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.
35 #include "wined3d_private.h"
37 WINE_DEFAULT_DEBUG_CHANNEL(d3d_shader
);
38 WINE_DECLARE_DEBUG_CHANNEL(d3d
);
39 WINE_DECLARE_DEBUG_CHANNEL(winediag
);
41 #define WINED3D_GLSL_SAMPLE_PROJECTED 0x01
42 #define WINED3D_GLSL_SAMPLE_LOD 0x02
43 #define WINED3D_GLSL_SAMPLE_GRAD 0x04
44 #define WINED3D_GLSL_SAMPLE_LOAD 0x08
45 #define WINED3D_GLSL_SAMPLE_OFFSET 0x10
49 unsigned int coord_size
;
50 unsigned int resinfo_size
;
51 const char *type_part
;
53 resource_type_info
[] =
55 {0, 0, ""}, /* WINED3D_SHADER_RESOURCE_NONE */
56 {1, 1, "Buffer"}, /* WINED3D_SHADER_RESOURCE_BUFFER */
57 {1, 1, "1D"}, /* WINED3D_SHADER_RESOURCE_TEXTURE_1D */
58 {2, 2, "2D"}, /* WINED3D_SHADER_RESOURCE_TEXTURE_2D */
59 {2, 2, ""}, /* WINED3D_SHADER_RESOURCE_TEXTURE_2DMS */
60 {3, 3, "3D"}, /* WINED3D_SHADER_RESOURCE_TEXTURE_3D */
61 {3, 2, "Cube"}, /* WINED3D_SHADER_RESOURCE_TEXTURE_CUBE */
62 {2, 2, ""}, /* WINED3D_SHADER_RESOURCE_TEXTURE_1DARRAY */
63 {3, 3, "2DArray"}, /* WINED3D_SHADER_RESOURCE_TEXTURE_2DARRAY */
64 {3, 3, ""}, /* WINED3D_SHADER_RESOURCE_TEXTURE_2DMSARRAY */
65 {4, 3, ""}, /* WINED3D_SHADER_RESOURCE_TEXTURE_CUBEARRAY */
70 enum wined3d_data_type data_type
;
71 const char *glsl_scalar_type
;
72 const char *glsl_vector_type
;
74 component_type_info
[] =
76 {WINED3D_DATA_FLOAT
, "float", "vec"}, /* WINED3D_TYPE_UNKNOWN */
77 {WINED3D_DATA_UINT
, "uint", "uvec"}, /* WINED3D_TYPE_UINT */
78 {WINED3D_DATA_INT
, "int", "ivec"}, /* WINED3D_TYPE_INT */
79 {WINED3D_DATA_FLOAT
, "float", "vec"}, /* WINED3D_TYPE_FLOAT */
93 struct glsl_sample_function
95 struct wined3d_string_buffer
*name
;
96 unsigned int coord_mask
;
97 unsigned int deriv_mask
;
98 enum wined3d_data_type data_type
;
99 BOOL output_single_component
;
100 unsigned int offset_size
;
105 HEAP_NODE_TRAVERSE_LEFT
,
106 HEAP_NODE_TRAVERSE_RIGHT
,
110 struct constant_entry
113 unsigned int version
;
118 struct constant_entry
*entries
;
120 unsigned int *positions
;
124 /* GLSL shader private data */
125 struct shader_glsl_priv
127 struct wined3d_string_buffer shader_buffer
;
128 struct wined3d_string_buffer_list string_buffers
;
129 struct wine_rb_tree program_lookup
;
130 struct constant_heap vconst_heap
;
131 struct constant_heap pconst_heap
;
132 unsigned char *stack
;
133 UINT next_constant_version
;
135 const struct wined3d_vertex_pipe_ops
*vertex_pipe
;
136 const struct wined3d_fragment_pipe_ops
*fragment_pipe
;
137 struct wine_rb_tree ffp_vertex_shaders
;
138 struct wine_rb_tree ffp_fragment_shaders
;
139 BOOL ffp_proj_control
;
140 BOOL legacy_lighting
;
143 struct glsl_vs_program
145 struct list shader_entry
;
147 GLenum vertex_color_clamp
;
148 GLint uniform_f_locations
[WINED3D_MAX_VS_CONSTS_F
];
149 GLint uniform_i_locations
[WINED3D_MAX_CONSTS_I
];
150 GLint uniform_b_locations
[WINED3D_MAX_CONSTS_B
];
151 GLint pos_fixup_location
;
152 GLint base_vertex_id_location
;
154 GLint modelview_matrix_location
[MAX_VERTEX_BLENDS
];
155 GLint projection_matrix_location
;
156 GLint normal_matrix_location
;
157 GLint texture_matrix_location
[WINED3D_MAX_TEXTURES
];
158 GLint material_ambient_location
;
159 GLint material_diffuse_location
;
160 GLint material_specular_location
;
161 GLint material_emissive_location
;
162 GLint material_shininess_location
;
163 GLint light_ambient_location
;
178 } light_location
[WINED3D_MAX_ACTIVE_LIGHTS
];
179 GLint pointsize_location
;
180 GLint pointsize_min_location
;
181 GLint pointsize_max_location
;
182 GLint pointsize_c_att_location
;
183 GLint pointsize_l_att_location
;
184 GLint pointsize_q_att_location
;
185 GLint clip_planes_location
;
188 struct glsl_hs_program
190 struct list shader_entry
;
194 struct glsl_ds_program
196 struct list shader_entry
;
199 GLint pos_fixup_location
;
202 struct glsl_gs_program
204 struct list shader_entry
;
207 GLint pos_fixup_location
;
210 struct glsl_ps_program
212 struct list shader_entry
;
214 GLint uniform_f_locations
[WINED3D_MAX_PS_CONSTS_F
];
215 GLint uniform_i_locations
[WINED3D_MAX_CONSTS_I
];
216 GLint uniform_b_locations
[WINED3D_MAX_CONSTS_B
];
217 GLint bumpenv_mat_location
[WINED3D_MAX_TEXTURES
];
218 GLint bumpenv_lum_scale_location
[WINED3D_MAX_TEXTURES
];
219 GLint bumpenv_lum_offset_location
[WINED3D_MAX_TEXTURES
];
220 GLint tss_constant_location
[WINED3D_MAX_TEXTURES
];
221 GLint tex_factor_location
;
222 GLint specular_enable_location
;
223 GLint fog_color_location
;
224 GLint fog_density_location
;
225 GLint fog_end_location
;
226 GLint fog_scale_location
;
227 GLint alpha_test_ref_location
;
228 GLint ycorrection_location
;
229 GLint np2_fixup_location
;
230 GLint color_key_location
;
231 const struct ps_np2fixup_info
*np2_fixup_info
;
234 struct glsl_cs_program
236 struct list shader_entry
;
240 /* Struct to maintain data about a linked GLSL program */
241 struct glsl_shader_prog_link
243 struct wine_rb_entry program_lookup_entry
;
244 struct glsl_vs_program vs
;
245 struct glsl_hs_program hs
;
246 struct glsl_ds_program ds
;
247 struct glsl_gs_program gs
;
248 struct glsl_ps_program ps
;
249 struct glsl_cs_program cs
;
251 DWORD constant_update_mask
;
252 unsigned int constant_version
;
253 DWORD shader_controlled_clip_distances
: 1;
254 DWORD clip_distance_mask
: 8; /* WINED3D_MAX_CLIP_DISTANCES, 8 */
258 struct glsl_program_key
268 struct shader_glsl_ctx_priv
270 const struct wined3d_gl_info
*gl_info
;
271 const struct vs_compile_args
*cur_vs_args
;
272 const struct ds_compile_args
*cur_ds_args
;
273 const struct ps_compile_args
*cur_ps_args
;
274 struct ps_np2fixup_info
*cur_np2fixup_info
;
275 struct wined3d_string_buffer_list
*string_buffers
;
278 struct glsl_context_data
280 struct glsl_shader_prog_link
*glsl_program
;
281 GLenum vertex_color_clamp
;
282 BOOL rasterization_disabled
;
285 struct glsl_ps_compiled_shader
287 struct ps_compile_args args
;
288 struct ps_np2fixup_info np2fixup
;
292 struct glsl_vs_compiled_shader
294 struct vs_compile_args args
;
298 struct glsl_hs_compiled_shader
303 struct glsl_ds_compiled_shader
305 struct ds_compile_args args
;
309 struct glsl_gs_compiled_shader
311 struct gs_compile_args args
;
315 struct glsl_cs_compiled_shader
320 struct glsl_shader_private
324 struct glsl_vs_compiled_shader
*vs
;
325 struct glsl_hs_compiled_shader
*hs
;
326 struct glsl_ds_compiled_shader
*ds
;
327 struct glsl_gs_compiled_shader
*gs
;
328 struct glsl_ps_compiled_shader
*ps
;
329 struct glsl_cs_compiled_shader
*cs
;
331 unsigned int num_gl_shaders
, shader_array_size
;
334 struct glsl_ffp_vertex_shader
336 struct wined3d_ffp_vs_desc desc
;
338 struct list linked_programs
;
341 struct glsl_ffp_fragment_shader
343 struct ffp_frag_desc entry
;
345 struct list linked_programs
;
348 struct glsl_ffp_destroy_ctx
350 struct shader_glsl_priv
*priv
;
351 const struct wined3d_context_gl
*context_gl
;
354 static void shader_glsl_generate_shader_epilogue(const struct wined3d_shader_context
*ctx
);
356 static const char *debug_gl_shader_type(GLenum type
)
360 #define WINED3D_TO_STR(u) case u: return #u
361 WINED3D_TO_STR(GL_VERTEX_SHADER
);
362 WINED3D_TO_STR(GL_TESS_CONTROL_SHADER
);
363 WINED3D_TO_STR(GL_TESS_EVALUATION_SHADER
);
364 WINED3D_TO_STR(GL_GEOMETRY_SHADER
);
365 WINED3D_TO_STR(GL_FRAGMENT_SHADER
);
366 WINED3D_TO_STR(GL_COMPUTE_SHADER
);
367 #undef WINED3D_TO_STR
369 return wine_dbg_sprintf("UNKNOWN(%#x)", type
);
373 static const char *shader_glsl_get_prefix(enum wined3d_shader_type type
)
377 case WINED3D_SHADER_TYPE_VERTEX
:
380 case WINED3D_SHADER_TYPE_HULL
:
383 case WINED3D_SHADER_TYPE_DOMAIN
:
386 case WINED3D_SHADER_TYPE_GEOMETRY
:
389 case WINED3D_SHADER_TYPE_PIXEL
:
392 case WINED3D_SHADER_TYPE_COMPUTE
:
396 FIXME("Unhandled shader type %#x.\n", type
);
401 static unsigned int shader_glsl_get_version(const struct wined3d_gl_info
*gl_info
)
403 if (gl_info
->glsl_version
>= MAKEDWORD_VERSION(4, 40))
405 else if (gl_info
->glsl_version
>= MAKEDWORD_VERSION(1, 50))
407 else if (gl_info
->glsl_version
>= MAKEDWORD_VERSION(1, 30))
413 static void shader_glsl_add_version_declaration(struct wined3d_string_buffer
*buffer
,
414 const struct wined3d_gl_info
*gl_info
)
416 shader_addline(buffer
, "#version %u\n", shader_glsl_get_version(gl_info
));
419 static unsigned int shader_glsl_full_ffp_varyings(const struct wined3d_gl_info
*gl_info
)
421 /* On core profile we have to also count diffuse and specular colours and
422 * the fog coordinate. */
423 return gl_info
->limits
.glsl_varyings
>= (gl_info
->supported
[WINED3D_GL_LEGACY_CONTEXT
]
424 ? WINED3D_MAX_TEXTURES
* 4 : (WINED3D_MAX_TEXTURES
+ 2) * 4 + 1);
427 static void shader_glsl_append_imm_vec(struct wined3d_string_buffer
*buffer
,
428 const float *values
, unsigned int size
, const struct wined3d_gl_info
*gl_info
)
430 const int *int_values
= (const int *)values
;
434 if (!gl_info
->supported
[ARB_SHADER_BIT_ENCODING
])
437 shader_addline(buffer
, "vec%u(", size
);
439 for (i
= 0; i
< size
; ++i
)
441 wined3d_ftoa(values
[i
], str
);
442 shader_addline(buffer
, i
? ", %s" : "%s", str
);
446 shader_addline(buffer
, ")");
451 shader_addline(buffer
, "intBitsToFloat(");
453 shader_addline(buffer
, "ivec%u(", size
);
455 for (i
= 0; i
< size
; ++i
)
457 wined3d_ftoa(values
[i
], str
);
458 shader_addline(buffer
, i
? ", %#x /* %s */" : "%#x /* %s */", int_values
[i
], str
);
462 shader_addline(buffer
, ")");
463 shader_addline(buffer
, ")");
466 static void shader_glsl_append_imm_ivec(struct wined3d_string_buffer
*buffer
,
467 const int *values
, unsigned int size
)
471 if (!size
|| size
> 4)
473 ERR("Invalid vector size %u.\n", size
);
478 shader_addline(buffer
, "ivec%u(", size
);
480 for (i
= 0; i
< size
; ++i
)
481 shader_addline(buffer
, i
? ", %#x" : "%#x", values
[i
]);
484 shader_addline(buffer
, ")");
487 static const char *get_info_log_line(const char **ptr
)
492 if (!(q
= strstr(p
, "\n")))
494 if (!*p
) return NULL
;
503 /* Context activation is done by the caller. */
504 void print_glsl_info_log(const struct wined3d_gl_info
*gl_info
, GLuint id
, BOOL program
)
509 if (!WARN_ON(d3d_shader
) && !FIXME_ON(d3d_shader
))
513 GL_EXTCALL(glGetProgramiv(id
, GL_INFO_LOG_LENGTH
, &length
));
515 GL_EXTCALL(glGetShaderiv(id
, GL_INFO_LOG_LENGTH
, &length
));
517 /* A size of 1 is just a null-terminated string, so the log should be bigger than
518 * that if there are errors. */
521 const char *ptr
, *line
;
523 log
= heap_alloc(length
);
524 /* The info log is supposed to be zero-terminated, but at least some
525 * versions of fglrx don't terminate the string properly. The reported
526 * length does include the terminator, so explicitly set it to zero
530 GL_EXTCALL(glGetProgramInfoLog(id
, length
, NULL
, log
));
532 GL_EXTCALL(glGetShaderInfoLog(id
, length
, NULL
, log
));
535 if (gl_info
->quirks
& WINED3D_QUIRK_INFO_LOG_SPAM
)
537 WARN("Info log received from GLSL shader #%u:\n", id
);
538 while ((line
= get_info_log_line(&ptr
))) WARN(" %.*s", (int)(ptr
- line
), line
);
542 FIXME("Info log received from GLSL shader #%u:\n", id
);
543 while ((line
= get_info_log_line(&ptr
))) FIXME(" %.*s", (int)(ptr
- line
), line
);
549 /* Context activation is done by the caller. */
550 static void shader_glsl_compile(const struct wined3d_gl_info
*gl_info
, GLuint shader
, const char *src
)
552 const char *ptr
, *line
;
554 TRACE("Compiling shader object %u.\n", shader
);
556 if (TRACE_ON(d3d_shader
))
559 while ((line
= get_info_log_line(&ptr
))) TRACE_(d3d_shader
)(" %.*s", (int)(ptr
- line
), line
);
562 GL_EXTCALL(glShaderSource(shader
, 1, &src
, NULL
));
563 checkGLcall("glShaderSource");
564 GL_EXTCALL(glCompileShader(shader
));
565 checkGLcall("glCompileShader");
566 print_glsl_info_log(gl_info
, shader
, FALSE
);
569 /* Context activation is done by the caller. */
570 static void shader_glsl_dump_program_source(const struct wined3d_gl_info
*gl_info
, GLuint program
)
572 GLint i
, shader_count
, source_size
= -1;
576 GL_EXTCALL(glGetProgramiv(program
, GL_ATTACHED_SHADERS
, &shader_count
));
577 if (!(shaders
= heap_calloc(shader_count
, sizeof(*shaders
))))
579 ERR("Failed to allocate shader array memory.\n");
583 GL_EXTCALL(glGetAttachedShaders(program
, shader_count
, NULL
, shaders
));
584 for (i
= 0; i
< shader_count
; ++i
)
586 const char *ptr
, *line
;
589 GL_EXTCALL(glGetShaderiv(shaders
[i
], GL_SHADER_SOURCE_LENGTH
, &tmp
));
591 if (source_size
< tmp
)
595 if (!(source
= heap_alloc_zero(tmp
)))
597 ERR("Failed to allocate %d bytes for shader source.\n", tmp
);
604 FIXME("Shader %u:\n", shaders
[i
]);
605 GL_EXTCALL(glGetShaderiv(shaders
[i
], GL_SHADER_TYPE
, &tmp
));
606 FIXME(" GL_SHADER_TYPE: %s.\n", debug_gl_shader_type(tmp
));
607 GL_EXTCALL(glGetShaderiv(shaders
[i
], GL_COMPILE_STATUS
, &tmp
));
608 FIXME(" GL_COMPILE_STATUS: %d.\n", tmp
);
612 GL_EXTCALL(glGetShaderSource(shaders
[i
], source_size
, NULL
, source
));
613 while ((line
= get_info_log_line(&ptr
))) FIXME(" %.*s", (int)(ptr
- line
), line
);
621 /* Context activation is done by the caller. */
622 void shader_glsl_validate_link(const struct wined3d_gl_info
*gl_info
, GLuint program
)
626 if (!TRACE_ON(d3d_shader
) && !FIXME_ON(d3d_shader
))
629 GL_EXTCALL(glGetProgramiv(program
, GL_LINK_STATUS
, &tmp
));
632 FIXME("Program %u link status invalid.\n", program
);
633 shader_glsl_dump_program_source(gl_info
, program
);
636 print_glsl_info_log(gl_info
, program
, TRUE
);
639 static BOOL
shader_glsl_use_layout_qualifier(const struct wined3d_gl_info
*gl_info
)
641 /* Layout qualifiers were introduced in GLSL 1.40. The Nvidia Legacy GPU
642 * driver (series 340.xx) doesn't parse layout qualifiers in older GLSL
644 return shader_glsl_get_version(gl_info
) >= 140;
647 static BOOL
shader_glsl_use_layout_binding_qualifier(const struct wined3d_gl_info
*gl_info
)
649 return gl_info
->supported
[ARB_SHADING_LANGUAGE_420PACK
] && shader_glsl_use_layout_qualifier(gl_info
);
652 static void shader_glsl_init_uniform_block_bindings(const struct wined3d_gl_info
*gl_info
,
653 struct shader_glsl_priv
*priv
, GLuint program_id
,
654 const struct wined3d_shader_reg_maps
*reg_maps
)
656 const char *prefix
= shader_glsl_get_prefix(reg_maps
->shader_version
.type
);
657 struct wined3d_string_buffer
*name
;
658 unsigned int i
, base
, count
;
661 if (shader_glsl_use_layout_binding_qualifier(gl_info
))
664 name
= string_buffer_get(&priv
->string_buffers
);
665 wined3d_gl_limits_get_uniform_block_range(&gl_info
->limits
, reg_maps
->shader_version
.type
, &base
, &count
);
666 for (i
= 0; i
< count
; ++i
)
668 if (!reg_maps
->cb_sizes
[i
])
671 string_buffer_sprintf(name
, "block_%s_cb%u", prefix
, i
);
672 block_idx
= GL_EXTCALL(glGetUniformBlockIndex(program_id
, name
->buffer
));
673 GL_EXTCALL(glUniformBlockBinding(program_id
, block_idx
, base
+ i
));
675 checkGLcall("glUniformBlockBinding");
676 string_buffer_release(&priv
->string_buffers
, name
);
679 /* Context activation is done by the caller. */
680 static void shader_glsl_load_samplers_range(const struct wined3d_gl_info
*gl_info
,
681 struct shader_glsl_priv
*priv
, GLuint program_id
, const char *prefix
,
682 unsigned int base
, unsigned int count
, const DWORD
*tex_unit_map
)
684 struct wined3d_string_buffer
*sampler_name
= string_buffer_get(&priv
->string_buffers
);
685 unsigned int i
, mapped_unit
;
688 for (i
= 0; i
< count
; ++i
)
690 string_buffer_sprintf(sampler_name
, "%s_sampler%u", prefix
, i
);
691 name_loc
= GL_EXTCALL(glGetUniformLocation(program_id
, sampler_name
->buffer
));
695 mapped_unit
= tex_unit_map
? tex_unit_map
[base
+ i
] : base
+ i
;
696 if (mapped_unit
== WINED3D_UNMAPPED_STAGE
|| mapped_unit
>= gl_info
->limits
.combined_samplers
)
698 ERR("Trying to load sampler %s on unsupported unit %u.\n", sampler_name
->buffer
, mapped_unit
);
702 TRACE("Loading sampler %s on unit %u.\n", sampler_name
->buffer
, mapped_unit
);
703 GL_EXTCALL(glUniform1i(name_loc
, mapped_unit
));
705 checkGLcall("Load sampler bindings");
706 string_buffer_release(&priv
->string_buffers
, sampler_name
);
709 static unsigned int shader_glsl_map_tex_unit(const struct wined3d_context
*context
,
710 const struct wined3d_shader_version
*shader_version
, unsigned int sampler_idx
)
712 const struct wined3d_context_gl
*context_gl
= wined3d_context_gl_const(context
);
713 const unsigned int *tex_unit_map
;
714 unsigned int base
, count
;
716 tex_unit_map
= wined3d_context_gl_get_tex_unit_mapping(context_gl
, shader_version
, &base
, &count
);
717 if (sampler_idx
>= count
)
718 return WINED3D_UNMAPPED_STAGE
;
720 return base
+ sampler_idx
;
721 return tex_unit_map
[base
+ sampler_idx
];
724 static void shader_glsl_append_sampler_binding_qualifier(struct wined3d_string_buffer
*buffer
,
725 const struct wined3d_context
*context
, const struct wined3d_shader_version
*shader_version
,
726 unsigned int sampler_idx
)
728 unsigned int mapped_unit
= shader_glsl_map_tex_unit(context
, shader_version
, sampler_idx
);
729 if (mapped_unit
!= WINED3D_UNMAPPED_STAGE
)
730 shader_addline(buffer
, "layout(binding = %u)\n", mapped_unit
);
732 ERR("Unmapped sampler %u.\n", sampler_idx
);
735 /* Context activation is done by the caller. */
736 static void shader_glsl_load_samplers(const struct wined3d_context
*context
,
737 struct shader_glsl_priv
*priv
, GLuint program_id
, const struct wined3d_shader_reg_maps
*reg_maps
)
739 const struct wined3d_context_gl
*context_gl
= wined3d_context_gl_const(context
);
740 const struct wined3d_gl_info
*gl_info
= context_gl
->gl_info
;
741 const struct wined3d_shader_version
*shader_version
;
742 const unsigned int *tex_unit_map
;
743 unsigned int base
, count
;
746 if (shader_glsl_use_layout_binding_qualifier(gl_info
))
749 shader_version
= reg_maps
? ®_maps
->shader_version
: NULL
;
750 prefix
= shader_glsl_get_prefix(shader_version
? shader_version
->type
: WINED3D_SHADER_TYPE_PIXEL
);
751 tex_unit_map
= wined3d_context_gl_get_tex_unit_mapping(context_gl
, shader_version
, &base
, &count
);
752 shader_glsl_load_samplers_range(gl_info
, priv
, program_id
, prefix
, base
, count
, tex_unit_map
);
755 static void shader_glsl_load_icb(const struct wined3d_gl_info
*gl_info
, struct shader_glsl_priv
*priv
,
756 GLuint program_id
, const struct wined3d_shader_reg_maps
*reg_maps
)
758 const struct wined3d_shader_immediate_constant_buffer
*icb
= reg_maps
->icb
;
762 struct wined3d_string_buffer
*icb_name
= string_buffer_get(&priv
->string_buffers
);
763 const char *prefix
= shader_glsl_get_prefix(reg_maps
->shader_version
.type
);
766 string_buffer_sprintf(icb_name
, "%s_icb", prefix
);
767 icb_location
= GL_EXTCALL(glGetUniformLocation(program_id
, icb_name
->buffer
));
768 GL_EXTCALL(glUniform4fv(icb_location
, icb
->vec4_count
, (const GLfloat
*)icb
->data
));
769 checkGLcall("Load immediate constant buffer");
771 string_buffer_release(&priv
->string_buffers
, icb_name
);
775 /* Context activation is done by the caller. */
776 static void shader_glsl_load_images(const struct wined3d_gl_info
*gl_info
, struct shader_glsl_priv
*priv
,
777 GLuint program_id
, const struct wined3d_shader_reg_maps
*reg_maps
)
779 const char *prefix
= shader_glsl_get_prefix(reg_maps
->shader_version
.type
);
780 struct wined3d_string_buffer
*name
;
784 if (shader_glsl_use_layout_binding_qualifier(gl_info
))
787 name
= string_buffer_get(&priv
->string_buffers
);
788 for (i
= 0; i
< MAX_UNORDERED_ACCESS_VIEWS
; ++i
)
790 if (!reg_maps
->uav_resource_info
[i
].type
)
793 string_buffer_sprintf(name
, "%s_image%u", prefix
, i
);
794 location
= GL_EXTCALL(glGetUniformLocation(program_id
, name
->buffer
));
798 TRACE("Loading image %s on unit %u.\n", name
->buffer
, i
);
799 GL_EXTCALL(glUniform1i(location
, i
));
801 checkGLcall("Load image bindings");
802 string_buffer_release(&priv
->string_buffers
, name
);
805 /* Context activation is done by the caller. */
806 static void shader_glsl_load_program_resources(const struct wined3d_context_gl
*context_gl
,
807 struct shader_glsl_priv
*priv
, GLuint program_id
, const struct wined3d_shader
*shader
)
809 const struct wined3d_shader_reg_maps
*reg_maps
= &shader
->reg_maps
;
811 shader_glsl_init_uniform_block_bindings(context_gl
->gl_info
, priv
, program_id
, reg_maps
);
812 shader_glsl_load_icb(context_gl
->gl_info
, priv
, program_id
, reg_maps
);
813 /* Texture unit mapping is set up to be the same each time the shader
814 * program is used so we can hardcode the sampler uniform values. */
815 shader_glsl_load_samplers(&context_gl
->c
, priv
, program_id
, reg_maps
);
818 static void append_transform_feedback_varying(const char **varyings
, unsigned int *varying_count
,
819 char **strings
, unsigned int *strings_length
, struct wined3d_string_buffer
*buffer
)
821 if (varyings
&& *strings
)
823 char *ptr
= *strings
;
825 varyings
[*varying_count
] = ptr
;
827 memcpy(ptr
, buffer
->buffer
, buffer
->content_size
+ 1);
828 ptr
+= buffer
->content_size
+ 1;
833 *strings_length
+= buffer
->content_size
+ 1;
837 static void append_transform_feedback_skip_components(const char **varyings
,
838 unsigned int *varying_count
, char **strings
, unsigned int *strings_length
,
839 struct wined3d_string_buffer
*buffer
, unsigned int component_count
)
843 for (j
= 0; j
< component_count
/ 4; ++j
)
845 string_buffer_sprintf(buffer
, "gl_SkipComponents4");
846 append_transform_feedback_varying(varyings
, varying_count
, strings
, strings_length
, buffer
);
848 if (component_count
% 4)
850 string_buffer_sprintf(buffer
, "gl_SkipComponents%u", component_count
% 4);
851 append_transform_feedback_varying(varyings
, varying_count
, strings
, strings_length
, buffer
);
855 static BOOL
shader_glsl_generate_transform_feedback_varyings(struct wined3d_string_buffer
*buffer
,
856 const char **varyings
, unsigned int *varying_count
, char *strings
, unsigned int *strings_length
,
857 GLenum buffer_mode
, struct wined3d_shader
*shader
)
859 const struct wined3d_stream_output_desc
*so_desc
= shader
->u
.gs
.so_desc
;
860 unsigned int buffer_idx
, count
, length
, highest_output_slot
, stride
;
861 unsigned int i
, register_idx
, component_idx
;
862 BOOL have_varyings_to_record
= FALSE
;
865 highest_output_slot
= 0;
866 for (buffer_idx
= 0; buffer_idx
< WINED3D_MAX_STREAM_OUTPUT_BUFFERS
; ++buffer_idx
)
870 for (i
= 0; i
< so_desc
->element_count
; ++i
)
872 const struct wined3d_stream_output_element
*e
= &so_desc
->elements
[i
];
874 highest_output_slot
= max(highest_output_slot
, e
->output_slot
);
875 if (e
->output_slot
!= buffer_idx
)
880 FIXME("Unhandled stream %u.\n", e
->stream_idx
);
884 stride
+= e
->component_count
;
886 if (!e
->semantic_name
)
888 append_transform_feedback_skip_components(varyings
, &count
,
889 &strings
, &length
, buffer
, e
->component_count
);
893 if (!shader_get_stream_output_register_info(shader
, e
, ®ister_idx
, &component_idx
))
896 if (component_idx
|| e
->component_count
!= 4)
898 if (so_desc
->rasterizer_stream_idx
!= WINED3D_NO_RASTERIZER_STREAM
)
900 FIXME("Unsupported component range %u-%u.\n", component_idx
, e
->component_count
);
901 append_transform_feedback_skip_components(varyings
, &count
,
902 &strings
, &length
, buffer
, e
->component_count
);
906 string_buffer_sprintf(buffer
, "shader_in_out.reg%u_%u_%u",
907 register_idx
, component_idx
, component_idx
+ e
->component_count
- 1);
908 append_transform_feedback_varying(varyings
, &count
, &strings
, &length
, buffer
);
912 string_buffer_sprintf(buffer
, "shader_in_out.reg%u", register_idx
);
913 append_transform_feedback_varying(varyings
, &count
, &strings
, &length
, buffer
);
916 have_varyings_to_record
= TRUE
;
919 if (buffer_idx
< so_desc
->buffer_stride_count
920 && stride
< so_desc
->buffer_strides
[buffer_idx
] / 4)
922 unsigned int component_count
= so_desc
->buffer_strides
[buffer_idx
] / 4 - stride
;
923 append_transform_feedback_skip_components(varyings
, &count
,
924 &strings
, &length
, buffer
, component_count
);
927 if (highest_output_slot
<= buffer_idx
)
930 if (buffer_mode
== GL_INTERLEAVED_ATTRIBS
)
932 string_buffer_sprintf(buffer
, "gl_NextBuffer");
933 append_transform_feedback_varying(varyings
, &count
, &strings
, &length
, buffer
);
938 *varying_count
= count
;
940 *strings_length
= length
;
942 return have_varyings_to_record
;
945 static void shader_glsl_init_transform_feedback(const struct wined3d_context_gl
*context_gl
,
946 struct shader_glsl_priv
*priv
, GLuint program_id
, struct wined3d_shader
*shader
)
948 const struct wined3d_stream_output_desc
*so_desc
= shader
->u
.gs
.so_desc
;
949 const struct wined3d_gl_info
*gl_info
= context_gl
->gl_info
;
950 struct wined3d_string_buffer
*buffer
;
951 unsigned int i
, count
, length
;
952 const char **varyings
;
959 if (gl_info
->supported
[ARB_TRANSFORM_FEEDBACK3
])
961 mode
= GL_INTERLEAVED_ATTRIBS
;
965 unsigned int element_count
[WINED3D_MAX_STREAM_OUTPUT_BUFFERS
] = {0};
967 for (i
= 0; i
< so_desc
->element_count
; ++i
)
969 if (!so_desc
->elements
[i
].semantic_name
)
971 FIXME("ARB_transform_feedback3 is needed for stream output gaps.\n");
974 ++element_count
[so_desc
->elements
[i
].output_slot
];
977 if (element_count
[0] == so_desc
->element_count
)
979 mode
= GL_INTERLEAVED_ATTRIBS
;
983 mode
= GL_SEPARATE_ATTRIBS
;
984 for (i
= 0; i
< ARRAY_SIZE(element_count
); ++i
)
986 if (element_count
[i
] != 1)
989 for (; i
< ARRAY_SIZE(element_count
); ++i
)
991 if (element_count
[i
])
993 FIXME("Only single element per buffer is allowed in separate mode.\n");
1000 buffer
= string_buffer_get(&priv
->string_buffers
);
1002 if (!shader_glsl_generate_transform_feedback_varyings(buffer
, NULL
, &count
, NULL
, &length
, mode
, shader
))
1004 FIXME("No varyings to record, disabling transform feedback.\n");
1005 shader
->u
.gs
.so_desc
= NULL
;
1006 string_buffer_release(&priv
->string_buffers
, buffer
);
1010 if (!(varyings
= heap_calloc(count
, sizeof(*varyings
))))
1012 ERR("Out of memory.\n");
1013 string_buffer_release(&priv
->string_buffers
, buffer
);
1016 if (!(strings
= heap_calloc(length
, sizeof(*strings
))))
1018 ERR("Out of memory.\n");
1019 heap_free(varyings
);
1020 string_buffer_release(&priv
->string_buffers
, buffer
);
1024 shader_glsl_generate_transform_feedback_varyings(buffer
, varyings
, NULL
, strings
, NULL
, mode
, shader
);
1025 GL_EXTCALL(glTransformFeedbackVaryings(program_id
, count
, varyings
, mode
));
1026 checkGLcall("glTransformFeedbackVaryings");
1028 heap_free(varyings
);
1030 string_buffer_release(&priv
->string_buffers
, buffer
);
1033 /* Context activation is done by the caller. */
1034 static void walk_constant_heap(const struct wined3d_gl_info
*gl_info
, const struct wined3d_vec4
*constants
,
1035 const GLint
*constant_locations
, const struct constant_heap
*heap
, unsigned char *stack
, DWORD version
)
1037 unsigned int start
= ~0U, end
= 0;
1039 unsigned int heap_idx
= 1;
1042 if (heap
->entries
[heap_idx
].version
<= version
) return;
1044 idx
= heap
->entries
[heap_idx
].idx
;
1045 if (constant_locations
[idx
] != -1)
1047 stack
[stack_idx
] = HEAP_NODE_TRAVERSE_LEFT
;
1049 while (stack_idx
>= 0)
1051 /* Note that we fall through to the next case statement. */
1052 switch(stack
[stack_idx
])
1054 case HEAP_NODE_TRAVERSE_LEFT
:
1056 unsigned int left_idx
= heap_idx
<< 1;
1057 if (left_idx
< heap
->size
&& heap
->entries
[left_idx
].version
> version
)
1059 heap_idx
= left_idx
;
1060 idx
= heap
->entries
[heap_idx
].idx
;
1061 if (constant_locations
[idx
] != -1)
1069 stack
[stack_idx
++] = HEAP_NODE_TRAVERSE_RIGHT
;
1070 stack
[stack_idx
] = HEAP_NODE_TRAVERSE_LEFT
;
1075 case HEAP_NODE_TRAVERSE_RIGHT
:
1077 unsigned int right_idx
= (heap_idx
<< 1) + 1;
1078 if (right_idx
< heap
->size
&& heap
->entries
[right_idx
].version
> version
)
1080 heap_idx
= right_idx
;
1081 idx
= heap
->entries
[heap_idx
].idx
;
1082 if (constant_locations
[idx
] != -1)
1090 stack
[stack_idx
++] = HEAP_NODE_POP
;
1091 stack
[stack_idx
] = HEAP_NODE_TRAVERSE_LEFT
;
1103 GL_EXTCALL(glUniform4fv(constant_locations
[start
], end
- start
+ 1, &constants
[start
].x
));
1104 checkGLcall("walk_constant_heap()");
1107 /* Context activation is done by the caller. */
1108 static void apply_clamped_constant(const struct wined3d_gl_info
*gl_info
,
1109 GLint location
, const struct wined3d_vec4
*data
)
1111 GLfloat clamped_constant
[4];
1113 if (location
== -1) return;
1115 clamped_constant
[0] = data
->x
< -1.0f
? -1.0f
: data
->x
> 1.0f
? 1.0f
: data
->x
;
1116 clamped_constant
[1] = data
->y
< -1.0f
? -1.0f
: data
->y
> 1.0f
? 1.0f
: data
->y
;
1117 clamped_constant
[2] = data
->z
< -1.0f
? -1.0f
: data
->z
> 1.0f
? 1.0f
: data
->z
;
1118 clamped_constant
[3] = data
->w
< -1.0f
? -1.0f
: data
->w
> 1.0f
? 1.0f
: data
->w
;
1120 GL_EXTCALL(glUniform4fv(location
, 1, clamped_constant
));
1123 /* Context activation is done by the caller. */
1124 static void walk_constant_heap_clamped(const struct wined3d_gl_info
*gl_info
,
1125 const struct wined3d_vec4
*constants
, const GLint
*constant_locations
,
1126 const struct constant_heap
*heap
, unsigned char *stack
, DWORD version
)
1129 unsigned int heap_idx
= 1;
1132 if (heap
->entries
[heap_idx
].version
<= version
) return;
1134 idx
= heap
->entries
[heap_idx
].idx
;
1135 apply_clamped_constant(gl_info
, constant_locations
[idx
], &constants
[idx
]);
1136 stack
[stack_idx
] = HEAP_NODE_TRAVERSE_LEFT
;
1138 while (stack_idx
>= 0)
1140 /* Note that we fall through to the next case statement. */
1141 switch(stack
[stack_idx
])
1143 case HEAP_NODE_TRAVERSE_LEFT
:
1145 unsigned int left_idx
= heap_idx
<< 1;
1146 if (left_idx
< heap
->size
&& heap
->entries
[left_idx
].version
> version
)
1148 heap_idx
= left_idx
;
1149 idx
= heap
->entries
[heap_idx
].idx
;
1150 apply_clamped_constant(gl_info
, constant_locations
[idx
], &constants
[idx
]);
1152 stack
[stack_idx
++] = HEAP_NODE_TRAVERSE_RIGHT
;
1153 stack
[stack_idx
] = HEAP_NODE_TRAVERSE_LEFT
;
1158 case HEAP_NODE_TRAVERSE_RIGHT
:
1160 unsigned int right_idx
= (heap_idx
<< 1) + 1;
1161 if (right_idx
< heap
->size
&& heap
->entries
[right_idx
].version
> version
)
1163 heap_idx
= right_idx
;
1164 idx
= heap
->entries
[heap_idx
].idx
;
1165 apply_clamped_constant(gl_info
, constant_locations
[idx
], &constants
[idx
]);
1167 stack
[stack_idx
++] = HEAP_NODE_POP
;
1168 stack
[stack_idx
] = HEAP_NODE_TRAVERSE_LEFT
;
1179 checkGLcall("walk_constant_heap_clamped()");
1182 /* Context activation is done by the caller. */
1183 static void shader_glsl_load_constants_f(const struct wined3d_shader
*shader
, const struct wined3d_gl_info
*gl_info
,
1184 const struct wined3d_vec4
*constants
, const GLint
*constant_locations
, const struct constant_heap
*heap
,
1185 unsigned char *stack
, unsigned int version
)
1187 const struct wined3d_shader_lconst
*lconst
;
1189 /* 1.X pshaders have the constants clamped to [-1;1] implicitly. */
1190 if (shader
->reg_maps
.shader_version
.major
== 1
1191 && shader
->reg_maps
.shader_version
.type
== WINED3D_SHADER_TYPE_PIXEL
)
1192 walk_constant_heap_clamped(gl_info
, constants
, constant_locations
, heap
, stack
, version
);
1194 walk_constant_heap(gl_info
, constants
, constant_locations
, heap
, stack
, version
);
1196 if (!shader
->load_local_constsF
)
1198 TRACE("No need to load local float constants for this shader.\n");
1202 /* Immediate constants are clamped to [-1;1] at shader creation time if needed */
1203 LIST_FOR_EACH_ENTRY(lconst
, &shader
->constantsF
, struct wined3d_shader_lconst
, entry
)
1205 GL_EXTCALL(glUniform4fv(constant_locations
[lconst
->idx
], 1, (const GLfloat
*)lconst
->value
));
1207 checkGLcall("glUniform4fv()");
1210 /* Context activation is done by the caller. */
1211 static void shader_glsl_load_constants_i(const struct wined3d_shader
*shader
, const struct wined3d_gl_info
*gl_info
,
1212 const struct wined3d_ivec4
*constants
, const GLint locations
[WINED3D_MAX_CONSTS_I
], uint32_t constants_set
)
1217 while (constants_set
)
1219 /* We found this uniform name in the program - go ahead and send the data */
1220 i
= wined3d_bit_scan(&constants_set
);
1221 GL_EXTCALL(glUniform4iv(locations
[i
], 1, &constants
[i
].x
));
1224 /* Load immediate constants */
1225 ptr
= list_head(&shader
->constantsI
);
1228 const struct wined3d_shader_lconst
*lconst
= LIST_ENTRY(ptr
, const struct wined3d_shader_lconst
, entry
);
1229 unsigned int idx
= lconst
->idx
;
1230 const GLint
*values
= (const GLint
*)lconst
->value
;
1232 /* We found this uniform name in the program - go ahead and send the data */
1233 GL_EXTCALL(glUniform4iv(locations
[idx
], 1, values
));
1234 ptr
= list_next(&shader
->constantsI
, ptr
);
1236 checkGLcall("glUniform4iv()");
1239 /* Context activation is done by the caller. */
1240 static void shader_glsl_load_constants_b(const struct wined3d_shader
*shader
, const struct wined3d_gl_info
*gl_info
,
1241 const BOOL
*constants
, const GLint locations
[WINED3D_MAX_CONSTS_B
], uint32_t constants_set
)
1246 while (constants_set
)
1248 i
= wined3d_bit_scan(&constants_set
);
1249 GL_EXTCALL(glUniform1iv(locations
[i
], 1, &constants
[i
]));
1252 /* Load immediate constants */
1253 ptr
= list_head(&shader
->constantsB
);
1256 const struct wined3d_shader_lconst
*lconst
= LIST_ENTRY(ptr
, const struct wined3d_shader_lconst
, entry
);
1257 unsigned int idx
= lconst
->idx
;
1258 const GLint
*values
= (const GLint
*)lconst
->value
;
1260 GL_EXTCALL(glUniform1iv(locations
[idx
], 1, values
));
1261 ptr
= list_next(&shader
->constantsB
, ptr
);
1263 checkGLcall("glUniform1iv()");
1266 static void reset_program_constant_version(struct wine_rb_entry
*entry
, void *context
)
1268 WINE_RB_ENTRY_VALUE(entry
, struct glsl_shader_prog_link
, program_lookup_entry
)->constant_version
= 0;
1271 /* Context activation is done by the caller (state handler). */
1272 static void shader_glsl_load_np2fixup_constants(const struct glsl_ps_program
*ps
,
1273 const struct wined3d_gl_info
*gl_info
, const struct wined3d_state
*state
)
1279 np2fixup_constants
[WINED3D_MAX_FRAGMENT_SAMPLERS
];
1280 UINT fixup
= ps
->np2_fixup_info
->active
;
1283 for (i
= 0; fixup
; fixup
>>= 1, ++i
)
1285 const struct wined3d_texture
*tex
= state
->textures
[i
];
1286 unsigned char idx
= ps
->np2_fixup_info
->idx
[i
];
1290 ERR("Nonexistent texture is flagged for NP2 texcoord fixup.\n");
1294 np2fixup_constants
[idx
].sx
= tex
->pow2_matrix
[0];
1295 np2fixup_constants
[idx
].sy
= tex
->pow2_matrix
[5];
1298 GL_EXTCALL(glUniform4fv(ps
->np2_fixup_location
, ps
->np2_fixup_info
->num_consts
, &np2fixup_constants
[0].sx
));
1301 static void transpose_matrix(struct wined3d_matrix
*out
, const struct wined3d_matrix
*m
)
1303 struct wined3d_matrix temp
;
1306 for (i
= 0; i
< 4; ++i
)
1307 for (j
= 0; j
< 4; ++j
)
1308 (&temp
._11
)[4 * j
+ i
] = (&m
->_11
)[4 * i
+ j
];
1313 static void shader_glsl_ffp_vertex_normalmatrix_uniform(const struct wined3d_context_gl
*context_gl
,
1314 const struct wined3d_state
*state
, struct glsl_shader_prog_link
*prog
)
1316 const struct wined3d_gl_info
*gl_info
= context_gl
->gl_info
;
1317 struct wined3d_matrix mv
;
1320 if (prog
->vs
.normal_matrix_location
== -1)
1323 get_modelview_matrix(&context_gl
->c
, state
, 0, &mv
);
1324 compute_normal_matrix(mat
, context_gl
->c
.d3d_info
->wined3d_creation_flags
& WINED3D_LEGACY_FFP_LIGHTING
, &mv
);
1326 GL_EXTCALL(glUniformMatrix3fv(prog
->vs
.normal_matrix_location
, 1, FALSE
, mat
));
1327 checkGLcall("glUniformMatrix3fv");
1330 static void shader_glsl_ffp_vertex_texmatrix_uniform(const struct wined3d_context_gl
*context_gl
,
1331 const struct wined3d_state
*state
, unsigned int tex
, struct glsl_shader_prog_link
*prog
)
1333 const struct wined3d_gl_info
*gl_info
= context_gl
->gl_info
;
1334 struct wined3d_matrix mat
;
1336 if (tex
>= WINED3D_MAX_TEXTURES
)
1338 if (prog
->vs
.texture_matrix_location
[tex
] == -1)
1341 get_texture_matrix(&context_gl
->c
, state
, tex
, &mat
);
1342 GL_EXTCALL(glUniformMatrix4fv(prog
->vs
.texture_matrix_location
[tex
], 1, FALSE
, &mat
._11
));
1343 checkGLcall("glUniformMatrix4fv");
1346 static void shader_glsl_ffp_vertex_material_uniform(const struct wined3d_context_gl
*context_gl
,
1347 const struct wined3d_state
*state
, struct glsl_shader_prog_link
*prog
)
1349 const struct wined3d_gl_info
*gl_info
= context_gl
->gl_info
;
1351 if (state
->render_states
[WINED3D_RS_SPECULARENABLE
])
1353 GL_EXTCALL(glUniform4fv(prog
->vs
.material_specular_location
, 1, &state
->material
.specular
.r
));
1354 GL_EXTCALL(glUniform1f(prog
->vs
.material_shininess_location
, state
->material
.power
));
1358 static const float black
[] = {0.0f
, 0.0f
, 0.0f
, 0.0f
};
1360 GL_EXTCALL(glUniform4fv(prog
->vs
.material_specular_location
, 1, black
));
1362 GL_EXTCALL(glUniform4fv(prog
->vs
.material_ambient_location
, 1, &state
->material
.ambient
.r
));
1363 GL_EXTCALL(glUniform4fv(prog
->vs
.material_diffuse_location
, 1, &state
->material
.diffuse
.r
));
1364 GL_EXTCALL(glUniform4fv(prog
->vs
.material_emissive_location
, 1, &state
->material
.emissive
.r
));
1365 checkGLcall("setting FFP material uniforms");
1368 static void shader_glsl_ffp_vertex_lightambient_uniform(const struct wined3d_context_gl
*context_gl
,
1369 const struct wined3d_state
*state
, struct glsl_shader_prog_link
*prog
)
1371 const struct wined3d_gl_info
*gl_info
= context_gl
->gl_info
;
1372 struct wined3d_color color
;
1374 wined3d_color_from_d3dcolor(&color
, state
->render_states
[WINED3D_RS_AMBIENT
]);
1375 GL_EXTCALL(glUniform3fv(prog
->vs
.light_ambient_location
, 1, &color
.r
));
1376 checkGLcall("glUniform3fv");
1379 static void shader_glsl_ffp_vertex_light_uniform(const struct wined3d_context_gl
*context_gl
,
1380 const struct wined3d_state
*state
, unsigned int light
, const struct wined3d_light_info
*light_info
,
1381 struct glsl_shader_prog_link
*prog
)
1383 const struct wined3d_matrix
*view
= &state
->transforms
[WINED3D_TS_VIEW
];
1384 const struct wined3d_gl_info
*gl_info
= context_gl
->gl_info
;
1385 struct wined3d_vec4 vec4
;
1387 GL_EXTCALL(glUniform4fv(prog
->vs
.light_location
[light
].diffuse
, 1, &light_info
->OriginalParms
.diffuse
.r
));
1388 GL_EXTCALL(glUniform4fv(prog
->vs
.light_location
[light
].specular
, 1, &light_info
->OriginalParms
.specular
.r
));
1389 GL_EXTCALL(glUniform4fv(prog
->vs
.light_location
[light
].ambient
, 1, &light_info
->OriginalParms
.ambient
.r
));
1391 switch (light_info
->OriginalParms
.type
)
1393 case WINED3D_LIGHT_POINT
:
1394 wined3d_vec4_transform(&vec4
, &light_info
->position
, view
);
1395 GL_EXTCALL(glUniform4fv(prog
->vs
.light_location
[light
].position
, 1, &vec4
.x
));
1396 GL_EXTCALL(glUniform1f(prog
->vs
.light_location
[light
].range
, light_info
->OriginalParms
.range
));
1397 GL_EXTCALL(glUniform1f(prog
->vs
.light_location
[light
].c_att
, light_info
->OriginalParms
.attenuation0
));
1398 GL_EXTCALL(glUniform1f(prog
->vs
.light_location
[light
].l_att
, light_info
->OriginalParms
.attenuation1
));
1399 GL_EXTCALL(glUniform1f(prog
->vs
.light_location
[light
].q_att
, light_info
->OriginalParms
.attenuation2
));
1402 case WINED3D_LIGHT_SPOT
:
1403 wined3d_vec4_transform(&vec4
, &light_info
->position
, view
);
1404 GL_EXTCALL(glUniform4fv(prog
->vs
.light_location
[light
].position
, 1, &vec4
.x
));
1406 wined3d_vec4_transform(&vec4
, &light_info
->direction
, view
);
1407 GL_EXTCALL(glUniform3fv(prog
->vs
.light_location
[light
].direction
, 1, &vec4
.x
));
1409 GL_EXTCALL(glUniform1f(prog
->vs
.light_location
[light
].range
, light_info
->OriginalParms
.range
));
1410 GL_EXTCALL(glUniform1f(prog
->vs
.light_location
[light
].falloff
, light_info
->OriginalParms
.falloff
));
1411 GL_EXTCALL(glUniform1f(prog
->vs
.light_location
[light
].c_att
, light_info
->OriginalParms
.attenuation0
));
1412 GL_EXTCALL(glUniform1f(prog
->vs
.light_location
[light
].l_att
, light_info
->OriginalParms
.attenuation1
));
1413 GL_EXTCALL(glUniform1f(prog
->vs
.light_location
[light
].q_att
, light_info
->OriginalParms
.attenuation2
));
1414 GL_EXTCALL(glUniform1f(prog
->vs
.light_location
[light
].cos_htheta
, cosf(light_info
->OriginalParms
.theta
/ 2.0f
)));
1415 GL_EXTCALL(glUniform1f(prog
->vs
.light_location
[light
].cos_hphi
, cosf(light_info
->OriginalParms
.phi
/ 2.0f
)));
1418 case WINED3D_LIGHT_DIRECTIONAL
:
1419 wined3d_vec4_transform(&vec4
, &light_info
->direction
, view
);
1420 GL_EXTCALL(glUniform3fv(prog
->vs
.light_location
[light
].direction
, 1, &vec4
.x
));
1423 case WINED3D_LIGHT_PARALLELPOINT
:
1424 wined3d_vec4_transform(&vec4
, &light_info
->position
, view
);
1425 GL_EXTCALL(glUniform4fv(prog
->vs
.light_location
[light
].position
, 1, &vec4
.x
));
1429 FIXME("Unrecognized light type %#x.\n", light_info
->OriginalParms
.type
);
1431 checkGLcall("setting FFP lights uniforms");
1434 static void shader_glsl_pointsize_uniform(const struct wined3d_context_gl
*context_gl
,
1435 const struct wined3d_state
*state
, struct glsl_shader_prog_link
*prog
)
1437 const struct wined3d_gl_info
*gl_info
= context_gl
->gl_info
;
1441 get_pointsize_minmax(&context_gl
->c
, state
, &min
, &max
);
1443 GL_EXTCALL(glUniform1f(prog
->vs
.pointsize_min_location
, min
));
1444 checkGLcall("glUniform1f");
1445 GL_EXTCALL(glUniform1f(prog
->vs
.pointsize_max_location
, max
));
1446 checkGLcall("glUniform1f");
1448 get_pointsize(&context_gl
->c
, state
, &size
, att
);
1450 GL_EXTCALL(glUniform1f(prog
->vs
.pointsize_location
, size
));
1451 checkGLcall("glUniform1f");
1452 GL_EXTCALL(glUniform1f(prog
->vs
.pointsize_c_att_location
, att
[0]));
1453 checkGLcall("glUniform1f");
1454 GL_EXTCALL(glUniform1f(prog
->vs
.pointsize_l_att_location
, att
[1]));
1455 checkGLcall("glUniform1f");
1456 GL_EXTCALL(glUniform1f(prog
->vs
.pointsize_q_att_location
, att
[2]));
1457 checkGLcall("glUniform1f");
1460 static void shader_glsl_load_fog_uniform(const struct wined3d_context_gl
*context_gl
,
1461 const struct wined3d_state
*state
, struct glsl_shader_prog_link
*prog
)
1463 const struct wined3d_gl_info
*gl_info
= context_gl
->gl_info
;
1464 struct wined3d_color color
;
1465 float start
, end
, scale
;
1472 wined3d_color_from_d3dcolor(&color
, state
->render_states
[WINED3D_RS_FOGCOLOR
]);
1473 GL_EXTCALL(glUniform4fv(prog
->ps
.fog_color_location
, 1, &color
.r
));
1474 tmpvalue
.d
= state
->render_states
[WINED3D_RS_FOGDENSITY
];
1475 GL_EXTCALL(glUniform1f(prog
->ps
.fog_density_location
, tmpvalue
.f
));
1476 get_fog_start_end(&context_gl
->c
, state
, &start
, &end
);
1477 scale
= 1.0f
/ (end
- start
);
1478 GL_EXTCALL(glUniform1f(prog
->ps
.fog_end_location
, end
));
1479 GL_EXTCALL(glUniform1f(prog
->ps
.fog_scale_location
, scale
));
1480 checkGLcall("fog emulation uniforms");
1483 static void shader_glsl_clip_plane_uniform(const struct wined3d_context_gl
*context_gl
,
1484 const struct wined3d_state
*state
, unsigned int index
, struct glsl_shader_prog_link
*prog
)
1486 const struct wined3d_gl_info
*gl_info
= context_gl
->gl_info
;
1487 struct wined3d_matrix matrix
;
1488 struct wined3d_vec4 plane
;
1490 plane
= state
->clip_planes
[index
];
1492 /* Clip planes are affected by the view transform in d3d for FFP draws. */
1495 invert_matrix(&matrix
, &state
->transforms
[WINED3D_TS_VIEW
]);
1496 transpose_matrix(&matrix
, &matrix
);
1497 wined3d_vec4_transform(&plane
, &plane
, &matrix
);
1500 GL_EXTCALL(glUniform4fv(prog
->vs
.clip_planes_location
+ index
, 1, &plane
.x
));
1503 /* Context activation is done by the caller (state handler). */
1504 static void shader_glsl_load_color_key_constant(const struct glsl_ps_program
*ps
,
1505 const struct wined3d_gl_info
*gl_info
, const struct wined3d_state
*state
)
1507 struct wined3d_color float_key
[2];
1508 const struct wined3d_texture
*texture
= state
->textures
[0];
1510 wined3d_format_get_float_color_key(texture
->resource
.format
, &texture
->async
.src_blt_color_key
, float_key
);
1511 GL_EXTCALL(glUniform4fv(ps
->color_key_location
, 2, &float_key
[0].r
));
1514 /* Context activation is done by the caller (state handler). */
1515 static void shader_glsl_load_constants(void *shader_priv
, struct wined3d_context
*context
,
1516 const struct wined3d_state
*state
)
1518 const struct wined3d_shader
*vshader
= state
->shader
[WINED3D_SHADER_TYPE_VERTEX
];
1519 const struct wined3d_shader
*pshader
= state
->shader
[WINED3D_SHADER_TYPE_PIXEL
];
1520 const struct glsl_context_data
*ctx_data
= context
->shader_backend_data
;
1521 struct wined3d_context_gl
*context_gl
= wined3d_context_gl(context
);
1522 struct glsl_shader_prog_link
*prog
= ctx_data
->glsl_program
;
1523 const struct wined3d_gl_info
*gl_info
= context_gl
->gl_info
;
1524 float position_fixup
[4 * WINED3D_MAX_VIEWPORTS
];
1525 struct shader_glsl_priv
*priv
= shader_priv
;
1526 unsigned int constant_version
;
1530 /* No GLSL program set - nothing to do. */
1534 constant_version
= prog
->constant_version
;
1535 update_mask
= context
->constant_update_mask
& prog
->constant_update_mask
;
1537 if (update_mask
& WINED3D_SHADER_CONST_VS_F
)
1538 shader_glsl_load_constants_f(vshader
, gl_info
, state
->vs_consts_f
,
1539 prog
->vs
.uniform_f_locations
, &priv
->vconst_heap
, priv
->stack
, constant_version
);
1541 if (update_mask
& WINED3D_SHADER_CONST_VS_I
)
1542 shader_glsl_load_constants_i(vshader
, gl_info
, state
->vs_consts_i
,
1543 prog
->vs
.uniform_i_locations
, vshader
->reg_maps
.integer_constants
);
1545 if (update_mask
& WINED3D_SHADER_CONST_VS_B
)
1546 shader_glsl_load_constants_b(vshader
, gl_info
, state
->vs_consts_b
,
1547 prog
->vs
.uniform_b_locations
, vshader
->reg_maps
.boolean_constants
);
1549 if (update_mask
& WINED3D_SHADER_CONST_VS_CLIP_PLANES
)
1551 for (i
= 0; i
< gl_info
->limits
.user_clip_distances
; ++i
)
1552 shader_glsl_clip_plane_uniform(context_gl
, state
, i
, prog
);
1555 if (update_mask
& WINED3D_SHADER_CONST_VS_POINTSIZE
)
1556 shader_glsl_pointsize_uniform(context_gl
, state
, prog
);
1558 if (update_mask
& WINED3D_SHADER_CONST_POS_FIXUP
)
1560 unsigned int fixup_count
= state
->shader
[WINED3D_SHADER_TYPE_GEOMETRY
] ?
1561 max(state
->viewport_count
, 1) : 1;
1562 shader_get_position_fixup(context
, state
, fixup_count
, position_fixup
);
1563 if (state
->shader
[WINED3D_SHADER_TYPE_GEOMETRY
])
1564 GL_EXTCALL(glUniform4fv(prog
->gs
.pos_fixup_location
, fixup_count
, position_fixup
));
1565 else if (state
->shader
[WINED3D_SHADER_TYPE_DOMAIN
])
1566 GL_EXTCALL(glUniform4fv(prog
->ds
.pos_fixup_location
, 1, position_fixup
));
1568 GL_EXTCALL(glUniform4fv(prog
->vs
.pos_fixup_location
, 1, position_fixup
));
1569 checkGLcall("glUniform4fv");
1572 if (update_mask
& WINED3D_SHADER_CONST_BASE_VERTEX_ID
)
1574 GL_EXTCALL(glUniform1i(prog
->vs
.base_vertex_id_location
, state
->base_vertex_index
));
1575 checkGLcall("base vertex id");
1578 if (update_mask
& WINED3D_SHADER_CONST_FFP_MODELVIEW
)
1580 struct wined3d_matrix mat
;
1582 get_modelview_matrix(context
, state
, 0, &mat
);
1583 GL_EXTCALL(glUniformMatrix4fv(prog
->vs
.modelview_matrix_location
[0], 1, FALSE
, &mat
._11
));
1584 checkGLcall("glUniformMatrix4fv");
1586 shader_glsl_ffp_vertex_normalmatrix_uniform(context_gl
, state
, prog
);
1589 if (update_mask
& WINED3D_SHADER_CONST_FFP_VERTEXBLEND
)
1591 struct wined3d_matrix mat
;
1593 for (i
= 1; i
< MAX_VERTEX_BLENDS
; ++i
)
1595 if (prog
->vs
.modelview_matrix_location
[i
] == -1)
1598 get_modelview_matrix(context
, state
, i
, &mat
);
1599 GL_EXTCALL(glUniformMatrix4fv(prog
->vs
.modelview_matrix_location
[i
], 1, FALSE
, &mat
._11
));
1600 checkGLcall("glUniformMatrix4fv");
1604 if (update_mask
& WINED3D_SHADER_CONST_FFP_PROJ
)
1606 struct wined3d_matrix projection
;
1608 get_projection_matrix(context
, state
, &projection
);
1609 GL_EXTCALL(glUniformMatrix4fv(prog
->vs
.projection_matrix_location
, 1, FALSE
, &projection
._11
));
1610 checkGLcall("glUniformMatrix4fv");
1613 if (update_mask
& WINED3D_SHADER_CONST_FFP_TEXMATRIX
)
1615 for (i
= 0; i
< WINED3D_MAX_TEXTURES
; ++i
)
1616 shader_glsl_ffp_vertex_texmatrix_uniform(context_gl
, state
, i
, prog
);
1619 if (update_mask
& WINED3D_SHADER_CONST_FFP_MATERIAL
)
1620 shader_glsl_ffp_vertex_material_uniform(context_gl
, state
, prog
);
1622 if (update_mask
& WINED3D_SHADER_CONST_FFP_LIGHTS
)
1624 unsigned int point_idx
, spot_idx
, directional_idx
, parallel_point_idx
;
1625 DWORD point_count
= 0;
1626 DWORD spot_count
= 0;
1627 DWORD directional_count
= 0;
1628 DWORD parallel_point_count
= 0;
1630 for (i
= 0; i
< WINED3D_MAX_ACTIVE_LIGHTS
; ++i
)
1632 if (!state
->light_state
.lights
[i
])
1635 switch (state
->light_state
.lights
[i
]->OriginalParms
.type
)
1637 case WINED3D_LIGHT_POINT
:
1640 case WINED3D_LIGHT_SPOT
:
1643 case WINED3D_LIGHT_DIRECTIONAL
:
1644 ++directional_count
;
1646 case WINED3D_LIGHT_PARALLELPOINT
:
1647 ++parallel_point_count
;
1650 FIXME("Unhandled light type %#x.\n", state
->light_state
.lights
[i
]->OriginalParms
.type
);
1655 spot_idx
= point_idx
+ point_count
;
1656 directional_idx
= spot_idx
+ spot_count
;
1657 parallel_point_idx
= directional_idx
+ directional_count
;
1659 shader_glsl_ffp_vertex_lightambient_uniform(context_gl
, state
, prog
);
1660 for (i
= 0; i
< WINED3D_MAX_ACTIVE_LIGHTS
; ++i
)
1662 const struct wined3d_light_info
*light_info
= state
->light_state
.lights
[i
];
1668 switch (light_info
->OriginalParms
.type
)
1670 case WINED3D_LIGHT_POINT
:
1673 case WINED3D_LIGHT_SPOT
:
1676 case WINED3D_LIGHT_DIRECTIONAL
:
1677 idx
= directional_idx
++;
1679 case WINED3D_LIGHT_PARALLELPOINT
:
1680 idx
= parallel_point_idx
++;
1683 FIXME("Unhandled light type %#x.\n", light_info
->OriginalParms
.type
);
1686 shader_glsl_ffp_vertex_light_uniform(context_gl
, state
, idx
, light_info
, prog
);
1690 if (update_mask
& WINED3D_SHADER_CONST_PS_F
)
1691 shader_glsl_load_constants_f(pshader
, gl_info
, state
->ps_consts_f
,
1692 prog
->ps
.uniform_f_locations
, &priv
->pconst_heap
, priv
->stack
, constant_version
);
1694 if (update_mask
& WINED3D_SHADER_CONST_PS_I
)
1695 shader_glsl_load_constants_i(pshader
, gl_info
, state
->ps_consts_i
,
1696 prog
->ps
.uniform_i_locations
, pshader
->reg_maps
.integer_constants
);
1698 if (update_mask
& WINED3D_SHADER_CONST_PS_B
)
1699 shader_glsl_load_constants_b(pshader
, gl_info
, state
->ps_consts_b
,
1700 prog
->ps
.uniform_b_locations
, pshader
->reg_maps
.boolean_constants
);
1702 if (update_mask
& WINED3D_SHADER_CONST_PS_BUMP_ENV
)
1704 for (i
= 0; i
< WINED3D_MAX_TEXTURES
; ++i
)
1706 if (prog
->ps
.bumpenv_mat_location
[i
] == -1)
1709 GL_EXTCALL(glUniformMatrix2fv(prog
->ps
.bumpenv_mat_location
[i
], 1, 0,
1710 (const GLfloat
*)&state
->texture_states
[i
][WINED3D_TSS_BUMPENV_MAT00
]));
1712 if (prog
->ps
.bumpenv_lum_scale_location
[i
] != -1)
1714 GL_EXTCALL(glUniform1fv(prog
->ps
.bumpenv_lum_scale_location
[i
], 1,
1715 (const GLfloat
*)&state
->texture_states
[i
][WINED3D_TSS_BUMPENV_LSCALE
]));
1716 GL_EXTCALL(glUniform1fv(prog
->ps
.bumpenv_lum_offset_location
[i
], 1,
1717 (const GLfloat
*)&state
->texture_states
[i
][WINED3D_TSS_BUMPENV_LOFFSET
]));
1721 checkGLcall("bump env uniforms");
1724 if (update_mask
& WINED3D_SHADER_CONST_PS_Y_CORR
)
1726 const struct wined3d_vec4 correction_params
=
1728 /* Position is relative to the framebuffer, not the viewport. */
1729 context
->render_offscreen
? 0.0f
: (float)state
->fb
.render_targets
[0]->height
,
1730 context
->render_offscreen
? 1.0f
: -1.0f
,
1735 GL_EXTCALL(glUniform4fv(prog
->ps
.ycorrection_location
, 1, &correction_params
.x
));
1738 if (update_mask
& WINED3D_SHADER_CONST_PS_NP2_FIXUP
)
1739 shader_glsl_load_np2fixup_constants(&prog
->ps
, gl_info
, state
);
1740 if (update_mask
& WINED3D_SHADER_CONST_FFP_COLOR_KEY
)
1741 shader_glsl_load_color_key_constant(&prog
->ps
, gl_info
, state
);
1743 if (update_mask
& WINED3D_SHADER_CONST_FFP_PS
)
1745 struct wined3d_color color
;
1747 if (prog
->ps
.tex_factor_location
!= -1)
1749 wined3d_color_from_d3dcolor(&color
, state
->render_states
[WINED3D_RS_TEXTUREFACTOR
]);
1750 GL_EXTCALL(glUniform4fv(prog
->ps
.tex_factor_location
, 1, &color
.r
));
1753 if (state
->render_states
[WINED3D_RS_SPECULARENABLE
])
1754 GL_EXTCALL(glUniform4f(prog
->ps
.specular_enable_location
, 1.0f
, 1.0f
, 1.0f
, 0.0f
));
1756 GL_EXTCALL(glUniform4f(prog
->ps
.specular_enable_location
, 0.0f
, 0.0f
, 0.0f
, 0.0f
));
1758 for (i
= 0; i
< WINED3D_MAX_TEXTURES
; ++i
)
1760 if (prog
->ps
.tss_constant_location
[i
] == -1)
1763 wined3d_color_from_d3dcolor(&color
, state
->texture_states
[i
][WINED3D_TSS_CONSTANT
]);
1764 GL_EXTCALL(glUniform4fv(prog
->ps
.tss_constant_location
[i
], 1, &color
.r
));
1767 checkGLcall("fixed function uniforms");
1770 if (update_mask
& WINED3D_SHADER_CONST_PS_FOG
)
1771 shader_glsl_load_fog_uniform(context_gl
, state
, prog
);
1773 if (update_mask
& WINED3D_SHADER_CONST_PS_ALPHA_TEST
)
1775 float ref
= wined3d_alpha_ref(state
);
1777 GL_EXTCALL(glUniform1f(prog
->ps
.alpha_test_ref_location
, ref
));
1778 checkGLcall("alpha test emulation uniform");
1781 if (priv
->next_constant_version
== UINT_MAX
)
1783 TRACE("Max constant version reached, resetting to 0.\n");
1784 wine_rb_for_each_entry(&priv
->program_lookup
, reset_program_constant_version
, NULL
);
1785 priv
->next_constant_version
= 1;
1789 prog
->constant_version
= priv
->next_constant_version
++;
1793 static void update_heap_entry(struct constant_heap
*heap
, unsigned int idx
, DWORD new_version
)
1795 struct constant_entry
*entries
= heap
->entries
;
1796 unsigned int *positions
= heap
->positions
;
1797 unsigned int heap_idx
, parent_idx
;
1799 if (!heap
->contained
[idx
])
1801 heap_idx
= heap
->size
++;
1802 heap
->contained
[idx
] = TRUE
;
1806 heap_idx
= positions
[idx
];
1809 while (heap_idx
> 1)
1811 parent_idx
= heap_idx
>> 1;
1813 if (new_version
<= entries
[parent_idx
].version
) break;
1815 entries
[heap_idx
] = entries
[parent_idx
];
1816 positions
[entries
[parent_idx
].idx
] = heap_idx
;
1817 heap_idx
= parent_idx
;
1820 entries
[heap_idx
].version
= new_version
;
1821 entries
[heap_idx
].idx
= idx
;
1822 positions
[idx
] = heap_idx
;
1825 static void shader_glsl_update_float_vertex_constants(struct wined3d_device
*device
, UINT start
, UINT count
)
1827 struct shader_glsl_priv
*priv
= device
->shader_priv
;
1828 struct constant_heap
*heap
= &priv
->vconst_heap
;
1831 for (i
= start
; i
< count
+ start
; ++i
)
1833 update_heap_entry(heap
, i
, priv
->next_constant_version
);
1837 static void shader_glsl_update_float_pixel_constants(struct wined3d_device
*device
, UINT start
, UINT count
)
1839 struct shader_glsl_priv
*priv
= device
->shader_priv
;
1840 struct constant_heap
*heap
= &priv
->pconst_heap
;
1843 for (i
= start
; i
< count
+ start
; ++i
)
1845 update_heap_entry(heap
, i
, priv
->next_constant_version
);
1849 static unsigned int vec4_varyings(DWORD shader_major
, const struct wined3d_gl_info
*gl_info
)
1851 unsigned int ret
= gl_info
->limits
.glsl_varyings
/ 4;
1852 /* 4.0 shaders do not write clip coords because d3d10 does not support user clipplanes */
1853 if(shader_major
> 3) return ret
;
1855 /* 3.0 shaders may need an extra varying for the clip coord on some cards(mostly dx10 ones) */
1856 if (gl_info
->quirks
& WINED3D_QUIRK_GLSL_CLIP_VARYING
) ret
-= 1;
1860 static BOOL
needs_legacy_glsl_syntax(const struct wined3d_gl_info
*gl_info
)
1862 return gl_info
->glsl_version
< MAKEDWORD_VERSION(1, 30);
1865 static BOOL
shader_glsl_use_explicit_attrib_location(const struct wined3d_gl_info
*gl_info
)
1867 return gl_info
->supported
[ARB_EXPLICIT_ATTRIB_LOCATION
]
1868 && shader_glsl_use_layout_qualifier(gl_info
) && !needs_legacy_glsl_syntax(gl_info
);
1871 static BOOL
shader_glsl_use_interface_blocks(const struct wined3d_gl_info
*gl_info
)
1873 return shader_glsl_get_version(gl_info
) >= 150;
1876 static const char *get_attribute_keyword(const struct wined3d_gl_info
*gl_info
)
1878 return needs_legacy_glsl_syntax(gl_info
) ? "attribute" : "in";
1881 static void PRINTF_ATTR(4, 5) declare_in_varying(const struct wined3d_gl_info
*gl_info
,
1882 struct wined3d_string_buffer
*buffer
, BOOL flat
, const char *format
, ...)
1887 shader_addline(buffer
, "%s%s ", flat
? "flat " : "",
1888 needs_legacy_glsl_syntax(gl_info
) ? "varying" : "in");
1891 va_start(args
, format
);
1892 ret
= shader_vaddline(buffer
, format
, args
);
1896 if (!string_buffer_resize(buffer
, ret
))
1901 static void PRINTF_ATTR(4, 5) declare_out_varying(const struct wined3d_gl_info
*gl_info
,
1902 struct wined3d_string_buffer
*buffer
, BOOL flat
, const char *format
, ...)
1907 shader_addline(buffer
, "%s%s ", flat
? "flat " : "",
1908 needs_legacy_glsl_syntax(gl_info
) ? "varying" : "out");
1911 va_start(args
, format
);
1912 ret
= shader_vaddline(buffer
, format
, args
);
1916 if (!string_buffer_resize(buffer
, ret
))
1921 static const char *shader_glsl_shader_input_name(const struct wined3d_gl_info
*gl_info
)
1923 return shader_glsl_use_interface_blocks(gl_info
) ? "shader_in.reg" : "ps_link";
1926 static const char *shader_glsl_shader_output_name(const struct wined3d_gl_info
*gl_info
)
1928 return shader_glsl_use_interface_blocks(gl_info
) ? "shader_out.reg" : "ps_link";
1931 static const char *shader_glsl_interpolation_qualifiers(enum wined3d_shader_interpolation_mode mode
)
1935 case WINED3DSIM_CONSTANT
:
1937 case WINED3DSIM_LINEAR_NOPERSPECTIVE
:
1938 return "noperspective ";
1940 FIXME("Unhandled interpolation mode %#x.\n", mode
);
1941 case WINED3DSIM_NONE
:
1942 case WINED3DSIM_LINEAR
:
1947 static enum wined3d_shader_interpolation_mode
wined3d_extract_interpolation_mode(
1948 const DWORD
*packed_interpolation_mode
, unsigned int register_idx
)
1950 return wined3d_extract_bits(packed_interpolation_mode
,
1951 register_idx
* WINED3D_PACKED_INTERPOLATION_BIT_COUNT
, WINED3D_PACKED_INTERPOLATION_BIT_COUNT
);
1954 static void shader_glsl_declare_shader_inputs(const struct wined3d_gl_info
*gl_info
,
1955 struct wined3d_string_buffer
*buffer
, unsigned int element_count
,
1956 const DWORD
*interpolation_mode
, BOOL unroll
)
1958 enum wined3d_shader_interpolation_mode mode
;
1961 if (shader_glsl_use_interface_blocks(gl_info
))
1965 shader_addline(buffer
, "in shader_in_out {\n");
1966 for (i
= 0; i
< element_count
; ++i
)
1968 mode
= wined3d_extract_interpolation_mode(interpolation_mode
, i
);
1969 shader_addline(buffer
, " %svec4 reg%u;\n", shader_glsl_interpolation_qualifiers(mode
), i
);
1971 shader_addline(buffer
, "} shader_in;\n");
1975 shader_addline(buffer
, "in shader_in_out { vec4 reg[%u]; } shader_in;\n", element_count
);
1980 declare_in_varying(gl_info
, buffer
, FALSE
, "vec4 ps_link[%u];\n", element_count
);
1984 static BOOL
needs_interpolation_qualifiers_for_shader_outputs(const struct wined3d_gl_info
*gl_info
)
1986 /* In GLSL 4.40+ it is fine to specify interpolation qualifiers only in
1987 * fragment shaders. In older GLSL versions interpolation qualifiers must
1988 * match between shader stages. */
1989 return gl_info
->glsl_version
< MAKEDWORD_VERSION(4, 40);
1992 static void shader_glsl_declare_shader_outputs(const struct wined3d_gl_info
*gl_info
,
1993 struct wined3d_string_buffer
*buffer
, unsigned int element_count
, BOOL rasterizer_setup
,
1994 const DWORD
*interpolation_mode
)
1996 enum wined3d_shader_interpolation_mode mode
;
1999 if (shader_glsl_use_interface_blocks(gl_info
))
2001 if (rasterizer_setup
)
2003 shader_addline(buffer
, "out shader_in_out {\n");
2004 for (i
= 0; i
< element_count
; ++i
)
2006 const char *interpolation_qualifiers
= "";
2007 if (needs_interpolation_qualifiers_for_shader_outputs(gl_info
))
2009 mode
= wined3d_extract_interpolation_mode(interpolation_mode
, i
);
2010 interpolation_qualifiers
= shader_glsl_interpolation_qualifiers(mode
);
2012 shader_addline(buffer
, " %svec4 reg%u;\n", interpolation_qualifiers
, i
);
2014 shader_addline(buffer
, "} shader_out;\n");
2018 shader_addline(buffer
, "out shader_in_out { vec4 reg[%u]; } shader_out;\n", element_count
);
2023 declare_out_varying(gl_info
, buffer
, FALSE
, "vec4 ps_link[%u];\n", element_count
);
2027 static BOOL
use_legacy_fragment_output(const struct wined3d_gl_info
*gl_info
)
2029 /* Technically 1.30 does support user-defined fragment shader outputs but
2030 * we might not have glBindFragDataLocation() available (i.e. GL version
2031 * might be < 3.0). */
2032 return gl_info
->glsl_version
<= MAKEDWORD_VERSION(1, 30);
2035 static const char *get_fragment_output(const struct wined3d_gl_info
*gl_info
)
2037 return use_legacy_fragment_output(gl_info
) ? "gl_FragData" : "ps_out";
2040 static const char *glsl_primitive_type_from_d3d(enum wined3d_primitive_type primitive_type
)
2042 switch (primitive_type
)
2044 case WINED3D_PT_POINTLIST
:
2047 case WINED3D_PT_LINELIST
:
2050 case WINED3D_PT_LINESTRIP
:
2051 return "line_strip";
2053 case WINED3D_PT_TRIANGLELIST
:
2056 case WINED3D_PT_TRIANGLESTRIP
:
2057 return "triangle_strip";
2059 case WINED3D_PT_LINELIST_ADJ
:
2060 return "lines_adjacency";
2062 case WINED3D_PT_TRIANGLELIST_ADJ
:
2063 return "triangles_adjacency";
2066 FIXME("Unhandled primitive type %s.\n", debug_d3dprimitivetype(primitive_type
));
2071 static BOOL
glsl_is_color_reg_read(const struct wined3d_shader
*shader
, unsigned int idx
)
2073 const struct wined3d_shader_signature
*input_signature
= &shader
->input_signature
;
2074 const struct wined3d_shader_reg_maps
*reg_maps
= &shader
->reg_maps
;
2075 DWORD input_reg_used
= shader
->u
.ps
.input_reg_used
;
2078 if (reg_maps
->shader_version
.major
< 3)
2079 return input_reg_used
& (1u << idx
);
2081 for (i
= 0; i
< input_signature
->element_count
; ++i
)
2083 const struct wined3d_shader_signature_element
*input
= &input_signature
->elements
[i
];
2085 if (!(reg_maps
->input_registers
& (1u << input
->register_idx
)))
2088 if (shader_match_semantic(input
->semantic_name
, WINED3D_DECL_USAGE_COLOR
)
2089 && input
->semantic_idx
== idx
)
2090 return input_reg_used
& (1u << input
->register_idx
);
2095 static BOOL
glsl_is_shadow_sampler(const struct wined3d_shader
*shader
,
2096 const struct ps_compile_args
*ps_args
, unsigned int resource_idx
, unsigned int sampler_idx
)
2098 const struct wined3d_shader_version
*version
= &shader
->reg_maps
.shader_version
;
2100 if (version
->major
>= 4)
2101 return shader
->reg_maps
.sampler_comparison_mode
& (1u << sampler_idx
);
2103 return version
->type
== WINED3D_SHADER_TYPE_PIXEL
&& (ps_args
->shadow
& (1u << resource_idx
));
2106 static void shader_glsl_declare_typed_vertex_attribute(struct wined3d_string_buffer
*buffer
,
2107 const struct wined3d_gl_info
*gl_info
, const char *vector_type
, const char *scalar_type
,
2110 shader_addline(buffer
, "%s %s4 vs_in_%s%u;\n",
2111 get_attribute_keyword(gl_info
), vector_type
, scalar_type
, index
);
2112 shader_addline(buffer
, "vec4 vs_in%u = %sBitsToFloat(vs_in_%s%u);\n",
2113 index
, scalar_type
, scalar_type
, index
);
2116 static void shader_glsl_declare_generic_vertex_attribute(struct wined3d_string_buffer
*buffer
,
2117 const struct wined3d_gl_info
*gl_info
, const struct wined3d_shader_signature_element
*e
)
2119 unsigned int index
= e
->register_idx
;
2120 enum wined3d_component_type type
;
2122 if (e
->sysval_semantic
== WINED3D_SV_VERTEX_ID
)
2124 shader_addline(buffer
, "uniform int base_vertex_id;\n");
2125 shader_addline(buffer
, "vec4 vs_in%u = vec4(intBitsToFloat(gl_VertexID - base_vertex_id), 0.0, 0.0, 0.0);\n", index
);
2128 if (e
->sysval_semantic
== WINED3D_SV_INSTANCE_ID
)
2130 shader_addline(buffer
, "vec4 vs_in%u = vec4(intBitsToFloat(gl_InstanceID), 0.0, 0.0, 0.0);\n",
2134 if (e
->sysval_semantic
&& e
->sysval_semantic
!= WINED3D_SV_POSITION
)
2135 FIXME("Unhandled sysval semantic %#x.\n", e
->sysval_semantic
);
2137 if (shader_glsl_use_explicit_attrib_location(gl_info
))
2138 shader_addline(buffer
, "layout(location = %u) ", index
);
2140 type
= e
->component_type
;
2141 if ((unsigned int)type
>= ARRAY_SIZE(component_type_info
))
2143 FIXME("Unhandled type %#x.\n", type
);
2144 type
= WINED3D_TYPE_FLOAT
;
2146 if (type
== WINED3D_TYPE_FLOAT
|| type
== WINED3D_TYPE_UNKNOWN
)
2147 shader_addline(buffer
, "%s vec4 vs_in%u;\n", get_attribute_keyword(gl_info
), index
);
2149 shader_glsl_declare_typed_vertex_attribute(buffer
, gl_info
,
2150 component_type_info
[type
].glsl_vector_type
,
2151 component_type_info
[type
].glsl_scalar_type
, index
);
2154 /** Generate the variable & register declarations for the GLSL output target */
2155 static void shader_generate_glsl_declarations(const struct wined3d_context_gl
*context_gl
,
2156 struct wined3d_string_buffer
*buffer
, const struct wined3d_shader
*shader
,
2157 const struct wined3d_shader_reg_maps
*reg_maps
, const struct shader_glsl_ctx_priv
*ctx_priv
)
2159 const struct wined3d_shader_version
*version
= ®_maps
->shader_version
;
2160 const struct vs_compile_args
*vs_args
= ctx_priv
->cur_vs_args
;
2161 const struct ps_compile_args
*ps_args
= ctx_priv
->cur_ps_args
;
2162 const struct wined3d_gl_info
*gl_info
= context_gl
->gl_info
;
2163 const struct wined3d_shader_indexable_temp
*idx_temp_reg
;
2164 unsigned int uniform_block_base
, uniform_block_count
;
2165 enum wined3d_shader_resource_type resource_type
;
2166 const struct wined3d_shader_lconst
*lconst
;
2171 if (wined3d_settings
.strict_shader_math
)
2172 shader_addline(buffer
, "#pragma optionNV(fastmath off)\n");
2174 prefix
= shader_glsl_get_prefix(version
->type
);
2176 /* Prototype the subroutines */
2177 map
= reg_maps
->labels
;
2180 i
= wined3d_bit_scan(&map
);
2181 shader_addline(buffer
, "void subroutine%u();\n", i
);
2184 if (version
->type
!= WINED3D_SHADER_TYPE_PIXEL
&& version
->type
!= WINED3D_SHADER_TYPE_COMPUTE
)
2186 if (version
->type
== WINED3D_SHADER_TYPE_HULL
)
2187 shader_addline(buffer
, "out gl_PerVertex { invariant vec4 gl_Position; } gl_out[];\n");
2189 shader_addline(buffer
, "invariant gl_Position;\n");
2192 /* Declare the constants (aka uniforms) */
2193 if (shader
->limits
->constant_float
> 0)
2195 unsigned max_constantsF
;
2197 /* Unless the shader uses indirect addressing, always declare the
2198 * maximum array size and ignore that we need some uniforms privately.
2199 * E.g. if GL supports 256 uniforms, and we need 2 for the pos fixup
2200 * and immediate values, still declare VC[256]. If the shader needs
2201 * more uniforms than we have it won't work in any case. If it uses
2202 * less, the compiler will figure out which uniforms are really used
2203 * and strip them out. This allows a shader to use c255 on a dx9 card,
2204 * as long as it doesn't also use all the other constants.
2206 * If the shader uses indirect addressing the compiler must assume
2207 * that all declared uniforms are used. In this case, declare only the
2208 * amount that we're assured to have.
2210 * Thus we run into problems in these two cases:
2211 * 1) The shader really uses more uniforms than supported.
2212 * 2) The shader uses indirect addressing, less constants than
2213 * supported, but uses a constant index > #supported consts. */
2214 if (version
->type
== WINED3D_SHADER_TYPE_PIXEL
)
2216 /* No indirect addressing here. */
2217 max_constantsF
= gl_info
->limits
.glsl_ps_float_constants
;
2221 if (reg_maps
->usesrelconstF
)
2223 /* Subtract the other potential uniforms from the max
2224 * available (bools, ints, and 1 row of projection matrix).
2225 * Subtract another uniform for immediate values, which have
2226 * to be loaded via uniform by the driver as well. The shader
2227 * code only uses 0.5, 2.0, 1.0, 128 and -128 in vertex
2228 * shader code, so one vec4 should be enough. (Unfortunately
2229 * the Nvidia driver doesn't store 128 and -128 in one float).
2231 * Writing gl_ClipVertex requires one uniform for each
2232 * clipplane as well. */
2233 max_constantsF
= gl_info
->limits
.glsl_vs_float_constants
- 3;
2234 if (vs_args
->clip_enabled
)
2235 max_constantsF
-= gl_info
->limits
.user_clip_distances
;
2236 max_constantsF
-= wined3d_popcount(reg_maps
->integer_constants
);
2237 /* Strictly speaking a bool only uses one scalar, but the nvidia(Linux) compiler doesn't pack them properly,
2238 * so each scalar requires a full vec4. We could work around this by packing the booleans ourselves, but
2239 * for now take this into account when calculating the number of available constants
2241 max_constantsF
-= wined3d_popcount(reg_maps
->boolean_constants
);
2242 /* Set by driver quirks in directx.c */
2243 max_constantsF
-= gl_info
->reserved_glsl_constants
;
2245 if (max_constantsF
< shader
->limits
->constant_float
)
2247 static unsigned int once
;
2250 ERR_(winediag
)("The hardware does not support enough uniform components to run this shader,"
2251 " it may not render correctly.\n");
2253 WARN("The hardware does not support enough uniform components to run this shader.\n");
2258 max_constantsF
= gl_info
->limits
.glsl_vs_float_constants
;
2261 max_constantsF
= min(shader
->limits
->constant_float
, max_constantsF
);
2262 shader_addline(buffer
, "uniform vec4 %s_c[%u];\n", prefix
, max_constantsF
);
2265 /* Always declare the full set of constants, the compiler can remove the
2266 * unused ones because d3d doesn't (yet) support indirect int and bool
2267 * constant addressing. This avoids problems if the app uses e.g. i0 and i9. */
2268 if (shader
->limits
->constant_int
> 0 && reg_maps
->integer_constants
)
2269 shader_addline(buffer
, "uniform ivec4 %s_i[%u];\n", prefix
, shader
->limits
->constant_int
);
2271 if (shader
->limits
->constant_bool
> 0 && reg_maps
->boolean_constants
)
2272 shader_addline(buffer
, "uniform bool %s_b[%u];\n", prefix
, shader
->limits
->constant_bool
);
2274 /* Declare immediate constant buffer */
2276 shader_addline(buffer
, "uniform vec4 %s_icb[%u];\n", prefix
, reg_maps
->icb
->vec4_count
);
2278 /* Declare constant buffers */
2279 wined3d_gl_limits_get_uniform_block_range(&gl_info
->limits
, version
->type
,
2280 &uniform_block_base
, &uniform_block_count
);
2281 for (i
= 0; i
< min(uniform_block_count
, WINED3D_MAX_CBS
); ++i
)
2283 if (reg_maps
->cb_sizes
[i
])
2285 shader_addline(buffer
, "layout(std140");
2286 if (shader_glsl_use_layout_binding_qualifier(gl_info
))
2287 shader_addline(buffer
, ", binding = %u", uniform_block_base
+ i
);
2288 shader_addline(buffer
, ") uniform block_%s_cb%u { vec4 %s_cb%u[%u]; };\n",
2289 prefix
, i
, prefix
, i
, reg_maps
->cb_sizes
[i
]);
2293 /* Declare texture samplers */
2294 for (i
= 0; i
< reg_maps
->sampler_map
.count
; ++i
)
2296 struct wined3d_shader_sampler_map_entry
*entry
;
2297 const char *sampler_type_prefix
, *sampler_type
;
2298 BOOL shadow_sampler
, tex_rect
;
2300 entry
= ®_maps
->sampler_map
.entries
[i
];
2302 if (entry
->resource_idx
>= ARRAY_SIZE(reg_maps
->resource_info
))
2304 ERR("Invalid resource index %u.\n", entry
->resource_idx
);
2308 switch (reg_maps
->resource_info
[entry
->resource_idx
].data_type
)
2310 case WINED3D_DATA_FLOAT
:
2311 case WINED3D_DATA_UNORM
:
2312 case WINED3D_DATA_SNORM
:
2313 sampler_type_prefix
= "";
2316 case WINED3D_DATA_INT
:
2317 sampler_type_prefix
= "i";
2320 case WINED3D_DATA_UINT
:
2321 sampler_type_prefix
= "u";
2325 sampler_type_prefix
= "";
2326 ERR("Unhandled resource data type %#x.\n", reg_maps
->resource_info
[i
].data_type
);
2330 shadow_sampler
= glsl_is_shadow_sampler(shader
, ps_args
, entry
->resource_idx
, entry
->sampler_idx
);
2331 resource_type
= version
->type
== WINED3D_SHADER_TYPE_PIXEL
2332 ? pixelshader_get_resource_type(reg_maps
, entry
->resource_idx
, ps_args
->tex_types
)
2333 : reg_maps
->resource_info
[entry
->resource_idx
].type
;
2335 switch (resource_type
)
2337 case WINED3D_SHADER_RESOURCE_BUFFER
:
2338 sampler_type
= "samplerBuffer";
2341 case WINED3D_SHADER_RESOURCE_TEXTURE_1D
:
2343 sampler_type
= "sampler1DShadow";
2345 sampler_type
= "sampler1D";
2348 case WINED3D_SHADER_RESOURCE_TEXTURE_2D
:
2349 tex_rect
= version
->type
== WINED3D_SHADER_TYPE_PIXEL
2350 && (ps_args
->np2_fixup
& (1u << entry
->resource_idx
))
2351 && gl_info
->supported
[ARB_TEXTURE_RECTANGLE
];
2355 sampler_type
= "sampler2DRectShadow";
2357 sampler_type
= "sampler2DShadow";
2362 sampler_type
= "sampler2DRect";
2364 sampler_type
= "sampler2D";
2368 case WINED3D_SHADER_RESOURCE_TEXTURE_3D
:
2370 FIXME("Unsupported 3D shadow sampler.\n");
2371 sampler_type
= "sampler3D";
2374 case WINED3D_SHADER_RESOURCE_TEXTURE_CUBE
:
2376 sampler_type
= "samplerCubeShadow";
2378 sampler_type
= "samplerCube";
2381 case WINED3D_SHADER_RESOURCE_TEXTURE_1DARRAY
:
2383 sampler_type
= "sampler1DArrayShadow";
2385 sampler_type
= "sampler1DArray";
2388 case WINED3D_SHADER_RESOURCE_TEXTURE_2DARRAY
:
2390 sampler_type
= "sampler2DArrayShadow";
2392 sampler_type
= "sampler2DArray";
2395 case WINED3D_SHADER_RESOURCE_TEXTURE_CUBEARRAY
:
2397 sampler_type
= "samplerCubeArrayShadow";
2399 sampler_type
= "samplerCubeArray";
2402 case WINED3D_SHADER_RESOURCE_TEXTURE_2DMS
:
2403 sampler_type
= "sampler2DMS";
2406 case WINED3D_SHADER_RESOURCE_TEXTURE_2DMSARRAY
:
2407 sampler_type
= "sampler2DMSArray";
2411 sampler_type
= "unsupported_sampler";
2412 FIXME("Unhandled resource type %#x.\n", resource_type
);
2416 if (shader_glsl_use_layout_binding_qualifier(gl_info
))
2417 shader_glsl_append_sampler_binding_qualifier(buffer
, &context_gl
->c
, version
, entry
->bind_idx
);
2418 shader_addline(buffer
, "uniform %s%s %s_sampler%u;\n",
2419 sampler_type_prefix
, sampler_type
, prefix
, entry
->bind_idx
);
2422 /* Declare images */
2423 for (i
= 0; i
< ARRAY_SIZE(reg_maps
->uav_resource_info
); ++i
)
2425 const char *image_type_prefix
, *image_type
, *read_format
;
2427 if (!reg_maps
->uav_resource_info
[i
].type
)
2430 switch (reg_maps
->uav_resource_info
[i
].data_type
)
2432 case WINED3D_DATA_FLOAT
:
2433 case WINED3D_DATA_UNORM
:
2434 case WINED3D_DATA_SNORM
:
2435 image_type_prefix
= "";
2436 read_format
= "r32f";
2439 case WINED3D_DATA_INT
:
2440 image_type_prefix
= "i";
2441 read_format
= "r32i";
2444 case WINED3D_DATA_UINT
:
2445 image_type_prefix
= "u";
2446 read_format
= "r32ui";
2450 image_type_prefix
= "";
2452 ERR("Unhandled resource data type %#x.\n", reg_maps
->uav_resource_info
[i
].data_type
);
2456 switch (reg_maps
->uav_resource_info
[i
].type
)
2458 case WINED3D_SHADER_RESOURCE_BUFFER
:
2459 image_type
= "imageBuffer";
2462 case WINED3D_SHADER_RESOURCE_TEXTURE_1D
:
2463 image_type
= "image1D";
2466 case WINED3D_SHADER_RESOURCE_TEXTURE_2D
:
2467 image_type
= "image2D";
2470 case WINED3D_SHADER_RESOURCE_TEXTURE_3D
:
2471 image_type
= "image3D";
2474 case WINED3D_SHADER_RESOURCE_TEXTURE_1DARRAY
:
2475 image_type
= "image1DArray";
2478 case WINED3D_SHADER_RESOURCE_TEXTURE_2DARRAY
:
2479 image_type
= "image2DArray";
2483 image_type
= "unsupported_image";
2484 FIXME("Unhandled resource type %#x.\n", reg_maps
->uav_resource_info
[i
].type
);
2488 if (shader_glsl_use_layout_binding_qualifier(gl_info
))
2489 shader_addline(buffer
, "layout(binding = %u)\n", i
);
2490 if (reg_maps
->uav_read_mask
& (1u << i
))
2491 shader_addline(buffer
, "layout(%s) uniform %s%s %s_image%u;\n",
2492 read_format
, image_type_prefix
, image_type
, prefix
, i
);
2494 shader_addline(buffer
, "writeonly uniform %s%s %s_image%u;\n",
2495 image_type_prefix
, image_type
, prefix
, i
);
2497 if (reg_maps
->uav_counter_mask
& (1u << i
))
2498 shader_addline(buffer
, "layout(binding = %u) uniform atomic_uint %s_counter%u;\n",
2502 /* Declare address variables */
2503 map
= reg_maps
->address
;
2506 i
= wined3d_bit_scan(&map
);
2507 shader_addline(buffer
, "ivec4 A%u;\n", i
);
2510 /* Declare output register temporaries */
2511 if (shader
->limits
->packed_output
)
2512 shader_addline(buffer
, "vec4 %s_out[%u];\n", prefix
, shader
->limits
->packed_output
);
2514 /* Declare temporary variables */
2515 if (reg_maps
->temporary_count
)
2517 for (i
= 0; i
< reg_maps
->temporary_count
; ++i
)
2518 shader_addline(buffer
, "vec4 R%u;\n", i
);
2520 else if (version
->major
< 4)
2522 map
= reg_maps
->temporary
;
2525 i
= wined3d_bit_scan(&map
);
2526 shader_addline(buffer
, "vec4 R%u;\n", i
);
2530 /* Declare indexable temporary variables */
2531 LIST_FOR_EACH_ENTRY(idx_temp_reg
, ®_maps
->indexable_temps
, struct wined3d_shader_indexable_temp
, entry
)
2533 if (idx_temp_reg
->component_count
!= 4)
2534 FIXME("Ignoring component count %u.\n", idx_temp_reg
->component_count
);
2535 shader_addline(buffer
, "vec4 X%u[%u];\n", idx_temp_reg
->register_idx
, idx_temp_reg
->register_size
);
2538 /* Declare loop registers aLx */
2539 if (version
->major
< 4)
2541 for (i
= 0; i
< reg_maps
->loop_depth
; ++i
)
2543 shader_addline(buffer
, "int aL%u;\n", i
);
2544 shader_addline(buffer
, "int tmpInt%u;\n", i
);
2548 /* Temporary variables for matrix operations */
2549 shader_addline(buffer
, "vec4 tmp0;\n");
2550 shader_addline(buffer
, "vec4 tmp1;\n");
2551 if (gl_info
->supported
[ARB_GPU_SHADER5
])
2552 shader_addline(buffer
, "precise vec4 tmp_precise[2];\n");
2554 shader_addline(buffer
, "/* precise */ vec4 tmp_precise[2];\n");
2556 if (!shader
->load_local_constsF
)
2558 LIST_FOR_EACH_ENTRY(lconst
, &shader
->constantsF
, struct wined3d_shader_lconst
, entry
)
2560 shader_addline(buffer
, "const vec4 %s_lc%u = ", prefix
, lconst
->idx
);
2561 shader_glsl_append_imm_vec(buffer
, (const float *)lconst
->value
, ARRAY_SIZE(lconst
->value
), gl_info
);
2562 shader_addline(buffer
, ";\n");
2568 static void shader_glsl_add_src_param_ext(const struct wined3d_shader_context
*ctx
,
2569 const struct wined3d_shader_src_param
*wined3d_src
, DWORD mask
, struct glsl_src_param
*glsl_src
,
2570 enum wined3d_data_type data_type
);
2572 /** Used for opcode modifiers - They multiply the result by the specified amount */
2573 static const char * const shift_glsl_tab
[] = {
2575 "2.0 * ", /* 1 (x2) */
2576 "4.0 * ", /* 2 (x4) */
2577 "8.0 * ", /* 3 (x8) */
2578 "16.0 * ", /* 4 (x16) */
2579 "32.0 * ", /* 5 (x32) */
2586 "0.0625 * ", /* 12 (d16) */
2587 "0.125 * ", /* 13 (d8) */
2588 "0.25 * ", /* 14 (d4) */
2589 "0.5 * " /* 15 (d2) */
2592 /* Generate a GLSL parameter that does the input modifier computation and return the input register/mask to use */
2593 static void shader_glsl_gen_modifier(enum wined3d_shader_src_modifier src_modifier
,
2594 const char *in_reg
, const char *in_regswizzle
, char *out_str
)
2596 switch (src_modifier
)
2598 case WINED3DSPSM_DZ
: /* Need to handle this in the instructions itself (texld & texcrd). */
2599 case WINED3DSPSM_DW
:
2600 case WINED3DSPSM_NONE
:
2601 sprintf(out_str
, "%s%s", in_reg
, in_regswizzle
);
2603 case WINED3DSPSM_NEG
:
2604 sprintf(out_str
, "-%s%s", in_reg
, in_regswizzle
);
2606 case WINED3DSPSM_NOT
:
2607 sprintf(out_str
, "!%s%s", in_reg
, in_regswizzle
);
2609 case WINED3DSPSM_BIAS
:
2610 sprintf(out_str
, "(%s%s - vec4(0.5)%s)", in_reg
, in_regswizzle
, in_regswizzle
);
2612 case WINED3DSPSM_BIASNEG
:
2613 sprintf(out_str
, "-(%s%s - vec4(0.5)%s)", in_reg
, in_regswizzle
, in_regswizzle
);
2615 case WINED3DSPSM_SIGN
:
2616 sprintf(out_str
, "(2.0 * (%s%s - 0.5))", in_reg
, in_regswizzle
);
2618 case WINED3DSPSM_SIGNNEG
:
2619 sprintf(out_str
, "-(2.0 * (%s%s - 0.5))", in_reg
, in_regswizzle
);
2621 case WINED3DSPSM_COMP
:
2622 sprintf(out_str
, "(1.0 - %s%s)", in_reg
, in_regswizzle
);
2624 case WINED3DSPSM_X2
:
2625 sprintf(out_str
, "(2.0 * %s%s)", in_reg
, in_regswizzle
);
2627 case WINED3DSPSM_X2NEG
:
2628 sprintf(out_str
, "-(2.0 * %s%s)", in_reg
, in_regswizzle
);
2630 case WINED3DSPSM_ABS
:
2631 sprintf(out_str
, "abs(%s%s)", in_reg
, in_regswizzle
);
2633 case WINED3DSPSM_ABSNEG
:
2634 sprintf(out_str
, "-abs(%s%s)", in_reg
, in_regswizzle
);
2637 FIXME("Unhandled modifier %u\n", src_modifier
);
2638 sprintf(out_str
, "%s%s", in_reg
, in_regswizzle
);
2642 static void shader_glsl_fixup_scalar_register_variable(struct wined3d_string_buffer
*register_name
,
2643 const char *glsl_variable
, const struct wined3d_gl_info
*gl_info
)
2645 /* The ARB_shading_language_420pack extension allows swizzle operations on
2647 if (gl_info
->supported
[ARB_SHADING_LANGUAGE_420PACK
])
2648 string_buffer_sprintf(register_name
, "%s", glsl_variable
);
2650 string_buffer_sprintf(register_name
, "ivec2(%s, 0)", glsl_variable
);
2653 /** Writes the GLSL variable name that corresponds to the register that the
2654 * DX opcode parameter is trying to access */
2655 static void shader_glsl_get_register_name(const struct wined3d_shader_register
*reg
,
2656 enum wined3d_data_type data_type
, struct wined3d_string_buffer
*register_name
,
2657 BOOL
*is_swizzled
, const struct wined3d_shader_context
*ctx
)
2659 /* oPos, oFog and oPts in D3D */
2660 static const char * const hwrastout_reg_names
[] = {"vs_out[10]", "vs_out[11].x", "vs_out[11].y"};
2662 const struct wined3d_shader
*shader
= ctx
->shader
;
2663 const struct wined3d_shader_reg_maps
*reg_maps
= ctx
->reg_maps
;
2664 const struct wined3d_shader_version
*version
= ®_maps
->shader_version
;
2665 const struct shader_glsl_ctx_priv
*priv
= ctx
->backend_data
;
2666 const char *prefix
= shader_glsl_get_prefix(version
->type
);
2667 const struct wined3d_gl_info
*gl_info
= priv
->gl_info
;
2668 struct glsl_src_param rel_param0
, rel_param1
;
2670 if (reg
->idx
[0].offset
!= ~0u && reg
->idx
[0].rel_addr
)
2671 shader_glsl_add_src_param_ext(ctx
, reg
->idx
[0].rel_addr
, WINED3DSP_WRITEMASK_0
,
2672 &rel_param0
, reg
->idx
[0].rel_addr
->reg
.data_type
);
2673 if (reg
->idx
[1].offset
!= ~0u && reg
->idx
[1].rel_addr
)
2674 shader_glsl_add_src_param_ext(ctx
, reg
->idx
[1].rel_addr
, WINED3DSP_WRITEMASK_0
,
2675 &rel_param1
, reg
->idx
[1].rel_addr
->reg
.data_type
);
2677 *is_swizzled
= FALSE
;
2681 case WINED3DSPR_TEMP
:
2682 string_buffer_sprintf(register_name
, "R%u", reg
->idx
[0].offset
);
2685 case WINED3DSPR_INPUT
:
2686 case WINED3DSPR_INCONTROLPOINT
:
2687 if (version
->type
== WINED3D_SHADER_TYPE_VERTEX
)
2689 if (reg
->idx
[0].rel_addr
)
2690 FIXME("VS3 input registers relative addressing.\n");
2691 if (is_swizzled
&& priv
->cur_vs_args
->swizzle_map
& (1u << reg
->idx
[0].offset
))
2692 *is_swizzled
= TRUE
;
2693 if (reg
->idx
[0].rel_addr
)
2695 string_buffer_sprintf(register_name
, "%s_in[%s + %u]",
2696 prefix
, rel_param0
.param_str
, reg
->idx
[0].offset
);
2700 string_buffer_sprintf(register_name
, "%s_in%u", prefix
, reg
->idx
[0].offset
);
2705 if (version
->type
== WINED3D_SHADER_TYPE_HULL
2706 || version
->type
== WINED3D_SHADER_TYPE_DOMAIN
2707 || version
->type
== WINED3D_SHADER_TYPE_GEOMETRY
)
2709 if (reg
->idx
[0].rel_addr
)
2711 if (reg
->idx
[1].rel_addr
)
2712 string_buffer_sprintf(register_name
, "shader_in[%s + %u].reg[%s + %u]",
2713 rel_param0
.param_str
, reg
->idx
[0].offset
,
2714 rel_param1
.param_str
, reg
->idx
[1].offset
);
2716 string_buffer_sprintf(register_name
, "shader_in[%s + %u].reg[%u]",
2717 rel_param0
.param_str
, reg
->idx
[0].offset
,
2718 reg
->idx
[1].offset
);
2720 else if (reg
->idx
[1].rel_addr
)
2721 string_buffer_sprintf(register_name
, "shader_in[%u].reg[%s + %u]", reg
->idx
[0].offset
,
2722 rel_param1
.param_str
, reg
->idx
[1].offset
);
2724 string_buffer_sprintf(register_name
, "shader_in[%u].reg[%u]",
2725 reg
->idx
[0].offset
, reg
->idx
[1].offset
);
2729 /* pixel shaders >= 3.0 */
2730 if (version
->major
>= 3)
2732 DWORD idx
= shader
->u
.ps
.input_reg_map
[reg
->idx
[0].offset
];
2733 unsigned int in_count
= vec4_varyings(version
->major
, gl_info
);
2735 if (reg
->idx
[0].rel_addr
)
2737 /* Removing a + 0 would be an obvious optimization, but
2738 * OS X doesn't see the NOP operation there. */
2741 if (needs_legacy_glsl_syntax(gl_info
)
2742 && shader
->u
.ps
.declared_in_count
> in_count
)
2744 string_buffer_sprintf(register_name
,
2745 "((%s + %u) > %u ? (%s + %u) > %u ? gl_SecondaryColor : gl_Color : %s_in[%s + %u])",
2746 rel_param0
.param_str
, idx
, in_count
- 1, rel_param0
.param_str
, idx
, in_count
,
2747 prefix
, rel_param0
.param_str
, idx
);
2751 string_buffer_sprintf(register_name
, "%s_in[%s + %u]", prefix
, rel_param0
.param_str
, idx
);
2756 if (needs_legacy_glsl_syntax(gl_info
)
2757 && shader
->u
.ps
.declared_in_count
> in_count
)
2759 string_buffer_sprintf(register_name
,
2760 "((%s) > %u ? (%s) > %u ? gl_SecondaryColor : gl_Color : %s_in[%s])",
2761 rel_param0
.param_str
, in_count
- 1, rel_param0
.param_str
, in_count
,
2762 prefix
, rel_param0
.param_str
);
2766 string_buffer_sprintf(register_name
, "%s_in[%s]", prefix
, rel_param0
.param_str
);
2772 if (idx
== in_count
)
2773 string_buffer_sprintf(register_name
, "gl_Color");
2774 else if (idx
== in_count
+ 1)
2775 string_buffer_sprintf(register_name
, "gl_SecondaryColor");
2777 string_buffer_sprintf(register_name
, "%s_in[%u]", prefix
, idx
);
2782 if (!reg
->idx
[0].offset
)
2783 string_buffer_sprintf(register_name
, "ffp_varying_diffuse");
2785 string_buffer_sprintf(register_name
, "ffp_varying_specular");
2790 case WINED3DSPR_CONST
:
2792 /* Relative addressing */
2793 if (reg
->idx
[0].rel_addr
)
2795 if (wined3d_settings
.check_float_constants
)
2796 string_buffer_sprintf(register_name
,
2797 "(%s + %u >= 0 && %s + %u < %u ? %s_c[%s + %u] : vec4(0.0))",
2798 rel_param0
.param_str
, reg
->idx
[0].offset
,
2799 rel_param0
.param_str
, reg
->idx
[0].offset
, shader
->limits
->constant_float
,
2800 prefix
, rel_param0
.param_str
, reg
->idx
[0].offset
);
2801 else if (reg
->idx
[0].offset
)
2802 string_buffer_sprintf(register_name
, "%s_c[%s + %u]",
2803 prefix
, rel_param0
.param_str
, reg
->idx
[0].offset
);
2805 string_buffer_sprintf(register_name
, "%s_c[%s]", prefix
, rel_param0
.param_str
);
2809 if (shader_constant_is_local(shader
, reg
->idx
[0].offset
))
2810 string_buffer_sprintf(register_name
, "%s_lc%u", prefix
, reg
->idx
[0].offset
);
2812 string_buffer_sprintf(register_name
, "%s_c[%u]", prefix
, reg
->idx
[0].offset
);
2817 case WINED3DSPR_CONSTINT
:
2818 string_buffer_sprintf(register_name
, "%s_i[%u]", prefix
, reg
->idx
[0].offset
);
2821 case WINED3DSPR_CONSTBOOL
:
2822 string_buffer_sprintf(register_name
, "%s_b[%u]", prefix
, reg
->idx
[0].offset
);
2825 case WINED3DSPR_TEXTURE
: /* case WINED3DSPR_ADDR: */
2826 if (version
->type
== WINED3D_SHADER_TYPE_PIXEL
)
2827 string_buffer_sprintf(register_name
, "T%u", reg
->idx
[0].offset
);
2829 string_buffer_sprintf(register_name
, "A%u", reg
->idx
[0].offset
);
2832 case WINED3DSPR_LOOP
:
2833 string_buffer_sprintf(register_name
, "aL%u", ctx
->state
->current_loop_reg
- 1);
2836 case WINED3DSPR_SAMPLER
:
2837 string_buffer_sprintf(register_name
, "%s_sampler%u", prefix
, reg
->idx
[0].offset
);
2840 case WINED3DSPR_COLOROUT
:
2841 if (reg
->idx
[0].offset
>= gl_info
->limits
.buffers
)
2842 WARN("Write to render target %u, only %d supported.\n",
2843 reg
->idx
[0].offset
, gl_info
->limits
.buffers
);
2845 string_buffer_sprintf(register_name
, "%s[%u]", get_fragment_output(gl_info
), reg
->idx
[0].offset
);
2848 case WINED3DSPR_RASTOUT
:
2849 string_buffer_sprintf(register_name
, "%s", hwrastout_reg_names
[reg
->idx
[0].offset
]);
2852 case WINED3DSPR_DEPTHOUT
:
2853 case WINED3DSPR_DEPTHOUTGE
:
2854 case WINED3DSPR_DEPTHOUTLE
:
2855 string_buffer_sprintf(register_name
, "gl_FragDepth");
2858 case WINED3DSPR_ATTROUT
:
2859 if (!reg
->idx
[0].offset
)
2860 string_buffer_sprintf(register_name
, "%s_out[8]", prefix
);
2862 string_buffer_sprintf(register_name
, "%s_out[9]", prefix
);
2865 case WINED3DSPR_TEXCRDOUT
:
2866 /* Vertex shaders >= 3.0: WINED3DSPR_OUTPUT */
2867 if (reg
->idx
[0].rel_addr
)
2868 string_buffer_sprintf(register_name
, "%s_out[%s + %u]",
2869 prefix
, rel_param0
.param_str
, reg
->idx
[0].offset
);
2871 string_buffer_sprintf(register_name
, "%s_out[%u]", prefix
, reg
->idx
[0].offset
);
2874 case WINED3DSPR_MISCTYPE
:
2875 if (!reg
->idx
[0].offset
)
2878 string_buffer_sprintf(register_name
, "vpos");
2880 else if (reg
->idx
[0].offset
== 1)
2882 /* Note that gl_FrontFacing is a bool, while vFace is
2883 * a float for which the sign determines front/back */
2884 string_buffer_sprintf(register_name
, "(gl_FrontFacing ? 1.0 : -1.0)");
2888 FIXME("Unhandled misctype register %u.\n", reg
->idx
[0].offset
);
2889 string_buffer_sprintf(register_name
, "unrecognized_register");
2893 case WINED3DSPR_IMMCONST
:
2894 switch (reg
->immconst_type
)
2896 case WINED3D_IMMCONST_SCALAR
:
2899 case WINED3D_DATA_UNORM
:
2900 case WINED3D_DATA_SNORM
:
2901 case WINED3D_DATA_FLOAT
:
2902 string_buffer_clear(register_name
);
2903 shader_glsl_append_imm_vec(register_name
, (const float *)reg
->u
.immconst_data
, 1, gl_info
);
2905 case WINED3D_DATA_INT
:
2906 string_buffer_sprintf(register_name
, "%#x", reg
->u
.immconst_data
[0]);
2908 case WINED3D_DATA_RESOURCE
:
2909 case WINED3D_DATA_SAMPLER
:
2910 case WINED3D_DATA_UINT
:
2911 string_buffer_sprintf(register_name
, "%#xu", reg
->u
.immconst_data
[0]);
2914 string_buffer_sprintf(register_name
, "<unhandled data type %#x>", data_type
);
2919 case WINED3D_IMMCONST_VEC4
:
2922 case WINED3D_DATA_UNORM
:
2923 case WINED3D_DATA_SNORM
:
2924 case WINED3D_DATA_FLOAT
:
2925 string_buffer_clear(register_name
);
2926 shader_glsl_append_imm_vec(register_name
, (const float *)reg
->u
.immconst_data
, 4, gl_info
);
2928 case WINED3D_DATA_INT
:
2929 string_buffer_sprintf(register_name
, "ivec4(%#x, %#x, %#x, %#x)",
2930 reg
->u
.immconst_data
[0], reg
->u
.immconst_data
[1],
2931 reg
->u
.immconst_data
[2], reg
->u
.immconst_data
[3]);
2933 case WINED3D_DATA_RESOURCE
:
2934 case WINED3D_DATA_SAMPLER
:
2935 case WINED3D_DATA_UINT
:
2936 string_buffer_sprintf(register_name
, "uvec4(%#xu, %#xu, %#xu, %#xu)",
2937 reg
->u
.immconst_data
[0], reg
->u
.immconst_data
[1],
2938 reg
->u
.immconst_data
[2], reg
->u
.immconst_data
[3]);
2941 string_buffer_sprintf(register_name
, "<unhandled data type %#x>", data_type
);
2947 FIXME("Unhandled immconst type %#x\n", reg
->immconst_type
);
2948 string_buffer_sprintf(register_name
, "<unhandled_immconst_type %#x>", reg
->immconst_type
);
2952 case WINED3DSPR_CONSTBUFFER
:
2953 if (reg
->idx
[1].rel_addr
)
2954 string_buffer_sprintf(register_name
, "%s_cb%u[%s + %u]",
2955 prefix
, reg
->idx
[0].offset
, rel_param1
.param_str
, reg
->idx
[1].offset
);
2957 string_buffer_sprintf(register_name
, "%s_cb%u[%u]", prefix
, reg
->idx
[0].offset
, reg
->idx
[1].offset
);
2960 case WINED3DSPR_IMMCONSTBUFFER
:
2961 if (reg
->idx
[0].rel_addr
)
2962 string_buffer_sprintf(register_name
, "%s_icb[%s + %u]",
2963 prefix
, rel_param0
.param_str
, reg
->idx
[0].offset
);
2965 string_buffer_sprintf(register_name
, "%s_icb[%u]", prefix
, reg
->idx
[0].offset
);
2968 case WINED3DSPR_PRIMID
:
2969 if (version
->type
== WINED3D_SHADER_TYPE_GEOMETRY
)
2970 string_buffer_sprintf(register_name
, "gl_PrimitiveIDIn");
2972 string_buffer_sprintf(register_name
, "gl_PrimitiveID");
2975 case WINED3DSPR_IDXTEMP
:
2976 if (reg
->idx
[1].rel_addr
)
2977 string_buffer_sprintf(register_name
, "X%u[%s + %u]",
2978 reg
->idx
[0].offset
, rel_param1
.param_str
, reg
->idx
[1].offset
);
2980 string_buffer_sprintf(register_name
, "X%u[%u]", reg
->idx
[0].offset
, reg
->idx
[1].offset
);
2983 case WINED3DSPR_LOCALTHREADINDEX
:
2984 shader_glsl_fixup_scalar_register_variable(register_name
,
2985 "int(gl_LocalInvocationIndex)", gl_info
);
2988 case WINED3DSPR_GSINSTID
:
2989 case WINED3DSPR_OUTPOINTID
:
2990 shader_glsl_fixup_scalar_register_variable(register_name
,
2991 "gl_InvocationID", gl_info
);
2994 case WINED3DSPR_THREADID
:
2995 string_buffer_sprintf(register_name
, "ivec3(gl_GlobalInvocationID)");
2998 case WINED3DSPR_THREADGROUPID
:
2999 string_buffer_sprintf(register_name
, "ivec3(gl_WorkGroupID)");
3002 case WINED3DSPR_LOCALTHREADID
:
3003 string_buffer_sprintf(register_name
, "ivec3(gl_LocalInvocationID)");
3006 case WINED3DSPR_FORKINSTID
:
3007 case WINED3DSPR_JOININSTID
:
3008 shader_glsl_fixup_scalar_register_variable(register_name
,
3009 "phase_instance_id", gl_info
);
3012 case WINED3DSPR_TESSCOORD
:
3013 string_buffer_sprintf(register_name
, "gl_TessCoord");
3016 case WINED3DSPR_OUTCONTROLPOINT
:
3017 if (reg
->idx
[0].rel_addr
)
3019 if (reg
->idx
[1].rel_addr
)
3020 string_buffer_sprintf(register_name
, "shader_out[%s + %u].reg[%s + %u]",
3021 rel_param0
.param_str
, reg
->idx
[0].offset
,
3022 rel_param1
.param_str
, reg
->idx
[1].offset
);
3024 string_buffer_sprintf(register_name
, "shader_out[%s + %u].reg[%u]",
3025 rel_param0
.param_str
, reg
->idx
[0].offset
,
3026 reg
->idx
[1].offset
);
3028 else if (reg
->idx
[1].rel_addr
)
3030 string_buffer_sprintf(register_name
, "shader_out[%u].reg[%s + %u]",
3031 reg
->idx
[0].offset
, rel_param1
.param_str
,
3032 reg
->idx
[1].offset
);
3036 string_buffer_sprintf(register_name
, "shader_out[%u].reg[%u]",
3037 reg
->idx
[0].offset
, reg
->idx
[1].offset
);
3041 case WINED3DSPR_PATCHCONST
:
3042 if (version
->type
== WINED3D_SHADER_TYPE_HULL
)
3043 string_buffer_sprintf(register_name
, "hs_out[%u]", reg
->idx
[0].offset
);
3045 string_buffer_sprintf(register_name
, "vpc[%u]", reg
->idx
[0].offset
);
3048 case WINED3DSPR_COVERAGE
:
3049 string_buffer_sprintf(register_name
, "gl_SampleMaskIn[0]");
3052 case WINED3DSPR_SAMPLEMASK
:
3053 string_buffer_sprintf(register_name
, "sample_mask");
3057 FIXME("Unhandled register type %#x.\n", reg
->type
);
3058 string_buffer_sprintf(register_name
, "unrecognised_register");
3063 static void shader_glsl_write_mask_to_str(DWORD write_mask
, char *str
)
3066 if (write_mask
& WINED3DSP_WRITEMASK_0
) *str
++ = 'x';
3067 if (write_mask
& WINED3DSP_WRITEMASK_1
) *str
++ = 'y';
3068 if (write_mask
& WINED3DSP_WRITEMASK_2
) *str
++ = 'z';
3069 if (write_mask
& WINED3DSP_WRITEMASK_3
) *str
++ = 'w';
3073 /* Get the GLSL write mask for the destination register */
3074 static DWORD
shader_glsl_get_write_mask(const struct wined3d_shader_dst_param
*param
, char *write_mask
)
3076 DWORD mask
= param
->write_mask
;
3078 if (shader_is_scalar(¶m
->reg
))
3080 mask
= WINED3DSP_WRITEMASK_0
;
3085 shader_glsl_write_mask_to_str(mask
, write_mask
);
3091 static unsigned int shader_glsl_get_write_mask_size(DWORD write_mask
)
3093 unsigned int size
= 0;
3095 if (write_mask
& WINED3DSP_WRITEMASK_0
) ++size
;
3096 if (write_mask
& WINED3DSP_WRITEMASK_1
) ++size
;
3097 if (write_mask
& WINED3DSP_WRITEMASK_2
) ++size
;
3098 if (write_mask
& WINED3DSP_WRITEMASK_3
) ++size
;
3103 static unsigned int shader_glsl_swizzle_get_component(DWORD swizzle
,
3104 unsigned int component_idx
)
3106 /* swizzle bits fields: wwzzyyxx */
3107 return (swizzle
>> (2 * component_idx
)) & 0x3;
3110 static void shader_glsl_swizzle_to_str(DWORD swizzle
, BOOL fixup
, DWORD mask
, char *str
)
3112 /* For registers of type WINED3DDECLTYPE_D3DCOLOR, data is stored as "bgra",
3113 * but addressed as "rgba". To fix this we need to swap the register's x
3114 * and z components. */
3115 const char *swizzle_chars
= fixup
? "zyxw" : "xyzw";
3119 for (i
= 0; i
< 4; ++i
)
3121 if (mask
& (WINED3DSP_WRITEMASK_0
<< i
))
3122 *str
++ = swizzle_chars
[shader_glsl_swizzle_get_component(swizzle
, i
)];
3127 static void shader_glsl_get_swizzle(const struct wined3d_shader_src_param
*param
,
3128 BOOL fixup
, DWORD mask
, char *swizzle_str
)
3130 if (shader_is_scalar(¶m
->reg
))
3131 *swizzle_str
= '\0';
3133 shader_glsl_swizzle_to_str(param
->swizzle
, fixup
, mask
, swizzle_str
);
3136 static void shader_glsl_sprintf_cast(struct wined3d_string_buffer
*dst_param
, const char *src_param
,
3137 enum wined3d_data_type dst_data_type
, enum wined3d_data_type src_data_type
, unsigned int size
)
3139 if (dst_data_type
== WINED3D_DATA_UNORM
|| dst_data_type
== WINED3D_DATA_SNORM
)
3140 dst_data_type
= WINED3D_DATA_FLOAT
;
3141 if (src_data_type
== WINED3D_DATA_UNORM
|| src_data_type
== WINED3D_DATA_SNORM
)
3142 src_data_type
= WINED3D_DATA_FLOAT
;
3144 if (dst_data_type
== src_data_type
)
3146 string_buffer_sprintf(dst_param
, "%s", src_param
);
3150 if (src_data_type
== WINED3D_DATA_FLOAT
)
3152 switch (dst_data_type
)
3154 case WINED3D_DATA_INT
:
3155 string_buffer_sprintf(dst_param
, "floatBitsToInt(%s)", src_param
);
3157 case WINED3D_DATA_RESOURCE
:
3158 case WINED3D_DATA_SAMPLER
:
3159 case WINED3D_DATA_UINT
:
3160 string_buffer_sprintf(dst_param
, "floatBitsToUint(%s)", src_param
);
3167 if (src_data_type
== WINED3D_DATA_UINT
&& dst_data_type
== WINED3D_DATA_FLOAT
)
3169 string_buffer_sprintf(dst_param
, "uintBitsToFloat(%s)", src_param
);
3173 if (src_data_type
== WINED3D_DATA_INT
)
3175 switch (dst_data_type
)
3177 case WINED3D_DATA_FLOAT
:
3178 string_buffer_sprintf(dst_param
, "intBitsToFloat(%s)", src_param
);
3180 case WINED3D_DATA_UINT
:
3182 string_buffer_sprintf(dst_param
, "uint(%s)", src_param
);
3184 string_buffer_sprintf(dst_param
, "uvec%u(%s)", size
, src_param
);
3191 FIXME("Unhandled cast from %#x to %#x.\n", src_data_type
, dst_data_type
);
3192 string_buffer_sprintf(dst_param
, "%s", src_param
);
3195 /* From a given parameter token, generate the corresponding GLSL string.
3196 * Also, return the actual register name and swizzle in case the
3197 * caller needs this information as well. */
3198 static void shader_glsl_add_src_param_ext(const struct wined3d_shader_context
*ctx
,
3199 const struct wined3d_shader_src_param
*wined3d_src
, DWORD mask
, struct glsl_src_param
*glsl_src
,
3200 enum wined3d_data_type data_type
)
3202 struct shader_glsl_ctx_priv
*priv
= ctx
->backend_data
;
3203 struct wined3d_string_buffer
*param_str
= string_buffer_get(priv
->string_buffers
);
3204 struct wined3d_string_buffer
*reg_name
= string_buffer_get(priv
->string_buffers
);
3205 enum wined3d_data_type param_data_type
;
3206 BOOL is_color
= FALSE
;
3207 char swizzle_str
[6];
3210 glsl_src
->param_str
[0] = '\0';
3211 swizzle_str
[0] = '\0';
3213 shader_glsl_get_register_name(&wined3d_src
->reg
, data_type
, reg_name
, &is_color
, ctx
);
3214 shader_glsl_get_swizzle(wined3d_src
, is_color
, mask
, swizzle_str
);
3216 switch (wined3d_src
->reg
.type
)
3218 case WINED3DSPR_IMMCONST
:
3219 param_data_type
= data_type
;
3220 size
= wined3d_src
->reg
.immconst_type
== WINED3D_IMMCONST_SCALAR
? 1 : 4;
3222 case WINED3DSPR_FORKINSTID
:
3223 case WINED3DSPR_GSINSTID
:
3224 case WINED3DSPR_JOININSTID
:
3225 case WINED3DSPR_LOCALTHREADINDEX
:
3226 case WINED3DSPR_OUTPOINTID
:
3227 case WINED3DSPR_PRIMID
:
3228 param_data_type
= WINED3D_DATA_INT
;
3231 case WINED3DSPR_LOCALTHREADID
:
3232 case WINED3DSPR_THREADGROUPID
:
3233 case WINED3DSPR_THREADID
:
3234 param_data_type
= WINED3D_DATA_INT
;
3238 param_data_type
= WINED3D_DATA_FLOAT
;
3243 shader_glsl_sprintf_cast(param_str
, reg_name
->buffer
, data_type
, param_data_type
, size
);
3244 shader_glsl_gen_modifier(wined3d_src
->modifiers
, param_str
->buffer
, swizzle_str
, glsl_src
->param_str
);
3246 string_buffer_release(priv
->string_buffers
, reg_name
);
3247 string_buffer_release(priv
->string_buffers
, param_str
);
3250 static void shader_glsl_add_src_param(const struct wined3d_shader_instruction
*ins
,
3251 const struct wined3d_shader_src_param
*wined3d_src
, DWORD mask
, struct glsl_src_param
*glsl_src
)
3253 shader_glsl_add_src_param_ext(ins
->ctx
, wined3d_src
, mask
, glsl_src
, wined3d_src
->reg
.data_type
);
3256 /* From a given parameter token, generate the corresponding GLSL string.
3257 * Also, return the actual register name and swizzle in case the
3258 * caller needs this information as well. */
3259 static DWORD
shader_glsl_add_dst_param(const struct wined3d_shader_instruction
*ins
,
3260 const struct wined3d_shader_dst_param
*wined3d_dst
, struct glsl_dst_param
*glsl_dst
)
3262 struct shader_glsl_ctx_priv
*priv
= ins
->ctx
->backend_data
;
3263 struct wined3d_string_buffer
*reg_name
;
3266 glsl_dst
->mask_str
[0] = '\0';
3268 reg_name
= string_buffer_get(priv
->string_buffers
);
3269 shader_glsl_get_register_name(&wined3d_dst
->reg
, wined3d_dst
->reg
.data_type
, reg_name
, NULL
, ins
->ctx
);
3270 len
= min(reg_name
->content_size
, ARRAY_SIZE(glsl_dst
->reg_name
) - 1);
3271 memcpy(glsl_dst
->reg_name
, reg_name
->buffer
, len
);
3272 glsl_dst
->reg_name
[len
] = '\0';
3273 string_buffer_release(priv
->string_buffers
, reg_name
);
3275 return shader_glsl_get_write_mask(wined3d_dst
, glsl_dst
->mask_str
);
3278 /* Append the destination part of the instruction to the buffer, return the effective write mask */
3279 static DWORD
shader_glsl_append_dst_ext(struct wined3d_string_buffer
*buffer
,
3280 const struct wined3d_shader_instruction
*ins
, const struct wined3d_shader_dst_param
*dst
,
3281 unsigned int dst_idx
, enum wined3d_data_type data_type
)
3283 struct glsl_dst_param glsl_dst
;
3286 if ((mask
= shader_glsl_add_dst_param(ins
, dst
, &glsl_dst
)))
3288 if (ins
->flags
& WINED3DSI_PRECISE_XYZW
)
3289 sprintf(glsl_dst
.reg_name
, "tmp_precise[%u]", dst_idx
);
3293 case WINED3D_DATA_FLOAT
:
3294 case WINED3D_DATA_UNORM
:
3295 case WINED3D_DATA_SNORM
:
3296 shader_addline(buffer
, "%s%s = %s(",
3297 glsl_dst
.reg_name
, glsl_dst
.mask_str
, shift_glsl_tab
[dst
->shift
]);
3299 case WINED3D_DATA_INT
:
3300 shader_addline(buffer
, "%s%s = %sintBitsToFloat(",
3301 glsl_dst
.reg_name
, glsl_dst
.mask_str
, shift_glsl_tab
[dst
->shift
]);
3303 case WINED3D_DATA_RESOURCE
:
3304 case WINED3D_DATA_SAMPLER
:
3305 case WINED3D_DATA_UINT
:
3306 shader_addline(buffer
, "%s%s = %suintBitsToFloat(",
3307 glsl_dst
.reg_name
, glsl_dst
.mask_str
, shift_glsl_tab
[dst
->shift
]);
3310 FIXME("Unhandled data type %#x.\n", data_type
);
3311 shader_addline(buffer
, "%s%s = %s(",
3312 glsl_dst
.reg_name
, glsl_dst
.mask_str
, shift_glsl_tab
[dst
->shift
]);
3320 /* Append the destination part of the instruction to the buffer, return the effective write mask */
3321 static DWORD
shader_glsl_append_dst(struct wined3d_string_buffer
*buffer
, const struct wined3d_shader_instruction
*ins
)
3323 return shader_glsl_append_dst_ext(buffer
, ins
, &ins
->dst
[0], 0, ins
->dst
[0].reg
.data_type
);
3326 /** Process GLSL instruction modifiers */
3327 static void shader_glsl_add_instruction_modifiers(const struct wined3d_shader_instruction
*ins
)
3329 const struct wined3d_shader_dst_param
*dst
;
3330 struct glsl_dst_param dst_param
;
3334 for (i
= 0; i
< ins
->dst_count
; ++i
)
3336 if ((dst
= &ins
->dst
[i
])->reg
.type
== WINED3DSPR_NULL
)
3339 if (ins
->flags
& WINED3DSI_PRECISE_XYZW
)
3341 shader_glsl_add_dst_param(ins
, dst
, &dst_param
);
3342 shader_addline(ins
->ctx
->buffer
, "%s%s = tmp_precise[%u]%s;\n",
3343 dst_param
.reg_name
, dst_param
.mask_str
, i
, dst_param
.mask_str
);
3346 if (!(modifiers
= dst
->modifiers
))
3349 shader_glsl_add_dst_param(ins
, dst
, &dst_param
);
3351 if (modifiers
& WINED3DSPDM_SATURATE
)
3353 /* _SAT means to clamp the value of the register to between 0 and 1 */
3354 shader_addline(ins
->ctx
->buffer
, "%s%s = clamp(%s%s, 0.0, 1.0);\n", dst_param
.reg_name
,
3355 dst_param
.mask_str
, dst_param
.reg_name
, dst_param
.mask_str
);
3358 if (modifiers
& WINED3DSPDM_MSAMPCENTROID
)
3360 FIXME("_centroid modifier not handled\n");
3363 if (modifiers
& WINED3DSPDM_PARTIALPRECISION
)
3365 /* MSDN says this modifier can be safely ignored, so that's what we'll do. */
3370 static const char *shader_glsl_get_rel_op(enum wined3d_shader_rel_op op
)
3374 case WINED3D_SHADER_REL_OP_GT
: return ">";
3375 case WINED3D_SHADER_REL_OP_EQ
: return "==";
3376 case WINED3D_SHADER_REL_OP_GE
: return ">=";
3377 case WINED3D_SHADER_REL_OP_LT
: return "<";
3378 case WINED3D_SHADER_REL_OP_NE
: return "!=";
3379 case WINED3D_SHADER_REL_OP_LE
: return "<=";
3381 FIXME("Unrecognized operator %#x.\n", op
);
3386 static BOOL
shader_glsl_has_core_grad(const struct wined3d_gl_info
*gl_info
)
3388 return shader_glsl_get_version(gl_info
) >= 130 || gl_info
->supported
[EXT_GPU_SHADER4
];
3391 static void shader_glsl_get_coord_size(enum wined3d_shader_resource_type resource_type
,
3392 unsigned int *coord_size
, unsigned int *deriv_size
)
3394 const BOOL is_array
= resource_type
== WINED3D_SHADER_RESOURCE_TEXTURE_1DARRAY
3395 || resource_type
== WINED3D_SHADER_RESOURCE_TEXTURE_2DARRAY
;
3397 *coord_size
= resource_type_info
[resource_type
].coord_size
;
3398 *deriv_size
= *coord_size
;
3403 static void shader_glsl_get_sample_function(const struct wined3d_shader_context
*ctx
,
3404 DWORD resource_idx
, DWORD sampler_idx
, DWORD flags
, struct glsl_sample_function
*sample_function
)
3406 enum wined3d_shader_resource_type resource_type
;
3407 struct shader_glsl_ctx_priv
*priv
= ctx
->backend_data
;
3408 const struct wined3d_gl_info
*gl_info
= priv
->gl_info
;
3409 BOOL shadow
= glsl_is_shadow_sampler(ctx
->shader
, priv
->cur_ps_args
, resource_idx
, sampler_idx
);
3410 BOOL projected
= flags
& WINED3D_GLSL_SAMPLE_PROJECTED
;
3411 BOOL texrect
= ctx
->reg_maps
->shader_version
.type
== WINED3D_SHADER_TYPE_PIXEL
3412 && priv
->cur_ps_args
->np2_fixup
& (1u << resource_idx
)
3413 && gl_info
->supported
[ARB_TEXTURE_RECTANGLE
];
3414 BOOL lod
= flags
& WINED3D_GLSL_SAMPLE_LOD
;
3415 BOOL grad
= flags
& WINED3D_GLSL_SAMPLE_GRAD
;
3416 BOOL offset
= flags
& WINED3D_GLSL_SAMPLE_OFFSET
;
3417 const char *base
= "texture", *type_part
= "", *suffix
= "";
3418 unsigned int coord_size
, deriv_size
;
3420 resource_type
= ctx
->reg_maps
->shader_version
.type
== WINED3D_SHADER_TYPE_PIXEL
3421 ? pixelshader_get_resource_type(ctx
->reg_maps
, resource_idx
, priv
->cur_ps_args
->tex_types
)
3422 : ctx
->reg_maps
->resource_info
[resource_idx
].type
;
3424 sample_function
->data_type
= ctx
->reg_maps
->resource_info
[resource_idx
].data_type
;
3426 if (resource_type
>= ARRAY_SIZE(resource_type_info
))
3428 ERR("Unexpected resource type %#x.\n", resource_type
);
3429 resource_type
= WINED3D_SHADER_RESOURCE_TEXTURE_2D
;
3432 /* Note that there's no such thing as a projected cube texture. */
3433 if (resource_type
== WINED3D_SHADER_RESOURCE_TEXTURE_CUBE
)
3436 if (needs_legacy_glsl_syntax(gl_info
))
3441 type_part
= resource_type_info
[resource_type
].type_part
;
3442 if (resource_type
== WINED3D_SHADER_RESOURCE_TEXTURE_2D
&& texrect
)
3443 type_part
= "2DRect";
3444 if (!type_part
[0] && resource_type
!= WINED3D_SHADER_RESOURCE_TEXTURE_CUBEARRAY
)
3445 FIXME("Unhandled resource type %#x.\n", resource_type
);
3447 if (!lod
&& grad
&& !shader_glsl_has_core_grad(gl_info
))
3449 if (gl_info
->supported
[ARB_SHADER_TEXTURE_LOD
])
3452 FIXME("Unsupported grad function.\n");
3456 if (flags
& WINED3D_GLSL_SAMPLE_LOAD
)
3458 static const DWORD texel_fetch_flags
= WINED3D_GLSL_SAMPLE_LOAD
| WINED3D_GLSL_SAMPLE_OFFSET
;
3459 if (flags
& ~texel_fetch_flags
)
3460 ERR("Unexpected flags %#x for texelFetch.\n", flags
& ~texel_fetch_flags
);
3462 base
= "texelFetch";
3466 sample_function
->name
= string_buffer_get(priv
->string_buffers
);
3467 string_buffer_sprintf(sample_function
->name
, "%s%s%s%s%s%s", base
, type_part
, projected
? "Proj" : "",
3468 lod
? "Lod" : grad
? "Grad" : "", offset
? "Offset" : "", suffix
);
3470 shader_glsl_get_coord_size(resource_type
, &coord_size
, &deriv_size
);
3473 sample_function
->offset_size
= offset
? deriv_size
: 0;
3474 sample_function
->coord_mask
= wined3d_mask_from_size(coord_size
);
3475 sample_function
->deriv_mask
= wined3d_mask_from_size(deriv_size
);
3476 sample_function
->output_single_component
= shadow
&& !needs_legacy_glsl_syntax(gl_info
);
3479 static void shader_glsl_release_sample_function(const struct wined3d_shader_context
*ctx
,
3480 struct glsl_sample_function
*sample_function
)
3482 const struct shader_glsl_ctx_priv
*priv
= ctx
->backend_data
;
3484 string_buffer_release(priv
->string_buffers
, sample_function
->name
);
3487 static void shader_glsl_append_fixup_arg(char *arguments
, const char *reg_name
,
3488 BOOL sign_fixup
, enum fixup_channel_source channel_source
)
3490 switch(channel_source
)
3492 case CHANNEL_SOURCE_ZERO
:
3493 strcat(arguments
, "0.0");
3496 case CHANNEL_SOURCE_ONE
:
3497 strcat(arguments
, "1.0");
3500 case CHANNEL_SOURCE_X
:
3501 strcat(arguments
, reg_name
);
3502 strcat(arguments
, ".x");
3505 case CHANNEL_SOURCE_Y
:
3506 strcat(arguments
, reg_name
);
3507 strcat(arguments
, ".y");
3510 case CHANNEL_SOURCE_Z
:
3511 strcat(arguments
, reg_name
);
3512 strcat(arguments
, ".z");
3515 case CHANNEL_SOURCE_W
:
3516 strcat(arguments
, reg_name
);
3517 strcat(arguments
, ".w");
3521 FIXME("Unhandled channel source %#x\n", channel_source
);
3522 strcat(arguments
, "undefined");
3526 if (sign_fixup
) strcat(arguments
, " * 2.0 - 1.0");
3529 static void shader_glsl_color_correction_ext(struct wined3d_string_buffer
*buffer
,
3530 const char *reg_name
, DWORD mask
, struct color_fixup_desc fixup
)
3532 unsigned int mask_size
, remaining
;
3533 DWORD fixup_mask
= 0;
3534 char arguments
[256];
3537 if (fixup
.x_sign_fixup
|| fixup
.x_source
!= CHANNEL_SOURCE_X
) fixup_mask
|= WINED3DSP_WRITEMASK_0
;
3538 if (fixup
.y_sign_fixup
|| fixup
.y_source
!= CHANNEL_SOURCE_Y
) fixup_mask
|= WINED3DSP_WRITEMASK_1
;
3539 if (fixup
.z_sign_fixup
|| fixup
.z_source
!= CHANNEL_SOURCE_Z
) fixup_mask
|= WINED3DSP_WRITEMASK_2
;
3540 if (fixup
.w_sign_fixup
|| fixup
.w_source
!= CHANNEL_SOURCE_W
) fixup_mask
|= WINED3DSP_WRITEMASK_3
;
3541 if (!(mask
&= fixup_mask
))
3544 if (is_complex_fixup(fixup
))
3546 enum complex_fixup complex_fixup
= get_complex_fixup(fixup
);
3547 FIXME("Complex fixup (%#x) not supported\n",complex_fixup
);
3551 shader_glsl_write_mask_to_str(mask
, mask_str
);
3552 mask_size
= shader_glsl_get_write_mask_size(mask
);
3554 arguments
[0] = '\0';
3555 remaining
= mask_size
;
3556 if (mask
& WINED3DSP_WRITEMASK_0
)
3558 shader_glsl_append_fixup_arg(arguments
, reg_name
, fixup
.x_sign_fixup
, fixup
.x_source
);
3559 if (--remaining
) strcat(arguments
, ", ");
3561 if (mask
& WINED3DSP_WRITEMASK_1
)
3563 shader_glsl_append_fixup_arg(arguments
, reg_name
, fixup
.y_sign_fixup
, fixup
.y_source
);
3564 if (--remaining
) strcat(arguments
, ", ");
3566 if (mask
& WINED3DSP_WRITEMASK_2
)
3568 shader_glsl_append_fixup_arg(arguments
, reg_name
, fixup
.z_sign_fixup
, fixup
.z_source
);
3569 if (--remaining
) strcat(arguments
, ", ");
3571 if (mask
& WINED3DSP_WRITEMASK_3
)
3573 shader_glsl_append_fixup_arg(arguments
, reg_name
, fixup
.w_sign_fixup
, fixup
.w_source
);
3574 if (--remaining
) strcat(arguments
, ", ");
3578 shader_addline(buffer
, "%s%s = vec%u(%s);\n", reg_name
, mask_str
, mask_size
, arguments
);
3580 shader_addline(buffer
, "%s%s = %s;\n", reg_name
, mask_str
, arguments
);
3583 static void shader_glsl_color_correction(const struct wined3d_shader_instruction
*ins
, struct color_fixup_desc fixup
)
3585 struct shader_glsl_ctx_priv
*priv
= ins
->ctx
->backend_data
;
3586 struct wined3d_string_buffer
*reg_name
;
3588 reg_name
= string_buffer_get(priv
->string_buffers
);
3589 shader_glsl_get_register_name(&ins
->dst
[0].reg
, ins
->dst
[0].reg
.data_type
, reg_name
, NULL
, ins
->ctx
);
3590 shader_glsl_color_correction_ext(ins
->ctx
->buffer
, reg_name
->buffer
, ins
->dst
[0].write_mask
, fixup
);
3591 string_buffer_release(priv
->string_buffers
, reg_name
);
3594 static void PRINTF_ATTR(9, 10) shader_glsl_gen_sample_code(const struct wined3d_shader_instruction
*ins
,
3595 unsigned int sampler_bind_idx
, const struct glsl_sample_function
*sample_function
, DWORD swizzle
,
3596 const char *dx
, const char *dy
, const char *bias
, const struct wined3d_shader_texel_offset
*offset
,
3597 const char *coord_reg_fmt
, ...)
3599 const struct wined3d_shader_version
*version
= &ins
->ctx
->reg_maps
->shader_version
;
3600 char dst_swizzle
[6];
3601 struct color_fixup_desc fixup
;
3602 BOOL np2_fixup
= FALSE
;
3606 shader_glsl_swizzle_to_str(swizzle
, FALSE
, ins
->dst
[0].write_mask
, dst_swizzle
);
3608 /* If ARB_texture_swizzle is supported we don't need to do anything here.
3609 * We actually rely on it for vertex shaders and SM4+. */
3610 if (version
->type
== WINED3D_SHADER_TYPE_PIXEL
&& version
->major
< 4)
3612 const struct shader_glsl_ctx_priv
*priv
= ins
->ctx
->backend_data
;
3613 fixup
= priv
->cur_ps_args
->color_fixup
[sampler_bind_idx
];
3615 if (priv
->cur_ps_args
->np2_fixup
& (1u << sampler_bind_idx
))
3620 fixup
= COLOR_FIXUP_IDENTITY
;
3623 shader_glsl_append_dst_ext(ins
->ctx
->buffer
, ins
, &ins
->dst
[0], 0, sample_function
->data_type
);
3625 if (sample_function
->output_single_component
)
3626 shader_addline(ins
->ctx
->buffer
, "vec4(");
3628 shader_addline(ins
->ctx
->buffer
, "%s(%s_sampler%u, ",
3629 sample_function
->name
->buffer
, shader_glsl_get_prefix(version
->type
), sampler_bind_idx
);
3633 va_start(args
, coord_reg_fmt
);
3634 ret
= shader_vaddline(ins
->ctx
->buffer
, coord_reg_fmt
, args
);
3638 if (!string_buffer_resize(ins
->ctx
->buffer
, ret
))
3644 const struct shader_glsl_ctx_priv
*priv
= ins
->ctx
->backend_data
;
3645 const unsigned char idx
= priv
->cur_np2fixup_info
->idx
[sampler_bind_idx
];
3647 switch (shader_glsl_get_write_mask_size(sample_function
->coord_mask
))
3650 shader_addline(ins
->ctx
->buffer
, " * ps_samplerNP2Fixup[%u].%s",
3651 idx
>> 1, (idx
% 2) ? "z" : "x");
3654 shader_addline(ins
->ctx
->buffer
, " * ps_samplerNP2Fixup[%u].%s",
3655 idx
>> 1, (idx
% 2) ? "zw" : "xy");
3658 shader_addline(ins
->ctx
->buffer
, " * vec3(ps_samplerNP2Fixup[%u].%s, 1.0)",
3659 idx
>> 1, (idx
% 2) ? "zw" : "xy");
3662 shader_addline(ins
->ctx
->buffer
, " * vec4(ps_samplerNP2Fixup[%u].%s, 1.0, 1.0)",
3663 idx
>> 1, (idx
% 2) ? "zw" : "xy");
3668 shader_addline(ins
->ctx
->buffer
, ", %s, %s", dx
, dy
);
3670 shader_addline(ins
->ctx
->buffer
, ", %s", bias
);
3671 if (sample_function
->offset_size
)
3673 int offset_immdata
[4] = {offset
->u
, offset
->v
, offset
->w
};
3674 shader_addline(ins
->ctx
->buffer
, ", ");
3675 shader_glsl_append_imm_ivec(ins
->ctx
->buffer
, offset_immdata
, sample_function
->offset_size
);
3677 shader_addline(ins
->ctx
->buffer
, ")");
3679 if (sample_function
->output_single_component
)
3680 shader_addline(ins
->ctx
->buffer
, ")");
3682 shader_addline(ins
->ctx
->buffer
, "%s);\n", dst_swizzle
);
3684 if (!is_identity_fixup(fixup
))
3685 shader_glsl_color_correction(ins
, fixup
);
3688 static void shader_glsl_fixup_position(struct wined3d_string_buffer
*buffer
, BOOL use_viewport_index
)
3690 /* Write the final position.
3692 * OpenGL coordinates specify the center of the pixel while D3D coords
3693 * specify the corner. The offsets are stored in z and w in
3694 * pos_fixup. pos_fixup.y contains 1.0 or -1.0 to turn the rendering
3695 * upside down for offscreen rendering. pos_fixup.x contains 1.0 to allow
3697 if (use_viewport_index
)
3699 shader_addline(buffer
, "gl_Position.y = gl_Position.y * pos_fixup[gl_ViewportIndex].y;\n");
3700 shader_addline(buffer
, "gl_Position.xy += pos_fixup[gl_ViewportIndex].zw * gl_Position.ww;\n");
3704 shader_addline(buffer
, "gl_Position.y = gl_Position.y * pos_fixup.y;\n");
3705 shader_addline(buffer
, "gl_Position.xy += pos_fixup.zw * gl_Position.ww;\n");
3708 /* Z coord [0;1]->[-1;1] mapping, see comment in get_projection_matrix()
3711 * Basically we want (in homogeneous coordinates) z = z * 2 - 1. However,
3712 * shaders are run before the homogeneous divide, so we have to take the w
3713 * into account: z = ((z / w) * 2 - 1) * w, which is the same as
3715 shader_addline(buffer
, "gl_Position.z = gl_Position.z * 2.0 - gl_Position.w;\n");
3718 /*****************************************************************************
3719 * Begin processing individual instruction opcodes
3720 ****************************************************************************/
3722 static void shader_glsl_binop(const struct wined3d_shader_instruction
*ins
)
3724 struct wined3d_string_buffer
*buffer
= ins
->ctx
->buffer
;
3725 struct glsl_src_param src0_param
;
3726 struct glsl_src_param src1_param
;
3730 /* Determine the GLSL operator to use based on the opcode */
3731 switch (ins
->handler_idx
)
3733 case WINED3DSIH_ADD
: op
= "+"; break;
3734 case WINED3DSIH_AND
: op
= "&"; break;
3735 case WINED3DSIH_DIV
: op
= "/"; break;
3736 case WINED3DSIH_IADD
: op
= "+"; break;
3737 case WINED3DSIH_ISHL
: op
= "<<"; break;
3738 case WINED3DSIH_ISHR
: op
= ">>"; break;
3739 case WINED3DSIH_MUL
: op
= "*"; break;
3740 case WINED3DSIH_OR
: op
= "|"; break;
3741 case WINED3DSIH_SUB
: op
= "-"; break;
3742 case WINED3DSIH_USHR
: op
= ">>"; break;
3743 case WINED3DSIH_XOR
: op
= "^"; break;
3745 op
= "<unhandled operator>";
3746 FIXME("Opcode %s not yet handled in GLSL.\n", debug_d3dshaderinstructionhandler(ins
->handler_idx
));
3750 write_mask
= shader_glsl_append_dst(buffer
, ins
);
3751 shader_glsl_add_src_param(ins
, &ins
->src
[0], write_mask
, &src0_param
);
3752 shader_glsl_add_src_param(ins
, &ins
->src
[1], write_mask
, &src1_param
);
3753 shader_addline(buffer
, "%s %s %s);\n", src0_param
.param_str
, op
, src1_param
.param_str
);
3756 static void shader_glsl_relop(const struct wined3d_shader_instruction
*ins
)
3758 struct wined3d_string_buffer
*buffer
= ins
->ctx
->buffer
;
3759 struct glsl_src_param src0_param
;
3760 struct glsl_src_param src1_param
;
3761 unsigned int mask_size
;
3765 write_mask
= shader_glsl_append_dst(buffer
, ins
);
3766 mask_size
= shader_glsl_get_write_mask_size(write_mask
);
3767 shader_glsl_add_src_param(ins
, &ins
->src
[0], write_mask
, &src0_param
);
3768 shader_glsl_add_src_param(ins
, &ins
->src
[1], write_mask
, &src1_param
);
3772 switch (ins
->handler_idx
)
3774 case WINED3DSIH_EQ
: op
= "equal"; break;
3775 case WINED3DSIH_IEQ
: op
= "equal"; break;
3776 case WINED3DSIH_GE
: op
= "greaterThanEqual"; break;
3777 case WINED3DSIH_IGE
: op
= "greaterThanEqual"; break;
3778 case WINED3DSIH_UGE
: op
= "greaterThanEqual"; break;
3779 case WINED3DSIH_LT
: op
= "lessThan"; break;
3780 case WINED3DSIH_ILT
: op
= "lessThan"; break;
3781 case WINED3DSIH_ULT
: op
= "lessThan"; break;
3782 case WINED3DSIH_NE
: op
= "notEqual"; break;
3783 case WINED3DSIH_INE
: op
= "notEqual"; break;
3785 op
= "<unhandled operator>";
3786 ERR("Unhandled opcode %#x.\n", ins
->handler_idx
);
3790 shader_addline(buffer
, "uvec%u(%s(%s, %s)) * 0xffffffffu);\n",
3791 mask_size
, op
, src0_param
.param_str
, src1_param
.param_str
);
3795 switch (ins
->handler_idx
)
3797 case WINED3DSIH_EQ
: op
= "=="; break;
3798 case WINED3DSIH_IEQ
: op
= "=="; break;
3799 case WINED3DSIH_GE
: op
= ">="; break;
3800 case WINED3DSIH_IGE
: op
= ">="; break;
3801 case WINED3DSIH_UGE
: op
= ">="; break;
3802 case WINED3DSIH_LT
: op
= "<"; break;
3803 case WINED3DSIH_ILT
: op
= "<"; break;
3804 case WINED3DSIH_ULT
: op
= "<"; break;
3805 case WINED3DSIH_NE
: op
= "!="; break;
3806 case WINED3DSIH_INE
: op
= "!="; break;
3808 op
= "<unhandled operator>";
3809 ERR("Unhandled opcode %#x.\n", ins
->handler_idx
);
3813 shader_addline(buffer
, "%s %s %s ? 0xffffffffu : 0u);\n",
3814 src0_param
.param_str
, op
, src1_param
.param_str
);
3818 static void shader_glsl_unary_op(const struct wined3d_shader_instruction
*ins
)
3820 struct glsl_src_param src_param
;
3824 switch (ins
->handler_idx
)
3826 case WINED3DSIH_INEG
: op
= "-"; break;
3827 case WINED3DSIH_NOT
: op
= "~"; break;
3829 op
= "<unhandled operator>";
3830 ERR("Unhandled opcode %s.\n",
3831 debug_d3dshaderinstructionhandler(ins
->handler_idx
));
3835 write_mask
= shader_glsl_append_dst(ins
->ctx
->buffer
, ins
);
3836 shader_glsl_add_src_param(ins
, &ins
->src
[0], write_mask
, &src_param
);
3837 shader_addline(ins
->ctx
->buffer
, "%s%s);\n", op
, src_param
.param_str
);
3840 static void shader_glsl_mul_extended(const struct wined3d_shader_instruction
*ins
)
3842 struct wined3d_string_buffer
*buffer
= ins
->ctx
->buffer
;
3843 struct glsl_src_param src0_param
;
3844 struct glsl_src_param src1_param
;
3847 /* If we have ARB_gpu_shader5, we can use imulExtended() / umulExtended().
3848 * If not, we can emulate it. */
3849 if (ins
->dst
[0].reg
.type
!= WINED3DSPR_NULL
)
3850 FIXME("64-bit integer multiplies not implemented.\n");
3852 if (ins
->dst
[1].reg
.type
!= WINED3DSPR_NULL
)
3854 write_mask
= shader_glsl_append_dst_ext(buffer
, ins
, &ins
->dst
[1], 1, ins
->dst
[1].reg
.data_type
);
3855 shader_glsl_add_src_param(ins
, &ins
->src
[0], write_mask
, &src0_param
);
3856 shader_glsl_add_src_param(ins
, &ins
->src
[1], write_mask
, &src1_param
);
3858 shader_addline(ins
->ctx
->buffer
, "%s * %s);\n",
3859 src0_param
.param_str
, src1_param
.param_str
);
3863 static void shader_glsl_udiv(const struct wined3d_shader_instruction
*ins
)
3865 struct wined3d_string_buffer
*buffer
= ins
->ctx
->buffer
;
3866 struct glsl_src_param src0_param
, src1_param
;
3869 if (ins
->dst
[0].reg
.type
!= WINED3DSPR_NULL
)
3871 if (ins
->dst
[1].reg
.type
!= WINED3DSPR_NULL
)
3875 write_mask
= shader_glsl_get_write_mask(&ins
->dst
[0], dst_mask
);
3876 shader_glsl_add_src_param(ins
, &ins
->src
[0], write_mask
, &src0_param
);
3877 shader_glsl_add_src_param(ins
, &ins
->src
[1], write_mask
, &src1_param
);
3878 shader_addline(buffer
, "tmp0%s = uintBitsToFloat(%s / %s);\n",
3879 dst_mask
, src0_param
.param_str
, src1_param
.param_str
);
3881 write_mask
= shader_glsl_append_dst_ext(buffer
, ins
, &ins
->dst
[1], 1, ins
->dst
[1].reg
.data_type
);
3882 shader_glsl_add_src_param(ins
, &ins
->src
[0], write_mask
, &src0_param
);
3883 shader_glsl_add_src_param(ins
, &ins
->src
[1], write_mask
, &src1_param
);
3884 shader_addline(buffer
, "%s %% %s);\n", src0_param
.param_str
, src1_param
.param_str
);
3886 shader_glsl_append_dst_ext(buffer
, ins
, &ins
->dst
[0], 0, WINED3D_DATA_FLOAT
);
3887 shader_addline(buffer
, "tmp0%s);\n", dst_mask
);
3891 write_mask
= shader_glsl_append_dst_ext(buffer
, ins
, &ins
->dst
[0], 0, ins
->dst
[0].reg
.data_type
);
3892 shader_glsl_add_src_param(ins
, &ins
->src
[0], write_mask
, &src0_param
);
3893 shader_glsl_add_src_param(ins
, &ins
->src
[1], write_mask
, &src1_param
);
3894 shader_addline(buffer
, "%s / %s);\n", src0_param
.param_str
, src1_param
.param_str
);
3897 else if (ins
->dst
[1].reg
.type
!= WINED3DSPR_NULL
)
3899 write_mask
= shader_glsl_append_dst_ext(buffer
, ins
, &ins
->dst
[1], 1, ins
->dst
[1].reg
.data_type
);
3900 shader_glsl_add_src_param(ins
, &ins
->src
[0], write_mask
, &src0_param
);
3901 shader_glsl_add_src_param(ins
, &ins
->src
[1], write_mask
, &src1_param
);
3902 shader_addline(buffer
, "%s %% %s);\n", src0_param
.param_str
, src1_param
.param_str
);
3906 /* Process the WINED3DSIO_MOV opcode using GLSL (dst = src) */
3907 static void shader_glsl_mov(const struct wined3d_shader_instruction
*ins
)
3909 const struct shader_glsl_ctx_priv
*priv
= ins
->ctx
->backend_data
;
3910 struct wined3d_string_buffer
*buffer
= ins
->ctx
->buffer
;
3911 const struct wined3d_gl_info
*gl_info
= priv
->gl_info
;
3912 struct glsl_src_param src0_param
;
3915 write_mask
= shader_glsl_append_dst(buffer
, ins
);
3916 shader_glsl_add_src_param(ins
, &ins
->src
[0], write_mask
, &src0_param
);
3918 /* In vs_1_1 WINED3DSIO_MOV can write to the address register. In later
3919 * shader versions WINED3DSIO_MOVA is used for this. */
3920 if (ins
->ctx
->reg_maps
->shader_version
.major
== 1
3921 && ins
->ctx
->reg_maps
->shader_version
.type
== WINED3D_SHADER_TYPE_VERTEX
3922 && ins
->dst
[0].reg
.type
== WINED3DSPR_ADDR
)
3924 /* This is a simple floor() */
3925 unsigned int mask_size
= shader_glsl_get_write_mask_size(write_mask
);
3926 if (mask_size
> 1) {
3927 shader_addline(buffer
, "ivec%d(floor(%s)));\n", mask_size
, src0_param
.param_str
);
3929 shader_addline(buffer
, "int(floor(%s)));\n", src0_param
.param_str
);
3932 else if (ins
->handler_idx
== WINED3DSIH_MOVA
)
3934 unsigned int mask_size
= shader_glsl_get_write_mask_size(write_mask
);
3936 if (shader_glsl_get_version(gl_info
) >= 130 || gl_info
->supported
[EXT_GPU_SHADER4
])
3939 shader_addline(buffer
, "ivec%d(round(%s)));\n", mask_size
, src0_param
.param_str
);
3941 shader_addline(buffer
, "int(round(%s)));\n", src0_param
.param_str
);
3946 shader_addline(buffer
, "ivec%d(floor(abs(%s) + vec%d(0.5)) * sign(%s)));\n",
3947 mask_size
, src0_param
.param_str
, mask_size
, src0_param
.param_str
);
3949 shader_addline(buffer
, "int(floor(abs(%s) + 0.5) * sign(%s)));\n",
3950 src0_param
.param_str
, src0_param
.param_str
);
3955 shader_addline(buffer
, "%s);\n", src0_param
.param_str
);
3959 /* Process the dot product operators DP3 and DP4 in GLSL (dst = dot(src0, src1)) */
3960 static void shader_glsl_dot(const struct wined3d_shader_instruction
*ins
)
3962 struct wined3d_string_buffer
*buffer
= ins
->ctx
->buffer
;
3963 struct glsl_src_param src0_param
;
3964 struct glsl_src_param src1_param
;
3965 DWORD dst_write_mask
, src_write_mask
;
3966 unsigned int dst_size
;
3968 dst_write_mask
= shader_glsl_append_dst(buffer
, ins
);
3969 dst_size
= shader_glsl_get_write_mask_size(dst_write_mask
);
3971 /* dp4 works on vec4, dp3 on vec3, etc. */
3972 if (ins
->handler_idx
== WINED3DSIH_DP4
)
3973 src_write_mask
= WINED3DSP_WRITEMASK_ALL
;
3974 else if (ins
->handler_idx
== WINED3DSIH_DP3
)
3975 src_write_mask
= WINED3DSP_WRITEMASK_0
| WINED3DSP_WRITEMASK_1
| WINED3DSP_WRITEMASK_2
;
3977 src_write_mask
= WINED3DSP_WRITEMASK_0
| WINED3DSP_WRITEMASK_1
;
3979 shader_glsl_add_src_param(ins
, &ins
->src
[0], src_write_mask
, &src0_param
);
3980 shader_glsl_add_src_param(ins
, &ins
->src
[1], src_write_mask
, &src1_param
);
3983 shader_addline(buffer
, "vec%d(dot(%s, %s)));\n", dst_size
, src0_param
.param_str
, src1_param
.param_str
);
3985 shader_addline(buffer
, "dot(%s, %s));\n", src0_param
.param_str
, src1_param
.param_str
);
3989 /* Note that this instruction has some restrictions. The destination write mask
3990 * can't contain the w component, and the source swizzles have to be .xyzw */
3991 static void shader_glsl_cross(const struct wined3d_shader_instruction
*ins
)
3993 DWORD src_mask
= WINED3DSP_WRITEMASK_0
| WINED3DSP_WRITEMASK_1
| WINED3DSP_WRITEMASK_2
;
3994 struct glsl_src_param src0_param
;
3995 struct glsl_src_param src1_param
;
3998 shader_glsl_get_write_mask(&ins
->dst
[0], dst_mask
);
3999 shader_glsl_append_dst(ins
->ctx
->buffer
, ins
);
4000 shader_glsl_add_src_param(ins
, &ins
->src
[0], src_mask
, &src0_param
);
4001 shader_glsl_add_src_param(ins
, &ins
->src
[1], src_mask
, &src1_param
);
4002 shader_addline(ins
->ctx
->buffer
, "cross(%s, %s)%s);\n", src0_param
.param_str
, src1_param
.param_str
, dst_mask
);
4005 static void shader_glsl_cut(const struct wined3d_shader_instruction
*ins
)
4007 unsigned int stream
= ins
->handler_idx
== WINED3DSIH_CUT
? 0 : ins
->src
[0].reg
.idx
[0].offset
;
4010 shader_addline(ins
->ctx
->buffer
, "EndPrimitive();\n");
4012 FIXME("Unhandled primitive stream %u.\n", stream
);
4015 /* Process the WINED3DSIO_POW instruction in GLSL (dst = |src0|^src1)
4016 * Src0 and src1 are scalars. Note that D3D uses the absolute of src0, while
4017 * GLSL uses the value as-is. */
4018 static void shader_glsl_pow(const struct wined3d_shader_instruction
*ins
)
4020 struct wined3d_string_buffer
*buffer
= ins
->ctx
->buffer
;
4021 struct glsl_src_param src0_param
;
4022 struct glsl_src_param src1_param
;
4023 DWORD dst_write_mask
;
4024 unsigned int dst_size
;
4026 dst_write_mask
= shader_glsl_append_dst(buffer
, ins
);
4027 dst_size
= shader_glsl_get_write_mask_size(dst_write_mask
);
4029 shader_glsl_add_src_param(ins
, &ins
->src
[0], WINED3DSP_WRITEMASK_0
, &src0_param
);
4030 shader_glsl_add_src_param(ins
, &ins
->src
[1], WINED3DSP_WRITEMASK_0
, &src1_param
);
4034 shader_addline(buffer
, "vec%u(%s == 0.0 ? 1.0 : pow(abs(%s), %s)));\n",
4035 dst_size
, src1_param
.param_str
, src0_param
.param_str
, src1_param
.param_str
);
4039 shader_addline(buffer
, "%s == 0.0 ? 1.0 : pow(abs(%s), %s));\n",
4040 src1_param
.param_str
, src0_param
.param_str
, src1_param
.param_str
);
4044 /* Map the opcode 1-to-1 to the GL code (arg->dst = instruction(src0, src1, ...) */
4045 static void shader_glsl_map2gl(const struct wined3d_shader_instruction
*ins
)
4047 const struct shader_glsl_ctx_priv
*priv
= ins
->ctx
->backend_data
;
4048 bool y_correction
= ins
->ctx
->reg_maps
->shader_version
.type
== WINED3D_SHADER_TYPE_PIXEL
4049 ? priv
->cur_ps_args
->y_correction
: false;
4050 struct wined3d_string_buffer
*buffer
= ins
->ctx
->buffer
;
4051 struct glsl_src_param src_param
;
4052 const char *instruction
;
4056 /* Determine the GLSL function to use based on the opcode */
4057 /* TODO: Possibly make this a table for faster lookups */
4058 switch (ins
->handler_idx
)
4060 case WINED3DSIH_ABS
: instruction
= "abs"; break;
4061 case WINED3DSIH_BFREV
: instruction
= "bitfieldReverse"; break;
4062 case WINED3DSIH_COUNTBITS
: instruction
= "bitCount"; break;
4063 case WINED3DSIH_DSX
: instruction
= "dFdx"; break;
4064 case WINED3DSIH_DSX_COARSE
: instruction
= "dFdxCoarse"; break;
4065 case WINED3DSIH_DSX_FINE
: instruction
= "dFdxFine"; break;
4066 case WINED3DSIH_DSY
: instruction
= y_correction
? "ycorrection.y * dFdy" : "dFdy"; break;
4067 case WINED3DSIH_DSY_COARSE
: instruction
= y_correction
? "ycorrection.y * dFdyCoarse" : "dFdyCoarse"; break;
4068 case WINED3DSIH_DSY_FINE
: instruction
= y_correction
? "ycorrection.y * dFdyFine" : "dFdyFine"; break;
4069 case WINED3DSIH_FIRSTBIT_HI
: instruction
= "findMSB"; break;
4070 case WINED3DSIH_FIRSTBIT_LO
: instruction
= "findLSB"; break;
4071 case WINED3DSIH_FIRSTBIT_SHI
: instruction
= "findMSB"; break;
4072 case WINED3DSIH_FRC
: instruction
= "fract"; break;
4073 case WINED3DSIH_IMAX
: instruction
= "max"; break;
4074 case WINED3DSIH_IMIN
: instruction
= "min"; break;
4075 case WINED3DSIH_MAX
: instruction
= "max"; break;
4076 case WINED3DSIH_MIN
: instruction
= "min"; break;
4077 case WINED3DSIH_ROUND_NE
: instruction
= "roundEven"; break;
4078 case WINED3DSIH_ROUND_NI
: instruction
= "floor"; break;
4079 case WINED3DSIH_ROUND_PI
: instruction
= "ceil"; break;
4080 case WINED3DSIH_ROUND_Z
: instruction
= "trunc"; break;
4081 case WINED3DSIH_SQRT
: instruction
= "sqrt"; break;
4082 case WINED3DSIH_UMAX
: instruction
= "max"; break;
4083 case WINED3DSIH_UMIN
: instruction
= "min"; break;
4084 default: instruction
= "";
4085 ERR("Opcode %s not yet handled in GLSL.\n", debug_d3dshaderinstructionhandler(ins
->handler_idx
));
4089 write_mask
= shader_glsl_append_dst(buffer
, ins
);
4091 /* In D3D bits are numbered from the most significant bit. */
4092 if (ins
->handler_idx
== WINED3DSIH_FIRSTBIT_HI
|| ins
->handler_idx
== WINED3DSIH_FIRSTBIT_SHI
)
4093 shader_addline(buffer
, "31 - ");
4094 shader_addline(buffer
, "%s(", instruction
);
4098 shader_glsl_add_src_param(ins
, &ins
->src
[0], write_mask
, &src_param
);
4099 shader_addline(buffer
, "%s", src_param
.param_str
);
4100 for (i
= 1; i
< ins
->src_count
; ++i
)
4102 shader_glsl_add_src_param(ins
, &ins
->src
[i
], write_mask
, &src_param
);
4103 shader_addline(buffer
, ", %s", src_param
.param_str
);
4107 shader_addline(buffer
, "));\n");
4110 static void shader_glsl_float16(const struct wined3d_shader_instruction
*ins
)
4112 struct wined3d_shader_dst_param dst
;
4113 struct glsl_src_param src
;
4118 fmt
= ins
->handler_idx
== WINED3DSIH_F16TOF32
4119 ? "unpackHalf2x16(%s).x);\n" : "packHalf2x16(vec2(%s, 0.0)));\n";
4122 for (i
= 0; i
< 4; ++i
)
4124 dst
.write_mask
= ins
->dst
[0].write_mask
& (WINED3DSP_WRITEMASK_0
<< i
);
4125 if (!(write_mask
= shader_glsl_append_dst_ext(ins
->ctx
->buffer
, ins
,
4126 &dst
, 0, dst
.reg
.data_type
)))
4129 shader_glsl_add_src_param(ins
, &ins
->src
[0], write_mask
, &src
);
4130 shader_addline(ins
->ctx
->buffer
, fmt
, src
.param_str
);
4134 static void shader_glsl_bitwise_op(const struct wined3d_shader_instruction
*ins
)
4136 struct wined3d_string_buffer
*buffer
= ins
->ctx
->buffer
;
4137 struct wined3d_shader_dst_param dst
;
4138 struct glsl_src_param src
[4];
4139 const char *instruction
;
4140 BOOL tmp_dst
= FALSE
;
4145 switch (ins
->handler_idx
)
4147 case WINED3DSIH_BFI
: instruction
= "bitfieldInsert"; break;
4148 case WINED3DSIH_IBFE
: instruction
= "bitfieldExtract"; break;
4149 case WINED3DSIH_UBFE
: instruction
= "bitfieldExtract"; break;
4151 ERR("Unhandled opcode %#x.\n", ins
->handler_idx
);
4155 for (i
= 0; i
< ins
->src_count
; ++i
)
4157 if (ins
->dst
[0].reg
.idx
[0].offset
== ins
->src
[i
].reg
.idx
[0].offset
4158 && ins
->dst
[0].reg
.type
== ins
->src
[i
].reg
.type
)
4163 for (i
= 0; i
< 4; ++i
)
4165 dst
.write_mask
= ins
->dst
[0].write_mask
& (WINED3DSP_WRITEMASK_0
<< i
);
4166 if (tmp_dst
&& (write_mask
= shader_glsl_get_write_mask(&dst
, mask_char
)))
4167 shader_addline(buffer
, "tmp0%s = %sBitsToFloat(", mask_char
,
4168 dst
.reg
.data_type
== WINED3D_DATA_INT
? "int" : "uint");
4169 else if (!(write_mask
= shader_glsl_append_dst_ext(buffer
, ins
, &dst
, 0, dst
.reg
.data_type
)))
4172 for (j
= 0; j
< ins
->src_count
; ++j
)
4173 shader_glsl_add_src_param(ins
, &ins
->src
[j
], write_mask
, &src
[j
]);
4174 shader_addline(buffer
, "%s(", instruction
);
4175 for (j
= 0; j
< ins
->src_count
- 2; ++j
)
4176 shader_addline(buffer
, "%s, ", src
[ins
->src_count
- j
- 1].param_str
);
4177 shader_addline(buffer
, "%s & 0x1f, %s & 0x1f));\n", src
[1].param_str
, src
[0].param_str
);
4182 shader_glsl_append_dst_ext(buffer
, ins
, &ins
->dst
[0], 0, WINED3D_DATA_FLOAT
);
4183 shader_glsl_get_write_mask(&ins
->dst
[0], mask_char
);
4184 shader_addline(buffer
, "tmp0%s);\n", mask_char
);
4188 static void shader_glsl_nop(const struct wined3d_shader_instruction
*ins
) {}
4190 static void shader_glsl_nrm(const struct wined3d_shader_instruction
*ins
)
4192 struct wined3d_string_buffer
*buffer
= ins
->ctx
->buffer
;
4193 struct glsl_src_param src_param
;
4194 unsigned int mask_size
;
4198 write_mask
= shader_glsl_get_write_mask(ins
->dst
, dst_mask
);
4199 mask_size
= shader_glsl_get_write_mask_size(write_mask
);
4200 shader_glsl_add_src_param(ins
, &ins
->src
[0], write_mask
, &src_param
);
4203 shader_addline(buffer
, "tmp0.x = dot(vec3(%s), vec3(%s));\n",
4204 src_param
.param_str
, src_param
.param_str
);
4206 shader_addline(buffer
, "tmp0.x = dot(%s, %s);\n",
4207 src_param
.param_str
, src_param
.param_str
);
4208 shader_glsl_append_dst(buffer
, ins
);
4210 shader_addline(buffer
, "tmp0.x == 0.0 ? %s : (%s * inversesqrt(tmp0.x)));\n",
4211 src_param
.param_str
, src_param
.param_str
);
4214 static void shader_glsl_scalar_op(const struct wined3d_shader_instruction
*ins
)
4216 DWORD shader_version
= WINED3D_SHADER_VERSION(ins
->ctx
->reg_maps
->shader_version
.major
,
4217 ins
->ctx
->reg_maps
->shader_version
.minor
);
4218 struct wined3d_string_buffer
*buffer
= ins
->ctx
->buffer
;
4219 struct glsl_src_param src0_param
;
4220 const char *prefix
, *suffix
;
4221 unsigned int dst_size
;
4222 DWORD dst_write_mask
;
4224 dst_write_mask
= shader_glsl_append_dst(buffer
, ins
);
4225 dst_size
= shader_glsl_get_write_mask_size(dst_write_mask
);
4227 if (shader_version
< WINED3D_SHADER_VERSION(4, 0))
4228 dst_write_mask
= WINED3DSP_WRITEMASK_3
;
4230 shader_glsl_add_src_param(ins
, &ins
->src
[0], dst_write_mask
, &src0_param
);
4232 switch (ins
->handler_idx
)
4234 case WINED3DSIH_EXP
:
4235 case WINED3DSIH_EXPP
:
4240 case WINED3DSIH_LOG
:
4241 case WINED3DSIH_LOGP
:
4242 prefix
= "log2(abs(";
4246 case WINED3DSIH_RCP
:
4251 case WINED3DSIH_RSQ
:
4252 prefix
= "inversesqrt(abs(";
4259 FIXME("Unhandled instruction %#x.\n", ins
->handler_idx
);
4263 if (dst_size
> 1 && shader_version
< WINED3D_SHADER_VERSION(4, 0))
4264 shader_addline(buffer
, "vec%u(%s%s%s));\n", dst_size
, prefix
, src0_param
.param_str
, suffix
);
4266 shader_addline(buffer
, "%s%s%s);\n", prefix
, src0_param
.param_str
, suffix
);
4269 /** Process the WINED3DSIO_EXPP instruction in GLSL:
4270 * For shader model 1.x, do the following (and honor the writemask, so use a temporary variable):
4271 * dst.x = 2^(floor(src))
4272 * dst.y = src - floor(src)
4273 * dst.z = 2^src (partial precision is allowed, but optional)
4275 * For 2.0 shaders, just do this (honoring writemask and swizzle):
4276 * dst = 2^src; (partial precision is allowed, but optional)
4278 static void shader_glsl_expp(const struct wined3d_shader_instruction
*ins
)
4280 if (ins
->ctx
->reg_maps
->shader_version
.major
< 2)
4282 struct glsl_src_param src_param
;
4285 shader_glsl_add_src_param(ins
, &ins
->src
[0], WINED3DSP_WRITEMASK_3
, &src_param
);
4287 shader_addline(ins
->ctx
->buffer
, "tmp0.x = exp2(floor(%s));\n", src_param
.param_str
);
4288 shader_addline(ins
->ctx
->buffer
, "tmp0.y = %s - floor(%s);\n", src_param
.param_str
, src_param
.param_str
);
4289 shader_addline(ins
->ctx
->buffer
, "tmp0.z = exp2(%s);\n", src_param
.param_str
);
4290 shader_addline(ins
->ctx
->buffer
, "tmp0.w = 1.0;\n");
4292 shader_glsl_append_dst(ins
->ctx
->buffer
, ins
);
4293 shader_glsl_get_write_mask(&ins
->dst
[0], dst_mask
);
4294 shader_addline(ins
->ctx
->buffer
, "tmp0%s);\n", dst_mask
);
4298 shader_glsl_scalar_op(ins
);
4301 static void shader_glsl_cast(const struct wined3d_shader_instruction
*ins
,
4302 const char *vector_constructor
, const char *scalar_constructor
)
4304 struct wined3d_string_buffer
*buffer
= ins
->ctx
->buffer
;
4305 struct glsl_src_param src_param
;
4306 unsigned int mask_size
;
4309 write_mask
= shader_glsl_append_dst(buffer
, ins
);
4310 mask_size
= shader_glsl_get_write_mask_size(write_mask
);
4311 shader_glsl_add_src_param(ins
, &ins
->src
[0], write_mask
, &src_param
);
4314 shader_addline(buffer
, "%s%u(%s));\n", vector_constructor
, mask_size
, src_param
.param_str
);
4316 shader_addline(buffer
, "%s(%s));\n", scalar_constructor
, src_param
.param_str
);
4319 static void shader_glsl_to_int(const struct wined3d_shader_instruction
*ins
)
4321 shader_glsl_cast(ins
, "ivec", "int");
4324 static void shader_glsl_to_uint(const struct wined3d_shader_instruction
*ins
)
4326 shader_glsl_cast(ins
, "uvec", "uint");
4329 static void shader_glsl_to_float(const struct wined3d_shader_instruction
*ins
)
4331 shader_glsl_cast(ins
, "vec", "float");
4334 /** Process signed comparison opcodes in GLSL. */
4335 static void shader_glsl_compare(const struct wined3d_shader_instruction
*ins
)
4337 struct glsl_src_param src0_param
;
4338 struct glsl_src_param src1_param
;
4340 unsigned int mask_size
;
4342 write_mask
= shader_glsl_append_dst(ins
->ctx
->buffer
, ins
);
4343 mask_size
= shader_glsl_get_write_mask_size(write_mask
);
4344 shader_glsl_add_src_param(ins
, &ins
->src
[0], write_mask
, &src0_param
);
4345 shader_glsl_add_src_param(ins
, &ins
->src
[1], write_mask
, &src1_param
);
4347 if (mask_size
> 1) {
4348 const char *compare
;
4350 switch(ins
->handler_idx
)
4352 case WINED3DSIH_SLT
: compare
= "lessThan"; break;
4353 case WINED3DSIH_SGE
: compare
= "greaterThanEqual"; break;
4354 default: compare
= "";
4355 FIXME("Can't handle opcode %s.\n", debug_d3dshaderinstructionhandler(ins
->handler_idx
));
4358 shader_addline(ins
->ctx
->buffer
, "vec%d(%s(%s, %s)));\n", mask_size
, compare
,
4359 src0_param
.param_str
, src1_param
.param_str
);
4361 switch(ins
->handler_idx
)
4363 case WINED3DSIH_SLT
:
4364 /* Step(src0, src1) is not suitable here because if src0 == src1 SLT is supposed,
4365 * to return 0.0 but step returns 1.0 because step is not < x
4366 * An alternative is a bvec compare padded with an unused second component.
4367 * step(src1 * -1.0, src0 * -1.0) is not an option because it suffers from the same
4368 * issue. Playing with not() is not possible either because not() does not accept
4371 shader_addline(ins
->ctx
->buffer
, "(%s < %s) ? 1.0 : 0.0);\n",
4372 src0_param
.param_str
, src1_param
.param_str
);
4374 case WINED3DSIH_SGE
:
4375 /* Here we can use the step() function and safe a conditional */
4376 shader_addline(ins
->ctx
->buffer
, "step(%s, %s));\n", src1_param
.param_str
, src0_param
.param_str
);
4379 FIXME("Can't handle opcode %s.\n", debug_d3dshaderinstructionhandler(ins
->handler_idx
));
4385 static void shader_glsl_swapc(const struct wined3d_shader_instruction
*ins
)
4387 struct wined3d_string_buffer
*buffer
= ins
->ctx
->buffer
;
4388 struct wined3d_shader_dst_param dst
[2];
4389 struct glsl_src_param src
[3];
4390 unsigned int i
, j
, k
;
4395 for (i
= 0; i
< ins
->dst_count
; ++i
)
4398 for (j
= 0; j
< ins
->src_count
; ++j
)
4400 if (ins
->dst
[i
].reg
.idx
[0].offset
== ins
->src
[j
].reg
.idx
[0].offset
4401 && ins
->dst
[i
].reg
.type
== ins
->src
[j
].reg
.type
)
4406 dst
[0] = ins
->dst
[0];
4407 dst
[1] = ins
->dst
[1];
4408 for (i
= 0; i
< 4; ++i
)
4410 for (j
= 0; j
< ARRAY_SIZE(dst
); ++j
)
4412 dst
[j
].write_mask
= ins
->dst
[j
].write_mask
& (WINED3DSP_WRITEMASK_0
<< i
);
4413 if (tmp_dst
[j
] && (write_mask
= shader_glsl_get_write_mask(&dst
[j
], mask_char
)))
4414 shader_addline(buffer
, "tmp%u%s = (", j
, mask_char
);
4415 else if (!(write_mask
= shader_glsl_append_dst_ext(buffer
, ins
, &dst
[j
], j
, dst
[j
].reg
.data_type
)))
4418 for (k
= 0; k
< ARRAY_SIZE(src
); ++k
)
4419 shader_glsl_add_src_param(ins
, &ins
->src
[k
], write_mask
, &src
[k
]);
4421 shader_addline(buffer
, "%sbool(%s) ? %s : %s);\n", !j
? "!" : "",
4422 src
[0].param_str
, src
[1].param_str
, src
[2].param_str
);
4426 for (i
= 0; i
< ARRAY_SIZE(tmp_dst
); ++i
)
4430 shader_glsl_get_write_mask(&ins
->dst
[i
], mask_char
);
4431 shader_glsl_append_dst_ext(buffer
, ins
, &ins
->dst
[i
], i
, ins
->dst
[i
].reg
.data_type
);
4432 shader_addline(buffer
, "tmp%u%s);\n", i
, mask_char
);
4437 static void shader_glsl_conditional_move(const struct wined3d_shader_instruction
*ins
)
4439 const char *condition_prefix
, *condition_suffix
;
4440 struct wined3d_shader_dst_param dst
;
4441 struct glsl_src_param src0_param
;
4442 struct glsl_src_param src1_param
;
4443 struct glsl_src_param src2_param
;
4444 BOOL temp_destination
= FALSE
;
4445 DWORD cmp_channel
= 0;
4450 switch (ins
->handler_idx
)
4452 case WINED3DSIH_CMP
:
4453 condition_prefix
= "";
4454 condition_suffix
= " >= 0.0";
4457 case WINED3DSIH_CND
:
4458 condition_prefix
= "";
4459 condition_suffix
= " > 0.5";
4462 case WINED3DSIH_MOVC
:
4463 condition_prefix
= "bool(";
4464 condition_suffix
= ")";
4468 FIXME("Unhandled instruction %#x.\n", ins
->handler_idx
);
4469 condition_prefix
= "<unhandled prefix>";
4470 condition_suffix
= "<unhandled suffix>";
4474 if (shader_is_scalar(&ins
->dst
[0].reg
) || shader_is_scalar(&ins
->src
[0].reg
))
4476 write_mask
= shader_glsl_append_dst(ins
->ctx
->buffer
, ins
);
4477 shader_glsl_add_src_param(ins
, &ins
->src
[0], write_mask
, &src0_param
);
4478 shader_glsl_add_src_param(ins
, &ins
->src
[1], write_mask
, &src1_param
);
4479 shader_glsl_add_src_param(ins
, &ins
->src
[2], write_mask
, &src2_param
);
4481 shader_addline(ins
->ctx
->buffer
, "%s%s%s ? %s : %s);\n",
4482 condition_prefix
, src0_param
.param_str
, condition_suffix
,
4483 src1_param
.param_str
, src2_param
.param_str
);
4489 /* Splitting the instruction up in multiple lines imposes a problem:
4490 * The first lines may overwrite source parameters of the following lines.
4491 * Deal with that by using a temporary destination register if needed. */
4492 if ((ins
->src
[0].reg
.idx
[0].offset
== dst
.reg
.idx
[0].offset
4493 && ins
->src
[0].reg
.type
== dst
.reg
.type
)
4494 || (ins
->src
[1].reg
.idx
[0].offset
== dst
.reg
.idx
[0].offset
4495 && ins
->src
[1].reg
.type
== dst
.reg
.type
)
4496 || (ins
->src
[2].reg
.idx
[0].offset
== dst
.reg
.idx
[0].offset
4497 && ins
->src
[2].reg
.type
== dst
.reg
.type
))
4498 temp_destination
= TRUE
;
4500 /* Cycle through all source0 channels. */
4501 for (i
= 0; i
< 4; ++i
)
4504 /* Find the destination channels which use the current source0 channel. */
4505 for (j
= 0; j
< 4; ++j
)
4507 if (shader_glsl_swizzle_get_component(ins
->src
[0].swizzle
, j
) == i
)
4509 write_mask
|= WINED3DSP_WRITEMASK_0
<< j
;
4510 cmp_channel
= WINED3DSP_WRITEMASK_0
<< j
;
4513 dst
.write_mask
= ins
->dst
[0].write_mask
& write_mask
;
4515 if (temp_destination
)
4517 if (!(write_mask
= shader_glsl_get_write_mask(&dst
, mask_char
)))
4519 shader_addline(ins
->ctx
->buffer
, "tmp0%s = (", mask_char
);
4521 else if (!(write_mask
= shader_glsl_append_dst_ext(ins
->ctx
->buffer
, ins
, &dst
, 0, dst
.reg
.data_type
)))
4524 shader_glsl_add_src_param(ins
, &ins
->src
[0], cmp_channel
, &src0_param
);
4525 shader_glsl_add_src_param(ins
, &ins
->src
[1], write_mask
, &src1_param
);
4526 shader_glsl_add_src_param(ins
, &ins
->src
[2], write_mask
, &src2_param
);
4528 shader_addline(ins
->ctx
->buffer
, "%s%s%s ? %s : %s);\n",
4529 condition_prefix
, src0_param
.param_str
, condition_suffix
,
4530 src1_param
.param_str
, src2_param
.param_str
);
4533 if (temp_destination
)
4535 shader_glsl_get_write_mask(&ins
->dst
[0], mask_char
);
4536 shader_glsl_append_dst(ins
->ctx
->buffer
, ins
);
4537 shader_addline(ins
->ctx
->buffer
, "tmp0%s);\n", mask_char
);
4541 /** Process the CND opcode in GLSL (dst = (src0 > 0.5) ? src1 : src2) */
4542 /* For ps 1.1-1.3, only a single component of src0 is used. For ps 1.4
4543 * the compare is done per component of src0. */
4544 static void shader_glsl_cnd(const struct wined3d_shader_instruction
*ins
)
4546 struct glsl_src_param src0_param
;
4547 struct glsl_src_param src1_param
;
4548 struct glsl_src_param src2_param
;
4550 DWORD shader_version
= WINED3D_SHADER_VERSION(ins
->ctx
->reg_maps
->shader_version
.major
,
4551 ins
->ctx
->reg_maps
->shader_version
.minor
);
4553 if (shader_version
< WINED3D_SHADER_VERSION(1, 4))
4555 write_mask
= shader_glsl_append_dst(ins
->ctx
->buffer
, ins
);
4556 shader_glsl_add_src_param(ins
, &ins
->src
[0], WINED3DSP_WRITEMASK_0
, &src0_param
);
4557 shader_glsl_add_src_param(ins
, &ins
->src
[1], write_mask
, &src1_param
);
4558 shader_glsl_add_src_param(ins
, &ins
->src
[2], write_mask
, &src2_param
);
4560 if (ins
->coissue
&& ins
->dst
->write_mask
!= WINED3DSP_WRITEMASK_3
)
4561 shader_addline(ins
->ctx
->buffer
, "%s /* COISSUE! */);\n", src1_param
.param_str
);
4563 shader_addline(ins
->ctx
->buffer
, "%s > 0.5 ? %s : %s);\n",
4564 src0_param
.param_str
, src1_param
.param_str
, src2_param
.param_str
);
4568 shader_glsl_conditional_move(ins
);
4571 /** GLSL code generation for WINED3DSIO_MAD: Multiply the first 2 opcodes, then add the last */
4572 static void shader_glsl_mad(const struct wined3d_shader_instruction
*ins
)
4574 struct glsl_src_param src0_param
;
4575 struct glsl_src_param src1_param
;
4576 struct glsl_src_param src2_param
;
4579 write_mask
= shader_glsl_append_dst(ins
->ctx
->buffer
, ins
);
4580 shader_glsl_add_src_param(ins
, &ins
->src
[0], write_mask
, &src0_param
);
4581 shader_glsl_add_src_param(ins
, &ins
->src
[1], write_mask
, &src1_param
);
4582 shader_glsl_add_src_param(ins
, &ins
->src
[2], write_mask
, &src2_param
);
4583 shader_addline(ins
->ctx
->buffer
, "(%s * %s) + %s);\n",
4584 src0_param
.param_str
, src1_param
.param_str
, src2_param
.param_str
);
4587 /* Handles transforming all WINED3DSIO_M?x? opcodes for
4588 Vertex shaders to GLSL codes */
4589 static void shader_glsl_mnxn(const struct wined3d_shader_instruction
*ins
)
4592 int nComponents
= 0;
4593 struct wined3d_shader_dst_param tmp_dst
= {{0}};
4594 struct wined3d_shader_src_param tmp_src
[2] = {{{0}}};
4595 struct wined3d_shader_instruction tmp_ins
;
4597 memset(&tmp_ins
, 0, sizeof(tmp_ins
));
4599 /* Set constants for the temporary argument */
4600 tmp_ins
.ctx
= ins
->ctx
;
4601 tmp_ins
.dst_count
= 1;
4602 tmp_ins
.dst
= &tmp_dst
;
4603 tmp_ins
.src_count
= 2;
4604 tmp_ins
.src
= tmp_src
;
4606 switch(ins
->handler_idx
)
4608 case WINED3DSIH_M4x4
:
4610 tmp_ins
.handler_idx
= WINED3DSIH_DP4
;
4612 case WINED3DSIH_M4x3
:
4614 tmp_ins
.handler_idx
= WINED3DSIH_DP4
;
4616 case WINED3DSIH_M3x4
:
4618 tmp_ins
.handler_idx
= WINED3DSIH_DP3
;
4620 case WINED3DSIH_M3x3
:
4622 tmp_ins
.handler_idx
= WINED3DSIH_DP3
;
4624 case WINED3DSIH_M3x2
:
4626 tmp_ins
.handler_idx
= WINED3DSIH_DP3
;
4632 tmp_dst
= ins
->dst
[0];
4633 tmp_src
[0] = ins
->src
[0];
4634 tmp_src
[1] = ins
->src
[1];
4635 for (i
= 0; i
< nComponents
; ++i
)
4637 tmp_dst
.write_mask
= WINED3DSP_WRITEMASK_0
<< i
;
4638 shader_glsl_dot(&tmp_ins
);
4639 ++tmp_src
[1].reg
.idx
[0].offset
;
4644 The LRP instruction performs a component-wise linear interpolation
4645 between the second and third operands using the first operand as the
4646 blend factor. Equation: (dst = src2 + src0 * (src1 - src2))
4647 This is equivalent to mix(src2, src1, src0);
4649 static void shader_glsl_lrp(const struct wined3d_shader_instruction
*ins
)
4651 struct glsl_src_param src0_param
;
4652 struct glsl_src_param src1_param
;
4653 struct glsl_src_param src2_param
;
4656 write_mask
= shader_glsl_append_dst(ins
->ctx
->buffer
, ins
);
4658 shader_glsl_add_src_param(ins
, &ins
->src
[0], write_mask
, &src0_param
);
4659 shader_glsl_add_src_param(ins
, &ins
->src
[1], write_mask
, &src1_param
);
4660 shader_glsl_add_src_param(ins
, &ins
->src
[2], write_mask
, &src2_param
);
4662 shader_addline(ins
->ctx
->buffer
, "mix(%s, %s, %s));\n",
4663 src2_param
.param_str
, src1_param
.param_str
, src0_param
.param_str
);
4666 /** Process the WINED3DSIO_LIT instruction in GLSL:
4667 * dst.x = dst.w = 1.0
4668 * dst.y = (src0.x > 0) ? src0.x
4669 * dst.z = (src0.x > 0) ? ((src0.y > 0) ? pow(src0.y, src.w) : 0) : 0
4670 * where src.w is clamped at +- 128
4672 static void shader_glsl_lit(const struct wined3d_shader_instruction
*ins
)
4674 struct glsl_src_param src0_param
;
4675 struct glsl_src_param src1_param
;
4676 struct glsl_src_param src3_param
;
4679 shader_glsl_append_dst(ins
->ctx
->buffer
, ins
);
4680 shader_glsl_get_write_mask(&ins
->dst
[0], dst_mask
);
4682 shader_glsl_add_src_param(ins
, &ins
->src
[0], WINED3DSP_WRITEMASK_0
, &src0_param
);
4683 shader_glsl_add_src_param(ins
, &ins
->src
[0], WINED3DSP_WRITEMASK_1
, &src1_param
);
4684 shader_glsl_add_src_param(ins
, &ins
->src
[0], WINED3DSP_WRITEMASK_3
, &src3_param
);
4686 /* The sdk specifies the instruction like this
4688 * if(src.x > 0.0) dst.y = src.x
4690 * if(src.x > 0.0 && src.y > 0.0) dst.z = pow(src.y, power);
4693 * (where power = src.w clamped between -128 and 128)
4695 * Obviously that has quite a few conditionals in it which we don't like. So the first step is this:
4696 * dst.x = 1.0 ... No further explanation needed
4697 * dst.y = max(src.y, 0.0); ... If x < 0.0, use 0.0, otherwise x. Same as the conditional
4698 * dst.z = x > 0.0 ? pow(max(y, 0.0), p) : 0; ... 0 ^ power is 0, and otherwise we use y anyway
4699 * dst.w = 1.0. ... Nothing fancy.
4701 * So we still have one conditional in there. So do this:
4702 * dst.z = pow(max(0.0, src.y) * step(0.0, src.x), power);
4704 * 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),
4705 * 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.
4706 * if both x and y are > 0, we get pow(y * 1.0, power), as it is supposed to.
4708 * Unfortunately pow(0.0 ^ 0.0) returns NaN on most GPUs, but lit with src.y = 0 and src.w = 0 returns
4709 * a non-NaN value in dst.z. What we return doesn't matter, as long as it is not NaN. Return 0, which is
4710 * what all Windows HW drivers and GL_ARB_vertex_program's LIT do.
4712 shader_addline(ins
->ctx
->buffer
,
4713 "vec4(1.0, max(%s, 0.0), %s == 0.0 ? 0.0 : "
4714 "pow(max(0.0, %s) * step(0.0, %s), clamp(%s, -128.0, 128.0)), 1.0)%s);\n",
4715 src0_param
.param_str
, src3_param
.param_str
, src1_param
.param_str
,
4716 src0_param
.param_str
, src3_param
.param_str
, dst_mask
);
4719 /** Process the WINED3DSIO_DST instruction in GLSL:
4721 * dst.y = src0.x * src0.y
4725 static void shader_glsl_dst(const struct wined3d_shader_instruction
*ins
)
4727 struct glsl_src_param src0y_param
;
4728 struct glsl_src_param src0z_param
;
4729 struct glsl_src_param src1y_param
;
4730 struct glsl_src_param src1w_param
;
4733 shader_glsl_append_dst(ins
->ctx
->buffer
, ins
);
4734 shader_glsl_get_write_mask(&ins
->dst
[0], dst_mask
);
4736 shader_glsl_add_src_param(ins
, &ins
->src
[0], WINED3DSP_WRITEMASK_1
, &src0y_param
);
4737 shader_glsl_add_src_param(ins
, &ins
->src
[0], WINED3DSP_WRITEMASK_2
, &src0z_param
);
4738 shader_glsl_add_src_param(ins
, &ins
->src
[1], WINED3DSP_WRITEMASK_1
, &src1y_param
);
4739 shader_glsl_add_src_param(ins
, &ins
->src
[1], WINED3DSP_WRITEMASK_3
, &src1w_param
);
4741 shader_addline(ins
->ctx
->buffer
, "vec4(1.0, %s * %s, %s, %s))%s;\n",
4742 src0y_param
.param_str
, src1y_param
.param_str
, src0z_param
.param_str
, src1w_param
.param_str
, dst_mask
);
4745 /** Process the WINED3DSIO_SINCOS instruction in GLSL:
4746 * VS 2.0 requires that specific cosine and sine constants be passed to this instruction so the hardware
4747 * can handle it. But, these functions are built-in for GLSL, so we can just ignore the last 2 params.
4749 * dst.x = cos(src0.?)
4750 * dst.y = sin(src0.?)
4754 static void shader_glsl_sincos(const struct wined3d_shader_instruction
*ins
)
4756 struct wined3d_string_buffer
*buffer
= ins
->ctx
->buffer
;
4757 struct glsl_src_param src0_param
;
4760 if (ins
->ctx
->reg_maps
->shader_version
.major
< 4)
4762 shader_glsl_add_src_param(ins
, &ins
->src
[0], WINED3DSP_WRITEMASK_0
, &src0_param
);
4764 write_mask
= shader_glsl_append_dst(buffer
, ins
);
4767 case WINED3DSP_WRITEMASK_0
:
4768 shader_addline(buffer
, "cos(%s));\n", src0_param
.param_str
);
4771 case WINED3DSP_WRITEMASK_1
:
4772 shader_addline(buffer
, "sin(%s));\n", src0_param
.param_str
);
4775 case (WINED3DSP_WRITEMASK_0
| WINED3DSP_WRITEMASK_1
):
4776 shader_addline(buffer
, "vec2(cos(%s), sin(%s)));\n",
4777 src0_param
.param_str
, src0_param
.param_str
);
4781 ERR("Write mask should be .x, .y or .xy\n");
4788 if (ins
->dst
[0].reg
.type
!= WINED3DSPR_NULL
)
4791 if (ins
->dst
[1].reg
.type
!= WINED3DSPR_NULL
)
4795 write_mask
= shader_glsl_get_write_mask(&ins
->dst
[0], dst_mask
);
4796 shader_glsl_add_src_param(ins
, &ins
->src
[0], write_mask
, &src0_param
);
4797 shader_addline(buffer
, "tmp0%s = sin(%s);\n", dst_mask
, src0_param
.param_str
);
4799 write_mask
= shader_glsl_append_dst_ext(buffer
, ins
, &ins
->dst
[1], 1, ins
->dst
[1].reg
.data_type
);
4800 shader_glsl_add_src_param(ins
, &ins
->src
[0], write_mask
, &src0_param
);
4801 shader_addline(buffer
, "cos(%s));\n", src0_param
.param_str
);
4803 shader_glsl_append_dst_ext(buffer
, ins
, &ins
->dst
[0], 0, ins
->dst
[0].reg
.data_type
);
4804 shader_addline(buffer
, "tmp0%s);\n", dst_mask
);
4808 write_mask
= shader_glsl_append_dst_ext(buffer
, ins
, &ins
->dst
[0], 0, ins
->dst
[0].reg
.data_type
);
4809 shader_glsl_add_src_param(ins
, &ins
->src
[0], write_mask
, &src0_param
);
4810 shader_addline(buffer
, "sin(%s));\n", src0_param
.param_str
);
4813 else if (ins
->dst
[1].reg
.type
!= WINED3DSPR_NULL
)
4815 write_mask
= shader_glsl_append_dst_ext(buffer
, ins
, &ins
->dst
[1], 1, ins
->dst
[1].reg
.data_type
);
4816 shader_glsl_add_src_param(ins
, &ins
->src
[0], write_mask
, &src0_param
);
4817 shader_addline(buffer
, "cos(%s));\n", src0_param
.param_str
);
4821 /* sgn in vs_2_0 has 2 extra parameters(registers for temporary storage) which we don't use
4822 * here. But those extra parameters require a dedicated function for sgn, since map2gl would
4823 * generate invalid code
4825 static void shader_glsl_sgn(const struct wined3d_shader_instruction
*ins
)
4827 struct glsl_src_param src0_param
;
4830 write_mask
= shader_glsl_append_dst(ins
->ctx
->buffer
, ins
);
4831 shader_glsl_add_src_param(ins
, &ins
->src
[0], write_mask
, &src0_param
);
4833 shader_addline(ins
->ctx
->buffer
, "sign(%s));\n", src0_param
.param_str
);
4836 /** Process the WINED3DSIO_LOOP instruction in GLSL:
4837 * Start a for() loop where src1.y is the initial value of aL,
4838 * increment aL by src1.z for a total of src1.x iterations.
4839 * Need to use a temporary variable for this operation.
4841 /* FIXME: I don't think nested loops will work correctly this way. */
4842 static void shader_glsl_loop(const struct wined3d_shader_instruction
*ins
)
4844 struct wined3d_shader_parser_state
*state
= ins
->ctx
->state
;
4845 struct shader_glsl_ctx_priv
*priv
= ins
->ctx
->backend_data
;
4846 struct wined3d_string_buffer
*buffer
= ins
->ctx
->buffer
;
4847 const struct wined3d_shader
*shader
= ins
->ctx
->shader
;
4848 const struct wined3d_shader_lconst
*constant
;
4849 struct wined3d_string_buffer
*reg_name
;
4850 const DWORD
*control_values
= NULL
;
4852 if (ins
->ctx
->reg_maps
->shader_version
.major
< 4)
4854 /* Try to hardcode the loop control parameters if possible. Direct3D 9
4855 * class hardware doesn't support real varying indexing, but Microsoft
4856 * designed this feature for Shader model 2.x+. If the loop control is
4857 * known at compile time, the GLSL compiler can unroll the loop, and
4858 * replace indirect addressing with direct addressing. */
4859 if (ins
->src
[1].reg
.type
== WINED3DSPR_CONSTINT
)
4861 LIST_FOR_EACH_ENTRY(constant
, &shader
->constantsI
, struct wined3d_shader_lconst
, entry
)
4863 if (constant
->idx
== ins
->src
[1].reg
.idx
[0].offset
)
4865 control_values
= constant
->value
;
4873 struct wined3d_shader_loop_control loop_control
;
4874 loop_control
.count
= control_values
[0];
4875 loop_control
.start
= control_values
[1];
4876 loop_control
.step
= (int)control_values
[2];
4878 if (loop_control
.step
> 0)
4880 shader_addline(buffer
, "for (aL%u = %u; aL%u < (%u * %d + %u); aL%u += %d)\n{\n",
4881 state
->current_loop_depth
, loop_control
.start
,
4882 state
->current_loop_depth
, loop_control
.count
, loop_control
.step
, loop_control
.start
,
4883 state
->current_loop_depth
, loop_control
.step
);
4885 else if (loop_control
.step
< 0)
4887 shader_addline(buffer
, "for (aL%u = %u; aL%u > (%u * %d + %u); aL%u += %d)\n{\n",
4888 state
->current_loop_depth
, loop_control
.start
,
4889 state
->current_loop_depth
, loop_control
.count
, loop_control
.step
, loop_control
.start
,
4890 state
->current_loop_depth
, loop_control
.step
);
4894 shader_addline(buffer
, "for (aL%u = %u, tmpInt%u = 0; tmpInt%u < %u; tmpInt%u++)\n{\n",
4895 state
->current_loop_depth
, loop_control
.start
, state
->current_loop_depth
,
4896 state
->current_loop_depth
, loop_control
.count
,
4897 state
->current_loop_depth
);
4902 reg_name
= string_buffer_get(priv
->string_buffers
);
4903 shader_glsl_get_register_name(&ins
->src
[1].reg
, ins
->src
[1].reg
.data_type
, reg_name
, NULL
, ins
->ctx
);
4905 shader_addline(buffer
, "for (tmpInt%u = 0, aL%u = %s.y; tmpInt%u < %s.x; tmpInt%u++, aL%u += %s.z)\n{\n",
4906 state
->current_loop_depth
, state
->current_loop_reg
, reg_name
->buffer
,
4907 state
->current_loop_depth
, reg_name
->buffer
,
4908 state
->current_loop_depth
, state
->current_loop_reg
, reg_name
->buffer
);
4910 string_buffer_release(priv
->string_buffers
, reg_name
);
4914 ++state
->current_loop_reg
;
4918 shader_addline(buffer
, "for (;;)\n{\n");
4921 ++state
->current_loop_depth
;
4924 static void shader_glsl_end(const struct wined3d_shader_instruction
*ins
)
4926 struct wined3d_shader_parser_state
*state
= ins
->ctx
->state
;
4928 shader_addline(ins
->ctx
->buffer
, "}\n");
4930 if (ins
->handler_idx
== WINED3DSIH_ENDLOOP
)
4932 --state
->current_loop_depth
;
4933 --state
->current_loop_reg
;
4936 if (ins
->handler_idx
== WINED3DSIH_ENDREP
)
4938 --state
->current_loop_depth
;
4942 static void shader_glsl_rep(const struct wined3d_shader_instruction
*ins
)
4944 struct wined3d_shader_parser_state
*state
= ins
->ctx
->state
;
4945 const struct wined3d_shader
*shader
= ins
->ctx
->shader
;
4946 const struct wined3d_shader_lconst
*constant
;
4947 struct glsl_src_param src0_param
;
4948 const DWORD
*control_values
= NULL
;
4950 /* Try to hardcode local values to help the GLSL compiler to unroll and optimize the loop */
4951 if (ins
->src
[0].reg
.type
== WINED3DSPR_CONSTINT
)
4953 LIST_FOR_EACH_ENTRY(constant
, &shader
->constantsI
, struct wined3d_shader_lconst
, entry
)
4955 if (constant
->idx
== ins
->src
[0].reg
.idx
[0].offset
)
4957 control_values
= constant
->value
;
4965 shader_addline(ins
->ctx
->buffer
, "for (tmpInt%d = 0; tmpInt%d < %d; tmpInt%d++) {\n",
4966 state
->current_loop_depth
, state
->current_loop_depth
,
4967 control_values
[0], state
->current_loop_depth
);
4971 shader_glsl_add_src_param(ins
, &ins
->src
[0], WINED3DSP_WRITEMASK_0
, &src0_param
);
4972 shader_addline(ins
->ctx
->buffer
, "for (tmpInt%d = 0; tmpInt%d < %s; tmpInt%d++) {\n",
4973 state
->current_loop_depth
, state
->current_loop_depth
,
4974 src0_param
.param_str
, state
->current_loop_depth
);
4977 ++state
->current_loop_depth
;
4980 static void shader_glsl_switch(const struct wined3d_shader_instruction
*ins
)
4982 struct glsl_src_param src0_param
;
4984 shader_glsl_add_src_param(ins
, &ins
->src
[0], WINED3DSP_WRITEMASK_0
, &src0_param
);
4985 shader_addline(ins
->ctx
->buffer
, "switch (%s)\n{\n", src0_param
.param_str
);
4988 static void shader_glsl_case(const struct wined3d_shader_instruction
*ins
)
4990 struct glsl_src_param src0_param
;
4992 shader_glsl_add_src_param(ins
, &ins
->src
[0], WINED3DSP_WRITEMASK_0
, &src0_param
);
4993 shader_addline(ins
->ctx
->buffer
, "case %s:\n", src0_param
.param_str
);
4996 static void shader_glsl_default(const struct wined3d_shader_instruction
*ins
)
4998 shader_addline(ins
->ctx
->buffer
, "default:\n");
5001 static void shader_glsl_generate_condition(const struct wined3d_shader_instruction
*ins
)
5003 struct glsl_src_param src_param
;
5004 const char *condition
;
5006 condition
= ins
->flags
== WINED3D_SHADER_CONDITIONAL_OP_NZ
? "bool" : "!bool";
5007 shader_glsl_add_src_param(ins
, &ins
->src
[0], WINED3DSP_WRITEMASK_0
, &src_param
);
5008 shader_addline(ins
->ctx
->buffer
, "if (%s(%s))\n", condition
, src_param
.param_str
);
5011 static void shader_glsl_if(const struct wined3d_shader_instruction
*ins
)
5013 shader_glsl_generate_condition(ins
);
5014 shader_addline(ins
->ctx
->buffer
, "{\n");
5017 static void shader_glsl_ifc(const struct wined3d_shader_instruction
*ins
)
5019 struct glsl_src_param src0_param
;
5020 struct glsl_src_param src1_param
;
5022 shader_glsl_add_src_param(ins
, &ins
->src
[0], WINED3DSP_WRITEMASK_0
, &src0_param
);
5023 shader_glsl_add_src_param(ins
, &ins
->src
[1], WINED3DSP_WRITEMASK_0
, &src1_param
);
5025 shader_addline(ins
->ctx
->buffer
, "if (%s %s %s) {\n",
5026 src0_param
.param_str
, shader_glsl_get_rel_op(ins
->flags
), src1_param
.param_str
);
5029 static void shader_glsl_else(const struct wined3d_shader_instruction
*ins
)
5031 shader_addline(ins
->ctx
->buffer
, "} else {\n");
5034 static void shader_glsl_emit(const struct wined3d_shader_instruction
*ins
)
5036 unsigned int stream
= ins
->handler_idx
== WINED3DSIH_EMIT
? 0 : ins
->src
[0].reg
.idx
[0].offset
;
5037 const struct wined3d_shader_reg_maps
*reg_maps
= ins
->ctx
->reg_maps
;
5038 const struct shader_glsl_ctx_priv
*priv
= ins
->ctx
->backend_data
;
5040 shader_addline(ins
->ctx
->buffer
, "setup_gs_output(gs_out);\n");
5041 if (!priv
->gl_info
->supported
[ARB_CLIP_CONTROL
])
5042 shader_glsl_fixup_position(ins
->ctx
->buffer
, reg_maps
->viewport_array
);
5045 shader_addline(ins
->ctx
->buffer
, "EmitVertex();\n");
5047 FIXME("Unhandled primitive stream %u.\n", stream
);
5050 static void shader_glsl_break(const struct wined3d_shader_instruction
*ins
)
5052 shader_addline(ins
->ctx
->buffer
, "break;\n");
5055 /* FIXME: According to MSDN the compare is done per component. */
5056 static void shader_glsl_breakc(const struct wined3d_shader_instruction
*ins
)
5058 struct glsl_src_param src0_param
;
5059 struct glsl_src_param src1_param
;
5061 shader_glsl_add_src_param(ins
, &ins
->src
[0], WINED3DSP_WRITEMASK_0
, &src0_param
);
5062 shader_glsl_add_src_param(ins
, &ins
->src
[1], WINED3DSP_WRITEMASK_0
, &src1_param
);
5064 shader_addline(ins
->ctx
->buffer
, "if (%s %s %s) break;\n",
5065 src0_param
.param_str
, shader_glsl_get_rel_op(ins
->flags
), src1_param
.param_str
);
5068 static void shader_glsl_conditional_op(const struct wined3d_shader_instruction
*ins
)
5072 switch (ins
->handler_idx
)
5074 case WINED3DSIH_BREAKP
:
5077 case WINED3DSIH_CONTINUEP
:
5080 case WINED3DSIH_RETP
:
5084 ERR("Unhandled opcode %#x.\n", ins
->handler_idx
);
5088 shader_glsl_generate_condition(ins
);
5089 if (ins
->handler_idx
== WINED3DSIH_RETP
)
5091 shader_addline(ins
->ctx
->buffer
, "{\n");
5092 shader_glsl_generate_shader_epilogue(ins
->ctx
);
5094 shader_addline(ins
->ctx
->buffer
, " %s\n", op
);
5095 if (ins
->handler_idx
== WINED3DSIH_RETP
)
5096 shader_addline(ins
->ctx
->buffer
, "}\n");
5099 static void shader_glsl_continue(const struct wined3d_shader_instruction
*ins
)
5101 shader_addline(ins
->ctx
->buffer
, "continue;\n");
5104 static void shader_glsl_label(const struct wined3d_shader_instruction
*ins
)
5106 shader_addline(ins
->ctx
->buffer
, "}\n");
5107 shader_addline(ins
->ctx
->buffer
, "void subroutine%u()\n{\n", ins
->src
[0].reg
.idx
[0].offset
);
5109 /* Subroutines appear at the end of the shader. */
5110 ins
->ctx
->state
->in_subroutine
= TRUE
;
5113 static void shader_glsl_call(const struct wined3d_shader_instruction
*ins
)
5115 shader_addline(ins
->ctx
->buffer
, "subroutine%u();\n", ins
->src
[0].reg
.idx
[0].offset
);
5118 static void shader_glsl_callnz(const struct wined3d_shader_instruction
*ins
)
5120 struct glsl_src_param src1_param
;
5122 shader_glsl_add_src_param(ins
, &ins
->src
[1], WINED3DSP_WRITEMASK_0
, &src1_param
);
5123 shader_addline(ins
->ctx
->buffer
, "if (%s) subroutine%u();\n",
5124 src1_param
.param_str
, ins
->src
[0].reg
.idx
[0].offset
);
5127 static void shader_glsl_ret(const struct wined3d_shader_instruction
*ins
)
5129 const struct wined3d_shader_version
*version
= &ins
->ctx
->shader
->reg_maps
.shader_version
;
5131 if (version
->major
>= 4 && !ins
->ctx
->state
->in_subroutine
)
5133 shader_glsl_generate_shader_epilogue(ins
->ctx
);
5134 shader_addline(ins
->ctx
->buffer
, "return;\n");
5138 static void shader_glsl_tex(const struct wined3d_shader_instruction
*ins
)
5140 DWORD shader_version
= WINED3D_SHADER_VERSION(ins
->ctx
->reg_maps
->shader_version
.major
,
5141 ins
->ctx
->reg_maps
->shader_version
.minor
);
5142 struct glsl_sample_function sample_function
;
5143 DWORD sample_flags
= 0;
5145 DWORD mask
= 0, swizzle
;
5146 const struct shader_glsl_ctx_priv
*priv
= ins
->ctx
->backend_data
;
5147 enum wined3d_shader_resource_type resource_type
;
5149 /* 1.0-1.4: Use destination register as sampler source.
5150 * 2.0+: Use provided sampler source. */
5151 if (shader_version
< WINED3D_SHADER_VERSION(2,0))
5152 resource_idx
= ins
->dst
[0].reg
.idx
[0].offset
;
5154 resource_idx
= ins
->src
[1].reg
.idx
[0].offset
;
5156 resource_type
= ins
->ctx
->reg_maps
->shader_version
.type
== WINED3D_SHADER_TYPE_PIXEL
5157 ? pixelshader_get_resource_type(ins
->ctx
->reg_maps
, resource_idx
, priv
->cur_ps_args
->tex_types
)
5158 : ins
->ctx
->reg_maps
->resource_info
[resource_idx
].type
;
5160 if (shader_version
< WINED3D_SHADER_VERSION(1,4))
5162 DWORD flags
= (priv
->cur_ps_args
->tex_transform
>> resource_idx
* WINED3D_PSARGS_TEXTRANSFORM_SHIFT
)
5163 & WINED3D_PSARGS_TEXTRANSFORM_MASK
;
5165 /* Projected cube textures don't make a lot of sense, the resulting coordinates stay the same. */
5166 if (flags
& WINED3D_PSARGS_PROJECTED
&& resource_type
!= WINED3D_SHADER_RESOURCE_TEXTURE_CUBE
)
5168 sample_flags
|= WINED3D_GLSL_SAMPLE_PROJECTED
;
5169 switch (flags
& ~WINED3D_PSARGS_PROJECTED
)
5171 case WINED3D_TTFF_COUNT1
:
5172 FIXME("WINED3D_TTFF_PROJECTED with WINED3D_TTFF_COUNT1?\n");
5174 case WINED3D_TTFF_COUNT2
:
5175 mask
= WINED3DSP_WRITEMASK_1
;
5177 case WINED3D_TTFF_COUNT3
:
5178 mask
= WINED3DSP_WRITEMASK_2
;
5180 case WINED3D_TTFF_COUNT4
:
5181 case WINED3D_TTFF_DISABLE
:
5182 mask
= WINED3DSP_WRITEMASK_3
;
5187 else if (shader_version
< WINED3D_SHADER_VERSION(2,0))
5189 enum wined3d_shader_src_modifier src_mod
= ins
->src
[0].modifiers
;
5191 if (src_mod
== WINED3DSPSM_DZ
) {
5192 sample_flags
|= WINED3D_GLSL_SAMPLE_PROJECTED
;
5193 mask
= WINED3DSP_WRITEMASK_2
;
5194 } else if (src_mod
== WINED3DSPSM_DW
) {
5195 sample_flags
|= WINED3D_GLSL_SAMPLE_PROJECTED
;
5196 mask
= WINED3DSP_WRITEMASK_3
;
5201 if ((ins
->flags
& WINED3DSI_TEXLD_PROJECT
)
5202 && resource_type
!= WINED3D_SHADER_RESOURCE_TEXTURE_CUBE
)
5204 /* ps 2.0 texldp instruction always divides by the fourth component. */
5205 sample_flags
|= WINED3D_GLSL_SAMPLE_PROJECTED
;
5206 mask
= WINED3DSP_WRITEMASK_3
;
5210 shader_glsl_get_sample_function(ins
->ctx
, resource_idx
, resource_idx
, sample_flags
, &sample_function
);
5211 mask
|= sample_function
.coord_mask
;
5212 sample_function
.coord_mask
= mask
;
5214 if (shader_version
< WINED3D_SHADER_VERSION(2,0)) swizzle
= WINED3DSP_NOSWIZZLE
;
5215 else swizzle
= ins
->src
[1].swizzle
;
5217 /* 1.0-1.3: Use destination register as coordinate source.
5218 1.4+: Use provided coordinate source register. */
5219 if (shader_version
< WINED3D_SHADER_VERSION(1,4))
5222 shader_glsl_write_mask_to_str(mask
, coord_mask
);
5223 shader_glsl_gen_sample_code(ins
, resource_idx
, &sample_function
, swizzle
, NULL
, NULL
, NULL
, NULL
,
5224 "T%u%s", resource_idx
, coord_mask
);
5228 struct glsl_src_param coord_param
;
5229 shader_glsl_add_src_param(ins
, &ins
->src
[0], mask
, &coord_param
);
5230 if (ins
->flags
& WINED3DSI_TEXLD_BIAS
)
5232 struct glsl_src_param bias
;
5233 shader_glsl_add_src_param(ins
, &ins
->src
[0], WINED3DSP_WRITEMASK_3
, &bias
);
5234 shader_glsl_gen_sample_code(ins
, resource_idx
, &sample_function
, swizzle
, NULL
, NULL
, bias
.param_str
,
5235 NULL
, "%s", coord_param
.param_str
);
5237 shader_glsl_gen_sample_code(ins
, resource_idx
, &sample_function
, swizzle
, NULL
, NULL
, NULL
, NULL
,
5238 "%s", coord_param
.param_str
);
5241 shader_glsl_release_sample_function(ins
->ctx
, &sample_function
);
5244 static void shader_glsl_texldd(const struct wined3d_shader_instruction
*ins
)
5246 const struct shader_glsl_ctx_priv
*priv
= ins
->ctx
->backend_data
;
5247 const struct wined3d_gl_info
*gl_info
= priv
->gl_info
;
5248 struct glsl_src_param coord_param
, dx_param
, dy_param
;
5249 struct glsl_sample_function sample_function
;
5251 DWORD swizzle
= ins
->src
[1].swizzle
;
5253 if (!shader_glsl_has_core_grad(gl_info
) && !gl_info
->supported
[ARB_SHADER_TEXTURE_LOD
])
5255 FIXME("texldd used, but not supported by hardware. Falling back to regular tex.\n");
5256 shader_glsl_tex(ins
);
5260 sampler_idx
= ins
->src
[1].reg
.idx
[0].offset
;
5262 shader_glsl_get_sample_function(ins
->ctx
, sampler_idx
, sampler_idx
, WINED3D_GLSL_SAMPLE_GRAD
, &sample_function
);
5263 shader_glsl_add_src_param(ins
, &ins
->src
[0], sample_function
.coord_mask
, &coord_param
);
5264 shader_glsl_add_src_param(ins
, &ins
->src
[2], sample_function
.deriv_mask
, &dx_param
);
5265 shader_glsl_add_src_param(ins
, &ins
->src
[3], sample_function
.deriv_mask
, &dy_param
);
5267 shader_glsl_gen_sample_code(ins
, sampler_idx
, &sample_function
, swizzle
, dx_param
.param_str
, dy_param
.param_str
,
5268 NULL
, NULL
, "%s", coord_param
.param_str
);
5269 shader_glsl_release_sample_function(ins
->ctx
, &sample_function
);
5272 static void shader_glsl_texldl(const struct wined3d_shader_instruction
*ins
)
5274 const struct wined3d_shader_version
*shader_version
= &ins
->ctx
->reg_maps
->shader_version
;
5275 const struct shader_glsl_ctx_priv
*priv
= ins
->ctx
->backend_data
;
5276 const struct wined3d_gl_info
*gl_info
= priv
->gl_info
;
5277 struct glsl_src_param coord_param
, lod_param
;
5278 struct glsl_sample_function sample_function
;
5279 DWORD swizzle
= ins
->src
[1].swizzle
;
5282 sampler_idx
= ins
->src
[1].reg
.idx
[0].offset
;
5284 shader_glsl_get_sample_function(ins
->ctx
, sampler_idx
, sampler_idx
, WINED3D_GLSL_SAMPLE_LOD
, &sample_function
);
5285 shader_glsl_add_src_param(ins
, &ins
->src
[0], sample_function
.coord_mask
, &coord_param
);
5287 shader_glsl_add_src_param(ins
, &ins
->src
[0], WINED3DSP_WRITEMASK_3
, &lod_param
);
5289 if (shader_version
->type
== WINED3D_SHADER_TYPE_PIXEL
&& !shader_glsl_has_core_grad(gl_info
)
5290 && !gl_info
->supported
[ARB_SHADER_TEXTURE_LOD
])
5292 /* Plain GLSL only supports Lod sampling functions in vertex shaders.
5293 * However, the NVIDIA drivers allow them in fragment shaders as well,
5294 * even without the appropriate extension. */
5295 WARN("Using %s in fragment shader.\n", sample_function
.name
->buffer
);
5297 shader_glsl_gen_sample_code(ins
, sampler_idx
, &sample_function
, swizzle
, NULL
, NULL
, lod_param
.param_str
, NULL
,
5298 "%s", coord_param
.param_str
);
5299 shader_glsl_release_sample_function(ins
->ctx
, &sample_function
);
5302 static unsigned int shader_glsl_find_sampler(const struct wined3d_shader_sampler_map
*sampler_map
,
5303 unsigned int resource_idx
, unsigned int sampler_idx
)
5305 struct wined3d_shader_sampler_map_entry
*entries
= sampler_map
->entries
;
5308 for (i
= 0; i
< sampler_map
->count
; ++i
)
5310 if (entries
[i
].resource_idx
== resource_idx
&& entries
[i
].sampler_idx
== sampler_idx
)
5311 return entries
[i
].bind_idx
;
5314 ERR("No GLSL sampler found for resource %u / sampler %u.\n", resource_idx
, sampler_idx
);
5319 static void shader_glsl_atomic(const struct wined3d_shader_instruction
*ins
)
5321 const BOOL is_imm_instruction
= WINED3DSIH_IMM_ATOMIC_AND
<= ins
->handler_idx
5322 && ins
->handler_idx
<= WINED3DSIH_IMM_ATOMIC_XOR
;
5323 const struct wined3d_shader_reg_maps
*reg_maps
= ins
->ctx
->reg_maps
;
5324 const struct wined3d_shader_version
*version
= ®_maps
->shader_version
;
5325 struct shader_glsl_ctx_priv
*priv
= ins
->ctx
->backend_data
;
5326 struct glsl_src_param structure_idx
, offset
, data
, data2
;
5327 struct wined3d_string_buffer
*buffer
= ins
->ctx
->buffer
;
5328 enum wined3d_shader_resource_type resource_type
;
5329 struct wined3d_string_buffer
*address
;
5330 enum wined3d_data_type data_type
;
5331 unsigned int resource_idx
, stride
;
5332 const char *op
, *resource
;
5336 resource_idx
= ins
->dst
[is_imm_instruction
].reg
.idx
[0].offset
;
5337 is_tgsm
= ins
->dst
[is_imm_instruction
].reg
.type
== WINED3DSPR_GROUPSHAREDMEM
;
5340 if (resource_idx
>= reg_maps
->tgsm_count
)
5342 ERR("Invalid TGSM index %u.\n", resource_idx
);
5346 data_type
= WINED3D_DATA_UINT
;
5348 stride
= reg_maps
->tgsm
[resource_idx
].stride
;
5352 if (resource_idx
>= ARRAY_SIZE(reg_maps
->uav_resource_info
))
5354 ERR("Invalid UAV index %u.\n", resource_idx
);
5357 resource_type
= reg_maps
->uav_resource_info
[resource_idx
].type
;
5358 if (resource_type
>= ARRAY_SIZE(resource_type_info
))
5360 ERR("Unexpected resource type %#x.\n", resource_type
);
5364 data_type
= reg_maps
->uav_resource_info
[resource_idx
].data_type
;
5365 coord_mask
= wined3d_mask_from_size(resource_type_info
[resource_type
].coord_size
);
5366 stride
= reg_maps
->uav_resource_info
[resource_idx
].stride
;
5369 switch (ins
->handler_idx
)
5371 case WINED3DSIH_ATOMIC_AND
:
5372 case WINED3DSIH_IMM_ATOMIC_AND
:
5376 op
= "imageAtomicAnd";
5378 case WINED3DSIH_ATOMIC_CMP_STORE
:
5379 case WINED3DSIH_IMM_ATOMIC_CMP_EXCH
:
5381 op
= "atomicCompSwap";
5383 op
= "imageAtomicCompSwap";
5385 case WINED3DSIH_ATOMIC_IADD
:
5386 case WINED3DSIH_IMM_ATOMIC_IADD
:
5390 op
= "imageAtomicAdd";
5392 case WINED3DSIH_ATOMIC_IMAX
:
5393 case WINED3DSIH_IMM_ATOMIC_IMAX
:
5397 op
= "imageAtomicMax";
5398 if (data_type
!= WINED3D_DATA_INT
)
5400 FIXME("Unhandled opcode %#x for unsigned integers.\n", ins
->handler_idx
);
5404 case WINED3DSIH_ATOMIC_IMIN
:
5405 case WINED3DSIH_IMM_ATOMIC_IMIN
:
5409 op
= "imageAtomicMin";
5410 if (data_type
!= WINED3D_DATA_INT
)
5412 FIXME("Unhandled opcode %#x for unsigned integers.\n", ins
->handler_idx
);
5416 case WINED3DSIH_ATOMIC_OR
:
5417 case WINED3DSIH_IMM_ATOMIC_OR
:
5421 op
= "imageAtomicOr";
5423 case WINED3DSIH_ATOMIC_UMAX
:
5424 case WINED3DSIH_IMM_ATOMIC_UMAX
:
5428 op
= "imageAtomicMax";
5429 if (data_type
!= WINED3D_DATA_UINT
)
5431 FIXME("Unhandled opcode %#x for signed integers.\n", ins
->handler_idx
);
5435 case WINED3DSIH_ATOMIC_UMIN
:
5436 case WINED3DSIH_IMM_ATOMIC_UMIN
:
5440 op
= "imageAtomicMin";
5441 if (data_type
!= WINED3D_DATA_UINT
)
5443 FIXME("Unhandled opcode %#x for signed integers.\n", ins
->handler_idx
);
5447 case WINED3DSIH_ATOMIC_XOR
:
5448 case WINED3DSIH_IMM_ATOMIC_XOR
:
5452 op
= "imageAtomicXor";
5454 case WINED3DSIH_IMM_ATOMIC_EXCH
:
5456 op
= "atomicExchange";
5458 op
= "imageAtomicExchange";
5461 ERR("Unhandled opcode %#x.\n", ins
->handler_idx
);
5465 address
= string_buffer_get(priv
->string_buffers
);
5468 shader_glsl_add_src_param(ins
, &ins
->src
[0], WINED3DSP_WRITEMASK_0
, &structure_idx
);
5469 shader_glsl_add_src_param(ins
, &ins
->src
[0], WINED3DSP_WRITEMASK_1
, &offset
);
5470 string_buffer_sprintf(address
, "%s * %u + %s / 4", structure_idx
.param_str
, stride
, offset
.param_str
);
5474 shader_glsl_add_src_param(ins
, &ins
->src
[0], coord_mask
, &offset
);
5475 string_buffer_sprintf(address
, "%s", offset
.param_str
);
5476 if (is_tgsm
|| (reg_maps
->uav_resource_info
[resource_idx
].flags
& WINED3D_VIEW_BUFFER_RAW
))
5477 shader_addline(address
, "/ 4");
5480 if (is_imm_instruction
)
5481 shader_glsl_append_dst_ext(ins
->ctx
->buffer
, ins
, &ins
->dst
[0], 0, data_type
);
5484 shader_addline(buffer
, "%s(%s_%s%u[%s], ",
5485 op
, shader_glsl_get_prefix(version
->type
), resource
, resource_idx
, address
->buffer
);
5487 shader_addline(buffer
, "%s(%s_%s%u, %s, ",
5488 op
, shader_glsl_get_prefix(version
->type
), resource
, resource_idx
, address
->buffer
);
5490 shader_glsl_add_src_param_ext(ins
->ctx
, &ins
->src
[1], WINED3DSP_WRITEMASK_0
, &data
, data_type
);
5491 shader_addline(buffer
, "%s", data
.param_str
);
5492 if (ins
->src_count
>= 3)
5494 shader_glsl_add_src_param_ext(ins
->ctx
, &ins
->src
[2], WINED3DSP_WRITEMASK_0
, &data2
, data_type
);
5495 shader_addline(buffer
, ", %s", data2
.param_str
);
5498 if (is_imm_instruction
)
5499 shader_addline(buffer
, ")");
5500 shader_addline(buffer
, ");\n");
5502 string_buffer_release(priv
->string_buffers
, address
);
5505 static void shader_glsl_uav_counter(const struct wined3d_shader_instruction
*ins
)
5507 const char *prefix
= shader_glsl_get_prefix(ins
->ctx
->reg_maps
->shader_version
.type
);
5510 if (ins
->handler_idx
== WINED3DSIH_IMM_ATOMIC_ALLOC
)
5511 op
= "atomicCounterIncrement";
5513 op
= "atomicCounterDecrement";
5515 shader_glsl_append_dst(ins
->ctx
->buffer
, ins
);
5516 shader_addline(ins
->ctx
->buffer
, "%s(%s_counter%u));\n", op
, prefix
, ins
->src
[0].reg
.idx
[0].offset
);
5519 static void shader_glsl_ld_uav(const struct wined3d_shader_instruction
*ins
)
5521 const struct wined3d_shader_reg_maps
*reg_maps
= ins
->ctx
->reg_maps
;
5522 const struct wined3d_shader_version
*version
= ®_maps
->shader_version
;
5523 enum wined3d_shader_resource_type resource_type
;
5524 struct glsl_src_param image_coord_param
;
5525 enum wined3d_data_type data_type
;
5526 DWORD coord_mask
, write_mask
;
5527 unsigned int uav_idx
;
5528 char dst_swizzle
[6];
5530 uav_idx
= ins
->src
[1].reg
.idx
[0].offset
;
5531 if (uav_idx
>= ARRAY_SIZE(reg_maps
->uav_resource_info
))
5533 ERR("Invalid UAV index %u.\n", uav_idx
);
5536 resource_type
= reg_maps
->uav_resource_info
[uav_idx
].type
;
5537 if (resource_type
>= ARRAY_SIZE(resource_type_info
))
5539 ERR("Unexpected resource type %#x.\n", resource_type
);
5540 resource_type
= WINED3D_SHADER_RESOURCE_TEXTURE_2D
;
5542 data_type
= reg_maps
->uav_resource_info
[uav_idx
].data_type
;
5543 coord_mask
= wined3d_mask_from_size(resource_type_info
[resource_type
].coord_size
);
5545 write_mask
= shader_glsl_append_dst_ext(ins
->ctx
->buffer
, ins
, &ins
->dst
[0], 0, data_type
);
5546 shader_glsl_get_swizzle(&ins
->src
[1], FALSE
, write_mask
, dst_swizzle
);
5548 shader_glsl_add_src_param(ins
, &ins
->src
[0], coord_mask
, &image_coord_param
);
5549 shader_addline(ins
->ctx
->buffer
, "imageLoad(%s_image%u, %s)%s);\n",
5550 shader_glsl_get_prefix(version
->type
), uav_idx
, image_coord_param
.param_str
, dst_swizzle
);
5553 static void shader_glsl_ld_raw_structured(const struct wined3d_shader_instruction
*ins
)
5555 const char *prefix
= shader_glsl_get_prefix(ins
->ctx
->reg_maps
->shader_version
.type
);
5556 const struct wined3d_shader_src_param
*src
= &ins
->src
[ins
->src_count
- 1];
5557 unsigned int i
, swizzle
, resource_idx
, bind_idx
, stride
, src_idx
= 0;
5558 const struct wined3d_shader_reg_maps
*reg_maps
= ins
->ctx
->reg_maps
;
5559 struct shader_glsl_ctx_priv
*priv
= ins
->ctx
->backend_data
;
5560 struct wined3d_string_buffer
*buffer
= ins
->ctx
->buffer
;
5561 struct glsl_src_param structure_idx
, offset
;
5562 struct wined3d_string_buffer
*address
;
5563 struct wined3d_shader_dst_param dst
;
5564 const char *function
, *resource
;
5566 resource_idx
= src
->reg
.idx
[0].offset
;
5567 if (src
->reg
.type
== WINED3DSPR_RESOURCE
)
5569 if (resource_idx
>= ARRAY_SIZE(reg_maps
->resource_info
))
5571 ERR("Invalid resource index %u.\n", resource_idx
);
5574 stride
= reg_maps
->resource_info
[resource_idx
].stride
;
5575 bind_idx
= shader_glsl_find_sampler(®_maps
->sampler_map
, resource_idx
, WINED3D_SAMPLER_DEFAULT
);
5576 function
= "texelFetch";
5577 resource
= "sampler";
5579 else if (src
->reg
.type
== WINED3DSPR_UAV
)
5581 if (resource_idx
>= ARRAY_SIZE(reg_maps
->uav_resource_info
))
5583 ERR("Invalid UAV index %u.\n", resource_idx
);
5586 stride
= reg_maps
->uav_resource_info
[resource_idx
].stride
;
5587 bind_idx
= resource_idx
;
5588 function
= "imageLoad";
5593 if (resource_idx
>= reg_maps
->tgsm_count
)
5595 ERR("Invalid TGSM index %u.\n", resource_idx
);
5598 stride
= reg_maps
->tgsm
[resource_idx
].stride
;
5599 bind_idx
= resource_idx
;
5604 address
= string_buffer_get(priv
->string_buffers
);
5605 if (ins
->handler_idx
== WINED3DSIH_LD_STRUCTURED
)
5607 shader_glsl_add_src_param(ins
, &ins
->src
[src_idx
++], WINED3DSP_WRITEMASK_0
, &structure_idx
);
5608 shader_addline(address
, "%s * %u + ", structure_idx
.param_str
, stride
);
5610 shader_glsl_add_src_param(ins
, &ins
->src
[src_idx
++], WINED3DSP_WRITEMASK_0
, &offset
);
5611 shader_addline(address
, "%s / 4", offset
.param_str
);
5614 if (shader_glsl_get_write_mask_size(dst
.write_mask
) > 1)
5616 /* The instruction is split into multiple lines. The first lines may
5617 * overwrite source parameters of the following lines. */
5618 shader_addline(buffer
, "tmp0.x = intBitsToFloat(%s);\n", address
->buffer
);
5619 string_buffer_sprintf(address
, "floatBitsToInt(tmp0.x)");
5622 for (i
= 0; i
< 4; ++i
)
5624 dst
.write_mask
= ins
->dst
[0].write_mask
& (WINED3DSP_WRITEMASK_0
<< i
);
5625 if (!shader_glsl_append_dst_ext(ins
->ctx
->buffer
, ins
, &dst
, 0, dst
.reg
.data_type
))
5628 swizzle
= shader_glsl_swizzle_get_component(src
->swizzle
, i
);
5630 shader_addline(buffer
, "%s(%s_%s%u, %s + %u).x);\n",
5631 function
, prefix
, resource
, bind_idx
, address
->buffer
, swizzle
);
5633 shader_addline(buffer
, "%s_%s%u[%s + %u]);\n",
5634 prefix
, resource
, bind_idx
, address
->buffer
, swizzle
);
5637 string_buffer_release(priv
->string_buffers
, address
);
5640 static void shader_glsl_store_uav(const struct wined3d_shader_instruction
*ins
)
5642 const struct wined3d_shader_reg_maps
*reg_maps
= ins
->ctx
->reg_maps
;
5643 const struct wined3d_shader_version
*version
= ®_maps
->shader_version
;
5644 struct glsl_src_param image_coord_param
, image_data_param
;
5645 enum wined3d_shader_resource_type resource_type
;
5646 enum wined3d_data_type data_type
;
5647 unsigned int uav_idx
;
5650 uav_idx
= ins
->dst
[0].reg
.idx
[0].offset
;
5651 if (uav_idx
>= ARRAY_SIZE(reg_maps
->uav_resource_info
))
5653 ERR("Invalid UAV index %u.\n", uav_idx
);
5656 resource_type
= reg_maps
->uav_resource_info
[uav_idx
].type
;
5657 if (resource_type
>= ARRAY_SIZE(resource_type_info
))
5659 ERR("Unexpected resource type %#x.\n", resource_type
);
5662 data_type
= reg_maps
->uav_resource_info
[uav_idx
].data_type
;
5663 coord_mask
= wined3d_mask_from_size(resource_type_info
[resource_type
].coord_size
);
5665 shader_glsl_add_src_param(ins
, &ins
->src
[0], coord_mask
, &image_coord_param
);
5666 shader_glsl_add_src_param_ext(ins
->ctx
, &ins
->src
[1], WINED3DSP_WRITEMASK_ALL
, &image_data_param
, data_type
);
5667 shader_addline(ins
->ctx
->buffer
, "imageStore(%s_image%u, %s, %s);\n",
5668 shader_glsl_get_prefix(version
->type
), uav_idx
,
5669 image_coord_param
.param_str
, image_data_param
.param_str
);
5672 static void shader_glsl_store_raw_structured(const struct wined3d_shader_instruction
*ins
)
5674 const char *prefix
= shader_glsl_get_prefix(ins
->ctx
->reg_maps
->shader_version
.type
);
5675 const struct wined3d_shader_reg_maps
*reg_maps
= ins
->ctx
->reg_maps
;
5676 struct shader_glsl_ctx_priv
*priv
= ins
->ctx
->backend_data
;
5677 struct wined3d_string_buffer
*buffer
= ins
->ctx
->buffer
;
5678 struct glsl_src_param structure_idx
, offset
, data
;
5679 unsigned int i
, resource_idx
, stride
, src_idx
= 0;
5680 struct wined3d_string_buffer
*address
;
5684 resource_idx
= ins
->dst
[0].reg
.idx
[0].offset
;
5685 is_tgsm
= ins
->dst
[0].reg
.type
== WINED3DSPR_GROUPSHAREDMEM
;
5688 if (resource_idx
>= reg_maps
->tgsm_count
)
5690 ERR("Invalid TGSM index %u.\n", resource_idx
);
5693 stride
= reg_maps
->tgsm
[resource_idx
].stride
;
5697 if (resource_idx
>= ARRAY_SIZE(reg_maps
->uav_resource_info
))
5699 ERR("Invalid UAV index %u.\n", resource_idx
);
5702 stride
= reg_maps
->uav_resource_info
[resource_idx
].stride
;
5705 address
= string_buffer_get(priv
->string_buffers
);
5706 if (ins
->handler_idx
== WINED3DSIH_STORE_STRUCTURED
)
5708 shader_glsl_add_src_param(ins
, &ins
->src
[src_idx
++], WINED3DSP_WRITEMASK_0
, &structure_idx
);
5709 shader_addline(address
, "%s * %u + ", structure_idx
.param_str
, stride
);
5711 shader_glsl_add_src_param(ins
, &ins
->src
[src_idx
++], WINED3DSP_WRITEMASK_0
, &offset
);
5712 shader_addline(address
, "%s / 4", offset
.param_str
);
5714 for (i
= 0; i
< 4; ++i
)
5716 if (!(write_mask
= ins
->dst
[0].write_mask
& (WINED3DSP_WRITEMASK_0
<< i
)))
5719 shader_glsl_add_src_param(ins
, &ins
->src
[src_idx
], write_mask
, &data
);
5722 shader_addline(buffer
, "%s_g%u[%s + %u] = %s;\n",
5723 prefix
, resource_idx
, address
->buffer
, i
, data
.param_str
);
5725 shader_addline(buffer
, "imageStore(%s_image%u, %s + %u, uvec4(%s, 0, 0, 0));\n",
5726 prefix
, resource_idx
, address
->buffer
, i
, data
.param_str
);
5729 string_buffer_release(priv
->string_buffers
, address
);
5732 static void shader_glsl_sync(const struct wined3d_shader_instruction
*ins
)
5734 struct wined3d_string_buffer
*buffer
= ins
->ctx
->buffer
;
5735 unsigned int sync_flags
= ins
->flags
;
5737 if (sync_flags
& WINED3DSSF_THREAD_GROUP
)
5739 shader_addline(buffer
, "barrier();\n");
5740 sync_flags
&= ~(WINED3DSSF_THREAD_GROUP
| WINED3DSSF_GROUP_SHARED_MEMORY
);
5743 if (sync_flags
& WINED3DSSF_GROUP_SHARED_MEMORY
)
5745 shader_addline(buffer
, "memoryBarrierShared();\n");
5746 sync_flags
&= ~WINED3DSSF_GROUP_SHARED_MEMORY
;
5749 if (sync_flags
& WINED3DSSF_GLOBAL_UAV
)
5751 shader_addline(buffer
, "memoryBarrier();\n");
5752 sync_flags
&= ~WINED3DSSF_GLOBAL_UAV
;
5756 FIXME("Unhandled sync flags %#x.\n", sync_flags
);
5759 static const struct wined3d_shader_resource_info
*shader_glsl_get_resource_info(
5760 const struct wined3d_shader_instruction
*ins
, const struct wined3d_shader_register
*reg
)
5762 const struct wined3d_shader_reg_maps
*reg_maps
= ins
->ctx
->reg_maps
;
5763 unsigned int idx
= reg
->idx
[0].offset
;
5765 if (reg
->type
== WINED3DSPR_RESOURCE
)
5767 if (idx
>= ARRAY_SIZE(reg_maps
->resource_info
))
5769 ERR("Invalid resource index %u.\n", idx
);
5772 return ®_maps
->resource_info
[idx
];
5775 if (reg
->type
== WINED3DSPR_UAV
)
5777 if (idx
>= ARRAY_SIZE(reg_maps
->uav_resource_info
))
5779 ERR("Invalid UAV index %u.\n", idx
);
5782 return ®_maps
->uav_resource_info
[idx
];
5785 FIXME("Unhandled register type %#x.\n", reg
->type
);
5789 static void shader_glsl_bufinfo(const struct wined3d_shader_instruction
*ins
)
5791 const char *prefix
= shader_glsl_get_prefix(ins
->ctx
->reg_maps
->shader_version
.type
);
5792 const struct wined3d_shader_resource_info
*resource_info
;
5793 struct wined3d_string_buffer
*buffer
= ins
->ctx
->buffer
;
5794 unsigned int resource_idx
;
5795 char dst_swizzle
[6];
5798 write_mask
= shader_glsl_append_dst(buffer
, ins
);
5799 shader_glsl_get_swizzle(&ins
->src
[0], FALSE
, write_mask
, dst_swizzle
);
5801 if (!(resource_info
= shader_glsl_get_resource_info(ins
, &ins
->src
[0].reg
)))
5803 resource_idx
= ins
->src
[0].reg
.idx
[0].offset
;
5805 shader_addline(buffer
, "ivec2(");
5806 if (ins
->src
[0].reg
.type
== WINED3DSPR_RESOURCE
)
5808 unsigned int bind_idx
= shader_glsl_find_sampler(&ins
->ctx
->reg_maps
->sampler_map
,
5809 resource_idx
, WINED3D_SAMPLER_DEFAULT
);
5810 shader_addline(buffer
, "textureSize(%s_sampler%u)", prefix
, bind_idx
);
5814 shader_addline(buffer
, "imageSize(%s_image%u)", prefix
, resource_idx
);
5816 if (resource_info
->stride
)
5817 shader_addline(buffer
, " / %u", resource_info
->stride
);
5818 else if (resource_info
->flags
& WINED3D_VIEW_BUFFER_RAW
)
5819 shader_addline(buffer
, " * 4");
5820 shader_addline(buffer
, ", %u)%s);\n", resource_info
->stride
, dst_swizzle
);
5823 static BOOL
is_multisampled(enum wined3d_shader_resource_type resource_type
)
5825 return resource_type
== WINED3D_SHADER_RESOURCE_TEXTURE_2DMS
5826 || resource_type
== WINED3D_SHADER_RESOURCE_TEXTURE_2DMSARRAY
;
5829 static BOOL
is_mipmapped(enum wined3d_shader_resource_type resource_type
)
5831 return resource_type
!= WINED3D_SHADER_RESOURCE_BUFFER
&& !is_multisampled(resource_type
);
5834 static void shader_glsl_resinfo(const struct wined3d_shader_instruction
*ins
)
5836 const struct wined3d_shader_version
*version
= &ins
->ctx
->reg_maps
->shader_version
;
5837 const struct shader_glsl_ctx_priv
*priv
= ins
->ctx
->backend_data
;
5838 struct wined3d_string_buffer
*buffer
= ins
->ctx
->buffer
;
5839 const struct wined3d_gl_info
*gl_info
= priv
->gl_info
;
5840 enum wined3d_shader_resource_type resource_type
;
5841 enum wined3d_shader_register_type reg_type
;
5842 unsigned int resource_idx
, bind_idx
, i
;
5843 enum wined3d_data_type dst_data_type
;
5844 struct glsl_src_param lod_param
;
5845 BOOL supports_mipmaps
;
5846 char dst_swizzle
[6];
5849 dst_data_type
= ins
->dst
[0].reg
.data_type
;
5850 if (ins
->flags
== WINED3DSI_RESINFO_UINT
)
5851 dst_data_type
= WINED3D_DATA_UINT
;
5852 else if (ins
->flags
)
5853 FIXME("Unhandled flags %#x.\n", ins
->flags
);
5855 reg_type
= ins
->src
[1].reg
.type
;
5856 resource_idx
= ins
->src
[1].reg
.idx
[0].offset
;
5857 shader_glsl_add_src_param(ins
, &ins
->src
[0], WINED3DSP_WRITEMASK_0
, &lod_param
);
5858 if (reg_type
== WINED3DSPR_RESOURCE
)
5860 resource_type
= ins
->ctx
->reg_maps
->resource_info
[resource_idx
].type
;
5861 bind_idx
= shader_glsl_find_sampler(&ins
->ctx
->reg_maps
->sampler_map
,
5862 resource_idx
, WINED3D_SAMPLER_DEFAULT
);
5866 resource_type
= ins
->ctx
->reg_maps
->uav_resource_info
[resource_idx
].type
;
5867 bind_idx
= resource_idx
;
5870 if (resource_type
>= ARRAY_SIZE(resource_type_info
))
5872 ERR("Unexpected resource type %#x.\n", resource_type
);
5876 write_mask
= shader_glsl_append_dst_ext(buffer
, ins
, &ins
->dst
[0], 0, dst_data_type
);
5877 shader_glsl_get_swizzle(&ins
->src
[1], FALSE
, write_mask
, dst_swizzle
);
5879 if (dst_data_type
== WINED3D_DATA_UINT
)
5880 shader_addline(buffer
, "uvec4(");
5882 shader_addline(buffer
, "vec4(");
5884 if (reg_type
== WINED3DSPR_RESOURCE
)
5886 shader_addline(buffer
, "textureSize(%s_sampler%u",
5887 shader_glsl_get_prefix(version
->type
), bind_idx
);
5891 shader_addline(buffer
, "imageSize(%s_image%u",
5892 shader_glsl_get_prefix(version
->type
), bind_idx
);
5895 supports_mipmaps
= is_mipmapped(resource_type
) && reg_type
!= WINED3DSPR_UAV
;
5896 if (supports_mipmaps
)
5897 shader_addline(buffer
, ", %s", lod_param
.param_str
);
5898 shader_addline(buffer
, "), ");
5900 for (i
= 0; i
< 3 - resource_type_info
[resource_type
].resinfo_size
; ++i
)
5901 shader_addline(buffer
, "0, ");
5903 if (supports_mipmaps
)
5905 if (gl_info
->supported
[ARB_TEXTURE_QUERY_LEVELS
])
5907 shader_addline(buffer
, "textureQueryLevels(%s_sampler%u)",
5908 shader_glsl_get_prefix(version
->type
), bind_idx
);
5912 FIXME("textureQueryLevels is not supported, returning 1 level.\n");
5913 shader_addline(buffer
, "1");
5918 shader_addline(buffer
, "1");
5921 shader_addline(buffer
, ")%s);\n", dst_swizzle
);
5924 static void shader_glsl_sample_info(const struct wined3d_shader_instruction
*ins
)
5926 const struct wined3d_shader_reg_maps
*reg_maps
= ins
->ctx
->reg_maps
;
5927 const struct shader_glsl_ctx_priv
*priv
= ins
->ctx
->backend_data
;
5928 const struct wined3d_gl_info
*gl_info
= priv
->gl_info
;
5929 struct wined3d_string_buffer
*buffer
= ins
->ctx
->buffer
;
5930 const struct wined3d_shader_dst_param
*dst
= ins
->dst
;
5931 const struct wined3d_shader_src_param
*src
= ins
->src
;
5932 enum wined3d_shader_resource_type resource_type
;
5933 enum wined3d_data_type dst_data_type
;
5934 unsigned int resource_idx
, bind_idx
;
5935 char dst_swizzle
[6];
5938 dst_data_type
= dst
->reg
.data_type
;
5939 if (ins
->flags
== WINED3DSI_SAMPLE_INFO_UINT
)
5940 dst_data_type
= WINED3D_DATA_UINT
;
5941 else if (ins
->flags
)
5942 FIXME("Unhandled flags %#x.\n", ins
->flags
);
5944 write_mask
= shader_glsl_append_dst_ext(buffer
, ins
, dst
, 0, dst_data_type
);
5945 shader_glsl_get_swizzle(src
, FALSE
, write_mask
, dst_swizzle
);
5947 if (dst_data_type
== WINED3D_DATA_UINT
)
5948 shader_addline(buffer
, "uvec4(");
5950 shader_addline(buffer
, "vec4(");
5952 if (src
->reg
.type
== WINED3DSPR_RASTERIZER
)
5954 if (gl_info
->supported
[ARB_SAMPLE_SHADING
])
5956 shader_addline(buffer
, "gl_NumSamples");
5960 FIXME("OpenGL implementation does not support ARB_sample_shading.\n");
5961 shader_addline(buffer
, "1");
5966 resource_idx
= src
->reg
.idx
[0].offset
;
5967 resource_type
= reg_maps
->resource_info
[resource_idx
].type
;
5968 if (resource_type
>= ARRAY_SIZE(resource_type_info
))
5970 ERR("Unexpected resource type %#x.\n", resource_type
);
5973 bind_idx
= shader_glsl_find_sampler(®_maps
->sampler_map
, resource_idx
, WINED3D_SAMPLER_DEFAULT
);
5975 if (gl_info
->supported
[ARB_SHADER_TEXTURE_IMAGE_SAMPLES
])
5977 shader_addline(buffer
, "textureSamples(%s_sampler%u)",
5978 shader_glsl_get_prefix(reg_maps
->shader_version
.type
), bind_idx
);
5982 FIXME("textureSamples() is not supported.\n");
5983 shader_addline(buffer
, "1");
5987 shader_addline(buffer
, ", 0, 0, 0)%s);\n", dst_swizzle
);
5990 static void shader_glsl_interpolate(const struct wined3d_shader_instruction
*ins
)
5992 const struct wined3d_shader_src_param
*input
= &ins
->src
[0];
5993 struct wined3d_string_buffer
*buffer
= ins
->ctx
->buffer
;
5994 struct glsl_src_param sample_param
;
5995 char dst_swizzle
[6];
5998 write_mask
= shader_glsl_append_dst(buffer
, ins
);
5999 shader_glsl_get_swizzle(input
, FALSE
, write_mask
, dst_swizzle
);
6001 shader_glsl_add_src_param(ins
, &ins
->src
[1], WINED3DSP_WRITEMASK_0
, &sample_param
);
6002 shader_addline(buffer
, "interpolateAtSample(shader_in.reg%u, %s)%s);\n",
6003 input
->reg
.idx
[0].offset
, sample_param
.param_str
, dst_swizzle
);
6006 static void shader_glsl_ld(const struct wined3d_shader_instruction
*ins
)
6008 const struct wined3d_shader_reg_maps
*reg_maps
= ins
->ctx
->reg_maps
;
6009 struct glsl_src_param coord_param
, lod_param
, sample_param
;
6010 unsigned int resource_idx
, sampler_idx
, sampler_bind_idx
;
6011 struct glsl_sample_function sample_function
;
6012 DWORD flags
= WINED3D_GLSL_SAMPLE_LOAD
;
6015 if (wined3d_shader_instruction_has_texel_offset(ins
))
6016 flags
|= WINED3D_GLSL_SAMPLE_OFFSET
;
6018 resource_idx
= ins
->src
[1].reg
.idx
[0].offset
;
6019 sampler_idx
= WINED3D_SAMPLER_DEFAULT
;
6021 if (resource_idx
>= ARRAY_SIZE(reg_maps
->resource_info
))
6023 ERR("Invalid resource index %u.\n", resource_idx
);
6026 has_lod_param
= is_mipmapped(reg_maps
->resource_info
[resource_idx
].type
);
6028 shader_glsl_get_sample_function(ins
->ctx
, resource_idx
, sampler_idx
, flags
, &sample_function
);
6029 shader_glsl_add_src_param(ins
, &ins
->src
[0], sample_function
.coord_mask
, &coord_param
);
6030 shader_glsl_add_src_param(ins
, &ins
->src
[0], WINED3DSP_WRITEMASK_3
, &lod_param
);
6031 sampler_bind_idx
= shader_glsl_find_sampler(®_maps
->sampler_map
, resource_idx
, sampler_idx
);
6032 if (is_multisampled(reg_maps
->resource_info
[resource_idx
].type
))
6034 shader_glsl_add_src_param(ins
, &ins
->src
[2], WINED3DSP_WRITEMASK_0
, &sample_param
);
6035 shader_glsl_gen_sample_code(ins
, sampler_bind_idx
, &sample_function
, ins
->src
[1].swizzle
,
6036 NULL
, NULL
, NULL
, &ins
->texel_offset
, "%s, %s", coord_param
.param_str
, sample_param
.param_str
);
6040 shader_glsl_gen_sample_code(ins
, sampler_bind_idx
, &sample_function
, ins
->src
[1].swizzle
,
6041 NULL
, NULL
, has_lod_param
? lod_param
.param_str
: NULL
, &ins
->texel_offset
,
6042 "%s", coord_param
.param_str
);
6044 shader_glsl_release_sample_function(ins
->ctx
, &sample_function
);
6047 static void shader_glsl_sample(const struct wined3d_shader_instruction
*ins
)
6049 const char *lod_param_str
= NULL
, *dx_param_str
= NULL
, *dy_param_str
= NULL
;
6050 struct glsl_src_param coord_param
, lod_param
, dx_param
, dy_param
;
6051 unsigned int resource_idx
, sampler_idx
, sampler_bind_idx
;
6052 struct glsl_sample_function sample_function
;
6055 if (ins
->handler_idx
== WINED3DSIH_SAMPLE_GRAD
)
6056 flags
|= WINED3D_GLSL_SAMPLE_GRAD
;
6057 if (ins
->handler_idx
== WINED3DSIH_SAMPLE_LOD
)
6058 flags
|= WINED3D_GLSL_SAMPLE_LOD
;
6059 if (wined3d_shader_instruction_has_texel_offset(ins
))
6060 flags
|= WINED3D_GLSL_SAMPLE_OFFSET
;
6062 resource_idx
= ins
->src
[1].reg
.idx
[0].offset
;
6063 sampler_idx
= ins
->src
[2].reg
.idx
[0].offset
;
6065 shader_glsl_get_sample_function(ins
->ctx
, resource_idx
, sampler_idx
, flags
, &sample_function
);
6066 shader_glsl_add_src_param(ins
, &ins
->src
[0], sample_function
.coord_mask
, &coord_param
);
6068 switch (ins
->handler_idx
)
6070 case WINED3DSIH_SAMPLE
:
6072 case WINED3DSIH_SAMPLE_B
:
6073 shader_glsl_add_src_param(ins
, &ins
->src
[3], WINED3DSP_WRITEMASK_0
, &lod_param
);
6074 lod_param_str
= lod_param
.param_str
;
6076 case WINED3DSIH_SAMPLE_GRAD
:
6077 shader_glsl_add_src_param(ins
, &ins
->src
[3], sample_function
.deriv_mask
, &dx_param
);
6078 shader_glsl_add_src_param(ins
, &ins
->src
[4], sample_function
.deriv_mask
, &dy_param
);
6079 dx_param_str
= dx_param
.param_str
;
6080 dy_param_str
= dy_param
.param_str
;
6082 case WINED3DSIH_SAMPLE_LOD
:
6083 shader_glsl_add_src_param(ins
, &ins
->src
[3], WINED3DSP_WRITEMASK_0
, &lod_param
);
6084 lod_param_str
= lod_param
.param_str
;
6087 ERR("Unhandled opcode %s.\n", debug_d3dshaderinstructionhandler(ins
->handler_idx
));
6091 sampler_bind_idx
= shader_glsl_find_sampler(&ins
->ctx
->reg_maps
->sampler_map
, resource_idx
, sampler_idx
);
6092 shader_glsl_gen_sample_code(ins
, sampler_bind_idx
, &sample_function
, ins
->src
[1].swizzle
,
6093 dx_param_str
, dy_param_str
, lod_param_str
, &ins
->texel_offset
, "%s", coord_param
.param_str
);
6094 shader_glsl_release_sample_function(ins
->ctx
, &sample_function
);
6097 /* Unless EXT_texture_shadow_lod is available, GLSL doesn't provide a function
6098 * to sample from level zero with depth comparison for array textures and cube
6099 * textures. We use textureGrad*() to implement sample_c_lz in that case. */
6100 static void shader_glsl_gen_sample_c_lz_emulation(const struct wined3d_shader_instruction
*ins
,
6101 unsigned int sampler_bind_idx
, const struct glsl_sample_function
*sample_function
,
6102 unsigned int coord_size
, const char *coord_param
, const char *ref_param
)
6104 const struct wined3d_shader_version
*version
= &ins
->ctx
->reg_maps
->shader_version
;
6105 unsigned int deriv_size
= wined3d_popcount(sample_function
->deriv_mask
);
6106 const struct wined3d_shader_texel_offset
*offset
= &ins
->texel_offset
;
6107 struct wined3d_string_buffer
*buffer
= ins
->ctx
->buffer
;
6108 char dst_swizzle
[6];
6110 WARN("Emitting textureGrad() for sample_c_lz.\n");
6112 shader_glsl_swizzle_to_str(WINED3DSP_NOSWIZZLE
, FALSE
, ins
->dst
[0].write_mask
, dst_swizzle
);
6113 shader_glsl_append_dst_ext(buffer
, ins
, &ins
->dst
[0], 0, sample_function
->data_type
);
6114 shader_addline(buffer
, "vec4(textureGrad%s(%s_sampler%u, vec%u(%s, %s), vec%u(0.0), vec%u(0.0)",
6115 sample_function
->offset_size
? "Offset" : "",
6116 shader_glsl_get_prefix(version
->type
), sampler_bind_idx
,
6117 coord_size
, coord_param
, ref_param
, deriv_size
, deriv_size
);
6118 if (sample_function
->offset_size
)
6120 int offset_immdata
[4] = {offset
->u
, offset
->v
, offset
->w
};
6121 shader_addline(buffer
, ", ");
6122 shader_glsl_append_imm_ivec(buffer
, offset_immdata
, sample_function
->offset_size
);
6124 shader_addline(buffer
, "))%s);\n", dst_swizzle
);
6127 static void shader_glsl_sample_c(const struct wined3d_shader_instruction
*ins
)
6129 const struct shader_glsl_ctx_priv
*priv
= ins
->ctx
->backend_data
;
6130 unsigned int resource_idx
, sampler_idx
, sampler_bind_idx
;
6131 const struct wined3d_shader_resource_info
*resource_info
;
6132 const struct wined3d_gl_info
*gl_info
= priv
->gl_info
;
6133 struct glsl_src_param coord_param
, compare_param
;
6134 struct glsl_sample_function sample_function
;
6135 const char *lod_param
= NULL
;
6136 unsigned int coord_size
;
6139 if (ins
->handler_idx
== WINED3DSIH_SAMPLE_C_LZ
)
6142 flags
|= WINED3D_GLSL_SAMPLE_LOD
;
6145 if (wined3d_shader_instruction_has_texel_offset(ins
))
6146 flags
|= WINED3D_GLSL_SAMPLE_OFFSET
;
6148 if (!(resource_info
= shader_glsl_get_resource_info(ins
, &ins
->src
[1].reg
)))
6150 resource_idx
= ins
->src
[1].reg
.idx
[0].offset
;
6151 sampler_idx
= ins
->src
[2].reg
.idx
[0].offset
;
6153 shader_glsl_get_sample_function(ins
->ctx
, resource_idx
, sampler_idx
, flags
, &sample_function
);
6154 coord_size
= shader_glsl_get_write_mask_size(sample_function
.coord_mask
);
6155 shader_glsl_add_src_param(ins
, &ins
->src
[0], sample_function
.coord_mask
>> 1, &coord_param
);
6156 shader_glsl_add_src_param(ins
, &ins
->src
[3], WINED3DSP_WRITEMASK_0
, &compare_param
);
6157 sampler_bind_idx
= shader_glsl_find_sampler(&ins
->ctx
->reg_maps
->sampler_map
, resource_idx
, sampler_idx
);
6158 if (ins
->handler_idx
== WINED3DSIH_SAMPLE_C_LZ
6159 && !gl_info
->supported
[EXT_TEXTURE_SHADOW_LOD
]
6160 && (resource_info
->type
== WINED3D_SHADER_RESOURCE_TEXTURE_2DARRAY
6161 || resource_info
->type
== WINED3D_SHADER_RESOURCE_TEXTURE_CUBE
))
6163 shader_glsl_gen_sample_c_lz_emulation(ins
, sampler_bind_idx
, &sample_function
,
6164 coord_size
, coord_param
.param_str
, compare_param
.param_str
);
6168 shader_glsl_gen_sample_code(ins
, sampler_bind_idx
, &sample_function
, WINED3DSP_NOSWIZZLE
,
6169 NULL
, NULL
, lod_param
, &ins
->texel_offset
, "vec%u(%s, %s)",
6170 coord_size
, coord_param
.param_str
, compare_param
.param_str
);
6172 shader_glsl_release_sample_function(ins
->ctx
, &sample_function
);
6175 static void shader_glsl_gather4(const struct wined3d_shader_instruction
*ins
)
6177 unsigned int resource_param_idx
, resource_idx
, sampler_idx
, sampler_bind_idx
, component_idx
;
6178 const struct wined3d_shader_reg_maps
*reg_maps
= ins
->ctx
->reg_maps
;
6179 const char *prefix
= shader_glsl_get_prefix(reg_maps
->shader_version
.type
);
6180 const struct shader_glsl_ctx_priv
*priv
= ins
->ctx
->backend_data
;
6181 struct glsl_src_param coord_param
, compare_param
, offset_param
;
6182 const struct wined3d_gl_info
*gl_info
= priv
->gl_info
;
6183 const struct wined3d_shader_resource_info
*resource_info
;
6184 struct wined3d_string_buffer
*buffer
= ins
->ctx
->buffer
;
6185 unsigned int coord_size
, offset_size
;
6186 char dst_swizzle
[6];
6189 if (!gl_info
->supported
[ARB_TEXTURE_GATHER
])
6191 FIXME("OpenGL implementation does not support textureGather.\n");
6195 has_offset
= ins
->handler_idx
== WINED3DSIH_GATHER4_PO
6196 || ins
->handler_idx
== WINED3DSIH_GATHER4_PO_C
6197 || wined3d_shader_instruction_has_texel_offset(ins
);
6199 resource_param_idx
=
6200 (ins
->handler_idx
== WINED3DSIH_GATHER4_PO
|| ins
->handler_idx
== WINED3DSIH_GATHER4_PO_C
) ? 2 : 1;
6201 resource_idx
= ins
->src
[resource_param_idx
].reg
.idx
[0].offset
;
6202 sampler_idx
= ins
->src
[resource_param_idx
+ 1].reg
.idx
[0].offset
;
6203 component_idx
= shader_glsl_swizzle_get_component(ins
->src
[resource_param_idx
+ 1].swizzle
, 0);
6204 sampler_bind_idx
= shader_glsl_find_sampler(®_maps
->sampler_map
, resource_idx
, sampler_idx
);
6206 if (!(resource_info
= shader_glsl_get_resource_info(ins
, &ins
->src
[resource_param_idx
].reg
)))
6209 if (resource_info
->type
>= ARRAY_SIZE(resource_type_info
))
6211 ERR("Unexpected resource type %#x.\n", resource_info
->type
);
6214 shader_glsl_get_coord_size(resource_info
->type
, &coord_size
, &offset_size
);
6216 shader_glsl_swizzle_to_str(ins
->src
[resource_param_idx
].swizzle
, FALSE
, ins
->dst
[0].write_mask
, dst_swizzle
);
6217 shader_glsl_append_dst_ext(buffer
, ins
, &ins
->dst
[0], 0, resource_info
->data_type
);
6219 shader_glsl_add_src_param(ins
, &ins
->src
[0], wined3d_mask_from_size(coord_size
), &coord_param
);
6221 shader_addline(buffer
, "textureGather%s(%s_sampler%u, %s",
6222 has_offset
? "Offset" : "", prefix
, sampler_bind_idx
, coord_param
.param_str
);
6223 if (ins
->handler_idx
== WINED3DSIH_GATHER4_C
|| ins
->handler_idx
== WINED3DSIH_GATHER4_PO_C
)
6225 shader_glsl_add_src_param(ins
, &ins
->src
[resource_param_idx
+ 2], WINED3DSP_WRITEMASK_0
, &compare_param
);
6226 shader_addline(buffer
, ", %s", compare_param
.param_str
);
6228 if (ins
->handler_idx
== WINED3DSIH_GATHER4_PO
|| ins
->handler_idx
== WINED3DSIH_GATHER4_PO_C
)
6230 shader_glsl_add_src_param(ins
, &ins
->src
[1], wined3d_mask_from_size(offset_size
), &offset_param
);
6231 shader_addline(buffer
, ", %s", offset_param
.param_str
);
6233 else if (has_offset
)
6235 int offset_immdata
[4] = {ins
->texel_offset
.u
, ins
->texel_offset
.v
, ins
->texel_offset
.w
};
6236 shader_addline(buffer
, ", ");
6237 shader_glsl_append_imm_ivec(buffer
, offset_immdata
, offset_size
);
6240 shader_addline(buffer
, ", %u", component_idx
);
6242 shader_addline(buffer
, ")%s);\n", dst_swizzle
);
6245 static void shader_glsl_texcoord(const struct wined3d_shader_instruction
*ins
)
6247 /* FIXME: Make this work for more than just 2D textures */
6248 struct wined3d_string_buffer
*buffer
= ins
->ctx
->buffer
;
6249 DWORD write_mask
= shader_glsl_append_dst(ins
->ctx
->buffer
, ins
);
6251 if (!(ins
->ctx
->reg_maps
->shader_version
.major
== 1 && ins
->ctx
->reg_maps
->shader_version
.minor
== 4))
6255 shader_glsl_get_write_mask(&ins
->dst
[0], dst_mask
);
6256 shader_addline(buffer
, "clamp(ffp_texcoord[%u], 0.0, 1.0)%s);\n",
6257 ins
->dst
[0].reg
.idx
[0].offset
, dst_mask
);
6261 enum wined3d_shader_src_modifier src_mod
= ins
->src
[0].modifiers
;
6262 DWORD reg
= ins
->src
[0].reg
.idx
[0].offset
;
6263 char dst_swizzle
[6];
6265 shader_glsl_get_swizzle(&ins
->src
[0], FALSE
, write_mask
, dst_swizzle
);
6267 if (src_mod
== WINED3DSPSM_DZ
|| src_mod
== WINED3DSPSM_DW
)
6269 unsigned int mask_size
= shader_glsl_get_write_mask_size(write_mask
);
6270 struct glsl_src_param div_param
;
6271 DWORD src_writemask
= src_mod
== WINED3DSPSM_DZ
? WINED3DSP_WRITEMASK_2
: WINED3DSP_WRITEMASK_3
;
6273 shader_glsl_add_src_param(ins
, &ins
->src
[0], src_writemask
, &div_param
);
6276 shader_addline(buffer
, "ffp_texcoord[%u]%s / vec%d(%s));\n", reg
, dst_swizzle
, mask_size
, div_param
.param_str
);
6278 shader_addline(buffer
, "ffp_texcoord[%u]%s / %s);\n", reg
, dst_swizzle
, div_param
.param_str
);
6282 shader_addline(buffer
, "ffp_texcoord[%u]%s);\n", reg
, dst_swizzle
);
6287 /** Process the WINED3DSIO_TEXDP3TEX instruction in GLSL:
6288 * Take a 3-component dot product of the TexCoord[dstreg] and src,
6289 * then perform a 1D texture lookup from stage dstregnum, place into dst. */
6290 static void shader_glsl_texdp3tex(const struct wined3d_shader_instruction
*ins
)
6292 DWORD src_mask
= WINED3DSP_WRITEMASK_0
| WINED3DSP_WRITEMASK_1
| WINED3DSP_WRITEMASK_2
;
6293 DWORD sampler_idx
= ins
->dst
[0].reg
.idx
[0].offset
;
6294 struct glsl_sample_function sample_function
;
6295 struct glsl_src_param src0_param
;
6298 shader_glsl_add_src_param(ins
, &ins
->src
[0], src_mask
, &src0_param
);
6300 /* Do I have to take care about the projected bit? I don't think so, since the dp3 returns only one
6301 * scalar, and projected sampling would require 4.
6303 * It is a dependent read - not valid with conditional NP2 textures
6305 shader_glsl_get_sample_function(ins
->ctx
, sampler_idx
, sampler_idx
, 0, &sample_function
);
6306 mask_size
= shader_glsl_get_write_mask_size(sample_function
.coord_mask
);
6311 shader_glsl_gen_sample_code(ins
, sampler_idx
, &sample_function
, WINED3DSP_NOSWIZZLE
, NULL
, NULL
, NULL
,
6312 NULL
, "dot(ffp_texcoord[%u].xyz, %s)", sampler_idx
, src0_param
.param_str
);
6316 shader_glsl_gen_sample_code(ins
, sampler_idx
, &sample_function
, WINED3DSP_NOSWIZZLE
, NULL
, NULL
, NULL
,
6317 NULL
, "vec2(dot(ffp_texcoord[%u].xyz, %s), 0.0)", sampler_idx
, src0_param
.param_str
);
6321 shader_glsl_gen_sample_code(ins
, sampler_idx
, &sample_function
, WINED3DSP_NOSWIZZLE
, NULL
, NULL
, NULL
,
6322 NULL
, "vec3(dot(ffp_texcoord[%u].xyz, %s), 0.0, 0.0)", sampler_idx
, src0_param
.param_str
);
6326 FIXME("Unexpected mask size %u\n", mask_size
);
6329 shader_glsl_release_sample_function(ins
->ctx
, &sample_function
);
6332 /** Process the WINED3DSIO_TEXDP3 instruction in GLSL:
6333 * Take a 3-component dot product of the TexCoord[dstreg] and src. */
6334 static void shader_glsl_texdp3(const struct wined3d_shader_instruction
*ins
)
6336 DWORD src_mask
= WINED3DSP_WRITEMASK_0
| WINED3DSP_WRITEMASK_1
| WINED3DSP_WRITEMASK_2
;
6337 DWORD dstreg
= ins
->dst
[0].reg
.idx
[0].offset
;
6338 struct glsl_src_param src0_param
;
6340 unsigned int mask_size
;
6342 dst_mask
= shader_glsl_append_dst(ins
->ctx
->buffer
, ins
);
6343 mask_size
= shader_glsl_get_write_mask_size(dst_mask
);
6344 shader_glsl_add_src_param(ins
, &ins
->src
[0], src_mask
, &src0_param
);
6346 if (mask_size
> 1) {
6347 shader_addline(ins
->ctx
->buffer
, "vec%d(dot(T%u.xyz, %s)));\n", mask_size
, dstreg
, src0_param
.param_str
);
6349 shader_addline(ins
->ctx
->buffer
, "dot(T%u.xyz, %s));\n", dstreg
, src0_param
.param_str
);
6353 /** Process the WINED3DSIO_TEXDEPTH instruction in GLSL:
6354 * Calculate the depth as dst.x / dst.y */
6355 static void shader_glsl_texdepth(const struct wined3d_shader_instruction
*ins
)
6357 struct glsl_dst_param dst_param
;
6359 shader_glsl_add_dst_param(ins
, &ins
->dst
[0], &dst_param
);
6361 /* Tests show that texdepth never returns anything below 0.0, and that r5.y is clamped to 1.0.
6362 * Negative input is accepted, -0.25 / -0.5 returns 0.5. GL should clamp gl_FragDepth to [0;1], but
6363 * this doesn't always work, so clamp the results manually. Whether or not the x value is clamped at 1
6364 * too is irrelevant, since if x = 0, any y value < 1.0 (and > 1.0 is not allowed) results in a result
6367 shader_addline(ins
->ctx
->buffer
, "gl_FragDepth = clamp((%s.x / min(%s.y, 1.0)), 0.0, 1.0);\n",
6368 dst_param
.reg_name
, dst_param
.reg_name
);
6371 /** Process the WINED3DSIO_TEXM3X2DEPTH instruction in GLSL:
6372 * Last row of a 3x2 matrix multiply, use the result to calculate the depth:
6373 * Calculate tmp0.y = TexCoord[dstreg] . src.xyz; (tmp0.x has already been calculated)
6374 * depth = (tmp0.y == 0.0) ? 1.0 : tmp0.x / tmp0.y
6376 static void shader_glsl_texm3x2depth(const struct wined3d_shader_instruction
*ins
)
6378 DWORD src_mask
= WINED3DSP_WRITEMASK_0
| WINED3DSP_WRITEMASK_1
| WINED3DSP_WRITEMASK_2
;
6379 DWORD dstreg
= ins
->dst
[0].reg
.idx
[0].offset
;
6380 struct glsl_src_param src0_param
;
6382 shader_glsl_add_src_param(ins
, &ins
->src
[0], src_mask
, &src0_param
);
6384 shader_addline(ins
->ctx
->buffer
, "tmp0.y = dot(T%u.xyz, %s);\n", dstreg
, src0_param
.param_str
);
6385 shader_addline(ins
->ctx
->buffer
, "gl_FragDepth = (tmp0.y == 0.0) ? 1.0 : clamp(tmp0.x / tmp0.y, 0.0, 1.0);\n");
6388 /** Process the WINED3DSIO_TEXM3X2PAD instruction in GLSL
6389 * Calculate the 1st of a 2-row matrix multiplication. */
6390 static void shader_glsl_texm3x2pad(const struct wined3d_shader_instruction
*ins
)
6392 DWORD src_mask
= WINED3DSP_WRITEMASK_0
| WINED3DSP_WRITEMASK_1
| WINED3DSP_WRITEMASK_2
;
6393 DWORD reg
= ins
->dst
[0].reg
.idx
[0].offset
;
6394 struct wined3d_string_buffer
*buffer
= ins
->ctx
->buffer
;
6395 struct glsl_src_param src0_param
;
6397 shader_glsl_add_src_param(ins
, &ins
->src
[0], src_mask
, &src0_param
);
6398 shader_addline(buffer
, "tmp0.x = dot(T%u.xyz, %s);\n", reg
, src0_param
.param_str
);
6401 /** Process the WINED3DSIO_TEXM3X3PAD instruction in GLSL
6402 * Calculate the 1st or 2nd row of a 3-row matrix multiplication. */
6403 static void shader_glsl_texm3x3pad(const struct wined3d_shader_instruction
*ins
)
6405 DWORD src_mask
= WINED3DSP_WRITEMASK_0
| WINED3DSP_WRITEMASK_1
| WINED3DSP_WRITEMASK_2
;
6406 struct wined3d_string_buffer
*buffer
= ins
->ctx
->buffer
;
6407 struct wined3d_shader_tex_mx
*tex_mx
= ins
->ctx
->tex_mx
;
6408 DWORD reg
= ins
->dst
[0].reg
.idx
[0].offset
;
6409 struct glsl_src_param src0_param
;
6411 shader_glsl_add_src_param(ins
, &ins
->src
[0], src_mask
, &src0_param
);
6412 shader_addline(buffer
, "tmp0.%c = dot(T%u.xyz, %s);\n", 'x' + tex_mx
->current_row
, reg
, src0_param
.param_str
);
6413 tex_mx
->texcoord_w
[tex_mx
->current_row
++] = reg
;
6416 static void shader_glsl_texm3x2tex(const struct wined3d_shader_instruction
*ins
)
6418 DWORD src_mask
= WINED3DSP_WRITEMASK_0
| WINED3DSP_WRITEMASK_1
| WINED3DSP_WRITEMASK_2
;
6419 struct wined3d_string_buffer
*buffer
= ins
->ctx
->buffer
;
6420 struct glsl_sample_function sample_function
;
6421 DWORD reg
= ins
->dst
[0].reg
.idx
[0].offset
;
6422 struct glsl_src_param src0_param
;
6424 shader_glsl_add_src_param(ins
, &ins
->src
[0], src_mask
, &src0_param
);
6425 shader_addline(buffer
, "tmp0.y = dot(T%u.xyz, %s);\n", reg
, src0_param
.param_str
);
6427 shader_glsl_get_sample_function(ins
->ctx
, reg
, reg
, 0, &sample_function
);
6429 /* Sample the texture using the calculated coordinates */
6430 shader_glsl_gen_sample_code(ins
, reg
, &sample_function
, WINED3DSP_NOSWIZZLE
, NULL
, NULL
, NULL
, NULL
, "tmp0.xy");
6431 shader_glsl_release_sample_function(ins
->ctx
, &sample_function
);
6434 /** Process the WINED3DSIO_TEXM3X3TEX instruction in GLSL
6435 * Perform the 3rd row of a 3x3 matrix multiply, then sample the texture using the calculated coordinates */
6436 static void shader_glsl_texm3x3tex(const struct wined3d_shader_instruction
*ins
)
6438 DWORD src_mask
= WINED3DSP_WRITEMASK_0
| WINED3DSP_WRITEMASK_1
| WINED3DSP_WRITEMASK_2
;
6439 struct wined3d_shader_tex_mx
*tex_mx
= ins
->ctx
->tex_mx
;
6440 struct glsl_sample_function sample_function
;
6441 DWORD reg
= ins
->dst
[0].reg
.idx
[0].offset
;
6442 struct glsl_src_param src0_param
;
6444 shader_glsl_add_src_param(ins
, &ins
->src
[0], src_mask
, &src0_param
);
6445 shader_addline(ins
->ctx
->buffer
, "tmp0.z = dot(T%u.xyz, %s);\n", reg
, src0_param
.param_str
);
6447 /* Dependent read, not valid with conditional NP2 */
6448 shader_glsl_get_sample_function(ins
->ctx
, reg
, reg
, 0, &sample_function
);
6450 /* Sample the texture using the calculated coordinates */
6451 shader_glsl_gen_sample_code(ins
, reg
, &sample_function
, WINED3DSP_NOSWIZZLE
, NULL
, NULL
, NULL
, NULL
, "tmp0.xyz");
6452 shader_glsl_release_sample_function(ins
->ctx
, &sample_function
);
6454 tex_mx
->current_row
= 0;
6457 /** Process the WINED3DSIO_TEXM3X3 instruction in GLSL
6458 * Perform the 3rd row of a 3x3 matrix multiply */
6459 static void shader_glsl_texm3x3(const struct wined3d_shader_instruction
*ins
)
6461 DWORD src_mask
= WINED3DSP_WRITEMASK_0
| WINED3DSP_WRITEMASK_1
| WINED3DSP_WRITEMASK_2
;
6462 struct wined3d_shader_tex_mx
*tex_mx
= ins
->ctx
->tex_mx
;
6463 DWORD reg
= ins
->dst
[0].reg
.idx
[0].offset
;
6464 struct glsl_src_param src0_param
;
6467 shader_glsl_add_src_param(ins
, &ins
->src
[0], src_mask
, &src0_param
);
6469 shader_glsl_append_dst(ins
->ctx
->buffer
, ins
);
6470 shader_glsl_get_write_mask(&ins
->dst
[0], dst_mask
);
6471 shader_addline(ins
->ctx
->buffer
, "vec4(tmp0.xy, dot(T%u.xyz, %s), 1.0)%s);\n", reg
, src0_param
.param_str
, dst_mask
);
6473 tex_mx
->current_row
= 0;
6476 /* Process the WINED3DSIO_TEXM3X3SPEC instruction in GLSL
6477 * Perform the final texture lookup based on the previous 2 3x3 matrix multiplies */
6478 static void shader_glsl_texm3x3spec(const struct wined3d_shader_instruction
*ins
)
6480 struct glsl_src_param src0_param
;
6481 struct glsl_src_param src1_param
;
6482 struct wined3d_string_buffer
*buffer
= ins
->ctx
->buffer
;
6483 struct wined3d_shader_tex_mx
*tex_mx
= ins
->ctx
->tex_mx
;
6484 DWORD src_mask
= WINED3DSP_WRITEMASK_0
| WINED3DSP_WRITEMASK_1
| WINED3DSP_WRITEMASK_2
;
6485 struct glsl_sample_function sample_function
;
6486 DWORD reg
= ins
->dst
[0].reg
.idx
[0].offset
;
6489 shader_glsl_add_src_param(ins
, &ins
->src
[0], src_mask
, &src0_param
);
6490 shader_glsl_add_src_param(ins
, &ins
->src
[1], src_mask
, &src1_param
);
6492 /* Perform the last matrix multiply operation */
6493 shader_addline(buffer
, "tmp0.z = dot(T%u.xyz, %s);\n", reg
, src0_param
.param_str
);
6494 /* Reflection calculation */
6495 shader_addline(buffer
, "tmp0.xyz = -reflect((%s), normalize(tmp0.xyz));\n", src1_param
.param_str
);
6497 /* Dependent read, not valid with conditional NP2 */
6498 shader_glsl_get_sample_function(ins
->ctx
, reg
, reg
, 0, &sample_function
);
6499 shader_glsl_write_mask_to_str(sample_function
.coord_mask
, coord_mask
);
6501 /* Sample the texture */
6502 shader_glsl_gen_sample_code(ins
, reg
, &sample_function
, WINED3DSP_NOSWIZZLE
,
6503 NULL
, NULL
, NULL
, NULL
, "tmp0%s", coord_mask
);
6504 shader_glsl_release_sample_function(ins
->ctx
, &sample_function
);
6506 tex_mx
->current_row
= 0;
6509 /* Process the WINED3DSIO_TEXM3X3VSPEC instruction in GLSL
6510 * Perform the final texture lookup based on the previous 2 3x3 matrix multiplies */
6511 static void shader_glsl_texm3x3vspec(const struct wined3d_shader_instruction
*ins
)
6513 struct wined3d_string_buffer
*buffer
= ins
->ctx
->buffer
;
6514 struct wined3d_shader_tex_mx
*tex_mx
= ins
->ctx
->tex_mx
;
6515 DWORD src_mask
= WINED3DSP_WRITEMASK_0
| WINED3DSP_WRITEMASK_1
| WINED3DSP_WRITEMASK_2
;
6516 struct glsl_sample_function sample_function
;
6517 DWORD reg
= ins
->dst
[0].reg
.idx
[0].offset
;
6518 struct glsl_src_param src0_param
;
6521 shader_glsl_add_src_param(ins
, &ins
->src
[0], src_mask
, &src0_param
);
6523 /* Perform the last matrix multiply operation */
6524 shader_addline(buffer
, "tmp0.z = dot(vec3(T%u), vec3(%s));\n", reg
, src0_param
.param_str
);
6526 /* Construct the eye-ray vector from w coordinates */
6527 shader_addline(buffer
, "tmp1.xyz = normalize(vec3(ffp_texcoord[%u].w, ffp_texcoord[%u].w, ffp_texcoord[%u].w));\n",
6528 tex_mx
->texcoord_w
[0], tex_mx
->texcoord_w
[1], reg
);
6529 shader_addline(buffer
, "tmp0.xyz = -reflect(tmp1.xyz, normalize(tmp0.xyz));\n");
6531 /* Dependent read, not valid with conditional NP2 */
6532 shader_glsl_get_sample_function(ins
->ctx
, reg
, reg
, 0, &sample_function
);
6533 shader_glsl_write_mask_to_str(sample_function
.coord_mask
, coord_mask
);
6535 /* Sample the texture using the calculated coordinates */
6536 shader_glsl_gen_sample_code(ins
, reg
, &sample_function
, WINED3DSP_NOSWIZZLE
,
6537 NULL
, NULL
, NULL
, NULL
, "tmp0%s", coord_mask
);
6538 shader_glsl_release_sample_function(ins
->ctx
, &sample_function
);
6540 tex_mx
->current_row
= 0;
6543 /** Process the WINED3DSIO_TEXBEM instruction in GLSL.
6544 * Apply a fake bump map transform.
6545 * texbem is pshader <= 1.3 only, this saves a few version checks
6547 static void shader_glsl_texbem(const struct wined3d_shader_instruction
*ins
)
6549 const struct shader_glsl_ctx_priv
*priv
= ins
->ctx
->backend_data
;
6550 struct glsl_sample_function sample_function
;
6551 struct glsl_src_param coord_param
;
6557 sampler_idx
= ins
->dst
[0].reg
.idx
[0].offset
;
6558 flags
= (priv
->cur_ps_args
->tex_transform
>> sampler_idx
* WINED3D_PSARGS_TEXTRANSFORM_SHIFT
)
6559 & WINED3D_PSARGS_TEXTRANSFORM_MASK
;
6561 /* Dependent read, not valid with conditional NP2 */
6562 shader_glsl_get_sample_function(ins
->ctx
, sampler_idx
, sampler_idx
, 0, &sample_function
);
6563 mask
= sample_function
.coord_mask
;
6565 shader_glsl_write_mask_to_str(mask
, coord_mask
);
6567 /* With projected textures, texbem only divides the static texture coord,
6568 * not the displacement, so we can't let GL handle this. */
6569 if (flags
& WINED3D_PSARGS_PROJECTED
)
6572 char coord_div_mask
[3];
6573 switch (flags
& ~WINED3D_PSARGS_PROJECTED
)
6575 case WINED3D_TTFF_COUNT1
:
6576 FIXME("WINED3D_TTFF_PROJECTED with WINED3D_TTFF_COUNT1?\n");
6578 case WINED3D_TTFF_COUNT2
:
6579 div_mask
= WINED3DSP_WRITEMASK_1
;
6581 case WINED3D_TTFF_COUNT3
:
6582 div_mask
= WINED3DSP_WRITEMASK_2
;
6584 case WINED3D_TTFF_COUNT4
:
6585 case WINED3D_TTFF_DISABLE
:
6586 div_mask
= WINED3DSP_WRITEMASK_3
;
6589 shader_glsl_write_mask_to_str(div_mask
, coord_div_mask
);
6590 shader_addline(ins
->ctx
->buffer
, "T%u%s /= T%u%s;\n", sampler_idx
, coord_mask
, sampler_idx
, coord_div_mask
);
6593 shader_glsl_add_src_param(ins
, &ins
->src
[0], WINED3DSP_WRITEMASK_0
| WINED3DSP_WRITEMASK_1
, &coord_param
);
6595 shader_glsl_gen_sample_code(ins
, sampler_idx
, &sample_function
, WINED3DSP_NOSWIZZLE
, NULL
, NULL
, NULL
, NULL
,
6596 "T%u%s + vec4(bumpenv_mat%u * %s, 0.0, 0.0)%s", sampler_idx
, coord_mask
, sampler_idx
,
6597 coord_param
.param_str
, coord_mask
);
6599 if (ins
->handler_idx
== WINED3DSIH_TEXBEML
)
6601 struct glsl_src_param luminance_param
;
6602 struct glsl_dst_param dst_param
;
6604 shader_glsl_add_src_param(ins
, &ins
->src
[0], WINED3DSP_WRITEMASK_2
, &luminance_param
);
6605 shader_glsl_add_dst_param(ins
, &ins
->dst
[0], &dst_param
);
6607 shader_addline(ins
->ctx
->buffer
, "%s%s *= (%s * bumpenv_lum_scale%u + bumpenv_lum_offset%u);\n",
6608 dst_param
.reg_name
, dst_param
.mask_str
,
6609 luminance_param
.param_str
, sampler_idx
, sampler_idx
);
6611 shader_glsl_release_sample_function(ins
->ctx
, &sample_function
);
6614 static void shader_glsl_bem(const struct wined3d_shader_instruction
*ins
)
6616 DWORD sampler_idx
= ins
->dst
[0].reg
.idx
[0].offset
;
6617 struct glsl_src_param src0_param
, src1_param
;
6619 shader_glsl_add_src_param(ins
, &ins
->src
[0], WINED3DSP_WRITEMASK_0
| WINED3DSP_WRITEMASK_1
, &src0_param
);
6620 shader_glsl_add_src_param(ins
, &ins
->src
[1], WINED3DSP_WRITEMASK_0
| WINED3DSP_WRITEMASK_1
, &src1_param
);
6622 shader_glsl_append_dst(ins
->ctx
->buffer
, ins
);
6623 shader_addline(ins
->ctx
->buffer
, "%s + bumpenv_mat%u * %s);\n",
6624 src0_param
.param_str
, sampler_idx
, src1_param
.param_str
);
6627 /** Process the WINED3DSIO_TEXREG2AR instruction in GLSL
6628 * Sample 2D texture at dst using the alpha & red (wx) components of src as texture coordinates */
6629 static void shader_glsl_texreg2ar(const struct wined3d_shader_instruction
*ins
)
6631 struct shader_glsl_ctx_priv
*priv
= ins
->ctx
->backend_data
;
6632 DWORD sampler_idx
= ins
->dst
[0].reg
.idx
[0].offset
;
6633 struct glsl_sample_function sample_function
;
6634 struct wined3d_string_buffer
*reg_name
;
6636 reg_name
= string_buffer_get(priv
->string_buffers
);
6637 shader_glsl_get_register_name(&ins
->src
[0].reg
, ins
->src
[0].reg
.data_type
, reg_name
, NULL
, ins
->ctx
);
6639 shader_glsl_get_sample_function(ins
->ctx
, sampler_idx
, sampler_idx
, 0, &sample_function
);
6640 shader_glsl_gen_sample_code(ins
, sampler_idx
, &sample_function
, WINED3DSP_NOSWIZZLE
, NULL
, NULL
, NULL
, NULL
,
6641 "%s.wx", reg_name
->buffer
);
6642 shader_glsl_release_sample_function(ins
->ctx
, &sample_function
);
6644 string_buffer_release(priv
->string_buffers
, reg_name
);
6647 /** Process the WINED3DSIO_TEXREG2GB instruction in GLSL
6648 * Sample 2D texture at dst using the green & blue (yz) components of src as texture coordinates */
6649 static void shader_glsl_texreg2gb(const struct wined3d_shader_instruction
*ins
)
6651 struct shader_glsl_ctx_priv
*priv
= ins
->ctx
->backend_data
;
6652 DWORD sampler_idx
= ins
->dst
[0].reg
.idx
[0].offset
;
6653 struct glsl_sample_function sample_function
;
6654 struct wined3d_string_buffer
*reg_name
;
6656 reg_name
= string_buffer_get(priv
->string_buffers
);
6657 shader_glsl_get_register_name(&ins
->src
[0].reg
, ins
->src
[0].reg
.data_type
, reg_name
, NULL
, ins
->ctx
);
6659 shader_glsl_get_sample_function(ins
->ctx
, sampler_idx
, sampler_idx
, 0, &sample_function
);
6660 shader_glsl_gen_sample_code(ins
, sampler_idx
, &sample_function
, WINED3DSP_NOSWIZZLE
, NULL
, NULL
, NULL
, NULL
,
6661 "%s.yz", reg_name
->buffer
);
6662 shader_glsl_release_sample_function(ins
->ctx
, &sample_function
);
6664 string_buffer_release(priv
->string_buffers
, reg_name
);
6667 /** Process the WINED3DSIO_TEXREG2RGB instruction in GLSL
6668 * Sample texture at dst using the rgb (xyz) components of src as texture coordinates */
6669 static void shader_glsl_texreg2rgb(const struct wined3d_shader_instruction
*ins
)
6671 DWORD sampler_idx
= ins
->dst
[0].reg
.idx
[0].offset
;
6672 struct glsl_sample_function sample_function
;
6673 struct glsl_src_param src0_param
;
6675 /* Dependent read, not valid with conditional NP2 */
6676 shader_glsl_get_sample_function(ins
->ctx
, sampler_idx
, sampler_idx
, 0, &sample_function
);
6677 shader_glsl_add_src_param(ins
, &ins
->src
[0], sample_function
.coord_mask
, &src0_param
);
6679 shader_glsl_gen_sample_code(ins
, sampler_idx
, &sample_function
, WINED3DSP_NOSWIZZLE
, NULL
, NULL
, NULL
, NULL
,
6680 "%s", src0_param
.param_str
);
6681 shader_glsl_release_sample_function(ins
->ctx
, &sample_function
);
6684 /** Process the WINED3DSIO_TEXKILL instruction in GLSL.
6685 * If any of the first 3 components are < 0, discard this pixel */
6686 static void shader_glsl_texkill(const struct wined3d_shader_instruction
*ins
)
6688 if (ins
->ctx
->reg_maps
->shader_version
.major
>= 4)
6690 shader_glsl_generate_condition(ins
);
6691 shader_addline(ins
->ctx
->buffer
, " discard;\n");
6695 struct glsl_dst_param dst_param
;
6697 /* The argument is a destination parameter, and no writemasks are allowed */
6698 shader_glsl_add_dst_param(ins
, &ins
->dst
[0], &dst_param
);
6700 /* 2.0 shaders compare all 4 components in texkill. */
6701 if (ins
->ctx
->reg_maps
->shader_version
.major
>= 2)
6702 shader_addline(ins
->ctx
->buffer
, "if (any(lessThan(%s.xyzw, vec4(0.0)))) discard;\n", dst_param
.reg_name
);
6703 /* 1.x shaders only compare the first 3 components, probably due to
6704 * the nature of the texkill instruction as a tex* instruction, and
6705 * phase, which kills all .w components. Even if all 4 components are
6706 * defined, only the first 3 are used. */
6708 shader_addline(ins
->ctx
->buffer
, "if (any(lessThan(%s.xyz, vec3(0.0)))) discard;\n", dst_param
.reg_name
);
6712 /** Process the WINED3DSIO_DP2ADD instruction in GLSL.
6713 * dst = dot2(src0, src1) + src2 */
6714 static void shader_glsl_dp2add(const struct wined3d_shader_instruction
*ins
)
6716 struct glsl_src_param src0_param
;
6717 struct glsl_src_param src1_param
;
6718 struct glsl_src_param src2_param
;
6720 unsigned int mask_size
;
6722 write_mask
= shader_glsl_append_dst(ins
->ctx
->buffer
, ins
);
6723 mask_size
= shader_glsl_get_write_mask_size(write_mask
);
6725 shader_glsl_add_src_param(ins
, &ins
->src
[0], WINED3DSP_WRITEMASK_0
| WINED3DSP_WRITEMASK_1
, &src0_param
);
6726 shader_glsl_add_src_param(ins
, &ins
->src
[1], WINED3DSP_WRITEMASK_0
| WINED3DSP_WRITEMASK_1
, &src1_param
);
6727 shader_glsl_add_src_param(ins
, &ins
->src
[2], WINED3DSP_WRITEMASK_0
, &src2_param
);
6729 if (mask_size
> 1) {
6730 shader_addline(ins
->ctx
->buffer
, "vec%d(dot(%s, %s) + %s));\n",
6731 mask_size
, src0_param
.param_str
, src1_param
.param_str
, src2_param
.param_str
);
6733 shader_addline(ins
->ctx
->buffer
, "dot(%s, %s) + %s);\n",
6734 src0_param
.param_str
, src1_param
.param_str
, src2_param
.param_str
);
6738 static void shader_glsl_input_pack(const struct wined3d_shader
*shader
, struct wined3d_string_buffer
*buffer
,
6739 const struct wined3d_shader_signature
*input_signature
,
6740 const struct wined3d_shader_reg_maps
*reg_maps
,
6741 const struct ps_compile_args
*args
, const struct wined3d_gl_info
*gl_info
, BOOL unroll
)
6745 for (i
= 0; i
< input_signature
->element_count
; ++i
)
6747 const struct wined3d_shader_signature_element
*input
= &input_signature
->elements
[i
];
6748 const char *semantic_name
;
6753 if (!(reg_maps
->input_registers
& (1u << input
->register_idx
)))
6756 semantic_name
= input
->semantic_name
;
6757 semantic_idx
= input
->semantic_idx
;
6758 shader_glsl_write_mask_to_str(input
->mask
, reg_mask
);
6760 if (args
->vp_mode
== WINED3D_VP_MODE_SHADER
)
6762 if (input
->sysval_semantic
== WINED3D_SV_POSITION
&& !semantic_idx
)
6764 shader_addline(buffer
, "ps_in[%u]%s = vpos%s;\n",
6765 shader
->u
.ps
.input_reg_map
[input
->register_idx
], reg_mask
, reg_mask
);
6767 else if (args
->pointsprite
&& shader_match_semantic(semantic_name
, WINED3D_DECL_USAGE_TEXCOORD
))
6769 shader_addline(buffer
, "ps_in[%u] = vec4(gl_PointCoord.xy, 0.0, 0.0);\n", input
->register_idx
);
6771 else if (input
->sysval_semantic
== WINED3D_SV_IS_FRONT_FACE
)
6773 shader_addline(buffer
, "ps_in[%u]%s = uintBitsToFloat(gl_FrontFacing ? 0xffffffffu : 0u);\n",
6774 input
->register_idx
, reg_mask
);
6776 else if (input
->sysval_semantic
== WINED3D_SV_SAMPLE_INDEX
)
6778 if (gl_info
->supported
[ARB_SAMPLE_SHADING
])
6779 shader_addline(buffer
, "ps_in[%u]%s = intBitsToFloat(gl_SampleID);\n",
6780 input
->register_idx
, reg_mask
);
6782 FIXME("ARB_sample_shading is not supported.\n");
6784 else if (input
->sysval_semantic
== WINED3D_SV_RENDER_TARGET_ARRAY_INDEX
&& !semantic_idx
)
6786 if (gl_info
->supported
[ARB_FRAGMENT_LAYER_VIEWPORT
])
6787 shader_addline(buffer
, "ps_in[%u]%s = intBitsToFloat(gl_Layer);\n",
6788 input
->register_idx
, reg_mask
);
6790 FIXME("ARB_fragment_layer_viewport is not supported.\n");
6792 else if (input
->sysval_semantic
== WINED3D_SV_VIEWPORT_ARRAY_INDEX
&& !semantic_idx
)
6794 if (gl_info
->supported
[ARB_VIEWPORT_ARRAY
])
6795 shader_addline(buffer
, "ps_in[%u]%s = intBitsToFloat(gl_ViewportIndex);\n",
6796 input
->register_idx
, reg_mask
);
6798 FIXME("ARB_viewport_array is not supported.\n");
6802 if (input
->sysval_semantic
)
6803 FIXME("Unhandled sysval semantic %#x.\n", input
->sysval_semantic
);
6804 shader_addline(buffer
, unroll
? "ps_in[%u]%s = %s%u%s;\n" : "ps_in[%u]%s = %s[%u]%s;\n",
6805 shader
->u
.ps
.input_reg_map
[input
->register_idx
], reg_mask
,
6806 shader_glsl_shader_input_name(gl_info
),
6807 shader
->u
.ps
.input_reg_map
[input
->register_idx
], reg_mask
);
6810 else if (shader_match_semantic(semantic_name
, WINED3D_DECL_USAGE_TEXCOORD
))
6812 if (args
->pointsprite
)
6813 shader_addline(buffer
, "ps_in[%u] = vec4(gl_PointCoord.xy, 0.0, 0.0);\n",
6814 shader
->u
.ps
.input_reg_map
[input
->register_idx
]);
6815 else if (args
->vp_mode
== WINED3D_VP_MODE_NONE
&& args
->texcoords_initialized
& (1u << semantic_idx
))
6816 shader_addline(buffer
, "ps_in[%u]%s = %s[%u]%s;\n",
6817 shader
->u
.ps
.input_reg_map
[input
->register_idx
], reg_mask
,
6818 needs_legacy_glsl_syntax(gl_info
)
6819 ? "gl_TexCoord" : "ffp_varying_texcoord", semantic_idx
, reg_mask
);
6821 shader_addline(buffer
, "ps_in[%u]%s = vec4(0.0, 0.0, 0.0, 0.0)%s;\n",
6822 shader
->u
.ps
.input_reg_map
[input
->register_idx
], reg_mask
, reg_mask
);
6824 else if (shader_match_semantic(semantic_name
, WINED3D_DECL_USAGE_COLOR
))
6827 shader_addline(buffer
, "ps_in[%u]%s = vec4(ffp_varying_diffuse)%s;\n",
6828 shader
->u
.ps
.input_reg_map
[input
->register_idx
], reg_mask
, reg_mask
);
6829 else if (semantic_idx
== 1)
6830 shader_addline(buffer
, "ps_in[%u]%s = vec4(ffp_varying_specular)%s;\n",
6831 shader
->u
.ps
.input_reg_map
[input
->register_idx
], reg_mask
, reg_mask
);
6833 shader_addline(buffer
, "ps_in[%u]%s = vec4(0.0, 0.0, 0.0, 0.0)%s;\n",
6834 shader
->u
.ps
.input_reg_map
[input
->register_idx
], reg_mask
, reg_mask
);
6838 shader_addline(buffer
, "ps_in[%u]%s = vec4(0.0, 0.0, 0.0, 0.0)%s;\n",
6839 shader
->u
.ps
.input_reg_map
[input
->register_idx
], reg_mask
, reg_mask
);
6844 static void add_glsl_program_entry(struct shader_glsl_priv
*priv
, struct glsl_shader_prog_link
*entry
)
6846 struct glsl_program_key key
;
6848 key
.vs_id
= entry
->vs
.id
;
6849 key
.hs_id
= entry
->hs
.id
;
6850 key
.ds_id
= entry
->ds
.id
;
6851 key
.gs_id
= entry
->gs
.id
;
6852 key
.ps_id
= entry
->ps
.id
;
6853 key
.cs_id
= entry
->cs
.id
;
6855 if (wine_rb_put(&priv
->program_lookup
, &key
, &entry
->program_lookup_entry
) == -1)
6857 ERR("Failed to insert program entry.\n");
6861 static struct glsl_shader_prog_link
*get_glsl_program_entry(const struct shader_glsl_priv
*priv
,
6862 const struct glsl_program_key
*key
)
6864 struct wine_rb_entry
*entry
;
6866 entry
= wine_rb_get(&priv
->program_lookup
, key
);
6867 return entry
? WINE_RB_ENTRY_VALUE(entry
, struct glsl_shader_prog_link
, program_lookup_entry
) : NULL
;
6870 /* Context activation is done by the caller. */
6871 static void delete_glsl_program_entry(struct shader_glsl_priv
*priv
, const struct wined3d_gl_info
*gl_info
,
6872 struct glsl_shader_prog_link
*entry
)
6874 wine_rb_remove(&priv
->program_lookup
, &entry
->program_lookup_entry
);
6876 GL_EXTCALL(glDeleteProgram(entry
->id
));
6878 list_remove(&entry
->vs
.shader_entry
);
6880 list_remove(&entry
->hs
.shader_entry
);
6882 list_remove(&entry
->ds
.shader_entry
);
6884 list_remove(&entry
->gs
.shader_entry
);
6886 list_remove(&entry
->ps
.shader_entry
);
6888 list_remove(&entry
->cs
.shader_entry
);
6892 static void shader_glsl_setup_vs3_output(struct shader_glsl_priv
*priv
,
6893 const struct wined3d_gl_info
*gl_info
, const DWORD
*map
,
6894 const struct wined3d_shader_signature
*input_signature
,
6895 const struct wined3d_shader_reg_maps
*reg_maps_in
,
6896 const struct wined3d_shader_signature
*output_signature
,
6897 const struct wined3d_shader_reg_maps
*reg_maps_out
)
6899 struct wined3d_string_buffer
*destination
= string_buffer_get(&priv
->string_buffers
);
6900 const char *out_array_name
= shader_glsl_shader_output_name(gl_info
);
6901 struct wined3d_string_buffer
*buffer
= &priv
->shader_buffer
;
6902 unsigned int in_count
= vec4_varyings(3, gl_info
);
6903 unsigned int max_varyings
= needs_legacy_glsl_syntax(gl_info
) ? in_count
+ 2 : in_count
;
6904 DWORD in_idx
, *set
= NULL
;
6908 set
= heap_calloc(max_varyings
, sizeof(*set
));
6910 for (i
= 0; i
< input_signature
->element_count
; ++i
)
6912 const struct wined3d_shader_signature_element
*input
= &input_signature
->elements
[i
];
6914 if (!(reg_maps_in
->input_registers
& (1u << input
->register_idx
)))
6917 in_idx
= map
[input
->register_idx
];
6918 /* Declared, but not read register */
6921 if (in_idx
>= max_varyings
)
6923 FIXME("More input varyings declared than supported, expect issues.\n");
6927 if (in_idx
== in_count
)
6928 string_buffer_sprintf(destination
, "gl_FrontColor");
6929 else if (in_idx
== in_count
+ 1)
6930 string_buffer_sprintf(destination
, "gl_FrontSecondaryColor");
6932 string_buffer_sprintf(destination
, "%s[%u]", out_array_name
, in_idx
);
6937 for (j
= 0; j
< output_signature
->element_count
; ++j
)
6939 const struct wined3d_shader_signature_element
*output
= &output_signature
->elements
[j
];
6942 if (!(reg_maps_out
->output_registers
& (1u << output
->register_idx
))
6943 || input
->semantic_idx
!= output
->semantic_idx
6944 || strcmp(input
->semantic_name
, output
->semantic_name
)
6945 || !(mask
= input
->mask
& output
->mask
))
6948 if (set
[in_idx
] == ~0u)
6950 set
[in_idx
] |= mask
& reg_maps_out
->u
.output_registers_mask
[output
->register_idx
];
6951 shader_glsl_write_mask_to_str(mask
, reg_mask
);
6953 shader_addline(buffer
, "%s%s = outputs[%u]%s;\n",
6954 destination
->buffer
, reg_mask
, output
->register_idx
, reg_mask
);
6958 for (i
= 0; i
< max_varyings
; ++i
)
6962 if (!set
[i
] || set
[i
] == WINED3DSP_WRITEMASK_ALL
)
6969 if (!(set
[i
] & WINED3DSP_WRITEMASK_0
))
6970 reg_mask
[size
++] = 'x';
6971 if (!(set
[i
] & WINED3DSP_WRITEMASK_1
))
6972 reg_mask
[size
++] = 'y';
6973 if (!(set
[i
] & WINED3DSP_WRITEMASK_2
))
6974 reg_mask
[size
++] = 'z';
6975 if (!(set
[i
] & WINED3DSP_WRITEMASK_3
))
6976 reg_mask
[size
++] = 'w';
6977 reg_mask
[size
] = '\0';
6980 string_buffer_sprintf(destination
, "gl_FrontColor");
6981 else if (i
== in_count
+ 1)
6982 string_buffer_sprintf(destination
, "gl_FrontSecondaryColor");
6984 string_buffer_sprintf(destination
, "%s[%u]", out_array_name
, i
);
6987 shader_addline(buffer
, "%s.%s = 0.0;\n", destination
->buffer
, reg_mask
);
6989 shader_addline(buffer
, "%s.%s = vec%u(0.0);\n", destination
->buffer
, reg_mask
, size
);
6993 string_buffer_release(&priv
->string_buffers
, destination
);
6996 static void shader_glsl_setup_sm4_shader_output(struct shader_glsl_priv
*priv
,
6997 unsigned int input_count
, const struct wined3d_shader_signature
*output_signature
,
6998 const struct wined3d_shader_reg_maps
*reg_maps_out
, const char *output_variable_name
,
6999 BOOL rasterizer_setup
)
7001 struct wined3d_string_buffer
*buffer
= &priv
->shader_buffer
;
7005 for (i
= 0; i
< output_signature
->element_count
; ++i
)
7007 const struct wined3d_shader_signature_element
*output
= &output_signature
->elements
[i
];
7009 if (!(reg_maps_out
->output_registers
& (1u << output
->register_idx
)))
7012 if (output
->stream_idx
)
7015 if (output
->register_idx
>= input_count
)
7018 shader_glsl_write_mask_to_str(output
->mask
, reg_mask
);
7020 shader_addline(buffer
,
7021 rasterizer_setup
? "%s.reg%u%s = outputs[%u]%s;\n" : "%s.reg[%u]%s = outputs[%u]%s;\n",
7022 output_variable_name
, output
->register_idx
, reg_mask
, output
->register_idx
, reg_mask
);
7026 static void shader_glsl_generate_clip_or_cull_distances(struct wined3d_string_buffer
*buffer
,
7027 const struct wined3d_shader_signature_element
*element
, DWORD clip_or_cull_distance_mask
)
7029 unsigned int i
, clip_or_cull_index
;
7033 name
= element
->sysval_semantic
== WINED3D_SV_CLIP_DISTANCE
? "Clip" : "Cull";
7034 /* Assign consecutive indices starting from 0. */
7035 clip_or_cull_index
= element
->semantic_idx
? wined3d_popcount(clip_or_cull_distance_mask
& 0xf) : 0;
7036 for (i
= 0; i
< 4; ++i
)
7038 if (!(element
->mask
& (WINED3DSP_WRITEMASK_0
<< i
)))
7041 shader_glsl_write_mask_to_str(WINED3DSP_WRITEMASK_0
<< i
, reg_mask
);
7042 shader_addline(buffer
, "gl_%sDistance[%u] = outputs[%u]%s;\n",
7043 name
, clip_or_cull_index
, element
->register_idx
, reg_mask
);
7044 ++clip_or_cull_index
;
7048 static void shader_glsl_setup_sm3_rasterizer_input(struct shader_glsl_priv
*priv
,
7049 const struct wined3d_gl_info
*gl_info
, const DWORD
*map
,
7050 const struct wined3d_shader_signature
*input_signature
,
7051 const struct wined3d_shader_reg_maps
*reg_maps_in
, unsigned int input_count
,
7052 const struct wined3d_shader_signature
*output_signature
,
7053 const struct wined3d_shader_reg_maps
*reg_maps_out
, BOOL per_vertex_point_size
)
7055 struct wined3d_string_buffer
*buffer
= &priv
->shader_buffer
;
7056 const char *semantic_name
;
7057 unsigned int semantic_idx
;
7061 /* First, sort out position and point size system values. */
7062 for (i
= 0; i
< output_signature
->element_count
; ++i
)
7064 const struct wined3d_shader_signature_element
*output
= &output_signature
->elements
[i
];
7066 if (!(reg_maps_out
->output_registers
& (1u << output
->register_idx
)))
7069 if (output
->stream_idx
)
7072 semantic_name
= output
->semantic_name
;
7073 semantic_idx
= output
->semantic_idx
;
7074 shader_glsl_write_mask_to_str(output
->mask
, reg_mask
);
7076 if (output
->sysval_semantic
== WINED3D_SV_POSITION
&& !semantic_idx
)
7078 shader_addline(buffer
, "gl_Position%s = outputs[%u]%s;\n",
7079 reg_mask
, output
->register_idx
, reg_mask
);
7081 else if (shader_match_semantic(semantic_name
, WINED3D_DECL_USAGE_PSIZE
) && per_vertex_point_size
)
7083 shader_addline(buffer
, "gl_PointSize = clamp(outputs[%u].%c, "
7084 "ffp_point.size_min, ffp_point.size_max);\n", output
->register_idx
, reg_mask
[1]);
7086 else if (output
->sysval_semantic
== WINED3D_SV_RENDER_TARGET_ARRAY_INDEX
&& !semantic_idx
)
7088 shader_addline(buffer
, "gl_Layer = floatBitsToInt(outputs[%u])%s;\n",
7089 output
->register_idx
, reg_mask
);
7091 else if (output
->sysval_semantic
== WINED3D_SV_VIEWPORT_ARRAY_INDEX
&& !semantic_idx
)
7093 shader_addline(buffer
, "gl_ViewportIndex = floatBitsToInt(outputs[%u])%s;\n",
7094 output
->register_idx
, reg_mask
);
7096 else if (output
->sysval_semantic
== WINED3D_SV_CLIP_DISTANCE
)
7098 shader_glsl_generate_clip_or_cull_distances(buffer
, output
, reg_maps_out
->clip_distance_mask
);
7100 else if (output
->sysval_semantic
== WINED3D_SV_CULL_DISTANCE
)
7102 shader_glsl_generate_clip_or_cull_distances(buffer
, output
, reg_maps_out
->cull_distance_mask
);
7104 else if (output
->sysval_semantic
)
7106 FIXME("Unhandled sysval semantic %#x.\n", output
->sysval_semantic
);
7110 /* Then, setup the pixel shader input. */
7111 if (reg_maps_out
->shader_version
.major
< 4)
7112 shader_glsl_setup_vs3_output(priv
, gl_info
, map
, input_signature
, reg_maps_in
,
7113 output_signature
, reg_maps_out
);
7115 shader_glsl_setup_sm4_shader_output(priv
, input_count
, output_signature
, reg_maps_out
, "shader_out", TRUE
);
7118 /* Context activation is done by the caller. */
7119 static GLuint
shader_glsl_generate_vs3_rasterizer_input_setup(struct shader_glsl_priv
*priv
,
7120 const struct wined3d_shader
*vs
, const struct wined3d_shader
*ps
,
7121 BOOL per_vertex_point_size
, BOOL flatshading
, const struct wined3d_gl_info
*gl_info
)
7123 const BOOL legacy_syntax
= needs_legacy_glsl_syntax(gl_info
);
7124 DWORD ps_major
= ps
? ps
->reg_maps
.shader_version
.major
: 0;
7125 struct wined3d_string_buffer
*buffer
= &priv
->shader_buffer
;
7126 const char *semantic_name
;
7132 string_buffer_clear(buffer
);
7134 shader_glsl_add_version_declaration(buffer
, gl_info
);
7136 shader_addline(buffer
, "invariant gl_Position;\n");
7138 if (per_vertex_point_size
)
7140 shader_addline(buffer
, "uniform struct\n{\n");
7141 shader_addline(buffer
, " float size_min;\n");
7142 shader_addline(buffer
, " float size_max;\n");
7143 shader_addline(buffer
, "} ffp_point;\n");
7148 DWORD colors_written_mask
[2] = {0};
7149 DWORD texcoords_written_mask
[WINED3D_MAX_TEXTURES
] = {0};
7153 declare_out_varying(gl_info
, buffer
, flatshading
, "vec4 ffp_varying_diffuse;\n");
7154 declare_out_varying(gl_info
, buffer
, flatshading
, "vec4 ffp_varying_specular;\n");
7155 declare_out_varying(gl_info
, buffer
, FALSE
, "vec4 ffp_varying_texcoord[%u];\n", WINED3D_MAX_TEXTURES
);
7156 declare_out_varying(gl_info
, buffer
, FALSE
, "float ffp_varying_fogcoord;\n");
7159 shader_addline(buffer
, "void setup_vs_output(in vec4 outputs[%u])\n{\n", vs
->limits
->packed_output
);
7161 for (i
= 0; i
< vs
->output_signature
.element_count
; ++i
)
7163 const struct wined3d_shader_signature_element
*output
= &vs
->output_signature
.elements
[i
];
7166 if (!(vs
->reg_maps
.output_registers
& (1u << output
->register_idx
)))
7169 semantic_name
= output
->semantic_name
;
7170 semantic_idx
= output
->semantic_idx
;
7171 write_mask
= output
->mask
;
7172 shader_glsl_write_mask_to_str(write_mask
, reg_mask
);
7174 if (shader_match_semantic(semantic_name
, WINED3D_DECL_USAGE_COLOR
) && semantic_idx
< 2)
7177 shader_addline(buffer
, "gl_Front%sColor%s = outputs[%u]%s;\n",
7178 semantic_idx
? "Secondary" : "", reg_mask
, output
->register_idx
, reg_mask
);
7180 shader_addline(buffer
, "ffp_varying_%s%s = clamp(outputs[%u]%s, 0.0, 1.0);\n",
7181 semantic_idx
? "specular" : "diffuse", reg_mask
, output
->register_idx
, reg_mask
);
7183 colors_written_mask
[semantic_idx
] = write_mask
;
7185 else if (shader_match_semantic(semantic_name
, WINED3D_DECL_USAGE_POSITION
) && !semantic_idx
)
7187 shader_addline(buffer
, "gl_Position%s = outputs[%u]%s;\n",
7188 reg_mask
, output
->register_idx
, reg_mask
);
7190 else if (shader_match_semantic(semantic_name
, WINED3D_DECL_USAGE_TEXCOORD
))
7192 if (semantic_idx
< WINED3D_MAX_TEXTURES
)
7194 shader_addline(buffer
, "%s[%u]%s = outputs[%u]%s;\n",
7195 legacy_syntax
? "gl_TexCoord" : "ffp_varying_texcoord",
7196 semantic_idx
, reg_mask
, output
->register_idx
, reg_mask
);
7197 texcoords_written_mask
[semantic_idx
] = write_mask
;
7200 else if (shader_match_semantic(semantic_name
, WINED3D_DECL_USAGE_PSIZE
) && per_vertex_point_size
)
7202 shader_addline(buffer
, "gl_PointSize = clamp(outputs[%u].%c, "
7203 "ffp_point.size_min, ffp_point.size_max);\n", output
->register_idx
, reg_mask
[1]);
7205 else if (shader_match_semantic(semantic_name
, WINED3D_DECL_USAGE_FOG
))
7207 shader_addline(buffer
, "%s = clamp(outputs[%u].%c, 0.0, 1.0);\n",
7208 legacy_syntax
? "gl_FogFragCoord" : "ffp_varying_fogcoord",
7209 output
->register_idx
, reg_mask
[1]);
7213 for (i
= 0; i
< 2; ++i
)
7215 if (colors_written_mask
[i
] != WINED3DSP_WRITEMASK_ALL
)
7217 shader_glsl_write_mask_to_str(~colors_written_mask
[i
] & WINED3DSP_WRITEMASK_ALL
, reg_mask
);
7219 shader_addline(buffer
, "%s%s = vec4(1.0)%s;\n",
7220 legacy_syntax
? "gl_FrontColor" : "ffp_varying_diffuse",
7221 reg_mask
, reg_mask
);
7223 shader_addline(buffer
, "%s%s = vec4(0.0)%s;\n",
7224 legacy_syntax
? "gl_FrontSecondaryColor" : "ffp_varying_specular",
7225 reg_mask
, reg_mask
);
7228 for (i
= 0; i
< WINED3D_MAX_TEXTURES
; ++i
)
7230 if (ps
&& !(ps
->reg_maps
.texcoord
& (1u << i
)))
7233 if (texcoords_written_mask
[i
] != WINED3DSP_WRITEMASK_ALL
)
7235 if (!shader_glsl_full_ffp_varyings(gl_info
) && !texcoords_written_mask
[i
])
7238 shader_glsl_write_mask_to_str(~texcoords_written_mask
[i
] & WINED3DSP_WRITEMASK_ALL
, reg_mask
);
7239 shader_addline(buffer
, "%s[%u]%s = vec4(0.0)%s;\n",
7240 legacy_syntax
? "gl_TexCoord" : "ffp_varying_texcoord", i
, reg_mask
, reg_mask
);
7246 unsigned int in_count
= min(vec4_varyings(ps_major
, gl_info
), ps
->limits
->packed_input
);
7248 shader_glsl_declare_shader_outputs(gl_info
, buffer
, in_count
, FALSE
, NULL
);
7249 shader_addline(buffer
, "void setup_vs_output(in vec4 outputs[%u])\n{\n", vs
->limits
->packed_output
);
7250 shader_glsl_setup_sm3_rasterizer_input(priv
, gl_info
, ps
->u
.ps
.input_reg_map
, &ps
->input_signature
,
7251 &ps
->reg_maps
, 0, &vs
->output_signature
, &vs
->reg_maps
, per_vertex_point_size
);
7254 shader_addline(buffer
, "}\n");
7256 ret
= GL_EXTCALL(glCreateShader(GL_VERTEX_SHADER
));
7257 checkGLcall("glCreateShader(GL_VERTEX_SHADER)");
7258 shader_glsl_compile(gl_info
, ret
, buffer
->buffer
);
7263 static void shader_glsl_generate_stream_output_setup(struct wined3d_string_buffer
*buffer
,
7264 const struct wined3d_shader
*shader
)
7266 const struct wined3d_stream_output_desc
*so_desc
= shader
->u
.gs
.so_desc
;
7267 unsigned int i
, register_idx
, component_idx
;
7269 shader_addline(buffer
, "out shader_in_out\n{\n");
7270 for (i
= 0; i
< so_desc
->element_count
; ++i
)
7272 const struct wined3d_stream_output_element
*e
= &so_desc
->elements
[i
];
7276 FIXME("Unhandled stream %u.\n", e
->stream_idx
);
7279 if (!e
->semantic_name
)
7281 if (!shader_get_stream_output_register_info(shader
, e
, ®ister_idx
, &component_idx
))
7284 if (component_idx
|| e
->component_count
!= 4)
7286 if (e
->component_count
== 1)
7287 shader_addline(buffer
, "float");
7289 shader_addline(buffer
, "vec%u", e
->component_count
);
7291 shader_addline(buffer
, " reg%u_%u_%u;\n",
7292 register_idx
, component_idx
, component_idx
+ e
->component_count
- 1);
7296 shader_addline(buffer
, "vec4 reg%u;\n", register_idx
);
7299 shader_addline(buffer
, "} shader_out;\n");
7301 shader_addline(buffer
, "void setup_gs_output(in vec4 outputs[%u])\n{\n",
7302 shader
->limits
->packed_output
);
7303 for (i
= 0; i
< so_desc
->element_count
; ++i
)
7305 const struct wined3d_stream_output_element
*e
= &so_desc
->elements
[i
];
7309 FIXME("Unhandled stream %u.\n", e
->stream_idx
);
7312 if (!e
->semantic_name
)
7314 if (!shader_get_stream_output_register_info(shader
, e
, ®ister_idx
, &component_idx
))
7317 if (component_idx
|| e
->component_count
!= 4)
7322 write_mask
= wined3d_mask_from_size(e
->component_count
) << component_idx
;
7323 shader_glsl_write_mask_to_str(write_mask
, str_mask
);
7324 shader_addline(buffer
, "shader_out.reg%u_%u_%u = outputs[%u]%s;\n",
7325 register_idx
, component_idx
, component_idx
+ e
->component_count
- 1,
7326 register_idx
, str_mask
);
7330 shader_addline(buffer
, "shader_out.reg%u = outputs[%u];\n", register_idx
, register_idx
);
7333 shader_addline(buffer
, "}\n");
7336 static void shader_glsl_generate_sm4_output_setup(struct shader_glsl_priv
*priv
,
7337 const struct wined3d_shader
*shader
, unsigned int input_count
,
7338 const struct wined3d_gl_info
*gl_info
, BOOL rasterizer_setup
, const DWORD
*interpolation_mode
)
7340 const char *prefix
= shader_glsl_get_prefix(shader
->reg_maps
.shader_version
.type
);
7341 struct wined3d_string_buffer
*buffer
= &priv
->shader_buffer
;
7343 if (rasterizer_setup
)
7344 input_count
= min(vec4_varyings(4, gl_info
), input_count
);
7347 shader_glsl_declare_shader_outputs(gl_info
, buffer
, input_count
, rasterizer_setup
, interpolation_mode
);
7349 shader_addline(buffer
, "void setup_%s_output(in vec4 outputs[%u])\n{\n",
7350 prefix
, shader
->limits
->packed_output
);
7352 if (rasterizer_setup
)
7353 shader_glsl_setup_sm3_rasterizer_input(priv
, gl_info
, NULL
, NULL
,
7354 NULL
, input_count
, &shader
->output_signature
, &shader
->reg_maps
, FALSE
);
7356 shader_glsl_setup_sm4_shader_output(priv
, input_count
, &shader
->output_signature
,
7357 &shader
->reg_maps
, "shader_out", rasterizer_setup
);
7359 shader_addline(buffer
, "}\n");
7362 static void shader_glsl_generate_patch_constant_name(struct wined3d_string_buffer
*buffer
,
7363 const struct wined3d_shader_signature_element
*constant
, unsigned int *user_constant_idx
,
7364 const char *reg_mask
)
7366 if (!constant
->sysval_semantic
)
7368 shader_addline(buffer
, "user_patch_constant[%u]%s", (*user_constant_idx
)++, reg_mask
);
7372 switch (constant
->sysval_semantic
)
7374 case WINED3D_SV_TESS_FACTOR_QUADEDGE
:
7375 case WINED3D_SV_TESS_FACTOR_TRIEDGE
:
7376 case WINED3D_SV_TESS_FACTOR_LINEDET
:
7377 case WINED3D_SV_TESS_FACTOR_LINEDEN
:
7378 shader_addline(buffer
, "gl_TessLevelOuter[%u]", constant
->semantic_idx
);
7380 case WINED3D_SV_TESS_FACTOR_QUADINT
:
7381 case WINED3D_SV_TESS_FACTOR_TRIINT
:
7382 shader_addline(buffer
, "gl_TessLevelInner[%u]", constant
->semantic_idx
);
7385 FIXME("Unhandled sysval semantic %#x.\n", constant
->sysval_semantic
);
7386 shader_addline(buffer
, "vec4(0.0)%s", reg_mask
);
7390 static void shader_glsl_generate_patch_constant_setup(struct wined3d_string_buffer
*buffer
,
7391 const struct wined3d_shader_signature
*signature
, BOOL input_setup
)
7393 unsigned int i
, register_count
, user_constant_index
, user_constant_count
;
7395 register_count
= user_constant_count
= 0;
7396 for (i
= 0; i
< signature
->element_count
; ++i
)
7398 const struct wined3d_shader_signature_element
*constant
= &signature
->elements
[i
];
7399 register_count
= max(constant
->register_idx
+ 1, register_count
);
7400 if (!constant
->sysval_semantic
)
7401 ++user_constant_count
;
7404 if (user_constant_count
)
7405 shader_addline(buffer
, "patch %s vec4 user_patch_constant[%u];\n",
7406 input_setup
? "in" : "out", user_constant_count
);
7408 shader_addline(buffer
, "vec4 vpc[%u];\n", register_count
);
7410 shader_addline(buffer
, "void setup_patch_constant_%s()\n{\n", input_setup
? "input" : "output");
7411 for (i
= 0, user_constant_index
= 0; i
< signature
->element_count
; ++i
)
7413 const struct wined3d_shader_signature_element
*constant
= &signature
->elements
[i
];
7416 shader_glsl_write_mask_to_str(constant
->mask
, reg_mask
);
7419 shader_addline(buffer
, "vpc[%u]%s", constant
->register_idx
, reg_mask
);
7421 shader_glsl_generate_patch_constant_name(buffer
, constant
, &user_constant_index
, reg_mask
);
7423 shader_addline(buffer
, " = ");
7426 shader_glsl_generate_patch_constant_name(buffer
, constant
, &user_constant_index
, reg_mask
);
7428 shader_addline(buffer
, "hs_out[%u]%s", constant
->register_idx
, reg_mask
);
7430 shader_addline(buffer
, ";\n");
7432 shader_addline(buffer
, "}\n");
7435 static void shader_glsl_generate_srgb_write_correction(struct wined3d_string_buffer
*buffer
,
7436 const struct wined3d_gl_info
*gl_info
)
7438 const char *output
= get_fragment_output(gl_info
);
7440 shader_addline(buffer
, "tmp0.xyz = pow(%s[0].xyz, vec3(srgb_const0.x));\n", output
);
7441 shader_addline(buffer
, "tmp0.xyz = tmp0.xyz * vec3(srgb_const0.y) - vec3(srgb_const0.z);\n");
7442 shader_addline(buffer
, "tmp1.xyz = %s[0].xyz * vec3(srgb_const0.w);\n", output
);
7443 shader_addline(buffer
, "bvec3 srgb_compare = lessThan(%s[0].xyz, vec3(srgb_const1.x));\n", output
);
7444 shader_addline(buffer
, "%s[0].xyz = mix(tmp0.xyz, tmp1.xyz, vec3(srgb_compare));\n", output
);
7445 shader_addline(buffer
, "%s[0] = clamp(%s[0], 0.0, 1.0);\n", output
, output
);
7448 static void shader_glsl_generate_fog_code(struct wined3d_string_buffer
*buffer
,
7449 const struct wined3d_gl_info
*gl_info
, enum wined3d_ffp_ps_fog_mode mode
)
7451 const char *output
= get_fragment_output(gl_info
);
7455 case WINED3D_FFP_PS_FOG_OFF
:
7458 case WINED3D_FFP_PS_FOG_LINEAR
:
7459 shader_addline(buffer
, "float fog = (ffp_fog.end - ffp_varying_fogcoord) * ffp_fog.scale;\n");
7462 case WINED3D_FFP_PS_FOG_EXP
:
7463 shader_addline(buffer
, "float fog = exp(-ffp_fog.density * ffp_varying_fogcoord);\n");
7466 case WINED3D_FFP_PS_FOG_EXP2
:
7467 shader_addline(buffer
, "float fog = exp(-ffp_fog.density * ffp_fog.density"
7468 " * ffp_varying_fogcoord * ffp_varying_fogcoord);\n");
7472 ERR("Invalid fog mode %#x.\n", mode
);
7476 shader_addline(buffer
, "%s[0].xyz = mix(ffp_fog.color.xyz, %s[0].xyz, clamp(fog, 0.0, 1.0));\n",
7480 static void shader_glsl_generate_alpha_test(struct wined3d_string_buffer
*buffer
,
7481 const struct wined3d_gl_info
*gl_info
, enum wined3d_cmp_func alpha_func
)
7483 /* alpha_func is the PASS condition, not the DISCARD condition. Instead of
7484 * flipping all the operators here, just negate the comparison below. */
7485 static const char * const comparison_operator
[] =
7487 "", /* WINED3D_CMP_NEVER */
7488 "<", /* WINED3D_CMP_LESS */
7489 "==", /* WINED3D_CMP_EQUAL */
7490 "<=", /* WINED3D_CMP_LESSEQUAL */
7491 ">", /* WINED3D_CMP_GREATER */
7492 "!=", /* WINED3D_CMP_NOTEQUAL */
7493 ">=", /* WINED3D_CMP_GREATEREQUAL */
7494 "" /* WINED3D_CMP_ALWAYS */
7497 if (alpha_func
== WINED3D_CMP_ALWAYS
)
7500 if (alpha_func
!= WINED3D_CMP_NEVER
)
7501 shader_addline(buffer
, "if (!(%s[0].a %s alpha_test_ref))\n",
7502 get_fragment_output(gl_info
), comparison_operator
[alpha_func
- WINED3D_CMP_NEVER
]);
7503 shader_addline(buffer
, " discard;\n");
7506 static void shader_glsl_generate_colour_key_test(struct wined3d_string_buffer
*buffer
,
7507 const char *colour
, const char *colour_key_low
, const char *colour_key_high
)
7509 shader_addline(buffer
, "if (all(greaterThanEqual(%s, %s)) && all(lessThan(%s, %s)))\n",
7510 colour
, colour_key_low
, colour
, colour_key_high
);
7511 shader_addline(buffer
, " discard;\n");
7514 static void shader_glsl_enable_extensions(struct wined3d_string_buffer
*buffer
,
7515 const struct wined3d_gl_info
*gl_info
)
7517 if (gl_info
->supported
[ARB_CULL_DISTANCE
])
7518 shader_addline(buffer
, "#extension GL_ARB_cull_distance : enable\n");
7519 if (gl_info
->supported
[ARB_GPU_SHADER5
])
7520 shader_addline(buffer
, "#extension GL_ARB_gpu_shader5 : enable\n");
7521 if (gl_info
->supported
[ARB_SHADER_ATOMIC_COUNTERS
])
7522 shader_addline(buffer
, "#extension GL_ARB_shader_atomic_counters : enable\n");
7523 if (gl_info
->supported
[ARB_SHADER_BIT_ENCODING
])
7524 shader_addline(buffer
, "#extension GL_ARB_shader_bit_encoding : enable\n");
7525 if (gl_info
->supported
[ARB_SHADER_IMAGE_LOAD_STORE
])
7526 shader_addline(buffer
, "#extension GL_ARB_shader_image_load_store : enable\n");
7527 if (gl_info
->supported
[ARB_SHADER_IMAGE_SIZE
])
7528 shader_addline(buffer
, "#extension GL_ARB_shader_image_size : enable\n");
7529 if (gl_info
->supported
[ARB_SHADER_STORAGE_BUFFER_OBJECT
])
7530 shader_addline(buffer
, "#extension GL_ARB_shader_storage_buffer_object : enable\n");
7531 if (gl_info
->supported
[ARB_SHADER_TEXTURE_IMAGE_SAMPLES
])
7532 shader_addline(buffer
, "#extension GL_ARB_shader_texture_image_samples : enable\n");
7533 if (gl_info
->supported
[ARB_SHADING_LANGUAGE_420PACK
])
7534 shader_addline(buffer
, "#extension GL_ARB_shading_language_420pack : enable\n");
7535 if (gl_info
->supported
[ARB_SHADING_LANGUAGE_PACKING
])
7536 shader_addline(buffer
, "#extension GL_ARB_shading_language_packing : enable\n");
7537 if (gl_info
->supported
[ARB_TEXTURE_CUBE_MAP_ARRAY
])
7538 shader_addline(buffer
, "#extension GL_ARB_texture_cube_map_array : enable\n");
7539 if (gl_info
->supported
[ARB_TEXTURE_GATHER
])
7540 shader_addline(buffer
, "#extension GL_ARB_texture_gather : enable\n");
7541 if (gl_info
->supported
[ARB_TEXTURE_QUERY_LEVELS
])
7542 shader_addline(buffer
, "#extension GL_ARB_texture_query_levels : enable\n");
7543 if (gl_info
->supported
[ARB_UNIFORM_BUFFER_OBJECT
])
7544 shader_addline(buffer
, "#extension GL_ARB_uniform_buffer_object : enable\n");
7545 if (gl_info
->supported
[ARB_VIEWPORT_ARRAY
])
7546 shader_addline(buffer
, "#extension GL_ARB_viewport_array : enable\n");
7547 if (gl_info
->supported
[EXT_GPU_SHADER4
])
7548 shader_addline(buffer
, "#extension GL_EXT_gpu_shader4 : enable\n");
7549 if (gl_info
->supported
[EXT_TEXTURE_ARRAY
])
7550 shader_addline(buffer
, "#extension GL_EXT_texture_array : enable\n");
7551 if (gl_info
->supported
[EXT_TEXTURE_SHADOW_LOD
])
7552 shader_addline(buffer
, "#extension GL_EXT_texture_shadow_lod : enable\n");
7555 static void shader_glsl_generate_color_output(struct wined3d_string_buffer
*buffer
,
7556 const struct wined3d_gl_info
*gl_info
, const struct wined3d_shader
*shader
,
7557 const struct ps_compile_args
*args
, struct wined3d_string_buffer_list
*string_buffers
)
7559 const struct wined3d_shader_signature
*output_signature
= &shader
->output_signature
;
7560 struct wined3d_string_buffer
*src
, *assignment
;
7561 enum wined3d_data_type dst_data_type
;
7562 const char *swizzle
;
7565 if (output_signature
->element_count
)
7567 src
= string_buffer_get(string_buffers
);
7568 assignment
= string_buffer_get(string_buffers
);
7569 for (i
= 0; i
< output_signature
->element_count
; ++i
)
7571 const struct wined3d_shader_signature_element
*output
= &output_signature
->elements
[i
];
7573 /* register_idx is set to ~0u for non-color outputs. */
7574 if (output
->register_idx
== ~0u)
7576 if ((unsigned int)output
->component_type
>= ARRAY_SIZE(component_type_info
))
7578 FIXME("Unhandled component type %#x.\n", output
->component_type
);
7581 dst_data_type
= component_type_info
[output
->component_type
].data_type
;
7582 shader_addline(buffer
, "color_out%u = ", output
->semantic_idx
);
7583 string_buffer_sprintf(src
, "ps_out[%u]", output
->semantic_idx
);
7584 shader_glsl_sprintf_cast(assignment
, src
->buffer
, dst_data_type
, WINED3D_DATA_FLOAT
, 4);
7585 swizzle
= args
->rt_alpha_swizzle
& (1u << output
->semantic_idx
) ? ".argb" : "";
7586 shader_addline(buffer
, "%s%s;\n", assignment
->buffer
, swizzle
);
7588 string_buffer_release(string_buffers
, src
);
7589 string_buffer_release(string_buffers
, assignment
);
7593 DWORD mask
= shader
->reg_maps
.rt_mask
;
7597 i
= wined3d_bit_scan(&mask
);
7598 swizzle
= args
->rt_alpha_swizzle
& (1u << i
) ? ".argb" : "";
7599 shader_addline(buffer
, "color_out%u = ps_out[%u]%s;\n", i
, i
, swizzle
);
7604 static void shader_glsl_generate_ps_epilogue(const struct wined3d_gl_info
*gl_info
,
7605 struct wined3d_string_buffer
*buffer
, const struct wined3d_shader
*shader
,
7606 const struct ps_compile_args
*args
, struct wined3d_string_buffer_list
*string_buffers
)
7608 const struct wined3d_shader_reg_maps
*reg_maps
= &shader
->reg_maps
;
7610 /* Pixel shaders < 2.0 place the resulting color in R0 implicitly. */
7611 if (reg_maps
->shader_version
.major
< 2)
7612 shader_addline(buffer
, "%s[0] = R0;\n", get_fragment_output(gl_info
));
7614 if (args
->srgb_correction
)
7615 shader_glsl_generate_srgb_write_correction(buffer
, gl_info
);
7617 /* SM < 3 does not replace the fog stage. */
7618 if (reg_maps
->shader_version
.major
< 3)
7619 shader_glsl_generate_fog_code(buffer
, gl_info
, args
->fog
);
7621 shader_glsl_generate_alpha_test(buffer
, gl_info
, args
->alpha_test_func
+ 1);
7623 if (reg_maps
->sample_mask
)
7624 shader_addline(buffer
, "gl_SampleMask[0] = floatBitsToInt(sample_mask);\n");
7626 if (!use_legacy_fragment_output(gl_info
))
7627 shader_glsl_generate_color_output(buffer
, gl_info
, shader
, args
, string_buffers
);
7630 /* Context activation is done by the caller. */
7631 static GLuint
shader_glsl_generate_fragment_shader(const struct wined3d_context_gl
*context_gl
,
7632 struct wined3d_string_buffer
*buffer
, struct wined3d_string_buffer_list
*string_buffers
,
7633 const struct wined3d_shader
*shader
, const struct ps_compile_args
*args
,
7634 struct ps_np2fixup_info
*np2fixup_info
)
7636 const struct wined3d_shader_reg_maps
*reg_maps
= &shader
->reg_maps
;
7637 const struct wined3d_shader_version
*version
= ®_maps
->shader_version
;
7638 const struct wined3d_gl_info
*gl_info
= context_gl
->gl_info
;
7639 const char *prefix
= shader_glsl_get_prefix(version
->type
);
7640 BOOL legacy_syntax
= needs_legacy_glsl_syntax(gl_info
);
7641 unsigned int i
, extra_constants_needed
= 0;
7642 struct shader_glsl_ctx_priv priv_ctx
;
7646 memset(&priv_ctx
, 0, sizeof(priv_ctx
));
7647 priv_ctx
.gl_info
= gl_info
;
7648 priv_ctx
.cur_ps_args
= args
;
7649 priv_ctx
.cur_np2fixup_info
= np2fixup_info
;
7650 priv_ctx
.string_buffers
= string_buffers
;
7652 shader_glsl_add_version_declaration(buffer
, gl_info
);
7654 shader_glsl_enable_extensions(buffer
, gl_info
);
7655 if (gl_info
->supported
[ARB_CONSERVATIVE_DEPTH
])
7656 shader_addline(buffer
, "#extension GL_ARB_conservative_depth : enable\n");
7657 if (gl_info
->supported
[ARB_DERIVATIVE_CONTROL
])
7658 shader_addline(buffer
, "#extension GL_ARB_derivative_control : enable\n");
7659 if (shader_glsl_use_explicit_attrib_location(gl_info
))
7660 shader_addline(buffer
, "#extension GL_ARB_explicit_attrib_location : enable\n");
7661 if (gl_info
->supported
[ARB_FRAGMENT_COORD_CONVENTIONS
])
7662 shader_addline(buffer
, "#extension GL_ARB_fragment_coord_conventions : enable\n");
7663 if (gl_info
->supported
[ARB_FRAGMENT_LAYER_VIEWPORT
])
7664 shader_addline(buffer
, "#extension GL_ARB_fragment_layer_viewport : enable\n");
7665 if (gl_info
->supported
[ARB_SAMPLE_SHADING
])
7666 shader_addline(buffer
, "#extension GL_ARB_sample_shading : enable\n");
7667 if (gl_info
->supported
[ARB_SHADER_TEXTURE_LOD
])
7668 shader_addline(buffer
, "#extension GL_ARB_shader_texture_lod : enable\n");
7669 /* The spec says that it doesn't have to be explicitly enabled, but the
7670 * nvidia drivers write a warning if we don't do so. */
7671 if (gl_info
->supported
[ARB_TEXTURE_RECTANGLE
])
7672 shader_addline(buffer
, "#extension GL_ARB_texture_rectangle : enable\n");
7674 /* Base Declarations */
7675 shader_generate_glsl_declarations(context_gl
, buffer
, shader
, reg_maps
, &priv_ctx
);
7677 if (gl_info
->supported
[ARB_CONSERVATIVE_DEPTH
])
7679 if (shader
->u
.ps
.depth_output
== WINED3DSPR_DEPTHOUTGE
)
7680 shader_addline(buffer
, "layout (depth_greater) out float gl_FragDepth;\n");
7681 else if (shader
->u
.ps
.depth_output
== WINED3DSPR_DEPTHOUTLE
)
7682 shader_addline(buffer
, "layout (depth_less) out float gl_FragDepth;\n");
7685 /* Declare uniforms for NP2 texcoord fixup:
7686 * This is NOT done inside the loop that declares the texture samplers
7687 * since the NP2 fixup code is currently only used for the GeforceFX
7688 * series and when forcing the ARB_npot extension off. Modern cards just
7689 * skip the code anyway, so put it inside a separate loop. */
7690 if (args
->np2_fixup
)
7692 struct ps_np2fixup_info
*fixup
= priv_ctx
.cur_np2fixup_info
;
7693 unsigned int cur
= 0;
7695 /* NP2/RECT textures in OpenGL use texcoords in the range [0,width]x[0,height]
7696 * while D3D has them in the (normalized) [0,1]x[0,1] range.
7697 * samplerNP2Fixup stores texture dimensions and is updated through
7698 * shader_glsl_load_np2fixup_constants when the sampler changes. */
7700 for (i
= 0; i
< shader
->limits
->sampler
; ++i
)
7702 enum wined3d_shader_resource_type resource_type
;
7704 resource_type
= pixelshader_get_resource_type(reg_maps
, i
, args
->tex_types
);
7706 if (!resource_type
|| !(args
->np2_fixup
& (1u << i
)))
7709 if (resource_type
!= WINED3D_SHADER_RESOURCE_TEXTURE_2D
)
7711 FIXME("Non-2D texture is flagged for NP2 texcoord fixup.\n");
7715 fixup
->idx
[i
] = cur
++;
7718 fixup
->num_consts
= (cur
+ 1) >> 1;
7719 fixup
->active
= args
->np2_fixup
;
7720 shader_addline(buffer
, "uniform vec4 %s_samplerNP2Fixup[%u];\n", prefix
, fixup
->num_consts
);
7723 if (version
->major
< 3 || args
->vp_mode
!= WINED3D_VP_MODE_SHADER
)
7725 shader_addline(buffer
, "uniform struct\n{\n");
7726 shader_addline(buffer
, " vec4 color;\n");
7727 shader_addline(buffer
, " float density;\n");
7728 shader_addline(buffer
, " float end;\n");
7729 shader_addline(buffer
, " float scale;\n");
7730 shader_addline(buffer
, "} ffp_fog;\n");
7734 if (glsl_is_color_reg_read(shader
, 0))
7735 shader_addline(buffer
, "vec4 ffp_varying_diffuse;\n");
7736 if (glsl_is_color_reg_read(shader
, 1))
7737 shader_addline(buffer
, "vec4 ffp_varying_specular;\n");
7738 shader_addline(buffer
, "vec4 ffp_texcoord[%u];\n", WINED3D_MAX_TEXTURES
);
7739 shader_addline(buffer
, "float ffp_varying_fogcoord;\n");
7743 if (glsl_is_color_reg_read(shader
, 0))
7744 declare_in_varying(gl_info
, buffer
, args
->flatshading
, "vec4 ffp_varying_diffuse;\n");
7745 if (glsl_is_color_reg_read(shader
, 1))
7746 declare_in_varying(gl_info
, buffer
, args
->flatshading
, "vec4 ffp_varying_specular;\n");
7747 declare_in_varying(gl_info
, buffer
, FALSE
, "vec4 ffp_varying_texcoord[%u];\n", WINED3D_MAX_TEXTURES
);
7748 shader_addline(buffer
, "vec4 ffp_texcoord[%u];\n", WINED3D_MAX_TEXTURES
);
7749 declare_in_varying(gl_info
, buffer
, FALSE
, "float ffp_varying_fogcoord;\n");
7753 if (version
->major
>= 3)
7755 unsigned int in_count
= min(vec4_varyings(version
->major
, gl_info
), shader
->limits
->packed_input
);
7757 if (args
->vp_mode
== WINED3D_VP_MODE_SHADER
&& reg_maps
->input_registers
)
7758 shader_glsl_declare_shader_inputs(gl_info
, buffer
, in_count
,
7759 shader
->u
.ps
.interpolation_mode
, version
->major
>= 4);
7760 shader_addline(buffer
, "vec4 %s_in[%u];\n", prefix
, in_count
);
7763 map
= reg_maps
->bumpmat
;
7766 i
= wined3d_bit_scan(&map
);
7767 shader_addline(buffer
, "uniform mat2 bumpenv_mat%u;\n", i
);
7769 if (reg_maps
->luminanceparams
& (1u << i
))
7771 shader_addline(buffer
, "uniform float bumpenv_lum_scale%u;\n", i
);
7772 shader_addline(buffer
, "uniform float bumpenv_lum_offset%u;\n", i
);
7773 extra_constants_needed
++;
7776 extra_constants_needed
++;
7779 if (args
->srgb_correction
)
7781 shader_addline(buffer
, "const vec4 srgb_const0 = ");
7782 shader_glsl_append_imm_vec(buffer
, &wined3d_srgb_const
[0].x
, 4, gl_info
);
7783 shader_addline(buffer
, ";\n");
7784 shader_addline(buffer
, "const vec4 srgb_const1 = ");
7785 shader_glsl_append_imm_vec(buffer
, &wined3d_srgb_const
[1].x
, 4, gl_info
);
7786 shader_addline(buffer
, ";\n");
7788 if ((reg_maps
->usesdsy
&& wined3d_settings
.offscreen_rendering_mode
!= ORM_FBO
)
7789 || (reg_maps
->vpos
&& !gl_info
->supported
[ARB_FRAGMENT_COORD_CONVENTIONS
]))
7791 ++extra_constants_needed
;
7792 shader_addline(buffer
, "uniform vec4 ycorrection;\n");
7796 if (gl_info
->supported
[ARB_FRAGMENT_COORD_CONVENTIONS
])
7798 if (context_gl
->c
.d3d_info
->wined3d_creation_flags
& WINED3D_PIXEL_CENTER_INTEGER
)
7799 shader_addline(buffer
, "layout(%spixel_center_integer) in vec4 gl_FragCoord;\n",
7800 args
->y_correction
? "origin_upper_left, " : "");
7801 else if (args
->y_correction
)
7802 shader_addline(buffer
, "layout(origin_upper_left) in vec4 gl_FragCoord;\n");
7804 shader_addline(buffer
, "vec4 vpos;\n");
7807 if (args
->alpha_test_func
+ 1 != WINED3D_CMP_ALWAYS
)
7808 shader_addline(buffer
, "uniform float alpha_test_ref;\n");
7810 if (!use_legacy_fragment_output(gl_info
))
7812 const struct wined3d_shader_signature
*output_signature
= &shader
->output_signature
;
7814 if (args
->dual_source_blend
)
7815 shader_addline(buffer
, "vec4 ps_out[2];\n");
7817 shader_addline(buffer
, "vec4 ps_out[%u];\n", gl_info
->limits
.buffers
);
7818 if (output_signature
->element_count
)
7820 for (i
= 0; i
< output_signature
->element_count
; ++i
)
7822 const struct wined3d_shader_signature_element
*output
= &output_signature
->elements
[i
];
7824 if (output
->register_idx
== ~0u)
7826 if ((unsigned int)output
->component_type
>= ARRAY_SIZE(component_type_info
))
7828 FIXME("Unhandled component type %#x.\n", output
->component_type
);
7831 if (shader_glsl_use_explicit_attrib_location(gl_info
))
7833 if (args
->dual_source_blend
)
7834 shader_addline(buffer
, "layout(location = 0, index = %u) ", output
->semantic_idx
);
7836 shader_addline(buffer
, "layout(location = %u) ", output
->semantic_idx
);
7838 shader_addline(buffer
, "out %s4 color_out%u;\n",
7839 component_type_info
[output
->component_type
].glsl_vector_type
, output
->semantic_idx
);
7844 DWORD mask
= reg_maps
->rt_mask
;
7848 i
= wined3d_bit_scan(&mask
);
7849 if (shader_glsl_use_explicit_attrib_location(gl_info
))
7851 if (args
->dual_source_blend
)
7852 shader_addline(buffer
, "layout(location = 0, index = %u) ", i
);
7854 shader_addline(buffer
, "layout(location = %u) ", i
);
7856 shader_addline(buffer
, "out vec4 color_out%u;\n", i
);
7861 if (shader
->limits
->constant_float
+ extra_constants_needed
>= gl_info
->limits
.glsl_ps_float_constants
)
7862 FIXME("Insufficient uniforms to run this shader.\n");
7864 if (shader
->u
.ps
.force_early_depth_stencil
)
7865 shader_addline(buffer
, "layout(early_fragment_tests) in;\n");
7867 shader_addline(buffer
, "void main()\n{\n");
7869 if (reg_maps
->sample_mask
)
7870 shader_addline(buffer
, "float sample_mask = uintBitsToFloat(0xffffffffu);\n");
7872 /* Direct3D applications expect integer vPos values, while OpenGL drivers
7873 * add approximately 0.5. This causes off-by-one problems as spotted by
7874 * the vPos d3d9 visual test. Unfortunately ATI cards do not add exactly
7875 * 0.5, but rather something like 0.49999999 or 0.50000001, which still
7876 * causes precision troubles when we just subtract 0.5.
7878 * To deal with that, just floor() the position. This will eliminate the
7879 * fraction on all cards.
7881 * TODO: Test how this behaves with multisampling.
7883 * An advantage of floor is that it works even if the driver doesn't add
7884 * 0.5. It is somewhat questionable if 1.5, 2.5, ... are the proper values
7885 * to return in gl_FragCoord, even though coordinates specify the pixel
7886 * centers instead of the pixel corners. This code will behave correctly
7887 * on drivers that returns integer values. */
7890 if (gl_info
->supported
[ARB_FRAGMENT_COORD_CONVENTIONS
])
7891 shader_addline(buffer
, "vpos = gl_FragCoord;\n");
7892 else if (context_gl
->c
.d3d_info
->wined3d_creation_flags
& WINED3D_PIXEL_CENTER_INTEGER
)
7893 shader_addline(buffer
,
7894 "vpos = floor(vec4(0, ycorrection[0], 0, 0) + gl_FragCoord * vec4(1, ycorrection[1], 1, 1));\n");
7896 shader_addline(buffer
,
7897 "vpos = vec4(0, ycorrection[0], 0, 0) + gl_FragCoord * vec4(1, ycorrection[1], 1, 1);\n");
7900 if (reg_maps
->shader_version
.major
< 3 || args
->vp_mode
!= WINED3D_VP_MODE_SHADER
)
7906 if (glsl_is_color_reg_read(shader
, 0))
7907 shader_addline(buffer
, "ffp_varying_diffuse = gl_Color;\n");
7908 if (glsl_is_color_reg_read(shader
, 1))
7909 shader_addline(buffer
, "ffp_varying_specular = gl_SecondaryColor;\n");
7912 map
= reg_maps
->texcoord
;
7915 i
= wined3d_bit_scan(&map
);
7916 if (args
->pointsprite
)
7917 shader_addline(buffer
, "ffp_texcoord[%u] = vec4(gl_PointCoord.xy, 0.0, 0.0);\n", i
);
7918 else if (args
->texcoords_initialized
& (1u << i
))
7919 shader_addline(buffer
, "ffp_texcoord[%u] = %s[%u];\n", i
,
7920 legacy_syntax
? "gl_TexCoord" : "ffp_varying_texcoord", i
);
7922 shader_addline(buffer
, "ffp_texcoord[%u] = vec4(0.0);\n", i
);
7923 shader_addline(buffer
, "vec4 T%u = ffp_texcoord[%u];\n", i
, i
);
7927 shader_addline(buffer
, "ffp_varying_fogcoord = gl_FogFragCoord;\n");
7930 /* Pack 3.0 inputs */
7931 if (reg_maps
->shader_version
.major
>= 3)
7932 shader_glsl_input_pack(shader
, buffer
, &shader
->input_signature
, reg_maps
, args
, gl_info
,
7933 reg_maps
->shader_version
.major
>= 4);
7935 /* Base Shader Body */
7936 if (FAILED(shader_generate_code(shader
, buffer
, reg_maps
, &priv_ctx
, NULL
, NULL
)))
7939 /* In SM4+ the shader epilogue is generated by the "ret" instruction. */
7940 if (reg_maps
->shader_version
.major
< 4)
7941 shader_glsl_generate_ps_epilogue(gl_info
, buffer
, shader
, args
, string_buffers
);
7943 shader_addline(buffer
, "}\n");
7945 shader_id
= GL_EXTCALL(glCreateShader(GL_FRAGMENT_SHADER
));
7946 TRACE("Compiling shader object %u.\n", shader_id
);
7947 shader_glsl_compile(gl_info
, shader_id
, buffer
->buffer
);
7952 static void shader_glsl_generate_vs_epilogue(const struct wined3d_gl_info
*gl_info
,
7953 struct wined3d_string_buffer
*buffer
, const struct wined3d_shader
*shader
,
7954 const struct vs_compile_args
*args
)
7956 const struct wined3d_shader_reg_maps
*reg_maps
= &shader
->reg_maps
;
7957 const BOOL legacy_syntax
= needs_legacy_glsl_syntax(gl_info
);
7960 /* Unpack outputs. */
7961 shader_addline(buffer
, "setup_vs_output(vs_out);\n");
7963 /* The D3DRS_FOGTABLEMODE render state defines if the shader-generated fog coord is used
7964 * or if the fragment depth is used. If the fragment depth is used(FOGTABLEMODE != NONE),
7965 * the fog frag coord is thrown away. If the fog frag coord is used, but not written by
7966 * the shader, it is set to 0.0(fully fogged, since start = 1.0, end = 0.0).
7968 if (reg_maps
->shader_version
.major
< 3)
7970 if (args
->fog_src
== VS_FOG_Z
)
7971 shader_addline(buffer
, "%s = gl_Position.z;\n",
7972 legacy_syntax
? "gl_FogFragCoord" : "ffp_varying_fogcoord");
7973 else if (!reg_maps
->fog
)
7974 shader_addline(buffer
, "%s = 0.0;\n",
7975 legacy_syntax
? "gl_FogFragCoord" : "ffp_varying_fogcoord");
7978 /* We always store the clipplanes without y inversion. */
7979 if (args
->clip_enabled
)
7982 shader_addline(buffer
, "gl_ClipVertex = gl_Position;\n");
7984 for (i
= 0; i
< gl_info
->limits
.user_clip_distances
; ++i
)
7985 shader_addline(buffer
, "gl_ClipDistance[%u] = dot(gl_Position, clip_planes[%u]);\n", i
, i
);
7988 if (args
->point_size
&& !args
->per_vertex_point_size
)
7989 shader_addline(buffer
, "gl_PointSize = clamp(ffp_point.size, ffp_point.size_min, ffp_point.size_max);\n");
7991 if (args
->next_shader_type
== WINED3D_SHADER_TYPE_PIXEL
&& !gl_info
->supported
[ARB_CLIP_CONTROL
])
7992 shader_glsl_fixup_position(buffer
, FALSE
);
7995 /* Context activation is done by the caller. */
7996 static GLuint
shader_glsl_generate_vertex_shader(const struct wined3d_context_gl
*context_gl
,
7997 struct shader_glsl_priv
*priv
, const struct wined3d_shader
*shader
, const struct vs_compile_args
*args
)
7999 struct wined3d_string_buffer_list
*string_buffers
= &priv
->string_buffers
;
8000 const struct wined3d_shader_reg_maps
*reg_maps
= &shader
->reg_maps
;
8001 const struct wined3d_shader_version
*version
= ®_maps
->shader_version
;
8002 const struct wined3d_gl_info
*gl_info
= context_gl
->gl_info
;
8003 struct wined3d_string_buffer
*buffer
= &priv
->shader_buffer
;
8004 struct shader_glsl_ctx_priv priv_ctx
;
8008 memset(&priv_ctx
, 0, sizeof(priv_ctx
));
8009 priv_ctx
.gl_info
= gl_info
;
8010 priv_ctx
.cur_vs_args
= args
;
8011 priv_ctx
.string_buffers
= string_buffers
;
8013 shader_glsl_add_version_declaration(buffer
, gl_info
);
8015 shader_glsl_enable_extensions(buffer
, gl_info
);
8016 if (gl_info
->supported
[ARB_DRAW_INSTANCED
])
8017 shader_addline(buffer
, "#extension GL_ARB_draw_instanced : enable\n");
8018 if (shader_glsl_use_explicit_attrib_location(gl_info
))
8019 shader_addline(buffer
, "#extension GL_ARB_explicit_attrib_location : enable\n");
8020 if (gl_info
->supported
[ARB_SHADER_VIEWPORT_LAYER_ARRAY
])
8021 shader_addline(buffer
, "#extension GL_ARB_shader_viewport_layer_array : enable\n");
8023 /* Base Declarations */
8024 shader_generate_glsl_declarations(context_gl
, buffer
, shader
, reg_maps
, &priv_ctx
);
8026 for (i
= 0; i
< shader
->input_signature
.element_count
; ++i
)
8027 shader_glsl_declare_generic_vertex_attribute(buffer
, gl_info
, &shader
->input_signature
.elements
[i
]);
8029 if (args
->point_size
&& !args
->per_vertex_point_size
)
8031 shader_addline(buffer
, "uniform struct\n{\n");
8032 shader_addline(buffer
, " float size;\n");
8033 shader_addline(buffer
, " float size_min;\n");
8034 shader_addline(buffer
, " float size_max;\n");
8035 shader_addline(buffer
, "} ffp_point;\n");
8038 if (!needs_legacy_glsl_syntax(gl_info
))
8040 if (args
->clip_enabled
)
8041 shader_addline(buffer
, "uniform vec4 clip_planes[%u];\n", gl_info
->limits
.user_clip_distances
);
8043 if (version
->major
< 3)
8045 declare_out_varying(gl_info
, buffer
, args
->flatshading
, "vec4 ffp_varying_diffuse;\n");
8046 declare_out_varying(gl_info
, buffer
, args
->flatshading
, "vec4 ffp_varying_specular;\n");
8047 declare_out_varying(gl_info
, buffer
, FALSE
, "vec4 ffp_varying_texcoord[%u];\n", WINED3D_MAX_TEXTURES
);
8048 declare_out_varying(gl_info
, buffer
, FALSE
, "float ffp_varying_fogcoord;\n");
8052 if (version
->major
< 4)
8053 shader_addline(buffer
, "void setup_vs_output(in vec4[%u]);\n", shader
->limits
->packed_output
);
8055 if (args
->next_shader_type
== WINED3D_SHADER_TYPE_PIXEL
&& !gl_info
->supported
[ARB_CLIP_CONTROL
])
8056 shader_addline(buffer
, "uniform vec4 pos_fixup;\n");
8058 if (reg_maps
->shader_version
.major
>= 4)
8059 shader_glsl_generate_sm4_output_setup(priv
, shader
, args
->next_shader_input_count
,
8060 gl_info
, args
->next_shader_type
== WINED3D_SHADER_TYPE_PIXEL
, args
->interpolation_mode
);
8062 shader_addline(buffer
, "void main()\n{\n");
8064 if (reg_maps
->input_rel_addressing
)
8066 unsigned int highest_input_register
= wined3d_log2i(reg_maps
->input_registers
);
8067 shader_addline(buffer
, "vec4 vs_in[%u];\n", highest_input_register
+ 1);
8068 for (i
= 0; i
< shader
->input_signature
.element_count
; ++i
)
8070 const struct wined3d_shader_signature_element
*e
= &shader
->input_signature
.elements
[i
];
8071 shader_addline(buffer
, "vs_in[%u] = vs_in%u;\n", e
->register_idx
, e
->register_idx
);
8075 if (FAILED(shader_generate_code(shader
, buffer
, reg_maps
, &priv_ctx
, NULL
, NULL
)))
8078 /* In SM4+ the shader epilogue is generated by the "ret" instruction. */
8079 if (reg_maps
->shader_version
.major
< 4)
8080 shader_glsl_generate_vs_epilogue(gl_info
, buffer
, shader
, args
);
8082 shader_addline(buffer
, "}\n");
8084 shader_id
= GL_EXTCALL(glCreateShader(GL_VERTEX_SHADER
));
8085 TRACE("Compiling shader object %u.\n", shader_id
);
8086 shader_glsl_compile(gl_info
, shader_id
, buffer
->buffer
);
8091 static void shader_glsl_generate_default_control_point_phase(const struct wined3d_shader
*shader
,
8092 struct wined3d_string_buffer
*buffer
, const struct wined3d_shader_reg_maps
*reg_maps
)
8094 const struct wined3d_shader_signature
*output_signature
= &shader
->output_signature
;
8098 for (i
= 0; i
< output_signature
->element_count
; ++i
)
8100 const struct wined3d_shader_signature_element
*output
= &output_signature
->elements
[i
];
8102 shader_glsl_write_mask_to_str(output
->mask
, reg_mask
);
8103 shader_addline(buffer
, "shader_out[gl_InvocationID].reg[%u]%s = shader_in[gl_InvocationID].reg[%u]%s;\n",
8104 output
->register_idx
, reg_mask
, output
->register_idx
, reg_mask
);
8108 static HRESULT
shader_glsl_generate_shader_phase(const struct wined3d_shader
*shader
,
8109 struct wined3d_string_buffer
*buffer
, const struct wined3d_shader_reg_maps
*reg_maps
,
8110 struct shader_glsl_ctx_priv
*priv_ctx
, const struct wined3d_shader_phase
*phase
,
8111 const char *phase_name
, unsigned phase_idx
)
8116 shader_addline(buffer
, "void hs_%s_phase%u(%s)\n{\n",
8117 phase_name
, phase_idx
, phase
->instance_count
? "int phase_instance_id" : "");
8118 for (i
= 0; i
< phase
->temporary_count
; ++i
)
8119 shader_addline(buffer
, "vec4 R%u;\n", i
);
8120 hr
= shader_generate_code(shader
, buffer
, reg_maps
, priv_ctx
, phase
->start
, phase
->end
);
8121 shader_addline(buffer
, "}\n");
8125 static void shader_glsl_generate_shader_phase_invocation(struct wined3d_string_buffer
*buffer
,
8126 const struct wined3d_shader_phase
*phase
, const char *phase_name
, unsigned int phase_idx
)
8128 if (phase
->instance_count
)
8130 shader_addline(buffer
, "for (int i = 0; i < %u; ++i)\n{\n", phase
->instance_count
);
8131 shader_addline(buffer
, "hs_%s_phase%u(i);\n", phase_name
, phase_idx
);
8132 shader_addline(buffer
, "}\n");
8136 shader_addline(buffer
, "hs_%s_phase%u();\n", phase_name
, phase_idx
);
8140 static GLuint
shader_glsl_generate_hull_shader(const struct wined3d_context_gl
*context_gl
,
8141 struct shader_glsl_priv
*priv
, const struct wined3d_shader
*shader
)
8143 struct wined3d_string_buffer_list
*string_buffers
= &priv
->string_buffers
;
8144 const struct wined3d_shader_reg_maps
*reg_maps
= &shader
->reg_maps
;
8145 const struct wined3d_gl_info
*gl_info
= context_gl
->gl_info
;
8146 struct wined3d_string_buffer
*buffer
= &priv
->shader_buffer
;
8147 const struct wined3d_hull_shader
*hs
= &shader
->u
.hs
;
8148 const struct wined3d_shader_phase
*phase
;
8149 struct shader_glsl_ctx_priv priv_ctx
;
8153 memset(&priv_ctx
, 0, sizeof(priv_ctx
));
8154 priv_ctx
.gl_info
= gl_info
;
8155 priv_ctx
.string_buffers
= string_buffers
;
8157 shader_glsl_add_version_declaration(buffer
, gl_info
);
8159 shader_glsl_enable_extensions(buffer
, gl_info
);
8160 shader_addline(buffer
, "#extension GL_ARB_tessellation_shader : enable\n");
8162 shader_generate_glsl_declarations(context_gl
, buffer
, shader
, reg_maps
, &priv_ctx
);
8164 shader_addline(buffer
, "layout(vertices = %u) out;\n", hs
->output_vertex_count
);
8166 shader_addline(buffer
, "in shader_in_out { vec4 reg[%u]; } shader_in[];\n", shader
->limits
->packed_input
);
8167 shader_addline(buffer
, "out shader_in_out { vec4 reg[%u]; } shader_out[];\n", shader
->limits
->packed_output
);
8169 shader_glsl_generate_patch_constant_setup(buffer
, &shader
->patch_constant_signature
, FALSE
);
8171 if (hs
->phases
.control_point
)
8173 shader_addline(buffer
, "void setup_hs_output(in vec4 outputs[%u])\n{\n",
8174 shader
->limits
->packed_output
);
8175 shader_glsl_setup_sm4_shader_output(priv
, shader
->limits
->packed_output
, &shader
->output_signature
,
8176 &shader
->reg_maps
, "shader_out[gl_InvocationID]", FALSE
);
8177 shader_addline(buffer
, "}\n");
8180 shader_addline(buffer
, "void hs_control_point_phase()\n{\n");
8181 if ((phase
= hs
->phases
.control_point
))
8183 for (i
= 0; i
< phase
->temporary_count
; ++i
)
8184 shader_addline(buffer
, "vec4 R%u;\n", i
);
8185 if (FAILED(shader_generate_code(shader
, buffer
, reg_maps
, &priv_ctx
, phase
->start
, phase
->end
)))
8187 shader_addline(buffer
, "setup_hs_output(hs_out);\n");
8191 shader_glsl_generate_default_control_point_phase(shader
, buffer
, reg_maps
);
8193 shader_addline(buffer
, "}\n");
8195 for (i
= 0; i
< hs
->phases
.fork_count
; ++i
)
8197 if (FAILED(shader_glsl_generate_shader_phase(shader
, buffer
, reg_maps
, &priv_ctx
,
8198 &hs
->phases
.fork
[i
], "fork", i
)))
8202 for (i
= 0; i
< hs
->phases
.join_count
; ++i
)
8204 if (FAILED(shader_glsl_generate_shader_phase(shader
, buffer
, reg_maps
, &priv_ctx
,
8205 &hs
->phases
.join
[i
], "join", i
)))
8209 shader_addline(buffer
, "void main()\n{\n");
8210 shader_addline(buffer
, "hs_control_point_phase();\n");
8212 shader_addline(buffer
, "barrier();\n");
8213 for (i
= 0; i
< hs
->phases
.fork_count
; ++i
)
8214 shader_glsl_generate_shader_phase_invocation(buffer
, &hs
->phases
.fork
[i
], "fork", i
);
8215 for (i
= 0; i
< hs
->phases
.join_count
; ++i
)
8216 shader_glsl_generate_shader_phase_invocation(buffer
, &hs
->phases
.join
[i
], "join", i
);
8217 shader_addline(buffer
, "setup_patch_constant_output();\n");
8218 shader_addline(buffer
, "}\n");
8220 shader_id
= GL_EXTCALL(glCreateShader(GL_TESS_CONTROL_SHADER
));
8221 TRACE("Compiling shader object %u.\n", shader_id
);
8222 shader_glsl_compile(gl_info
, shader_id
, buffer
->buffer
);
8227 static void shader_glsl_generate_ds_epilogue(const struct wined3d_gl_info
*gl_info
,
8228 struct wined3d_string_buffer
*buffer
, const struct wined3d_shader
*shader
,
8229 const struct ds_compile_args
*args
)
8231 shader_addline(buffer
, "setup_ds_output(ds_out);\n");
8233 if (args
->next_shader_type
== WINED3D_SHADER_TYPE_PIXEL
&& !gl_info
->supported
[ARB_CLIP_CONTROL
])
8234 shader_glsl_fixup_position(buffer
, FALSE
);
8237 static GLuint
shader_glsl_generate_domain_shader(const struct wined3d_context_gl
*context_gl
,
8238 struct shader_glsl_priv
*priv
, const struct wined3d_shader
*shader
, const struct ds_compile_args
*args
)
8240 struct wined3d_string_buffer_list
*string_buffers
= &priv
->string_buffers
;
8241 const struct wined3d_shader_reg_maps
*reg_maps
= &shader
->reg_maps
;
8242 const struct wined3d_gl_info
*gl_info
= context_gl
->gl_info
;
8243 struct wined3d_string_buffer
*buffer
= &priv
->shader_buffer
;
8244 struct shader_glsl_ctx_priv priv_ctx
;
8247 memset(&priv_ctx
, 0, sizeof(priv_ctx
));
8248 priv_ctx
.gl_info
= gl_info
;
8249 priv_ctx
.cur_ds_args
= args
;
8250 priv_ctx
.string_buffers
= string_buffers
;
8252 shader_glsl_add_version_declaration(buffer
, gl_info
);
8254 shader_glsl_enable_extensions(buffer
, gl_info
);
8255 shader_addline(buffer
, "#extension GL_ARB_tessellation_shader : enable\n");
8257 shader_generate_glsl_declarations(context_gl
, buffer
, shader
, reg_maps
, &priv_ctx
);
8259 shader_addline(buffer
, "layout(");
8260 switch (shader
->u
.ds
.tessellator_domain
)
8262 case WINED3D_TESSELLATOR_DOMAIN_LINE
:
8263 shader_addline(buffer
, "isolines");
8265 case WINED3D_TESSELLATOR_DOMAIN_QUAD
:
8266 shader_addline(buffer
, "quads");
8268 case WINED3D_TESSELLATOR_DOMAIN_TRIANGLE
:
8269 shader_addline(buffer
, "triangles");
8272 switch (args
->tessellator_output_primitive
)
8274 case WINED3D_TESSELLATOR_OUTPUT_TRIANGLE_CW
:
8275 if (args
->render_offscreen
)
8276 shader_addline(buffer
, ", ccw");
8278 shader_addline(buffer
, ", cw");
8280 case WINED3D_TESSELLATOR_OUTPUT_TRIANGLE_CCW
:
8281 if (args
->render_offscreen
)
8282 shader_addline(buffer
, ", cw");
8284 shader_addline(buffer
, ", ccw");
8286 case WINED3D_TESSELLATOR_OUTPUT_POINT
:
8287 shader_addline(buffer
, ", point_mode");
8289 case WINED3D_TESSELLATOR_OUTPUT_LINE
:
8292 switch (args
->tessellator_partitioning
)
8294 case WINED3D_TESSELLATOR_PARTITIONING_FRACTIONAL_ODD
:
8295 shader_addline(buffer
, ", fractional_odd_spacing");
8297 case WINED3D_TESSELLATOR_PARTITIONING_FRACTIONAL_EVEN
:
8298 shader_addline(buffer
, ", fractional_even_spacing");
8300 case WINED3D_TESSELLATOR_PARTITIONING_INTEGER
:
8301 case WINED3D_TESSELLATOR_PARTITIONING_POW2
:
8302 shader_addline(buffer
, ", equal_spacing");
8305 shader_addline(buffer
, ") in;\n");
8307 shader_addline(buffer
, "in shader_in_out { vec4 reg[%u]; } shader_in[];\n", shader
->limits
->packed_input
);
8309 if (args
->next_shader_type
== WINED3D_SHADER_TYPE_PIXEL
&& !gl_info
->supported
[ARB_CLIP_CONTROL
])
8310 shader_addline(buffer
, "uniform vec4 pos_fixup;\n");
8312 shader_glsl_generate_sm4_output_setup(priv
, shader
, args
->output_count
, gl_info
,
8313 args
->next_shader_type
== WINED3D_SHADER_TYPE_PIXEL
, args
->interpolation_mode
);
8314 shader_glsl_generate_patch_constant_setup(buffer
, &shader
->patch_constant_signature
, TRUE
);
8316 shader_addline(buffer
, "void main()\n{\n");
8317 shader_addline(buffer
, "setup_patch_constant_input();\n");
8319 if (FAILED(shader_generate_code(shader
, buffer
, reg_maps
, &priv_ctx
, NULL
, NULL
)))
8322 shader_addline(buffer
, "}\n");
8324 shader_id
= GL_EXTCALL(glCreateShader(GL_TESS_EVALUATION_SHADER
));
8325 TRACE("Compiling shader object %u.\n", shader_id
);
8326 shader_glsl_compile(gl_info
, shader_id
, buffer
->buffer
);
8331 /* Context activation is done by the caller. */
8332 static GLuint
shader_glsl_generate_geometry_shader(const struct wined3d_context_gl
*context_gl
,
8333 struct shader_glsl_priv
*priv
, const struct wined3d_shader
*shader
, const struct gs_compile_args
*args
)
8335 struct wined3d_string_buffer_list
*string_buffers
= &priv
->string_buffers
;
8336 const struct wined3d_shader_reg_maps
*reg_maps
= &shader
->reg_maps
;
8337 const struct wined3d_gl_info
*gl_info
= context_gl
->gl_info
;
8338 struct wined3d_string_buffer
*buffer
= &priv
->shader_buffer
;
8339 const struct wined3d_shader_signature_element
*output
;
8340 enum wined3d_primitive_type primitive_type
;
8341 struct shader_glsl_ctx_priv priv_ctx
;
8342 unsigned int max_vertices
;
8346 memset(&priv_ctx
, 0, sizeof(priv_ctx
));
8347 priv_ctx
.gl_info
= gl_info
;
8348 priv_ctx
.string_buffers
= string_buffers
;
8350 shader_glsl_add_version_declaration(buffer
, gl_info
);
8352 shader_glsl_enable_extensions(buffer
, gl_info
);
8354 shader_generate_glsl_declarations(context_gl
, buffer
, shader
, reg_maps
, &priv_ctx
);
8356 primitive_type
= shader
->u
.gs
.input_type
? shader
->u
.gs
.input_type
: args
->primitive_type
;
8357 shader_addline(buffer
, "layout(%s", glsl_primitive_type_from_d3d(primitive_type
));
8358 if (shader
->u
.gs
.instance_count
> 1)
8359 shader_addline(buffer
, ", invocations = %u", shader
->u
.gs
.instance_count
);
8360 shader_addline(buffer
, ") in;\n");
8362 primitive_type
= shader
->u
.gs
.output_type
? shader
->u
.gs
.output_type
: args
->primitive_type
;
8363 if (!(max_vertices
= shader
->u
.gs
.vertices_out
))
8365 switch (args
->primitive_type
)
8367 case WINED3D_PT_POINTLIST
:
8370 case WINED3D_PT_LINELIST
:
8373 case WINED3D_PT_TRIANGLELIST
:
8377 FIXME("Unhandled primitive type %s.\n", debug_d3dprimitivetype(args
->primitive_type
));
8381 shader_addline(buffer
, "layout(%s, max_vertices = %u) out;\n",
8382 glsl_primitive_type_from_d3d(primitive_type
), max_vertices
);
8383 shader_addline(buffer
, "in shader_in_out { vec4 reg[%u]; } shader_in[];\n", shader
->limits
->packed_input
);
8385 if (!gl_info
->supported
[ARB_CLIP_CONTROL
])
8387 shader_addline(buffer
, "uniform vec4 pos_fixup");
8388 if (reg_maps
->viewport_array
)
8389 shader_addline(buffer
, "[%u]", WINED3D_MAX_VIEWPORTS
);
8390 shader_addline(buffer
, ";\n");
8393 if (is_rasterization_disabled(shader
))
8395 shader_glsl_generate_stream_output_setup(buffer
, shader
);
8399 shader_glsl_generate_sm4_output_setup(priv
, shader
, args
->output_count
,
8400 gl_info
, TRUE
, args
->interpolation_mode
);
8403 shader_addline(buffer
, "void main()\n{\n");
8404 if (shader
->function
)
8406 if (FAILED(shader_generate_code(shader
, buffer
, reg_maps
, &priv_ctx
, NULL
, NULL
)))
8411 for (i
= 0; i
< max_vertices
; ++i
)
8413 for (j
= 0; j
< shader
->output_signature
.element_count
; ++j
)
8415 output
= &shader
->output_signature
.elements
[j
];
8416 shader_addline(buffer
, "gs_out[%u] = shader_in[%u].reg[%u];\n",
8417 output
->register_idx
, i
, output
->register_idx
);
8419 shader_addline(buffer
, "setup_gs_output(gs_out);\n");
8420 if (!gl_info
->supported
[ARB_CLIP_CONTROL
])
8421 shader_glsl_fixup_position(buffer
, FALSE
);
8422 shader_addline(buffer
, "EmitVertex();\n");
8425 shader_addline(buffer
, "}\n");
8427 shader_id
= GL_EXTCALL(glCreateShader(GL_GEOMETRY_SHADER
));
8428 TRACE("Compiling shader object %u.\n", shader_id
);
8429 shader_glsl_compile(gl_info
, shader_id
, buffer
->buffer
);
8434 static void shader_glsl_generate_shader_epilogue(const struct wined3d_shader_context
*ctx
)
8436 const struct shader_glsl_ctx_priv
*priv
= ctx
->backend_data
;
8437 const struct wined3d_gl_info
*gl_info
= priv
->gl_info
;
8438 struct wined3d_string_buffer
*buffer
= ctx
->buffer
;
8439 const struct wined3d_shader
*shader
= ctx
->shader
;
8441 switch (shader
->reg_maps
.shader_version
.type
)
8443 case WINED3D_SHADER_TYPE_PIXEL
:
8444 shader_glsl_generate_ps_epilogue(gl_info
, buffer
, shader
, priv
->cur_ps_args
, priv
->string_buffers
);
8446 case WINED3D_SHADER_TYPE_VERTEX
:
8447 shader_glsl_generate_vs_epilogue(gl_info
, buffer
, shader
, priv
->cur_vs_args
);
8449 case WINED3D_SHADER_TYPE_DOMAIN
:
8450 shader_glsl_generate_ds_epilogue(gl_info
, buffer
, shader
, priv
->cur_ds_args
);
8452 case WINED3D_SHADER_TYPE_GEOMETRY
:
8453 case WINED3D_SHADER_TYPE_COMPUTE
:
8456 FIXME("Unhandled shader type %#x.\n", shader
->reg_maps
.shader_version
.type
);
8461 /* Context activation is done by the caller. */
8462 static GLuint
shader_glsl_generate_compute_shader(const struct wined3d_context_gl
*context_gl
,
8463 struct wined3d_string_buffer
*buffer
, struct wined3d_string_buffer_list
*string_buffers
,
8464 const struct wined3d_shader
*shader
)
8466 const struct wined3d_shader_thread_group_size
*thread_group_size
= &shader
->u
.cs
.thread_group_size
;
8467 const struct wined3d_shader_reg_maps
*reg_maps
= &shader
->reg_maps
;
8468 const struct wined3d_gl_info
*gl_info
= context_gl
->gl_info
;
8469 struct shader_glsl_ctx_priv priv_ctx
;
8473 memset(&priv_ctx
, 0, sizeof(priv_ctx
));
8474 priv_ctx
.gl_info
= gl_info
;
8475 priv_ctx
.string_buffers
= string_buffers
;
8477 shader_glsl_add_version_declaration(buffer
, gl_info
);
8479 shader_glsl_enable_extensions(buffer
, gl_info
);
8480 shader_addline(buffer
, "#extension GL_ARB_compute_shader : enable\n");
8482 shader_generate_glsl_declarations(context_gl
, buffer
, shader
, reg_maps
, &priv_ctx
);
8484 for (i
= 0; i
< reg_maps
->tgsm_count
; ++i
)
8486 if (reg_maps
->tgsm
[i
].size
)
8487 shader_addline(buffer
, "shared uint cs_g%u[%u];\n", i
, reg_maps
->tgsm
[i
].size
);
8490 shader_addline(buffer
, "layout(local_size_x = %u, local_size_y = %u, local_size_z = %u) in;\n",
8491 thread_group_size
->x
, thread_group_size
->y
, thread_group_size
->z
);
8493 shader_addline(buffer
, "void main()\n{\n");
8494 shader_generate_code(shader
, buffer
, reg_maps
, &priv_ctx
, NULL
, NULL
);
8495 shader_addline(buffer
, "}\n");
8497 shader_id
= GL_EXTCALL(glCreateShader(GL_COMPUTE_SHADER
));
8498 TRACE("Compiling shader object %u.\n", shader_id
);
8499 shader_glsl_compile(gl_info
, shader_id
, buffer
->buffer
);
8504 static GLuint
find_glsl_fragment_shader(const struct wined3d_context_gl
*context_gl
,
8505 struct wined3d_string_buffer
*buffer
, struct wined3d_string_buffer_list
*string_buffers
,
8506 struct wined3d_shader
*shader
,
8507 const struct ps_compile_args
*args
, const struct ps_np2fixup_info
**np2fixup_info
)
8509 struct glsl_ps_compiled_shader
*gl_shaders
, *new_array
;
8510 struct glsl_shader_private
*shader_data
;
8511 struct ps_np2fixup_info
*np2fixup
;
8516 if (!shader
->backend_data
)
8518 if (!(shader
->backend_data
= heap_alloc_zero(sizeof(*shader_data
))))
8520 ERR("Failed to allocate backend data.\n");
8524 shader_data
= shader
->backend_data
;
8525 gl_shaders
= shader_data
->gl_shaders
.ps
;
8527 /* Usually we have very few GL shaders for each d3d shader(just 1 or maybe 2),
8528 * so a linear search is more performant than a hashmap or a binary search
8529 * (cache coherency etc)
8531 for (i
= 0; i
< shader_data
->num_gl_shaders
; ++i
)
8533 if (!memcmp(&gl_shaders
[i
].args
, args
, sizeof(*args
)))
8535 if (args
->np2_fixup
)
8536 *np2fixup_info
= &gl_shaders
[i
].np2fixup
;
8537 return gl_shaders
[i
].id
;
8541 TRACE("No matching GL shader found for shader %p, compiling a new shader.\n", shader
);
8542 if (shader_data
->shader_array_size
== shader_data
->num_gl_shaders
)
8544 if (shader_data
->num_gl_shaders
)
8546 new_size
= shader_data
->shader_array_size
+ max(1, shader_data
->shader_array_size
/ 2);
8547 new_array
= heap_realloc(shader_data
->gl_shaders
.ps
, new_size
* sizeof(*gl_shaders
));
8551 new_array
= heap_alloc(sizeof(*gl_shaders
));
8556 ERR("Out of memory\n");
8559 shader_data
->gl_shaders
.ps
= new_array
;
8560 shader_data
->shader_array_size
= new_size
;
8561 gl_shaders
= new_array
;
8564 gl_shaders
[shader_data
->num_gl_shaders
].args
= *args
;
8566 np2fixup
= &gl_shaders
[shader_data
->num_gl_shaders
].np2fixup
;
8567 memset(np2fixup
, 0, sizeof(*np2fixup
));
8568 *np2fixup_info
= args
->np2_fixup
? np2fixup
: NULL
;
8570 string_buffer_clear(buffer
);
8571 ret
= shader_glsl_generate_fragment_shader(context_gl
, buffer
, string_buffers
, shader
, args
, np2fixup
);
8572 gl_shaders
[shader_data
->num_gl_shaders
++].id
= ret
;
8577 static BOOL
vs_args_equal(const struct vs_compile_args
*stored
, const struct vs_compile_args
*new,
8578 const DWORD use_map
)
8580 if ((stored
->swizzle_map
& use_map
) != new->swizzle_map
)
8582 if ((stored
->clip_enabled
) != new->clip_enabled
)
8584 if (stored
->point_size
!= new->point_size
)
8586 if (stored
->per_vertex_point_size
!= new->per_vertex_point_size
)
8588 if (stored
->flatshading
!= new->flatshading
)
8590 if (stored
->next_shader_type
!= new->next_shader_type
)
8592 if (stored
->next_shader_input_count
!= new->next_shader_input_count
)
8594 if (stored
->fog_src
!= new->fog_src
)
8596 return !memcmp(stored
->interpolation_mode
, new->interpolation_mode
, sizeof(new->interpolation_mode
));
8599 static GLuint
find_glsl_vertex_shader(const struct wined3d_context_gl
*context_gl
,
8600 struct shader_glsl_priv
*priv
, struct wined3d_shader
*shader
, const struct vs_compile_args
*args
)
8602 struct glsl_vs_compiled_shader
*gl_shaders
, *new_array
;
8603 uint32_t use_map
= context_gl
->c
.stream_info
.use_map
;
8604 struct glsl_shader_private
*shader_data
;
8605 unsigned int i
, new_size
;
8608 if (!shader
->backend_data
)
8610 if (!(shader
->backend_data
= heap_alloc_zero(sizeof(*shader_data
))))
8612 ERR("Failed to allocate backend data.\n");
8616 shader_data
= shader
->backend_data
;
8617 gl_shaders
= shader_data
->gl_shaders
.vs
;
8619 /* Usually we have very few GL shaders for each d3d shader(just 1 or maybe 2),
8620 * so a linear search is more performant than a hashmap or a binary search
8621 * (cache coherency etc)
8623 for (i
= 0; i
< shader_data
->num_gl_shaders
; ++i
)
8625 if (vs_args_equal(&gl_shaders
[i
].args
, args
, use_map
))
8626 return gl_shaders
[i
].id
;
8629 TRACE("No matching GL shader found for shader %p, compiling a new shader.\n", shader
);
8631 if (shader_data
->shader_array_size
== shader_data
->num_gl_shaders
)
8633 if (shader_data
->num_gl_shaders
)
8635 new_size
= shader_data
->shader_array_size
+ max(1, shader_data
->shader_array_size
/ 2);
8636 new_array
= heap_realloc(shader_data
->gl_shaders
.vs
, new_size
* sizeof(*gl_shaders
));
8640 new_array
= heap_alloc(sizeof(*gl_shaders
));
8645 ERR("Out of memory\n");
8648 shader_data
->gl_shaders
.vs
= new_array
;
8649 shader_data
->shader_array_size
= new_size
;
8650 gl_shaders
= new_array
;
8653 gl_shaders
[shader_data
->num_gl_shaders
].args
= *args
;
8655 string_buffer_clear(&priv
->shader_buffer
);
8656 ret
= shader_glsl_generate_vertex_shader(context_gl
, priv
, shader
, args
);
8657 gl_shaders
[shader_data
->num_gl_shaders
++].id
= ret
;
8662 static GLuint
find_glsl_hull_shader(const struct wined3d_context_gl
*context_gl
,
8663 struct shader_glsl_priv
*priv
, struct wined3d_shader
*shader
)
8665 struct glsl_hs_compiled_shader
*gl_shaders
, *new_array
;
8666 struct glsl_shader_private
*shader_data
;
8667 unsigned int new_size
;
8670 if (!shader
->backend_data
)
8672 if (!(shader
->backend_data
= heap_alloc_zero(sizeof(*shader_data
))))
8674 ERR("Failed to allocate backend data.\n");
8678 shader_data
= shader
->backend_data
;
8679 gl_shaders
= shader_data
->gl_shaders
.hs
;
8681 if (shader_data
->num_gl_shaders
> 0)
8683 assert(shader_data
->num_gl_shaders
== 1);
8684 return gl_shaders
[0].id
;
8687 TRACE("No matching GL shader found for shader %p, compiling a new shader.\n", shader
);
8689 assert(!shader_data
->gl_shaders
.hs
);
8691 if (!(new_array
= heap_alloc(sizeof(*new_array
))))
8693 ERR("Failed to allocate GL shaders array.\n");
8696 shader_data
->gl_shaders
.hs
= new_array
;
8697 shader_data
->shader_array_size
= new_size
;
8698 gl_shaders
= new_array
;
8700 string_buffer_clear(&priv
->shader_buffer
);
8701 ret
= shader_glsl_generate_hull_shader(context_gl
, priv
, shader
);
8702 gl_shaders
[shader_data
->num_gl_shaders
++].id
= ret
;
8707 static GLuint
find_glsl_domain_shader(const struct wined3d_context_gl
*context_gl
,
8708 struct shader_glsl_priv
*priv
, struct wined3d_shader
*shader
, const struct ds_compile_args
*args
)
8710 struct glsl_ds_compiled_shader
*gl_shaders
, *new_array
;
8711 struct glsl_shader_private
*shader_data
;
8712 unsigned int i
, new_size
;
8715 if (!shader
->backend_data
)
8717 if (!(shader
->backend_data
= heap_alloc_zero(sizeof(*shader_data
))))
8719 ERR("Failed to allocate backend data.\n");
8723 shader_data
= shader
->backend_data
;
8724 gl_shaders
= shader_data
->gl_shaders
.ds
;
8726 for (i
= 0; i
< shader_data
->num_gl_shaders
; ++i
)
8728 if (!memcmp(&gl_shaders
[i
].args
, args
, sizeof(*args
)))
8729 return gl_shaders
[i
].id
;
8732 TRACE("No matching GL shader found for shader %p, compiling a new shader.\n", shader
);
8734 if (shader_data
->num_gl_shaders
)
8736 new_size
= shader_data
->shader_array_size
+ 1;
8737 new_array
= heap_realloc(shader_data
->gl_shaders
.ds
, new_size
* sizeof(*new_array
));
8741 new_array
= heap_alloc(sizeof(*new_array
));
8747 ERR("Failed to allocate GL shaders array.\n");
8750 shader_data
->gl_shaders
.ds
= new_array
;
8751 shader_data
->shader_array_size
= new_size
;
8752 gl_shaders
= new_array
;
8754 string_buffer_clear(&priv
->shader_buffer
);
8755 ret
= shader_glsl_generate_domain_shader(context_gl
, priv
, shader
, args
);
8756 gl_shaders
[shader_data
->num_gl_shaders
].args
= *args
;
8757 gl_shaders
[shader_data
->num_gl_shaders
++].id
= ret
;
8762 static GLuint
find_glsl_geometry_shader(const struct wined3d_context_gl
*context_gl
,
8763 struct shader_glsl_priv
*priv
, struct wined3d_shader
*shader
, const struct gs_compile_args
*args
)
8765 struct glsl_gs_compiled_shader
*gl_shaders
, *new_array
;
8766 struct glsl_shader_private
*shader_data
;
8767 unsigned int i
, new_size
;
8770 if (!shader
->backend_data
)
8772 if (!(shader
->backend_data
= heap_alloc_zero(sizeof(*shader_data
))))
8774 ERR("Failed to allocate backend data.\n");
8778 shader_data
= shader
->backend_data
;
8779 gl_shaders
= shader_data
->gl_shaders
.gs
;
8781 for (i
= 0; i
< shader_data
->num_gl_shaders
; ++i
)
8783 if (!memcmp(&gl_shaders
[i
].args
, args
, sizeof(*args
)))
8784 return gl_shaders
[i
].id
;
8787 TRACE("No matching GL shader found for shader %p, compiling a new shader.\n", shader
);
8789 if (shader_data
->num_gl_shaders
)
8791 new_size
= shader_data
->shader_array_size
+ 1;
8792 new_array
= heap_realloc(shader_data
->gl_shaders
.gs
, new_size
* sizeof(*new_array
));
8796 new_array
= heap_alloc(sizeof(*new_array
));
8802 ERR("Failed to allocate GL shaders array.\n");
8805 shader_data
->gl_shaders
.gs
= new_array
;
8806 shader_data
->shader_array_size
= new_size
;
8807 gl_shaders
= new_array
;
8809 string_buffer_clear(&priv
->shader_buffer
);
8810 ret
= shader_glsl_generate_geometry_shader(context_gl
, priv
, shader
, args
);
8811 gl_shaders
[shader_data
->num_gl_shaders
].args
= *args
;
8812 gl_shaders
[shader_data
->num_gl_shaders
++].id
= ret
;
8817 static const char *shader_glsl_ffp_mcs(enum wined3d_material_color_source mcs
, const char *material
)
8821 case WINED3D_MCS_MATERIAL
:
8823 case WINED3D_MCS_COLOR1
:
8824 return "ffp_attrib_diffuse";
8825 case WINED3D_MCS_COLOR2
:
8826 return "ffp_attrib_specular";
8828 ERR("Invalid material color source %#x.\n", mcs
);
8833 static void shader_glsl_ffp_vertex_lighting_footer(struct wined3d_string_buffer
*buffer
,
8834 const struct wined3d_ffp_vs_settings
*settings
, unsigned int idx
, BOOL legacy_lighting
)
8836 shader_addline(buffer
, "diffuse += clamp(dot(dir, normal), 0.0, 1.0)"
8837 " * ffp_light[%u].diffuse.xyz * att;\n", idx
);
8838 if (settings
->localviewer
)
8839 shader_addline(buffer
, "t = dot(normal, normalize(dir - normalize(ec_pos.xyz)));\n");
8841 shader_addline(buffer
, "t = dot(normal, normalize(dir + vec3(0.0, 0.0, -1.0)));\n");
8842 shader_addline(buffer
, "if (dot(dir, normal) > 0.0 && t > 0.0%s) specular +="
8843 " pow(t, ffp_material.shininess) * ffp_light[%u].specular * att;\n",
8844 legacy_lighting
? " && ffp_material.shininess > 0.0" : "", idx
);
8847 static void shader_glsl_ffp_vertex_lighting(struct wined3d_string_buffer
*buffer
,
8848 const struct wined3d_ffp_vs_settings
*settings
, BOOL legacy_lighting
)
8850 const char *diffuse
, *specular
, *emissive
, *ambient
;
8851 unsigned int i
, idx
;
8853 if (!settings
->lighting
)
8855 shader_addline(buffer
, "ffp_varying_diffuse = ffp_attrib_diffuse;\n");
8856 shader_addline(buffer
, "ffp_varying_specular = ffp_attrib_specular;\n");
8860 shader_addline(buffer
, "vec3 ambient = ffp_light_ambient;\n");
8861 shader_addline(buffer
, "vec3 diffuse = vec3(0.0);\n");
8862 shader_addline(buffer
, "vec4 specular = vec4(0.0);\n");
8863 shader_addline(buffer
, "vec3 dir, dst;\n");
8864 shader_addline(buffer
, "float att, t;\n");
8866 ambient
= shader_glsl_ffp_mcs(settings
->ambient_source
, "ffp_material.ambient");
8867 diffuse
= shader_glsl_ffp_mcs(settings
->diffuse_source
, "ffp_material.diffuse");
8868 specular
= shader_glsl_ffp_mcs(settings
->specular_source
, "ffp_material.specular");
8869 emissive
= shader_glsl_ffp_mcs(settings
->emissive_source
, "ffp_material.emissive");
8872 for (i
= 0; i
< settings
->point_light_count
; ++i
, ++idx
)
8874 shader_addline(buffer
, "dir = ffp_light[%u].position.xyz - ec_pos.xyz;\n", idx
);
8875 shader_addline(buffer
, "dst.z = dot(dir, dir);\n");
8876 shader_addline(buffer
, "dst.y = sqrt(dst.z);\n");
8877 shader_addline(buffer
, "dst.x = 1.0;\n");
8878 if (legacy_lighting
)
8880 shader_addline(buffer
, "dst.y = (ffp_light[%u].range - dst.y) / ffp_light[%u].range;\n", idx
, idx
);
8881 shader_addline(buffer
, "dst.z = dst.y * dst.y;\n");
8882 shader_addline(buffer
, "if (dst.y > 0.0)\n{\n");
8886 shader_addline(buffer
, "if (dst.y <= ffp_light[%u].range)\n{\n", idx
);
8888 shader_addline(buffer
, "att = dot(dst.xyz, vec3(ffp_light[%u].c_att,"
8889 " ffp_light[%u].l_att, ffp_light[%u].q_att));\n", idx
, idx
, idx
);
8890 if (!legacy_lighting
)
8891 shader_addline(buffer
, "att = 1.0 / att;\n");
8892 shader_addline(buffer
, "ambient += ffp_light[%u].ambient.xyz * att;\n", idx
);
8893 if (!settings
->normal
)
8895 shader_addline(buffer
, "}\n");
8898 shader_addline(buffer
, "dir = normalize(dir);\n");
8899 shader_glsl_ffp_vertex_lighting_footer(buffer
, settings
, idx
, legacy_lighting
);
8900 shader_addline(buffer
, "}\n");
8903 for (i
= 0; i
< settings
->spot_light_count
; ++i
, ++idx
)
8905 shader_addline(buffer
, "dir = ffp_light[%u].position.xyz - ec_pos.xyz;\n", idx
);
8906 shader_addline(buffer
, "dst.z = dot(dir, dir);\n");
8907 shader_addline(buffer
, "dst.y = sqrt(dst.z);\n");
8908 shader_addline(buffer
, "dst.x = 1.0;\n");
8909 if (legacy_lighting
)
8911 shader_addline(buffer
, "dst.y = (ffp_light[%u].range - dst.y) / ffp_light[%u].range;\n", idx
, idx
);
8912 shader_addline(buffer
, "dst.z = dst.y * dst.y;\n");
8913 shader_addline(buffer
, "if (dst.y > 0.0)\n{\n");
8917 shader_addline(buffer
, "if (dst.y <= ffp_light[%u].range)\n{\n", idx
);
8919 shader_addline(buffer
, "dir = normalize(dir);\n");
8920 shader_addline(buffer
, "t = dot(-dir, normalize(ffp_light[%u].direction));\n", idx
);
8921 shader_addline(buffer
, "if (t > ffp_light[%u].cos_htheta) att = 1.0;\n", idx
);
8922 shader_addline(buffer
, "else if (t <= ffp_light[%u].cos_hphi) att = 0.0;\n", idx
);
8923 shader_addline(buffer
, "else att = pow((t - ffp_light[%u].cos_hphi)"
8924 " / (ffp_light[%u].cos_htheta - ffp_light[%u].cos_hphi), ffp_light[%u].falloff);\n",
8925 idx
, idx
, idx
, idx
);
8926 if (legacy_lighting
)
8927 shader_addline(buffer
, "att *= dot(dst.xyz, vec3(ffp_light[%u].c_att,"
8928 " ffp_light[%u].l_att, ffp_light[%u].q_att));\n",
8931 shader_addline(buffer
, "att /= dot(dst.xyz, vec3(ffp_light[%u].c_att,"
8932 " ffp_light[%u].l_att, ffp_light[%u].q_att));\n",
8934 shader_addline(buffer
, "ambient += ffp_light[%u].ambient.xyz * att;\n", idx
);
8935 if (!settings
->normal
)
8937 shader_addline(buffer
, "}\n");
8940 shader_glsl_ffp_vertex_lighting_footer(buffer
, settings
, idx
, legacy_lighting
);
8941 shader_addline(buffer
, "}\n");
8944 for (i
= 0; i
< settings
->directional_light_count
; ++i
, ++idx
)
8946 shader_addline(buffer
, "ambient += ffp_light[%u].ambient.xyz;\n", idx
);
8947 if (!settings
->normal
)
8949 shader_addline(buffer
, "att = 1.0;\n");
8950 shader_addline(buffer
, "dir = normalize(ffp_light[%u].direction.xyz);\n", idx
);
8951 shader_glsl_ffp_vertex_lighting_footer(buffer
, settings
, idx
, legacy_lighting
);
8954 for (i
= 0; i
< settings
->parallel_point_light_count
; ++i
, ++idx
)
8956 shader_addline(buffer
, "ambient += ffp_light[%u].ambient.xyz;\n", idx
);
8957 if (!settings
->normal
)
8959 shader_addline(buffer
, "att = 1.0;\n");
8960 shader_addline(buffer
, "dir = normalize(ffp_light[%u].position.xyz);\n", idx
);
8961 shader_glsl_ffp_vertex_lighting_footer(buffer
, settings
, idx
, legacy_lighting
);
8964 shader_addline(buffer
, "ffp_varying_diffuse.xyz = %s.xyz * ambient + %s.xyz * diffuse + %s.xyz;\n",
8965 ambient
, diffuse
, emissive
);
8966 shader_addline(buffer
, "ffp_varying_diffuse.w = %s.w;\n", diffuse
);
8967 shader_addline(buffer
, "ffp_varying_specular = %s * specular;\n", specular
);
8970 /* Context activation is done by the caller. */
8971 static GLuint
shader_glsl_generate_ffp_vertex_shader(struct shader_glsl_priv
*priv
,
8972 const struct wined3d_ffp_vs_settings
*settings
, const struct wined3d_gl_info
*gl_info
)
8974 static const struct attrib_info
8977 const char name
[24];
8981 {"vec4", "ffp_attrib_position"}, /* WINED3D_FFP_POSITION */
8982 {"vec4", "ffp_attrib_blendweight"}, /* WINED3D_FFP_BLENDWEIGHT */
8983 /* TODO: Indexed vertex blending */
8984 {"float", ""}, /* WINED3D_FFP_BLENDINDICES */
8985 {"vec3", "ffp_attrib_normal"}, /* WINED3D_FFP_NORMAL */
8986 {"float", "ffp_attrib_psize"}, /* WINED3D_FFP_PSIZE */
8987 {"vec4", "ffp_attrib_diffuse"}, /* WINED3D_FFP_DIFFUSE */
8988 {"vec4", "ffp_attrib_specular"}, /* WINED3D_FFP_SPECULAR */
8990 const BOOL legacy_syntax
= needs_legacy_glsl_syntax(gl_info
);
8991 struct wined3d_string_buffer
*buffer
= &priv
->shader_buffer
;
8992 BOOL output_legacy_fogcoord
= legacy_syntax
;
8993 BOOL legacy_lighting
= priv
->legacy_lighting
;
8997 string_buffer_clear(buffer
);
8999 shader_glsl_add_version_declaration(buffer
, gl_info
);
9001 if (shader_glsl_use_explicit_attrib_location(gl_info
))
9002 shader_addline(buffer
, "#extension GL_ARB_explicit_attrib_location : enable\n");
9004 shader_addline(buffer
, "invariant gl_Position;\n");
9006 for (i
= 0; i
< WINED3D_FFP_ATTRIBS_COUNT
; ++i
)
9008 const char *type
= i
< ARRAY_SIZE(attrib_info
) ? attrib_info
[i
].type
: "vec4";
9010 if (shader_glsl_use_explicit_attrib_location(gl_info
))
9011 shader_addline(buffer
, "layout(location = %u) ", i
);
9012 shader_addline(buffer
, "%s %s vs_in%u;\n", get_attribute_keyword(gl_info
), type
, i
);
9014 shader_addline(buffer
, "\n");
9016 shader_addline(buffer
, "uniform mat4 ffp_modelview_matrix[%u];\n", MAX_VERTEX_BLENDS
);
9017 shader_addline(buffer
, "uniform mat4 ffp_projection_matrix;\n");
9018 shader_addline(buffer
, "uniform mat3 ffp_normal_matrix;\n");
9019 shader_addline(buffer
, "uniform mat4 ffp_texture_matrix[%u];\n", WINED3D_MAX_TEXTURES
);
9021 shader_addline(buffer
, "uniform struct\n{\n");
9022 shader_addline(buffer
, " vec4 emissive;\n");
9023 shader_addline(buffer
, " vec4 ambient;\n");
9024 shader_addline(buffer
, " vec4 diffuse;\n");
9025 shader_addline(buffer
, " vec4 specular;\n");
9026 shader_addline(buffer
, " float shininess;\n");
9027 shader_addline(buffer
, "} ffp_material;\n");
9029 shader_addline(buffer
, "uniform vec3 ffp_light_ambient;\n");
9030 shader_addline(buffer
, "uniform struct\n{\n");
9031 shader_addline(buffer
, " vec4 diffuse;\n");
9032 shader_addline(buffer
, " vec4 specular;\n");
9033 shader_addline(buffer
, " vec4 ambient;\n");
9034 shader_addline(buffer
, " vec4 position;\n");
9035 shader_addline(buffer
, " vec3 direction;\n");
9036 shader_addline(buffer
, " float range;\n");
9037 shader_addline(buffer
, " float falloff;\n");
9038 shader_addline(buffer
, " float c_att;\n");
9039 shader_addline(buffer
, " float l_att;\n");
9040 shader_addline(buffer
, " float q_att;\n");
9041 shader_addline(buffer
, " float cos_htheta;\n");
9042 shader_addline(buffer
, " float cos_hphi;\n");
9043 shader_addline(buffer
, "} ffp_light[%u];\n", WINED3D_MAX_ACTIVE_LIGHTS
);
9045 if (settings
->point_size
)
9047 shader_addline(buffer
, "uniform struct\n{\n");
9048 shader_addline(buffer
, " float size;\n");
9049 shader_addline(buffer
, " float size_min;\n");
9050 shader_addline(buffer
, " float size_max;\n");
9051 shader_addline(buffer
, " float c_att;\n");
9052 shader_addline(buffer
, " float l_att;\n");
9053 shader_addline(buffer
, " float q_att;\n");
9054 shader_addline(buffer
, "} ffp_point;\n");
9059 shader_addline(buffer
, "vec4 ffp_varying_diffuse;\n");
9060 shader_addline(buffer
, "vec4 ffp_varying_specular;\n");
9061 shader_addline(buffer
, "vec4 ffp_varying_texcoord[%u];\n", WINED3D_MAX_TEXTURES
);
9062 shader_addline(buffer
, "float ffp_varying_fogcoord;\n");
9066 if (settings
->clipping
)
9067 shader_addline(buffer
, "uniform vec4 clip_planes[%u];\n", gl_info
->limits
.user_clip_distances
);
9069 declare_out_varying(gl_info
, buffer
, settings
->flatshading
, "vec4 ffp_varying_diffuse;\n");
9070 declare_out_varying(gl_info
, buffer
, settings
->flatshading
, "vec4 ffp_varying_specular;\n");
9071 declare_out_varying(gl_info
, buffer
, FALSE
, "vec4 ffp_varying_texcoord[%u];\n", WINED3D_MAX_TEXTURES
);
9072 declare_out_varying(gl_info
, buffer
, FALSE
, "float ffp_varying_fogcoord;\n");
9075 shader_addline(buffer
, "\nvoid main()\n{\n");
9076 shader_addline(buffer
, "float m;\n");
9077 shader_addline(buffer
, "vec3 r;\n");
9079 for (i
= 0; i
< ARRAY_SIZE(attrib_info
); ++i
)
9081 if (attrib_info
[i
].name
[0])
9082 shader_addline(buffer
, "%s %s = vs_in%u%s;\n", attrib_info
[i
].type
, attrib_info
[i
].name
,
9083 i
, settings
->swizzle_map
& (1u << i
) ? ".zyxw" : "");
9085 for (i
= 0; i
< WINED3D_MAX_TEXTURES
; ++i
)
9087 unsigned int coord_idx
= settings
->texgen
[i
] & 0x0000ffff;
9088 if ((settings
->texgen
[i
] & 0xffff0000) == WINED3DTSS_TCI_PASSTHRU
9089 && settings
->texcoords
& (1u << i
))
9090 shader_addline(buffer
, "vec4 ffp_attrib_texcoord%u = vs_in%u;\n", i
, coord_idx
+ WINED3D_FFP_TEXCOORD0
);
9093 shader_addline(buffer
, "ffp_attrib_blendweight[%u] = 1.0;\n", settings
->vertexblends
);
9095 if (settings
->transformed
)
9097 shader_addline(buffer
, "vec4 ec_pos = vec4(ffp_attrib_position.xyz, 1.0);\n");
9098 shader_addline(buffer
, "gl_Position = ffp_projection_matrix * ec_pos;\n");
9099 shader_addline(buffer
, "if (ffp_attrib_position.w != 0.0) gl_Position /= ffp_attrib_position.w;\n");
9103 for (i
= 0; i
< settings
->vertexblends
; ++i
)
9104 shader_addline(buffer
, "ffp_attrib_blendweight[%u] -= ffp_attrib_blendweight[%u];\n", settings
->vertexblends
, i
);
9106 shader_addline(buffer
, "vec4 ec_pos = vec4(0.0);\n");
9107 for (i
= 0; i
< settings
->vertexblends
+ 1; ++i
)
9108 shader_addline(buffer
, "ec_pos += ffp_attrib_blendweight[%u] * (ffp_modelview_matrix[%u] * ffp_attrib_position);\n", i
, i
);
9110 shader_addline(buffer
, "gl_Position = ffp_projection_matrix * ec_pos;\n");
9111 if (settings
->clipping
)
9114 shader_addline(buffer
, "gl_ClipVertex = ec_pos;\n");
9116 for (i
= 0; i
< gl_info
->limits
.user_clip_distances
; ++i
)
9117 shader_addline(buffer
, "gl_ClipDistance[%u] = dot(ec_pos, clip_planes[%u]);\n", i
, i
);
9119 shader_addline(buffer
, "ec_pos /= ec_pos.w;\n");
9122 shader_addline(buffer
, "vec3 normal = vec3(0.0);\n");
9123 if (settings
->normal
)
9125 if (!settings
->vertexblends
)
9127 shader_addline(buffer
, "normal = ffp_normal_matrix * ffp_attrib_normal;\n");
9131 for (i
= 0; i
< settings
->vertexblends
+ 1; ++i
)
9132 shader_addline(buffer
, "normal += ffp_attrib_blendweight[%u] * (mat3(ffp_modelview_matrix[%u]) * ffp_attrib_normal);\n", i
, i
);
9135 if (settings
->normalize
)
9136 shader_addline(buffer
, "normal = normalize(normal);\n");
9139 shader_glsl_ffp_vertex_lighting(buffer
, settings
, legacy_lighting
);
9142 shader_addline(buffer
, "gl_FrontColor = ffp_varying_diffuse;\n");
9143 shader_addline(buffer
, "gl_FrontSecondaryColor = ffp_varying_specular;\n");
9147 shader_addline(buffer
, "ffp_varying_diffuse = clamp(ffp_varying_diffuse, 0.0, 1.0);\n");
9148 shader_addline(buffer
, "ffp_varying_specular = clamp(ffp_varying_specular, 0.0, 1.0);\n");
9151 for (i
= 0; i
< WINED3D_MAX_TEXTURES
; ++i
)
9153 BOOL output_legacy_texcoord
= legacy_syntax
;
9155 switch (settings
->texgen
[i
] & 0xffff0000)
9157 case WINED3DTSS_TCI_PASSTHRU
:
9158 if (settings
->texcoords
& (1u << i
))
9159 shader_addline(buffer
, "ffp_varying_texcoord[%u] = ffp_texture_matrix[%u] * ffp_attrib_texcoord%u;\n",
9161 else if (shader_glsl_full_ffp_varyings(gl_info
))
9162 shader_addline(buffer
, "ffp_varying_texcoord[%u] = vec4(0.0);\n", i
);
9164 output_legacy_texcoord
= FALSE
;
9167 case WINED3DTSS_TCI_CAMERASPACENORMAL
:
9168 shader_addline(buffer
, "ffp_varying_texcoord[%u] = ffp_texture_matrix[%u] * vec4(normal, 1.0);\n", i
, i
);
9171 case WINED3DTSS_TCI_CAMERASPACEPOSITION
:
9172 shader_addline(buffer
, "ffp_varying_texcoord[%u] = ffp_texture_matrix[%u] * ec_pos;\n", i
, i
);
9175 case WINED3DTSS_TCI_CAMERASPACEREFLECTIONVECTOR
:
9176 shader_addline(buffer
, "ffp_varying_texcoord[%u] = ffp_texture_matrix[%u]"
9177 " * vec4(reflect(normalize(ec_pos.xyz), normal), 1.0);\n", i
, i
);
9180 case WINED3DTSS_TCI_SPHEREMAP
:
9181 shader_addline(buffer
, "r = reflect(normalize(ec_pos.xyz), normal);\n");
9182 shader_addline(buffer
, "m = 2.0 * length(vec3(r.x, r.y, r.z + 1.0));\n");
9183 shader_addline(buffer
, "ffp_varying_texcoord[%u] = ffp_texture_matrix[%u]"
9184 " * vec4(r.x / m + 0.5, r.y / m + 0.5, 0.0, 1.0);\n", i
, i
);
9188 ERR("Unhandled texgen %#x.\n", settings
->texgen
[i
]);
9191 if (output_legacy_texcoord
)
9192 shader_addline(buffer
, "gl_TexCoord[%u] = ffp_varying_texcoord[%u];\n", i
, i
);
9195 switch (settings
->fog_mode
)
9197 case WINED3D_FFP_VS_FOG_OFF
:
9198 output_legacy_fogcoord
= FALSE
;
9201 case WINED3D_FFP_VS_FOG_FOGCOORD
:
9202 shader_addline(buffer
, "ffp_varying_fogcoord = ffp_attrib_specular.w * 255.0;\n");
9205 case WINED3D_FFP_VS_FOG_RANGE
:
9206 shader_addline(buffer
, "ffp_varying_fogcoord = length(ec_pos.xyz);\n");
9209 case WINED3D_FFP_VS_FOG_DEPTH
:
9210 if (settings
->ortho_fog
)
9212 if (gl_info
->supported
[ARB_CLIP_CONTROL
])
9213 shader_addline(buffer
, "ffp_varying_fogcoord = gl_Position.z;\n");
9215 /* Need to undo the [0.0 - 1.0] -> [-1.0 - 1.0] transformation from D3D to GL coordinates. */
9216 shader_addline(buffer
, "ffp_varying_fogcoord = gl_Position.z * 0.5 + 0.5;\n");
9218 else if (settings
->transformed
)
9220 shader_addline(buffer
, "ffp_varying_fogcoord = ec_pos.z;\n");
9224 shader_addline(buffer
, "ffp_varying_fogcoord = abs(ec_pos.z);\n");
9229 ERR("Unhandled fog mode %#x.\n", settings
->fog_mode
);
9232 if (output_legacy_fogcoord
)
9233 shader_addline(buffer
, "gl_FogFragCoord = ffp_varying_fogcoord;\n");
9235 if (settings
->point_size
)
9237 shader_addline(buffer
, "gl_PointSize = %s / sqrt(ffp_point.c_att"
9238 " + ffp_point.l_att * length(ec_pos.xyz)"
9239 " + ffp_point.q_att * dot(ec_pos.xyz, ec_pos.xyz));\n",
9240 settings
->per_vertex_point_size
? "ffp_attrib_psize" : "ffp_point.size");
9241 shader_addline(buffer
, "gl_PointSize = clamp(gl_PointSize, ffp_point.size_min, ffp_point.size_max);\n");
9244 shader_addline(buffer
, "}\n");
9246 shader_obj
= GL_EXTCALL(glCreateShader(GL_VERTEX_SHADER
));
9247 shader_glsl_compile(gl_info
, shader_obj
, buffer
->buffer
);
9252 static const char *shader_glsl_get_ffp_fragment_op_arg(struct wined3d_string_buffer
*buffer
,
9253 DWORD argnum
, unsigned int stage
, DWORD arg
)
9257 if (arg
== ARG_UNUSED
)
9258 return "<unused arg>";
9260 switch (arg
& WINED3DTA_SELECTMASK
)
9262 case WINED3DTA_DIFFUSE
:
9263 ret
= "ffp_varying_diffuse";
9266 case WINED3DTA_CURRENT
:
9270 case WINED3DTA_TEXTURE
:
9273 case 0: ret
= "tex0"; break;
9274 case 1: ret
= "tex1"; break;
9275 case 2: ret
= "tex2"; break;
9276 case 3: ret
= "tex3"; break;
9277 case 4: ret
= "tex4"; break;
9278 case 5: ret
= "tex5"; break;
9279 case 6: ret
= "tex6"; break;
9280 case 7: ret
= "tex7"; break;
9282 ret
= "<invalid texture>";
9287 case WINED3DTA_TFACTOR
:
9291 case WINED3DTA_SPECULAR
:
9292 ret
= "ffp_varying_specular";
9295 case WINED3DTA_TEMP
:
9299 case WINED3DTA_CONSTANT
:
9302 case 0: ret
= "tss_const0"; break;
9303 case 1: ret
= "tss_const1"; break;
9304 case 2: ret
= "tss_const2"; break;
9305 case 3: ret
= "tss_const3"; break;
9306 case 4: ret
= "tss_const4"; break;
9307 case 5: ret
= "tss_const5"; break;
9308 case 6: ret
= "tss_const6"; break;
9309 case 7: ret
= "tss_const7"; break;
9311 ret
= "<invalid constant>";
9317 return "<unhandled arg>";
9320 if (arg
& WINED3DTA_COMPLEMENT
)
9322 shader_addline(buffer
, "arg%u = vec4(1.0) - %s;\n", argnum
, ret
);
9325 else if (argnum
== 1)
9327 else if (argnum
== 2)
9331 if (arg
& WINED3DTA_ALPHAREPLICATE
)
9333 shader_addline(buffer
, "arg%u = vec4(%s.w);\n", argnum
, ret
);
9336 else if (argnum
== 1)
9338 else if (argnum
== 2)
9345 static void shader_glsl_ffp_fragment_op(struct wined3d_string_buffer
*buffer
, unsigned int stage
, BOOL color
,
9346 BOOL alpha
, BOOL tmp_dst
, DWORD op
, DWORD dw_arg0
, DWORD dw_arg1
, DWORD dw_arg2
)
9348 const char *dstmask
, *dstreg
, *arg0
, *arg1
, *arg2
;
9357 dstreg
= tmp_dst
? "temp_reg" : "ret";
9359 arg0
= shader_glsl_get_ffp_fragment_op_arg(buffer
, 0, stage
, dw_arg0
);
9360 arg1
= shader_glsl_get_ffp_fragment_op_arg(buffer
, 1, stage
, dw_arg1
);
9361 arg2
= shader_glsl_get_ffp_fragment_op_arg(buffer
, 2, stage
, dw_arg2
);
9365 case WINED3D_TOP_DISABLE
:
9368 case WINED3D_TOP_SELECT_ARG1
:
9369 shader_addline(buffer
, "%s%s = %s%s;\n", dstreg
, dstmask
, arg1
, dstmask
);
9372 case WINED3D_TOP_SELECT_ARG2
:
9373 shader_addline(buffer
, "%s%s = %s%s;\n", dstreg
, dstmask
, arg2
, dstmask
);
9376 case WINED3D_TOP_MODULATE
:
9377 shader_addline(buffer
, "%s%s = %s%s * %s%s;\n", dstreg
, dstmask
, arg1
, dstmask
, arg2
, dstmask
);
9380 case WINED3D_TOP_MODULATE_4X
:
9381 shader_addline(buffer
, "%s%s = clamp(%s%s * %s%s * 4.0, 0.0, 1.0);\n",
9382 dstreg
, dstmask
, arg1
, dstmask
, arg2
, dstmask
);
9385 case WINED3D_TOP_MODULATE_2X
:
9386 shader_addline(buffer
, "%s%s = clamp(%s%s * %s%s * 2.0, 0.0, 1.0);\n",
9387 dstreg
, dstmask
, arg1
, dstmask
, arg2
, dstmask
);
9390 case WINED3D_TOP_ADD
:
9391 shader_addline(buffer
, "%s%s = clamp(%s%s + %s%s, 0.0, 1.0);\n",
9392 dstreg
, dstmask
, arg1
, dstmask
, arg2
, dstmask
);
9395 case WINED3D_TOP_ADD_SIGNED
:
9396 shader_addline(buffer
, "%s%s = clamp(%s%s + (%s - vec4(0.5))%s, 0.0, 1.0);\n",
9397 dstreg
, dstmask
, arg1
, dstmask
, arg2
, dstmask
);
9400 case WINED3D_TOP_ADD_SIGNED_2X
:
9401 shader_addline(buffer
, "%s%s = clamp((%s%s + (%s - vec4(0.5))%s) * 2.0, 0.0, 1.0);\n",
9402 dstreg
, dstmask
, arg1
, dstmask
, arg2
, dstmask
);
9405 case WINED3D_TOP_SUBTRACT
:
9406 shader_addline(buffer
, "%s%s = clamp(%s%s - %s%s, 0.0, 1.0);\n",
9407 dstreg
, dstmask
, arg1
, dstmask
, arg2
, dstmask
);
9410 case WINED3D_TOP_ADD_SMOOTH
:
9411 shader_addline(buffer
, "%s%s = clamp((vec4(1.0) - %s)%s * %s%s + %s%s, 0.0, 1.0);\n",
9412 dstreg
, dstmask
, arg1
, dstmask
, arg2
, dstmask
, arg1
, dstmask
);
9415 case WINED3D_TOP_BLEND_DIFFUSE_ALPHA
:
9416 arg0
= shader_glsl_get_ffp_fragment_op_arg(buffer
, 0, stage
, WINED3DTA_DIFFUSE
);
9417 shader_addline(buffer
, "%s%s = mix(%s%s, %s%s, %s.w);\n",
9418 dstreg
, dstmask
, arg2
, dstmask
, arg1
, dstmask
, arg0
);
9421 case WINED3D_TOP_BLEND_TEXTURE_ALPHA
:
9422 arg0
= shader_glsl_get_ffp_fragment_op_arg(buffer
, 0, stage
, WINED3DTA_TEXTURE
);
9423 shader_addline(buffer
, "%s%s = mix(%s%s, %s%s, %s.w);\n",
9424 dstreg
, dstmask
, arg2
, dstmask
, arg1
, dstmask
, arg0
);
9427 case WINED3D_TOP_BLEND_FACTOR_ALPHA
:
9428 arg0
= shader_glsl_get_ffp_fragment_op_arg(buffer
, 0, stage
, WINED3DTA_TFACTOR
);
9429 shader_addline(buffer
, "%s%s = mix(%s%s, %s%s, %s.w);\n",
9430 dstreg
, dstmask
, arg2
, dstmask
, arg1
, dstmask
, arg0
);
9433 case WINED3D_TOP_BLEND_TEXTURE_ALPHA_PM
:
9434 arg0
= shader_glsl_get_ffp_fragment_op_arg(buffer
, 0, stage
, WINED3DTA_TEXTURE
);
9435 shader_addline(buffer
, "%s%s = clamp(%s%s * (1.0 - %s.w) + %s%s, 0.0, 1.0);\n",
9436 dstreg
, dstmask
, arg2
, dstmask
, arg0
, arg1
, dstmask
);
9439 case WINED3D_TOP_BLEND_CURRENT_ALPHA
:
9440 arg0
= shader_glsl_get_ffp_fragment_op_arg(buffer
, 0, stage
, WINED3DTA_CURRENT
);
9441 shader_addline(buffer
, "%s%s = mix(%s%s, %s%s, %s.w);\n",
9442 dstreg
, dstmask
, arg2
, dstmask
, arg1
, dstmask
, arg0
);
9445 case WINED3D_TOP_MODULATE_ALPHA_ADD_COLOR
:
9446 shader_addline(buffer
, "%s%s = clamp(%s%s * %s.w + %s%s, 0.0, 1.0);\n",
9447 dstreg
, dstmask
, arg2
, dstmask
, arg1
, arg1
, dstmask
);
9450 case WINED3D_TOP_MODULATE_COLOR_ADD_ALPHA
:
9451 shader_addline(buffer
, "%s%s = clamp(%s%s * %s%s + %s.w, 0.0, 1.0);\n",
9452 dstreg
, dstmask
, arg1
, dstmask
, arg2
, dstmask
, arg1
);
9455 case WINED3D_TOP_MODULATE_INVALPHA_ADD_COLOR
:
9456 shader_addline(buffer
, "%s%s = clamp(%s%s * (1.0 - %s.w) + %s%s, 0.0, 1.0);\n",
9457 dstreg
, dstmask
, arg2
, dstmask
, arg1
, arg1
, dstmask
);
9459 case WINED3D_TOP_MODULATE_INVCOLOR_ADD_ALPHA
:
9460 shader_addline(buffer
, "%s%s = clamp((vec4(1.0) - %s)%s * %s%s + %s.w, 0.0, 1.0);\n",
9461 dstreg
, dstmask
, arg1
, dstmask
, arg2
, dstmask
, arg1
);
9464 case WINED3D_TOP_BUMPENVMAP
:
9465 case WINED3D_TOP_BUMPENVMAP_LUMINANCE
:
9466 /* These are handled in the first pass, nothing to do. */
9469 case WINED3D_TOP_DOTPRODUCT3
:
9470 shader_addline(buffer
, "%s%s = vec4(clamp(dot(%s.xyz - 0.5, %s.xyz - 0.5) * 4.0, 0.0, 1.0))%s;\n",
9471 dstreg
, dstmask
, arg1
, arg2
, dstmask
);
9474 case WINED3D_TOP_MULTIPLY_ADD
:
9475 shader_addline(buffer
, "%s%s = clamp(%s%s * %s%s + %s%s, 0.0, 1.0);\n",
9476 dstreg
, dstmask
, arg1
, dstmask
, arg2
, dstmask
, arg0
, dstmask
);
9479 case WINED3D_TOP_LERP
:
9480 /* MSDN isn't quite right here. */
9481 shader_addline(buffer
, "%s%s = mix(%s%s, %s%s, %s%s);\n",
9482 dstreg
, dstmask
, arg2
, dstmask
, arg1
, dstmask
, arg0
, dstmask
);
9486 FIXME("Unhandled operation %#x.\n", op
);
9491 /* Context activation is done by the caller. */
9492 static GLuint
shader_glsl_generate_ffp_fragment_shader(struct shader_glsl_priv
*priv
,
9493 const struct ffp_frag_settings
*settings
, const struct wined3d_context_gl
*context_gl
)
9495 struct wined3d_string_buffer
*tex_reg_name
= string_buffer_get(&priv
->string_buffers
);
9496 enum wined3d_cmp_func alpha_test_func
= settings
->alpha_test_func
+ 1;
9497 struct wined3d_string_buffer
*buffer
= &priv
->shader_buffer
;
9498 BYTE lum_map
= 0, bump_map
= 0, tex_map
= 0, tss_const_map
= 0;
9499 const struct wined3d_gl_info
*gl_info
= context_gl
->gl_info
;
9500 const BOOL legacy_syntax
= needs_legacy_glsl_syntax(gl_info
);
9501 BOOL tempreg_used
= FALSE
, tfactor_used
= FALSE
;
9502 UINT lowest_disabled_stage
;
9504 DWORD arg0
, arg1
, arg2
;
9507 string_buffer_clear(buffer
);
9509 /* Find out which textures are read */
9510 for (stage
= 0; stage
< WINED3D_MAX_TEXTURES
; ++stage
)
9512 if (settings
->op
[stage
].cop
== WINED3D_TOP_DISABLE
)
9515 arg0
= settings
->op
[stage
].carg0
& WINED3DTA_SELECTMASK
;
9516 arg1
= settings
->op
[stage
].carg1
& WINED3DTA_SELECTMASK
;
9517 arg2
= settings
->op
[stage
].carg2
& WINED3DTA_SELECTMASK
;
9519 if (arg0
== WINED3DTA_TEXTURE
|| arg1
== WINED3DTA_TEXTURE
|| arg2
== WINED3DTA_TEXTURE
9520 || (stage
== 0 && settings
->color_key_enabled
))
9521 tex_map
|= 1u << stage
;
9522 if (arg0
== WINED3DTA_TFACTOR
|| arg1
== WINED3DTA_TFACTOR
|| arg2
== WINED3DTA_TFACTOR
)
9523 tfactor_used
= TRUE
;
9524 if (arg0
== WINED3DTA_TEMP
|| arg1
== WINED3DTA_TEMP
|| arg2
== WINED3DTA_TEMP
)
9525 tempreg_used
= TRUE
;
9526 if (settings
->op
[stage
].tmp_dst
)
9527 tempreg_used
= TRUE
;
9528 if (arg0
== WINED3DTA_CONSTANT
|| arg1
== WINED3DTA_CONSTANT
|| arg2
== WINED3DTA_CONSTANT
)
9529 tss_const_map
|= 1u << stage
;
9531 switch (settings
->op
[stage
].cop
)
9533 case WINED3D_TOP_BUMPENVMAP_LUMINANCE
:
9534 lum_map
|= 1u << stage
;
9536 case WINED3D_TOP_BUMPENVMAP
:
9537 bump_map
|= 1u << stage
;
9539 case WINED3D_TOP_BLEND_TEXTURE_ALPHA
:
9540 case WINED3D_TOP_BLEND_TEXTURE_ALPHA_PM
:
9541 tex_map
|= 1u << stage
;
9544 case WINED3D_TOP_BLEND_FACTOR_ALPHA
:
9545 tfactor_used
= TRUE
;
9552 if (settings
->op
[stage
].aop
== WINED3D_TOP_DISABLE
)
9555 arg0
= settings
->op
[stage
].aarg0
& WINED3DTA_SELECTMASK
;
9556 arg1
= settings
->op
[stage
].aarg1
& WINED3DTA_SELECTMASK
;
9557 arg2
= settings
->op
[stage
].aarg2
& WINED3DTA_SELECTMASK
;
9559 if (arg0
== WINED3DTA_TEXTURE
|| arg1
== WINED3DTA_TEXTURE
|| arg2
== WINED3DTA_TEXTURE
)
9560 tex_map
|= 1u << stage
;
9561 if (arg0
== WINED3DTA_TFACTOR
|| arg1
== WINED3DTA_TFACTOR
|| arg2
== WINED3DTA_TFACTOR
)
9562 tfactor_used
= TRUE
;
9563 if (arg0
== WINED3DTA_TEMP
|| arg1
== WINED3DTA_TEMP
|| arg2
== WINED3DTA_TEMP
)
9564 tempreg_used
= TRUE
;
9565 if (arg0
== WINED3DTA_CONSTANT
|| arg1
== WINED3DTA_CONSTANT
|| arg2
== WINED3DTA_CONSTANT
)
9566 tss_const_map
|= 1u << stage
;
9568 lowest_disabled_stage
= stage
;
9570 shader_glsl_add_version_declaration(buffer
, gl_info
);
9572 if (shader_glsl_use_explicit_attrib_location(gl_info
))
9573 shader_addline(buffer
, "#extension GL_ARB_explicit_attrib_location : enable\n");
9574 if (gl_info
->supported
[ARB_SHADING_LANGUAGE_420PACK
])
9575 shader_addline(buffer
, "#extension GL_ARB_shading_language_420pack : enable\n");
9576 if (gl_info
->supported
[ARB_TEXTURE_RECTANGLE
])
9577 shader_addline(buffer
, "#extension GL_ARB_texture_rectangle : enable\n");
9579 if (!use_legacy_fragment_output(gl_info
))
9581 shader_addline(buffer
, "vec4 ps_out[1];\n");
9582 if (shader_glsl_use_explicit_attrib_location(gl_info
))
9583 shader_addline(buffer
, "layout(location = 0) ");
9584 shader_addline(buffer
, "out vec4 color_out0;\n");
9587 shader_addline(buffer
, "vec4 tmp0, tmp1;\n");
9588 shader_addline(buffer
, "vec4 ret;\n");
9589 if (tempreg_used
|| settings
->sRGB_write
)
9590 shader_addline(buffer
, "vec4 temp_reg = vec4(0.0);\n");
9591 shader_addline(buffer
, "vec4 arg0, arg1, arg2;\n");
9593 for (stage
= 0; stage
< WINED3D_MAX_TEXTURES
; ++stage
)
9595 const char *sampler_type
;
9597 if (tss_const_map
& (1u << stage
))
9598 shader_addline(buffer
, "uniform vec4 tss_const%u;\n", stage
);
9600 if (!(tex_map
& (1u << stage
)))
9603 switch (settings
->op
[stage
].tex_type
)
9605 case WINED3D_GL_RES_TYPE_TEX_1D
:
9606 sampler_type
= "1D";
9608 case WINED3D_GL_RES_TYPE_TEX_2D
:
9609 sampler_type
= "2D";
9611 case WINED3D_GL_RES_TYPE_TEX_3D
:
9612 sampler_type
= "3D";
9614 case WINED3D_GL_RES_TYPE_TEX_CUBE
:
9615 sampler_type
= "Cube";
9617 case WINED3D_GL_RES_TYPE_TEX_RECT
:
9618 sampler_type
= "2DRect";
9621 FIXME("Unhandled sampler type %#x.\n", settings
->op
[stage
].tex_type
);
9622 sampler_type
= NULL
;
9627 if (shader_glsl_use_layout_binding_qualifier(gl_info
))
9628 shader_glsl_append_sampler_binding_qualifier(buffer
, &context_gl
->c
, NULL
, stage
);
9629 shader_addline(buffer
, "uniform sampler%s ps_sampler%u;\n", sampler_type
, stage
);
9632 shader_addline(buffer
, "vec4 tex%u;\n", stage
);
9634 if (!(bump_map
& (1u << stage
)))
9636 shader_addline(buffer
, "uniform mat2 bumpenv_mat%u;\n", stage
);
9638 if (!(lum_map
& (1u << stage
)))
9640 shader_addline(buffer
, "uniform float bumpenv_lum_scale%u;\n", stage
);
9641 shader_addline(buffer
, "uniform float bumpenv_lum_offset%u;\n", stage
);
9644 shader_addline(buffer
, "uniform vec4 tex_factor;\n");
9645 if (settings
->color_key_enabled
)
9646 shader_addline(buffer
, "uniform vec4 color_key[2];\n");
9647 shader_addline(buffer
, "uniform vec4 specular_enable;\n");
9649 if (settings
->sRGB_write
)
9651 shader_addline(buffer
, "const vec4 srgb_const0 = ");
9652 shader_glsl_append_imm_vec(buffer
, &wined3d_srgb_const
[0].x
, 4, gl_info
);
9653 shader_addline(buffer
, ";\n");
9654 shader_addline(buffer
, "const vec4 srgb_const1 = ");
9655 shader_glsl_append_imm_vec(buffer
, &wined3d_srgb_const
[1].x
, 4, gl_info
);
9656 shader_addline(buffer
, ";\n");
9659 shader_addline(buffer
, "uniform struct\n{\n");
9660 shader_addline(buffer
, " vec4 color;\n");
9661 shader_addline(buffer
, " float density;\n");
9662 shader_addline(buffer
, " float end;\n");
9663 shader_addline(buffer
, " float scale;\n");
9664 shader_addline(buffer
, "} ffp_fog;\n");
9666 if (alpha_test_func
!= WINED3D_CMP_ALWAYS
)
9667 shader_addline(buffer
, "uniform float alpha_test_ref;\n");
9671 shader_addline(buffer
, "vec4 ffp_varying_diffuse;\n");
9672 shader_addline(buffer
, "vec4 ffp_varying_specular;\n");
9673 shader_addline(buffer
, "vec4 ffp_varying_texcoord[%u];\n", WINED3D_MAX_TEXTURES
);
9674 shader_addline(buffer
, "vec4 ffp_texcoord[%u];\n", WINED3D_MAX_TEXTURES
);
9675 shader_addline(buffer
, "float ffp_varying_fogcoord;\n");
9679 declare_in_varying(gl_info
, buffer
, settings
->flatshading
, "vec4 ffp_varying_diffuse;\n");
9680 declare_in_varying(gl_info
, buffer
, settings
->flatshading
, "vec4 ffp_varying_specular;\n");
9681 declare_in_varying(gl_info
, buffer
, FALSE
, "vec4 ffp_varying_texcoord[%u];\n", WINED3D_MAX_TEXTURES
);
9682 shader_addline(buffer
, "vec4 ffp_texcoord[%u];\n", WINED3D_MAX_TEXTURES
);
9683 declare_in_varying(gl_info
, buffer
, FALSE
, "float ffp_varying_fogcoord;\n");
9686 shader_addline(buffer
, "void main()\n{\n");
9690 shader_addline(buffer
, "ffp_varying_diffuse = gl_Color;\n");
9691 shader_addline(buffer
, "ffp_varying_specular = gl_SecondaryColor;\n");
9694 for (stage
= 0; stage
< WINED3D_MAX_TEXTURES
; ++stage
)
9696 if (tex_map
& (1u << stage
))
9698 if (settings
->pointsprite
)
9699 shader_addline(buffer
, "ffp_texcoord[%u] = vec4(gl_PointCoord.xy, 0.0, 0.0);\n", stage
);
9700 else if (settings
->texcoords_initialized
& (1u << stage
))
9701 shader_addline(buffer
, "ffp_texcoord[%u] = %s[%u];\n",
9702 stage
, legacy_syntax
? "gl_TexCoord" : "ffp_varying_texcoord", stage
);
9704 shader_addline(buffer
, "ffp_texcoord[%u] = vec4(0.0);\n", stage
);
9708 if (legacy_syntax
&& settings
->fog
!= WINED3D_FFP_PS_FOG_OFF
)
9709 shader_addline(buffer
, "ffp_varying_fogcoord = gl_FogFragCoord;\n");
9711 if (lowest_disabled_stage
< 7 && settings
->emul_clipplanes
)
9712 shader_addline(buffer
, "if (any(lessThan(ffp_texcoord[7], vec4(0.0)))) discard;\n");
9714 /* Generate texture sampling instructions */
9715 for (stage
= 0; stage
< WINED3D_MAX_TEXTURES
&& settings
->op
[stage
].cop
!= WINED3D_TOP_DISABLE
; ++stage
)
9717 const char *texture_function
, *coord_mask
;
9720 if (!(tex_map
& (1u << stage
)))
9723 if (settings
->op
[stage
].projected
== WINED3D_PROJECTION_NONE
)
9727 else if (settings
->op
[stage
].projected
== WINED3D_PROJECTION_COUNT4
9728 || settings
->op
[stage
].projected
== WINED3D_PROJECTION_COUNT3
)
9734 FIXME("Unexpected projection mode %d\n", settings
->op
[stage
].projected
);
9738 if (settings
->op
[stage
].tex_type
== WINED3D_GL_RES_TYPE_TEX_CUBE
)
9741 switch (settings
->op
[stage
].tex_type
)
9743 case WINED3D_GL_RES_TYPE_TEX_1D
:
9744 texture_function
= "texture1D";
9747 case WINED3D_GL_RES_TYPE_TEX_2D
:
9748 texture_function
= "texture2D";
9751 case WINED3D_GL_RES_TYPE_TEX_3D
:
9752 texture_function
= "texture3D";
9755 case WINED3D_GL_RES_TYPE_TEX_CUBE
:
9756 texture_function
= "textureCube";
9759 case WINED3D_GL_RES_TYPE_TEX_RECT
:
9760 texture_function
= "texture2DRect";
9764 FIXME("Unhandled texture type %#x.\n", settings
->op
[stage
].tex_type
);
9765 texture_function
= "";
9766 coord_mask
= "xyzw";
9771 texture_function
= "texture";
9774 && (settings
->op
[stage
- 1].cop
== WINED3D_TOP_BUMPENVMAP
9775 || settings
->op
[stage
- 1].cop
== WINED3D_TOP_BUMPENVMAP_LUMINANCE
))
9777 shader_addline(buffer
, "ret.xy = bumpenv_mat%u * tex%u.xy;\n", stage
- 1, stage
- 1);
9779 /* With projective textures, texbem only divides the static
9780 * texture coordinate, not the displacement, so multiply the
9781 * displacement with the dividing parameter before passing it to
9783 if (settings
->op
[stage
].projected
!= WINED3D_PROJECTION_NONE
)
9785 if (settings
->op
[stage
].projected
== WINED3D_PROJECTION_COUNT4
)
9787 shader_addline(buffer
, "ret.xy = (ret.xy * ffp_texcoord[%u].w) + ffp_texcoord[%u].xy;\n",
9789 shader_addline(buffer
, "ret.zw = ffp_texcoord[%u].ww;\n", stage
);
9793 shader_addline(buffer
, "ret.xy = (ret.xy * ffp_texcoord[%u].z) + ffp_texcoord[%u].xy;\n",
9795 shader_addline(buffer
, "ret.zw = ffp_texcoord[%u].zz;\n", stage
);
9800 shader_addline(buffer
, "ret = ffp_texcoord[%u] + ret.xyxy;\n", stage
);
9803 shader_addline(buffer
, "tex%u = %s%s(ps_sampler%u, ret.%s%s);\n",
9804 stage
, texture_function
, proj
? "Proj" : "", stage
, coord_mask
, proj
? "w" : "");
9806 if (settings
->op
[stage
- 1].cop
== WINED3D_TOP_BUMPENVMAP_LUMINANCE
)
9807 shader_addline(buffer
, "tex%u *= clamp(tex%u.z * bumpenv_lum_scale%u + bumpenv_lum_offset%u, 0.0, 1.0);\n",
9808 stage
, stage
- 1, stage
- 1, stage
- 1);
9810 else if (settings
->op
[stage
].projected
== WINED3D_PROJECTION_COUNT3
)
9812 shader_addline(buffer
, "tex%u = %s%s(ps_sampler%u, ffp_texcoord[%u].xyz);\n",
9813 stage
, texture_function
, proj
? "Proj" : "", stage
, stage
);
9817 shader_addline(buffer
, "tex%u = %s%s(ps_sampler%u, ffp_texcoord[%u].%s%s);\n",
9818 stage
, texture_function
, proj
? "Proj" : "", stage
, stage
, coord_mask
, proj
? "w" : "");
9821 string_buffer_sprintf(tex_reg_name
, "tex%u", stage
);
9822 shader_glsl_color_correction_ext(buffer
, tex_reg_name
->buffer
, WINED3DSP_WRITEMASK_ALL
,
9823 settings
->op
[stage
].color_fixup
);
9826 if (settings
->color_key_enabled
)
9827 shader_glsl_generate_colour_key_test(buffer
, "tex0", "color_key[0]", "color_key[1]");
9829 shader_addline(buffer
, "ret = ffp_varying_diffuse;\n");
9831 /* Generate the main shader */
9832 for (stage
= 0; stage
< WINED3D_MAX_TEXTURES
; ++stage
)
9836 if (settings
->op
[stage
].cop
== WINED3D_TOP_DISABLE
)
9839 if (settings
->op
[stage
].cop
== WINED3D_TOP_SELECT_ARG1
9840 && settings
->op
[stage
].aop
== WINED3D_TOP_SELECT_ARG1
)
9841 op_equal
= settings
->op
[stage
].carg1
== settings
->op
[stage
].aarg1
;
9842 else if (settings
->op
[stage
].cop
== WINED3D_TOP_SELECT_ARG1
9843 && settings
->op
[stage
].aop
== WINED3D_TOP_SELECT_ARG2
)
9844 op_equal
= settings
->op
[stage
].carg1
== settings
->op
[stage
].aarg2
;
9845 else if (settings
->op
[stage
].cop
== WINED3D_TOP_SELECT_ARG2
9846 && settings
->op
[stage
].aop
== WINED3D_TOP_SELECT_ARG1
)
9847 op_equal
= settings
->op
[stage
].carg2
== settings
->op
[stage
].aarg1
;
9848 else if (settings
->op
[stage
].cop
== WINED3D_TOP_SELECT_ARG2
9849 && settings
->op
[stage
].aop
== WINED3D_TOP_SELECT_ARG2
)
9850 op_equal
= settings
->op
[stage
].carg2
== settings
->op
[stage
].aarg2
;
9852 op_equal
= settings
->op
[stage
].aop
== settings
->op
[stage
].cop
9853 && settings
->op
[stage
].carg0
== settings
->op
[stage
].aarg0
9854 && settings
->op
[stage
].carg1
== settings
->op
[stage
].aarg1
9855 && settings
->op
[stage
].carg2
== settings
->op
[stage
].aarg2
;
9857 if (settings
->op
[stage
].aop
== WINED3D_TOP_DISABLE
)
9859 shader_glsl_ffp_fragment_op(buffer
, stage
, TRUE
, FALSE
, settings
->op
[stage
].tmp_dst
,
9860 settings
->op
[stage
].cop
, settings
->op
[stage
].carg0
,
9861 settings
->op
[stage
].carg1
, settings
->op
[stage
].carg2
);
9865 shader_glsl_ffp_fragment_op(buffer
, stage
, TRUE
, TRUE
, settings
->op
[stage
].tmp_dst
,
9866 settings
->op
[stage
].cop
, settings
->op
[stage
].carg0
,
9867 settings
->op
[stage
].carg1
, settings
->op
[stage
].carg2
);
9869 else if (settings
->op
[stage
].cop
!= WINED3D_TOP_BUMPENVMAP
9870 && settings
->op
[stage
].cop
!= WINED3D_TOP_BUMPENVMAP_LUMINANCE
)
9872 shader_glsl_ffp_fragment_op(buffer
, stage
, TRUE
, FALSE
, settings
->op
[stage
].tmp_dst
,
9873 settings
->op
[stage
].cop
, settings
->op
[stage
].carg0
,
9874 settings
->op
[stage
].carg1
, settings
->op
[stage
].carg2
);
9875 shader_glsl_ffp_fragment_op(buffer
, stage
, FALSE
, TRUE
, settings
->op
[stage
].tmp_dst
,
9876 settings
->op
[stage
].aop
, settings
->op
[stage
].aarg0
,
9877 settings
->op
[stage
].aarg1
, settings
->op
[stage
].aarg2
);
9881 shader_addline(buffer
, "%s[0] = ffp_varying_specular * specular_enable + ret;\n",
9882 get_fragment_output(gl_info
));
9884 if (settings
->sRGB_write
)
9885 shader_glsl_generate_srgb_write_correction(buffer
, gl_info
);
9887 shader_glsl_generate_fog_code(buffer
, gl_info
, settings
->fog
);
9889 shader_glsl_generate_alpha_test(buffer
, gl_info
, alpha_test_func
);
9890 if (!use_legacy_fragment_output(gl_info
))
9891 shader_addline(buffer
, "color_out0 = ps_out[0];\n");
9893 shader_addline(buffer
, "}\n");
9895 shader_id
= GL_EXTCALL(glCreateShader(GL_FRAGMENT_SHADER
));
9896 shader_glsl_compile(gl_info
, shader_id
, buffer
->buffer
);
9898 string_buffer_release(&priv
->string_buffers
, tex_reg_name
);
9902 static struct glsl_ffp_vertex_shader
*shader_glsl_find_ffp_vertex_shader(struct shader_glsl_priv
*priv
,
9903 const struct wined3d_gl_info
*gl_info
, const struct wined3d_ffp_vs_settings
*settings
)
9905 struct glsl_ffp_vertex_shader
*shader
;
9906 const struct wine_rb_entry
*entry
;
9908 if ((entry
= wine_rb_get(&priv
->ffp_vertex_shaders
, settings
)))
9909 return WINE_RB_ENTRY_VALUE(entry
, struct glsl_ffp_vertex_shader
, desc
.entry
);
9911 if (!(shader
= heap_alloc(sizeof(*shader
))))
9914 shader
->desc
.settings
= *settings
;
9915 shader
->id
= shader_glsl_generate_ffp_vertex_shader(priv
, settings
, gl_info
);
9916 list_init(&shader
->linked_programs
);
9917 if (wine_rb_put(&priv
->ffp_vertex_shaders
, &shader
->desc
.settings
, &shader
->desc
.entry
) == -1)
9918 ERR("Failed to insert ffp vertex shader.\n");
9923 static struct glsl_ffp_fragment_shader
*shader_glsl_find_ffp_fragment_shader(struct shader_glsl_priv
*priv
,
9924 const struct ffp_frag_settings
*args
, const struct wined3d_context_gl
*context_gl
)
9926 struct glsl_ffp_fragment_shader
*glsl_desc
;
9927 const struct ffp_frag_desc
*desc
;
9929 if ((desc
= find_ffp_frag_shader(&priv
->ffp_fragment_shaders
, args
)))
9930 return CONTAINING_RECORD(desc
, struct glsl_ffp_fragment_shader
, entry
);
9932 if (!(glsl_desc
= heap_alloc(sizeof(*glsl_desc
))))
9935 glsl_desc
->entry
.settings
= *args
;
9936 glsl_desc
->id
= shader_glsl_generate_ffp_fragment_shader(priv
, args
, context_gl
);
9937 list_init(&glsl_desc
->linked_programs
);
9938 add_ffp_frag_shader(&priv
->ffp_fragment_shaders
, &glsl_desc
->entry
);
9944 static void shader_glsl_init_vs_uniform_locations(const struct wined3d_gl_info
*gl_info
,
9945 struct shader_glsl_priv
*priv
, GLuint program_id
, struct glsl_vs_program
*vs
, unsigned int vs_c_count
)
9948 struct wined3d_string_buffer
*name
= string_buffer_get(&priv
->string_buffers
);
9950 for (i
= 0; i
< vs_c_count
; ++i
)
9952 string_buffer_sprintf(name
, "vs_c[%u]", i
);
9953 vs
->uniform_f_locations
[i
] = GL_EXTCALL(glGetUniformLocation(program_id
, name
->buffer
));
9955 memset(&vs
->uniform_f_locations
[vs_c_count
], 0xff, (WINED3D_MAX_VS_CONSTS_F
- vs_c_count
) * sizeof(GLuint
));
9957 for (i
= 0; i
< WINED3D_MAX_CONSTS_I
; ++i
)
9959 string_buffer_sprintf(name
, "vs_i[%u]", i
);
9960 vs
->uniform_i_locations
[i
] = GL_EXTCALL(glGetUniformLocation(program_id
, name
->buffer
));
9963 for (i
= 0; i
< WINED3D_MAX_CONSTS_B
; ++i
)
9965 string_buffer_sprintf(name
, "vs_b[%u]", i
);
9966 vs
->uniform_b_locations
[i
] = GL_EXTCALL(glGetUniformLocation(program_id
, name
->buffer
));
9969 vs
->pos_fixup_location
= GL_EXTCALL(glGetUniformLocation(program_id
, "pos_fixup"));
9970 vs
->base_vertex_id_location
= GL_EXTCALL(glGetUniformLocation(program_id
, "base_vertex_id"));
9972 for (i
= 0; i
< MAX_VERTEX_BLENDS
; ++i
)
9974 string_buffer_sprintf(name
, "ffp_modelview_matrix[%u]", i
);
9975 vs
->modelview_matrix_location
[i
] = GL_EXTCALL(glGetUniformLocation(program_id
, name
->buffer
));
9977 vs
->projection_matrix_location
= GL_EXTCALL(glGetUniformLocation(program_id
, "ffp_projection_matrix"));
9978 vs
->normal_matrix_location
= GL_EXTCALL(glGetUniformLocation(program_id
, "ffp_normal_matrix"));
9979 for (i
= 0; i
< WINED3D_MAX_TEXTURES
; ++i
)
9981 string_buffer_sprintf(name
, "ffp_texture_matrix[%u]", i
);
9982 vs
->texture_matrix_location
[i
] = GL_EXTCALL(glGetUniformLocation(program_id
, name
->buffer
));
9984 vs
->material_ambient_location
= GL_EXTCALL(glGetUniformLocation(program_id
, "ffp_material.ambient"));
9985 vs
->material_diffuse_location
= GL_EXTCALL(glGetUniformLocation(program_id
, "ffp_material.diffuse"));
9986 vs
->material_specular_location
= GL_EXTCALL(glGetUniformLocation(program_id
, "ffp_material.specular"));
9987 vs
->material_emissive_location
= GL_EXTCALL(glGetUniformLocation(program_id
, "ffp_material.emissive"));
9988 vs
->material_shininess_location
= GL_EXTCALL(glGetUniformLocation(program_id
, "ffp_material.shininess"));
9989 vs
->light_ambient_location
= GL_EXTCALL(glGetUniformLocation(program_id
, "ffp_light_ambient"));
9990 for (i
= 0; i
< WINED3D_MAX_ACTIVE_LIGHTS
; ++i
)
9992 string_buffer_sprintf(name
, "ffp_light[%u].diffuse", i
);
9993 vs
->light_location
[i
].diffuse
= GL_EXTCALL(glGetUniformLocation(program_id
, name
->buffer
));
9994 string_buffer_sprintf(name
, "ffp_light[%u].specular", i
);
9995 vs
->light_location
[i
].specular
= GL_EXTCALL(glGetUniformLocation(program_id
, name
->buffer
));
9996 string_buffer_sprintf(name
, "ffp_light[%u].ambient", i
);
9997 vs
->light_location
[i
].ambient
= GL_EXTCALL(glGetUniformLocation(program_id
, name
->buffer
));
9998 string_buffer_sprintf(name
, "ffp_light[%u].position", i
);
9999 vs
->light_location
[i
].position
= GL_EXTCALL(glGetUniformLocation(program_id
, name
->buffer
));
10000 string_buffer_sprintf(name
, "ffp_light[%u].direction", i
);
10001 vs
->light_location
[i
].direction
= GL_EXTCALL(glGetUniformLocation(program_id
, name
->buffer
));
10002 string_buffer_sprintf(name
, "ffp_light[%u].range", i
);
10003 vs
->light_location
[i
].range
= GL_EXTCALL(glGetUniformLocation(program_id
, name
->buffer
));
10004 string_buffer_sprintf(name
, "ffp_light[%u].falloff", i
);
10005 vs
->light_location
[i
].falloff
= GL_EXTCALL(glGetUniformLocation(program_id
, name
->buffer
));
10006 string_buffer_sprintf(name
, "ffp_light[%u].c_att", i
);
10007 vs
->light_location
[i
].c_att
= GL_EXTCALL(glGetUniformLocation(program_id
, name
->buffer
));
10008 string_buffer_sprintf(name
, "ffp_light[%u].l_att", i
);
10009 vs
->light_location
[i
].l_att
= GL_EXTCALL(glGetUniformLocation(program_id
, name
->buffer
));
10010 string_buffer_sprintf(name
, "ffp_light[%u].q_att", i
);
10011 vs
->light_location
[i
].q_att
= GL_EXTCALL(glGetUniformLocation(program_id
, name
->buffer
));
10012 string_buffer_sprintf(name
, "ffp_light[%u].cos_htheta", i
);
10013 vs
->light_location
[i
].cos_htheta
= GL_EXTCALL(glGetUniformLocation(program_id
, name
->buffer
));
10014 string_buffer_sprintf(name
, "ffp_light[%u].cos_hphi", i
);
10015 vs
->light_location
[i
].cos_hphi
= GL_EXTCALL(glGetUniformLocation(program_id
, name
->buffer
));
10017 vs
->pointsize_location
= GL_EXTCALL(glGetUniformLocation(program_id
, "ffp_point.size"));
10018 vs
->pointsize_min_location
= GL_EXTCALL(glGetUniformLocation(program_id
, "ffp_point.size_min"));
10019 vs
->pointsize_max_location
= GL_EXTCALL(glGetUniformLocation(program_id
, "ffp_point.size_max"));
10020 vs
->pointsize_c_att_location
= GL_EXTCALL(glGetUniformLocation(program_id
, "ffp_point.c_att"));
10021 vs
->pointsize_l_att_location
= GL_EXTCALL(glGetUniformLocation(program_id
, "ffp_point.l_att"));
10022 vs
->pointsize_q_att_location
= GL_EXTCALL(glGetUniformLocation(program_id
, "ffp_point.q_att"));
10023 vs
->clip_planes_location
= GL_EXTCALL(glGetUniformLocation(program_id
, "clip_planes"));
10025 string_buffer_release(&priv
->string_buffers
, name
);
10028 static void shader_glsl_init_ds_uniform_locations(const struct wined3d_gl_info
*gl_info
,
10029 struct shader_glsl_priv
*priv
, GLuint program_id
, struct glsl_ds_program
*ds
)
10031 ds
->pos_fixup_location
= GL_EXTCALL(glGetUniformLocation(program_id
, "pos_fixup"));
10034 static void shader_glsl_init_gs_uniform_locations(const struct wined3d_gl_info
*gl_info
,
10035 struct shader_glsl_priv
*priv
, GLuint program_id
, struct glsl_gs_program
*gs
)
10037 gs
->pos_fixup_location
= GL_EXTCALL(glGetUniformLocation(program_id
, "pos_fixup"));
10040 static void shader_glsl_init_ps_uniform_locations(const struct wined3d_gl_info
*gl_info
,
10041 struct shader_glsl_priv
*priv
, GLuint program_id
, struct glsl_ps_program
*ps
, unsigned int ps_c_count
)
10044 struct wined3d_string_buffer
*name
= string_buffer_get(&priv
->string_buffers
);
10046 for (i
= 0; i
< ps_c_count
; ++i
)
10048 string_buffer_sprintf(name
, "ps_c[%u]", i
);
10049 ps
->uniform_f_locations
[i
] = GL_EXTCALL(glGetUniformLocation(program_id
, name
->buffer
));
10051 memset(&ps
->uniform_f_locations
[ps_c_count
], 0xff, (WINED3D_MAX_PS_CONSTS_F
- ps_c_count
) * sizeof(GLuint
));
10053 for (i
= 0; i
< WINED3D_MAX_CONSTS_I
; ++i
)
10055 string_buffer_sprintf(name
, "ps_i[%u]", i
);
10056 ps
->uniform_i_locations
[i
] = GL_EXTCALL(glGetUniformLocation(program_id
, name
->buffer
));
10059 for (i
= 0; i
< WINED3D_MAX_CONSTS_B
; ++i
)
10061 string_buffer_sprintf(name
, "ps_b[%u]", i
);
10062 ps
->uniform_b_locations
[i
] = GL_EXTCALL(glGetUniformLocation(program_id
, name
->buffer
));
10065 for (i
= 0; i
< WINED3D_MAX_TEXTURES
; ++i
)
10067 string_buffer_sprintf(name
, "bumpenv_mat%u", i
);
10068 ps
->bumpenv_mat_location
[i
] = GL_EXTCALL(glGetUniformLocation(program_id
, name
->buffer
));
10069 string_buffer_sprintf(name
, "bumpenv_lum_scale%u", i
);
10070 ps
->bumpenv_lum_scale_location
[i
] = GL_EXTCALL(glGetUniformLocation(program_id
, name
->buffer
));
10071 string_buffer_sprintf(name
, "bumpenv_lum_offset%u", i
);
10072 ps
->bumpenv_lum_offset_location
[i
] = GL_EXTCALL(glGetUniformLocation(program_id
, name
->buffer
));
10073 string_buffer_sprintf(name
, "tss_const%u", i
);
10074 ps
->tss_constant_location
[i
] = GL_EXTCALL(glGetUniformLocation(program_id
, name
->buffer
));
10077 ps
->tex_factor_location
= GL_EXTCALL(glGetUniformLocation(program_id
, "tex_factor"));
10078 ps
->specular_enable_location
= GL_EXTCALL(glGetUniformLocation(program_id
, "specular_enable"));
10080 ps
->fog_color_location
= GL_EXTCALL(glGetUniformLocation(program_id
, "ffp_fog.color"));
10081 ps
->fog_density_location
= GL_EXTCALL(glGetUniformLocation(program_id
, "ffp_fog.density"));
10082 ps
->fog_end_location
= GL_EXTCALL(glGetUniformLocation(program_id
, "ffp_fog.end"));
10083 ps
->fog_scale_location
= GL_EXTCALL(glGetUniformLocation(program_id
, "ffp_fog.scale"));
10085 ps
->alpha_test_ref_location
= GL_EXTCALL(glGetUniformLocation(program_id
, "alpha_test_ref"));
10087 ps
->np2_fixup_location
= GL_EXTCALL(glGetUniformLocation(program_id
, "ps_samplerNP2Fixup"));
10088 ps
->ycorrection_location
= GL_EXTCALL(glGetUniformLocation(program_id
, "ycorrection"));
10089 ps
->color_key_location
= GL_EXTCALL(glGetUniformLocation(program_id
, "color_key"));
10091 string_buffer_release(&priv
->string_buffers
, name
);
10094 static HRESULT
shader_glsl_compile_compute_shader(struct shader_glsl_priv
*priv
,
10095 const struct wined3d_context_gl
*context_gl
, struct wined3d_shader
*shader
)
10097 struct glsl_context_data
*ctx_data
= context_gl
->c
.shader_backend_data
;
10098 const struct wined3d_gl_info
*gl_info
= context_gl
->gl_info
;
10099 struct wined3d_string_buffer
*buffer
= &priv
->shader_buffer
;
10100 struct glsl_cs_compiled_shader
*gl_shaders
;
10101 struct glsl_shader_private
*shader_data
;
10102 struct glsl_shader_prog_link
*entry
;
10103 GLuint shader_id
, program_id
;
10105 if (!(entry
= heap_alloc(sizeof(*entry
))))
10107 ERR("Out of memory.\n");
10108 return E_OUTOFMEMORY
;
10111 if (!(shader
->backend_data
= heap_alloc_zero(sizeof(*shader_data
))))
10113 ERR("Failed to allocate backend data.\n");
10115 return E_OUTOFMEMORY
;
10117 shader_data
= shader
->backend_data
;
10119 if (!(shader_data
->gl_shaders
.cs
= heap_alloc(sizeof(*gl_shaders
))))
10121 ERR("Failed to allocate GL shader array.\n");
10123 heap_free(shader
->backend_data
);
10124 shader
->backend_data
= NULL
;
10125 return E_OUTOFMEMORY
;
10127 shader_data
->shader_array_size
= 1;
10128 gl_shaders
= shader_data
->gl_shaders
.cs
;
10130 TRACE("Compiling compute shader %p.\n", shader
);
10132 string_buffer_clear(buffer
);
10133 shader_id
= shader_glsl_generate_compute_shader(context_gl
, buffer
, &priv
->string_buffers
, shader
);
10134 gl_shaders
[shader_data
->num_gl_shaders
++].id
= shader_id
;
10136 program_id
= GL_EXTCALL(glCreateProgram());
10137 TRACE("Created new GLSL shader program %u.\n", program_id
);
10139 entry
->id
= program_id
;
10145 entry
->cs
.id
= shader_id
;
10146 entry
->constant_version
= 0;
10147 entry
->shader_controlled_clip_distances
= 0;
10148 entry
->ps
.np2_fixup_info
= NULL
;
10149 add_glsl_program_entry(priv
, entry
);
10151 TRACE("Attaching GLSL shader object %u to program %u.\n", shader_id
, program_id
);
10152 GL_EXTCALL(glAttachShader(program_id
, shader_id
));
10153 checkGLcall("glAttachShader");
10155 list_add_head(&shader
->linked_programs
, &entry
->cs
.shader_entry
);
10157 TRACE("Linking GLSL shader program %u.\n", program_id
);
10158 GL_EXTCALL(glLinkProgram(program_id
));
10159 shader_glsl_validate_link(gl_info
, program_id
);
10161 GL_EXTCALL(glUseProgram(program_id
));
10162 checkGLcall("glUseProgram");
10163 shader_glsl_load_program_resources(context_gl
, priv
, program_id
, shader
);
10164 shader_glsl_load_images(gl_info
, priv
, program_id
, &shader
->reg_maps
);
10166 entry
->constant_update_mask
= 0;
10168 GL_EXTCALL(glUseProgram(ctx_data
->glsl_program
? ctx_data
->glsl_program
->id
: 0));
10169 checkGLcall("glUseProgram");
10173 static GLuint
find_glsl_compute_shader(const struct wined3d_context_gl
*context_gl
,
10174 struct shader_glsl_priv
*priv
, struct wined3d_shader
*shader
)
10176 struct glsl_shader_private
*shader_data
;
10178 if (!shader
->backend_data
)
10180 WARN("Failed to find GLSL program for compute shader %p.\n", shader
);
10181 if (FAILED(shader_glsl_compile_compute_shader(priv
, context_gl
, shader
)))
10183 ERR("Failed to compile compute shader %p.\n", shader
);
10187 shader_data
= shader
->backend_data
;
10188 return shader_data
->gl_shaders
.cs
[0].id
;
10191 /* Context activation is done by the caller. */
10192 static void set_glsl_compute_shader_program(const struct wined3d_context_gl
*context_gl
,
10193 const struct wined3d_state
*state
, struct shader_glsl_priv
*priv
, struct glsl_context_data
*ctx_data
)
10195 struct glsl_shader_prog_link
*entry
;
10196 struct wined3d_shader
*shader
;
10197 struct glsl_program_key key
;
10200 if (!(context_gl
->c
.shader_update_mask
& (1u << WINED3D_SHADER_TYPE_COMPUTE
)))
10203 if (!(shader
= state
->shader
[WINED3D_SHADER_TYPE_COMPUTE
]))
10205 WARN("Compute shader is NULL.\n");
10206 ctx_data
->glsl_program
= NULL
;
10210 cs_id
= find_glsl_compute_shader(context_gl
, priv
, shader
);
10211 memset(&key
, 0, sizeof(key
));
10213 if (!(entry
= get_glsl_program_entry(priv
, &key
)))
10214 ERR("Failed to find GLSL program for compute shader %p.\n", shader
);
10215 ctx_data
->glsl_program
= entry
;
10218 /* Context activation is done by the caller. */
10219 static void set_glsl_shader_program(const struct wined3d_context_gl
*context_gl
, const struct wined3d_state
*state
,
10220 struct shader_glsl_priv
*priv
, struct glsl_context_data
*ctx_data
)
10222 const struct wined3d_d3d_info
*d3d_info
= context_gl
->c
.d3d_info
;
10223 const struct wined3d_gl_info
*gl_info
= context_gl
->gl_info
;
10224 const struct wined3d_shader
*pre_rasterization_shader
;
10225 const struct ps_np2fixup_info
*np2fixup_info
= NULL
;
10226 struct wined3d_shader
*hshader
, *dshader
, *gshader
;
10227 struct glsl_shader_prog_link
*entry
= NULL
;
10228 struct wined3d_shader
*vshader
= NULL
;
10229 struct wined3d_shader
*pshader
= NULL
;
10230 GLuint reorder_shader_id
= 0;
10231 struct glsl_program_key key
;
10232 uint32_t attribs_map
;
10240 struct list
*ps_list
, *vs_list
;
10241 struct wined3d_string_buffer
*tmp_name
;
10243 if (!(context_gl
->c
.shader_update_mask
& (1u << WINED3D_SHADER_TYPE_VERTEX
)) && ctx_data
->glsl_program
)
10245 vs_id
= ctx_data
->glsl_program
->vs
.id
;
10246 vs_list
= &ctx_data
->glsl_program
->vs
.shader_entry
;
10249 vshader
= state
->shader
[WINED3D_SHADER_TYPE_VERTEX
];
10251 else if (use_vs(state
))
10253 struct vs_compile_args vs_compile_args
;
10255 vshader
= state
->shader
[WINED3D_SHADER_TYPE_VERTEX
];
10257 find_vs_compile_args(state
, vshader
, &vs_compile_args
, &context_gl
->c
);
10258 vs_id
= find_glsl_vertex_shader(context_gl
, priv
, vshader
, &vs_compile_args
);
10259 vs_list
= &vshader
->linked_programs
;
10261 else if (priv
->vertex_pipe
== &glsl_vertex_pipe
)
10263 struct glsl_ffp_vertex_shader
*ffp_shader
;
10264 struct wined3d_ffp_vs_settings settings
;
10266 wined3d_ffp_get_vs_settings(&context_gl
->c
, state
, &settings
);
10267 ffp_shader
= shader_glsl_find_ffp_vertex_shader(priv
, gl_info
, &settings
);
10268 vs_id
= ffp_shader
->id
;
10269 vs_list
= &ffp_shader
->linked_programs
;
10272 hshader
= state
->shader
[WINED3D_SHADER_TYPE_HULL
];
10273 if (!(context_gl
->c
.shader_update_mask
& (1u << WINED3D_SHADER_TYPE_HULL
)) && ctx_data
->glsl_program
)
10274 hs_id
= ctx_data
->glsl_program
->hs
.id
;
10276 hs_id
= find_glsl_hull_shader(context_gl
, priv
, hshader
);
10278 dshader
= state
->shader
[WINED3D_SHADER_TYPE_DOMAIN
];
10279 if (!(context_gl
->c
.shader_update_mask
& (1u << WINED3D_SHADER_TYPE_DOMAIN
)) && ctx_data
->glsl_program
)
10281 ds_id
= ctx_data
->glsl_program
->ds
.id
;
10285 struct ds_compile_args args
;
10287 find_ds_compile_args(state
, dshader
, &args
, &context_gl
->c
);
10288 ds_id
= find_glsl_domain_shader(context_gl
, priv
, dshader
, &args
);
10291 gshader
= state
->shader
[WINED3D_SHADER_TYPE_GEOMETRY
];
10292 if (!(context_gl
->c
.shader_update_mask
& (1u << WINED3D_SHADER_TYPE_GEOMETRY
)) && ctx_data
->glsl_program
)
10294 gs_id
= ctx_data
->glsl_program
->gs
.id
;
10298 struct gs_compile_args args
;
10300 find_gs_compile_args(state
, gshader
, &args
, &context_gl
->c
);
10301 gs_id
= find_glsl_geometry_shader(context_gl
, priv
, gshader
, &args
);
10304 /* A pixel shader is not used when rasterization is disabled. */
10305 if (is_rasterization_disabled(gshader
))
10310 else if (!(context_gl
->c
.shader_update_mask
& (1u << WINED3D_SHADER_TYPE_PIXEL
)) && ctx_data
->glsl_program
)
10312 ps_id
= ctx_data
->glsl_program
->ps
.id
;
10313 ps_list
= &ctx_data
->glsl_program
->ps
.shader_entry
;
10316 pshader
= state
->shader
[WINED3D_SHADER_TYPE_PIXEL
];
10318 else if (use_ps(state
))
10320 struct ps_compile_args ps_compile_args
;
10321 pshader
= state
->shader
[WINED3D_SHADER_TYPE_PIXEL
];
10322 find_ps_compile_args(state
, pshader
, context_gl
->c
.stream_info
.position_transformed
,
10323 &ps_compile_args
, &context_gl
->c
);
10324 ps_id
= find_glsl_fragment_shader(context_gl
, &priv
->shader_buffer
, &priv
->string_buffers
,
10325 pshader
, &ps_compile_args
, &np2fixup_info
);
10326 ps_list
= &pshader
->linked_programs
;
10328 else if (priv
->fragment_pipe
== &glsl_fragment_pipe
10329 && !(vshader
&& vshader
->reg_maps
.shader_version
.major
>= 4))
10331 struct glsl_ffp_fragment_shader
*ffp_shader
;
10332 struct ffp_frag_settings settings
;
10334 wined3d_ffp_get_fs_settings(&context_gl
->c
, state
, &settings
, FALSE
);
10335 ffp_shader
= shader_glsl_find_ffp_fragment_shader(priv
, &settings
, context_gl
);
10336 ps_id
= ffp_shader
->id
;
10337 ps_list
= &ffp_shader
->linked_programs
;
10346 if ((!vs_id
&& !hs_id
&& !ds_id
&& !gs_id
&& !ps_id
) || (entry
= get_glsl_program_entry(priv
, &key
)))
10348 ctx_data
->glsl_program
= entry
;
10352 /* If we get to this point, then no matching program exists, so we create one */
10353 program_id
= GL_EXTCALL(glCreateProgram());
10354 TRACE("Created new GLSL shader program %u.\n", program_id
);
10356 /* Create the entry */
10357 entry
= heap_alloc(sizeof(*entry
));
10358 entry
->id
= program_id
;
10359 entry
->vs
.id
= vs_id
;
10360 entry
->hs
.id
= hs_id
;
10361 entry
->ds
.id
= ds_id
;
10362 entry
->gs
.id
= gs_id
;
10363 entry
->ps
.id
= ps_id
;
10365 entry
->constant_version
= 0;
10366 entry
->shader_controlled_clip_distances
= 0;
10367 entry
->ps
.np2_fixup_info
= np2fixup_info
;
10368 /* Add the hash table entry */
10369 add_glsl_program_entry(priv
, entry
);
10371 /* Set the current program */
10372 ctx_data
->glsl_program
= entry
;
10374 /* Attach GLSL vshader */
10377 TRACE("Attaching GLSL shader object %u to program %u.\n", vs_id
, program_id
);
10378 GL_EXTCALL(glAttachShader(program_id
, vs_id
));
10379 checkGLcall("glAttachShader");
10381 list_add_head(vs_list
, &entry
->vs
.shader_entry
);
10386 attribs_map
= vshader
->reg_maps
.input_registers
;
10387 if (vshader
->reg_maps
.shader_version
.major
< 4)
10389 reorder_shader_id
= shader_glsl_generate_vs3_rasterizer_input_setup(priv
, vshader
, pshader
,
10390 state
->primitive_type
== WINED3D_PT_POINTLIST
&& vshader
->reg_maps
.point_size
,
10391 d3d_info
->emulated_flatshading
10392 && state
->render_states
[WINED3D_RS_SHADEMODE
] == WINED3D_SHADE_FLAT
, gl_info
);
10393 TRACE("Attaching GLSL shader object %u to program %u.\n", reorder_shader_id
, program_id
);
10394 GL_EXTCALL(glAttachShader(program_id
, reorder_shader_id
));
10395 checkGLcall("glAttachShader");
10396 /* Flag the reorder function for deletion, it will be freed
10397 * automatically when the program is destroyed. */
10398 GL_EXTCALL(glDeleteShader(reorder_shader_id
));
10403 attribs_map
= (1u << WINED3D_FFP_ATTRIBS_COUNT
) - 1;
10406 if (!shader_glsl_use_explicit_attrib_location(gl_info
))
10408 /* Bind vertex attributes to a corresponding index number to match
10409 * the same index numbers as ARB_vertex_programs (makes loading
10410 * vertex attributes simpler). With this method, we can use the
10411 * exact same code to load the attributes later for both ARB and
10414 * We have to do this here because we need to know the Program ID
10415 * in order to make the bindings work, and it has to be done prior
10416 * to linking the GLSL program. */
10417 tmp_name
= string_buffer_get(&priv
->string_buffers
);
10418 while (attribs_map
)
10420 i
= wined3d_bit_scan(&attribs_map
);
10421 string_buffer_sprintf(tmp_name
, "vs_in%u", i
);
10422 GL_EXTCALL(glBindAttribLocation(program_id
, i
, tmp_name
->buffer
));
10423 if (vshader
&& vshader
->reg_maps
.shader_version
.major
>= 4)
10425 string_buffer_sprintf(tmp_name
, "vs_in_uint%u", i
);
10426 GL_EXTCALL(glBindAttribLocation(program_id
, i
, tmp_name
->buffer
));
10427 string_buffer_sprintf(tmp_name
, "vs_in_int%u", i
);
10428 GL_EXTCALL(glBindAttribLocation(program_id
, i
, tmp_name
->buffer
));
10431 checkGLcall("glBindAttribLocation");
10433 if (!use_legacy_fragment_output(gl_info
))
10435 for (i
= 0; i
< WINED3D_MAX_RENDER_TARGETS
; ++i
)
10437 string_buffer_sprintf(tmp_name
, "color_out%u", i
);
10438 if (state
->blend_state
&& state
->blend_state
->dual_source
)
10439 GL_EXTCALL(glBindFragDataLocationIndexed(program_id
, 0, i
, tmp_name
->buffer
));
10441 GL_EXTCALL(glBindFragDataLocation(program_id
, i
, tmp_name
->buffer
));
10442 checkGLcall("glBindFragDataLocation");
10445 string_buffer_release(&priv
->string_buffers
, tmp_name
);
10450 TRACE("Attaching GLSL tessellation control shader object %u to program %u.\n", hs_id
, program_id
);
10451 GL_EXTCALL(glAttachShader(program_id
, hs_id
));
10452 checkGLcall("glAttachShader");
10454 list_add_head(&hshader
->linked_programs
, &entry
->hs
.shader_entry
);
10459 TRACE("Attaching GLSL tessellation evaluation shader object %u to program %u.\n", ds_id
, program_id
);
10460 GL_EXTCALL(glAttachShader(program_id
, ds_id
));
10461 checkGLcall("glAttachShader");
10463 list_add_head(&dshader
->linked_programs
, &entry
->ds
.shader_entry
);
10468 TRACE("Attaching GLSL geometry shader object %u to program %u.\n", gs_id
, program_id
);
10469 GL_EXTCALL(glAttachShader(program_id
, gs_id
));
10470 checkGLcall("glAttachShader");
10472 shader_glsl_init_transform_feedback(context_gl
, priv
, program_id
, gshader
);
10474 list_add_head(&gshader
->linked_programs
, &entry
->gs
.shader_entry
);
10477 /* Attach GLSL pshader */
10480 TRACE("Attaching GLSL shader object %u to program %u.\n", ps_id
, program_id
);
10481 GL_EXTCALL(glAttachShader(program_id
, ps_id
));
10482 checkGLcall("glAttachShader");
10484 list_add_head(ps_list
, &entry
->ps
.shader_entry
);
10487 /* Link the program */
10488 TRACE("Linking GLSL shader program %u.\n", program_id
);
10489 GL_EXTCALL(glLinkProgram(program_id
));
10490 shader_glsl_validate_link(gl_info
, program_id
);
10492 shader_glsl_init_vs_uniform_locations(gl_info
, priv
, program_id
, &entry
->vs
,
10493 vshader
? vshader
->limits
->constant_float
: 0);
10494 shader_glsl_init_ds_uniform_locations(gl_info
, priv
, program_id
, &entry
->ds
);
10495 shader_glsl_init_gs_uniform_locations(gl_info
, priv
, program_id
, &entry
->gs
);
10496 shader_glsl_init_ps_uniform_locations(gl_info
, priv
, program_id
, &entry
->ps
,
10497 pshader
? pshader
->limits
->constant_float
: 0);
10498 checkGLcall("find glsl program uniform locations");
10500 pre_rasterization_shader
= gshader
? gshader
: dshader
? dshader
: vshader
;
10501 if (pre_rasterization_shader
&& pre_rasterization_shader
->reg_maps
.shader_version
.major
>= 4)
10503 unsigned int clip_distance_count
= wined3d_popcount(pre_rasterization_shader
->reg_maps
.clip_distance_mask
);
10504 entry
->shader_controlled_clip_distances
= 1;
10505 entry
->clip_distance_mask
= wined3d_mask_from_size(clip_distance_count
);
10508 if (needs_legacy_glsl_syntax(gl_info
))
10510 if (pshader
&& pshader
->reg_maps
.shader_version
.major
>= 3
10511 && pshader
->u
.ps
.declared_in_count
> vec4_varyings(3, gl_info
))
10513 TRACE("Shader %d needs vertex color clamping disabled.\n", program_id
);
10514 entry
->vs
.vertex_color_clamp
= GL_FALSE
;
10518 entry
->vs
.vertex_color_clamp
= GL_FIXED_ONLY_ARB
;
10523 /* With core profile we never change vertex_color_clamp from
10524 * GL_FIXED_ONLY_MODE (which is also the initial value) so we never call
10525 * glClampColorARB(). */
10526 entry
->vs
.vertex_color_clamp
= GL_FIXED_ONLY_ARB
;
10529 /* Set the shader to allow uniform loading on it */
10530 GL_EXTCALL(glUseProgram(program_id
));
10531 checkGLcall("glUseProgram");
10533 entry
->constant_update_mask
= 0;
10536 entry
->constant_update_mask
|= WINED3D_SHADER_CONST_VS_F
;
10537 if (vshader
->reg_maps
.integer_constants
)
10538 entry
->constant_update_mask
|= WINED3D_SHADER_CONST_VS_I
;
10539 if (vshader
->reg_maps
.boolean_constants
)
10540 entry
->constant_update_mask
|= WINED3D_SHADER_CONST_VS_B
;
10541 if (entry
->vs
.pos_fixup_location
!= -1)
10542 entry
->constant_update_mask
|= WINED3D_SHADER_CONST_POS_FIXUP
;
10543 if (entry
->vs
.base_vertex_id_location
!= -1)
10544 entry
->constant_update_mask
|= WINED3D_SHADER_CONST_BASE_VERTEX_ID
;
10546 shader_glsl_load_program_resources(context_gl
, priv
, program_id
, vshader
);
10550 entry
->constant_update_mask
|= WINED3D_SHADER_CONST_FFP_MODELVIEW
10551 | WINED3D_SHADER_CONST_FFP_PROJ
;
10553 for (i
= 1; i
< MAX_VERTEX_BLENDS
; ++i
)
10555 if (entry
->vs
.modelview_matrix_location
[i
] != -1)
10557 entry
->constant_update_mask
|= WINED3D_SHADER_CONST_FFP_VERTEXBLEND
;
10562 for (i
= 0; i
< WINED3D_MAX_TEXTURES
; ++i
)
10564 if (entry
->vs
.texture_matrix_location
[i
] != -1)
10566 entry
->constant_update_mask
|= WINED3D_SHADER_CONST_FFP_TEXMATRIX
;
10570 if (entry
->vs
.material_ambient_location
!= -1 || entry
->vs
.material_diffuse_location
!= -1
10571 || entry
->vs
.material_specular_location
!= -1
10572 || entry
->vs
.material_emissive_location
!= -1
10573 || entry
->vs
.material_shininess_location
!= -1)
10574 entry
->constant_update_mask
|= WINED3D_SHADER_CONST_FFP_MATERIAL
;
10575 if (entry
->vs
.light_ambient_location
!= -1)
10576 entry
->constant_update_mask
|= WINED3D_SHADER_CONST_FFP_LIGHTS
;
10578 if (entry
->vs
.clip_planes_location
!= -1)
10579 entry
->constant_update_mask
|= WINED3D_SHADER_CONST_VS_CLIP_PLANES
;
10580 if (entry
->vs
.pointsize_min_location
!= -1)
10581 entry
->constant_update_mask
|= WINED3D_SHADER_CONST_VS_POINTSIZE
;
10584 shader_glsl_load_program_resources(context_gl
, priv
, program_id
, hshader
);
10588 if (entry
->ds
.pos_fixup_location
!= -1)
10589 entry
->constant_update_mask
|= WINED3D_SHADER_CONST_POS_FIXUP
;
10591 shader_glsl_load_program_resources(context_gl
, priv
, program_id
, dshader
);
10596 if (entry
->gs
.pos_fixup_location
!= -1)
10597 entry
->constant_update_mask
|= WINED3D_SHADER_CONST_POS_FIXUP
;
10599 shader_glsl_load_program_resources(context_gl
, priv
, program_id
, gshader
);
10606 entry
->constant_update_mask
|= WINED3D_SHADER_CONST_PS_F
;
10607 if (pshader
->reg_maps
.integer_constants
)
10608 entry
->constant_update_mask
|= WINED3D_SHADER_CONST_PS_I
;
10609 if (pshader
->reg_maps
.boolean_constants
)
10610 entry
->constant_update_mask
|= WINED3D_SHADER_CONST_PS_B
;
10611 if (entry
->ps
.ycorrection_location
!= -1)
10612 entry
->constant_update_mask
|= WINED3D_SHADER_CONST_PS_Y_CORR
;
10614 shader_glsl_load_program_resources(context_gl
, priv
, program_id
, pshader
);
10615 shader_glsl_load_images(gl_info
, priv
, program_id
, &pshader
->reg_maps
);
10619 entry
->constant_update_mask
|= WINED3D_SHADER_CONST_FFP_PS
;
10621 shader_glsl_load_samplers(&context_gl
->c
, priv
, program_id
, NULL
);
10624 for (i
= 0; i
< WINED3D_MAX_TEXTURES
; ++i
)
10626 if (entry
->ps
.bumpenv_mat_location
[i
] != -1)
10628 entry
->constant_update_mask
|= WINED3D_SHADER_CONST_PS_BUMP_ENV
;
10633 if (entry
->ps
.fog_color_location
!= -1)
10634 entry
->constant_update_mask
|= WINED3D_SHADER_CONST_PS_FOG
;
10635 if (entry
->ps
.alpha_test_ref_location
!= -1)
10636 entry
->constant_update_mask
|= WINED3D_SHADER_CONST_PS_ALPHA_TEST
;
10637 if (entry
->ps
.np2_fixup_location
!= -1)
10638 entry
->constant_update_mask
|= WINED3D_SHADER_CONST_PS_NP2_FIXUP
;
10639 if (entry
->ps
.color_key_location
!= -1)
10640 entry
->constant_update_mask
|= WINED3D_SHADER_CONST_FFP_COLOR_KEY
;
10644 static void shader_glsl_precompile(void *shader_priv
, struct wined3d_shader
*shader
)
10646 struct wined3d_device
*device
= shader
->device
;
10647 struct wined3d_context
*context
;
10649 if (shader
->reg_maps
.shader_version
.type
== WINED3D_SHADER_TYPE_COMPUTE
)
10651 context
= context_acquire(device
, NULL
, 0);
10652 shader_glsl_compile_compute_shader(shader_priv
, wined3d_context_gl(context
), shader
);
10653 context_release(context
);
10657 /* Context activation is done by the caller. */
10658 static void shader_glsl_select(void *shader_priv
, struct wined3d_context
*context
,
10659 const struct wined3d_state
*state
)
10661 struct wined3d_context_gl
*context_gl
= wined3d_context_gl(context
);
10662 struct glsl_context_data
*ctx_data
= context
->shader_backend_data
;
10663 const struct wined3d_gl_info
*gl_info
= context_gl
->gl_info
;
10664 struct shader_glsl_priv
*priv
= shader_priv
;
10665 struct glsl_shader_prog_link
*glsl_program
;
10666 GLenum current_vertex_color_clamp
;
10667 GLuint program_id
, prev_id
;
10669 priv
->vertex_pipe
->vp_enable(context
, !use_vs(state
));
10670 priv
->fragment_pipe
->fp_enable(context
, !use_ps(state
));
10672 prev_id
= ctx_data
->glsl_program
? ctx_data
->glsl_program
->id
: 0;
10673 set_glsl_shader_program(context_gl
, state
, priv
, ctx_data
);
10674 glsl_program
= ctx_data
->glsl_program
;
10678 program_id
= glsl_program
->id
;
10679 current_vertex_color_clamp
= glsl_program
->vs
.vertex_color_clamp
;
10680 if (glsl_program
->shader_controlled_clip_distances
)
10681 wined3d_context_gl_enable_clip_distances(context_gl
, glsl_program
->clip_distance_mask
);
10686 current_vertex_color_clamp
= GL_FIXED_ONLY_ARB
;
10689 if (ctx_data
->vertex_color_clamp
!= current_vertex_color_clamp
)
10691 ctx_data
->vertex_color_clamp
= current_vertex_color_clamp
;
10692 if (gl_info
->supported
[ARB_COLOR_BUFFER_FLOAT
])
10694 GL_EXTCALL(glClampColorARB(GL_CLAMP_VERTEX_COLOR_ARB
, current_vertex_color_clamp
));
10695 checkGLcall("glClampColorARB");
10699 FIXME("Vertex color clamp needs to be changed, but extension not supported.\n");
10703 TRACE("Using GLSL program %u.\n", program_id
);
10705 if (prev_id
!= program_id
)
10707 GL_EXTCALL(glUseProgram(program_id
));
10708 checkGLcall("glUseProgram");
10711 context
->constant_update_mask
|= glsl_program
->constant_update_mask
;
10714 context
->shader_update_mask
|= (1u << WINED3D_SHADER_TYPE_COMPUTE
);
10717 /* Context activation is done by the caller. */
10718 static void shader_glsl_select_compute(void *shader_priv
, struct wined3d_context
*context
,
10719 const struct wined3d_state
*state
)
10721 struct wined3d_context_gl
*context_gl
= wined3d_context_gl(context
);
10722 struct glsl_context_data
*ctx_data
= context
->shader_backend_data
;
10723 const struct wined3d_gl_info
*gl_info
= context_gl
->gl_info
;
10724 struct shader_glsl_priv
*priv
= shader_priv
;
10725 GLuint program_id
, prev_id
;
10727 prev_id
= ctx_data
->glsl_program
? ctx_data
->glsl_program
->id
: 0;
10728 set_glsl_compute_shader_program(context_gl
, state
, priv
, ctx_data
);
10729 program_id
= ctx_data
->glsl_program
? ctx_data
->glsl_program
->id
: 0;
10731 TRACE("Using GLSL program %u.\n", program_id
);
10733 if (prev_id
!= program_id
)
10735 GL_EXTCALL(glUseProgram(program_id
));
10736 checkGLcall("glUseProgram");
10739 context
->shader_update_mask
|= (1u << WINED3D_SHADER_TYPE_PIXEL
)
10740 | (1u << WINED3D_SHADER_TYPE_VERTEX
)
10741 | (1u << WINED3D_SHADER_TYPE_GEOMETRY
)
10742 | (1u << WINED3D_SHADER_TYPE_HULL
)
10743 | (1u << WINED3D_SHADER_TYPE_DOMAIN
);
10746 /* "context" is not necessarily the currently active context. */
10747 static void shader_glsl_invalidate_current_program(struct wined3d_context
*context
)
10749 struct glsl_context_data
*ctx_data
= context
->shader_backend_data
;
10751 ctx_data
->glsl_program
= NULL
;
10752 context
->shader_update_mask
= (1u << WINED3D_SHADER_TYPE_PIXEL
)
10753 | (1u << WINED3D_SHADER_TYPE_VERTEX
)
10754 | (1u << WINED3D_SHADER_TYPE_GEOMETRY
)
10755 | (1u << WINED3D_SHADER_TYPE_HULL
)
10756 | (1u << WINED3D_SHADER_TYPE_DOMAIN
)
10757 | (1u << WINED3D_SHADER_TYPE_COMPUTE
);
10760 /* Context activation is done by the caller. */
10761 static void shader_glsl_disable(void *shader_priv
, struct wined3d_context
*context
)
10763 struct wined3d_context_gl
*context_gl
= wined3d_context_gl(context
);
10764 struct glsl_context_data
*ctx_data
= context
->shader_backend_data
;
10765 const struct wined3d_gl_info
*gl_info
= context_gl
->gl_info
;
10766 struct shader_glsl_priv
*priv
= shader_priv
;
10768 shader_glsl_invalidate_current_program(context
);
10769 GL_EXTCALL(glUseProgram(0));
10770 checkGLcall("glUseProgram");
10772 priv
->vertex_pipe
->vp_enable(context
, FALSE
);
10773 priv
->fragment_pipe
->fp_enable(context
, FALSE
);
10775 if (needs_legacy_glsl_syntax(gl_info
) && gl_info
->supported
[ARB_COLOR_BUFFER_FLOAT
])
10777 ctx_data
->vertex_color_clamp
= GL_FIXED_ONLY_ARB
;
10778 GL_EXTCALL(glClampColorARB(GL_CLAMP_VERTEX_COLOR_ARB
, GL_FIXED_ONLY_ARB
));
10779 checkGLcall("glClampColorARB");
10783 static void shader_glsl_invalidate_contexts_program(struct wined3d_device
*device
,
10784 const struct glsl_shader_prog_link
*program
)
10786 const struct glsl_context_data
*ctx_data
;
10787 struct wined3d_context
*context
;
10790 for (i
= 0; i
< device
->context_count
; ++i
)
10792 context
= device
->contexts
[i
];
10793 ctx_data
= context
->shader_backend_data
;
10795 if (ctx_data
->glsl_program
== program
)
10796 shader_glsl_invalidate_current_program(context
);
10800 static void shader_glsl_destroy(struct wined3d_shader
*shader
)
10802 struct glsl_shader_private
*shader_data
= shader
->backend_data
;
10803 struct wined3d_device
*device
= shader
->device
;
10804 struct shader_glsl_priv
*priv
= device
->shader_priv
;
10805 const struct wined3d_gl_info
*gl_info
;
10806 const struct list
*linked_programs
;
10807 struct wined3d_context
*context
;
10809 if (!shader_data
|| !shader_data
->num_gl_shaders
)
10811 heap_free(shader_data
);
10812 shader
->backend_data
= NULL
;
10816 context
= context_acquire(device
, NULL
, 0);
10817 gl_info
= wined3d_context_gl(context
)->gl_info
;
10819 TRACE("Deleting linked programs.\n");
10820 linked_programs
= &shader
->linked_programs
;
10821 if (!list_empty(linked_programs
))
10823 struct glsl_shader_prog_link
*entry
, *entry2
;
10826 switch (shader
->reg_maps
.shader_version
.type
)
10828 case WINED3D_SHADER_TYPE_PIXEL
:
10830 struct glsl_ps_compiled_shader
*gl_shaders
= shader_data
->gl_shaders
.ps
;
10832 for (i
= 0; i
< shader_data
->num_gl_shaders
; ++i
)
10834 TRACE("Deleting pixel shader %u.\n", gl_shaders
[i
].id
);
10835 GL_EXTCALL(glDeleteShader(gl_shaders
[i
].id
));
10836 checkGLcall("glDeleteShader");
10838 heap_free(shader_data
->gl_shaders
.ps
);
10840 LIST_FOR_EACH_ENTRY_SAFE(entry
, entry2
, linked_programs
,
10841 struct glsl_shader_prog_link
, ps
.shader_entry
)
10843 shader_glsl_invalidate_contexts_program(device
, entry
);
10844 delete_glsl_program_entry(priv
, gl_info
, entry
);
10850 case WINED3D_SHADER_TYPE_VERTEX
:
10852 struct glsl_vs_compiled_shader
*gl_shaders
= shader_data
->gl_shaders
.vs
;
10854 for (i
= 0; i
< shader_data
->num_gl_shaders
; ++i
)
10856 TRACE("Deleting vertex shader %u.\n", gl_shaders
[i
].id
);
10857 GL_EXTCALL(glDeleteShader(gl_shaders
[i
].id
));
10858 checkGLcall("glDeleteShader");
10860 heap_free(shader_data
->gl_shaders
.vs
);
10862 LIST_FOR_EACH_ENTRY_SAFE(entry
, entry2
, linked_programs
,
10863 struct glsl_shader_prog_link
, vs
.shader_entry
)
10865 shader_glsl_invalidate_contexts_program(device
, entry
);
10866 delete_glsl_program_entry(priv
, gl_info
, entry
);
10872 case WINED3D_SHADER_TYPE_HULL
:
10874 struct glsl_hs_compiled_shader
*gl_shaders
= shader_data
->gl_shaders
.hs
;
10876 for (i
= 0; i
< shader_data
->num_gl_shaders
; ++i
)
10878 TRACE("Deleting hull shader %u.\n", gl_shaders
[i
].id
);
10879 GL_EXTCALL(glDeleteShader(gl_shaders
[i
].id
));
10880 checkGLcall("glDeleteShader");
10882 heap_free(shader_data
->gl_shaders
.hs
);
10884 LIST_FOR_EACH_ENTRY_SAFE(entry
, entry2
, linked_programs
,
10885 struct glsl_shader_prog_link
, hs
.shader_entry
)
10887 shader_glsl_invalidate_contexts_program(device
, entry
);
10888 delete_glsl_program_entry(priv
, gl_info
, entry
);
10894 case WINED3D_SHADER_TYPE_DOMAIN
:
10896 struct glsl_ds_compiled_shader
*gl_shaders
= shader_data
->gl_shaders
.ds
;
10898 for (i
= 0; i
< shader_data
->num_gl_shaders
; ++i
)
10900 TRACE("Deleting domain shader %u.\n", gl_shaders
[i
].id
);
10901 GL_EXTCALL(glDeleteShader(gl_shaders
[i
].id
));
10902 checkGLcall("glDeleteShader");
10904 heap_free(shader_data
->gl_shaders
.ds
);
10906 LIST_FOR_EACH_ENTRY_SAFE(entry
, entry2
, linked_programs
,
10907 struct glsl_shader_prog_link
, ds
.shader_entry
)
10909 shader_glsl_invalidate_contexts_program(device
, entry
);
10910 delete_glsl_program_entry(priv
, gl_info
, entry
);
10916 case WINED3D_SHADER_TYPE_GEOMETRY
:
10918 struct glsl_gs_compiled_shader
*gl_shaders
= shader_data
->gl_shaders
.gs
;
10920 for (i
= 0; i
< shader_data
->num_gl_shaders
; ++i
)
10922 TRACE("Deleting geometry shader %u.\n", gl_shaders
[i
].id
);
10923 GL_EXTCALL(glDeleteShader(gl_shaders
[i
].id
));
10924 checkGLcall("glDeleteShader");
10926 heap_free(shader_data
->gl_shaders
.gs
);
10928 LIST_FOR_EACH_ENTRY_SAFE(entry
, entry2
, linked_programs
,
10929 struct glsl_shader_prog_link
, gs
.shader_entry
)
10931 shader_glsl_invalidate_contexts_program(device
, entry
);
10932 delete_glsl_program_entry(priv
, gl_info
, entry
);
10938 case WINED3D_SHADER_TYPE_COMPUTE
:
10940 struct glsl_cs_compiled_shader
*gl_shaders
= shader_data
->gl_shaders
.cs
;
10942 for (i
= 0; i
< shader_data
->num_gl_shaders
; ++i
)
10944 TRACE("Deleting compute shader %u.\n", gl_shaders
[i
].id
);
10945 GL_EXTCALL(glDeleteShader(gl_shaders
[i
].id
));
10946 checkGLcall("glDeleteShader");
10948 heap_free(shader_data
->gl_shaders
.cs
);
10950 LIST_FOR_EACH_ENTRY_SAFE(entry
, entry2
, linked_programs
,
10951 struct glsl_shader_prog_link
, cs
.shader_entry
)
10953 shader_glsl_invalidate_contexts_program(device
, entry
);
10954 delete_glsl_program_entry(priv
, gl_info
, entry
);
10961 ERR("Unhandled shader type %#x.\n", shader
->reg_maps
.shader_version
.type
);
10966 heap_free(shader
->backend_data
);
10967 shader
->backend_data
= NULL
;
10969 context_release(context
);
10972 static int glsl_program_key_compare(const void *key
, const struct wine_rb_entry
*entry
)
10974 const struct glsl_program_key
*k
= key
;
10975 const struct glsl_shader_prog_link
*prog
= WINE_RB_ENTRY_VALUE(entry
,
10976 const struct glsl_shader_prog_link
, program_lookup_entry
);
10979 if ((ret
= wined3d_uint32_compare(k
->vs_id
, prog
->vs
.id
)))
10981 if ((ret
= wined3d_uint32_compare(k
->gs_id
, prog
->gs
.id
)))
10983 if ((ret
= wined3d_uint32_compare(k
->ps_id
, prog
->ps
.id
)))
10985 if ((ret
= wined3d_uint32_compare(k
->hs_id
, prog
->hs
.id
)))
10987 if ((ret
= wined3d_uint32_compare(k
->ds_id
, prog
->ds
.id
)))
10989 if ((ret
= wined3d_uint32_compare(k
->cs_id
, prog
->cs
.id
)))
10995 static BOOL
constant_heap_init(struct constant_heap
*heap
, unsigned int constant_count
)
10997 SIZE_T size
= (constant_count
+ 1) * sizeof(*heap
->entries
)
10998 + constant_count
* sizeof(*heap
->contained
)
10999 + constant_count
* sizeof(*heap
->positions
);
11002 if (!(mem
= heap_alloc(size
)))
11004 ERR("Failed to allocate memory\n");
11008 heap
->entries
= mem
;
11009 heap
->entries
[1].version
= 0;
11010 heap
->contained
= (BOOL
*)(heap
->entries
+ constant_count
+ 1);
11011 memset(heap
->contained
, 0, constant_count
* sizeof(*heap
->contained
));
11012 heap
->positions
= (unsigned int *)(heap
->contained
+ constant_count
);
11018 static void constant_heap_free(struct constant_heap
*heap
)
11020 heap_free(heap
->entries
);
11023 static HRESULT
shader_glsl_alloc(struct wined3d_device
*device
, const struct wined3d_vertex_pipe_ops
*vertex_pipe
,
11024 const struct wined3d_fragment_pipe_ops
*fragment_pipe
)
11026 SIZE_T stack_size
= wined3d_log2i(max(WINED3D_MAX_VS_CONSTS_F
, WINED3D_MAX_PS_CONSTS_F
)) + 1;
11027 struct fragment_caps fragment_caps
;
11028 void *vertex_priv
, *fragment_priv
;
11029 struct shader_glsl_priv
*priv
;
11031 if (!(priv
= heap_alloc_zero(sizeof(*priv
))))
11032 return E_OUTOFMEMORY
;
11034 string_buffer_list_init(&priv
->string_buffers
);
11036 if (!(vertex_priv
= vertex_pipe
->vp_alloc(&glsl_shader_backend
, priv
)))
11038 ERR("Failed to initialize vertex pipe.\n");
11043 if (!(fragment_priv
= fragment_pipe
->alloc_private(&glsl_shader_backend
, priv
)))
11045 ERR("Failed to initialize fragment pipe.\n");
11046 vertex_pipe
->vp_free(device
, NULL
);
11051 if (!string_buffer_init(&priv
->shader_buffer
))
11053 ERR("Failed to initialize shader buffer.\n");
11057 if (!(priv
->stack
= heap_calloc(stack_size
, sizeof(*priv
->stack
))))
11059 ERR("Failed to allocate memory.\n");
11063 if (!constant_heap_init(&priv
->vconst_heap
, WINED3D_MAX_VS_CONSTS_F
))
11065 ERR("Failed to initialize vertex shader constant heap\n");
11069 if (!constant_heap_init(&priv
->pconst_heap
, WINED3D_MAX_PS_CONSTS_F
))
11071 ERR("Failed to initialize pixel shader constant heap\n");
11075 wine_rb_init(&priv
->program_lookup
, glsl_program_key_compare
);
11077 priv
->next_constant_version
= 1;
11078 priv
->vertex_pipe
= vertex_pipe
;
11079 priv
->fragment_pipe
= fragment_pipe
;
11080 fragment_pipe
->get_caps(device
->adapter
, &fragment_caps
);
11081 priv
->ffp_proj_control
= fragment_caps
.wined3d_caps
& WINED3D_FRAGMENT_CAP_PROJ_CONTROL
;
11082 priv
->legacy_lighting
= device
->wined3d
->flags
& WINED3D_LEGACY_FFP_LIGHTING
;
11084 device
->vertex_priv
= vertex_priv
;
11085 device
->fragment_priv
= fragment_priv
;
11086 device
->shader_priv
= priv
;
11091 constant_heap_free(&priv
->pconst_heap
);
11092 constant_heap_free(&priv
->vconst_heap
);
11093 heap_free(priv
->stack
);
11094 string_buffer_free(&priv
->shader_buffer
);
11095 fragment_pipe
->free_private(device
, NULL
);
11096 vertex_pipe
->vp_free(device
, NULL
);
11098 return E_OUTOFMEMORY
;
11101 /* Context activation is done by the caller. */
11102 static void shader_glsl_free(struct wined3d_device
*device
, struct wined3d_context
*context
)
11104 struct shader_glsl_priv
*priv
= device
->shader_priv
;
11106 wine_rb_destroy(&priv
->program_lookup
, NULL
, NULL
);
11107 constant_heap_free(&priv
->pconst_heap
);
11108 constant_heap_free(&priv
->vconst_heap
);
11109 heap_free(priv
->stack
);
11110 string_buffer_list_cleanup(&priv
->string_buffers
);
11111 string_buffer_free(&priv
->shader_buffer
);
11112 priv
->fragment_pipe
->free_private(device
, context
);
11113 priv
->vertex_pipe
->vp_free(device
, context
);
11115 heap_free(device
->shader_priv
);
11116 device
->shader_priv
= NULL
;
11119 static BOOL
shader_glsl_allocate_context_data(struct wined3d_context
*context
)
11121 struct glsl_context_data
*ctx_data
;
11123 if (!(ctx_data
= heap_alloc_zero(sizeof(*ctx_data
))))
11125 ctx_data
->vertex_color_clamp
= GL_FIXED_ONLY_ARB
;
11126 context
->shader_backend_data
= ctx_data
;
11130 static void shader_glsl_free_context_data(struct wined3d_context
*context
)
11132 heap_free(context
->shader_backend_data
);
11135 static void shader_glsl_init_context_state(struct wined3d_context
*context
)
11137 struct wined3d_context_gl
*context_gl
= wined3d_context_gl(context
);
11138 const struct wined3d_gl_info
*gl_info
= context_gl
->gl_info
;
11140 gl_info
->gl_ops
.gl
.p_glEnable(GL_PROGRAM_POINT_SIZE
);
11141 checkGLcall("GL_PROGRAM_POINT_SIZE");
11144 static unsigned int shader_glsl_get_shader_model(const struct wined3d_gl_info
*gl_info
)
11146 BOOL shader_model_4
= gl_info
->glsl_version
>= MAKEDWORD_VERSION(1, 50)
11147 && gl_info
->supported
[ARB_SHADER_BIT_ENCODING
]
11148 && gl_info
->supported
[ARB_TEXTURE_SWIZZLE
];
11151 && gl_info
->supported
[ARB_COMPUTE_SHADER
]
11152 && gl_info
->supported
[ARB_CULL_DISTANCE
]
11153 && gl_info
->supported
[ARB_DERIVATIVE_CONTROL
]
11154 && gl_info
->supported
[ARB_GPU_SHADER5
]
11155 && gl_info
->supported
[ARB_SHADER_ATOMIC_COUNTERS
]
11156 && gl_info
->supported
[ARB_SHADER_IMAGE_LOAD_STORE
]
11157 && gl_info
->supported
[ARB_SHADER_IMAGE_SIZE
]
11158 && gl_info
->supported
[ARB_SHADING_LANGUAGE_PACKING
]
11159 && gl_info
->supported
[ARB_TESSELLATION_SHADER
]
11160 && gl_info
->supported
[ARB_TEXTURE_GATHER
]
11161 && gl_info
->supported
[ARB_TRANSFORM_FEEDBACK3
])
11164 if (shader_model_4
)
11167 /* Support for texldd and texldl instructions in pixel shaders is required
11169 if (shader_glsl_has_core_grad(gl_info
) || gl_info
->supported
[ARB_SHADER_TEXTURE_LOD
])
11175 static void shader_glsl_get_caps(const struct wined3d_adapter
*adapter
, struct shader_caps
*caps
)
11177 const struct wined3d_gl_info
*gl_info
= &adapter
->gl_info
;
11178 unsigned int shader_model
= shader_glsl_get_shader_model(gl_info
);
11180 TRACE("Shader model %u.\n", shader_model
);
11182 caps
->vs_version
= min(wined3d_settings
.max_sm_vs
, shader_model
);
11183 caps
->hs_version
= min(wined3d_settings
.max_sm_hs
, shader_model
);
11184 caps
->ds_version
= min(wined3d_settings
.max_sm_ds
, shader_model
);
11185 caps
->gs_version
= min(wined3d_settings
.max_sm_gs
, shader_model
);
11186 caps
->ps_version
= min(wined3d_settings
.max_sm_ps
, shader_model
);
11187 caps
->cs_version
= min(wined3d_settings
.max_sm_cs
, shader_model
);
11189 caps
->vs_version
= gl_info
->supported
[ARB_VERTEX_SHADER
] ? caps
->vs_version
: 0;
11190 caps
->ps_version
= gl_info
->supported
[ARB_FRAGMENT_SHADER
] ? caps
->ps_version
: 0;
11192 caps
->vs_uniform_count
= min(WINED3D_MAX_VS_CONSTS_F
, gl_info
->limits
.glsl_vs_float_constants
);
11193 caps
->ps_uniform_count
= min(WINED3D_MAX_PS_CONSTS_F
, gl_info
->limits
.glsl_ps_float_constants
);
11194 caps
->varying_count
= gl_info
->limits
.glsl_varyings
;
11196 /* FIXME: The following line is card dependent. -8.0 to 8.0 is the
11197 * Direct3D minimum requirement.
11199 * Both GL_ARB_fragment_program and GLSL require a "maximum representable magnitude"
11200 * of colors to be 2^10, and 2^32 for other floats. Should we use 1024 here?
11202 * The problem is that the refrast clamps temporary results in the shader to
11203 * [-MaxValue;+MaxValue]. If the card's max value is bigger than the one we advertize here,
11204 * then applications may miss the clamping behavior. On the other hand, if it is smaller,
11205 * the shader will generate incorrect results too. Unfortunately, GL deliberately doesn't
11206 * offer a way to query this.
11208 if (shader_model
>= 4)
11209 caps
->ps_1x_max_value
= FLT_MAX
;
11211 caps
->ps_1x_max_value
= 1024.0f
;
11213 /* Ideally we'd only set caps like sRGB writes here if supported by both
11214 * the shader backend and the fragment pipe, but we can get called before
11215 * shader_glsl_alloc(). */
11216 caps
->wined3d_caps
= WINED3D_SHADER_CAP_VS_CLIPPING
11217 | WINED3D_SHADER_CAP_SRGB_WRITE
;
11218 if (needs_interpolation_qualifiers_for_shader_outputs(gl_info
))
11219 caps
->wined3d_caps
|= WINED3D_SHADER_CAP_OUTPUT_INTERPOLATION
;
11220 if (shader_glsl_full_ffp_varyings(gl_info
))
11221 caps
->wined3d_caps
|= WINED3D_SHADER_CAP_FULL_FFP_VARYINGS
;
11224 static BOOL
shader_glsl_color_fixup_supported(struct color_fixup_desc fixup
)
11226 /* We support everything except YUV conversions. */
11227 return !is_complex_fixup(fixup
);
11230 static const SHADER_HANDLER shader_glsl_instruction_handler_table
[WINED3DSIH_TABLE_SIZE
] =
11232 /* WINED3DSIH_ABS */ shader_glsl_map2gl
,
11233 /* WINED3DSIH_ADD */ shader_glsl_binop
,
11234 /* WINED3DSIH_AND */ shader_glsl_binop
,
11235 /* WINED3DSIH_ATOMIC_AND */ shader_glsl_atomic
,
11236 /* WINED3DSIH_ATOMIC_CMP_STORE */ shader_glsl_atomic
,
11237 /* WINED3DSIH_ATOMIC_IADD */ shader_glsl_atomic
,
11238 /* WINED3DSIH_ATOMIC_IMAX */ shader_glsl_atomic
,
11239 /* WINED3DSIH_ATOMIC_IMIN */ shader_glsl_atomic
,
11240 /* WINED3DSIH_ATOMIC_OR */ shader_glsl_atomic
,
11241 /* WINED3DSIH_ATOMIC_UMAX */ shader_glsl_atomic
,
11242 /* WINED3DSIH_ATOMIC_UMIN */ shader_glsl_atomic
,
11243 /* WINED3DSIH_ATOMIC_XOR */ shader_glsl_atomic
,
11244 /* WINED3DSIH_BEM */ shader_glsl_bem
,
11245 /* WINED3DSIH_BFI */ shader_glsl_bitwise_op
,
11246 /* WINED3DSIH_BFREV */ shader_glsl_map2gl
,
11247 /* WINED3DSIH_BREAK */ shader_glsl_break
,
11248 /* WINED3DSIH_BREAKC */ shader_glsl_breakc
,
11249 /* WINED3DSIH_BREAKP */ shader_glsl_conditional_op
,
11250 /* WINED3DSIH_BUFINFO */ shader_glsl_bufinfo
,
11251 /* WINED3DSIH_CALL */ shader_glsl_call
,
11252 /* WINED3DSIH_CALLNZ */ shader_glsl_callnz
,
11253 /* WINED3DSIH_CASE */ shader_glsl_case
,
11254 /* WINED3DSIH_CMP */ shader_glsl_conditional_move
,
11255 /* WINED3DSIH_CND */ shader_glsl_cnd
,
11256 /* WINED3DSIH_CONTINUE */ shader_glsl_continue
,
11257 /* WINED3DSIH_CONTINUEP */ shader_glsl_conditional_op
,
11258 /* WINED3DSIH_COUNTBITS */ shader_glsl_map2gl
,
11259 /* WINED3DSIH_CRS */ shader_glsl_cross
,
11260 /* WINED3DSIH_CUT */ shader_glsl_cut
,
11261 /* WINED3DSIH_CUT_STREAM */ shader_glsl_cut
,
11262 /* WINED3DSIH_DCL */ shader_glsl_nop
,
11263 /* WINED3DSIH_DCL_CONSTANT_BUFFER */ shader_glsl_nop
,
11264 /* WINED3DSIH_DCL_FUNCTION_BODY */ NULL
,
11265 /* WINED3DSIH_DCL_FUNCTION_TABLE */ NULL
,
11266 /* WINED3DSIH_DCL_GLOBAL_FLAGS */ shader_glsl_nop
,
11267 /* WINED3DSIH_DCL_GS_INSTANCES */ shader_glsl_nop
,
11268 /* WINED3DSIH_DCL_HS_FORK_PHASE_INSTANCE_COUNT */ shader_glsl_nop
,
11269 /* WINED3DSIH_DCL_HS_JOIN_PHASE_INSTANCE_COUNT */ shader_glsl_nop
,
11270 /* WINED3DSIH_DCL_HS_MAX_TESSFACTOR */ shader_glsl_nop
,
11271 /* WINED3DSIH_DCL_IMMEDIATE_CONSTANT_BUFFER */ shader_glsl_nop
,
11272 /* WINED3DSIH_DCL_INDEX_RANGE */ shader_glsl_nop
,
11273 /* WINED3DSIH_DCL_INDEXABLE_TEMP */ shader_glsl_nop
,
11274 /* WINED3DSIH_DCL_INPUT */ shader_glsl_nop
,
11275 /* WINED3DSIH_DCL_INPUT_CONTROL_POINT_COUNT */ shader_glsl_nop
,
11276 /* WINED3DSIH_DCL_INPUT_PRIMITIVE */ shader_glsl_nop
,
11277 /* WINED3DSIH_DCL_INPUT_PS */ shader_glsl_nop
,
11278 /* WINED3DSIH_DCL_INPUT_PS_SGV */ shader_glsl_nop
,
11279 /* WINED3DSIH_DCL_INPUT_PS_SIV */ shader_glsl_nop
,
11280 /* WINED3DSIH_DCL_INPUT_SGV */ shader_glsl_nop
,
11281 /* WINED3DSIH_DCL_INPUT_SIV */ shader_glsl_nop
,
11282 /* WINED3DSIH_DCL_INTERFACE */ NULL
,
11283 /* WINED3DSIH_DCL_OUTPUT */ shader_glsl_nop
,
11284 /* WINED3DSIH_DCL_OUTPUT_CONTROL_POINT_COUNT */ shader_glsl_nop
,
11285 /* WINED3DSIH_DCL_OUTPUT_SIV */ shader_glsl_nop
,
11286 /* WINED3DSIH_DCL_OUTPUT_TOPOLOGY */ shader_glsl_nop
,
11287 /* WINED3DSIH_DCL_RESOURCE_RAW */ shader_glsl_nop
,
11288 /* WINED3DSIH_DCL_RESOURCE_STRUCTURED */ shader_glsl_nop
,
11289 /* WINED3DSIH_DCL_SAMPLER */ shader_glsl_nop
,
11290 /* WINED3DSIH_DCL_STREAM */ NULL
,
11291 /* WINED3DSIH_DCL_TEMPS */ shader_glsl_nop
,
11292 /* WINED3DSIH_DCL_TESSELLATOR_DOMAIN */ shader_glsl_nop
,
11293 /* WINED3DSIH_DCL_TESSELLATOR_OUTPUT_PRIMITIVE */ shader_glsl_nop
,
11294 /* WINED3DSIH_DCL_TESSELLATOR_PARTITIONING */ shader_glsl_nop
,
11295 /* WINED3DSIH_DCL_TGSM_RAW */ shader_glsl_nop
,
11296 /* WINED3DSIH_DCL_TGSM_STRUCTURED */ shader_glsl_nop
,
11297 /* WINED3DSIH_DCL_THREAD_GROUP */ shader_glsl_nop
,
11298 /* WINED3DSIH_DCL_UAV_RAW */ shader_glsl_nop
,
11299 /* WINED3DSIH_DCL_UAV_STRUCTURED */ shader_glsl_nop
,
11300 /* WINED3DSIH_DCL_UAV_TYPED */ shader_glsl_nop
,
11301 /* WINED3DSIH_DCL_VERTICES_OUT */ shader_glsl_nop
,
11302 /* WINED3DSIH_DEF */ shader_glsl_nop
,
11303 /* WINED3DSIH_DEFAULT */ shader_glsl_default
,
11304 /* WINED3DSIH_DEFB */ shader_glsl_nop
,
11305 /* WINED3DSIH_DEFI */ shader_glsl_nop
,
11306 /* WINED3DSIH_DIV */ shader_glsl_binop
,
11307 /* WINED3DSIH_DP2 */ shader_glsl_dot
,
11308 /* WINED3DSIH_DP2ADD */ shader_glsl_dp2add
,
11309 /* WINED3DSIH_DP3 */ shader_glsl_dot
,
11310 /* WINED3DSIH_DP4 */ shader_glsl_dot
,
11311 /* WINED3DSIH_DST */ shader_glsl_dst
,
11312 /* WINED3DSIH_DSX */ shader_glsl_map2gl
,
11313 /* WINED3DSIH_DSX_COARSE */ shader_glsl_map2gl
,
11314 /* WINED3DSIH_DSX_FINE */ shader_glsl_map2gl
,
11315 /* WINED3DSIH_DSY */ shader_glsl_map2gl
,
11316 /* WINED3DSIH_DSY_COARSE */ shader_glsl_map2gl
,
11317 /* WINED3DSIH_DSY_FINE */ shader_glsl_map2gl
,
11318 /* WINED3DSIH_ELSE */ shader_glsl_else
,
11319 /* WINED3DSIH_EMIT */ shader_glsl_emit
,
11320 /* WINED3DSIH_EMIT_STREAM */ shader_glsl_emit
,
11321 /* WINED3DSIH_ENDIF */ shader_glsl_end
,
11322 /* WINED3DSIH_ENDLOOP */ shader_glsl_end
,
11323 /* WINED3DSIH_ENDREP */ shader_glsl_end
,
11324 /* WINED3DSIH_ENDSWITCH */ shader_glsl_end
,
11325 /* WINED3DSIH_EQ */ shader_glsl_relop
,
11326 /* WINED3DSIH_EVAL_SAMPLE_INDEX */ shader_glsl_interpolate
,
11327 /* WINED3DSIH_EXP */ shader_glsl_scalar_op
,
11328 /* WINED3DSIH_EXPP */ shader_glsl_expp
,
11329 /* WINED3DSIH_F16TOF32 */ shader_glsl_float16
,
11330 /* WINED3DSIH_F32TOF16 */ shader_glsl_float16
,
11331 /* WINED3DSIH_FCALL */ NULL
,
11332 /* WINED3DSIH_FIRSTBIT_HI */ shader_glsl_map2gl
,
11333 /* WINED3DSIH_FIRSTBIT_LO */ shader_glsl_map2gl
,
11334 /* WINED3DSIH_FIRSTBIT_SHI */ shader_glsl_map2gl
,
11335 /* WINED3DSIH_FRC */ shader_glsl_map2gl
,
11336 /* WINED3DSIH_FTOI */ shader_glsl_to_int
,
11337 /* WINED3DSIH_FTOU */ shader_glsl_to_uint
,
11338 /* WINED3DSIH_GATHER4 */ shader_glsl_gather4
,
11339 /* WINED3DSIH_GATHER4_C */ shader_glsl_gather4
,
11340 /* WINED3DSIH_GATHER4_PO */ shader_glsl_gather4
,
11341 /* WINED3DSIH_GATHER4_PO_C */ shader_glsl_gather4
,
11342 /* WINED3DSIH_GE */ shader_glsl_relop
,
11343 /* WINED3DSIH_HS_CONTROL_POINT_PHASE */ shader_glsl_nop
,
11344 /* WINED3DSIH_HS_DECLS */ shader_glsl_nop
,
11345 /* WINED3DSIH_HS_FORK_PHASE */ shader_glsl_nop
,
11346 /* WINED3DSIH_HS_JOIN_PHASE */ shader_glsl_nop
,
11347 /* WINED3DSIH_IADD */ shader_glsl_binop
,
11348 /* WINED3DSIH_IBFE */ shader_glsl_bitwise_op
,
11349 /* WINED3DSIH_IEQ */ shader_glsl_relop
,
11350 /* WINED3DSIH_IF */ shader_glsl_if
,
11351 /* WINED3DSIH_IFC */ shader_glsl_ifc
,
11352 /* WINED3DSIH_IGE */ shader_glsl_relop
,
11353 /* WINED3DSIH_ILT */ shader_glsl_relop
,
11354 /* WINED3DSIH_IMAD */ shader_glsl_mad
,
11355 /* WINED3DSIH_IMAX */ shader_glsl_map2gl
,
11356 /* WINED3DSIH_IMIN */ shader_glsl_map2gl
,
11357 /* WINED3DSIH_IMM_ATOMIC_ALLOC */ shader_glsl_uav_counter
,
11358 /* WINED3DSIH_IMM_ATOMIC_AND */ shader_glsl_atomic
,
11359 /* WINED3DSIH_IMM_ATOMIC_CMP_EXCH */ shader_glsl_atomic
,
11360 /* WINED3DSIH_IMM_ATOMIC_CONSUME */ shader_glsl_uav_counter
,
11361 /* WINED3DSIH_IMM_ATOMIC_EXCH */ shader_glsl_atomic
,
11362 /* WINED3DSIH_IMM_ATOMIC_IADD */ shader_glsl_atomic
,
11363 /* WINED3DSIH_IMM_ATOMIC_IMAX */ shader_glsl_atomic
,
11364 /* WINED3DSIH_IMM_ATOMIC_IMIN */ shader_glsl_atomic
,
11365 /* WINED3DSIH_IMM_ATOMIC_OR */ shader_glsl_atomic
,
11366 /* WINED3DSIH_IMM_ATOMIC_UMAX */ shader_glsl_atomic
,
11367 /* WINED3DSIH_IMM_ATOMIC_UMIN */ shader_glsl_atomic
,
11368 /* WINED3DSIH_IMM_ATOMIC_XOR */ shader_glsl_atomic
,
11369 /* WINED3DSIH_IMUL */ shader_glsl_mul_extended
,
11370 /* WINED3DSIH_INE */ shader_glsl_relop
,
11371 /* WINED3DSIH_INEG */ shader_glsl_unary_op
,
11372 /* WINED3DSIH_ISHL */ shader_glsl_binop
,
11373 /* WINED3DSIH_ISHR */ shader_glsl_binop
,
11374 /* WINED3DSIH_ITOF */ shader_glsl_to_float
,
11375 /* WINED3DSIH_LABEL */ shader_glsl_label
,
11376 /* WINED3DSIH_LD */ shader_glsl_ld
,
11377 /* WINED3DSIH_LD2DMS */ shader_glsl_ld
,
11378 /* WINED3DSIH_LD_RAW */ shader_glsl_ld_raw_structured
,
11379 /* WINED3DSIH_LD_STRUCTURED */ shader_glsl_ld_raw_structured
,
11380 /* WINED3DSIH_LD_UAV_TYPED */ shader_glsl_ld_uav
,
11381 /* WINED3DSIH_LIT */ shader_glsl_lit
,
11382 /* WINED3DSIH_LOD */ NULL
,
11383 /* WINED3DSIH_LOG */ shader_glsl_scalar_op
,
11384 /* WINED3DSIH_LOGP */ shader_glsl_scalar_op
,
11385 /* WINED3DSIH_LOOP */ shader_glsl_loop
,
11386 /* WINED3DSIH_LRP */ shader_glsl_lrp
,
11387 /* WINED3DSIH_LT */ shader_glsl_relop
,
11388 /* WINED3DSIH_M3x2 */ shader_glsl_mnxn
,
11389 /* WINED3DSIH_M3x3 */ shader_glsl_mnxn
,
11390 /* WINED3DSIH_M3x4 */ shader_glsl_mnxn
,
11391 /* WINED3DSIH_M4x3 */ shader_glsl_mnxn
,
11392 /* WINED3DSIH_M4x4 */ shader_glsl_mnxn
,
11393 /* WINED3DSIH_MAD */ shader_glsl_mad
,
11394 /* WINED3DSIH_MAX */ shader_glsl_map2gl
,
11395 /* WINED3DSIH_MIN */ shader_glsl_map2gl
,
11396 /* WINED3DSIH_MOV */ shader_glsl_mov
,
11397 /* WINED3DSIH_MOVA */ shader_glsl_mov
,
11398 /* WINED3DSIH_MOVC */ shader_glsl_conditional_move
,
11399 /* WINED3DSIH_MUL */ shader_glsl_binop
,
11400 /* WINED3DSIH_NE */ shader_glsl_relop
,
11401 /* WINED3DSIH_NOP */ shader_glsl_nop
,
11402 /* WINED3DSIH_NOT */ shader_glsl_unary_op
,
11403 /* WINED3DSIH_NRM */ shader_glsl_nrm
,
11404 /* WINED3DSIH_OR */ shader_glsl_binop
,
11405 /* WINED3DSIH_PHASE */ shader_glsl_nop
,
11406 /* WINED3DSIH_POW */ shader_glsl_pow
,
11407 /* WINED3DSIH_RCP */ shader_glsl_scalar_op
,
11408 /* WINED3DSIH_REP */ shader_glsl_rep
,
11409 /* WINED3DSIH_RESINFO */ shader_glsl_resinfo
,
11410 /* WINED3DSIH_RET */ shader_glsl_ret
,
11411 /* WINED3DSIH_RETP */ shader_glsl_conditional_op
,
11412 /* WINED3DSIH_ROUND_NE */ shader_glsl_map2gl
,
11413 /* WINED3DSIH_ROUND_NI */ shader_glsl_map2gl
,
11414 /* WINED3DSIH_ROUND_PI */ shader_glsl_map2gl
,
11415 /* WINED3DSIH_ROUND_Z */ shader_glsl_map2gl
,
11416 /* WINED3DSIH_RSQ */ shader_glsl_scalar_op
,
11417 /* WINED3DSIH_SAMPLE */ shader_glsl_sample
,
11418 /* WINED3DSIH_SAMPLE_B */ shader_glsl_sample
,
11419 /* WINED3DSIH_SAMPLE_C */ shader_glsl_sample_c
,
11420 /* WINED3DSIH_SAMPLE_C_LZ */ shader_glsl_sample_c
,
11421 /* WINED3DSIH_SAMPLE_GRAD */ shader_glsl_sample
,
11422 /* WINED3DSIH_SAMPLE_INFO */ shader_glsl_sample_info
,
11423 /* WINED3DSIH_SAMPLE_LOD */ shader_glsl_sample
,
11424 /* WINED3DSIH_SAMPLE_POS */ NULL
,
11425 /* WINED3DSIH_SETP */ NULL
,
11426 /* WINED3DSIH_SGE */ shader_glsl_compare
,
11427 /* WINED3DSIH_SGN */ shader_glsl_sgn
,
11428 /* WINED3DSIH_SINCOS */ shader_glsl_sincos
,
11429 /* WINED3DSIH_SLT */ shader_glsl_compare
,
11430 /* WINED3DSIH_SQRT */ shader_glsl_map2gl
,
11431 /* WINED3DSIH_STORE_RAW */ shader_glsl_store_raw_structured
,
11432 /* WINED3DSIH_STORE_STRUCTURED */ shader_glsl_store_raw_structured
,
11433 /* WINED3DSIH_STORE_UAV_TYPED */ shader_glsl_store_uav
,
11434 /* WINED3DSIH_SUB */ shader_glsl_binop
,
11435 /* WINED3DSIH_SWAPC */ shader_glsl_swapc
,
11436 /* WINED3DSIH_SWITCH */ shader_glsl_switch
,
11437 /* WINED3DSIH_SYNC */ shader_glsl_sync
,
11438 /* WINED3DSIH_TEX */ shader_glsl_tex
,
11439 /* WINED3DSIH_TEXBEM */ shader_glsl_texbem
,
11440 /* WINED3DSIH_TEXBEML */ shader_glsl_texbem
,
11441 /* WINED3DSIH_TEXCOORD */ shader_glsl_texcoord
,
11442 /* WINED3DSIH_TEXDEPTH */ shader_glsl_texdepth
,
11443 /* WINED3DSIH_TEXDP3 */ shader_glsl_texdp3
,
11444 /* WINED3DSIH_TEXDP3TEX */ shader_glsl_texdp3tex
,
11445 /* WINED3DSIH_TEXKILL */ shader_glsl_texkill
,
11446 /* WINED3DSIH_TEXLDD */ shader_glsl_texldd
,
11447 /* WINED3DSIH_TEXLDL */ shader_glsl_texldl
,
11448 /* WINED3DSIH_TEXM3x2DEPTH */ shader_glsl_texm3x2depth
,
11449 /* WINED3DSIH_TEXM3x2PAD */ shader_glsl_texm3x2pad
,
11450 /* WINED3DSIH_TEXM3x2TEX */ shader_glsl_texm3x2tex
,
11451 /* WINED3DSIH_TEXM3x3 */ shader_glsl_texm3x3
,
11452 /* WINED3DSIH_TEXM3x3DIFF */ NULL
,
11453 /* WINED3DSIH_TEXM3x3PAD */ shader_glsl_texm3x3pad
,
11454 /* WINED3DSIH_TEXM3x3SPEC */ shader_glsl_texm3x3spec
,
11455 /* WINED3DSIH_TEXM3x3TEX */ shader_glsl_texm3x3tex
,
11456 /* WINED3DSIH_TEXM3x3VSPEC */ shader_glsl_texm3x3vspec
,
11457 /* WINED3DSIH_TEXREG2AR */ shader_glsl_texreg2ar
,
11458 /* WINED3DSIH_TEXREG2GB */ shader_glsl_texreg2gb
,
11459 /* WINED3DSIH_TEXREG2RGB */ shader_glsl_texreg2rgb
,
11460 /* WINED3DSIH_UBFE */ shader_glsl_bitwise_op
,
11461 /* WINED3DSIH_UDIV */ shader_glsl_udiv
,
11462 /* WINED3DSIH_UGE */ shader_glsl_relop
,
11463 /* WINED3DSIH_ULT */ shader_glsl_relop
,
11464 /* WINED3DSIH_UMAX */ shader_glsl_map2gl
,
11465 /* WINED3DSIH_UMIN */ shader_glsl_map2gl
,
11466 /* WINED3DSIH_UMUL */ shader_glsl_mul_extended
,
11467 /* WINED3DSIH_USHR */ shader_glsl_binop
,
11468 /* WINED3DSIH_UTOF */ shader_glsl_to_float
,
11469 /* WINED3DSIH_XOR */ shader_glsl_binop
,
11472 static void shader_glsl_handle_instruction(const struct wined3d_shader_instruction
*ins
) {
11473 SHADER_HANDLER hw_fct
;
11475 /* Select handler */
11476 hw_fct
= shader_glsl_instruction_handler_table
[ins
->handler_idx
];
11478 /* Unhandled opcode */
11481 FIXME("Backend can't handle opcode %s.\n", debug_d3dshaderinstructionhandler(ins
->handler_idx
));
11486 shader_glsl_add_instruction_modifiers(ins
);
11489 static BOOL
shader_glsl_has_ffp_proj_control(void *shader_priv
)
11491 struct shader_glsl_priv
*priv
= shader_priv
;
11493 return priv
->ffp_proj_control
;
11496 static uint64_t shader_glsl_shader_compile(struct wined3d_context
*context
, const struct wined3d_shader_desc
*shader_desc
,
11497 enum wined3d_shader_type shader_type
)
11499 ERR("Not implemented.\n");
11503 const struct wined3d_shader_backend_ops glsl_shader_backend
=
11505 shader_glsl_handle_instruction
,
11506 shader_glsl_precompile
,
11507 shader_glsl_select
,
11508 shader_glsl_select_compute
,
11509 shader_glsl_disable
,
11510 shader_glsl_update_float_vertex_constants
,
11511 shader_glsl_update_float_pixel_constants
,
11512 shader_glsl_load_constants
,
11513 shader_glsl_destroy
,
11516 shader_glsl_allocate_context_data
,
11517 shader_glsl_free_context_data
,
11518 shader_glsl_init_context_state
,
11519 shader_glsl_get_caps
,
11520 shader_glsl_color_fixup_supported
,
11521 shader_glsl_has_ffp_proj_control
,
11522 shader_glsl_shader_compile
,
11525 static void glsl_vertex_pipe_vp_enable(const struct wined3d_context
*context
, BOOL enable
) {}
11527 static void glsl_vertex_pipe_vp_get_caps(const struct wined3d_adapter
*adapter
, struct wined3d_vertex_caps
*caps
)
11529 const struct wined3d_gl_info
*gl_info
= &adapter
->gl_info
;
11531 caps
->xyzrhw
= TRUE
;
11532 caps
->emulated_flatshading
= !needs_legacy_glsl_syntax(gl_info
);
11533 caps
->ffp_generic_attributes
= TRUE
;
11534 caps
->max_active_lights
= WINED3D_MAX_ACTIVE_LIGHTS
;
11535 caps
->max_vertex_blend_matrices
= MAX_VERTEX_BLENDS
;
11536 caps
->max_vertex_blend_matrix_index
= 0;
11537 caps
->vertex_processing_caps
= WINED3DVTXPCAPS_TEXGEN
11538 | WINED3DVTXPCAPS_MATERIALSOURCE7
11539 | WINED3DVTXPCAPS_VERTEXFOG
11540 | WINED3DVTXPCAPS_DIRECTIONALLIGHTS
11541 | WINED3DVTXPCAPS_POSITIONALLIGHTS
11542 | WINED3DVTXPCAPS_LOCALVIEWER
11543 | WINED3DVTXPCAPS_TEXGEN_SPHEREMAP
;
11544 caps
->fvf_caps
= WINED3DFVFCAPS_PSIZE
| 8; /* 8 texture coordinates. */
11545 caps
->max_user_clip_planes
= gl_info
->limits
.user_clip_distances
;
11546 caps
->raster_caps
= WINED3DPRASTERCAPS_FOGRANGE
;
11549 static DWORD
glsl_vertex_pipe_vp_get_emul_mask(const struct wined3d_gl_info
*gl_info
)
11551 if (gl_info
->supported
[WINED3D_GL_LEGACY_CONTEXT
])
11552 return GL_EXT_EMUL_ARB_MULTITEXTURE
;
11556 static void *glsl_vertex_pipe_vp_alloc(const struct wined3d_shader_backend_ops
*shader_backend
, void *shader_priv
)
11558 struct shader_glsl_priv
*priv
;
11560 if (shader_backend
== &glsl_shader_backend
)
11562 priv
= shader_priv
;
11563 wine_rb_init(&priv
->ffp_vertex_shaders
, wined3d_ffp_vertex_program_key_compare
);
11567 FIXME("GLSL vertex pipe without GLSL shader backend not implemented.\n");
11572 static void shader_glsl_free_ffp_vertex_shader(struct wine_rb_entry
*entry
, void *param
)
11574 struct glsl_ffp_vertex_shader
*shader
= WINE_RB_ENTRY_VALUE(entry
,
11575 struct glsl_ffp_vertex_shader
, desc
.entry
);
11576 struct glsl_shader_prog_link
*program
, *program2
;
11577 struct glsl_ffp_destroy_ctx
*ctx
= param
;
11578 const struct wined3d_gl_info
*gl_info
;
11580 gl_info
= ctx
->context_gl
->gl_info
;
11581 LIST_FOR_EACH_ENTRY_SAFE(program
, program2
, &shader
->linked_programs
,
11582 struct glsl_shader_prog_link
, vs
.shader_entry
)
11584 delete_glsl_program_entry(ctx
->priv
, gl_info
, program
);
11586 GL_EXTCALL(glDeleteShader(shader
->id
));
11590 /* Context activation is done by the caller. */
11591 static void glsl_vertex_pipe_vp_free(struct wined3d_device
*device
, struct wined3d_context
*context
)
11593 struct wined3d_context_gl
*context_gl
= wined3d_context_gl(context
);
11594 struct shader_glsl_priv
*priv
= device
->vertex_priv
;
11595 struct glsl_ffp_destroy_ctx ctx
;
11598 ctx
.context_gl
= context_gl
;
11599 wine_rb_destroy(&priv
->ffp_vertex_shaders
, shader_glsl_free_ffp_vertex_shader
, &ctx
);
11602 static void glsl_vertex_pipe_nop(struct wined3d_context
*context
,
11603 const struct wined3d_state
*state
, DWORD state_id
) {}
11605 static void glsl_vertex_pipe_shader(struct wined3d_context
*context
,
11606 const struct wined3d_state
*state
, DWORD state_id
)
11608 context
->shader_update_mask
|= 1u << WINED3D_SHADER_TYPE_VERTEX
;
11611 static void glsl_vertex_pipe_vdecl(struct wined3d_context
*context
,
11612 const struct wined3d_state
*state
, DWORD state_id
)
11614 struct wined3d_context_gl
*context_gl
= wined3d_context_gl(context
);
11615 const struct wined3d_gl_info
*gl_info
= context_gl
->gl_info
;
11616 BOOL specular
= !!(context
->stream_info
.use_map
& (1u << WINED3D_FFP_SPECULAR
));
11617 BOOL diffuse
= !!(context
->stream_info
.use_map
& (1u << WINED3D_FFP_DIFFUSE
));
11618 BOOL normal
= !!(context
->stream_info
.use_map
& (1u << WINED3D_FFP_NORMAL
));
11619 const BOOL legacy_clip_planes
= needs_legacy_glsl_syntax(gl_info
);
11620 BOOL transformed
= context
->stream_info
.position_transformed
;
11621 BOOL wasrhw
= context
->last_was_rhw
;
11624 context
->last_was_rhw
= transformed
;
11626 /* If the vertex declaration contains a transformed position attribute,
11627 * the draw uses the fixed function vertex pipeline regardless of any
11628 * vertex shader set by the application. */
11629 if (transformed
!= wasrhw
11630 || context
->stream_info
.swizzle_map
!= context
->last_swizzle_map
)
11631 context
->shader_update_mask
|= 1u << WINED3D_SHADER_TYPE_VERTEX
;
11633 context
->last_swizzle_map
= context
->stream_info
.swizzle_map
;
11635 if (!use_vs(state
))
11637 if (context
->last_was_vshader
)
11639 if (legacy_clip_planes
)
11640 for (i
= 0; i
< gl_info
->limits
.user_clip_distances
; ++i
)
11641 clipplane(context
, state
, STATE_CLIPPLANE(i
));
11643 context
->constant_update_mask
|= WINED3D_SHADER_CONST_VS_CLIP_PLANES
;
11646 context
->constant_update_mask
|= WINED3D_SHADER_CONST_FFP_TEXMATRIX
;
11648 /* Because of settings->texcoords, we have to regenerate the vertex
11649 * shader on a vdecl change if there aren't enough varyings to just
11650 * always output all the texture coordinates.
11652 * Likewise, we have to invalidate the shader when using per-vertex
11653 * colours and diffuse/specular attribute presence changes, or when
11654 * normal presence changes. */
11655 if (!shader_glsl_full_ffp_varyings(gl_info
) || (state
->render_states
[WINED3D_RS_COLORVERTEX
]
11656 && (diffuse
!= context
->last_was_diffuse
|| specular
!= context
->last_was_specular
))
11657 || normal
!= context
->last_was_normal
)
11658 context
->shader_update_mask
|= 1u << WINED3D_SHADER_TYPE_VERTEX
;
11661 && state
->shader
[WINED3D_SHADER_TYPE_PIXEL
]->reg_maps
.shader_version
.major
== 1
11662 && state
->shader
[WINED3D_SHADER_TYPE_PIXEL
]->reg_maps
.shader_version
.minor
<= 3)
11663 context
->shader_update_mask
|= 1u << WINED3D_SHADER_TYPE_PIXEL
;
11667 if (!context
->last_was_vshader
)
11669 /* Vertex shader clipping ignores the view matrix. Update all clip planes. */
11670 if (legacy_clip_planes
)
11671 for (i
= 0; i
< gl_info
->limits
.user_clip_distances
; ++i
)
11672 clipplane(context
, state
, STATE_CLIPPLANE(i
));
11674 context
->constant_update_mask
|= WINED3D_SHADER_CONST_VS_CLIP_PLANES
;
11678 context
->last_was_vshader
= use_vs(state
);
11679 context
->last_was_diffuse
= diffuse
;
11680 context
->last_was_specular
= specular
;
11681 context
->last_was_normal
= normal
;
11684 static void glsl_vertex_pipe_vs(struct wined3d_context
*context
,
11685 const struct wined3d_state
*state
, DWORD state_id
)
11687 context
->shader_update_mask
|= 1u << WINED3D_SHADER_TYPE_VERTEX
;
11688 /* Different vertex shaders potentially require a different vertex attributes setup. */
11689 if (!isStateDirty(context
, STATE_VDECL
))
11690 context_apply_state(context
, state
, STATE_VDECL
);
11693 static void glsl_vertex_pipe_hs(struct wined3d_context
*context
,
11694 const struct wined3d_state
*state
, DWORD state_id
)
11696 /* In Direct3D tessellator options (e.g. output primitive type, primitive
11697 * winding) are defined in Hull Shaders, while in GLSL those are
11698 * specified in Tessellation Evaluation Shaders. */
11699 context
->shader_update_mask
|= 1u << WINED3D_SHADER_TYPE_DOMAIN
;
11701 if (state
->shader
[WINED3D_SHADER_TYPE_VERTEX
])
11702 context
->shader_update_mask
|= 1u << WINED3D_SHADER_TYPE_VERTEX
;
11705 static void glsl_vertex_pipe_geometry_shader(struct wined3d_context
*context
,
11706 const struct wined3d_state
*state
, DWORD state_id
)
11708 struct glsl_context_data
*ctx_data
= context
->shader_backend_data
;
11709 BOOL rasterization_disabled
;
11711 rasterization_disabled
= is_rasterization_disabled(state
->shader
[WINED3D_SHADER_TYPE_GEOMETRY
]);
11712 if (ctx_data
->rasterization_disabled
!= rasterization_disabled
)
11713 context
->shader_update_mask
|= 1u << WINED3D_SHADER_TYPE_PIXEL
;
11714 ctx_data
->rasterization_disabled
= rasterization_disabled
;
11716 if (state
->shader
[WINED3D_SHADER_TYPE_DOMAIN
])
11717 context
->shader_update_mask
|= 1u << WINED3D_SHADER_TYPE_DOMAIN
;
11718 else if (state
->shader
[WINED3D_SHADER_TYPE_VERTEX
]
11719 && state
->shader
[WINED3D_SHADER_TYPE_VERTEX
]->reg_maps
.shader_version
.major
>= 4)
11720 context
->shader_update_mask
|= 1u << WINED3D_SHADER_TYPE_VERTEX
;
11723 static void glsl_vertex_pipe_pixel_shader(struct wined3d_context
*context
,
11724 const struct wined3d_state
*state
, DWORD state_id
)
11726 if (state
->shader
[WINED3D_SHADER_TYPE_GEOMETRY
])
11727 context
->shader_update_mask
|= 1u << WINED3D_SHADER_TYPE_GEOMETRY
;
11728 else if (state
->shader
[WINED3D_SHADER_TYPE_DOMAIN
])
11729 context
->shader_update_mask
|= 1u << WINED3D_SHADER_TYPE_DOMAIN
;
11730 else if (state
->shader
[WINED3D_SHADER_TYPE_VERTEX
]
11731 && state
->shader
[WINED3D_SHADER_TYPE_VERTEX
]->reg_maps
.shader_version
.major
>= 4)
11732 context
->shader_update_mask
|= 1u << WINED3D_SHADER_TYPE_VERTEX
;
11735 static void glsl_vertex_pipe_world(struct wined3d_context
*context
,
11736 const struct wined3d_state
*state
, DWORD state_id
)
11738 context
->constant_update_mask
|= WINED3D_SHADER_CONST_FFP_MODELVIEW
;
11741 static void glsl_vertex_pipe_vertexblend(struct wined3d_context
*context
,
11742 const struct wined3d_state
*state
, DWORD state_id
)
11744 context
->constant_update_mask
|= WINED3D_SHADER_CONST_FFP_VERTEXBLEND
;
11747 static void glsl_vertex_pipe_view(struct wined3d_context
*context
, const struct wined3d_state
*state
, DWORD state_id
)
11749 struct wined3d_context_gl
*context_gl
= wined3d_context_gl(context
);
11750 const struct wined3d_gl_info
*gl_info
= context_gl
->gl_info
;
11753 context
->constant_update_mask
|= WINED3D_SHADER_CONST_FFP_MODELVIEW
11754 | WINED3D_SHADER_CONST_FFP_LIGHTS
11755 | WINED3D_SHADER_CONST_FFP_VERTEXBLEND
;
11757 if (needs_legacy_glsl_syntax(gl_info
))
11759 for (k
= 0; k
< gl_info
->limits
.user_clip_distances
; ++k
)
11761 if (!isStateDirty(context
, STATE_CLIPPLANE(k
)))
11762 clipplane(context
, state
, STATE_CLIPPLANE(k
));
11767 context
->constant_update_mask
|= WINED3D_SHADER_CONST_VS_CLIP_PLANES
;
11771 static void glsl_vertex_pipe_projection(struct wined3d_context
*context
,
11772 const struct wined3d_state
*state
, DWORD state_id
)
11774 /* Table fog behavior depends on the projection matrix. */
11775 if (state
->render_states
[WINED3D_RS_FOGENABLE
]
11776 && state
->render_states
[WINED3D_RS_FOGTABLEMODE
] != WINED3D_FOG_NONE
)
11777 context
->shader_update_mask
|= 1u << WINED3D_SHADER_TYPE_VERTEX
;
11778 context
->constant_update_mask
|= WINED3D_SHADER_CONST_FFP_PROJ
;
11781 static void glsl_vertex_pipe_viewport(struct wined3d_context
*context
,
11782 const struct wined3d_state
*state
, DWORD state_id
)
11784 if (!isStateDirty(context
, STATE_TRANSFORM(WINED3D_TS_PROJECTION
)))
11785 glsl_vertex_pipe_projection(context
, state
, STATE_TRANSFORM(WINED3D_TS_PROJECTION
));
11786 if (!isStateDirty(context
, STATE_RENDER(WINED3D_RS_POINTSCALEENABLE
))
11787 && state
->render_states
[WINED3D_RS_POINTSCALEENABLE
])
11788 context
->constant_update_mask
|= WINED3D_SHADER_CONST_VS_POINTSIZE
;
11789 context
->constant_update_mask
|= WINED3D_SHADER_CONST_POS_FIXUP
;
11792 static void glsl_vertex_pipe_texmatrix(struct wined3d_context
*context
,
11793 const struct wined3d_state
*state
, DWORD state_id
)
11795 context
->constant_update_mask
|= WINED3D_SHADER_CONST_FFP_TEXMATRIX
;
11798 static void glsl_vertex_pipe_texmatrix_np2(struct wined3d_context
*context
,
11799 const struct wined3d_state
*state
, DWORD state_id
)
11801 DWORD sampler
= state_id
- STATE_SAMPLER(0);
11802 const struct wined3d_texture
*texture
= state
->textures
[sampler
];
11808 if (sampler
>= WINED3D_MAX_TEXTURES
)
11811 if ((np2
= !(texture
->flags
& WINED3D_TEXTURE_POW2_MAT_IDENT
))
11812 || context
->lastWasPow2Texture
& (1u << sampler
))
11815 context
->lastWasPow2Texture
|= 1u << sampler
;
11817 context
->lastWasPow2Texture
&= ~(1u << sampler
);
11819 context
->constant_update_mask
|= WINED3D_SHADER_CONST_FFP_TEXMATRIX
;
11823 static void glsl_vertex_pipe_material(struct wined3d_context
*context
,
11824 const struct wined3d_state
*state
, DWORD state_id
)
11826 context
->constant_update_mask
|= WINED3D_SHADER_CONST_FFP_MATERIAL
;
11829 static void glsl_vertex_pipe_light(struct wined3d_context
*context
,
11830 const struct wined3d_state
*state
, DWORD state_id
)
11832 context
->constant_update_mask
|= WINED3D_SHADER_CONST_FFP_LIGHTS
;
11835 static void glsl_vertex_pipe_pointsize(struct wined3d_context
*context
,
11836 const struct wined3d_state
*state
, DWORD state_id
)
11838 context
->constant_update_mask
|= WINED3D_SHADER_CONST_VS_POINTSIZE
;
11841 static void glsl_vertex_pipe_pointscale(struct wined3d_context
*context
,
11842 const struct wined3d_state
*state
, DWORD state_id
)
11844 if (!use_vs(state
))
11845 context
->constant_update_mask
|= WINED3D_SHADER_CONST_VS_POINTSIZE
;
11848 static void glsl_vertex_pointsprite_core(struct wined3d_context
*context
,
11849 const struct wined3d_state
*state
, DWORD state_id
)
11851 static unsigned int once
;
11853 if (state
->primitive_type
== WINED3D_PT_POINTLIST
11854 && !state
->render_states
[WINED3D_RS_POINTSPRITEENABLE
] && !once
++)
11855 FIXME("Non-point sprite points not supported in core profile.\n");
11858 static void glsl_vertex_pipe_shademode(struct wined3d_context
*context
,
11859 const struct wined3d_state
*state
, DWORD state_id
)
11861 context
->shader_update_mask
|= 1u << WINED3D_SHADER_TYPE_VERTEX
;
11864 static void glsl_vertex_pipe_clip_plane(struct wined3d_context
*context
,
11865 const struct wined3d_state
*state
, DWORD state_id
)
11867 struct wined3d_context_gl
*context_gl
= wined3d_context_gl(context
);
11868 const struct wined3d_gl_info
*gl_info
= context_gl
->gl_info
;
11869 UINT index
= state_id
- STATE_CLIPPLANE(0);
11871 if (index
>= gl_info
->limits
.user_clip_distances
)
11874 context
->constant_update_mask
|= WINED3D_SHADER_CONST_VS_CLIP_PLANES
;
11877 static const struct wined3d_state_entry_template glsl_vertex_pipe_vp_states
[] =
11879 {STATE_VDECL
, {STATE_VDECL
, glsl_vertex_pipe_vdecl
}, WINED3D_GL_EXT_NONE
},
11880 {STATE_SHADER(WINED3D_SHADER_TYPE_VERTEX
), {STATE_SHADER(WINED3D_SHADER_TYPE_VERTEX
), glsl_vertex_pipe_vs
}, WINED3D_GL_EXT_NONE
},
11881 {STATE_SHADER(WINED3D_SHADER_TYPE_HULL
), {STATE_SHADER(WINED3D_SHADER_TYPE_HULL
), glsl_vertex_pipe_hs
}, WINED3D_GL_EXT_NONE
},
11882 {STATE_SHADER(WINED3D_SHADER_TYPE_GEOMETRY
), {STATE_SHADER(WINED3D_SHADER_TYPE_GEOMETRY
), glsl_vertex_pipe_geometry_shader
}, WINED3D_GL_EXT_NONE
},
11883 {STATE_SHADER(WINED3D_SHADER_TYPE_PIXEL
), {STATE_SHADER(WINED3D_SHADER_TYPE_PIXEL
), glsl_vertex_pipe_pixel_shader
}, WINED3D_GL_EXT_NONE
},
11884 {STATE_MATERIAL
, {STATE_RENDER(WINED3D_RS_SPECULARENABLE
), NULL
}, WINED3D_GL_EXT_NONE
},
11885 {STATE_RENDER(WINED3D_RS_SPECULARENABLE
), {STATE_RENDER(WINED3D_RS_SPECULARENABLE
), glsl_vertex_pipe_material
}, WINED3D_GL_EXT_NONE
},
11887 {STATE_CLIPPLANE(0), {STATE_CLIPPLANE(0), glsl_vertex_pipe_clip_plane
}, WINED3D_GLSL_130
},
11888 {STATE_CLIPPLANE(0), {STATE_CLIPPLANE(0), clipplane
}, WINED3D_GL_EXT_NONE
},
11889 {STATE_CLIPPLANE(1), {STATE_CLIPPLANE(1), glsl_vertex_pipe_clip_plane
}, WINED3D_GLSL_130
},
11890 {STATE_CLIPPLANE(1), {STATE_CLIPPLANE(1), clipplane
}, WINED3D_GL_EXT_NONE
},
11891 {STATE_CLIPPLANE(2), {STATE_CLIPPLANE(2), glsl_vertex_pipe_clip_plane
}, WINED3D_GLSL_130
},
11892 {STATE_CLIPPLANE(2), {STATE_CLIPPLANE(2), clipplane
}, WINED3D_GL_EXT_NONE
},
11893 {STATE_CLIPPLANE(3), {STATE_CLIPPLANE(3), glsl_vertex_pipe_clip_plane
}, WINED3D_GLSL_130
},
11894 {STATE_CLIPPLANE(3), {STATE_CLIPPLANE(3), clipplane
}, WINED3D_GL_EXT_NONE
},
11895 {STATE_CLIPPLANE(4), {STATE_CLIPPLANE(4), glsl_vertex_pipe_clip_plane
}, WINED3D_GLSL_130
},
11896 {STATE_CLIPPLANE(4), {STATE_CLIPPLANE(4), clipplane
}, WINED3D_GL_EXT_NONE
},
11897 {STATE_CLIPPLANE(5), {STATE_CLIPPLANE(5), glsl_vertex_pipe_clip_plane
}, WINED3D_GLSL_130
},
11898 {STATE_CLIPPLANE(5), {STATE_CLIPPLANE(5), clipplane
}, WINED3D_GL_EXT_NONE
},
11899 {STATE_CLIPPLANE(6), {STATE_CLIPPLANE(6), glsl_vertex_pipe_clip_plane
}, WINED3D_GLSL_130
},
11900 {STATE_CLIPPLANE(6), {STATE_CLIPPLANE(6), clipplane
}, WINED3D_GL_EXT_NONE
},
11901 {STATE_CLIPPLANE(7), {STATE_CLIPPLANE(7), glsl_vertex_pipe_clip_plane
}, WINED3D_GLSL_130
},
11902 {STATE_CLIPPLANE(7), {STATE_CLIPPLANE(7), clipplane
}, WINED3D_GL_EXT_NONE
},
11904 {STATE_LIGHT_TYPE
, {STATE_RENDER(WINED3D_RS_FOGENABLE
), NULL
}, WINED3D_GL_EXT_NONE
},
11905 {STATE_ACTIVELIGHT(0), {STATE_ACTIVELIGHT(0), glsl_vertex_pipe_light
}, WINED3D_GL_EXT_NONE
},
11906 {STATE_ACTIVELIGHT(1), {STATE_ACTIVELIGHT(1), glsl_vertex_pipe_light
}, WINED3D_GL_EXT_NONE
},
11907 {STATE_ACTIVELIGHT(2), {STATE_ACTIVELIGHT(2), glsl_vertex_pipe_light
}, WINED3D_GL_EXT_NONE
},
11908 {STATE_ACTIVELIGHT(3), {STATE_ACTIVELIGHT(3), glsl_vertex_pipe_light
}, WINED3D_GL_EXT_NONE
},
11909 {STATE_ACTIVELIGHT(4), {STATE_ACTIVELIGHT(4), glsl_vertex_pipe_light
}, WINED3D_GL_EXT_NONE
},
11910 {STATE_ACTIVELIGHT(5), {STATE_ACTIVELIGHT(5), glsl_vertex_pipe_light
}, WINED3D_GL_EXT_NONE
},
11911 {STATE_ACTIVELIGHT(6), {STATE_ACTIVELIGHT(6), glsl_vertex_pipe_light
}, WINED3D_GL_EXT_NONE
},
11912 {STATE_ACTIVELIGHT(7), {STATE_ACTIVELIGHT(7), glsl_vertex_pipe_light
}, WINED3D_GL_EXT_NONE
},
11914 {STATE_VIEWPORT
, {STATE_VIEWPORT
, glsl_vertex_pipe_viewport
}, WINED3D_GL_EXT_NONE
},
11915 /* Transform states */
11916 {STATE_TRANSFORM(WINED3D_TS_VIEW
), {STATE_TRANSFORM(WINED3D_TS_VIEW
), glsl_vertex_pipe_view
}, WINED3D_GL_EXT_NONE
},
11917 {STATE_TRANSFORM(WINED3D_TS_PROJECTION
), {STATE_TRANSFORM(WINED3D_TS_PROJECTION
), glsl_vertex_pipe_projection
}, WINED3D_GL_EXT_NONE
},
11918 {STATE_TRANSFORM(WINED3D_TS_TEXTURE0
), {STATE_TEXTURESTAGE(0, WINED3D_TSS_TEXTURE_TRANSFORM_FLAGS
), NULL
}, WINED3D_GL_EXT_NONE
},
11919 {STATE_TRANSFORM(WINED3D_TS_TEXTURE1
), {STATE_TEXTURESTAGE(1, WINED3D_TSS_TEXTURE_TRANSFORM_FLAGS
), NULL
}, WINED3D_GL_EXT_NONE
},
11920 {STATE_TRANSFORM(WINED3D_TS_TEXTURE2
), {STATE_TEXTURESTAGE(2, WINED3D_TSS_TEXTURE_TRANSFORM_FLAGS
), NULL
}, WINED3D_GL_EXT_NONE
},
11921 {STATE_TRANSFORM(WINED3D_TS_TEXTURE3
), {STATE_TEXTURESTAGE(3, WINED3D_TSS_TEXTURE_TRANSFORM_FLAGS
), NULL
}, WINED3D_GL_EXT_NONE
},
11922 {STATE_TRANSFORM(WINED3D_TS_TEXTURE4
), {STATE_TEXTURESTAGE(4, WINED3D_TSS_TEXTURE_TRANSFORM_FLAGS
), NULL
}, WINED3D_GL_EXT_NONE
},
11923 {STATE_TRANSFORM(WINED3D_TS_TEXTURE5
), {STATE_TEXTURESTAGE(5, WINED3D_TSS_TEXTURE_TRANSFORM_FLAGS
), NULL
}, WINED3D_GL_EXT_NONE
},
11924 {STATE_TRANSFORM(WINED3D_TS_TEXTURE6
), {STATE_TEXTURESTAGE(6, WINED3D_TSS_TEXTURE_TRANSFORM_FLAGS
), NULL
}, WINED3D_GL_EXT_NONE
},
11925 {STATE_TRANSFORM(WINED3D_TS_TEXTURE7
), {STATE_TEXTURESTAGE(7, WINED3D_TSS_TEXTURE_TRANSFORM_FLAGS
), NULL
}, WINED3D_GL_EXT_NONE
},
11926 {STATE_TRANSFORM(WINED3D_TS_WORLD_MATRIX(0)), {STATE_TRANSFORM(WINED3D_TS_WORLD_MATRIX(0)), glsl_vertex_pipe_world
}, WINED3D_GL_EXT_NONE
},
11927 {STATE_TRANSFORM(WINED3D_TS_WORLD_MATRIX(1)), {STATE_TRANSFORM(WINED3D_TS_WORLD_MATRIX(1)), glsl_vertex_pipe_vertexblend
}, WINED3D_GL_EXT_NONE
},
11928 {STATE_TRANSFORM(WINED3D_TS_WORLD_MATRIX(2)), {STATE_TRANSFORM(WINED3D_TS_WORLD_MATRIX(2)), glsl_vertex_pipe_vertexblend
}, WINED3D_GL_EXT_NONE
},
11929 {STATE_TRANSFORM(WINED3D_TS_WORLD_MATRIX(3)), {STATE_TRANSFORM(WINED3D_TS_WORLD_MATRIX(3)), glsl_vertex_pipe_vertexblend
}, WINED3D_GL_EXT_NONE
},
11930 {STATE_TEXTURESTAGE(0, WINED3D_TSS_TEXTURE_TRANSFORM_FLAGS
), {STATE_TEXTURESTAGE(0, WINED3D_TSS_TEXTURE_TRANSFORM_FLAGS
), glsl_vertex_pipe_texmatrix
}, WINED3D_GL_EXT_NONE
},
11931 {STATE_TEXTURESTAGE(1, WINED3D_TSS_TEXTURE_TRANSFORM_FLAGS
), {STATE_TEXTURESTAGE(1, WINED3D_TSS_TEXTURE_TRANSFORM_FLAGS
), glsl_vertex_pipe_texmatrix
}, WINED3D_GL_EXT_NONE
},
11932 {STATE_TEXTURESTAGE(2, WINED3D_TSS_TEXTURE_TRANSFORM_FLAGS
), {STATE_TEXTURESTAGE(2, WINED3D_TSS_TEXTURE_TRANSFORM_FLAGS
), glsl_vertex_pipe_texmatrix
}, WINED3D_GL_EXT_NONE
},
11933 {STATE_TEXTURESTAGE(3, WINED3D_TSS_TEXTURE_TRANSFORM_FLAGS
), {STATE_TEXTURESTAGE(3, WINED3D_TSS_TEXTURE_TRANSFORM_FLAGS
), glsl_vertex_pipe_texmatrix
}, WINED3D_GL_EXT_NONE
},
11934 {STATE_TEXTURESTAGE(4, WINED3D_TSS_TEXTURE_TRANSFORM_FLAGS
), {STATE_TEXTURESTAGE(4, WINED3D_TSS_TEXTURE_TRANSFORM_FLAGS
), glsl_vertex_pipe_texmatrix
}, WINED3D_GL_EXT_NONE
},
11935 {STATE_TEXTURESTAGE(5, WINED3D_TSS_TEXTURE_TRANSFORM_FLAGS
), {STATE_TEXTURESTAGE(5, WINED3D_TSS_TEXTURE_TRANSFORM_FLAGS
), glsl_vertex_pipe_texmatrix
}, WINED3D_GL_EXT_NONE
},
11936 {STATE_TEXTURESTAGE(6, WINED3D_TSS_TEXTURE_TRANSFORM_FLAGS
), {STATE_TEXTURESTAGE(6, WINED3D_TSS_TEXTURE_TRANSFORM_FLAGS
), glsl_vertex_pipe_texmatrix
}, WINED3D_GL_EXT_NONE
},
11937 {STATE_TEXTURESTAGE(7, WINED3D_TSS_TEXTURE_TRANSFORM_FLAGS
), {STATE_TEXTURESTAGE(7, WINED3D_TSS_TEXTURE_TRANSFORM_FLAGS
), glsl_vertex_pipe_texmatrix
}, WINED3D_GL_EXT_NONE
},
11938 {STATE_TEXTURESTAGE(0, WINED3D_TSS_TEXCOORD_INDEX
), {STATE_SHADER(WINED3D_SHADER_TYPE_VERTEX
), NULL
}, WINED3D_GL_EXT_NONE
},
11939 {STATE_TEXTURESTAGE(1, WINED3D_TSS_TEXCOORD_INDEX
), {STATE_SHADER(WINED3D_SHADER_TYPE_VERTEX
), NULL
}, WINED3D_GL_EXT_NONE
},
11940 {STATE_TEXTURESTAGE(2, WINED3D_TSS_TEXCOORD_INDEX
), {STATE_SHADER(WINED3D_SHADER_TYPE_VERTEX
), NULL
}, WINED3D_GL_EXT_NONE
},
11941 {STATE_TEXTURESTAGE(3, WINED3D_TSS_TEXCOORD_INDEX
), {STATE_SHADER(WINED3D_SHADER_TYPE_VERTEX
), NULL
}, WINED3D_GL_EXT_NONE
},
11942 {STATE_TEXTURESTAGE(4, WINED3D_TSS_TEXCOORD_INDEX
), {STATE_SHADER(WINED3D_SHADER_TYPE_VERTEX
), NULL
}, WINED3D_GL_EXT_NONE
},
11943 {STATE_TEXTURESTAGE(5, WINED3D_TSS_TEXCOORD_INDEX
), {STATE_SHADER(WINED3D_SHADER_TYPE_VERTEX
), NULL
}, WINED3D_GL_EXT_NONE
},
11944 {STATE_TEXTURESTAGE(6, WINED3D_TSS_TEXCOORD_INDEX
), {STATE_SHADER(WINED3D_SHADER_TYPE_VERTEX
), NULL
}, WINED3D_GL_EXT_NONE
},
11945 {STATE_TEXTURESTAGE(7, WINED3D_TSS_TEXCOORD_INDEX
), {STATE_SHADER(WINED3D_SHADER_TYPE_VERTEX
), NULL
}, WINED3D_GL_EXT_NONE
},
11947 {STATE_RENDER(WINED3D_RS_FOGENABLE
), {STATE_RENDER(WINED3D_RS_FOGENABLE
), glsl_vertex_pipe_shader
}, WINED3D_GL_EXT_NONE
},
11948 {STATE_RENDER(WINED3D_RS_FOGTABLEMODE
), {STATE_RENDER(WINED3D_RS_FOGENABLE
), NULL
}, WINED3D_GL_EXT_NONE
},
11949 {STATE_RENDER(WINED3D_RS_FOGVERTEXMODE
), {STATE_RENDER(WINED3D_RS_FOGENABLE
), NULL
}, WINED3D_GL_EXT_NONE
},
11950 {STATE_RENDER(WINED3D_RS_RANGEFOGENABLE
), {STATE_RENDER(WINED3D_RS_FOGENABLE
), NULL
}, WINED3D_GL_EXT_NONE
},
11951 {STATE_RENDER(WINED3D_RS_CLIPPING
), {STATE_RENDER(WINED3D_RS_CLIPPING
), state_clipping
}, WINED3D_GL_EXT_NONE
},
11952 {STATE_RENDER(WINED3D_RS_CLIPPLANEENABLE
), {STATE_RENDER(WINED3D_RS_CLIPPING
), NULL
}, WINED3D_GL_EXT_NONE
},
11953 {STATE_RENDER(WINED3D_RS_LIGHTING
), {STATE_SHADER(WINED3D_SHADER_TYPE_VERTEX
), NULL
}, WINED3D_GL_EXT_NONE
},
11954 {STATE_RENDER(WINED3D_RS_AMBIENT
), {STATE_RENDER(WINED3D_RS_AMBIENT
), glsl_vertex_pipe_light
}, WINED3D_GL_EXT_NONE
},
11955 {STATE_RENDER(WINED3D_RS_COLORVERTEX
), {STATE_RENDER(WINED3D_RS_COLORVERTEX
), glsl_vertex_pipe_shader
}, WINED3D_GL_EXT_NONE
},
11956 {STATE_RENDER(WINED3D_RS_LOCALVIEWER
), {STATE_SHADER(WINED3D_SHADER_TYPE_VERTEX
), NULL
}, WINED3D_GL_EXT_NONE
},
11957 {STATE_RENDER(WINED3D_RS_NORMALIZENORMALS
), {STATE_SHADER(WINED3D_SHADER_TYPE_VERTEX
), NULL
}, WINED3D_GL_EXT_NONE
},
11958 {STATE_RENDER(WINED3D_RS_DIFFUSEMATERIALSOURCE
), {STATE_SHADER(WINED3D_SHADER_TYPE_VERTEX
), NULL
}, WINED3D_GL_EXT_NONE
},
11959 {STATE_RENDER(WINED3D_RS_SPECULARMATERIALSOURCE
), {STATE_SHADER(WINED3D_SHADER_TYPE_VERTEX
), NULL
}, WINED3D_GL_EXT_NONE
},
11960 {STATE_RENDER(WINED3D_RS_AMBIENTMATERIALSOURCE
), {STATE_SHADER(WINED3D_SHADER_TYPE_VERTEX
), NULL
}, WINED3D_GL_EXT_NONE
},
11961 {STATE_RENDER(WINED3D_RS_EMISSIVEMATERIALSOURCE
), {STATE_SHADER(WINED3D_SHADER_TYPE_VERTEX
), NULL
}, WINED3D_GL_EXT_NONE
},
11962 {STATE_RENDER(WINED3D_RS_VERTEXBLEND
), {STATE_SHADER(WINED3D_SHADER_TYPE_VERTEX
), NULL
}, WINED3D_GL_EXT_NONE
},
11963 {STATE_RENDER(WINED3D_RS_POINTSIZE
), {STATE_RENDER(WINED3D_RS_POINTSIZE_MIN
), NULL
}, WINED3D_GL_EXT_NONE
},
11964 {STATE_RENDER(WINED3D_RS_POINTSIZE_MIN
), {STATE_RENDER(WINED3D_RS_POINTSIZE_MIN
), glsl_vertex_pipe_pointsize
}, WINED3D_GL_EXT_NONE
},
11965 {STATE_RENDER(WINED3D_RS_POINTSPRITEENABLE
), {STATE_RENDER(WINED3D_RS_POINTSPRITEENABLE
), state_pointsprite
}, ARB_POINT_SPRITE
},
11966 {STATE_RENDER(WINED3D_RS_POINTSPRITEENABLE
), {STATE_RENDER(WINED3D_RS_POINTSPRITEENABLE
), state_pointsprite_w
}, WINED3D_GL_LEGACY_CONTEXT
},
11967 {STATE_RENDER(WINED3D_RS_POINTSPRITEENABLE
), {STATE_RENDER(WINED3D_RS_POINTSPRITEENABLE
), glsl_vertex_pointsprite_core
}, WINED3D_GL_EXT_NONE
},
11968 {STATE_RENDER(WINED3D_RS_POINTSCALEENABLE
), {STATE_RENDER(WINED3D_RS_POINTSCALEENABLE
), glsl_vertex_pipe_pointscale
}, WINED3D_GL_EXT_NONE
},
11969 {STATE_RENDER(WINED3D_RS_POINTSCALE_A
), {STATE_RENDER(WINED3D_RS_POINTSCALEENABLE
), NULL
}, WINED3D_GL_EXT_NONE
},
11970 {STATE_RENDER(WINED3D_RS_POINTSCALE_B
), {STATE_RENDER(WINED3D_RS_POINTSCALEENABLE
), NULL
}, WINED3D_GL_EXT_NONE
},
11971 {STATE_RENDER(WINED3D_RS_POINTSCALE_C
), {STATE_RENDER(WINED3D_RS_POINTSCALEENABLE
), NULL
}, WINED3D_GL_EXT_NONE
},
11972 {STATE_RENDER(WINED3D_RS_POINTSIZE_MAX
), {STATE_RENDER(WINED3D_RS_POINTSIZE_MIN
), NULL
}, WINED3D_GL_EXT_NONE
},
11973 {STATE_RENDER(WINED3D_RS_TWEENFACTOR
), {STATE_SHADER(WINED3D_SHADER_TYPE_VERTEX
), NULL
}, WINED3D_GL_EXT_NONE
},
11974 {STATE_RENDER(WINED3D_RS_INDEXEDVERTEXBLENDENABLE
), {STATE_SHADER(WINED3D_SHADER_TYPE_VERTEX
), NULL
}, WINED3D_GL_EXT_NONE
},
11975 /* NP2 texture matrix fixups. They are not needed if
11976 * GL_ARB_texture_non_power_of_two is supported. Otherwise, register
11977 * glsl_vertex_pipe_texmatrix(), which takes care of updating the texture
11979 {STATE_SAMPLER(0), {0, NULL
}, ARB_TEXTURE_NON_POWER_OF_TWO
},
11980 {STATE_SAMPLER(0), {0, NULL
}, WINED3D_GL_NORMALIZED_TEXRECT
},
11981 {STATE_SAMPLER(0), {STATE_SAMPLER(0), glsl_vertex_pipe_texmatrix_np2
}, WINED3D_GL_EXT_NONE
},
11982 {STATE_SAMPLER(1), {0, NULL
}, ARB_TEXTURE_NON_POWER_OF_TWO
},
11983 {STATE_SAMPLER(1), {0, NULL
}, WINED3D_GL_NORMALIZED_TEXRECT
},
11984 {STATE_SAMPLER(1), {STATE_SAMPLER(1), glsl_vertex_pipe_texmatrix_np2
}, WINED3D_GL_EXT_NONE
},
11985 {STATE_SAMPLER(2), {0, NULL
}, ARB_TEXTURE_NON_POWER_OF_TWO
},
11986 {STATE_SAMPLER(2), {0, NULL
}, WINED3D_GL_NORMALIZED_TEXRECT
},
11987 {STATE_SAMPLER(2), {STATE_SAMPLER(2), glsl_vertex_pipe_texmatrix_np2
}, WINED3D_GL_EXT_NONE
},
11988 {STATE_SAMPLER(3), {0, NULL
}, ARB_TEXTURE_NON_POWER_OF_TWO
},
11989 {STATE_SAMPLER(3), {0, NULL
}, WINED3D_GL_NORMALIZED_TEXRECT
},
11990 {STATE_SAMPLER(3), {STATE_SAMPLER(3), glsl_vertex_pipe_texmatrix_np2
}, WINED3D_GL_EXT_NONE
},
11991 {STATE_SAMPLER(4), {0, NULL
}, ARB_TEXTURE_NON_POWER_OF_TWO
},
11992 {STATE_SAMPLER(4), {0, NULL
}, WINED3D_GL_NORMALIZED_TEXRECT
},
11993 {STATE_SAMPLER(4), {STATE_SAMPLER(4), glsl_vertex_pipe_texmatrix_np2
}, WINED3D_GL_EXT_NONE
},
11994 {STATE_SAMPLER(5), {0, NULL
}, ARB_TEXTURE_NON_POWER_OF_TWO
},
11995 {STATE_SAMPLER(5), {0, NULL
}, WINED3D_GL_NORMALIZED_TEXRECT
},
11996 {STATE_SAMPLER(5), {STATE_SAMPLER(5), glsl_vertex_pipe_texmatrix_np2
}, WINED3D_GL_EXT_NONE
},
11997 {STATE_SAMPLER(6), {0, NULL
}, ARB_TEXTURE_NON_POWER_OF_TWO
},
11998 {STATE_SAMPLER(6), {0, NULL
}, WINED3D_GL_NORMALIZED_TEXRECT
},
11999 {STATE_SAMPLER(6), {STATE_SAMPLER(6), glsl_vertex_pipe_texmatrix_np2
}, WINED3D_GL_EXT_NONE
},
12000 {STATE_SAMPLER(7), {0, NULL
}, ARB_TEXTURE_NON_POWER_OF_TWO
},
12001 {STATE_SAMPLER(7), {0, NULL
}, WINED3D_GL_NORMALIZED_TEXRECT
},
12002 {STATE_SAMPLER(7), {STATE_SAMPLER(7), glsl_vertex_pipe_texmatrix_np2
}, WINED3D_GL_EXT_NONE
},
12003 {STATE_POINT_ENABLE
, {STATE_POINT_ENABLE
, glsl_vertex_pipe_shader
}, WINED3D_GL_EXT_NONE
},
12004 {STATE_RENDER(WINED3D_RS_SHADEMODE
), {STATE_RENDER(WINED3D_RS_SHADEMODE
), glsl_vertex_pipe_shademode
}, WINED3D_GLSL_130
},
12005 {STATE_RENDER(WINED3D_RS_SHADEMODE
), {STATE_RENDER(WINED3D_RS_SHADEMODE
), glsl_vertex_pipe_nop
}, WINED3D_GL_EXT_NONE
},
12006 {0 /* Terminate */, {0, NULL
}, WINED3D_GL_EXT_NONE
},
12010 * - Implement vertex tweening. */
12011 const struct wined3d_vertex_pipe_ops glsl_vertex_pipe
=
12013 glsl_vertex_pipe_vp_enable
,
12014 glsl_vertex_pipe_vp_get_caps
,
12015 glsl_vertex_pipe_vp_get_emul_mask
,
12016 glsl_vertex_pipe_vp_alloc
,
12017 glsl_vertex_pipe_vp_free
,
12018 glsl_vertex_pipe_vp_states
,
12021 static void glsl_fragment_pipe_enable(const struct wined3d_context
*context
, BOOL enable
)
12023 /* Nothing to do. */
12026 static void glsl_fragment_pipe_get_caps(const struct wined3d_adapter
*adapter
, struct fragment_caps
*caps
)
12028 const struct wined3d_gl_info
*gl_info
= &adapter
->gl_info
;
12030 caps
->wined3d_caps
= WINED3D_FRAGMENT_CAP_PROJ_CONTROL
12031 | WINED3D_FRAGMENT_CAP_SRGB_WRITE
12032 | WINED3D_FRAGMENT_CAP_COLOR_KEY
;
12033 caps
->PrimitiveMiscCaps
= WINED3DPMISCCAPS_TSSARGTEMP
12034 | WINED3DPMISCCAPS_PERSTAGECONSTANT
;
12035 caps
->TextureOpCaps
= WINED3DTEXOPCAPS_DISABLE
12036 | WINED3DTEXOPCAPS_SELECTARG1
12037 | WINED3DTEXOPCAPS_SELECTARG2
12038 | WINED3DTEXOPCAPS_MODULATE4X
12039 | WINED3DTEXOPCAPS_MODULATE2X
12040 | WINED3DTEXOPCAPS_MODULATE
12041 | WINED3DTEXOPCAPS_ADDSIGNED2X
12042 | WINED3DTEXOPCAPS_ADDSIGNED
12043 | WINED3DTEXOPCAPS_ADD
12044 | WINED3DTEXOPCAPS_SUBTRACT
12045 | WINED3DTEXOPCAPS_ADDSMOOTH
12046 | WINED3DTEXOPCAPS_BLENDCURRENTALPHA
12047 | WINED3DTEXOPCAPS_BLENDFACTORALPHA
12048 | WINED3DTEXOPCAPS_BLENDTEXTUREALPHA
12049 | WINED3DTEXOPCAPS_BLENDDIFFUSEALPHA
12050 | WINED3DTEXOPCAPS_BLENDTEXTUREALPHAPM
12051 | WINED3DTEXOPCAPS_MODULATEALPHA_ADDCOLOR
12052 | WINED3DTEXOPCAPS_MODULATECOLOR_ADDALPHA
12053 | WINED3DTEXOPCAPS_MODULATEINVCOLOR_ADDALPHA
12054 | WINED3DTEXOPCAPS_MODULATEINVALPHA_ADDCOLOR
12055 | WINED3DTEXOPCAPS_DOTPRODUCT3
12056 | WINED3DTEXOPCAPS_MULTIPLYADD
12057 | WINED3DTEXOPCAPS_LERP
12058 | WINED3DTEXOPCAPS_BUMPENVMAP
12059 | WINED3DTEXOPCAPS_BUMPENVMAPLUMINANCE
;
12060 caps
->MaxTextureBlendStages
= WINED3D_MAX_TEXTURES
;
12061 caps
->MaxSimultaneousTextures
= min(gl_info
->limits
.samplers
[WINED3D_SHADER_TYPE_PIXEL
], WINED3D_MAX_TEXTURES
);
12064 static DWORD
glsl_fragment_pipe_get_emul_mask(const struct wined3d_gl_info
*gl_info
)
12066 if (gl_info
->supported
[WINED3D_GL_LEGACY_CONTEXT
])
12067 return GL_EXT_EMUL_ARB_MULTITEXTURE
;
12071 static void *glsl_fragment_pipe_alloc(const struct wined3d_shader_backend_ops
*shader_backend
, void *shader_priv
)
12073 struct shader_glsl_priv
*priv
;
12075 if (shader_backend
== &glsl_shader_backend
)
12077 priv
= shader_priv
;
12078 wine_rb_init(&priv
->ffp_fragment_shaders
, wined3d_ffp_frag_program_key_compare
);
12082 FIXME("GLSL fragment pipe without GLSL shader backend not implemented.\n");
12087 static void shader_glsl_free_ffp_fragment_shader(struct wine_rb_entry
*entry
, void *param
)
12089 struct glsl_ffp_fragment_shader
*shader
= WINE_RB_ENTRY_VALUE(entry
,
12090 struct glsl_ffp_fragment_shader
, entry
.entry
);
12091 struct glsl_shader_prog_link
*program
, *program2
;
12092 struct glsl_ffp_destroy_ctx
*ctx
= param
;
12093 const struct wined3d_gl_info
*gl_info
;
12095 gl_info
= ctx
->context_gl
->gl_info
;
12096 LIST_FOR_EACH_ENTRY_SAFE(program
, program2
, &shader
->linked_programs
,
12097 struct glsl_shader_prog_link
, ps
.shader_entry
)
12099 delete_glsl_program_entry(ctx
->priv
, gl_info
, program
);
12101 GL_EXTCALL(glDeleteShader(shader
->id
));
12105 /* Context activation is done by the caller. */
12106 static void glsl_fragment_pipe_free(struct wined3d_device
*device
, struct wined3d_context
*context
)
12108 struct wined3d_context_gl
*context_gl
= wined3d_context_gl(context
);
12109 struct shader_glsl_priv
*priv
= device
->fragment_priv
;
12110 struct glsl_ffp_destroy_ctx ctx
;
12113 ctx
.context_gl
= context_gl
;
12114 wine_rb_destroy(&priv
->ffp_fragment_shaders
, shader_glsl_free_ffp_fragment_shader
, &ctx
);
12117 static void glsl_fragment_pipe_shader(struct wined3d_context
*context
,
12118 const struct wined3d_state
*state
, DWORD state_id
)
12120 context
->last_was_pshader
= use_ps(state
);
12122 context
->shader_update_mask
|= 1u << WINED3D_SHADER_TYPE_PIXEL
;
12125 static void glsl_fragment_pipe_fogparams(struct wined3d_context
*context
,
12126 const struct wined3d_state
*state
, DWORD state_id
)
12128 context
->constant_update_mask
|= WINED3D_SHADER_CONST_PS_FOG
;
12131 static void glsl_fragment_pipe_fog(struct wined3d_context
*context
,
12132 const struct wined3d_state
*state
, DWORD state_id
)
12134 BOOL use_vshader
= use_vs(state
);
12135 enum fogsource new_source
;
12136 DWORD fogstart
= state
->render_states
[WINED3D_RS_FOGSTART
];
12137 DWORD fogend
= state
->render_states
[WINED3D_RS_FOGEND
];
12139 context
->shader_update_mask
|= 1u << WINED3D_SHADER_TYPE_PIXEL
;
12141 if (!state
->render_states
[WINED3D_RS_FOGENABLE
])
12144 if (state
->render_states
[WINED3D_RS_FOGTABLEMODE
] == WINED3D_FOG_NONE
)
12147 new_source
= FOGSOURCE_VS
;
12148 else if (state
->render_states
[WINED3D_RS_FOGVERTEXMODE
] == WINED3D_FOG_NONE
|| context
->stream_info
.position_transformed
)
12149 new_source
= FOGSOURCE_COORD
;
12151 new_source
= FOGSOURCE_FFP
;
12155 new_source
= FOGSOURCE_FFP
;
12158 if (new_source
!= context
->fog_source
|| fogstart
== fogend
)
12160 context
->fog_source
= new_source
;
12161 context
->constant_update_mask
|= WINED3D_SHADER_CONST_PS_FOG
;
12165 static void glsl_fragment_pipe_vdecl(struct wined3d_context
*context
,
12166 const struct wined3d_state
*state
, DWORD state_id
)
12168 const struct wined3d_gl_info
*gl_info
= wined3d_context_gl(context
)->gl_info
;
12170 /* Because of settings->texcoords_initialized and args->texcoords_initialized. */
12171 if (!shader_glsl_full_ffp_varyings(gl_info
))
12172 context
->shader_update_mask
|= 1u << WINED3D_SHADER_TYPE_PIXEL
;
12174 if (!isStateDirty(context
, STATE_RENDER(WINED3D_RS_FOGENABLE
)))
12175 glsl_fragment_pipe_fog(context
, state
, state_id
);
12178 static void glsl_fragment_pipe_vs(struct wined3d_context
*context
,
12179 const struct wined3d_state
*state
, DWORD state_id
)
12181 const struct wined3d_gl_info
*gl_info
= wined3d_context_gl(context
)->gl_info
;
12183 /* Because of settings->texcoords_initialized and args->texcoords_initialized. */
12184 if (!shader_glsl_full_ffp_varyings(gl_info
))
12185 context
->shader_update_mask
|= 1u << WINED3D_SHADER_TYPE_PIXEL
;
12188 static void glsl_fragment_pipe_tex_transform(struct wined3d_context
*context
,
12189 const struct wined3d_state
*state
, DWORD state_id
)
12191 context
->shader_update_mask
|= 1u << WINED3D_SHADER_TYPE_PIXEL
;
12194 static void glsl_fragment_pipe_invalidate_constants(struct wined3d_context
*context
,
12195 const struct wined3d_state
*state
, DWORD state_id
)
12197 context
->constant_update_mask
|= WINED3D_SHADER_CONST_FFP_PS
;
12200 static void glsl_fragment_pipe_alpha_test_func(struct wined3d_context
*context
,
12201 const struct wined3d_state
*state
, DWORD state_id
)
12203 const struct wined3d_gl_info
*gl_info
= wined3d_context_gl(context
)->gl_info
;
12204 GLint func
= wined3d_gl_compare_func(state
->render_states
[WINED3D_RS_ALPHAFUNC
]);
12205 float ref
= wined3d_alpha_ref(state
);
12209 gl_info
->gl_ops
.gl
.p_glAlphaFunc(func
, ref
);
12210 checkGLcall("glAlphaFunc");
12214 static void glsl_fragment_pipe_core_alpha_test(struct wined3d_context
*context
,
12215 const struct wined3d_state
*state
, DWORD state_id
)
12217 context
->shader_update_mask
|= 1u << WINED3D_SHADER_TYPE_PIXEL
;
12220 static void glsl_fragment_pipe_alpha_test(struct wined3d_context
*context
,
12221 const struct wined3d_state
*state
, DWORD state_id
)
12223 const struct wined3d_gl_info
*gl_info
= wined3d_context_gl(context
)->gl_info
;
12225 if (state
->render_states
[WINED3D_RS_ALPHATESTENABLE
])
12227 gl_info
->gl_ops
.gl
.p_glEnable(GL_ALPHA_TEST
);
12228 checkGLcall("glEnable(GL_ALPHA_TEST)");
12232 gl_info
->gl_ops
.gl
.p_glDisable(GL_ALPHA_TEST
);
12233 checkGLcall("glDisable(GL_ALPHA_TEST)");
12237 static void glsl_fragment_pipe_core_alpha_test_ref(struct wined3d_context
*context
,
12238 const struct wined3d_state
*state
, DWORD state_id
)
12240 context
->constant_update_mask
|= WINED3D_SHADER_CONST_PS_ALPHA_TEST
;
12243 static void glsl_fragment_pipe_color_key(struct wined3d_context
*context
,
12244 const struct wined3d_state
*state
, DWORD state_id
)
12246 context
->constant_update_mask
|= WINED3D_SHADER_CONST_FFP_COLOR_KEY
;
12249 static void glsl_fragment_pipe_shademode(struct wined3d_context
*context
,
12250 const struct wined3d_state
*state
, DWORD state_id
)
12252 context
->shader_update_mask
|= 1u << WINED3D_SHADER_TYPE_PIXEL
;
12255 static const struct wined3d_state_entry_template glsl_fragment_pipe_state_template
[] =
12257 {STATE_VDECL
, {STATE_VDECL
, glsl_fragment_pipe_vdecl
}, WINED3D_GL_EXT_NONE
},
12258 {STATE_SHADER(WINED3D_SHADER_TYPE_VERTEX
), {STATE_SHADER(WINED3D_SHADER_TYPE_VERTEX
), glsl_fragment_pipe_vs
}, WINED3D_GL_EXT_NONE
},
12259 {STATE_RENDER(WINED3D_RS_TEXTUREFACTOR
), {STATE_RENDER(WINED3D_RS_TEXTUREFACTOR
), glsl_fragment_pipe_invalidate_constants
}, WINED3D_GL_EXT_NONE
},
12260 {STATE_TEXTURESTAGE(0, WINED3D_TSS_COLOR_OP
), {STATE_SHADER(WINED3D_SHADER_TYPE_PIXEL
), NULL
}, WINED3D_GL_EXT_NONE
},
12261 {STATE_TEXTURESTAGE(0, WINED3D_TSS_COLOR_ARG1
), {STATE_SHADER(WINED3D_SHADER_TYPE_PIXEL
), NULL
}, WINED3D_GL_EXT_NONE
},
12262 {STATE_TEXTURESTAGE(0, WINED3D_TSS_COLOR_ARG2
), {STATE_SHADER(WINED3D_SHADER_TYPE_PIXEL
), NULL
}, WINED3D_GL_EXT_NONE
},
12263 {STATE_TEXTURESTAGE(0, WINED3D_TSS_COLOR_ARG0
), {STATE_SHADER(WINED3D_SHADER_TYPE_PIXEL
), NULL
}, WINED3D_GL_EXT_NONE
},
12264 {STATE_TEXTURESTAGE(0, WINED3D_TSS_ALPHA_OP
), {STATE_SHADER(WINED3D_SHADER_TYPE_PIXEL
), NULL
}, WINED3D_GL_EXT_NONE
},
12265 {STATE_TEXTURESTAGE(0, WINED3D_TSS_ALPHA_ARG1
), {STATE_SHADER(WINED3D_SHADER_TYPE_PIXEL
), NULL
}, WINED3D_GL_EXT_NONE
},
12266 {STATE_TEXTURESTAGE(0, WINED3D_TSS_ALPHA_ARG2
), {STATE_SHADER(WINED3D_SHADER_TYPE_PIXEL
), NULL
}, WINED3D_GL_EXT_NONE
},
12267 {STATE_TEXTURESTAGE(0, WINED3D_TSS_ALPHA_ARG0
), {STATE_SHADER(WINED3D_SHADER_TYPE_PIXEL
), NULL
}, WINED3D_GL_EXT_NONE
},
12268 {STATE_TEXTURESTAGE(0, WINED3D_TSS_RESULT_ARG
), {STATE_SHADER(WINED3D_SHADER_TYPE_PIXEL
), NULL
}, WINED3D_GL_EXT_NONE
},
12269 {STATE_TEXTURESTAGE(1, WINED3D_TSS_COLOR_OP
), {STATE_SHADER(WINED3D_SHADER_TYPE_PIXEL
), NULL
}, WINED3D_GL_EXT_NONE
},
12270 {STATE_TEXTURESTAGE(1, WINED3D_TSS_COLOR_ARG1
), {STATE_SHADER(WINED3D_SHADER_TYPE_PIXEL
), NULL
}, WINED3D_GL_EXT_NONE
},
12271 {STATE_TEXTURESTAGE(1, WINED3D_TSS_COLOR_ARG2
), {STATE_SHADER(WINED3D_SHADER_TYPE_PIXEL
), NULL
}, WINED3D_GL_EXT_NONE
},
12272 {STATE_TEXTURESTAGE(1, WINED3D_TSS_COLOR_ARG0
), {STATE_SHADER(WINED3D_SHADER_TYPE_PIXEL
), NULL
}, WINED3D_GL_EXT_NONE
},
12273 {STATE_TEXTURESTAGE(1, WINED3D_TSS_ALPHA_OP
), {STATE_SHADER(WINED3D_SHADER_TYPE_PIXEL
), NULL
}, WINED3D_GL_EXT_NONE
},
12274 {STATE_TEXTURESTAGE(1, WINED3D_TSS_ALPHA_ARG1
), {STATE_SHADER(WINED3D_SHADER_TYPE_PIXEL
), NULL
}, WINED3D_GL_EXT_NONE
},
12275 {STATE_TEXTURESTAGE(1, WINED3D_TSS_ALPHA_ARG2
), {STATE_SHADER(WINED3D_SHADER_TYPE_PIXEL
), NULL
}, WINED3D_GL_EXT_NONE
},
12276 {STATE_TEXTURESTAGE(1, WINED3D_TSS_ALPHA_ARG0
), {STATE_SHADER(WINED3D_SHADER_TYPE_PIXEL
), NULL
}, WINED3D_GL_EXT_NONE
},
12277 {STATE_TEXTURESTAGE(1, WINED3D_TSS_RESULT_ARG
), {STATE_SHADER(WINED3D_SHADER_TYPE_PIXEL
), NULL
}, WINED3D_GL_EXT_NONE
},
12278 {STATE_TEXTURESTAGE(2, WINED3D_TSS_COLOR_OP
), {STATE_SHADER(WINED3D_SHADER_TYPE_PIXEL
), NULL
}, WINED3D_GL_EXT_NONE
},
12279 {STATE_TEXTURESTAGE(2, WINED3D_TSS_COLOR_ARG1
), {STATE_SHADER(WINED3D_SHADER_TYPE_PIXEL
), NULL
}, WINED3D_GL_EXT_NONE
},
12280 {STATE_TEXTURESTAGE(2, WINED3D_TSS_COLOR_ARG2
), {STATE_SHADER(WINED3D_SHADER_TYPE_PIXEL
), NULL
}, WINED3D_GL_EXT_NONE
},
12281 {STATE_TEXTURESTAGE(2, WINED3D_TSS_COLOR_ARG0
), {STATE_SHADER(WINED3D_SHADER_TYPE_PIXEL
), NULL
}, WINED3D_GL_EXT_NONE
},
12282 {STATE_TEXTURESTAGE(2, WINED3D_TSS_ALPHA_OP
), {STATE_SHADER(WINED3D_SHADER_TYPE_PIXEL
), NULL
}, WINED3D_GL_EXT_NONE
},
12283 {STATE_TEXTURESTAGE(2, WINED3D_TSS_ALPHA_ARG1
), {STATE_SHADER(WINED3D_SHADER_TYPE_PIXEL
), NULL
}, WINED3D_GL_EXT_NONE
},
12284 {STATE_TEXTURESTAGE(2, WINED3D_TSS_ALPHA_ARG2
), {STATE_SHADER(WINED3D_SHADER_TYPE_PIXEL
), NULL
}, WINED3D_GL_EXT_NONE
},
12285 {STATE_TEXTURESTAGE(2, WINED3D_TSS_ALPHA_ARG0
), {STATE_SHADER(WINED3D_SHADER_TYPE_PIXEL
), NULL
}, WINED3D_GL_EXT_NONE
},
12286 {STATE_TEXTURESTAGE(2, WINED3D_TSS_RESULT_ARG
), {STATE_SHADER(WINED3D_SHADER_TYPE_PIXEL
), NULL
}, WINED3D_GL_EXT_NONE
},
12287 {STATE_TEXTURESTAGE(3, WINED3D_TSS_COLOR_OP
), {STATE_SHADER(WINED3D_SHADER_TYPE_PIXEL
), NULL
}, WINED3D_GL_EXT_NONE
},
12288 {STATE_TEXTURESTAGE(3, WINED3D_TSS_COLOR_ARG1
), {STATE_SHADER(WINED3D_SHADER_TYPE_PIXEL
), NULL
}, WINED3D_GL_EXT_NONE
},
12289 {STATE_TEXTURESTAGE(3, WINED3D_TSS_COLOR_ARG2
), {STATE_SHADER(WINED3D_SHADER_TYPE_PIXEL
), NULL
}, WINED3D_GL_EXT_NONE
},
12290 {STATE_TEXTURESTAGE(3, WINED3D_TSS_COLOR_ARG0
), {STATE_SHADER(WINED3D_SHADER_TYPE_PIXEL
), NULL
}, WINED3D_GL_EXT_NONE
},
12291 {STATE_TEXTURESTAGE(3, WINED3D_TSS_ALPHA_OP
), {STATE_SHADER(WINED3D_SHADER_TYPE_PIXEL
), NULL
}, WINED3D_GL_EXT_NONE
},
12292 {STATE_TEXTURESTAGE(3, WINED3D_TSS_ALPHA_ARG1
), {STATE_SHADER(WINED3D_SHADER_TYPE_PIXEL
), NULL
}, WINED3D_GL_EXT_NONE
},
12293 {STATE_TEXTURESTAGE(3, WINED3D_TSS_ALPHA_ARG2
), {STATE_SHADER(WINED3D_SHADER_TYPE_PIXEL
), NULL
}, WINED3D_GL_EXT_NONE
},
12294 {STATE_TEXTURESTAGE(3, WINED3D_TSS_ALPHA_ARG0
), {STATE_SHADER(WINED3D_SHADER_TYPE_PIXEL
), NULL
}, WINED3D_GL_EXT_NONE
},
12295 {STATE_TEXTURESTAGE(3, WINED3D_TSS_RESULT_ARG
), {STATE_SHADER(WINED3D_SHADER_TYPE_PIXEL
), NULL
}, WINED3D_GL_EXT_NONE
},
12296 {STATE_TEXTURESTAGE(4, WINED3D_TSS_COLOR_OP
), {STATE_SHADER(WINED3D_SHADER_TYPE_PIXEL
), NULL
}, WINED3D_GL_EXT_NONE
},
12297 {STATE_TEXTURESTAGE(4, WINED3D_TSS_COLOR_ARG1
), {STATE_SHADER(WINED3D_SHADER_TYPE_PIXEL
), NULL
}, WINED3D_GL_EXT_NONE
},
12298 {STATE_TEXTURESTAGE(4, WINED3D_TSS_COLOR_ARG2
), {STATE_SHADER(WINED3D_SHADER_TYPE_PIXEL
), NULL
}, WINED3D_GL_EXT_NONE
},
12299 {STATE_TEXTURESTAGE(4, WINED3D_TSS_COLOR_ARG0
), {STATE_SHADER(WINED3D_SHADER_TYPE_PIXEL
), NULL
}, WINED3D_GL_EXT_NONE
},
12300 {STATE_TEXTURESTAGE(4, WINED3D_TSS_ALPHA_OP
), {STATE_SHADER(WINED3D_SHADER_TYPE_PIXEL
), NULL
}, WINED3D_GL_EXT_NONE
},
12301 {STATE_TEXTURESTAGE(4, WINED3D_TSS_ALPHA_ARG1
), {STATE_SHADER(WINED3D_SHADER_TYPE_PIXEL
), NULL
}, WINED3D_GL_EXT_NONE
},
12302 {STATE_TEXTURESTAGE(4, WINED3D_TSS_ALPHA_ARG2
), {STATE_SHADER(WINED3D_SHADER_TYPE_PIXEL
), NULL
}, WINED3D_GL_EXT_NONE
},
12303 {STATE_TEXTURESTAGE(4, WINED3D_TSS_ALPHA_ARG0
), {STATE_SHADER(WINED3D_SHADER_TYPE_PIXEL
), NULL
}, WINED3D_GL_EXT_NONE
},
12304 {STATE_TEXTURESTAGE(4, WINED3D_TSS_RESULT_ARG
), {STATE_SHADER(WINED3D_SHADER_TYPE_PIXEL
), NULL
}, WINED3D_GL_EXT_NONE
},
12305 {STATE_TEXTURESTAGE(5, WINED3D_TSS_COLOR_OP
), {STATE_SHADER(WINED3D_SHADER_TYPE_PIXEL
), NULL
}, WINED3D_GL_EXT_NONE
},
12306 {STATE_TEXTURESTAGE(5, WINED3D_TSS_COLOR_ARG1
), {STATE_SHADER(WINED3D_SHADER_TYPE_PIXEL
), NULL
}, WINED3D_GL_EXT_NONE
},
12307 {STATE_TEXTURESTAGE(5, WINED3D_TSS_COLOR_ARG2
), {STATE_SHADER(WINED3D_SHADER_TYPE_PIXEL
), NULL
}, WINED3D_GL_EXT_NONE
},
12308 {STATE_TEXTURESTAGE(5, WINED3D_TSS_COLOR_ARG0
), {STATE_SHADER(WINED3D_SHADER_TYPE_PIXEL
), NULL
}, WINED3D_GL_EXT_NONE
},
12309 {STATE_TEXTURESTAGE(5, WINED3D_TSS_ALPHA_OP
), {STATE_SHADER(WINED3D_SHADER_TYPE_PIXEL
), NULL
}, WINED3D_GL_EXT_NONE
},
12310 {STATE_TEXTURESTAGE(5, WINED3D_TSS_ALPHA_ARG1
), {STATE_SHADER(WINED3D_SHADER_TYPE_PIXEL
), NULL
}, WINED3D_GL_EXT_NONE
},
12311 {STATE_TEXTURESTAGE(5, WINED3D_TSS_ALPHA_ARG2
), {STATE_SHADER(WINED3D_SHADER_TYPE_PIXEL
), NULL
}, WINED3D_GL_EXT_NONE
},
12312 {STATE_TEXTURESTAGE(5, WINED3D_TSS_ALPHA_ARG0
), {STATE_SHADER(WINED3D_SHADER_TYPE_PIXEL
), NULL
}, WINED3D_GL_EXT_NONE
},
12313 {STATE_TEXTURESTAGE(5, WINED3D_TSS_RESULT_ARG
), {STATE_SHADER(WINED3D_SHADER_TYPE_PIXEL
), NULL
}, WINED3D_GL_EXT_NONE
},
12314 {STATE_TEXTURESTAGE(6, WINED3D_TSS_COLOR_OP
), {STATE_SHADER(WINED3D_SHADER_TYPE_PIXEL
), NULL
}, WINED3D_GL_EXT_NONE
},
12315 {STATE_TEXTURESTAGE(6, WINED3D_TSS_COLOR_ARG1
), {STATE_SHADER(WINED3D_SHADER_TYPE_PIXEL
), NULL
}, WINED3D_GL_EXT_NONE
},
12316 {STATE_TEXTURESTAGE(6, WINED3D_TSS_COLOR_ARG2
), {STATE_SHADER(WINED3D_SHADER_TYPE_PIXEL
), NULL
}, WINED3D_GL_EXT_NONE
},
12317 {STATE_TEXTURESTAGE(6, WINED3D_TSS_COLOR_ARG0
), {STATE_SHADER(WINED3D_SHADER_TYPE_PIXEL
), NULL
}, WINED3D_GL_EXT_NONE
},
12318 {STATE_TEXTURESTAGE(6, WINED3D_TSS_ALPHA_OP
), {STATE_SHADER(WINED3D_SHADER_TYPE_PIXEL
), NULL
}, WINED3D_GL_EXT_NONE
},
12319 {STATE_TEXTURESTAGE(6, WINED3D_TSS_ALPHA_ARG1
), {STATE_SHADER(WINED3D_SHADER_TYPE_PIXEL
), NULL
}, WINED3D_GL_EXT_NONE
},
12320 {STATE_TEXTURESTAGE(6, WINED3D_TSS_ALPHA_ARG2
), {STATE_SHADER(WINED3D_SHADER_TYPE_PIXEL
), NULL
}, WINED3D_GL_EXT_NONE
},
12321 {STATE_TEXTURESTAGE(6, WINED3D_TSS_ALPHA_ARG0
), {STATE_SHADER(WINED3D_SHADER_TYPE_PIXEL
), NULL
}, WINED3D_GL_EXT_NONE
},
12322 {STATE_TEXTURESTAGE(6, WINED3D_TSS_RESULT_ARG
), {STATE_SHADER(WINED3D_SHADER_TYPE_PIXEL
), NULL
}, WINED3D_GL_EXT_NONE
},
12323 {STATE_TEXTURESTAGE(7, WINED3D_TSS_COLOR_OP
), {STATE_SHADER(WINED3D_SHADER_TYPE_PIXEL
), NULL
}, WINED3D_GL_EXT_NONE
},
12324 {STATE_TEXTURESTAGE(7, WINED3D_TSS_COLOR_ARG1
), {STATE_SHADER(WINED3D_SHADER_TYPE_PIXEL
), NULL
}, WINED3D_GL_EXT_NONE
},
12325 {STATE_TEXTURESTAGE(7, WINED3D_TSS_COLOR_ARG2
), {STATE_SHADER(WINED3D_SHADER_TYPE_PIXEL
), NULL
}, WINED3D_GL_EXT_NONE
},
12326 {STATE_TEXTURESTAGE(7, WINED3D_TSS_COLOR_ARG0
), {STATE_SHADER(WINED3D_SHADER_TYPE_PIXEL
), NULL
}, WINED3D_GL_EXT_NONE
},
12327 {STATE_TEXTURESTAGE(7, WINED3D_TSS_ALPHA_OP
), {STATE_SHADER(WINED3D_SHADER_TYPE_PIXEL
), NULL
}, WINED3D_GL_EXT_NONE
},
12328 {STATE_TEXTURESTAGE(7, WINED3D_TSS_ALPHA_ARG1
), {STATE_SHADER(WINED3D_SHADER_TYPE_PIXEL
), NULL
}, WINED3D_GL_EXT_NONE
},
12329 {STATE_TEXTURESTAGE(7, WINED3D_TSS_ALPHA_ARG2
), {STATE_SHADER(WINED3D_SHADER_TYPE_PIXEL
), NULL
}, WINED3D_GL_EXT_NONE
},
12330 {STATE_TEXTURESTAGE(7, WINED3D_TSS_ALPHA_ARG0
), {STATE_SHADER(WINED3D_SHADER_TYPE_PIXEL
), NULL
}, WINED3D_GL_EXT_NONE
},
12331 {STATE_TEXTURESTAGE(7, WINED3D_TSS_RESULT_ARG
), {STATE_SHADER(WINED3D_SHADER_TYPE_PIXEL
), NULL
}, WINED3D_GL_EXT_NONE
},
12332 {STATE_SHADER(WINED3D_SHADER_TYPE_PIXEL
), {STATE_SHADER(WINED3D_SHADER_TYPE_PIXEL
), glsl_fragment_pipe_shader
}, WINED3D_GL_EXT_NONE
},
12333 {STATE_RENDER(WINED3D_RS_ALPHAFUNC
), {STATE_RENDER(WINED3D_RS_ALPHAFUNC
), glsl_fragment_pipe_alpha_test_func
}, WINED3D_GL_LEGACY_CONTEXT
},
12334 {STATE_RENDER(WINED3D_RS_ALPHAFUNC
), {STATE_RENDER(WINED3D_RS_ALPHATESTENABLE
), NULL
}, WINED3D_GL_EXT_NONE
},
12335 {STATE_RENDER(WINED3D_RS_ALPHAREF
), {STATE_RENDER(WINED3D_RS_ALPHAFUNC
), NULL
}, WINED3D_GL_LEGACY_CONTEXT
},
12336 {STATE_RENDER(WINED3D_RS_ALPHAREF
), {STATE_RENDER(WINED3D_RS_ALPHAREF
), glsl_fragment_pipe_core_alpha_test_ref
}, WINED3D_GL_EXT_NONE
},
12337 {STATE_RENDER(WINED3D_RS_ALPHATESTENABLE
), {STATE_RENDER(WINED3D_RS_ALPHATESTENABLE
), glsl_fragment_pipe_alpha_test
}, WINED3D_GL_LEGACY_CONTEXT
},
12338 {STATE_RENDER(WINED3D_RS_ALPHATESTENABLE
), {STATE_RENDER(WINED3D_RS_ALPHATESTENABLE
), glsl_fragment_pipe_core_alpha_test
}, WINED3D_GL_EXT_NONE
},
12339 {STATE_RENDER(WINED3D_RS_COLORKEYENABLE
), {STATE_SHADER(WINED3D_SHADER_TYPE_PIXEL
), NULL
}, WINED3D_GL_EXT_NONE
},
12340 {STATE_COLOR_KEY
, { STATE_COLOR_KEY
, glsl_fragment_pipe_color_key
}, WINED3D_GL_EXT_NONE
},
12341 {STATE_RENDER(WINED3D_RS_FOGENABLE
), {STATE_RENDER(WINED3D_RS_FOGENABLE
), glsl_fragment_pipe_fog
}, WINED3D_GL_EXT_NONE
},
12342 {STATE_RENDER(WINED3D_RS_FOGTABLEMODE
), {STATE_RENDER(WINED3D_RS_FOGENABLE
), NULL
}, WINED3D_GL_EXT_NONE
},
12343 {STATE_RENDER(WINED3D_RS_FOGVERTEXMODE
), {STATE_RENDER(WINED3D_RS_FOGENABLE
), NULL
}, WINED3D_GL_EXT_NONE
},
12344 {STATE_RENDER(WINED3D_RS_FOGSTART
), {STATE_RENDER(WINED3D_RS_FOGSTART
), glsl_fragment_pipe_fogparams
}, WINED3D_GL_EXT_NONE
},
12345 {STATE_RENDER(WINED3D_RS_FOGEND
), {STATE_RENDER(WINED3D_RS_FOGSTART
), NULL
}, WINED3D_GL_EXT_NONE
},
12346 {STATE_RENDER(WINED3D_RS_SRGBWRITEENABLE
), {STATE_RENDER(WINED3D_RS_SRGBWRITEENABLE
), state_srgbwrite
}, ARB_FRAMEBUFFER_SRGB
},
12347 {STATE_RENDER(WINED3D_RS_SRGBWRITEENABLE
), {STATE_SHADER(WINED3D_SHADER_TYPE_PIXEL
), NULL
}, WINED3D_GL_EXT_NONE
},
12348 {STATE_RENDER(WINED3D_RS_FOGCOLOR
), {STATE_RENDER(WINED3D_RS_FOGCOLOR
), glsl_fragment_pipe_fogparams
}, WINED3D_GL_EXT_NONE
},
12349 {STATE_RENDER(WINED3D_RS_FOGDENSITY
), {STATE_RENDER(WINED3D_RS_FOGDENSITY
), glsl_fragment_pipe_fogparams
}, WINED3D_GL_EXT_NONE
},
12350 {STATE_RENDER(WINED3D_RS_POINTSPRITEENABLE
), {STATE_RENDER(WINED3D_RS_POINTSPRITEENABLE
), glsl_fragment_pipe_shader
}, ARB_POINT_SPRITE
},
12351 {STATE_RENDER(WINED3D_RS_POINTSPRITEENABLE
), {STATE_RENDER(WINED3D_RS_POINTSPRITEENABLE
), glsl_fragment_pipe_shader
}, WINED3D_GL_VERSION_2_0
},
12352 {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
},
12353 {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
},
12354 {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
},
12355 {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
},
12356 {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
},
12357 {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
},
12358 {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
},
12359 {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
},
12360 {STATE_TEXTURESTAGE(0, WINED3D_TSS_CONSTANT
), {STATE_TEXTURESTAGE(0, WINED3D_TSS_CONSTANT
), glsl_fragment_pipe_invalidate_constants
}, WINED3D_GL_EXT_NONE
},
12361 {STATE_TEXTURESTAGE(1, WINED3D_TSS_CONSTANT
), {STATE_TEXTURESTAGE(1, WINED3D_TSS_CONSTANT
), glsl_fragment_pipe_invalidate_constants
}, WINED3D_GL_EXT_NONE
},
12362 {STATE_TEXTURESTAGE(2, WINED3D_TSS_CONSTANT
), {STATE_TEXTURESTAGE(2, WINED3D_TSS_CONSTANT
), glsl_fragment_pipe_invalidate_constants
}, WINED3D_GL_EXT_NONE
},
12363 {STATE_TEXTURESTAGE(3, WINED3D_TSS_CONSTANT
), {STATE_TEXTURESTAGE(3, WINED3D_TSS_CONSTANT
), glsl_fragment_pipe_invalidate_constants
}, WINED3D_GL_EXT_NONE
},
12364 {STATE_TEXTURESTAGE(4, WINED3D_TSS_CONSTANT
), {STATE_TEXTURESTAGE(4, WINED3D_TSS_CONSTANT
), glsl_fragment_pipe_invalidate_constants
}, WINED3D_GL_EXT_NONE
},
12365 {STATE_TEXTURESTAGE(5, WINED3D_TSS_CONSTANT
), {STATE_TEXTURESTAGE(5, WINED3D_TSS_CONSTANT
), glsl_fragment_pipe_invalidate_constants
}, WINED3D_GL_EXT_NONE
},
12366 {STATE_TEXTURESTAGE(6, WINED3D_TSS_CONSTANT
), {STATE_TEXTURESTAGE(6, WINED3D_TSS_CONSTANT
), glsl_fragment_pipe_invalidate_constants
}, WINED3D_GL_EXT_NONE
},
12367 {STATE_TEXTURESTAGE(7, WINED3D_TSS_CONSTANT
), {STATE_TEXTURESTAGE(7, WINED3D_TSS_CONSTANT
), glsl_fragment_pipe_invalidate_constants
}, WINED3D_GL_EXT_NONE
},
12368 {STATE_RENDER(WINED3D_RS_SPECULARENABLE
), {STATE_RENDER(WINED3D_RS_SPECULARENABLE
), glsl_fragment_pipe_invalidate_constants
}, WINED3D_GL_EXT_NONE
},
12369 {STATE_POINT_ENABLE
, {STATE_POINT_ENABLE
, glsl_fragment_pipe_shader
}, WINED3D_GL_EXT_NONE
},
12370 {STATE_RENDER(WINED3D_RS_SHADEMODE
), {STATE_RENDER(WINED3D_RS_SHADEMODE
), glsl_fragment_pipe_shademode
}, WINED3D_GLSL_130
},
12371 {STATE_RENDER(WINED3D_RS_SHADEMODE
), {STATE_RENDER(WINED3D_RS_SHADEMODE
), state_shademode
}, WINED3D_GL_EXT_NONE
},
12372 {0 /* Terminate */, {0, 0 }, WINED3D_GL_EXT_NONE
},
12375 static BOOL
glsl_fragment_pipe_alloc_context_data(struct wined3d_context
*context
)
12380 static void glsl_fragment_pipe_free_context_data(struct wined3d_context
*context
)
12384 const struct wined3d_fragment_pipe_ops glsl_fragment_pipe
=
12386 glsl_fragment_pipe_enable
,
12387 glsl_fragment_pipe_get_caps
,
12388 glsl_fragment_pipe_get_emul_mask
,
12389 glsl_fragment_pipe_alloc
,
12390 glsl_fragment_pipe_free
,
12391 glsl_fragment_pipe_alloc_context_data
,
12392 glsl_fragment_pipe_free_context_data
,
12393 shader_glsl_color_fixup_supported
,
12394 glsl_fragment_pipe_state_template
,
12397 struct glsl_blitter_args
12399 GLenum texture_type
;
12400 struct color_fixup_desc fixup
;
12401 unsigned short use_colour_key
;
12404 struct glsl_blitter_program
12406 struct wine_rb_entry entry
;
12407 struct glsl_blitter_args args
;
12411 struct wined3d_glsl_blitter
12413 struct wined3d_blitter blitter
;
12414 struct wined3d_string_buffer_list string_buffers
;
12415 struct wine_rb_tree programs
;
12416 GLuint palette_texture
;
12419 static int glsl_blitter_args_compare(const void *key
, const struct wine_rb_entry
*entry
)
12421 const struct glsl_blitter_args
*a
= key
;
12422 const struct glsl_blitter_args
*b
= &WINE_RB_ENTRY_VALUE(entry
, const struct glsl_blitter_program
, entry
)->args
;
12424 return memcmp(a
, b
, sizeof(*a
));
12427 /* Context activation is done by the caller. */
12428 static void glsl_free_blitter_program(struct wine_rb_entry
*entry
, void *ctx
)
12430 struct glsl_blitter_program
*program
= WINE_RB_ENTRY_VALUE(entry
, struct glsl_blitter_program
, entry
);
12431 struct wined3d_context_gl
*context_gl
= ctx
;
12432 const struct wined3d_gl_info
*gl_info
;
12434 gl_info
= context_gl
->gl_info
;
12435 GL_EXTCALL(glDeleteProgram(program
->id
));
12436 checkGLcall("glDeleteProgram()");
12437 heap_free(program
);
12440 /* Context activation is done by the caller. */
12441 static void glsl_blitter_destroy(struct wined3d_blitter
*blitter
, struct wined3d_context
*context
)
12443 struct wined3d_context_gl
*context_gl
= wined3d_context_gl(context
);
12444 const struct wined3d_gl_info
*gl_info
= context_gl
->gl_info
;
12445 struct wined3d_glsl_blitter
*glsl_blitter
;
12446 struct wined3d_blitter
*next
;
12448 if ((next
= blitter
->next
))
12449 next
->ops
->blitter_destroy(next
, context
);
12451 glsl_blitter
= CONTAINING_RECORD(blitter
, struct wined3d_glsl_blitter
, blitter
);
12453 if (glsl_blitter
->palette_texture
)
12454 gl_info
->gl_ops
.gl
.p_glDeleteTextures(1, &glsl_blitter
->palette_texture
);
12456 wine_rb_destroy(&glsl_blitter
->programs
, glsl_free_blitter_program
, context_gl
);
12457 string_buffer_list_cleanup(&glsl_blitter
->string_buffers
);
12459 heap_free(glsl_blitter
);
12462 static void glsl_blitter_generate_p8_shader(struct wined3d_string_buffer
*buffer
,
12463 const struct wined3d_gl_info
*gl_info
, const struct glsl_blitter_args
*args
,
12464 const char *output
, const char *tex_type
, const char *swizzle
)
12466 shader_addline(buffer
, "uniform sampler1D sampler_palette;\n");
12467 shader_addline(buffer
, "\nvoid main()\n{\n");
12468 /* The alpha-component contains the palette index. */
12469 shader_addline(buffer
, " float index = texture%s(sampler, out_texcoord.%s).%c;\n",
12470 needs_legacy_glsl_syntax(gl_info
) ? tex_type
: "", swizzle
,
12471 gl_info
->supported
[WINED3D_GL_LEGACY_CONTEXT
] ? 'w' : 'x');
12472 /* Scale the index by 255/256 and add a bias of 0.5 in order to sample in
12474 shader_addline(buffer
, " index = (index * 255.0 + 0.5) / 256.0;\n");
12475 shader_addline(buffer
, " %s = texture%s(sampler_palette, index);\n",
12476 output
, needs_legacy_glsl_syntax(gl_info
) ? "1D" : "");
12477 if (args
->use_colour_key
)
12478 shader_glsl_generate_colour_key_test(buffer
, output
, "colour_key.low", "colour_key.high");
12479 shader_addline(buffer
, "}\n");
12482 static void gen_packed_yuv_read(struct wined3d_string_buffer
*buffer
,
12483 const struct wined3d_gl_info
*gl_info
, const struct glsl_blitter_args
*args
,
12484 const char *tex_type
)
12486 enum complex_fixup complex_fixup
= get_complex_fixup(args
->fixup
);
12487 char chroma
, luminance
;
12490 /* The YUY2 and UYVY formats contain two pixels packed into a 32 bit
12491 * macropixel, giving effectively 16 bits per pixel. The color consists of
12492 * a luminance(Y) and two chroma(U and V) values. Each macropixel has two
12493 * luminance values, one for each single pixel it contains, and one U and
12494 * one V value shared between both pixels.
12496 * The data is loaded into an A8L8 texture. With YUY2, the luminance
12497 * component contains the luminance and alpha the chroma. With UYVY it is
12498 * vice versa. Thus take the format into account when generating the read
12501 * Reading the Y value is straightforward - just sample the texture. The
12502 * hardware takes care of filtering in the horizontal and vertical
12505 * Reading the U and V values is harder. We have to avoid filtering
12506 * horizontally, because that would mix the U and V values of one pixel or
12507 * two adjacent pixels. Thus floor the texture coordinate and add 0.5 to
12508 * get an unfiltered read, regardless of the filtering setting. Vertical
12509 * filtering works automatically though - the U and V values of two rows
12510 * are mixed nicely.
12512 * Apart of avoiding filtering issues, the code has to know which value it
12513 * just read, and where it can find the other one. To determine this, it
12514 * checks if it sampled an even or odd pixel, and shifts the 2nd read
12517 * Handling horizontal filtering of U and V values requires reading a 2nd
12518 * pair of pixels, extracting U and V and mixing them. This is not
12521 * An alternative implementation idea is to load the texture as A8R8G8B8
12522 * texture, with width / 2. This way one read gives all 3 values, finding
12523 * U and V is easy in an unfiltered situation. Finding the luminance on
12524 * the other hand requires finding out if it is an odd or even pixel. The
12525 * real drawback of this approach is filtering. This would have to be
12526 * emulated completely in the shader, reading up two 2 packed pixels in up
12527 * to 2 rows and interpolating both horizontally and vertically. Beyond
12528 * that it would require adjustments to the texture handling code to deal
12529 * with the width scaling. */
12531 if (complex_fixup
== COMPLEX_FIXUP_UYVY
)
12534 luminance
= gl_info
->supported
[WINED3D_GL_LEGACY_CONTEXT
] ? 'w' : 'y';
12538 chroma
= gl_info
->supported
[WINED3D_GL_LEGACY_CONTEXT
] ? 'w' : 'y';
12542 tex
= needs_legacy_glsl_syntax(gl_info
) ? tex_type
: "";
12544 /* First we have to read the chroma values. This means we need at least
12545 * two pixels (no filtering), or 4 pixels (with filtering). To get the
12546 * unmodified chroma, we have to rid ourselves of the filtering when we
12547 * sample the texture. */
12548 shader_addline(buffer
, " texcoord.xy = out_texcoord.xy;\n");
12549 /* We must not allow filtering between pixel x and x+1, this would mix U
12550 * and V. Vertical filtering is ok. However, bear in mind that the pixel
12551 * center is at 0.5, so add 0.5. */
12552 shader_addline(buffer
, " texcoord.x = (floor(texcoord.x * size.x) + 0.5) / size.x;\n");
12553 shader_addline(buffer
, " luminance = texture%s(sampler, texcoord.xy).%c;\n", tex
, chroma
);
12555 /* Multiply the x coordinate by 0.5 and get the fraction. This gives 0.25
12556 * and 0.75 for the even and odd pixels respectively. */
12557 /* Put the value into either of the chroma values. */
12558 shader_addline(buffer
, " bool even = fract(texcoord.x * size.x * 0.5) < 0.5;\n");
12559 shader_addline(buffer
, " if (even)\n");
12560 shader_addline(buffer
, " chroma.y = luminance;\n");
12561 shader_addline(buffer
, " else\n");
12562 shader_addline(buffer
, " chroma.x = luminance;\n");
12564 /* Sample pixel 2. If we read an even pixel, sample the pixel right to the
12565 * current one. Otherwise, sample the left pixel. */
12566 shader_addline(buffer
, " texcoord.x += even ? 1.0 / size.x : -1.0 / size.x;\n");
12567 shader_addline(buffer
, " luminance = texture%s(sampler, texcoord.xy).%c;\n", tex
, chroma
);
12569 /* Put the value into the other chroma. */
12570 shader_addline(buffer
, " if (even)\n");
12571 shader_addline(buffer
, " chroma.x = luminance;\n");
12572 shader_addline(buffer
, " else\n");
12573 shader_addline(buffer
, " chroma.y = luminance;\n");
12575 /* TODO: If filtering is enabled, sample a 2nd pair of pixels left or right of
12576 * the current one and lerp the two U and V values. */
12578 /* This gives the correctly filtered luminance value. */
12579 shader_addline(buffer
, " luminance = texture%s(sampler, out_texcoord.xy).%c;\n", tex
, luminance
);
12582 static void gen_yv12_read(struct wined3d_string_buffer
*buffer
,
12583 const struct wined3d_gl_info
*gl_info
, const char *tex_type
)
12585 char component
= gl_info
->supported
[WINED3D_GL_LEGACY_CONTEXT
] ? 'w' : 'x';
12586 const char *tex
= needs_legacy_glsl_syntax(gl_info
) ? tex_type
: "";
12588 /* YV12 surfaces contain a WxH sized luminance plane, followed by a
12589 * (W/2)x(H/2) V and a (W/2)x(H/2) U plane, each with 8 bit per pixel. So
12590 * the effective bitdepth is 12 bits per pixel. Since the U and V planes
12591 * have only half the pitch of the luminance plane, the packing into the
12592 * gl texture is a bit unfortunate. If the whole texture is interpreted as
12593 * luminance data it looks approximately like this:
12595 * +----------------------------------+----
12607 * +----------------+-----------------+----
12609 * | V even rows | V odd rows |
12611 * +----------------+------------------ -
12613 * | U even rows | U odd rows |
12615 * +----------------+-----------------+----
12619 * So it appears as if there are 4 chroma images, but in fact the odd rows
12620 * in the chroma images are in the same row as the even ones. So it is
12621 * kinda tricky to read. */
12623 /* First sample the chroma values. */
12624 shader_addline(buffer
, " texcoord.xy = out_texcoord.xy;\n");
12625 /* The chroma planes have only half the width. */
12626 shader_addline(buffer
, " texcoord.x *= 0.5;\n");
12628 /* The first value is between 2/3 and 5/6 of the texture's height, so
12629 * scale+bias the coordinate. Also read the right side of the image when
12630 * reading odd lines.
12632 * Don't forget to clamp the y values in into the range, otherwise we'll
12633 * get filtering bleeding. */
12635 /* Read odd lines from the right side (add 0.5 to the x coordinate). Keep
12636 * in mind that each line of the chroma plane corresponds to 2 lines of the
12637 * resulting image. */
12638 shader_addline(buffer
, " if (fract(texcoord.y * size.y * 0.25) >= 0.5)\n");
12639 shader_addline(buffer
, " texcoord.x += 0.5;\n");
12641 /* Clamp, keep the half pixel origin in mind. */
12642 shader_addline(buffer
, " texcoord.y = clamp(2.0 / 3.0 + texcoord.y / 6.0, "
12643 "2.0 / 3.0 + 0.5 / size.y, 5.0 / 6.0 - 0.5 / size.y);\n");
12645 shader_addline(buffer
, " chroma.x = texture%s(sampler, texcoord.xy).%c;\n", tex
, component
);
12647 /* The other chroma value is 1/6th of the texture lower, from 5/6th to
12648 * 6/6th No need to clamp because we're just reusing the already clamped
12649 * value from above. */
12650 shader_addline(buffer
, " texcoord.y += 1.0 / 6.0;\n");
12651 shader_addline(buffer
, " chroma.y = texture%s(sampler, texcoord.xy).%c;\n", tex
, component
);
12653 /* Sample the luminance value. It is in the top 2/3rd of the texture, so
12654 * scale the y coordinate. Clamp the y coordinate to prevent the chroma
12655 * values from bleeding into the sampled luminance values due to
12657 shader_addline(buffer
, " texcoord.xy = out_texcoord.xy;\n");
12658 /* Multiply the y coordinate by 2/3 and clamp it. */
12659 shader_addline(buffer
, " texcoord.y = min(texcoord.y * 2.0 / 3.0, 2.0 / 3.0 - 0.5 / size.y);\n");
12660 shader_addline(buffer
, " luminance = texture%s(sampler, texcoord.xy).%c;\n", tex
, component
);
12663 static void gen_nv12_read(struct wined3d_string_buffer
*buffer
,
12664 const struct wined3d_gl_info
*gl_info
, const char *tex_type
)
12666 char component
= gl_info
->supported
[WINED3D_GL_LEGACY_CONTEXT
] ? 'w' : 'x';
12667 const char *tex
= needs_legacy_glsl_syntax(gl_info
) ? tex_type
: "";
12669 /* NV12 surfaces contain a WxH sized luminance plane, followed by a
12670 * (W/2)x(H/2) sized plane where each component is an UV pair. So the
12671 * effective bitdepth is 12 bits per pixel. If the whole texture is
12672 * interpreted as luminance data it looks approximately like this:
12674 * +----------------------------------+----
12686 * +----------------------------------+----
12687 * |UVUVUVUVUVUVUVUVUVUVUVUVUVUVUVUVUV|
12688 * |UVUVUVUVUVUVUVUVUVUVUVUVUVUVUVUVUV|
12694 * +----------------------------------+---- */
12696 /* First sample the chroma values. */
12697 shader_addline(buffer
, " texcoord.xy = out_texcoord.xy;\n");
12698 /* We only have half the number of chroma pixels. */
12699 shader_addline(buffer
, " texcoord.x *= 0.5;\n");
12700 shader_addline(buffer
, " texcoord.y = (texcoord.y + 2.0) / 3.0;\n");
12702 /* We must not allow filtering horizontally, this would mix U and V.
12703 * Vertical filtering is ok. However, bear in mind that the pixel center
12704 * is at 0.5, so add 0.5. */
12706 /* Convert to non-normalised coordinates so we can find the individual
12708 shader_addline(buffer
, " texcoord.x = floor(texcoord.x * size.x);\n");
12709 /* Multiply by 2 since chroma components are stored in UV pixel pairs, add
12710 * 0.5 to hit the center of the pixel. Then convert back to normalised
12712 shader_addline(buffer
, " texcoord.x = (texcoord.x * 2.0 + 0.5) / size.x;\n");
12713 /* Clamp, keep the half pixel origin in mind. */
12714 shader_addline(buffer
, " texcoord.y = max(texcoord.y, 2.0 / 3.0 + 0.5 / size.y);\n");
12716 shader_addline(buffer
, " chroma.y = texture%s(sampler, texcoord.xy).%c;\n", tex
, component
);
12717 /* Add 1.0 / size.x to sample the adjacent texel. */
12718 shader_addline(buffer
, " texcoord.x += 1.0 / size.x;\n");
12719 shader_addline(buffer
, " chroma.x = texture%s(sampler, texcoord.xy).%c;\n", tex
, component
);
12721 /* Sample the luminance value. It is in the top 2/3rd of the texture, so
12722 * scale the y coordinate. Clamp the y coordinate to prevent the chroma
12723 * values from bleeding into the sampled luminance values due to
12725 shader_addline(buffer
, " texcoord.xy = out_texcoord.xy;\n");
12726 /* Multiply the y coordinate by 2/3 and clamp it. */
12727 shader_addline(buffer
, " texcoord.y = min(texcoord.y * 2.0 / 3.0, 2.0 / 3.0 - 0.5 / size.y);\n");
12728 shader_addline(buffer
, " luminance = texture%s(sampler, texcoord.xy).%c;\n", tex
, component
);
12731 static void glsl_blitter_generate_yuv_shader(struct wined3d_string_buffer
*buffer
,
12732 const struct wined3d_gl_info
*gl_info
, const struct glsl_blitter_args
*args
,
12733 const char *output
, const char *tex_type
, const char *swizzle
)
12735 enum complex_fixup complex_fixup
= get_complex_fixup(args
->fixup
);
12737 shader_addline(buffer
, "const vec4 yuv_coef = vec4(1.403, -0.344, -0.714, 1.770);\n");
12738 shader_addline(buffer
, "float luminance;\n");
12739 shader_addline(buffer
, "vec2 texcoord;\n");
12740 shader_addline(buffer
, "vec2 chroma;\n");
12741 shader_addline(buffer
, "uniform vec2 size;\n");
12743 shader_addline(buffer
, "\nvoid main()\n{\n");
12745 switch (complex_fixup
)
12747 case COMPLEX_FIXUP_UYVY
:
12748 case COMPLEX_FIXUP_YUY2
:
12749 gen_packed_yuv_read(buffer
, gl_info
, args
, tex_type
);
12752 case COMPLEX_FIXUP_YV12
:
12753 gen_yv12_read(buffer
, gl_info
, tex_type
);
12756 case COMPLEX_FIXUP_NV12
:
12757 gen_nv12_read(buffer
, gl_info
, tex_type
);
12760 case COMPLEX_FIXUP_YUV
:
12761 /* With APPLE_rgb_422, things are much simpler. The only thing we
12762 * have to do here is Y'CbCr to RGB conversion. */
12763 shader_addline(buffer
, " vec3 yuv = vec3(texture%s(sampler, out_texcoord.xy));\n",
12764 needs_legacy_glsl_syntax(gl_info
) ? tex_type
: "");
12765 shader_addline(buffer
, " luminance = yuv.y;\n");
12766 shader_addline(buffer
, " chroma = yuv.xz;\n");
12770 FIXME("Unsupported fixup %#x.\n", complex_fixup
);
12771 string_buffer_free(buffer
);
12775 /* Calculate the final result. Formula is taken from
12776 * http://www.fourcc.org/fccyvrgb.php. Note that the chroma
12777 * ranges from -0.5 to 0.5. */
12778 shader_addline(buffer
, "\n chroma.xy -= 0.5;\n");
12780 shader_addline(buffer
, " %s.x = luminance + chroma.x * yuv_coef.x;\n", output
);
12781 shader_addline(buffer
, " %s.y = luminance + chroma.y * yuv_coef.y + chroma.x * yuv_coef.z;\n", output
);
12782 shader_addline(buffer
, " %s.z = luminance + chroma.y * yuv_coef.w;\n", output
);
12783 if (args
->use_colour_key
)
12784 shader_glsl_generate_colour_key_test(buffer
, output
, "colour_key.low", "colour_key.high");
12786 shader_addline(buffer
, "}\n");
12789 static void glsl_blitter_generate_plain_shader(struct wined3d_string_buffer
*buffer
,
12790 const struct wined3d_gl_info
*gl_info
, const struct glsl_blitter_args
*args
,
12791 const char *output
, const char *tex_type
, const char *swizzle
)
12793 shader_addline(buffer
, "\nvoid main()\n{\n");
12794 shader_addline(buffer
, " %s = texture%s(sampler, out_texcoord.%s);\n",
12795 output
, needs_legacy_glsl_syntax(gl_info
) ? tex_type
: "", swizzle
);
12796 shader_glsl_color_correction_ext(buffer
, output
, WINED3DSP_WRITEMASK_ALL
, args
->fixup
);
12797 if (args
->use_colour_key
)
12798 shader_glsl_generate_colour_key_test(buffer
, output
, "colour_key.low", "colour_key.high");
12799 shader_addline(buffer
, "}\n");
12802 /* Context activation is done by the caller. */
12803 static GLuint
glsl_blitter_generate_program(struct wined3d_glsl_blitter
*blitter
,
12804 const struct wined3d_gl_info
*gl_info
, const struct glsl_blitter_args
*args
)
12806 static const struct
12808 GLenum texture_target
;
12809 const char texture_type
[7];
12810 const char texcoord_swizzle
[4];
12814 {GL_TEXTURE_2D
, "2D", "xy"},
12815 {GL_TEXTURE_CUBE_MAP
, "Cube", "xyz"},
12816 {GL_TEXTURE_RECTANGLE_ARB
, "2DRect", "xy"},
12818 static const char vshader_main
[] =
12822 " gl_Position = vec4(pos, 0.0, 1.0);\n"
12823 " out_texcoord = texcoord;\n"
12825 enum complex_fixup complex_fixup
= get_complex_fixup(args
->fixup
);
12826 struct wined3d_string_buffer
*buffer
, *output
;
12827 GLuint program
, vshader_id
, fshader_id
;
12828 const char *tex_type
= NULL
, *swizzle
= NULL
, *ptr
;
12832 for (i
= 0; i
< ARRAY_SIZE(texture_data
); ++i
)
12834 if (args
->texture_type
== texture_data
[i
].texture_target
)
12836 tex_type
= texture_data
[i
].texture_type
;
12837 swizzle
= texture_data
[i
].texcoord_swizzle
;
12841 if (i
== ARRAY_SIZE(texture_data
))
12843 FIXME("Unsupported texture type %#x.\n", args
->texture_type
);
12847 program
= GL_EXTCALL(glCreateProgram());
12849 vshader_id
= GL_EXTCALL(glCreateShader(GL_VERTEX_SHADER
));
12851 buffer
= string_buffer_get(&blitter
->string_buffers
);
12852 shader_glsl_add_version_declaration(buffer
, gl_info
);
12853 shader_addline(buffer
, "%s vec2 pos;\n", get_attribute_keyword(gl_info
));
12854 shader_addline(buffer
, "%s vec3 texcoord;\n", get_attribute_keyword(gl_info
));
12855 declare_out_varying(gl_info
, buffer
, FALSE
, "vec3 out_texcoord;\n");
12856 shader_addline(buffer
, vshader_main
);
12858 ptr
= buffer
->buffer
;
12859 GL_EXTCALL(glShaderSource(vshader_id
, 1, &ptr
, NULL
));
12860 GL_EXTCALL(glAttachShader(program
, vshader_id
));
12861 GL_EXTCALL(glDeleteShader(vshader_id
));
12863 fshader_id
= GL_EXTCALL(glCreateShader(GL_FRAGMENT_SHADER
));
12865 string_buffer_clear(buffer
);
12866 shader_glsl_add_version_declaration(buffer
, gl_info
);
12867 shader_addline(buffer
, "uniform sampler%s sampler;\n", tex_type
);
12868 if (args
->use_colour_key
)
12870 shader_addline(buffer
, "uniform struct\n{\n");
12871 shader_addline(buffer
, " vec4 low, high;\n");
12872 shader_addline(buffer
, "} colour_key;\n");
12874 declare_in_varying(gl_info
, buffer
, FALSE
, "vec3 out_texcoord;\n");
12875 /* TODO: Declare the out variable with the correct type (and put it in the
12876 * blitter args). */
12877 if (!use_legacy_fragment_output(gl_info
))
12878 shader_addline(buffer
, "out vec4 ps_out[1];\n");
12880 output
= string_buffer_get(&blitter
->string_buffers
);
12881 string_buffer_sprintf(output
, "%s[0]", get_fragment_output(gl_info
));
12883 switch (complex_fixup
)
12885 case COMPLEX_FIXUP_P8
:
12886 glsl_blitter_generate_p8_shader(buffer
, gl_info
, args
, output
->buffer
, tex_type
, swizzle
);
12888 case COMPLEX_FIXUP_YUY2
:
12889 case COMPLEX_FIXUP_UYVY
:
12890 case COMPLEX_FIXUP_YV12
:
12891 case COMPLEX_FIXUP_NV12
:
12892 case COMPLEX_FIXUP_YUV
:
12893 glsl_blitter_generate_yuv_shader(buffer
, gl_info
, args
, output
->buffer
, tex_type
, swizzle
);
12895 case COMPLEX_FIXUP_NONE
:
12896 glsl_blitter_generate_plain_shader(buffer
, gl_info
, args
, output
->buffer
, tex_type
, swizzle
);
12899 string_buffer_release(&blitter
->string_buffers
, output
);
12901 ptr
= buffer
->buffer
;
12902 GL_EXTCALL(glShaderSource(fshader_id
, 1, &ptr
, NULL
));
12903 string_buffer_release(&blitter
->string_buffers
, buffer
);
12904 GL_EXTCALL(glAttachShader(program
, fshader_id
));
12905 GL_EXTCALL(glDeleteShader(fshader_id
));
12907 GL_EXTCALL(glBindAttribLocation(program
, 0, "pos"));
12908 GL_EXTCALL(glBindAttribLocation(program
, 1, "texcoord"));
12910 if (!use_legacy_fragment_output(gl_info
))
12911 GL_EXTCALL(glBindFragDataLocation(program
, 0, "ps_out"));
12913 GL_EXTCALL(glCompileShader(vshader_id
));
12914 print_glsl_info_log(gl_info
, vshader_id
, FALSE
);
12915 GL_EXTCALL(glCompileShader(fshader_id
));
12916 print_glsl_info_log(gl_info
, fshader_id
, FALSE
);
12917 GL_EXTCALL(glLinkProgram(program
));
12918 shader_glsl_validate_link(gl_info
, program
);
12920 GL_EXTCALL(glUseProgram(program
));
12921 loc
= GL_EXTCALL(glGetUniformLocation(program
, "sampler"));
12922 GL_EXTCALL(glUniform1i(loc
, 0));
12923 if (complex_fixup
== COMPLEX_FIXUP_P8
)
12925 loc
= GL_EXTCALL(glGetUniformLocation(program
, "sampler_palette"));
12926 GL_EXTCALL(glUniform1i(loc
, 1));
12932 /* Context activation is done by the caller. */
12933 static void glsl_blitter_upload_palette(struct wined3d_glsl_blitter
*blitter
,
12934 struct wined3d_context_gl
*context_gl
, const struct wined3d_texture_gl
*texture_gl
)
12936 const struct wined3d_gl_info
*gl_info
= context_gl
->gl_info
;
12937 const struct wined3d_palette
*palette
;
12939 palette
= texture_gl
->t
.swapchain
? texture_gl
->t
.swapchain
->palette
: NULL
;
12941 if (!blitter
->palette_texture
)
12942 gl_info
->gl_ops
.gl
.p_glGenTextures(1, &blitter
->palette_texture
);
12944 wined3d_context_gl_active_texture(context_gl
, gl_info
, 1);
12945 gl_info
->gl_ops
.gl
.p_glBindTexture(GL_TEXTURE_1D
, blitter
->palette_texture
);
12946 gl_info
->gl_ops
.gl
.p_glTexParameteri(GL_TEXTURE_1D
, GL_TEXTURE_MIN_FILTER
, GL_NEAREST
);
12947 gl_info
->gl_ops
.gl
.p_glTexParameteri(GL_TEXTURE_1D
, GL_TEXTURE_MAG_FILTER
, GL_NEAREST
);
12948 gl_info
->gl_ops
.gl
.p_glTexParameteri(GL_TEXTURE_1D
, GL_TEXTURE_WRAP_S
, GL_CLAMP_TO_EDGE
);
12950 GL_EXTCALL(glBindBuffer(GL_PIXEL_UNPACK_BUFFER
, 0));
12951 checkGLcall("glBindBuffer");
12955 gl_info
->gl_ops
.gl
.p_glTexImage1D(GL_TEXTURE_1D
, 0, GL_RGB
, 256, 0, GL_BGRA
,
12956 GL_UNSIGNED_INT_8_8_8_8_REV
, palette
->colors
);
12960 static const DWORD black
;
12962 FIXME("P8 texture loaded without a palette.\n");
12963 gl_info
->gl_ops
.gl
.p_glTexImage1D(GL_TEXTURE_1D
, 0, GL_RGB
, 1, 0, GL_BGRA
,
12964 GL_UNSIGNED_INT_8_8_8_8_REV
, &black
);
12967 wined3d_context_gl_active_texture(context_gl
, gl_info
, 0);
12970 /* Context activation is done by the caller. */
12971 static struct glsl_blitter_program
*glsl_blitter_get_program(struct wined3d_glsl_blitter
*blitter
,
12972 struct wined3d_context_gl
*context_gl
, const struct wined3d_texture_gl
*texture_gl
, BOOL use_colour_key
)
12974 const struct wined3d_gl_info
*gl_info
= context_gl
->gl_info
;
12975 struct glsl_blitter_program
*program
;
12976 struct glsl_blitter_args args
;
12977 struct wine_rb_entry
*entry
;
12979 memset(&args
, 0, sizeof(args
));
12980 args
.texture_type
= texture_gl
->target
;
12981 args
.fixup
= texture_gl
->t
.resource
.format
->color_fixup
;
12982 args
.use_colour_key
= use_colour_key
;
12984 if ((entry
= wine_rb_get(&blitter
->programs
, &args
)))
12985 return WINE_RB_ENTRY_VALUE(entry
, struct glsl_blitter_program
, entry
);
12987 if (!(program
= heap_alloc(sizeof(*program
))))
12989 ERR("Failed to allocate blitter program memory.\n");
12993 program
->args
= args
;
12994 if (!(program
->id
= glsl_blitter_generate_program(blitter
, gl_info
, &args
)))
12996 WARN("Failed to generate blitter program.\n");
12997 heap_free(program
);
13001 if (wine_rb_put(&blitter
->programs
, &program
->args
, &program
->entry
) == -1)
13003 ERR("Failed to store blitter program.\n");
13004 GL_EXTCALL(glDeleteProgram(program
->id
));
13005 heap_free(program
);
13012 static BOOL
glsl_blitter_supported(enum wined3d_blit_op blit_op
, const struct wined3d_context
*context
,
13013 const struct wined3d_texture_gl
*src_texture
, DWORD src_location
,
13014 const struct wined3d_texture_gl
*dst_texture
, DWORD dst_location
)
13016 const struct wined3d_resource
*src_resource
= &src_texture
->t
.resource
;
13017 const struct wined3d_resource
*dst_resource
= &dst_texture
->t
.resource
;
13018 const struct wined3d_format
*src_format
= src_resource
->format
;
13019 const struct wined3d_format
*dst_format
= dst_resource
->format
;
13020 bool src_ds
, dst_ds
;
13023 if (blit_op
== WINED3D_BLIT_OP_RAW_BLIT
&& dst_format
->id
== src_format
->id
)
13025 src_ds
= src_format
->depth_size
|| src_format
->stencil_size
;
13026 dst_ds
= dst_format
->depth_size
|| dst_format
->stencil_size
;
13027 /* We could in principle support raw depth/stencil <-> colour blits as
13028 * well in some cases, but note that typeless formats like e.g.
13029 * R16_TYPELESS may use a normalised GL format for depth/stencil
13030 * resources, and a floating-point format for colour resources, */
13031 if (src_ds
&& dst_ds
)
13032 blit_op
= WINED3D_BLIT_OP_DEPTH_BLIT
;
13033 else if (!src_ds
&& !dst_ds
)
13034 blit_op
= WINED3D_BLIT_OP_COLOR_BLIT
;
13037 if (blit_op
!= WINED3D_BLIT_OP_COLOR_BLIT
&& blit_op
!= WINED3D_BLIT_OP_COLOR_BLIT_CKEY
13038 && blit_op
!= WINED3D_BLIT_OP_COLOR_BLIT_ALPHATEST
)
13040 TRACE("Unsupported blit_op %#x.\n", blit_op
);
13044 if (src_resource
->type
!= WINED3D_RTYPE_TEXTURE_2D
)
13047 if (src_texture
->target
== GL_TEXTURE_2D_MULTISAMPLE
13048 || src_texture
->target
== GL_TEXTURE_2D_MULTISAMPLE_ARRAY
)
13050 TRACE("Multi-sample source textures not supported.\n");
13054 if (!(dst_resource
->access
& WINED3D_RESOURCE_ACCESS_GPU
))
13056 TRACE("Destination resource does not have GPU access.\n");
13060 /* We don't necessarily want to blit from resources without
13061 * WINED3D_RESOURCE_ACCESS_GPU, but that may be the only way to decompress
13062 * compressed textures. */
13063 decompress
= (src_format
->flags
[WINED3D_GL_RES_TYPE_TEX_2D
] & WINED3DFMT_FLAG_COMPRESSED
)
13064 && !(dst_format
->flags
[WINED3D_GL_RES_TYPE_TEX_2D
] & WINED3DFMT_FLAG_COMPRESSED
);
13065 if (!decompress
&& !(src_resource
->access
& WINED3D_RESOURCE_ACCESS_GPU
))
13067 TRACE("Source resource does not have GPU access.\n");
13071 if (!is_identity_fixup(dst_format
->color_fixup
)
13072 && (dst_format
->id
!= src_format
->id
|| dst_location
!= WINED3D_LOCATION_DRAWABLE
))
13074 TRACE("Destination fixups are not supported.\n");
13078 if (wined3d_settings
.offscreen_rendering_mode
== ORM_FBO
13079 && !((dst_format
->flags
[WINED3D_GL_RES_TYPE_TEX_2D
] & WINED3DFMT_FLAG_FBO_ATTACHABLE
)
13080 || (dst_resource
->bind_flags
& WINED3D_BIND_RENDER_TARGET
)))
13082 TRACE("Destination texture is not FBO attachable.\n");
13086 TRACE("Returning supported.\n");
13090 static DWORD
glsl_blitter_blit(struct wined3d_blitter
*blitter
, enum wined3d_blit_op op
,
13091 struct wined3d_context
*context
, struct wined3d_texture
*src_texture
, unsigned int src_sub_resource_idx
,
13092 DWORD src_location
, const RECT
*src_rect
, struct wined3d_texture
*dst_texture
,
13093 unsigned int dst_sub_resource_idx
, DWORD dst_location
, const RECT
*dst_rect
,
13094 const struct wined3d_color_key
*colour_key
, enum wined3d_texture_filter_type filter
,
13095 const struct wined3d_format
*resolve_format
)
13097 struct wined3d_texture_gl
*src_texture_gl
= wined3d_texture_gl(src_texture
);
13098 struct wined3d_texture_gl
*dst_texture_gl
= wined3d_texture_gl(dst_texture
);
13099 struct wined3d_context_gl
*context_gl
= wined3d_context_gl(context
);
13100 struct wined3d_device
*device
= dst_texture
->resource
.device
;
13101 const struct wined3d_gl_info
*gl_info
= context_gl
->gl_info
;
13102 struct wined3d_texture
*staging_texture
= NULL
;
13103 struct wined3d_glsl_blitter
*glsl_blitter
;
13104 struct wined3d_color_key alpha_test_key
;
13105 struct glsl_blitter_program
*program
;
13106 struct wined3d_blitter
*next
;
13107 unsigned int src_level
;
13111 TRACE("blitter %p, op %#x, context %p, src_texture %p, src_sub_resource_idx %u, src_location %s, "
13112 "src_rect %s, dst_texture %p, dst_sub_resource_idx %u, dst_location %s, dst_rect %s, "
13113 "colour_key %p, filter %s, resolve format %p.\n",
13114 blitter
, op
, context
, src_texture
, src_sub_resource_idx
, wined3d_debug_location(src_location
),
13115 wine_dbgstr_rect(src_rect
), dst_texture
, dst_sub_resource_idx
, wined3d_debug_location(dst_location
),
13116 wine_dbgstr_rect(dst_rect
), colour_key
, debug_d3dtexturefiltertype(filter
), resolve_format
);
13118 if (!glsl_blitter_supported(op
, context
, src_texture_gl
, src_location
, dst_texture_gl
, dst_location
))
13120 if (!(next
= blitter
->next
))
13122 ERR("No blitter to handle blit op %#x.\n", op
);
13123 return dst_location
;
13126 TRACE("Forwarding to blitter %p.\n", next
);
13127 return next
->ops
->blitter_blit(next
, op
, context
, src_texture
, src_sub_resource_idx
, src_location
,
13128 src_rect
, dst_texture
, dst_sub_resource_idx
, dst_location
, dst_rect
, colour_key
, filter
,
13132 glsl_blitter
= CONTAINING_RECORD(blitter
, struct wined3d_glsl_blitter
, blitter
);
13134 if (!(src_texture
->resource
.access
& WINED3D_RESOURCE_ACCESS_GPU
))
13136 struct wined3d_resource_desc desc
;
13137 struct wined3d_box upload_box
;
13140 TRACE("Source texture is not GPU accessible, creating a staging texture.\n");
13142 src_level
= src_sub_resource_idx
% src_texture
->level_count
;
13143 desc
.resource_type
= WINED3D_RTYPE_TEXTURE_2D
;
13144 desc
.format
= src_texture
->resource
.format
->id
;
13145 desc
.multisample_type
= src_texture
->resource
.multisample_type
;
13146 desc
.multisample_quality
= src_texture
->resource
.multisample_quality
;
13147 desc
.usage
= WINED3DUSAGE_PRIVATE
;
13148 desc
.bind_flags
= 0;
13149 desc
.access
= WINED3D_RESOURCE_ACCESS_GPU
;
13150 desc
.width
= wined3d_texture_get_level_width(src_texture
, src_level
);
13151 desc
.height
= wined3d_texture_get_level_height(src_texture
, src_level
);
13155 if (FAILED(hr
= wined3d_texture_create(device
, &desc
, 1, 1, 0,
13156 NULL
, NULL
, &wined3d_null_parent_ops
, &staging_texture
)))
13158 ERR("Failed to create staging texture, hr %#x.\n", hr
);
13159 return dst_location
;
13162 wined3d_box_set(&upload_box
, 0, 0, desc
.width
, desc
.height
, 0, desc
.depth
);
13163 wined3d_texture_upload_from_texture(staging_texture
, 0, 0, 0, 0,
13164 src_texture
, src_sub_resource_idx
, &upload_box
);
13166 src_texture
= staging_texture
;
13167 src_texture_gl
= wined3d_texture_gl(src_texture
);
13168 src_sub_resource_idx
= 0;
13170 else if (wined3d_settings
.offscreen_rendering_mode
!= ORM_FBO
13171 && (src_texture
->sub_resources
[src_sub_resource_idx
].locations
13172 & (WINED3D_LOCATION_TEXTURE_RGB
| WINED3D_LOCATION_DRAWABLE
)) == WINED3D_LOCATION_DRAWABLE
13173 && !wined3d_resource_is_offscreen(&src_texture
->resource
))
13176 /* Without FBO blits transferring from the drawable to the texture is
13177 * expensive, because we have to flip the data in sysmem. Since we can
13178 * flip in the blitter, we don't actually need that flip anyway. So we
13179 * use the surface's texture as scratch texture, and flip the source
13180 * rectangle instead. */
13181 texture2d_load_fb_texture(src_texture_gl
, src_sub_resource_idx
, FALSE
, context
);
13184 src_level
= src_sub_resource_idx
% src_texture
->level_count
;
13185 s
.top
= wined3d_texture_get_level_height(src_texture
, src_level
) - s
.top
;
13186 s
.bottom
= wined3d_texture_get_level_height(src_texture
, src_level
) - s
.bottom
;
13191 wined3d_texture_load(src_texture
, context
, FALSE
);
13194 if (wined3d_texture_is_full_rect(dst_texture
, dst_sub_resource_idx
% dst_texture
->level_count
, dst_rect
))
13195 wined3d_texture_prepare_location(dst_texture
, dst_sub_resource_idx
, context
, dst_location
);
13197 wined3d_texture_load_location(dst_texture
, dst_sub_resource_idx
, context
, dst_location
);
13199 wined3d_context_gl_apply_blit_state(context_gl
, device
);
13201 if (dst_location
== WINED3D_LOCATION_DRAWABLE
)
13204 wined3d_texture_translate_drawable_coords(dst_texture
, context_gl
->window
, &d
);
13208 if (wined3d_settings
.offscreen_rendering_mode
== ORM_FBO
)
13212 if (dst_location
== WINED3D_LOCATION_DRAWABLE
)
13214 TRACE("Destination texture %p is onscreen.\n", dst_texture
);
13215 buffer
= wined3d_texture_get_gl_buffer(dst_texture
);
13219 TRACE("Destination texture %p is offscreen.\n", dst_texture
);
13220 buffer
= GL_COLOR_ATTACHMENT0
;
13222 wined3d_context_gl_apply_fbo_state_blit(context_gl
, GL_DRAW_FRAMEBUFFER
,
13223 &dst_texture
->resource
, dst_sub_resource_idx
, NULL
, 0, dst_location
);
13224 wined3d_context_gl_set_draw_buffer(context_gl
, buffer
);
13225 wined3d_context_gl_check_fbo_status(context_gl
, GL_DRAW_FRAMEBUFFER
);
13226 context_invalidate_state(context
, STATE_FRAMEBUFFER
);
13229 if (op
== WINED3D_BLIT_OP_COLOR_BLIT_ALPHATEST
)
13231 const struct wined3d_format
*f
= src_texture
->resource
.format
;
13233 alpha_test_key
.color_space_low_value
= 0;
13234 alpha_test_key
.color_space_high_value
= ~(wined3d_mask_from_size(f
->alpha_size
) << f
->alpha_offset
);
13235 colour_key
= &alpha_test_key
;
13237 else if (op
!= WINED3D_BLIT_OP_COLOR_BLIT_CKEY
)
13242 if (!(program
= glsl_blitter_get_program(glsl_blitter
, context_gl
, src_texture_gl
, !!colour_key
)))
13244 ERR("Failed to get blitter program.\n");
13245 return dst_location
;
13247 GL_EXTCALL(glUseProgram(program
->id
));
13248 switch (get_complex_fixup(program
->args
.fixup
))
13250 case COMPLEX_FIXUP_P8
:
13251 glsl_blitter_upload_palette(glsl_blitter
, context_gl
, src_texture_gl
);
13254 case COMPLEX_FIXUP_YUY2
:
13255 case COMPLEX_FIXUP_UYVY
:
13256 case COMPLEX_FIXUP_YV12
:
13257 case COMPLEX_FIXUP_NV12
:
13258 case COMPLEX_FIXUP_YUV
:
13259 src_level
= src_sub_resource_idx
% src_texture
->level_count
;
13260 location
= GL_EXTCALL(glGetUniformLocation(program
->id
, "size"));
13261 GL_EXTCALL(glUniform2f(location
, wined3d_texture_get_level_pow2_width(src_texture
, src_level
),
13262 wined3d_texture_get_level_pow2_height(src_texture
, src_level
)));
13270 struct wined3d_color float_key
[2];
13272 wined3d_format_get_float_color_key(src_texture
->resource
.format
, colour_key
, float_key
);
13273 location
= GL_EXTCALL(glGetUniformLocation(program
->id
, "colour_key.low"));
13274 GL_EXTCALL(glUniform4fv(location
, 1, &float_key
[0].r
));
13275 location
= GL_EXTCALL(glGetUniformLocation(program
->id
, "colour_key.high"));
13276 GL_EXTCALL(glUniform4fv(location
, 1, &float_key
[1].r
));
13278 wined3d_context_gl_draw_shaded_quad(context_gl
, src_texture_gl
, src_sub_resource_idx
, src_rect
, dst_rect
, filter
);
13279 GL_EXTCALL(glUseProgram(0));
13281 if (dst_texture
->swapchain
&& (dst_texture
->swapchain
->front_buffer
== dst_texture
))
13282 gl_info
->gl_ops
.gl
.p_glFlush();
13284 if (staging_texture
)
13285 wined3d_texture_decref(staging_texture
);
13287 return dst_location
;
13290 static void glsl_blitter_clear(struct wined3d_blitter
*blitter
, struct wined3d_device
*device
,
13291 unsigned int rt_count
, const struct wined3d_fb_state
*fb
, unsigned int rect_count
, const RECT
*clear_rects
,
13292 const RECT
*draw_rect
, DWORD flags
, const struct wined3d_color
*color
, float depth
, DWORD stencil
)
13294 struct wined3d_blitter
*next
;
13296 if ((next
= blitter
->next
))
13297 next
->ops
->blitter_clear(next
, device
, rt_count
, fb
, rect_count
,
13298 clear_rects
, draw_rect
, flags
, color
, depth
, stencil
);
13301 static const struct wined3d_blitter_ops glsl_blitter_ops
=
13303 glsl_blitter_destroy
,
13304 glsl_blitter_clear
,
13308 struct wined3d_blitter
*wined3d_glsl_blitter_create(struct wined3d_blitter
**next
,
13309 const struct wined3d_device
*device
)
13311 const struct wined3d_gl_info
*gl_info
= &device
->adapter
->gl_info
;
13312 struct wined3d_glsl_blitter
*blitter
;
13314 if (device
->shader_backend
!= &glsl_shader_backend
)
13317 if (!gl_info
->supported
[ARB_VERTEX_SHADER
] || !gl_info
->supported
[ARB_FRAGMENT_SHADER
])
13320 if (!(blitter
= heap_alloc(sizeof(*blitter
))))
13322 ERR("Failed to allocate blitter.\n");
13326 TRACE("Created blitter %p.\n", blitter
);
13328 blitter
->blitter
.ops
= &glsl_blitter_ops
;
13329 blitter
->blitter
.next
= *next
;
13330 string_buffer_list_init(&blitter
->string_buffers
);
13331 wine_rb_init(&blitter
->programs
, glsl_blitter_args_compare
);
13332 blitter
->palette_texture
= 0;
13333 *next
= &blitter
->blitter
;