Merge branch 'mesa_7_5_branch' into mesa_7_6_branch
[mesa.git] / src / gallium / drivers / softpipe / sp_quad_fs.c
1 /**************************************************************************
2 *
3 * Copyright 2007 Tungsten Graphics, Inc., Cedar Park, Texas.
4 * All Rights Reserved.
5 * Copyright 2008 VMware, Inc. All rights reserved.
6 *
7 * Permission is hereby granted, free of charge, to any person obtaining a
8 * copy of this software and associated documentation files (the
9 * "Software"), to deal in the Software without restriction, including
10 * without limitation the rights to use, copy, modify, merge, publish,
11 * distribute, sub license, and/or sell copies of the Software, and to
12 * permit persons to whom the Software is furnished to do so, subject to
13 * the following conditions:
14 *
15 * The above copyright notice and this permission notice (including the
16 * next paragraph) shall be included in all copies or substantial portions
17 * of the Software.
18 *
19 * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
20 * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
21 * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
22 * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR
23 * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
24 * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
25 * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
26 *
27 **************************************************************************/
28
29 /* Vertices are just an array of floats, with all the attributes
30 * packed. We currently assume a layout like:
31 *
32 * attr[0][0..3] - window position
33 * attr[1..n][0..3] - remaining attributes.
34 *
35 * Attributes are assumed to be 4 floats wide but are packed so that
36 * all the enabled attributes run contiguously.
37 */
38
39 #include "util/u_math.h"
40 #include "util/u_memory.h"
41 #include "pipe/p_defines.h"
42 #include "pipe/p_shader_tokens.h"
43
44 #include "sp_context.h"
45 #include "sp_state.h"
46 #include "sp_quad.h"
47 #include "sp_quad_pipe.h"
48 #include "sp_texture.h"
49 #include "sp_tex_sample.h"
50
51
52 struct quad_shade_stage
53 {
54 struct quad_stage stage; /**< base class */
55 struct tgsi_exec_machine *machine;
56 struct tgsi_exec_vector *inputs, *outputs;
57 };
58
59
60 /** cast wrapper */
61 static INLINE struct quad_shade_stage *
62 quad_shade_stage(struct quad_stage *qs)
63 {
64 return (struct quad_shade_stage *) qs;
65 }
66
67
68 /**
69 * Execute fragment shader for the four fragments in the quad.
70 */
71 static void
72 shade_quad(struct quad_stage *qs, struct quad_header *quad)
73 {
74 struct quad_shade_stage *qss = quad_shade_stage( qs );
75 struct softpipe_context *softpipe = qs->softpipe;
76 struct tgsi_exec_machine *machine = qss->machine;
77 boolean z_written;
78
79 /* Consts do not require 16 byte alignment. */
80 machine->Consts = softpipe->mapped_constants[PIPE_SHADER_FRAGMENT];
81
82 machine->InterpCoefs = quad->coef;
83
84 /* run shader */
85 quad->inout.mask &= softpipe->fs->run( softpipe->fs, machine, quad );
86
87 /* store outputs */
88 z_written = FALSE;
89 {
90 const ubyte *sem_name = softpipe->fs->info.output_semantic_name;
91 const ubyte *sem_index = softpipe->fs->info.output_semantic_index;
92 const uint n = qss->stage.softpipe->fs->info.num_outputs;
93 uint i;
94 for (i = 0; i < n; i++) {
95 switch (sem_name[i]) {
96 case TGSI_SEMANTIC_COLOR:
97 {
98 uint cbuf = sem_index[i];
99 memcpy(quad->output.color[cbuf],
100 &machine->Outputs[i].xyzw[0].f[0],
101 sizeof(quad->output.color[0]) );
102 }
103 break;
104 case TGSI_SEMANTIC_POSITION:
105 {
106 uint j;
107 for (j = 0; j < 4; j++) {
108 quad->output.depth[j] = machine->Outputs[0].xyzw[2].f[j];
109 }
110 z_written = TRUE;
111 }
112 break;
113 }
114 }
115 }
116
117 if (!z_written) {
118 /* compute Z values now, as in the quad earlyz stage */
119 /* XXX we should really only do this if the earlyz stage is not used */
120 const float fx = (float) quad->input.x0;
121 const float fy = (float) quad->input.y0;
122 const float dzdx = quad->posCoef->dadx[2];
123 const float dzdy = quad->posCoef->dady[2];
124 const float z0 = quad->posCoef->a0[2] + dzdx * fx + dzdy * fy;
125
126 quad->output.depth[0] = z0;
127 quad->output.depth[1] = z0 + dzdx;
128 quad->output.depth[2] = z0 + dzdy;
129 quad->output.depth[3] = z0 + dzdx + dzdy;
130 }
131
132 /* shader may cull fragments */
133 if (quad->inout.mask) {
134 qs->next->run( qs->next, quad );
135 }
136 }
137
138
139 /**
140 * Per-primitive (or per-begin?) setup
141 */
142 static void
143 shade_begin(struct quad_stage *qs)
144 {
145 struct quad_shade_stage *qss = quad_shade_stage(qs);
146 struct softpipe_context *softpipe = qs->softpipe;
147
148 softpipe->fs->prepare( softpipe->fs,
149 qss->machine,
150 (struct tgsi_sampler **)
151 softpipe->tgsi.frag_samplers_list );
152
153 qs->next->begin(qs->next);
154 }
155
156
157 static void
158 shade_destroy(struct quad_stage *qs)
159 {
160 struct quad_shade_stage *qss = (struct quad_shade_stage *) qs;
161
162 tgsi_exec_machine_destroy(qss->machine);
163
164 FREE( qs );
165 }
166
167
168 struct quad_stage *
169 sp_quad_shade_stage( struct softpipe_context *softpipe )
170 {
171 struct quad_shade_stage *qss = CALLOC_STRUCT(quad_shade_stage);
172 if (!qss)
173 goto fail;
174
175 qss->stage.softpipe = softpipe;
176 qss->stage.begin = shade_begin;
177 qss->stage.run = shade_quad;
178 qss->stage.destroy = shade_destroy;
179
180 qss->machine = tgsi_exec_machine_create();
181 if (!qss->machine)
182 goto fail;
183
184 return &qss->stage;
185
186 fail:
187 if (qss && qss->machine)
188 tgsi_exec_machine_destroy(qss->machine);
189
190 FREE(qss);
191 return NULL;
192 }