1 /**************************************************************************
3 * Copyright 2007 Tungsten Graphics, Inc., Cedar Park, Texas.
6 * Permission is hereby granted, free of charge, to any person obtaining a
7 * copy of this software and associated documentation files (the
8 * "Software"), to deal in the Software without restriction, including
9 * without limitation the rights to use, copy, modify, merge, publish,
10 * distribute, sub license, and/or sell copies of the Software, and to
11 * permit persons to whom the Software is furnished to do so, subject to
12 * the following conditions:
14 * The above copyright notice and this permission notice (including the
15 * next paragraph) shall be included in all copies or substantial portions
18 * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
19 * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
20 * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
21 * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR
22 * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
23 * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
24 * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
26 **************************************************************************/
28 /* Vertices are just an array of floats, with all the attributes
29 * packed. We currently assume a layout like:
31 * attr[0][0..3] - window position
32 * attr[1..n][0..3] - remaining attributes.
34 * Attributes are assumed to be 4 floats wide but are packed so that
35 * all the enabled attributes run contiguously.
38 #include "util/u_math.h"
39 #include "util/u_memory.h"
40 #include "pipe/p_defines.h"
41 #include "pipe/p_shader_tokens.h"
43 #include "sp_context.h"
45 #include "sp_headers.h"
47 #include "sp_texture.h"
48 #include "sp_tex_sample.h"
51 struct quad_shade_stage
53 struct quad_stage stage
;
54 struct tgsi_sampler samplers
[PIPE_MAX_SAMPLERS
];
55 struct tgsi_exec_machine machine
;
56 struct tgsi_exec_vector
*inputs
, *outputs
;
61 static INLINE
struct quad_shade_stage
*
62 quad_shade_stage(struct quad_stage
*qs
)
64 return (struct quad_shade_stage
*) qs
;
70 * Execute fragment shader for the four fragments in the quad.
74 struct quad_stage
*qs
,
75 struct quad_header
*quad
)
77 struct quad_shade_stage
*qss
= quad_shade_stage( qs
);
78 struct softpipe_context
*softpipe
= qs
->softpipe
;
79 struct tgsi_exec_machine
*machine
= &qss
->machine
;
82 /* Consts do not require 16 byte alignment. */
83 machine
->Consts
= softpipe
->mapped_constants
[PIPE_SHADER_FRAGMENT
];
85 machine
->InterpCoefs
= quad
->coef
;
88 quad
->mask
&= softpipe
->fs
->run( softpipe
->fs
,
95 const ubyte
*sem_name
= softpipe
->fs
->info
.output_semantic_name
;
96 const ubyte
*sem_index
= softpipe
->fs
->info
.output_semantic_index
;
97 const uint n
= qss
->stage
.softpipe
->fs
->info
.num_outputs
;
99 for (i
= 0; i
< n
; i
++) {
100 switch (sem_name
[i
]) {
101 case TGSI_SEMANTIC_COLOR
:
103 uint cbuf
= sem_index
[i
];
104 memcpy(quad
->outputs
.color
[cbuf
],
105 &machine
->Outputs
[i
].xyzw
[0].f
[0],
106 sizeof(quad
->outputs
.color
[0]) );
109 case TGSI_SEMANTIC_POSITION
:
112 for (j
= 0; j
< 4; j
++) {
113 quad
->outputs
.depth
[j
] = machine
->Outputs
[0].xyzw
[2].f
[j
];
123 /* compute Z values now, as in the quad earlyz stage */
124 /* XXX we should really only do this if the earlyz stage is not used */
125 const float fx
= (float) quad
->x0
;
126 const float fy
= (float) quad
->y0
;
127 const float dzdx
= quad
->posCoef
->dadx
[2];
128 const float dzdy
= quad
->posCoef
->dady
[2];
129 const float z0
= quad
->posCoef
->a0
[2] + dzdx
* fx
+ dzdy
* fy
;
131 quad
->outputs
.depth
[0] = z0
;
132 quad
->outputs
.depth
[1] = z0
+ dzdx
;
133 quad
->outputs
.depth
[2] = z0
+ dzdy
;
134 quad
->outputs
.depth
[3] = z0
+ dzdx
+ dzdy
;
137 /* shader may cull fragments */
139 qs
->next
->run( qs
->next
, quad
);
144 * Per-primitive (or per-begin?) setup
146 static void shade_begin(struct quad_stage
*qs
)
148 struct quad_shade_stage
*qss
= quad_shade_stage(qs
);
149 struct softpipe_context
*softpipe
= qs
->softpipe
;
151 unsigned num
= MAX2(softpipe
->num_textures
, softpipe
->num_samplers
);
153 /* set TGSI sampler state that varies */
154 for (i
= 0; i
< num
; i
++) {
155 qss
->samplers
[i
].state
= softpipe
->sampler
[i
];
156 qss
->samplers
[i
].texture
= softpipe
->texture
[i
];
159 softpipe
->fs
->prepare( softpipe
->fs
,
163 qs
->next
->begin(qs
->next
);
167 static void shade_destroy(struct quad_stage
*qs
)
169 struct quad_shade_stage
*qss
= (struct quad_shade_stage
*) qs
;
171 tgsi_exec_machine_free_data(&qss
->machine
);
173 FREE( qss
->outputs
);
178 struct quad_stage
*sp_quad_shade_stage( struct softpipe_context
*softpipe
)
180 struct quad_shade_stage
*qss
= CALLOC_STRUCT(quad_shade_stage
);
183 /* allocate storage for program inputs/outputs, aligned to 16 bytes */
184 qss
->inputs
= MALLOC(PIPE_MAX_ATTRIBS
* sizeof(*qss
->inputs
) + 16);
185 qss
->outputs
= MALLOC(PIPE_MAX_ATTRIBS
* sizeof(*qss
->outputs
) + 16);
186 qss
->machine
.Inputs
= align16(qss
->inputs
);
187 qss
->machine
.Outputs
= align16(qss
->outputs
);
189 qss
->stage
.softpipe
= softpipe
;
190 qss
->stage
.begin
= shade_begin
;
191 qss
->stage
.run
= shade_quad
;
192 qss
->stage
.destroy
= shade_destroy
;
194 /* set TGSI sampler state that's constant */
195 for (i
= 0; i
< PIPE_MAX_SAMPLERS
; i
++) {
196 assert(softpipe
->tex_cache
[i
]);
197 qss
->samplers
[i
].get_samples
= sp_get_samples
;
198 qss
->samplers
[i
].pipe
= &softpipe
->pipe
;
199 qss
->samplers
[i
].cache
= softpipe
->tex_cache
[i
];
202 tgsi_exec_machine_init( &qss
->machine
);