5538 case WINED3D_TOP_BLEND_TEXTURE_ALPHA_PM
:
5539 tex_map
|= 1 << stage
;
5542 case WINED3D_TOP_BLEND_FACTOR_ALPHA
:
5543 tfactor_used
= TRUE
;
5550 if (settings
->op
[stage
].aop
== WINED3D_TOP_DISABLE
)
5553 arg0
= settings
->op
[stage
].aarg0
& WINED3DTA_SELECTMASK
;
5554 arg1
= settings
->op
[stage
].aarg1
& WINED3DTA_SELECTMASK
;
5555 arg2
= settings
->op
[stage
].aarg2
& WINED3DTA_SELECTMASK
;
5557 if (arg0
== WINED3DTA_TEXTURE
|| arg1
== WINED3DTA_TEXTURE
|| arg2
== WINED3DTA_TEXTURE
)
5558 tex_map
|= 1 << stage
;
5559 if (arg0
== WINED3DTA_TFACTOR
|| arg1
== WINED3DTA_TFACTOR
|| arg2
== WINED3DTA_TFACTOR
)
5560 tfactor_used
= TRUE
;
5561 if (arg0
== WINED3DTA_TEMP
|| arg1
== WINED3DTA_TEMP
|| arg2
== WINED3DTA_TEMP
)
5562 tempreg_used
= TRUE
;
5563 if (arg0
== WINED3DTA_CONSTANT
|| arg1
== WINED3DTA_CONSTANT
|| arg2
== WINED3DTA_CONSTANT
)
5564 tss_const_map
|= 1 << stage
;
5566 lowest_disabled_stage
= stage
;
5568 shader_addline(buffer
, "#version 120\n");
5570 if (gl_info
->supported
[ARB_TEXTURE_RECTANGLE
])
5571 shader_addline(buffer
, "#extension GL_ARB_texture_rectangle : enable\n");
5573 shader_addline(buffer
, "vec4 tmp0, tmp1;\n");
5574 shader_addline(buffer
, "vec4 ret;\n");
5575 if (tempreg_used
|| settings
->sRGB_write
)
5576 shader_addline(buffer
, "vec4 temp_reg = vec4(0.0);\n");
5577 shader_addline(buffer
, "vec4 arg0, arg1, arg2;\n");
5579 for (stage
= 0; stage
< MAX_TEXTURES
; ++stage
)
5581 if (tss_const_map
& (1 << stage
))
5582 shader_addline(buffer
, "uniform vec4 tss_const%u;\n", stage
);
5584 if (!(tex_map
& (1 << stage
)))
5587 switch (settings
->op
[stage
].tex_type
)
5589 case WINED3D_GL_RES_TYPE_TEX_1D
:
5590 shader_addline(buffer
, "uniform sampler1D ps_sampler%u;\n", stage
);
5592 case WINED3D_GL_RES_TYPE_TEX_2D
:
5593 shader_addline(buffer
, "uniform sampler2D ps_sampler%u;\n", stage
);
5595 case WINED3D_GL_RES_TYPE_TEX_3D
:
5596 shader_addline(buffer
, "uniform sampler3D ps_sampler%u;\n", stage
);
5598 case WINED3D_GL_RES_TYPE_TEX_CUBE
:
5599 shader_addline(buffer
, "uniform samplerCube ps_sampler%u;\n", stage
);
5601 case WINED3D_GL_RES_TYPE_TEX_RECT
:
5602 shader_addline(buffer
, "uniform sampler2DRect ps_sampler%u;\n", stage
);
5605 FIXME("Unhandled sampler type %#x.\n", settings
->op
[stage
].tex_type
);
5609 shader_addline(buffer
, "vec4 tex%u;\n", stage
);
5611 if (!(bump_map
& (1 << stage
)))
5613 shader_addline(buffer
, "uniform mat2 bumpenv_mat%u;\n", stage
);
5615 if (!(lum_map
& (1 << stage
)))
5617 shader_addline(buffer
, "uniform float bumpenv_lum_scale%u;\n", stage
);
5618 shader_addline(buffer
, "uniform float bumpenv_lum_offset%u;\n", stage
);
5621 shader_addline(buffer
, "uniform vec4 tex_factor;\n");
5622 if (settings
->color_key_enabled
)
5623 shader_addline(buffer
, "uniform vec4 color_key;\n");
5624 shader_addline(buffer
, "uniform vec4 specular_enable;\n");
5626 if (settings
->sRGB_write
)
5628 shader_addline(buffer
, "const vec4 srgb_const0 = ");
5629 shader_glsl_append_imm_vec4(buffer
, wined3d_srgb_const0
);
5630 shader_addline(buffer
, ";\n");
5631 shader_addline(buffer
, "const vec4 srgb_const1 = ");
5632 shader_glsl_append_imm_vec4(buffer
, wined3d_srgb_const1
);
5633 shader_addline(buffer
, ";\n");
5636 shader_addline(buffer
, "void main()\n{\n");
5638 if (lowest_disabled_stage
< 7 && settings
->emul_clipplanes
)
5639 shader_addline(buffer
, "if (any(lessThan(gl_TexCoord[7], vec4(0.0)))) discard;\n");
5641 /* Generate texture sampling instructions) */
5642 for (stage
= 0; stage
< MAX_TEXTURES
&& settings
->op
[stage
].cop
!= WINED3D_TOP_DISABLE
; ++stage
)
5644 const char *texture_function
, *coord_mask
;
5645 char tex_reg_name
[8];
5648 if (!(tex_map
& (1 << stage
)))
5651 if (settings
->op
[stage
].projected
== proj_none
)
5655 else if (settings
->op
[stage
].projected
== proj_count4
5656 || settings
->op
[stage
].projected
== proj_count3
)
5662 FIXME("Unexpected projection mode %d\n", settings
->op
[stage
].projected
);
5666 switch (settings
->op
[stage
].tex_type
)
5668 case WINED3D_GL_RES_TYPE_TEX_1D
:
5671 texture_function
= "texture1DProj";
5676 texture_function
= "texture1D";
5680 case WINED3D_GL_RES_TYPE_TEX_2D
:
5683 texture_function
= "texture2DProj";
5688 texture_function
= "texture2D";
5692 case WINED3D_GL_RES_TYPE_TEX_3D
:
5695 texture_function
= "texture3DProj";
5696 coord_mask
= "xyzw";
5700 texture_function
= "texture3D";
5704 case WINED3D_GL_RES_TYPE_TEX_CUBE
:
5705 texture_function
= "textureCube";
5708 case WINED3D_GL_RES_TYPE_TEX_RECT
:
5711 texture_function
= "texture2DRectProj";
5716 texture_function
= "texture2DRect";
5721 FIXME("Unhandled texture type %#x.\n", settings
->op
[stage
].tex_type
);
5722 texture_function
= "";
5723 coord_mask
= "xyzw";
5728 && (settings
->op
[stage
- 1].cop
== WINED3D_TOP_BUMPENVMAP
5729 || settings
->op
[stage
- 1].cop
== WINED3D_TOP_BUMPENVMAP_LUMINANCE
))
5731 shader_addline(buffer
, "ret.xy = bumpenv_mat%u * tex%u.xy;\n", stage
- 1, stage
- 1);
5733 /* With projective textures, texbem only divides the static
5734 * texture coord, not the displacement, so multiply the
5735 * displacement with the dividing parameter before passing it to
5737 if (settings
->op
[stage
].projected
!= proj_none
)
5739 if (settings
->op
[stage
].projected
== proj_count4
)
5741 shader_addline(buffer
, "ret.xy = (ret.xy * gl_TexCoord[%u].w) + gl_TexCoord[%u].xy;\n",
5743 shader_addline(buffer
, "ret.zw = gl_TexCoord[%u].ww;\n", stage
);
5747 shader_addline(buffer
, "ret.xy = (ret.xy * gl_TexCoord[%u].z) + gl_TexCoord[%u].xy;\n",
5749 shader_addline(buffer
, "ret.zw = gl_TexCoord[%u].zz;\n", stage
);
5754 shader_addline(buffer
, "ret = gl_TexCoord[%u] + ret.xyxy;\n", stage
);
5757 shader_addline(buffer
, "tex%u = %s(ps_sampler%u, ret.%s);\n",
5758 stage
, texture_function
, stage
, coord_mask
);
5760 if (settings
->op
[stage
- 1].cop
== WINED3D_TOP_BUMPENVMAP_LUMINANCE
)
5761 shader_addline(buffer
, "tex%u *= clamp(tex%u.z * bumpenv_lum_scale%u + bumpenv_lum_offset%u, 0.0, 1.0);\n",
5762 stage
, stage
- 1, stage
- 1, stage
- 1);
5764 else if (settings
->op
[stage
].projected
== proj_count3
)
5766 shader_addline(buffer
, "tex%u = %s(ps_sampler%u, gl_TexCoord[%u].xyz);\n",
5767 stage
, texture_function
, stage
, stage
);
5771 shader_addline(buffer
, "tex%u = %s(ps_sampler%u, gl_TexCoord[%u].%s);\n",
5772 stage
, texture_function
, stage
, stage
, coord_mask
);
5775 sprintf(tex_reg_name
, "tex%u", stage
);
5776 shader_glsl_color_correction_ext(buffer
, tex_reg_name
, WINED3DSP_WRITEMASK_ALL
,
5777 settings
->op
[stage
].color_fixup
);
5780 if (settings
->color_key_enabled
)
5781 shader_addline(buffer
, "if (all(equal(tex0, color_key))) discard;\n");
5783 /* Generate the main shader */
5784 for (stage
= 0; stage
< MAX_TEXTURES
; ++stage
)
5788 if (settings
->op
[stage
].cop
== WINED3D_TOP_DISABLE
)
5791 final_combiner_src
= "gl_Color";
5795 if (settings
->op
[stage
].cop
== WINED3D_TOP_SELECT_ARG1
5796 && settings
->op
[stage
].aop
== WINED3D_TOP_SELECT_ARG1
)
5797 op_equal
= settings
->op
[stage
].carg1
== settings
->op
[stage
].aarg1
;
5798 else if (settings
->op
[stage
].cop
== WINED3D_TOP_SELECT_ARG1
5799 && settings
->op
[stage
].aop
== WINED3D_TOP_SELECT_ARG2
)
5800 op_equal
= settings
->op
[stage
].carg1
== settings
->op
[stage
].aarg2
;
5801 else if (settings
->op
[stage
].cop
== WINED3D_TOP_SELECT_ARG2
5802 && settings
->op
[stage
].aop
== WINED3D_TOP_SELECT_ARG1
)
5803 op_equal
= settings
->op
[stage
].carg2
== settings
->op
[stage
].aarg1
;
5804 else if (settings
->op
[stage
].cop
== WINED3D_TOP_SELECT_ARG2
5805 && settings
->op
[stage
].aop
== WINED3D_TOP_SELECT_ARG2
)
5806 op_equal
= settings
->op
[stage
].carg2
== settings
->op
[stage
].aarg2
;
5808 op_equal
= settings
->op
[stage
].aop
== settings
->op
[stage
].cop
5809 && settings
->op
[stage
].carg0
== settings
->op
[stage
].aarg0
5810 && settings
->op
[stage
].carg1
== settings
->op
[stage
].aarg1
5811 && settings
->op
[stage
].carg2
== settings
->op
[stage
].aarg2
;
5813 if (settings
->op
[stage
].aop
== WINED3D_TOP_DISABLE
)
5815 shader_glsl_ffp_fragment_op(buffer
, stage
, TRUE
, FALSE
, settings
->op
[stage
].dst
,
5816 settings
->op
[stage
].cop
, settings
->op
[stage
].carg0
,
5817 settings
->op
[stage
].carg1
, settings
->op
[stage
].carg2
);
5819 shader_addline(buffer
, "ret.w = gl_Color.w;\n");
5823 shader_glsl_ffp_fragment_op(buffer
, stage
, TRUE
, TRUE
, settings
->op
[stage
].dst
,
5824 settings
->op
[stage
].cop
, settings
->op
[stage
].carg0
,
5825 settings
->op
[stage
].carg1
, settings
->op
[stage
].carg2
);
5829 shader_glsl_ffp_fragment_op(buffer
, stage
, TRUE
, FALSE
, settings
->op
[stage
].dst
,
5830 settings
->op
[stage
].cop
, settings
->op
[stage
].carg0
,
5831 settings
->op
[stage
].carg1
, settings
->op
[stage
].carg2
);
5832 shader_glsl_ffp_fragment_op(buffer
, stage
, FALSE
, TRUE
, settings
->op
[stage
].dst
,
5833 settings
->op
[stage
].aop
, settings
->op
[stage
].aarg0
,
5834 settings
->op
[stage
].aarg1
, settings
->op
[stage
].aarg2
);
5838 shader_addline(buffer
, "gl_FragData[0] = gl_SecondaryColor * specular_enable + %s;\n", final_combiner_src
);
5840 if (settings
->sRGB_write
)
5841 shader_glsl_generate_srgb_write_correction(buffer
);
5843 shader_glsl_generate_fog_code(buffer
, settings
->fog
);
5845 shader_addline(buffer
, "}\n");
5847 shader_id
= GL_EXTCALL(glCreateShader(GL_FRAGMENT_SHADER
));
5848 shader_glsl_compile(gl_info
, shader_id
, buffer
->buffer
);
5852 static struct glsl_ffp_vertex_shader
*shader_glsl_find_ffp_vertex_shader(struct shader_glsl_priv
*priv
,
5853 const struct wined3d_gl_info
*gl_info
, const struct wined3d_ffp_vs_settings
*settings
)
5855 struct glsl_ffp_vertex_shader
*shader
;
5856 const struct wine_rb_entry
*entry
;
5858 if ((entry
= wine_rb_get(&priv
->ffp_vertex_shaders
, settings
)))
5859 return WINE_RB_ENTRY_VALUE(entry
, struct glsl_ffp_vertex_shader
, desc
.entry
);
5861 if (!(shader
= HeapAlloc(GetProcessHeap(), 0, sizeof(*shader
))))
5864 shader
->desc
.settings
= *settings
;
5865 shader
->id
= shader_glsl_generate_ffp_vertex_shader(&priv
->shader_buffer
, settings
, gl_info
);
5866 list_init(&shader
->linked_programs
);
5867 if (wine_rb_put(&priv
->ffp_vertex_shaders
, &shader
->desc
.settings
, &shader
->desc
.entry
) == -1)
5868 ERR("Failed to insert ffp vertex shader.\n");
5873 static struct glsl_ffp_fragment_shader
*shader_glsl_find_ffp_fragment_shader(struct shader_glsl_priv
*priv
,
5874 const struct wined3d_gl_info
*gl_info
, const struct ffp_frag_settings
*args
)
5876 struct glsl_ffp_fragment_shader
*glsl_desc
;
5877 const struct ffp_frag_desc
*desc
;
5879 if ((desc
= find_ffp_frag_shader(&priv
->ffp_fragment_shaders
, args
)))
5880 return CONTAINING_RECORD(desc
, struct glsl_ffp_fragment_shader
, entry
);
5882 if (!(glsl_desc
= HeapAlloc(GetProcessHeap(), 0, sizeof(*glsl_desc
))))
5885 glsl_desc
->entry
.settings
= *args
;
5886 glsl_desc
->id
= shader_glsl_generate_ffp_fragment_shader(&priv
->shader_buffer
, args
, gl_info
);
5887 list_init(&glsl_desc
->linked_programs
);
5888 add_ffp_frag_shader(&priv
->ffp_fragment_shaders
, &glsl_desc
->entry
);
5894 static void shader_glsl_init_vs_uniform_locations(const struct wined3d_gl_info
*gl_info
,
5895 GLuint program_id
, struct glsl_vs_program
*vs
, unsigned int vs_c_count
)
5900 vs
->uniform_f_locations
= HeapAlloc(GetProcessHeap(), 0,
5901 sizeof(GLuint
) * gl_info
->limits
.glsl_vs_float_constants
);
5902 for (i
= 0; i
< vs_c_count
; ++i
)
5904 snprintf(name
, sizeof(name
), "vs_c[%u]", i
);
5905 vs
->uniform_f_locations
[i
] = GL_EXTCALL(glGetUniformLocation(program_id
, name
));
5907 memset(&vs
->uniform_f_locations
[vs_c_count
], 0xff,
5908 (gl_info
->limits
.glsl_vs_float_constants
- vs_c_count
) * sizeof(GLuint
));
5910 for (i
= 0; i
< MAX_CONST_I
; ++i
)
5912 snprintf(name
, sizeof(name
), "vs_i[%u]", i
);
5913 vs
->uniform_i_locations
[i
] = GL_EXTCALL(glGetUniformLocation(program_id
, name
));
5916 for (i
= 0; i
< MAX_CONST_B
; ++i
)
5918 snprintf(name
, sizeof(name
), "vs_b[%u]", i
);
5919 vs
->uniform_b_locations
[i
] = GL_EXTCALL(glGetUniformLocation(program_id
, name
));
5922 vs
->pos_fixup_location
= GL_EXTCALL(glGetUniformLocation(program_id
, "posFixup"));
5924 vs
->modelview_matrix_location
= GL_EXTCALL(glGetUniformLocation(program_id
, "ffp_modelview_matrix"));
5925 vs
->projection_matrix_location
= GL_EXTCALL(glGetUniformLocation(program_id
, "ffp_projection_matrix"));
5926 vs
->normal_matrix_location
= GL_EXTCALL(glGetUniformLocation(program_id
, "ffp_normal_matrix"));
5929 static void shader_glsl_init_ps_uniform_locations(const struct wined3d_gl_info
*gl_info
,
5930 GLuint program_id
, struct glsl_ps_program
*ps
, unsigned int ps_c_count
)
5935 ps
->uniform_f_locations
= HeapAlloc(GetProcessHeap(), 0,
5936 sizeof(GLuint
) * gl_info
->limits
.glsl_ps_float_constants
);
5937 for (i
= 0; i
< ps_c_count
; ++i
)
5939 snprintf(name
, sizeof(name
), "ps_c[%u]", i
);
5940 ps
->uniform_f_locations
[i
] = GL_EXTCALL(glGetUniformLocation(program_id
, name
));
5942 memset(&ps
->uniform_f_locations
[ps_c_count
], 0xff,
5943 (gl_info
->limits
.glsl_ps_float_constants
- ps_c_count
) * sizeof(GLuint
));
5945 for (i
= 0; i
< MAX_CONST_I
; ++i
)
5947 snprintf(name
, sizeof(name
), "ps_i[%u]", i
);
5948 ps
->uniform_i_locations
[i
] = GL_EXTCALL(glGetUniformLocation(program_id
, name
));
5951 for (i
= 0; i
< MAX_CONST_B
; ++i
)
5953 snprintf(name
, sizeof(name
), "ps_b[%u]", i
);
5954 ps
->uniform_b_locations
[i
] = GL_EXTCALL(glGetUniformLocation(program_id
, name
));
5957 for (i
= 0; i
< MAX_TEXTURES
; ++i
)
5959 snprintf(name
, sizeof(name
), "bumpenv_mat%u", i
);
5960 ps
->bumpenv_mat_location
[i
] = GL_EXTCALL(glGetUniformLocation(program_id
, name
));
5961 snprintf(name
, sizeof(name
), "bumpenv_lum_scale%u", i
);
5962 ps
->bumpenv_lum_scale_location
[i
] = GL_EXTCALL(glGetUniformLocation(program_id
, name
));
5963 snprintf(name
, sizeof(name
), "bumpenv_lum_offset%u", i
);
5964 ps
->bumpenv_lum_offset_location
[i
] = GL_EXTCALL(glGetUniformLocation(program_id
, name
));
5965 snprintf(name
, sizeof(name
), "tss_const%u", i
);
5966 ps
->tss_constant_location
[i
] = GL_EXTCALL(glGetUniformLocation(program_id
, name
));
5969 ps
->tex_factor_location
= GL_EXTCALL(glGetUniformLocation(program_id
, "tex_factor"));
5970 ps
->specular_enable_location
= GL_EXTCALL(glGetUniformLocation(program_id
, "specular_enable"));
5971 ps
->np2_fixup_location
= GL_EXTCALL(glGetUniformLocation(program_id
, "ps_samplerNP2Fixup"));
5972 ps
->ycorrection_location
= GL_EXTCALL(glGetUniformLocation(program_id
, "ycorrection"));
5973 ps
->color_key_location
= GL_EXTCALL(glGetUniformLocation(program_id
, "color_key"));
5976 static void shader_glsl_init_uniform_block_bindings(const struct wined3d_gl_info
*gl_info
, GLuint program_id
,
5977 const struct wined3d_shader_reg_maps
*reg_maps
, unsigned int base
, unsigned int count
)
5979 const char *prefix
= shader_glsl_get_prefix(reg_maps
->shader_version
.type
);
5984 for (i
= 0; i
< count
; ++i
)
5986 if (!reg_maps
->cb_sizes
[i
])
5989 snprintf(name
, sizeof(name
), "block_%s_cb%u", prefix
, i
);
5990 block_idx
= GL_EXTCALL(glGetUniformBlockIndex(program_id
, name
));
5991 GL_EXTCALL(glUniformBlockBinding(program_id
, block_idx
, base
+ i
));
5993 checkGLcall("glUniformBlockBinding");
5996 /* Context activation is done by the caller. */
5997 static void set_glsl_shader_program(const struct wined3d_context
*context
, const struct wined3d_state
*state
,
5998 struct shader_glsl_priv
*priv
, struct glsl_context_data
*ctx_data
)
6000 const struct wined3d_gl_info
*gl_info
= context
->gl_info
;
6001 const struct ps_np2fixup_info
*np2fixup_info
= NULL
;
6002 struct glsl_shader_prog_link
*entry
= NULL
;
6003 struct wined3d_shader
*vshader
= NULL
;
6004 struct wined3d_shader
*gshader
= NULL
;
6005 struct wined3d_shader
*pshader
= NULL
;
6006 GLuint program_id
= 0;
6007 GLuint reorder_shader_id
= 0;
6012 struct list
*ps_list
, *vs_list
;
6014 if (!(context
->shader_update_mask
& (1 << WINED3D_SHADER_TYPE_VERTEX
)))
6016 vs_id
= ctx_data
->glsl_program
->vs
.id
;
6017 vs_list
= &ctx_data
->glsl_program
->vs
.shader_entry
;
6021 vshader
= state
->shader
[WINED3D_SHADER_TYPE_VERTEX
];
6022 gshader
= state
->shader
[WINED3D_SHADER_TYPE_GEOMETRY
];
6024 if (!(context
->shader_update_mask
& (1 << WINED3D_SHADER_TYPE_GEOMETRY
))
6025 && ctx_data
->glsl_program
->gs
.id
)
6026 gs_id
= ctx_data
->glsl_program
->gs
.id
;
6028 gs_id
= find_glsl_geometry_shader(context
, &priv
->shader_buffer
, gshader
);
6031 else if (use_vs(state
))
6033 struct vs_compile_args vs_compile_args
;
6034 vshader
= state
->shader
[WINED3D_SHADER_TYPE_VERTEX
];
6036 find_vs_compile_args(state
, vshader
, context
->stream_info
.swizzle_map
, &vs_compile_args
);
6037 vs_id
= find_glsl_vshader(context
, &priv
->shader_buffer
, vshader
, &vs_compile_args
);
6038 vs_list
= &vshader
->linked_programs
;
6040 if ((gshader
= state
->shader
[WINED3D_SHADER_TYPE_GEOMETRY
]))
6041 gs_id
= find_glsl_geometry_shader(context
, &priv
->shader_buffer
, gshader
);
6043 else if (priv
->vertex_pipe
== &glsl_vertex_pipe
)
6045 struct glsl_ffp_vertex_shader
*ffp_shader
;
6046 struct wined3d_ffp_vs_settings settings
;
6048 wined3d_ffp_get_vs_settings(state
, &context
->stream_info
, &settings
);
6049 ffp_shader
= shader_glsl_find_ffp_vertex_shader(priv
, gl_info
, &settings
);
6050 vs_id
= ffp_shader
->id
;
6051 vs_list
= &ffp_shader
->linked_programs
;
6054 if (!(context
->shader_update_mask
& (1 << WINED3D_SHADER_TYPE_PIXEL
)))
6056 ps_id
= ctx_data
->glsl_program
->ps
.id
;
6057 ps_list
= &ctx_data
->glsl_program
->ps
.shader_entry
;
6060 pshader
= state
->shader
[WINED3D_SHADER_TYPE_PIXEL
];
6062 else if (use_ps(state
))
6064 struct ps_compile_args ps_compile_args
;
6065 pshader
= state
->shader
[WINED3D_SHADER_TYPE_PIXEL
];
6066 find_ps_compile_args(state
, pshader
, context
->stream_info
.position_transformed
, &ps_compile_args
, gl_info
);
6067 ps_id
= find_glsl_pshader(context
, &priv
->shader_buffer
,
6068 pshader
, &ps_compile_args
, &np2fixup_info
);
6069 ps_list
= &pshader
->linked_programs
;
6071 else if (priv
->fragment_pipe
== &glsl_fragment_pipe
)
6073 struct glsl_ffp_fragment_shader
*ffp_shader
;
6074 struct ffp_frag_settings settings
;
6076 gen_ffp_frag_op(context
, state
, &settings
, FALSE
);
6077 ffp_shader
= shader_glsl_find_ffp_fragment_shader(priv
, gl_info
, &settings
);
6078 ps_id
= ffp_shader
->id
;
6079 ps_list
= &ffp_shader
->linked_programs
;
6082 if ((!vs_id
&& !gs_id
&& !ps_id
) || (entry
= get_glsl_program_entry(priv
, vs_id
, gs_id
, ps_id
)))
6084 ctx_data
->glsl_program
= entry
;
6088 /* If we get to this point, then no matching program exists, so we create one */
6089 program_id
= GL_EXTCALL(glCreateProgram());
6090 TRACE("Created new GLSL shader program %u.\n", program_id
);
6092 /* Create the entry */
6093 entry
= HeapAlloc(GetProcessHeap(), 0, sizeof(struct glsl_shader_prog_link
));
6094 entry
->id
= program_id
;
6095 entry
->vs
.id
= vs_id
;
6096 entry
->gs
.id
= gs_id
;
6097 entry
->ps
.id
= ps_id
;
6098 entry
->constant_version
= 0;
6099 entry
->ps
.np2_fixup_info
= np2fixup_info
;
6100 /* Add the hash table entry */
6101 add_glsl_program_entry(priv
, entry
);
6103 /* Set the current program */
6104 ctx_data
->glsl_program
= entry
;
6106 /* Attach GLSL vshader */
6109 TRACE("Attaching GLSL shader object %u to program %u.\n", vs_id
, program_id
);
6110 GL_EXTCALL(glAttachShader(program_id
, vs_id
));
6111 checkGLcall("glAttachShader");
6113 list_add_head(vs_list
, &entry
->vs
.shader_entry
);
6118 WORD map
= vshader
->reg_maps
.input_registers
;
6121 reorder_shader_id
= generate_param_reorder_function(&priv
->shader_buffer
, vshader
, pshader
, gl_info
);
6122 TRACE("Attaching GLSL shader object %u to program %u.\n", reorder_shader_id
, program_id
);
6123 GL_EXTCALL(glAttachShader(program_id
, reorder_shader_id
));
6124 checkGLcall("glAttachShader");
6125 /* Flag the reorder function for deletion, then it will be freed automatically when the program
6128 GL_EXTCALL(glDeleteShader(reorder_shader_id
));
6130 /* Bind vertex attributes to a corresponding index number to match
6131 * the same index numbers as ARB_vertex_programs (makes loading
6132 * vertex attributes simpler). With this method, we can use the
6133 * exact same code to load the attributes later for both ARB and
6136 * We have to do this here because we need to know the Program ID
6137 * in order to make the bindings work, and it has to be done prior
6138 * to linking the GLSL program. */
6139 for (i
= 0; map
; map
>>= 1, ++i
)
6141 if (!(map
& 1)) continue;
6143 snprintf(tmp_name
, sizeof(tmp_name
), "vs_in%u", i
);
6144 GL_EXTCALL(glBindAttribLocation(program_id
, i
, tmp_name
));
6146 checkGLcall("glBindAttribLocation");
6151 TRACE("Attaching GLSL geometry shader object %u to program %u.\n", gs_id
, program_id
);
6152 GL_EXTCALL(glAttachShader(program_id
, gs_id
));
6153 checkGLcall("glAttachShader");
6155 TRACE("input type %s, output type %s, vertices out %u.\n",
6156 debug_d3dprimitivetype(gshader
->u
.gs
.input_type
),
6157 debug_d3dprimitivetype(gshader
->u
.gs
.output_type
),
6158 gshader
->u
.gs
.vertices_out
);
6159 GL_EXTCALL(glProgramParameteriARB(program_id
, GL_GEOMETRY_INPUT_TYPE_ARB
,
6160 gl_primitive_type_from_d3d(gshader
->u
.gs
.input_type
)));
6161 GL_EXTCALL(glProgramParameteriARB(program_id
, GL_GEOMETRY_OUTPUT_TYPE_ARB
,
6162 gl_primitive_type_from_d3d(gshader
->u
.gs
.output_type
)));
6163 GL_EXTCALL(glProgramParameteriARB(program_id
, GL_GEOMETRY_VERTICES_OUT_ARB
,
6164 gshader
->u
.gs
.vertices_out
));
6165 checkGLcall("glProgramParameteriARB");
6167 list_add_head(&gshader
->linked_programs
, &entry
->gs
.shader_entry
);
6170 /* Attach GLSL pshader */
6173 TRACE("Attaching GLSL shader object %u to program %u.\n", ps_id
, program_id
);
6174 GL_EXTCALL(glAttachShader(program_id
, ps_id
));
6175 checkGLcall("glAttachShader");
6177 list_add_head(ps_list
, &entry
->ps
.shader_entry
);
6180 /* Link the program */
6181 TRACE("Linking GLSL shader program %u.\n", program_id
);
6182 GL_EXTCALL(glLinkProgram(program_id
));
6183 shader_glsl_validate_link(gl_info
, program_id
);
6185 shader_glsl_init_vs_uniform_locations(gl_info
, program_id
, &entry
->vs
,
6186 vshader
? min(vshader
->limits
->constant_float
, gl_info
->limits
.glsl_vs_float_constants
) : 0);
6187 shader_glsl_init_ps_uniform_locations(gl_info
, program_id
, &entry
->ps
,
6188 pshader
? min(pshader
->limits
->constant_float
, gl_info
->limits
.glsl_ps_float_constants
) : 0);
6189 checkGLcall("Find glsl program uniform locations");
6191 if (pshader
&& pshader
->reg_maps
.shader_version
.major
>= 3
6192 && pshader
->u
.ps
.declared_in_count
> vec4_varyings(3, gl_info
))
6194 TRACE("Shader %d needs vertex color clamping disabled.\n", program_id
);
6195 entry
->vs
.vertex_color_clamp
= GL_FALSE
;
6199 entry
->vs
.vertex_color_clamp
= GL_FIXED_ONLY_ARB
;
6202 /* Set the shader to allow uniform loading on it */
6203 GL_EXTCALL(glUseProgram(program_id
));
6204 checkGLcall("glUseProgram");
6206 /* Load the vertex and pixel samplers now. The function that finds the mappings makes sure
6207 * that it stays the same for each vertexshader-pixelshader pair(=linked glsl program). If
6208 * a pshader with fixed function pipeline is used there are no vertex samplers, and if a
6209 * vertex shader with fixed function pixel processing is used we make sure that the card
6210 * supports enough samplers to allow the max number of vertex samplers with all possible
6211 * fixed function fragment processing setups. So once the program is linked these samplers
6213 shader_glsl_load_samplers(gl_info
, context
->tex_unit_map
, program_id
);
6215 entry
->constant_update_mask
= 0;
6218 entry
->constant_update_mask
|= WINED3D_SHADER_CONST_VS_F
;
6219 if (vshader
->reg_maps
.integer_constants
)
6220 entry
->constant_update_mask
|= WINED3D_SHADER_CONST_VS_I
;
6221 if (vshader
->reg_maps
.boolean_constants
)
6222 entry
->constant_update_mask
|= WINED3D_SHADER_CONST_VS_B
;
6223 entry
->constant_update_mask
|= WINED3D_SHADER_CONST_VS_POS_FIXUP
;
6225 shader_glsl_init_uniform_block_bindings(gl_info
, program_id
, &vshader
->reg_maps
,
6226 0, gl_info
->limits
.vertex_uniform_blocks
);
6230 entry
->constant_update_mask
|= WINED3D_SHADER_CONST_FFP_MODELVIEW
6231 | WINED3D_SHADER_CONST_FFP_PROJ
;
6235 shader_glsl_init_uniform_block_bindings(gl_info
, program_id
, &gshader
->reg_maps
,
6236 gl_info
->limits
.vertex_uniform_blocks
, gl_info
->limits
.geometry_uniform_blocks
);
6242 entry
->constant_update_mask
|= WINED3D_SHADER_CONST_PS_F
;
6243 if (pshader
->reg_maps
.integer_constants
)
6244 entry
->constant_update_mask
|= WINED3D_SHADER_CONST_PS_I
;
6245 if (pshader
->reg_maps
.boolean_constants
)
6246 entry
->constant_update_mask
|= WINED3D_SHADER_CONST_PS_B
;
6247 if (entry
->ps
.ycorrection_location
!= -1)
6248 entry
->constant_update_mask
|= WINED3D_SHADER_CONST_PS_Y_CORR
;
6250 shader_glsl_init_uniform_block_bindings(gl_info
, program_id
, &pshader
->reg_maps
,
6251 gl_info
->limits
.vertex_uniform_blocks
+ gl_info
->limits
.geometry_uniform_blocks
,
6252 gl_info
->limits
.fragment_uniform_blocks
);
6256 entry
->constant_update_mask
|= WINED3D_SHADER_CONST_FFP_PS
;
6259 for (i
= 0; i
< MAX_TEXTURES
; ++i
)
6261 if (entry
->ps
.bumpenv_mat_location
[i
] != -1)
6263 entry
->constant_update_mask
|= WINED3D_SHADER_CONST_PS_BUMP_ENV
;
6268 if (entry
->ps
.np2_fixup_location
!= -1)
6269 entry
->constant_update_mask
|= WINED3D_SHADER_CONST_PS_NP2_FIXUP
;
6270 if (entry
->ps
.color_key_location
!= -1)
6271 entry
->constant_update_mask
|= WINED3D_SHADER_CONST_FFP_COLOR_KEY
;
6275 /* Context activation is done by the caller. */
6276 static GLuint
create_glsl_blt_shader(const struct wined3d_gl_info
*gl_info
, enum wined3d_gl_resource_type tex_type
,
6280 GLuint vshader_id
, pshader_id
;
6281 const char *blt_pshader
;
6283 static const char blt_vshader
[] =
6287 " gl_Position = gl_Vertex;\n"
6288 " gl_FrontColor = vec4(1.0);\n"
6289 " gl_TexCoord[0] = gl_MultiTexCoord0;\n"
6292 static const char * const blt_pshaders_full
[WINED3D_GL_RES_TYPE_COUNT
] =
6294 /* WINED3D_GL_RES_TYPE_TEX_1D */
6296 /* WINED3D_GL_RES_TYPE_TEX_2D */
6298 "uniform sampler2D sampler;\n"
6301 " gl_FragDepth = texture2D(sampler, gl_TexCoord[0].xy).x;\n"
6303 /* WINED3D_GL_RES_TYPE_TEX_3D */
6305 /* WINED3D_GL_RES_TYPE_TEX_CUBE */
6307 "uniform samplerCube sampler;\n"
6310 " gl_FragDepth = textureCube(sampler, gl_TexCoord[0].xyz).x;\n"
6312 /* WINED3D_GL_RES_TYPE_TEX_RECT */
6314 "#extension GL_ARB_texture_rectangle : enable\n"
6315 "uniform sampler2DRect sampler;\n"
6318 " gl_FragDepth = texture2DRect(sampler, gl_TexCoord[0].xy).x;\n"
6322 static const char * const blt_pshaders_masked
[WINED3D_GL_RES_TYPE_COUNT
] =
6324 /* WINED3D_GL_RES_TYPE_TEX_1D */
6326 /* WINED3D_GL_RES_TYPE_TEX_2D */
6328 "uniform sampler2D sampler;\n"
6329 "uniform vec4 mask;\n"
6332 " if (all(lessThan(gl_FragCoord.xy, mask.zw))) discard;\n"
6333 " gl_FragDepth = texture2D(sampler, gl_TexCoord[0].xy).x;\n"
6335 /* WINED3D_GL_RES_TYPE_TEX_3D */
6337 /* WINED3D_GL_RES_TYPE_TEX_CUBE */
6339 "uniform samplerCube sampler;\n"
6340 "uniform vec4 mask;\n"
6343 " if (all(lessThan(gl_FragCoord.xy, mask.zw))) discard;\n"
6344 " gl_FragDepth = textureCube(sampler, gl_TexCoord[0].xyz).x;\n"
6346 /* WINED3D_GL_RES_TYPE_TEX_RECT */
6348 "#extension GL_ARB_texture_rectangle : enable\n"
6349 "uniform sampler2DRect sampler;\n"
6350 "uniform vec4 mask;\n"
6353 " if (all(lessThan(gl_FragCoord.xy, mask.zw))) discard;\n"
6354 " gl_FragDepth = texture2DRect(sampler, gl_TexCoord[0].xy).x;\n"
6358 blt_pshader
= masked
? blt_pshaders_masked
[tex_type
] : blt_pshaders_full
[tex_type
];
6361 FIXME("tex_type %#x not supported\n", tex_type
);
6365 vshader_id
= GL_EXTCALL(glCreateShader(GL_VERTEX_SHADER
));
6366 shader_glsl_compile(gl_info
, vshader_id
, blt_vshader
);
6368 pshader_id
= GL_EXTCALL(glCreateShader(GL_FRAGMENT_SHADER
));
6369 shader_glsl_compile(gl_info
, pshader_id
, blt_pshader
);
6371 program_id
= GL_EXTCALL(glCreateProgram());
6372 GL_EXTCALL(glAttachShader(program_id
, vshader_id
));
6373 GL_EXTCALL(glAttachShader(program_id
, pshader_id
));
6374 GL_EXTCALL(glLinkProgram(program_id
));
6376 shader_glsl_validate_link(gl_info
, program_id
);
6378 /* Once linked we can mark the shaders for deletion. They will be deleted once the program
6381 GL_EXTCALL(glDeleteShader(vshader_id
));
6382 GL_EXTCALL(glDeleteShader(pshader_id
));
6386 /* Context activation is done by the caller. */
6387 static void shader_glsl_select(void *shader_priv
, struct wined3d_context
*context
,
6388 const struct wined3d_state
*state
)
6390 struct glsl_context_data
*ctx_data
= context
->shader_backend_data
;
6391 const struct wined3d_gl_info
*gl_info
= context
->gl_info
;
6392 struct shader_glsl_priv
*priv
= shader_priv
;
6393 GLuint program_id
= 0, prev_id
= 0;
6394 GLenum old_vertex_color_clamp
, current_vertex_color_clamp
;
6396 priv
->vertex_pipe
->vp_enable(gl_info
, !use_vs(state
));
6397 priv
->fragment_pipe
->enable_extension(gl_info
, !use_ps(state
));
6399 if (ctx_data
->glsl_program
)
6401 prev_id
= ctx_data
->glsl_program
->id
;
6402 old_vertex_color_clamp
= ctx_data
->glsl_program
->vs
.vertex_color_clamp
;
6407 old_vertex_color_clamp
= GL_FIXED_ONLY_ARB
;
6410 set_glsl_shader_program(context
, state
, priv
, ctx_data
);
6412 if (ctx_data
->glsl_program
)
6414 program_id
= ctx_data
->glsl_program
->id
;
6415 current_vertex_color_clamp
= ctx_data
->glsl_program
->vs
.vertex_color_clamp
;
6420 current_vertex_color_clamp
= GL_FIXED_ONLY_ARB
;
6423 if (old_vertex_color_clamp
!= current_vertex_color_clamp
)
6425 if (gl_info
->supported
[ARB_COLOR_BUFFER_FLOAT
])
6427 GL_EXTCALL(glClampColorARB(GL_CLAMP_VERTEX_COLOR_ARB
, current_vertex_color_clamp
));
6428 checkGLcall("glClampColorARB");
6432 FIXME("vertex color clamp needs to be changed, but extension not supported.\n");
6436 TRACE("Using GLSL program %u.\n", program_id
);
6438 if (prev_id
!= program_id
)
6440 GL_EXTCALL(glUseProgram(program_id
));
6441 checkGLcall("glUseProgram");
6444 context
->constant_update_mask
|= ctx_data
->glsl_program
->constant_update_mask
;
6448 /* "context" is not necessarily the currently active context. */
6449 static void shader_glsl_invalidate_current_program(struct wined3d_context
*context
)
6451 struct glsl_context_data
*ctx_data
= context
->shader_backend_data
;
6453 ctx_data
->glsl_program
= NULL
;
6454 context
->shader_update_mask
= (1 << WINED3D_SHADER_TYPE_PIXEL
)
6455 | (1 << WINED3D_SHADER_TYPE_VERTEX
)
6456 | (1 << WINED3D_SHADER_TYPE_GEOMETRY
);
6459 /* Context activation is done by the caller. */
6460 static void shader_glsl_disable(void *shader_priv
, struct wined3d_context
*context
)
6462 const struct wined3d_gl_info
*gl_info
= context
->gl_info
;
6463 struct shader_glsl_priv
*priv
= shader_priv
;
6465 shader_glsl_invalidate_current_program(context
);
6466 GL_EXTCALL(glUseProgram(0));
6467 checkGLcall("glUseProgram");
6469 priv
->vertex_pipe
->vp_enable(gl_info
, FALSE
);
6470 priv
->fragment_pipe
->enable_extension(gl_info
, FALSE
);
6472 if (gl_info
->supported
[ARB_COLOR_BUFFER_FLOAT
])
6474 GL_EXTCALL(glClampColorARB(GL_CLAMP_VERTEX_COLOR_ARB
, GL_FIXED_ONLY_ARB
));
6475 checkGLcall("glClampColorARB");
6479 /* Context activation is done by the caller. */
6480 static void shader_glsl_select_depth_blt(void *shader_priv
, const struct wined3d_gl_info
*gl_info
,
6481 enum wined3d_gl_resource_type tex_type
, const SIZE
*ds_mask_size
)
6483 BOOL masked
= ds_mask_size
->cx
&& ds_mask_size
->cy
;
6484 struct shader_glsl_priv
*priv
= shader_priv
;
6485 GLuint
*blt_program
;
6488 blt_program
= masked
? &priv
->depth_blt_program_masked
[tex_type
] : &priv
->depth_blt_program_full
[tex_type
];
6491 *blt_program
= create_glsl_blt_shader(gl_info
, tex_type
, masked
);
6492 loc
= GL_EXTCALL(glGetUniformLocation(*blt_program
, "sampler"));
6493 GL_EXTCALL(glUseProgram(*blt_program
));
6494 GL_EXTCALL(glUniform1i(loc
, 0));
6498 GL_EXTCALL(glUseProgram(*blt_program
));
6503 loc
= GL_EXTCALL(glGetUniformLocation(*blt_program
, "mask"));
6504 GL_EXTCALL(glUniform4f(loc
, 0.0f
, 0.0f
, (float)ds_mask_size
->cx
, (float)ds_mask_size
->cy
));
6508 /* Context activation is done by the caller. */
6509 static void shader_glsl_deselect_depth_blt(void *shader_priv
, const struct wined3d_gl_info
*gl_info
)
6511 const struct glsl_context_data
*ctx_data
= context_get_current()->shader_backend_data
;
6514 program_id
= ctx_data
->glsl_program
? ctx_data
->glsl_program
->id
: 0;
6515 if (program_id
) TRACE("Using GLSL program %u\n", program_id
);
6517 GL_EXTCALL(glUseProgram(program_id
));
6518 checkGLcall("glUseProgram");
6521 static void shader_glsl_invalidate_contexts_program(struct wined3d_device
*device
,
6522 const struct glsl_shader_prog_link
*program
)
6524 const struct glsl_context_data
*ctx_data
;
6525 struct wined3d_context
*context
;
6528 for (i
= 0; i
< device
->context_count
; ++i
)
6530 context
= device
->contexts
[i
];
6531 ctx_data
= context
->shader_backend_data
;
6533 if (ctx_data
->glsl_program
== program
)
6534 shader_glsl_invalidate_current_program(context
);
6538 static void shader_glsl_destroy(struct wined3d_shader
*shader
)
6540 struct glsl_shader_private
*shader_data
= shader
->backend_data
;
6541 struct wined3d_device
*device
= shader
->device
;
6542 struct shader_glsl_priv
*priv
= device
->shader_priv
;
6543 const struct wined3d_gl_info
*gl_info
;
6544 const struct list
*linked_programs
;
6545 struct wined3d_context
*context
;
6547 if (!shader_data
|| !shader_data
->num_gl_shaders
)
6549 HeapFree(GetProcessHeap(), 0, shader_data
);
6550 shader
->backend_data
= NULL
;
6554 context
= context_acquire(device
, NULL
);
6555 gl_info
= context
->gl_info
;
6557 TRACE("Deleting linked programs.\n");
6558 linked_programs
= &shader
->linked_programs
;
6559 if (linked_programs
->next
)
6561 struct glsl_shader_prog_link
*entry
, *entry2
;
6564 switch (shader
->reg_maps
.shader_version
.type
)
6566 case WINED3D_SHADER_TYPE_PIXEL
:
6568 struct glsl_ps_compiled_shader
*gl_shaders
= shader_data
->gl_shaders
.ps
;
6570 for (i
= 0; i
< shader_data
->num_gl_shaders
; ++i
)
6572 TRACE("Deleting pixel shader %u.\n", gl_shaders
[i
].id
);
6573 GL_EXTCALL(glDeleteShader(gl_shaders
[i
].id
));
6574 checkGLcall("glDeleteShader");
6576 HeapFree(GetProcessHeap(), 0, shader_data
->gl_shaders
.ps
);
6578 LIST_FOR_EACH_ENTRY_SAFE(entry
, entry2
, linked_programs
,
6579 struct glsl_shader_prog_link
, ps
.shader_entry
)
6581 shader_glsl_invalidate_contexts_program(device
, entry
);
6582 delete_glsl_program_entry(priv
, gl_info
, entry
);
6588 case WINED3D_SHADER_TYPE_VERTEX
:
6590 struct glsl_vs_compiled_shader
*gl_shaders
= shader_data
->gl_shaders
.vs
;
6592 for (i
= 0; i
< shader_data
->num_gl_shaders
; ++i
)
6594 TRACE("Deleting vertex shader %u.\n", gl_shaders
[i
].id
);
6595 GL_EXTCALL(glDeleteShader(gl_shaders
[i
].id
));
6596 checkGLcall("glDeleteShader");
6598 HeapFree(GetProcessHeap(), 0, shader_data
->gl_shaders
.vs
);
6600 LIST_FOR_EACH_ENTRY_SAFE(entry
, entry2
, linked_programs
,
6601 struct glsl_shader_prog_link
, vs
.shader_entry
)
6603 shader_glsl_invalidate_contexts_program(device
, entry
);
6604 delete_glsl_program_entry(priv
, gl_info
, entry
);
6610 case WINED3D_SHADER_TYPE_GEOMETRY
:
6612 struct glsl_gs_compiled_shader
*gl_shaders
= shader_data
->gl_shaders
.gs
;
6614 for (i
= 0; i
< shader_data
->num_gl_shaders
; ++i
)
6616 TRACE("Deleting geometry shader %u.\n", gl_shaders
[i
].id
);
6617 GL_EXTCALL(glDeleteShader(gl_shaders
[i
].id
));
6618 checkGLcall("glDeleteShader");
6620 HeapFree(GetProcessHeap(), 0, shader_data
->gl_shaders
.gs
);
6622 LIST_FOR_EACH_ENTRY_SAFE(entry
, entry2
, linked_programs
,
6623 struct glsl_shader_prog_link
, gs
.shader_entry
)
6625 shader_glsl_invalidate_contexts_program(device
, entry
);
6626 delete_glsl_program_entry(priv
, gl_info
, entry
);
6633 ERR("Unhandled shader type %#x.\n", shader
->reg_maps
.shader_version
.type
);
6638 HeapFree(GetProcessHeap(), 0, shader
->backend_data
);
6639 shader
->backend_data
= NULL
;
6641 context_release(context
);
6644 static int glsl_program_key_compare(const void *key
, const struct wine_rb_entry
*entry
)
6646 const struct glsl_program_key
*k
= key
;
6647 const struct glsl_shader_prog_link
*prog
= WINE_RB_ENTRY_VALUE(entry
,
6648 const struct glsl_shader_prog_link
, program_lookup_entry
);
6650 if (k
->vs_id
> prog
->vs
.id
) return 1;
6651 else if (k
->vs_id
< prog
->vs
.id
) return -1;
6653 if (k
->gs_id
> prog
->gs
.id
) return 1;
6654 else if (k
->gs_id
< prog
->gs
.id
) return -1;
6656 if (k
->ps_id
> prog
->ps
.id
) return 1;
6657 else if (k
->ps_id
< prog
->ps
.id
) return -1;
6662 static BOOL
constant_heap_init(struct constant_heap
*heap
, unsigned int constant_count
)
6664 SIZE_T size
= (constant_count
+ 1) * sizeof(*heap
->entries
)
6665 + constant_count
* sizeof(*heap
->contained
)
6666 + constant_count
* sizeof(*heap
->positions
);
6667 void *mem
= HeapAlloc(GetProcessHeap(), 0, size
);
6671 ERR("Failed to allocate memory\n");
6675 heap
->entries
= mem
;
6676 heap
->entries
[1].version
= 0;
6677 heap
->contained
= (BOOL
*)(heap
->entries
+ constant_count
+ 1);
6678 memset(heap
->contained
, 0, constant_count
* sizeof(*heap
->contained
));
6679 heap
->positions
= (unsigned int *)(heap
->contained
+ constant_count
);
6685 static void constant_heap_free(struct constant_heap
*heap
)
6687 HeapFree(GetProcessHeap(), 0, heap
->entries
);
6690 static const struct wine_rb_functions wined3d_glsl_program_rb_functions
=
6695 glsl_program_key_compare
,
6698 static HRESULT
shader_glsl_alloc(struct wined3d_device
*device
, const struct wined3d_vertex_pipe_ops
*vertex_pipe
,
6699 const struct fragment_pipeline
*fragment_pipe
)
6701 const struct wined3d_gl_info
*gl_info
= &device
->adapter
->gl_info
;
6702 struct shader_glsl_priv
*priv
= HeapAlloc(GetProcessHeap(), HEAP_ZERO_MEMORY
, sizeof(struct shader_glsl_priv
));
6703 SIZE_T stack_size
= wined3d_log2i(max(gl_info
->limits
.glsl_vs_float_constants
,
6704 gl_info
->limits
.glsl_ps_float_constants
)) + 1;
6705 struct fragment_caps fragment_caps
;
6706 void *vertex_priv
, *fragment_priv
;
6708 if (!(vertex_priv
= vertex_pipe
->vp_alloc(&glsl_shader_backend
, priv
)))
6710 ERR("Failed to initialize vertex pipe.\n");
6711 HeapFree(GetProcessHeap(), 0, priv
);
6715 if (!(fragment_priv
= fragment_pipe
->alloc_private(&glsl_shader_backend
, priv
)))
6717 ERR("Failed to initialize fragment pipe.\n");
6718 vertex_pipe
->vp_free(device
);
6719 HeapFree(GetProcessHeap(), 0, priv
);
6723 if (!shader_buffer_init(&priv
->shader_buffer
))
6725 ERR("Failed to initialize shader buffer.\n");
6729 priv
->stack
= HeapAlloc(GetProcessHeap(), 0, stack_size
* sizeof(*priv
->stack
));
6732 ERR("Failed to allocate memory.\n");
6736 if (!constant_heap_init(&priv
->vconst_heap
, gl_info
->limits
.glsl_vs_float_constants
))
6738 ERR("Failed to initialize vertex shader constant heap\n");
6742 if (!constant_heap_init(&priv
->pconst_heap
, gl_info
->limits
.glsl_ps_float_constants
))
6744 ERR("Failed to initialize pixel shader constant heap\n");
6748 if (wine_rb_init(&priv
->program_lookup
, &wined3d_glsl_program_rb_functions
) == -1)
6750 ERR("Failed to initialize rbtree.\n");
6754 priv
->next_constant_version
= 1;
6755 priv
->vertex_pipe
= vertex_pipe
;
6756 priv
->fragment_pipe
= fragment_pipe
;
6757 fragment_pipe
->get_caps(gl_info
, &fragment_caps
);
6758 priv
->ffp_proj_control
= fragment_caps
.wined3d_caps
& WINED3D_FRAGMENT_CAP_PROJ_CONTROL
;
6760 device
->vertex_priv
= vertex_priv
;
6761 device
->fragment_priv
= fragment_priv
;
6762 device
->shader_priv
= priv
;
6767 constant_heap_free(&priv
->pconst_heap
);
6768 constant_heap_free(&priv
->vconst_heap
);
6769 HeapFree(GetProcessHeap(), 0, priv
->stack
);
6770 shader_buffer_free(&priv
->shader_buffer
);
6771 fragment_pipe
->free_private(device
);
6772 vertex_pipe
->vp_free(device
);
6773 HeapFree(GetProcessHeap(), 0, priv
);
6774 return E_OUTOFMEMORY
;
6777 /* Context activation is done by the caller. */
6778 static void shader_glsl_free(struct wined3d_device
*device
)
6780 const struct wined3d_gl_info
*gl_info
= &device
->adapter
->gl_info
;
6781 struct shader_glsl_priv
*priv
= device
->shader_priv
;
6784 for (i
= 0; i
< WINED3D_GL_RES_TYPE_COUNT
; ++i
)
6786 if (priv
->depth_blt_program_full
[i
])
6788 GL_EXTCALL(glDeleteProgram(priv
->depth_blt_program_full
[i
]));
6790 if (priv
->depth_blt_program_masked
[i
])
6792 GL_EXTCALL(glDeleteProgram(priv
->depth_blt_program_masked
[i
]));
6796 wine_rb_destroy(&priv
->program_lookup
, NULL
, NULL
);
6797 constant_heap_free(&priv
->pconst_heap
);
6798 constant_heap_free(&priv
->vconst_heap
);
6799 HeapFree(GetProcessHeap(), 0, priv
->stack
);
6800 shader_buffer_free(&priv
->shader_buffer
);
6801 priv
->fragment_pipe
->free_private(device
);
6802 priv
->vertex_pipe
->vp_free(device
);
6804 HeapFree(GetProcessHeap(), 0, device
->shader_priv
);
6805 device
->shader_priv
= NULL
;
6808 static BOOL
shader_glsl_allocate_context_data(struct wined3d_context
*context
)
6810 return !!(context
->shader_backend_data
= HeapAlloc(GetProcessHeap(),
6811 HEAP_ZERO_MEMORY
, sizeof(struct glsl_context_data
)));
6814 static void shader_glsl_free_context_data(struct wined3d_context
*context
)
6816 HeapFree(GetProcessHeap(), 0, context
->shader_backend_data
);
6819 static void shader_glsl_get_caps(const struct wined3d_gl_info
*gl_info
, struct shader_caps
*caps
)
6823 if (gl_info
->supported
[EXT_GPU_SHADER4
] && gl_info
->supported
[ARB_SHADER_BIT_ENCODING
]
6824 && gl_info
->supported
[ARB_GEOMETRY_SHADER4
] && gl_info
->glsl_version
>= MAKEDWORD_VERSION(1, 50)
6825 && gl_info
->supported
[ARB_DRAW_ELEMENTS_BASE_VERTEX
] && gl_info
->supported
[ARB_DRAW_INSTANCED
]
6826 && gl_info
->supported
[ARB_TEXTURE_RG
] && gl_info
->supported
[ARB_SAMPLER_OBJECTS
])
6828 /* ARB_shader_texture_lod or EXT_gpu_shader4 is required for the SM3
6829 * texldd and texldl instructions. */
6830 else if (gl_info
->supported
[ARB_SHADER_TEXTURE_LOD
] || gl_info
->supported
[EXT_GPU_SHADER4
])
6834 TRACE("Shader model %u.\n", shader_model
);
6836 caps
->vs_version
= min(wined3d_settings
.max_sm_vs
, shader_model
);
6837 caps
->gs_version
= min(wined3d_settings
.max_sm_gs
, shader_model
);
6838 caps
->ps_version
= min(wined3d_settings
.max_sm_ps
, shader_model
);
6840 caps
->vs_uniform_count
= gl_info
->limits
.glsl_vs_float_constants
;
6841 caps
->ps_uniform_count
= gl_info
->limits
.glsl_ps_float_constants
;
6843 /* FIXME: The following line is card dependent. -8.0 to 8.0 is the
6844 * Direct3D minimum requirement.
6846 * Both GL_ARB_fragment_program and GLSL require a "maximum representable magnitude"
6847 * of colors to be 2^10, and 2^32 for other floats. Should we use 1024 here?
6849 * The problem is that the refrast clamps temporary results in the shader to
6850 * [-MaxValue;+MaxValue]. If the card's max value is bigger than the one we advertize here,
6851 * then applications may miss the clamping behavior. On the other hand, if it is smaller,
6852 * the shader will generate incorrect results too. Unfortunately, GL deliberately doesn't
6853 * offer a way to query this.
6855 if (shader_model
>= 4)
6856 caps
->ps_1x_max_value
= FLT_MAX
;
6858 caps
->ps_1x_max_value
= 1024.0f
;
6860 /* Ideally we'd only set caps like sRGB writes here if supported by both
6861 * the shader backend and the fragment pipe, but we can get called before
6862 * shader_glsl_alloc(). */
6863 caps
->wined3d_caps
= WINED3D_SHADER_CAP_VS_CLIPPING
6864 | WINED3D_SHADER_CAP_SRGB_WRITE
;
6867 static BOOL
shader_glsl_color_fixup_supported(struct color_fixup_desc fixup
)
6869 if (TRACE_ON(d3d_shader
) && TRACE_ON(d3d
))
6871 TRACE("Checking support for fixup:\n");
6872 dump_color_fixup_desc(fixup
);
6875 /* We support everything except YUV conversions. */
6876 if (!is_complex_fixup(fixup
))
6882 TRACE("[FAILED]\n");
6886 static const SHADER_HANDLER shader_glsl_instruction_handler_table
[WINED3DSIH_TABLE_SIZE
] =
6888 /* WINED3DSIH_ABS */ shader_glsl_map2gl
,
6889 /* WINED3DSIH_ADD */ shader_glsl_binop
,
6890 /* WINED3DSIH_AND */ shader_glsl_binop
,
6891 /* WINED3DSIH_BEM */ shader_glsl_bem
,
6892 /* WINED3DSIH_BREAK */ shader_glsl_break
,
6893 /* WINED3DSIH_BREAKC */ shader_glsl_breakc
,
6894 /* WINED3DSIH_BREAKP */ shader_glsl_breakp
,
6895 /* WINED3DSIH_CALL */ shader_glsl_call
,
6896 /* WINED3DSIH_CALLNZ */ shader_glsl_callnz
,
6897 /* WINED3DSIH_CMP */ shader_glsl_conditional_move
,
6898 /* WINED3DSIH_CND */ shader_glsl_cnd
,
6899 /* WINED3DSIH_CRS */ shader_glsl_cross
,
6900 /* WINED3DSIH_CUT */ shader_glsl_cut
,
6901 /* WINED3DSIH_DCL */ shader_glsl_nop
,
6902 /* WINED3DSIH_DCL_CONSTANT_BUFFER */ shader_glsl_nop
,
6903 /* WINED3DSIH_DCL_INPUT_PRIMITIVE */ shader_glsl_nop
,
6904 /* WINED3DSIH_DCL_OUTPUT_TOPOLOGY */ shader_glsl_nop
,
6905 /* WINED3DSIH_DCL_VERTICES_OUT */ shader_glsl_nop
,
6906 /* WINED3DSIH_DEF */ shader_glsl_nop
,
6907 /* WINED3DSIH_DEFB */ shader_glsl_nop
,
6908 /* WINED3DSIH_DEFI */ shader_glsl_nop
,
6909 /* WINED3DSIH_DIV */ shader_glsl_binop
,
6910 /* WINED3DSIH_DP2 */ shader_glsl_dot
,
6911 /* WINED3DSIH_DP2ADD */ shader_glsl_dp2add
,
6912 /* WINED3DSIH_DP3 */ shader_glsl_dot
,
6913 /* WINED3DSIH_DP4 */ shader_glsl_dot
,
6914 /* WINED3DSIH_DST */ shader_glsl_dst
,
6915 /* WINED3DSIH_DSX */ shader_glsl_map2gl
,
6916 /* WINED3DSIH_DSY */ shader_glsl_map2gl
,
6917 /* WINED3DSIH_ELSE */ shader_glsl_else
,
6918 /* WINED3DSIH_EMIT */ shader_glsl_emit
,
6919 /* WINED3DSIH_ENDIF */ shader_glsl_end
,
6920 /* WINED3DSIH_ENDLOOP */ shader_glsl_end
,
6921 /* WINED3DSIH_ENDREP */ shader_glsl_end
,
6922 /* WINED3DSIH_EQ */ shader_glsl_relop
,
6923 /* WINED3DSIH_EXP */ shader_glsl_scalar_op
,
6924 /* WINED3DSIH_EXPP */ shader_glsl_expp
,
6925 /* WINED3DSIH_FRC */ shader_glsl_map2gl
,
6926 /* WINED3DSIH_FTOI */ shader_glsl_to_int
,
6927 /* WINED3DSIH_GE */ shader_glsl_relop
,
6928 /* WINED3DSIH_IADD */ shader_glsl_binop
,
6929 /* WINED3DSIH_IEQ */ NULL
,
6930 /* WINED3DSIH_IF */ shader_glsl_if
,
6931 /* WINED3DSIH_IFC */ shader_glsl_ifc
,
6932 /* WINED3DSIH_IGE */ shader_glsl_relop
,
6933 /* WINED3DSIH_IMUL */ shader_glsl_imul
,
6934 /* WINED3DSIH_ISHL */ shader_glsl_binop
,
6935 /* WINED3DSIH_ITOF */ shader_glsl_to_float
,
6936 /* WINED3DSIH_LABEL */ shader_glsl_label
,
6937 /* WINED3DSIH_LD */ NULL
,
6938 /* WINED3DSIH_LIT */ shader_glsl_lit
,
6939 /* WINED3DSIH_LOG */ shader_glsl_scalar_op
,
6940 /* WINED3DSIH_LOGP */ shader_glsl_scalar_op
,
6941 /* WINED3DSIH_LOOP */ shader_glsl_loop
,
6942 /* WINED3DSIH_LRP */ shader_glsl_lrp
,
6943 /* WINED3DSIH_LT */ shader_glsl_relop
,
6944 /* WINED3DSIH_M3x2 */ shader_glsl_mnxn
,
6945 /* WINED3DSIH_M3x3 */ shader_glsl_mnxn
,
6946 /* WINED3DSIH_M3x4 */ shader_glsl_mnxn
,
6947 /* WINED3DSIH_M4x3 */ shader_glsl_mnxn
,
6948 /* WINED3DSIH_M4x4 */ shader_glsl_mnxn
,
6949 /* WINED3DSIH_MAD */ shader_glsl_mad
,
6950 /* WINED3DSIH_MAX */ shader_glsl_map2gl
,
6951 /* WINED3DSIH_MIN */ shader_glsl_map2gl
,
6952 /* WINED3DSIH_MOV */ shader_glsl_mov
,
6953 /* WINED3DSIH_MOVA */ shader_glsl_mov
,
6954 /* WINED3DSIH_MOVC */ shader_glsl_conditional_move
,
6955 /* WINED3DSIH_MUL */ shader_glsl_binop
,
6956 /* WINED3DSIH_NE */ shader_glsl_relop
,
6957 /* WINED3DSIH_NOP */ shader_glsl_nop
,
6958 /* WINED3DSIH_NRM */ shader_glsl_nrm
,
6959 /* WINED3DSIH_OR */ shader_glsl_binop
,
6960 /* WINED3DSIH_PHASE */ shader_glsl_nop
,
6961 /* WINED3DSIH_POW */ shader_glsl_pow
,
6962 /* WINED3DSIH_RCP */ shader_glsl_scalar_op
,
6963 /* WINED3DSIH_REP */ shader_glsl_rep
,
6964 /* WINED3DSIH_RET */ shader_glsl_ret
,
6965 /* WINED3DSIH_ROUND_NI */ shader_glsl_map2gl
,
6966 /* WINED3DSIH_RSQ */ shader_glsl_scalar_op
,
6967 /* WINED3DSIH_SAMPLE */ shader_glsl_sample
,
6968 /* WINED3DSIH_SAMPLE_GRAD */ NULL
,
6969 /* WINED3DSIH_SAMPLE_LOD */ NULL
,
6970 /* WINED3DSIH_SETP */ NULL
,
6971 /* WINED3DSIH_SGE */ shader_glsl_compare
,
6972 /* WINED3DSIH_SGN */ shader_glsl_sgn
,
6973 /* WINED3DSIH_SINCOS */ shader_glsl_sincos
,
6974 /* WINED3DSIH_SLT */ shader_glsl_compare
,
6975 /* WINED3DSIH_SQRT */ shader_glsl_map2gl
,
6976 /* WINED3DSIH_SUB */ shader_glsl_binop
,
6977 /* WINED3DSIH_TEX */ shader_glsl_tex
,
6978 /* WINED3DSIH_TEXBEM */ shader_glsl_texbem
,
6979 /* WINED3DSIH_TEXBEML */ shader_glsl_texbem
,
6980 /* WINED3DSIH_TEXCOORD */ shader_glsl_texcoord
,
6981 /* WINED3DSIH_TEXDEPTH */ shader_glsl_texdepth
,
6982 /* WINED3DSIH_TEXDP3 */ shader_glsl_texdp3
,
6983 /* WINED3DSIH_TEXDP3TEX */ shader_glsl_texdp3tex
,
6984 /* WINED3DSIH_TEXKILL */ shader_glsl_texkill
,
6985 /* WINED3DSIH_TEXLDD */ shader_glsl_texldd
,
6986 /* WINED3DSIH_TEXLDL */ shader_glsl_texldl
,
6987 /* WINED3DSIH_TEXM3x2DEPTH */ shader_glsl_texm3x2depth
,
6988 /* WINED3DSIH_TEXM3x2PAD */ shader_glsl_texm3x2pad
,
6989 /* WINED3DSIH_TEXM3x2TEX */ shader_glsl_texm3x2tex
,
6990 /* WINED3DSIH_TEXM3x3 */ shader_glsl_texm3x3
,
6991 /* WINED3DSIH_TEXM3x3DIFF */ NULL
,
6992 /* WINED3DSIH_TEXM3x3PAD */ shader_glsl_texm3x3pad
,
6993 /* WINED3DSIH_TEXM3x3SPEC */ shader_glsl_texm3x3spec
,
6994 /* WINED3DSIH_TEXM3x3TEX */ shader_glsl_texm3x3tex
,
6995 /* WINED3DSIH_TEXM3x3VSPEC */ shader_glsl_texm3x3vspec
,
6996 /* WINED3DSIH_TEXREG2AR */ shader_glsl_texreg2ar
,
6997 /* WINED3DSIH_TEXREG2GB */ shader_glsl_texreg2gb
,
6998 /* WINED3DSIH_TEXREG2RGB */ shader_glsl_texreg2rgb
,
6999 /* WINED3DSIH_UDIV */ shader_glsl_udiv
,
7000 /* WINED3DSIH_UGE */ shader_glsl_relop
,
7001 /* WINED3DSIH_USHR */ shader_glsl_binop
,
7002 /* WINED3DSIH_UTOF */ shader_glsl_to_float
,
7003 /* WINED3DSIH_XOR */ shader_glsl_binop
,
7006 static void shader_glsl_handle_instruction(const struct wined3d_shader_instruction
*ins
) {
7007 SHADER_HANDLER hw_fct
;
7009 /* Select handler */
7010 hw_fct
= shader_glsl_instruction_handler_table
[ins
->handler_idx
];
7012 /* Unhandled opcode */
7015 FIXME("Backend can't handle opcode %#x\n", ins
->handler_idx
);
7020 shader_glsl_add_instruction_modifiers(ins
);
7023 static BOOL
shader_glsl_has_ffp_proj_control(void *shader_priv
)
7025 struct shader_glsl_priv
*priv
= shader_priv
;
7027 return priv
->ffp_proj_control
;
7030 const struct wined3d_shader_backend_ops glsl_shader_backend
=
7032 shader_glsl_handle_instruction
,
7034 shader_glsl_disable
,
7035 shader_glsl_select_depth_blt
,
7036 shader_glsl_deselect_depth_blt
,
7037 shader_glsl_update_float_vertex_constants
,
7038 shader_glsl_update_float_pixel_constants
,
7039 shader_glsl_load_constants
,
7040 shader_glsl_destroy
,
7043 shader_glsl_allocate_context_data
,
7044 shader_glsl_free_context_data
,
7045 shader_glsl_get_caps
,
7046 shader_glsl_color_fixup_supported
,
7047 shader_glsl_has_ffp_proj_control
,
7050 static void glsl_vertex_pipe_vp_enable(const struct wined3d_gl_info
*gl_info
, BOOL enable
)
7053 gl_info
->gl_ops
.gl
.p_glEnable(GL_VERTEX_PROGRAM_POINT_SIZE_ARB
);
7055 gl_info
->gl_ops
.gl
.p_glDisable(GL_VERTEX_PROGRAM_POINT_SIZE_ARB
);
7056 checkGLcall("GL_VERTEX_PROGRAM_POINT_SIZE_ARB");
7059 static void glsl_vertex_pipe_vp_get_caps(const struct wined3d_gl_info
*gl_info
, struct wined3d_vertex_caps
*caps
)
7061 caps
->xyzrhw
= TRUE
;
7062 caps
->max_active_lights
= gl_info
->limits
.lights
;
7063 caps
->max_vertex_blend_matrices
= 1;
7064 caps
->max_vertex_blend_matrix_index
= 0;
7065 caps
->vertex_processing_caps
= WINED3DVTXPCAPS_TEXGEN
7066 | WINED3DVTXPCAPS_MATERIALSOURCE7
7067 | WINED3DVTXPCAPS_VERTEXFOG
7068 | WINED3DVTXPCAPS_DIRECTIONALLIGHTS
7069 | WINED3DVTXPCAPS_POSITIONALLIGHTS
7070 | WINED3DVTXPCAPS_LOCALVIEWER
7071 | WINED3DVTXPCAPS_TEXGEN_SPHEREMAP
;
7072 caps
->fvf_caps
= WINED3DFVFCAPS_PSIZE
| 8; /* 8 texture coordinates. */
7073 caps
->max_user_clip_planes
= gl_info
->limits
.clipplanes
;
7074 caps
->raster_caps
= WINED3DPRASTERCAPS_FOGRANGE
;
7077 static void *glsl_vertex_pipe_vp_alloc(const struct wined3d_shader_backend_ops
*shader_backend
, void *shader_priv
)
7079 struct shader_glsl_priv
*priv
;
7081 if (shader_backend
== &glsl_shader_backend
)
7085 if (wine_rb_init(&priv
->ffp_vertex_shaders
, &wined3d_ffp_vertex_program_rb_functions
) == -1)
7087 ERR("Failed to initialize rbtree.\n");
7094 FIXME("GLSL vertex pipe without GLSL shader backend not implemented.\n");
7099 static void shader_glsl_free_ffp_vertex_shader(struct wine_rb_entry
*entry
, void *context
)
7101 struct glsl_ffp_vertex_shader
*shader
= WINE_RB_ENTRY_VALUE(entry
,
7102 struct glsl_ffp_vertex_shader
, desc
.entry
);
7103 struct glsl_shader_prog_link
*program
, *program2
;
7104 struct glsl_ffp_destroy_ctx
*ctx
= context
;
7106 LIST_FOR_EACH_ENTRY_SAFE(program
, program2
, &shader
->linked_programs
,
7107 struct glsl_shader_prog_link
, vs
.shader_entry
)
7109 delete_glsl_program_entry(ctx
->priv
, ctx
->gl_info
, program
);
7111 ctx
->gl_info
->gl_ops
.ext
.p_glDeleteShader(shader
->id
);
7112 HeapFree(GetProcessHeap(), 0, shader
);
7115 /* Context activation is done by the caller. */
7116 static void glsl_vertex_pipe_vp_free(struct wined3d_device
*device
)
7118 struct shader_glsl_priv
*priv
= device
->vertex_priv
;
7119 struct glsl_ffp_destroy_ctx ctx
;
7122 ctx
.gl_info
= &device
->adapter
->gl_info
;
7123 wine_rb_destroy(&priv
->ffp_vertex_shaders
, shader_glsl_free_ffp_vertex_shader
, &ctx
);
7126 static void glsl_vertex_pipe_shader(struct wined3d_context
*context
,
7127 const struct wined3d_state
*state
, DWORD state_id
)
7129 context
->shader_update_mask
|= 1 << WINED3D_SHADER_TYPE_VERTEX
;
7132 static void glsl_vertex_pipe_vdecl(struct wined3d_context
*context
,
7133 const struct wined3d_state
*state
, DWORD state_id
)
7135 const struct wined3d_gl_info
*gl_info
= context
->gl_info
;
7136 BOOL transformed
= context
->stream_info
.position_transformed
;
7137 BOOL wasrhw
= context
->last_was_rhw
;
7140 context
->last_was_rhw
= transformed
;
7144 if (context
->last_was_vshader
)
7146 for (i
= 0; i
< gl_info
->limits
.clipplanes
; ++i
)
7147 clipplane(context
, state
, STATE_CLIPPLANE(i
));
7150 if (transformed
!= wasrhw
)
7151 context
->constant_update_mask
|= WINED3D_SHADER_CONST_FFP_PROJ
;
7153 for (i
= 0; i
< MAX_TEXTURES
; ++i
)
7155 if (!isStateDirty(context
, STATE_TRANSFORM(WINED3D_TS_TEXTURE0
+ i
)))
7156 transform_texture(context
, state
, STATE_TEXTURESTAGE(i
, WINED3D_TSS_TEXTURE_TRANSFORM_FLAGS
));
7159 if (!isStateDirty(context
, STATE_RENDER(WINED3D_RS_LIGHTING
)))
7160 state_lighting(context
, state
, STATE_RENDER(WINED3D_RS_LIGHTING
));
7161 if (!isStateDirty(context
, STATE_RENDER(WINED3D_RS_NORMALIZENORMALS
)))
7162 state_normalize(context
, state
, STATE_RENDER(WINED3D_RS_NORMALIZENORMALS
));
7164 /* Because of settings->texcoords, we have to always regenerate the
7165 * vertex shader on a vdecl change.
7166 * TODO: Just always output all the texcoords when there are enough
7167 * varyings available to drop the dependency. */
7168 context
->shader_update_mask
|= 1 << WINED3D_SHADER_TYPE_VERTEX
;
7171 && state
->shader
[WINED3D_SHADER_TYPE_PIXEL
]->reg_maps
.shader_version
.major
== 1
7172 && state
->shader
[WINED3D_SHADER_TYPE_PIXEL
]->reg_maps
.shader_version
.minor
<= 3)
7173 context
->shader_update_mask
|= 1 << WINED3D_SHADER_TYPE_PIXEL
;
7177 if (!context
->last_was_vshader
)
7179 /* Vertex shader clipping ignores the view matrix. Update all clipplanes. */
7180 for (i
= 0; i
< gl_info
->limits
.clipplanes
; ++i
)
7181 clipplane(context
, state
, STATE_CLIPPLANE(i
));
7185 context
->last_was_vshader
= use_vs(state
);
7188 static void glsl_vertex_pipe_vs(struct wined3d_context
*context
,
7189 const struct wined3d_state
*state
, DWORD state_id
)
7191 context
->shader_update_mask
|= 1 << WINED3D_SHADER_TYPE_VERTEX
;
7192 /* Different vertex shaders potentially require a different vertex attributes setup. */
7193 if (!isStateDirty(context
, STATE_VDECL
))
7194 context_apply_state(context
, state
, STATE_VDECL
);
7197 static void glsl_vertex_pipe_world(struct wined3d_context
*context
,
7198 const struct wined3d_state
*state
, DWORD state_id
)
7200 context
->constant_update_mask
|= WINED3D_SHADER_CONST_FFP_MODELVIEW
;
7203 void glsl_vertex_pipe_view(struct wined3d_context
*context
, const struct wined3d_state
*state
, DWORD state_id
)
7205 const struct wined3d_gl_info
*gl_info
= context
->gl_info
;
7206 const struct wined3d_light_info
*light
= NULL
;
7209 context
->constant_update_mask
|= WINED3D_SHADER_CONST_FFP_MODELVIEW
;
7211 /* Light settings are affected by the ModelView transform in OpenGL, the View transform in Direct3D. */
7212 gl_info
->gl_ops
.gl
.p_glMatrixMode(GL_MODELVIEW
);
7213 gl_info
->gl_ops
.gl
.p_glPushMatrix();
7214 gl_info
->gl_ops
.gl
.p_glLoadMatrixf(&state
->transforms
[WINED3D_TS_VIEW
]._11
);
7216 for (k
= 0; k
< gl_info
->limits
.lights
; ++k
)
7218 if (!(light
= state
->lights
[k
]))
7220 gl_info
->gl_ops
.gl
.p_glLightfv(GL_LIGHT0
+ light
->glIndex
, GL_POSITION
, light
->lightPosn
);
7221 checkGLcall("glLightfv posn");
7222 gl_info
->gl_ops
.gl
.p_glLightfv(GL_LIGHT0
+ light
->glIndex
, GL_SPOT_DIRECTION
, light
->lightDirn
);
7223 checkGLcall("glLightfv dirn");
7226 gl_info
->gl_ops
.gl
.p_glPopMatrix();
7228 for (k
= 0; k
< gl_info
->limits
.clipplanes
; ++k
)
7230 if (!isStateDirty(context
, STATE_CLIPPLANE(k
)))
7231 clipplane(context
, state
, STATE_CLIPPLANE(k
));
7234 if (context
->swapchain
->device
->vertexBlendUsed
)
7239 FIXME("Vertex blending emulation.\n");
7243 static void glsl_vertex_pipe_projection(struct wined3d_context
*context
,
7244 const struct wined3d_state
*state
, DWORD state_id
)
7246 /* Table fog behavior depends on the projection matrix. */
7247 if (state
->render_states
[WINED3D_RS_FOGENABLE
]
7248 && state
->render_states
[WINED3D_RS_FOGTABLEMODE
] != WINED3D_FOG_NONE
)
7249 context
->shader_update_mask
|= 1 << WINED3D_SHADER_TYPE_VERTEX
;
7250 context
->constant_update_mask
|= WINED3D_SHADER_CONST_FFP_PROJ
;
7253 static void glsl_vertex_pipe_viewport(struct wined3d_context
*context
,
7254 const struct wined3d_state
*state
, DWORD state_id
)
7256 if (!isStateDirty(context
, STATE_TRANSFORM(WINED3D_TS_PROJECTION
)))
7257 glsl_vertex_pipe_projection(context
, state
, STATE_TRANSFORM(WINED3D_TS_PROJECTION
));
7258 if (!isStateDirty(context
, STATE_RENDER(WINED3D_RS_POINTSCALEENABLE
))
7259 && state
->render_states
[WINED3D_RS_POINTSCALEENABLE
])
7260 state_pscale(context
, state
, STATE_RENDER(WINED3D_RS_POINTSCALEENABLE
));
7261 context
->constant_update_mask
|= WINED3D_SHADER_CONST_VS_POS_FIXUP
;
7264 static const struct StateEntryTemplate glsl_vertex_pipe_vp_states
[] =
7266 {STATE_VDECL
, {STATE_VDECL
, glsl_vertex_pipe_vdecl
}, WINED3D_GL_EXT_NONE
},
7267 {STATE_SHADER(WINED3D_SHADER_TYPE_VERTEX
), {STATE_SHADER(WINED3D_SHADER_TYPE_VERTEX
), glsl_vertex_pipe_vs
}, WINED3D_GL_EXT_NONE
},
7268 {STATE_MATERIAL
, {STATE_RENDER(WINED3D_RS_SPECULARENABLE
), NULL
}, WINED3D_GL_EXT_NONE
},
7269 {STATE_RENDER(WINED3D_RS_SPECULARENABLE
), {STATE_RENDER(WINED3D_RS_SPECULARENABLE
), state_specularenable
}, WINED3D_GL_EXT_NONE
},
7271 {STATE_CLIPPLANE(0), {STATE_CLIPPLANE(0), clipplane
}, WINED3D_GL_EXT_NONE
},
7272 {STATE_CLIPPLANE(1), {STATE_CLIPPLANE(1), clipplane
}, WINED3D_GL_EXT_NONE
},
7273 {STATE_CLIPPLANE(2), {STATE_CLIPPLANE(2), clipplane
}, WINED3D_GL_EXT_NONE
},
7274 {STATE_CLIPPLANE(3), {STATE_CLIPPLANE(3), clipplane
}, WINED3D_GL_EXT_NONE
},
7275 {STATE_CLIPPLANE(4), {STATE_CLIPPLANE(4), clipplane
}, WINED3D_GL_EXT_NONE
},
7276 {STATE_CLIPPLANE(5), {STATE_CLIPPLANE(5), clipplane
}, WINED3D_GL_EXT_NONE
},
7277 {STATE_CLIPPLANE(6), {STATE_CLIPPLANE(6), clipplane
}, WINED3D_GL_EXT_NONE
},
7278 {STATE_CLIPPLANE(7), {STATE_CLIPPLANE(7), clipplane
}, WINED3D_GL_EXT_NONE
},
7279 {STATE_CLIPPLANE(8), {STATE_CLIPPLANE(8), clipplane
}, WINED3D_GL_EXT_NONE
},
7280 {STATE_CLIPPLANE(9), {STATE_CLIPPLANE(9), clipplane
}, WINED3D_GL_EXT_NONE
},
7281 {STATE_CLIPPLANE(10), {STATE_CLIPPLANE(10), clipplane
}, WINED3D_GL_EXT_NONE
},
7282 {STATE_CLIPPLANE(11), {STATE_CLIPPLANE(11), clipplane
}, WINED3D_GL_EXT_NONE
},
7283 {STATE_CLIPPLANE(12), {STATE_CLIPPLANE(12), clipplane
}, WINED3D_GL_EXT_NONE
},
7284 {STATE_CLIPPLANE(13), {STATE_CLIPPLANE(13), clipplane
}, WINED3D_GL_EXT_NONE
},
7285 {STATE_CLIPPLANE(14), {STATE_CLIPPLANE(14), clipplane
}, WINED3D_GL_EXT_NONE
},
7286 {STATE_CLIPPLANE(15), {STATE_CLIPPLANE(15), clipplane
}, WINED3D_GL_EXT_NONE
},
7287 {STATE_CLIPPLANE(16), {STATE_CLIPPLANE(16), clipplane
}, WINED3D_GL_EXT_NONE
},
7288 {STATE_CLIPPLANE(17), {STATE_CLIPPLANE(17), clipplane
}, WINED3D_GL_EXT_NONE
},
7289 {STATE_CLIPPLANE(18), {STATE_CLIPPLANE(18), clipplane
}, WINED3D_GL_EXT_NONE
},
7290 {STATE_CLIPPLANE(19), {STATE_CLIPPLANE(19), clipplane
}, WINED3D_GL_EXT_NONE
},
7291 {STATE_CLIPPLANE(20), {STATE_CLIPPLANE(20), clipplane
}, WINED3D_GL_EXT_NONE
},
7292 {STATE_CLIPPLANE(21), {STATE_CLIPPLANE(21), clipplane
}, WINED3D_GL_EXT_NONE
},
7293 {STATE_CLIPPLANE(22), {STATE_CLIPPLANE(22), clipplane
}, WINED3D_GL_EXT_NONE
},
7294 {STATE_CLIPPLANE(23), {STATE_CLIPPLANE(23), clipplane
}, WINED3D_GL_EXT_NONE
},
7295 {STATE_CLIPPLANE(24), {STATE_CLIPPLANE(24), clipplane
}, WINED3D_GL_EXT_NONE
},
7296 {STATE_CLIPPLANE(25), {STATE_CLIPPLANE(25), clipplane
}, WINED3D_GL_EXT_NONE
},
7297 {STATE_CLIPPLANE(26), {STATE_CLIPPLANE(26), clipplane
}, WINED3D_GL_EXT_NONE
},
7298 {STATE_CLIPPLANE(27), {STATE_CLIPPLANE(27), clipplane
}, WINED3D_GL_EXT_NONE
},
7299 {STATE_CLIPPLANE(28), {STATE_CLIPPLANE(28), clipplane
}, WINED3D_GL_EXT_NONE
},
7300 {STATE_CLIPPLANE(29), {STATE_CLIPPLANE(29), clipplane
}, WINED3D_GL_EXT_NONE
},
7301 {STATE_CLIPPLANE(30), {STATE_CLIPPLANE(30), clipplane
}, WINED3D_GL_EXT_NONE
},
7302 {STATE_CLIPPLANE(31), {STATE_CLIPPLANE(31), clipplane
}, WINED3D_GL_EXT_NONE
},
7304 {STATE_LIGHT_TYPE
, {STATE_RENDER(WINED3D_RS_FOGENABLE
), NULL
}, WINED3D_GL_EXT_NONE
},
7305 {STATE_ACTIVELIGHT(0), {STATE_ACTIVELIGHT(0), light
}, WINED3D_GL_EXT_NONE
},
7306 {STATE_ACTIVELIGHT(1), {STATE_ACTIVELIGHT(1), light
}, WINED3D_GL_EXT_NONE
},
7307 {STATE_ACTIVELIGHT(2), {STATE_ACTIVELIGHT(2), light
}, WINED3D_GL_EXT_NONE
},
7308 {STATE_ACTIVELIGHT(3), {STATE_ACTIVELIGHT(3), light
}, WINED3D_GL_EXT_NONE
},
7309 {STATE_ACTIVELIGHT(4), {STATE_ACTIVELIGHT(4), light
}, WINED3D_GL_EXT_NONE
},
7310 {STATE_ACTIVELIGHT(5), {STATE_ACTIVELIGHT(5), light
}, WINED3D_GL_EXT_NONE
},
7311 {STATE_ACTIVELIGHT(6), {STATE_ACTIVELIGHT(6), light
}, WINED3D_GL_EXT_NONE
},
7312 {STATE_ACTIVELIGHT(7), {STATE_ACTIVELIGHT(7), light
}, WINED3D_GL_EXT_NONE
},
7314 {STATE_VIEWPORT
, {STATE_VIEWPORT
, glsl_vertex_pipe_viewport
}, WINED3D_GL_EXT_NONE
},
7315 /* Transform states */
7316 {STATE_TRANSFORM(WINED3D_TS_VIEW
), {STATE_TRANSFORM(WINED3D_TS_VIEW
), glsl_vertex_pipe_view
}, WINED3D_GL_EXT_NONE
},
7317 {STATE_TRANSFORM(WINED3D_TS_PROJECTION
), {STATE_TRANSFORM(WINED3D_TS_PROJECTION
), glsl_vertex_pipe_projection
}, WINED3D_GL_EXT_NONE
},
7318 {STATE_TRANSFORM(WINED3D_TS_TEXTURE0
), {STATE_TEXTURESTAGE(0, WINED3D_TSS_TEXTURE_TRANSFORM_FLAGS
), NULL
}, WINED3D_GL_EXT_NONE
},
7319 {STATE_TRANSFORM(WINED3D_TS_TEXTURE1
), {STATE_TEXTURESTAGE(1, WINED3D_TSS_TEXTURE_TRANSFORM_FLAGS
), NULL
}, WINED3D_GL_EXT_NONE
},
7320 {STATE_TRANSFORM(WINED3D_TS_TEXTURE2
), {STATE_TEXTURESTAGE(2, WINED3D_TSS_TEXTURE_TRANSFORM_FLAGS
), NULL
}, WINED3D_GL_EXT_NONE
},
7321 {STATE_TRANSFORM(WINED3D_TS_TEXTURE3
), {STATE_TEXTURESTAGE(3, WINED3D_TSS_TEXTURE_TRANSFORM_FLAGS
), NULL
}, WINED3D_GL_EXT_NONE
},
7322 {STATE_TRANSFORM(WINED3D_TS_TEXTURE4
), {STATE_TEXTURESTAGE(4, WINED3D_TSS_TEXTURE_TRANSFORM_FLAGS
), NULL
}, WINED3D_GL_EXT_NONE
},
7323 {STATE_TRANSFORM(WINED3D_TS_TEXTURE5
), {STATE_TEXTURESTAGE(5, WINED3D_TSS_TEXTURE_TRANSFORM_FLAGS
), NULL
}, WINED3D_GL_EXT_NONE
},
7324 {STATE_TRANSFORM(WINED3D_TS_TEXTURE6
), {STATE_TEXTURESTAGE(6, WINED3D_TSS_TEXTURE_TRANSFORM_FLAGS
), NULL
}, WINED3D_GL_EXT_NONE
},
7325 {STATE_TRANSFORM(WINED3D_TS_TEXTURE7
), {STATE_TEXTURESTAGE(7, WINED3D_TSS_TEXTURE_TRANSFORM_FLAGS
), NULL
}, WINED3D_GL_EXT_NONE
},
7326 {STATE_TRANSFORM(WINED3D_TS_WORLD_MATRIX(0)), {STATE_TRANSFORM(WINED3D_TS_WORLD_MATRIX(0)), glsl_vertex_pipe_world
}, WINED3D_GL_EXT_NONE
},
7327 {STATE_TEXTURESTAGE(0, WINED3D_TSS_TEXTURE_TRANSFORM_FLAGS
), {STATE_TEXTURESTAGE(0, WINED3D_TSS_TEXTURE_TRANSFORM_FLAGS
), transform_texture
}, WINED3D_GL_EXT_NONE
},
7328 {STATE_TEXTURESTAGE(1, WINED3D_TSS_TEXTURE_TRANSFORM_FLAGS
), {STATE_TEXTURESTAGE(1, WINED3D_TSS_TEXTURE_TRANSFORM_FLAGS
), transform_texture
}, WINED3D_GL_EXT_NONE
},
7329 {STATE_TEXTURESTAGE(2, WINED3D_TSS_TEXTURE_TRANSFORM_FLAGS
), {STATE_TEXTURESTAGE(2, WINED3D_TSS_TEXTURE_TRANSFORM_FLAGS
), transform_texture
}, WINED3D_GL_EXT_NONE
},
7330 {STATE_TEXTURESTAGE(3, WINED3D_TSS_TEXTURE_TRANSFORM_FLAGS
), {STATE_TEXTURESTAGE(3, WINED3D_TSS_TEXTURE_TRANSFORM_FLAGS
), transform_texture
}, WINED3D_GL_EXT_NONE
},
7331 {STATE_TEXTURESTAGE(4, WINED3D_TSS_TEXTURE_TRANSFORM_FLAGS
), {STATE_TEXTURESTAGE(4, WINED3D_TSS_TEXTURE_TRANSFORM_FLAGS
), transform_texture
}, WINED3D_GL_EXT_NONE
},
7332 {STATE_TEXTURESTAGE(5, WINED3D_TSS_TEXTURE_TRANSFORM_FLAGS
), {STATE_TEXTURESTAGE(5, WINED3D_TSS_TEXTURE_TRANSFORM_FLAGS
), transform_texture
}, WINED3D_GL_EXT_NONE
},
7333 {STATE_TEXTURESTAGE(6, WINED3D_TSS_TEXTURE_TRANSFORM_FLAGS
), {STATE_TEXTURESTAGE(6, WINED3D_TSS_TEXTURE_TRANSFORM_FLAGS
), transform_texture
}, WINED3D_GL_EXT_NONE
},
7334 {STATE_TEXTURESTAGE(7, WINED3D_TSS_TEXTURE_TRANSFORM_FLAGS
), {STATE_TEXTURESTAGE(7, WINED3D_TSS_TEXTURE_TRANSFORM_FLAGS
), transform_texture
}, WINED3D_GL_EXT_NONE
},
7335 {STATE_TEXTURESTAGE(0, WINED3D_TSS_TEXCOORD_INDEX
), {STATE_SHADER(WINED3D_SHADER_TYPE_VERTEX
), NULL
}, WINED3D_GL_EXT_NONE
},
7336 {STATE_TEXTURESTAGE(1, WINED3D_TSS_TEXCOORD_INDEX
), {STATE_SHADER(WINED3D_SHADER_TYPE_VERTEX
), NULL
}, WINED3D_GL_EXT_NONE
},
7337 {STATE_TEXTURESTAGE(2, WINED3D_TSS_TEXCOORD_INDEX
), {STATE_SHADER(WINED3D_SHADER_TYPE_VERTEX
), NULL
}, WINED3D_GL_EXT_NONE
},
7338 {STATE_TEXTURESTAGE(3, WINED3D_TSS_TEXCOORD_INDEX
), {STATE_SHADER(WINED3D_SHADER_TYPE_VERTEX
), NULL
}, WINED3D_GL_EXT_NONE
},
7339 {STATE_TEXTURESTAGE(4, WINED3D_TSS_TEXCOORD_INDEX
), {STATE_SHADER(WINED3D_SHADER_TYPE_VERTEX
), NULL
}, WINED3D_GL_EXT_NONE
},
7340 {STATE_TEXTURESTAGE(5, WINED3D_TSS_TEXCOORD_INDEX
), {STATE_SHADER(WINED3D_SHADER_TYPE_VERTEX
), NULL
}, WINED3D_GL_EXT_NONE
},
7341 {STATE_TEXTURESTAGE(6, WINED3D_TSS_TEXCOORD_INDEX
), {STATE_SHADER(WINED3D_SHADER_TYPE_VERTEX
), NULL
}, WINED3D_GL_EXT_NONE
},
7342 {STATE_TEXTURESTAGE(7, WINED3D_TSS_TEXCOORD_INDEX
), {STATE_SHADER(WINED3D_SHADER_TYPE_VERTEX
), NULL
}, WINED3D_GL_EXT_NONE
},
7344 {STATE_RENDER(WINED3D_RS_FOGENABLE
), {STATE_RENDER(WINED3D_RS_FOGENABLE
), glsl_vertex_pipe_shader
}, WINED3D_GL_EXT_NONE
},
7345 {STATE_RENDER(WINED3D_RS_FOGTABLEMODE
), {STATE_RENDER(WINED3D_RS_FOGENABLE
), NULL
}, WINED3D_GL_EXT_NONE
},
7346 {STATE_RENDER(WINED3D_RS_FOGVERTEXMODE
), {STATE_RENDER(WINED3D_RS_FOGENABLE
), NULL
}, WINED3D_GL_EXT_NONE
},
7347 {STATE_RENDER(WINED3D_RS_RANGEFOGENABLE
), {STATE_RENDER(WINED3D_RS_FOGENABLE
), NULL
}, WINED3D_GL_EXT_NONE
},
7348 {STATE_RENDER(WINED3D_RS_CLIPPING
), {STATE_RENDER(WINED3D_RS_CLIPPING
), state_clipping
}, WINED3D_GL_EXT_NONE
},
7349 {STATE_RENDER(WINED3D_RS_CLIPPLANEENABLE
), {STATE_RENDER(WINED3D_RS_CLIPPING
), NULL
}, WINED3D_GL_EXT_NONE
},
7350 {STATE_RENDER(WINED3D_RS_LIGHTING
), {STATE_SHADER(WINED3D_SHADER_TYPE_VERTEX
), NULL
}, WINED3D_GL_EXT_NONE
},
7351 {STATE_RENDER(WINED3D_RS_AMBIENT
), {STATE_RENDER(WINED3D_RS_AMBIENT
), state_ambient
}, WINED3D_GL_EXT_NONE
},
7352 {STATE_RENDER(WINED3D_RS_COLORVERTEX
), {STATE_RENDER(WINED3D_RS_COLORVERTEX
), glsl_vertex_pipe_shader
}, WINED3D_GL_EXT_NONE
},
7353 {STATE_RENDER(WINED3D_RS_LOCALVIEWER
), {STATE_SHADER(WINED3D_SHADER_TYPE_VERTEX
), NULL
}, WINED3D_GL_EXT_NONE
},
7354 {STATE_RENDER(WINED3D_RS_NORMALIZENORMALS
), {STATE_SHADER(WINED3D_SHADER_TYPE_VERTEX
), NULL
}, WINED3D_GL_EXT_NONE
},
7355 {STATE_RENDER(WINED3D_RS_DIFFUSEMATERIALSOURCE
), {STATE_SHADER(WINED3D_SHADER_TYPE_VERTEX
), NULL
}, WINED3D_GL_EXT_NONE
},
7356 {STATE_RENDER(WINED3D_RS_SPECULARMATERIALSOURCE
), {STATE_SHADER(WINED3D_SHADER_TYPE_VERTEX
), NULL
}, WINED3D_GL_EXT_NONE
},
7357 {STATE_RENDER(WINED3D_RS_AMBIENTMATERIALSOURCE
), {STATE_SHADER(WINED3D_SHADER_TYPE_VERTEX
), NULL
}, WINED3D_GL_EXT_NONE
},
7358 {STATE_RENDER(WINED3D_RS_EMISSIVEMATERIALSOURCE
), {STATE_SHADER(WINED3D_SHADER_TYPE_VERTEX
), NULL
}, WINED3D_GL_EXT_NONE
},
7359 {STATE_RENDER(WINED3D_RS_VERTEXBLEND
), {STATE_SHADER(WINED3D_SHADER_TYPE_VERTEX
), NULL
}, WINED3D_GL_EXT_NONE
},
7360 {STATE_RENDER(WINED3D_RS_POINTSIZE
), {STATE_RENDER(WINED3D_RS_POINTSCALEENABLE
), NULL
}, WINED3D_GL_EXT_NONE
},
7361 {STATE_RENDER(WINED3D_RS_POINTSIZE_MIN
), {STATE_RENDER(WINED3D_RS_POINTSIZE_MIN
), state_psizemin_arb
}, ARB_POINT_PARAMETERS
},
7362 {STATE_RENDER(WINED3D_RS_POINTSIZE_MIN
), {STATE_RENDER(WINED3D_RS_POINTSIZE_MIN
), state_psizemin_ext
}, EXT_POINT_PARAMETERS
},
7363 {STATE_RENDER(WINED3D_RS_POINTSIZE_MIN
), {STATE_RENDER(WINED3D_RS_POINTSIZE_MIN
), state_psizemin_w
}, WINED3D_GL_EXT_NONE
},
7364 {STATE_RENDER(WINED3D_RS_POINTSPRITEENABLE
), {STATE_RENDER(WINED3D_RS_POINTSPRITEENABLE
), state_pointsprite
}, ARB_POINT_SPRITE
},
7365 {STATE_RENDER(WINED3D_RS_POINTSPRITEENABLE
), {STATE_RENDER(WINED3D_RS_POINTSPRITEENABLE
), state_pointsprite_w
}, WINED3D_GL_EXT_NONE
},
7366 {STATE_RENDER(WINED3D_RS_POINTSCALEENABLE
), {STATE_RENDER(WINED3D_RS_POINTSCALEENABLE
), state_pscale
}, WINED3D_GL_EXT_NONE
},
7367 {STATE_RENDER(WINED3D_RS_POINTSCALE_A
), {STATE_RENDER(WINED3D_RS_POINTSCALEENABLE
), NULL
}, WINED3D_GL_EXT_NONE
},
7368 {STATE_RENDER(WINED3D_RS_POINTSCALE_B
), {STATE_RENDER(WINED3D_RS_POINTSCALEENABLE
), NULL
}, WINED3D_GL_EXT_NONE
},
7369 {STATE_RENDER(WINED3D_RS_POINTSCALE_C
), {STATE_RENDER(WINED3D_RS_POINTSCALEENABLE
), NULL
}, WINED3D_GL_EXT_NONE
},
7370 {STATE_RENDER(WINED3D_RS_POINTSIZE_MAX
), {STATE_RENDER(WINED3D_RS_POINTSIZE_MIN
), NULL
}, ARB_POINT_PARAMETERS
},
7371 {STATE_RENDER(WINED3D_RS_POINTSIZE_MAX
), {STATE_RENDER(WINED3D_RS_POINTSIZE_MIN
), NULL
}, EXT_POINT_PARAMETERS
},
7372 {STATE_RENDER(WINED3D_RS_POINTSIZE_MAX
), {STATE_RENDER(WINED3D_RS_POINTSIZE_MIN
), NULL
}, WINED3D_GL_EXT_NONE
},
7373 {STATE_RENDER(WINED3D_RS_TWEENFACTOR
), {STATE_SHADER(WINED3D_SHADER_TYPE_VERTEX
), NULL
}, WINED3D_GL_EXT_NONE
},
7374 {STATE_RENDER(WINED3D_RS_INDEXEDVERTEXBLENDENABLE
), {STATE_SHADER(WINED3D_SHADER_TYPE_VERTEX
), NULL
}, WINED3D_GL_EXT_NONE
},
7375 /* Samplers for NP2 texture matrix adjustions. They are not needed if
7376 * GL_ARB_texture_non_power_of_two is supported, so register a NULL state
7377 * handler in that case to get the vertex part of sampler() skipped (VTF
7378 * is handled in the misc states). Otherwise, register
7379 * sampler_texmatrix(), which takes care of updating the texture matrix. */
7380 {STATE_SAMPLER(0), {0, NULL
}, ARB_TEXTURE_NON_POWER_OF_TWO
},
7381 {STATE_SAMPLER(0), {0, NULL
}, WINED3D_GL_NORMALIZED_TEXRECT
},
7382 {STATE_SAMPLER(0), {STATE_SAMPLER(0), sampler_texmatrix
}, WINED3D_GL_EXT_NONE
},
7383 {STATE_SAMPLER(1), {0, NULL
}, ARB_TEXTURE_NON_POWER_OF_TWO
},
7384 {STATE_SAMPLER(1), {0, NULL
}, WINED3D_GL_NORMALIZED_TEXRECT
},
7385 {STATE_SAMPLER(1), {STATE_SAMPLER(1), sampler_texmatrix
}, WINED3D_GL_EXT_NONE
},
7386 {STATE_SAMPLER(2), {0, NULL
}, ARB_TEXTURE_NON_POWER_OF_TWO
},
7387 {STATE_SAMPLER(2), {0, NULL
}, WINED3D_GL_NORMALIZED_TEXRECT
},
7388 {STATE_SAMPLER(2), {STATE_SAMPLER(2), sampler_texmatrix
}, WINED3D_GL_EXT_NONE
},
7389 {STATE_SAMPLER(3), {0, NULL
}, ARB_TEXTURE_NON_POWER_OF_TWO
},
7390 {STATE_SAMPLER(3), {0, NULL
}, WINED3D_GL_NORMALIZED_TEXRECT
},
7391 {STATE_SAMPLER(3), {STATE_SAMPLER(3), sampler_texmatrix
}, WINED3D_GL_EXT_NONE
},
7392 {STATE_SAMPLER(4), {0, NULL
}, ARB_TEXTURE_NON_POWER_OF_TWO
},
7393 {STATE_SAMPLER(4), {0, NULL
}, WINED3D_GL_NORMALIZED_TEXRECT
},
7394 {STATE_SAMPLER(4), {STATE_SAMPLER(4), sampler_texmatrix
}, WINED3D_GL_EXT_NONE
},
7395 {STATE_SAMPLER(5), {0, NULL
}, ARB_TEXTURE_NON_POWER_OF_TWO
},
7396 {STATE_SAMPLER(5), {0, NULL
}, WINED3D_GL_NORMALIZED_TEXRECT
},
7397 {STATE_SAMPLER(5), {STATE_SAMPLER(5), sampler_texmatrix
}, WINED3D_GL_EXT_NONE
},
7398 {STATE_SAMPLER(6), {0, NULL
}, ARB_TEXTURE_NON_POWER_OF_TWO
},
7399 {STATE_SAMPLER(6), {0, NULL
}, WINED3D_GL_NORMALIZED_TEXRECT
},
7400 {STATE_SAMPLER(6), {STATE_SAMPLER(6), sampler_texmatrix
}, WINED3D_GL_EXT_NONE
},
7401 {STATE_SAMPLER(7), {0, NULL
}, ARB_TEXTURE_NON_POWER_OF_TWO
},
7402 {STATE_SAMPLER(7), {0, NULL
}, WINED3D_GL_NORMALIZED_TEXRECT
},
7403 {STATE_SAMPLER(7), {STATE_SAMPLER(7), sampler_texmatrix
}, WINED3D_GL_EXT_NONE
},
7404 {STATE_POINT_SIZE_ENABLE
, {STATE_RENDER(WINED3D_RS_FOGENABLE
), NULL
}, WINED3D_GL_EXT_NONE
},
7405 {0 /* Terminate */, {0, NULL
}, WINED3D_GL_EXT_NONE
},
7409 * - This currently depends on GL fixed function functions to set things
7410 * like light parameters. Ideally we'd use regular uniforms for that.
7411 * - In part because of the previous point, much of this is modelled after
7412 * GL fixed function, and has much of the same limitations. For example,
7413 * D3D spot lights are slightly different from GL spot lights.
7414 * - We can now implement drawing transformed vertices using the GLSL pipe,
7415 * instead of using the immediate mode fallback.
7416 * - Similarly, we don't need the fallback for certain combinations of
7417 * material sources anymore.
7418 * - Implement vertex blending and vertex tweening.
7419 * - Handle WINED3D_TSS_TEXCOORD_INDEX in the shader, instead of duplicating
7420 * attribute arrays in load_tex_coords().
7421 * - Per-vertex point sizes. */
7422 const struct wined3d_vertex_pipe_ops glsl_vertex_pipe
=
7424 glsl_vertex_pipe_vp_enable
,
7425 glsl_vertex_pipe_vp_get_caps
,
7426 glsl_vertex_pipe_vp_alloc
,
7427 glsl_vertex_pipe_vp_free
,
7428 glsl_vertex_pipe_vp_states
,
7431 static void glsl_fragment_pipe_enable(const struct wined3d_gl_info
*gl_info
, BOOL enable
)
7433 /* Nothing to do. */
7436 static void glsl_fragment_pipe_get_caps(const struct wined3d_gl_info
*gl_info
, struct fragment_caps
*caps
)
7438 caps
->wined3d_caps
= WINED3D_FRAGMENT_CAP_PROJ_CONTROL
7439 | WINED3D_FRAGMENT_CAP_SRGB_WRITE
7440 | WINED3D_FRAGMENT_CAP_COLOR_KEY
;
7441 caps
->PrimitiveMiscCaps
= WINED3DPMISCCAPS_TSSARGTEMP
7442 | WINED3DPMISCCAPS_PERSTAGECONSTANT
;
7443 caps
->TextureOpCaps
= WINED3DTEXOPCAPS_DISABLE
7444 | WINED3DTEXOPCAPS_SELECTARG1
7445 | WINED3DTEXOPCAPS_SELECTARG2
7446 | WINED3DTEXOPCAPS_MODULATE4X
7447 | WINED3DTEXOPCAPS_MODULATE2X
7448 | WINED3DTEXOPCAPS_MODULATE
7449 | WINED3DTEXOPCAPS_ADDSIGNED2X
7450 | WINED3DTEXOPCAPS_ADDSIGNED
7451 | WINED3DTEXOPCAPS_ADD
7452 | WINED3DTEXOPCAPS_SUBTRACT
7453 | WINED3DTEXOPCAPS_ADDSMOOTH
7454 | WINED3DTEXOPCAPS_BLENDCURRENTALPHA
7455 | WINED3DTEXOPCAPS_BLENDFACTORALPHA
7456 | WINED3DTEXOPCAPS_BLENDTEXTUREALPHA
7457 | WINED3DTEXOPCAPS_BLENDDIFFUSEALPHA
7458 | WINED3DTEXOPCAPS_BLENDTEXTUREALPHAPM
7459 | WINED3DTEXOPCAPS_MODULATEALPHA_ADDCOLOR
7460 | WINED3DTEXOPCAPS_MODULATECOLOR_ADDALPHA
7461 | WINED3DTEXOPCAPS_MODULATEINVCOLOR_ADDALPHA
7462 | WINED3DTEXOPCAPS_MODULATEINVALPHA_ADDCOLOR
7463 | WINED3DTEXOPCAPS_DOTPRODUCT3
7464 | WINED3DTEXOPCAPS_MULTIPLYADD
7465 | WINED3DTEXOPCAPS_LERP
7466 | WINED3DTEXOPCAPS_BUMPENVMAP
7467 | WINED3DTEXOPCAPS_BUMPENVMAPLUMINANCE
;
7468 caps
->MaxTextureBlendStages
= 8;
7469 caps
->MaxSimultaneousTextures
= min(gl_info
->limits
.fragment_samplers
, 8);
7472 static void *glsl_fragment_pipe_alloc(const struct wined3d_shader_backend_ops
*shader_backend
, void *shader_priv
)
7474 struct shader_glsl_priv
*priv
;
7476 if (shader_backend
== &glsl_shader_backend
)
7480 if (wine_rb_init(&priv
->ffp_fragment_shaders
, &wined3d_ffp_frag_program_rb_functions
) == -1)
7482 ERR("Failed to initialize rbtree.\n");
7489 FIXME("GLSL fragment pipe without GLSL shader backend not implemented.\n");
7494 static void shader_glsl_free_ffp_fragment_shader(struct wine_rb_entry
*entry
, void *context
)
7496 struct glsl_ffp_fragment_shader
*shader
= WINE_RB_ENTRY_VALUE(entry
,
7497 struct glsl_ffp_fragment_shader
, entry
.entry
);
7498 struct glsl_shader_prog_link
*program
, *program2
;
7499 struct glsl_ffp_destroy_ctx
*ctx
= context
;
7501 LIST_FOR_EACH_ENTRY_SAFE(program
, program2
, &shader
->linked_programs
,
7502 struct glsl_shader_prog_link
, ps
.shader_entry
)
7504 delete_glsl_program_entry(ctx
->priv
, ctx
->gl_info
, program
);
7506 ctx
->gl_info
->gl_ops
.ext
.p_glDeleteShader(shader
->id
);
7507 HeapFree(GetProcessHeap(), 0, shader
);
7510 /* Context activation is done by the caller. */
7511 static void glsl_fragment_pipe_free(struct wined3d_device
*device
)
7513 struct shader_glsl_priv
*priv
= device
->fragment_priv
;
7514 struct glsl_ffp_destroy_ctx ctx
;
7517 ctx
.gl_info
= &device
->adapter
->gl_info
;
7518 wine_rb_destroy(&priv
->ffp_fragment_shaders
, shader_glsl_free_ffp_fragment_shader
, &ctx
);
7521 static void glsl_fragment_pipe_shader(struct wined3d_context
*context
,
7522 const struct wined3d_state
*state
, DWORD state_id
)
7524 context
->last_was_pshader
= use_ps(state
);
7526 context
->shader_update_mask
|= 1 << WINED3D_SHADER_TYPE_PIXEL
;
7529 static void glsl_fragment_pipe_fog(struct wined3d_context
*context
,
7530 const struct wined3d_state
*state
, DWORD state_id
)
7532 BOOL use_vshader
= use_vs(state
);
7533 enum fogsource new_source
;
7534 DWORD fogstart
= state
->render_states
[WINED3D_RS_FOGSTART
];
7535 DWORD fogend
= state
->render_states
[WINED3D_RS_FOGEND
];
7537 context
->shader_update_mask
|= 1 << WINED3D_SHADER_TYPE_PIXEL
;
7539 if (!state
->render_states
[WINED3D_RS_FOGENABLE
])
7542 if (state
->render_states
[WINED3D_RS_FOGTABLEMODE
] == WINED3D_FOG_NONE
)
7545 new_source
= FOGSOURCE_VS
;
7546 else if (state
->render_states
[WINED3D_RS_FOGVERTEXMODE
] == WINED3D_FOG_NONE
|| context
->stream_info
.position_transformed
)
7547 new_source
= FOGSOURCE_COORD
;
7549 new_source
= FOGSOURCE_FFP
;
7553 new_source
= FOGSOURCE_FFP
;
7556 if (new_source
!= context
->fog_source
|| fogstart
== fogend
)
7558 context
->fog_source
= new_source
;
7559 state_fogstartend(context
, state
, STATE_RENDER(WINED3D_RS_FOGSTART
));
7563 static void glsl_fragment_pipe_vdecl(struct wined3d_context
*context
,
7564 const struct wined3d_state
*state
, DWORD state_id
)
7566 if (!isStateDirty(context
, STATE_RENDER(WINED3D_RS_FOGENABLE
)))
7567 glsl_fragment_pipe_fog(context
, state
, state_id
);
7570 static void glsl_fragment_pipe_tex_transform(struct wined3d_context
*context
,
7571 const struct wined3d_state
*state
, DWORD state_id
)
7573 context
->shader_update_mask
|= 1 << WINED3D_SHADER_TYPE_PIXEL
;
7576 static void glsl_fragment_pipe_invalidate_constants(struct wined3d_context
*context
,
7577 const struct wined3d_state
*state
, DWORD state_id
)
7579 context
->constant_update_mask
|= WINED3D_SHADER_CONST_FFP_PS
;
7582 static void glsl_fragment_pipe_alpha_test(struct wined3d_context
*context
,
7583 const struct wined3d_state
*state
, DWORD state_id
)
7585 const struct wined3d_gl_info
*gl_info
= context
->gl_info
;
7589 TRACE("context %p, state %p, state_id %#x.\n", context
, state
, state_id
);
7591 if (state
->render_states
[WINED3D_RS_ALPHATESTENABLE
])
7593 gl_info
->gl_ops
.gl
.p_glEnable(GL_ALPHA_TEST
);
7594 checkGLcall("glEnable GL_ALPHA_TEST");
7598 gl_info
->gl_ops
.gl
.p_glDisable(GL_ALPHA_TEST
);
7599 checkGLcall("glDisable GL_ALPHA_TEST");
7603 ref
= ((float)state
->render_states
[WINED3D_RS_ALPHAREF
]) / 255.0f
;
7604 glParm
= wined3d_gl_compare_func(state
->render_states
[WINED3D_RS_ALPHAFUNC
]);
7608 gl_info
->gl_ops
.gl
.p_glAlphaFunc(glParm
, ref
);
7609 checkGLcall("glAlphaFunc");
7613 static const struct StateEntryTemplate glsl_fragment_pipe_state_template
[] =
7615 {STATE_VDECL
, {STATE_VDECL
, glsl_fragment_pipe_vdecl
}, WINED3D_GL_EXT_NONE
},
7616 {STATE_RENDER(WINED3D_RS_TEXTUREFACTOR
), {STATE_RENDER(WINED3D_RS_TEXTUREFACTOR
), glsl_fragment_pipe_invalidate_constants
}, WINED3D_GL_EXT_NONE
},
7617 {STATE_TEXTURESTAGE(0, WINED3D_TSS_COLOR_OP
), {STATE_SHADER(WINED3D_SHADER_TYPE_PIXEL
), NULL
}, WINED3D_GL_EXT_NONE
},
7618 {STATE_TEXTURESTAGE(0, WINED3D_TSS_COLOR_ARG1
), {STATE_SHADER(WINED3D_SHADER_TYPE_PIXEL
), NULL
}, WINED3D_GL_EXT_NONE
},
7619 {STATE_TEXTURESTAGE(0, WINED3D_TSS_COLOR_ARG2
), {STATE_SHADER(WINED3D_SHADER_TYPE_PIXEL
), NULL
}, WINED3D_GL_EXT_NONE
},
7620 {STATE_TEXTURESTAGE(0, WINED3D_TSS_COLOR_ARG0
), {STATE_SHADER(WINED3D_SHADER_TYPE_PIXEL
), NULL
}, WINED3D_GL_EXT_NONE
},
7621 {STATE_TEXTURESTAGE(0, WINED3D_TSS_ALPHA_OP
), {STATE_SHADER(WINED3D_SHADER_TYPE_PIXEL
), NULL
}, WINED3D_GL_EXT_NONE
},
7622 {STATE_TEXTURESTAGE(0, WINED3D_TSS_ALPHA_ARG1
), {STATE_SHADER(WINED3D_SHADER_TYPE_PIXEL
), NULL
}, WINED3D_GL_EXT_NONE
},
7623 {STATE_TEXTURESTAGE(0, WINED3D_TSS_ALPHA_ARG2
), {STATE_SHADER(WINED3D_SHADER_TYPE_PIXEL
), NULL
}, WINED3D_GL_EXT_NONE
},
7624 {STATE_TEXTURESTAGE(0, WINED3D_TSS_ALPHA_ARG0
), {STATE_SHADER(WINED3D_SHADER_TYPE_PIXEL
), NULL
}, WINED3D_GL_EXT_NONE
},
7625 {STATE_TEXTURESTAGE(0, WINED3D_TSS_RESULT_ARG
), {STATE_SHADER(WINED3D_SHADER_TYPE_PIXEL
), NULL
}, WINED3D_GL_EXT_NONE
},
7626 {STATE_TEXTURESTAGE(1, WINED3D_TSS_COLOR_OP
), {STATE_SHADER(WINED3D_SHADER_TYPE_PIXEL
), NULL
}, WINED3D_GL_EXT_NONE
},
7627 {STATE_TEXTURESTAGE(1, WINED3D_TSS_COLOR_ARG1
), {STATE_SHADER(WINED3D_SHADER_TYPE_PIXEL
), NULL
}, WINED3D_GL_EXT_NONE
},
7628 {STATE_TEXTURESTAGE(1, WINED3D_TSS_COLOR_ARG2
), {STATE_SHADER(WINED3D_SHADER_TYPE_PIXEL
), NULL
}, WINED3D_GL_EXT_NONE
},
7629 {STATE_TEXTURESTAGE(1, WINED3D_TSS_COLOR_ARG0
), {STATE_SHADER(WINED3D_SHADER_TYPE_PIXEL
), NULL
}, WINED3D_GL_EXT_NONE
},
7630 {STATE_TEXTURESTAGE(1, WINED3D_TSS_ALPHA_OP
), {STATE_SHADER(WINED3D_SHADER_TYPE_PIXEL
), NULL
}, WINED3D_GL_EXT_NONE
},
7631 {STATE_TEXTURESTAGE(1, WINED3D_TSS_ALPHA_ARG1
), {STATE_SHADER(WINED3D_SHADER_TYPE_PIXEL
), NULL
}, WINED3D_GL_EXT_NONE
},
7632 {STATE_TEXTURESTAGE(1, WINED3D_TSS_ALPHA_ARG2
), {STATE_SHADER(WINED3D_SHADER_TYPE_PIXEL
), NULL
}, WINED3D_GL_EXT_NONE
},
7633 {STATE_TEXTURESTAGE(1, WINED3D_TSS_ALPHA_ARG0
), {STATE_SHADER(WINED3D_SHADER_TYPE_PIXEL
), NULL
}, WINED3D_GL_EXT_NONE
},
7634 {STATE_TEXTURESTAGE(1, WINED3D_TSS_RESULT_ARG
), {STATE_SHADER(WINED3D_SHADER_TYPE_PIXEL
), NULL
}, WINED3D_GL_EXT_NONE
},
7635 {STATE_TEXTURESTAGE(2, WINED3D_TSS_COLOR_OP
), {STATE_SHADER(WINED3D_SHADER_TYPE_PIXEL
), NULL
}, WINED3D_GL_EXT_NONE
},
7636 {STATE_TEXTURESTAGE(2, WINED3D_TSS_COLOR_ARG1
), {STATE_SHADER(WINED3D_SHADER_TYPE_PIXEL
), NULL
}, WINED3D_GL_EXT_NONE
},
7637 {STATE_TEXTURESTAGE(2, WINED3D_TSS_COLOR_ARG2
), {STATE_SHADER(WINED3D_SHADER_TYPE_PIXEL
), NULL
}, WINED3D_GL_EXT_NONE
},
7638 {STATE_TEXTURESTAGE(2, WINED3D_TSS_COLOR_ARG0
), {STATE_SHADER(WINED3D_SHADER_TYPE_PIXEL
), NULL
}, WINED3D_GL_EXT_NONE
},
7639 {STATE_TEXTURESTAGE(2, WINED3D_TSS_ALPHA_OP
), {STATE_SHADER(WINED3D_SHADER_TYPE_PIXEL
), NULL
}, WINED3D_GL_EXT_NONE
},
7640 {STATE_TEXTURESTAGE(2, WINED3D_TSS_ALPHA_ARG1
), {STATE_SHADER(WINED3D_SHADER_TYPE_PIXEL
), NULL
}, WINED3D_GL_EXT_NONE
},
7641 {STATE_TEXTURESTAGE(2, WINED3D_TSS_ALPHA_ARG2
), {STATE_SHADER(WINED3D_SHADER_TYPE_PIXEL
), NULL
}, WINED3D_GL_EXT_NONE
},
7642 {STATE_TEXTURESTAGE(2, WINED3D_TSS_ALPHA_ARG0
), {STATE_SHADER(WINED3D_SHADER_TYPE_PIXEL
), NULL
}, WINED3D_GL_EXT_NONE
},
7643 {STATE_TEXTURESTAGE(2, WINED3D_TSS_RESULT_ARG
), {STATE_SHADER(WINED3D_SHADER_TYPE_PIXEL
), NULL
}, WINED3D_GL_EXT_NONE
},
7644 {STATE_TEXTURESTAGE(3, WINED3D_TSS_COLOR_OP
), {STATE_SHADER(WINED3D_SHADER_TYPE_PIXEL
), NULL
}, WINED3D_GL_EXT_NONE
},
7645 {STATE_TEXTURESTAGE(3, WINED3D_TSS_COLOR_ARG1
), {STATE_SHADER(WINED3D_SHADER_TYPE_PIXEL
), NULL
}, WINED3D_GL_EXT_NONE
},
7646 {STATE_TEXTURESTAGE(3, WINED3D_TSS_COLOR_ARG2
), {STATE_SHADER(WINED3D_SHADER_TYPE_PIXEL
), NULL
}, WINED3D_GL_EXT_NONE
},
7647 {STATE_TEXTURESTAGE(3, WINED3D_TSS_COLOR_ARG0
), {STATE_SHADER(WINED3D_SHADER_TYPE_PIXEL
), NULL
}, WINED3D_GL_EXT_NONE
},
7648 {STATE_TEXTURESTAGE(3, WINED3D_TSS_ALPHA_OP
), {STATE_SHADER(WINED3D_SHADER_TYPE_PIXEL
), NULL
}, WINED3D_GL_EXT_NONE
},
7649 {STATE_TEXTURESTAGE(3, WINED3D_TSS_ALPHA_ARG1
), {STATE_SHADER(WINED3D_SHADER_TYPE_PIXEL
), NULL
}, WINED3D_GL_EXT_NONE
},
7650 {STATE_TEXTURESTAGE(3, WINED3D_TSS_ALPHA_ARG2
), {STATE_SHADER(WINED3D_SHADER_TYPE_PIXEL
), NULL
}, WINED3D_GL_EXT_NONE
},
7651 {STATE_TEXTURESTAGE(3, WINED3D_TSS_ALPHA_ARG0
), {STATE_SHADER(WINED3D_SHADER_TYPE_PIXEL
), NULL
}, WINED3D_GL_EXT_NONE
},
7652 {STATE_TEXTURESTAGE(3, WINED3D_TSS_RESULT_ARG
), {STATE_SHADER(WINED3D_SHADER_TYPE_PIXEL
), NULL
}, WINED3D_GL_EXT_NONE
},
7653 {STATE_TEXTURESTAGE(4, WINED3D_TSS_COLOR_OP
), {STATE_SHADER(WINED3D_SHADER_TYPE_PIXEL
), NULL
}, WINED3D_GL_EXT_NONE
},
7654 {STATE_TEXTURESTAGE(4, WINED3D_TSS_COLOR_ARG1
), {STATE_SHADER(WINED3D_SHADER_TYPE_PIXEL
), NULL
}, WINED3D_GL_EXT_NONE
},
7655 {STATE_TEXTURESTAGE(4, WINED3D_TSS_COLOR_ARG2
), {STATE_SHADER(WINED3D_SHADER_TYPE_PIXEL
), NULL
}, WINED3D_GL_EXT_NONE
},
7656 {STATE_TEXTURESTAGE(4, WINED3D_TSS_COLOR_ARG0
), {STATE_SHADER(WINED3D_SHADER_TYPE_PIXEL
), NULL
}, WINED3D_GL_EXT_NONE
},
7657 {STATE_TEXTURESTAGE(4, WINED3D_TSS_ALPHA_OP
), {STATE_SHADER(WINED3D_SHADER_TYPE_PIXEL
), NULL
}, WINED3D_GL_EXT_NONE
},
7658 {STATE_TEXTURESTAGE(4, WINED3D_TSS_ALPHA_ARG1
), {STATE_SHADER(WINED3D_SHADER_TYPE_PIXEL
), NULL
}, WINED3D_GL_EXT_NONE
},
7659 {STATE_TEXTURESTAGE(4, WINED3D_TSS_ALPHA_ARG2
), {STATE_SHADER(WINED3D_SHADER_TYPE_PIXEL
), NULL
}, WINED3D_GL_EXT_NONE
},
7660 {STATE_TEXTURESTAGE(4, WINED3D_TSS_ALPHA_ARG0
), {STATE_SHADER(WINED3D_SHADER_TYPE_PIXEL
), NULL
}, WINED3D_GL_EXT_NONE
},
7661 {STATE_TEXTURESTAGE(4, WINED3D_TSS_RESULT_ARG
), {STATE_SHADER(WINED3D_SHADER_TYPE_PIXEL
), NULL
}, WINED3D_GL_EXT_NONE
},
7662 {STATE_TEXTURESTAGE(5, WINED3D_TSS_COLOR_OP
), {STATE_SHADER(WINED3D_SHADER_TYPE_PIXEL
), NULL
}, WINED3D_GL_EXT_NONE
},
7663 {STATE_TEXTURESTAGE(5, WINED3D_TSS_COLOR_ARG1
), {STATE_SHADER(WINED3D_SHADER_TYPE_PIXEL
), NULL
}, WINED3D_GL_EXT_NONE
},
7664 {STATE_TEXTURESTAGE(5, WINED3D_TSS_COLOR_ARG2
), {STATE_SHADER(WINED3D_SHADER_TYPE_PIXEL
), NULL
}, WINED3D_GL_EXT_NONE
},
7665 {STATE_TEXTURESTAGE(5, WINED3D_TSS_COLOR_ARG0
), {STATE_SHADER(WINED3D_SHADER_TYPE_PIXEL
), NULL
}, WINED3D_GL_EXT_NONE
},
7666 {STATE_TEXTURESTAGE(5, WINED3D_TSS_ALPHA_OP
), {STATE_SHADER(WINED3D_SHADER_TYPE_PIXEL
), NULL
}, WINED3D_GL_EXT_NONE
},
7667 {STATE_TEXTURESTAGE(5, WINED3D_TSS_ALPHA_ARG1
), {STATE_SHADER(WINED3D_SHADER_TYPE_PIXEL
), NULL
}, WINED3D_GL_EXT_NONE
},
7668 {STATE_TEXTURESTAGE(5, WINED3D_TSS_ALPHA_ARG2
), {STATE_SHADER(WINED3D_SHADER_TYPE_PIXEL
), NULL
}, WINED3D_GL_EXT_NONE
},
7669 {STATE_TEXTURESTAGE(5, WINED3D_TSS_ALPHA_ARG0
), {STATE_SHADER(WINED3D_SHADER_TYPE_PIXEL
), NULL
}, WINED3D_GL_EXT_NONE
},
7670 {STATE_TEXTURESTAGE(5, WINED3D_TSS_RESULT_ARG
), {STATE_SHADER(WINED3D_SHADER_TYPE_PIXEL
), NULL
}, WINED3D_GL_EXT_NONE
},
7671 {STATE_TEXTURESTAGE(6, WINED3D_TSS_COLOR_OP
), {STATE_SHADER(WINED3D_SHADER_TYPE_PIXEL
), NULL
}, WINED3D_GL_EXT_NONE
},
7672 {STATE_TEXTURESTAGE(6, WINED3D_TSS_COLOR_ARG1
), {STATE_SHADER(WINED3D_SHADER_TYPE_PIXEL
), NULL
}, WINED3D_GL_EXT_NONE
},
7673 {STATE_TEXTURESTAGE(6, WINED3D_TSS_COLOR_ARG2
), {STATE_SHADER(WINED3D_SHADER_TYPE_PIXEL
), NULL
}, WINED3D_GL_EXT_NONE
},
7674 {STATE_TEXTURESTAGE(6, WINED3D_TSS_COLOR_ARG0
), {STATE_SHADER(WINED3D_SHADER_TYPE_PIXEL
), NULL
}, WINED3D_GL_EXT_NONE
},
7675 {STATE_TEXTURESTAGE(6, WINED3D_TSS_ALPHA_OP
), {STATE_SHADER(WINED3D_SHADER_TYPE_PIXEL
), NULL
}, WINED3D_GL_EXT_NONE
},
7676 {STATE_TEXTURESTAGE(6, WINED3D_TSS_ALPHA_ARG1
), {STATE_SHADER(WINED3D_SHADER_TYPE_PIXEL
), NULL
}, WINED3D_GL_EXT_NONE
},
7677 {STATE_TEXTURESTAGE(6, WINED3D_TSS_ALPHA_ARG2
), {STATE_SHADER(WINED3D_SHADER_TYPE_PIXEL
), NULL
}, WINED3D_GL_EXT_NONE
},
7678 {STATE_TEXTURESTAGE(6, WINED3D_TSS_ALPHA_ARG0
), {STATE_SHADER(WINED3D_SHADER_TYPE_PIXEL
), NULL
}, WINED3D_GL_EXT_NONE
},
7679 {STATE_TEXTURESTAGE(6, WINED3D_TSS_RESULT_ARG
), {STATE_SHADER(WINED3D_SHADER_TYPE_PIXEL
), NULL
}, WINED3D_GL_EXT_NONE
},
7680 {STATE_TEXTURESTAGE(7, WINED3D_TSS_COLOR_OP
), {STATE_SHADER(WINED3D_SHADER_TYPE_PIXEL
), NULL
}, WINED3D_GL_EXT_NONE
},
7681 {STATE_TEXTURESTAGE(7, WINED3D_TSS_COLOR_ARG1
), {STATE_SHADER(WINED3D_SHADER_TYPE_PIXEL
), NULL
}, WINED3D_GL_EXT_NONE
},
7682 {STATE_TEXTURESTAGE(7, WINED3D_TSS_COLOR_ARG2
), {STATE_SHADER(WINED3D_SHADER_TYPE_PIXEL
), NULL
}, WINED3D_GL_EXT_NONE
},
7683 {STATE_TEXTURESTAGE(7, WINED3D_TSS_COLOR_ARG0
), {STATE_SHADER(WINED3D_SHADER_TYPE_PIXEL
), NULL
}, WINED3D_GL_EXT_NONE
},
7684 {STATE_TEXTURESTAGE(7, WINED3D_TSS_ALPHA_OP
), {STATE_SHADER(WINED3D_SHADER_TYPE_PIXEL
), NULL
}, WINED3D_GL_EXT_NONE
},
7685 {STATE_TEXTURESTAGE(7, WINED3D_TSS_ALPHA_ARG1
), {STATE_SHADER(WINED3D_SHADER_TYPE_PIXEL
), NULL
}, WINED3D_GL_EXT_NONE
},
7686 {STATE_TEXTURESTAGE(7, WINED3D_TSS_ALPHA_ARG2
), {STATE_SHADER(WINED3D_SHADER_TYPE_PIXEL
), NULL
}, WINED3D_GL_EXT_NONE
},
7687 {STATE_TEXTURESTAGE(7, WINED3D_TSS_ALPHA_ARG0
), {STATE_SHADER(WINED3D_SHADER_TYPE_PIXEL
), NULL
}, WINED3D_GL_EXT_NONE
},
7688 {STATE_TEXTURESTAGE(7, WINED3D_TSS_RESULT_ARG
), {STATE_SHADER(WINED3D_SHADER_TYPE_PIXEL
), NULL
}, WINED3D_GL_EXT_NONE
},
7689 {STATE_SHADER(WINED3D_SHADER_TYPE_PIXEL
), {STATE_SHADER(WINED3D_SHADER_TYPE_PIXEL
), glsl_fragment_pipe_shader
}, WINED3D_GL_EXT_NONE
},
7690 {STATE_RENDER(WINED3D_RS_ALPHAFUNC
), {STATE_RENDER(WINED3D_RS_ALPHATESTENABLE
), NULL
}, WINED3D_GL_EXT_NONE
},
7691 {STATE_RENDER(WINED3D_RS_ALPHAREF
), {STATE_RENDER(WINED3D_RS_ALPHATESTENABLE
), NULL
}, WINED3D_GL_EXT_NONE
},
7692 {STATE_RENDER(WINED3D_RS_ALPHATESTENABLE
), {STATE_RENDER(WINED3D_RS_ALPHATESTENABLE
), glsl_fragment_pipe_alpha_test
}, WINED3D_GL_EXT_NONE
},
7693 {STATE_RENDER(WINED3D_RS_COLORKEYENABLE
), {STATE_SHADER(WINED3D_SHADER_TYPE_PIXEL
), NULL
}, WINED3D_GL_EXT_NONE
},
7694 {STATE_RENDER(WINED3D_RS_FOGENABLE
), {STATE_RENDER(WINED3D_RS_FOGENABLE
), glsl_fragment_pipe_fog
}, WINED3D_GL_EXT_NONE
},
7695 {STATE_RENDER(WINED3D_RS_FOGTABLEMODE
), {STATE_RENDER(WINED3D_RS_FOGENABLE
), NULL
}, WINED3D_GL_EXT_NONE
},
7696 {STATE_RENDER(WINED3D_RS_FOGVERTEXMODE
), {STATE_RENDER(WINED3D_RS_FOGENABLE
), NULL
}, WINED3D_GL_EXT_NONE
},
7697 {STATE_RENDER(WINED3D_RS_FOGSTART
), {STATE_RENDER(WINED3D_RS_FOGSTART
), state_fogstartend
}, WINED3D_GL_EXT_NONE
},
7698 {STATE_RENDER(WINED3D_RS_FOGEND
), {STATE_RENDER(WINED3D_RS_FOGSTART
), NULL
}, WINED3D_GL_EXT_NONE
},
7699 {STATE_RENDER(WINED3D_RS_SRGBWRITEENABLE
), {STATE_RENDER(WINED3D_RS_SRGBWRITEENABLE
), state_srgbwrite
}, ARB_FRAMEBUFFER_SRGB
},
7700 {STATE_RENDER(WINED3D_RS_SRGBWRITEENABLE
), {STATE_SHADER(WINED3D_SHADER_TYPE_PIXEL
), NULL
}, WINED3D_GL_EXT_NONE
},
7701 {STATE_RENDER(WINED3D_RS_FOGCOLOR
), {STATE_RENDER(WINED3D_RS_FOGCOLOR
), state_fogcolor
}, WINED3D_GL_EXT_NONE
},
7702 {STATE_RENDER(WINED3D_RS_FOGDENSITY
), {STATE_RENDER(WINED3D_RS_FOGDENSITY
), state_fogdensity
}, WINED3D_GL_EXT_NONE
},
7703 {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
},
7704 {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
},
7705 {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
},
7706 {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
},
7707 {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
},
7708 {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
},
7709 {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
},
7710 {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
},
7711 {STATE_TEXTURESTAGE(0, WINED3D_TSS_CONSTANT
), {STATE_TEXTURESTAGE(0, WINED3D_TSS_CONSTANT
), glsl_fragment_pipe_invalidate_constants
}, WINED3D_GL_EXT_NONE
},
7712 {STATE_TEXTURESTAGE(1, WINED3D_TSS_CONSTANT
), {STATE_TEXTURESTAGE(1, WINED3D_TSS_CONSTANT
), glsl_fragment_pipe_invalidate_constants
}, WINED3D_GL_EXT_NONE
},
7713 {STATE_TEXTURESTAGE(2, WINED3D_TSS_CONSTANT
), {STATE_TEXTURESTAGE(2, WINED3D_TSS_CONSTANT
), glsl_fragment_pipe_invalidate_constants
}, WINED3D_GL_EXT_NONE
},
7714 {STATE_TEXTURESTAGE(3, WINED3D_TSS_CONSTANT
), {STATE_TEXTURESTAGE(3, WINED3D_TSS_CONSTANT
), glsl_fragment_pipe_invalidate_constants
}, WINED3D_GL_EXT_NONE
},
7715 {STATE_TEXTURESTAGE(4, WINED3D_TSS_CONSTANT
), {STATE_TEXTURESTAGE(4, WINED3D_TSS_CONSTANT
), glsl_fragment_pipe_invalidate_constants
}, WINED3D_GL_EXT_NONE
},
7716 {STATE_TEXTURESTAGE(5, WINED3D_TSS_CONSTANT
), {STATE_TEXTURESTAGE(5, WINED3D_TSS_CONSTANT
), glsl_fragment_pipe_invalidate_constants
}, WINED3D_GL_EXT_NONE
},
7717 {STATE_TEXTURESTAGE(6, WINED3D_TSS_CONSTANT
), {STATE_TEXTURESTAGE(6, WINED3D_TSS_CONSTANT
), glsl_fragment_pipe_invalidate_constants
}, WINED3D_GL_EXT_NONE
},
7718 {STATE_TEXTURESTAGE(7, WINED3D_TSS_CONSTANT
), {STATE_TEXTURESTAGE(7, WINED3D_TSS_CONSTANT
), glsl_fragment_pipe_invalidate_constants
}, WINED3D_GL_EXT_NONE
},
7719 {STATE_RENDER(WINED3D_RS_SPECULARENABLE
), {STATE_RENDER(WINED3D_RS_SPECULARENABLE
), glsl_fragment_pipe_invalidate_constants
}, WINED3D_GL_EXT_NONE
},
7720 {0 /* Terminate */, {0, 0 }, WINED3D_GL_EXT_NONE
},
7723 static BOOL
glsl_fragment_pipe_alloc_context_data(struct wined3d_context
*context
)
7728 static void glsl_fragment_pipe_free_context_data(struct wined3d_context
*context
)
7732 const struct fragment_pipeline glsl_fragment_pipe
=
7734 glsl_fragment_pipe_enable
,
7735 glsl_fragment_pipe_get_caps
,
7736 glsl_fragment_pipe_alloc
,
7737 glsl_fragment_pipe_free
,
7738 glsl_fragment_pipe_alloc_context_data
,
7739 glsl_fragment_pipe_free_context_data
,
7740 shader_glsl_color_fixup_supported
,
7741 glsl_fragment_pipe_state_template
,