softpipe: add support for explicit early depth testing
[mesa.git] / src / gallium / drivers / softpipe / sp_fs_exec.c
1 /**************************************************************************
2 *
3 * Copyright 2007 VMware, Inc.
4 * All Rights Reserved.
5 *
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:
13 *
14 * The above copyright notice and this permission notice (including the
15 * next paragraph) shall be included in all copies or substantial portions
16 * of the Software.
17 *
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 VMWARE 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.
25 *
26 **************************************************************************/
27
28 /**
29 * Execute fragment shader using the TGSI interpreter.
30 */
31
32 #include "sp_context.h"
33 #include "sp_state.h"
34 #include "sp_fs.h"
35 #include "sp_quad.h"
36
37 #include "pipe/p_state.h"
38 #include "pipe/p_defines.h"
39 #include "util/u_memory.h"
40 #include "tgsi/tgsi_exec.h"
41 #include "tgsi/tgsi_parse.h"
42
43
44 /**
45 * Subclass of sp_fragment_shader_variant
46 */
47 struct sp_exec_fragment_shader
48 {
49 struct sp_fragment_shader_variant base;
50 /* No other members for now */
51 };
52
53
54 /** cast wrapper */
55 static inline struct sp_exec_fragment_shader *
56 sp_exec_fragment_shader(const struct sp_fragment_shader_variant *var)
57 {
58 return (struct sp_exec_fragment_shader *) var;
59 }
60
61
62 static void
63 exec_prepare( const struct sp_fragment_shader_variant *var,
64 struct tgsi_exec_machine *machine,
65 struct tgsi_sampler *sampler )
66 {
67 /*
68 * Bind tokens/shader to the interpreter's machine state.
69 */
70 tgsi_exec_machine_bind_shader(machine,
71 var->tokens,
72 sampler);
73 }
74
75
76
77 /**
78 * Compute quad X,Y,Z,W for the four fragments in a quad.
79 *
80 * This should really be part of the compiled shader.
81 */
82 static void
83 setup_pos_vector(const struct tgsi_interp_coef *coef,
84 float x, float y,
85 struct tgsi_exec_vector *quadpos)
86 {
87 uint chan;
88 /* do X */
89 quadpos->xyzw[0].f[0] = x;
90 quadpos->xyzw[0].f[1] = x + 1;
91 quadpos->xyzw[0].f[2] = x;
92 quadpos->xyzw[0].f[3] = x + 1;
93
94 /* do Y */
95 quadpos->xyzw[1].f[0] = y;
96 quadpos->xyzw[1].f[1] = y;
97 quadpos->xyzw[1].f[2] = y + 1;
98 quadpos->xyzw[1].f[3] = y + 1;
99
100 /* do Z and W for all fragments in the quad */
101 for (chan = 2; chan < 4; chan++) {
102 const float dadx = coef->dadx[chan];
103 const float dady = coef->dady[chan];
104 const float a0 = coef->a0[chan] + dadx * x + dady * y;
105 quadpos->xyzw[chan].f[0] = a0;
106 quadpos->xyzw[chan].f[1] = a0 + dadx;
107 quadpos->xyzw[chan].f[2] = a0 + dady;
108 quadpos->xyzw[chan].f[3] = a0 + dadx + dady;
109 }
110 }
111
112
113 /* TODO: hide the machine struct in here somewhere, remove from this
114 * interface:
115 */
116 static unsigned
117 exec_run( const struct sp_fragment_shader_variant *var,
118 struct tgsi_exec_machine *machine,
119 struct quad_header *quad,
120 bool early_depth_test )
121 {
122 /* Compute X, Y, Z, W vals for this quad */
123 setup_pos_vector(quad->posCoef,
124 (float)quad->input.x0, (float)quad->input.y0,
125 &machine->QuadPos);
126
127 /* convert 0 to 1.0 and 1 to -1.0 */
128 machine->Face = (float) (quad->input.facing * -2 + 1);
129
130 quad->inout.mask &= tgsi_exec_machine_run( machine );
131 if (quad->inout.mask == 0)
132 return FALSE;
133
134 /* store outputs */
135 {
136 const ubyte *sem_name = var->info.output_semantic_name;
137 const ubyte *sem_index = var->info.output_semantic_index;
138 const uint n = var->info.num_outputs;
139 uint i;
140 for (i = 0; i < n; i++) {
141 switch (sem_name[i]) {
142 case TGSI_SEMANTIC_COLOR:
143 {
144 uint cbuf = sem_index[i];
145
146 assert(sizeof(quad->output.color[cbuf]) ==
147 sizeof(machine->Outputs[i]));
148
149 /* copy float[4][4] result */
150 memcpy(quad->output.color[cbuf],
151 &machine->Outputs[i],
152 sizeof(quad->output.color[0]) );
153 }
154 break;
155 case TGSI_SEMANTIC_POSITION:
156 {
157 uint j;
158
159 if (!early_depth_test) {
160 for (j = 0; j < 4; j++)
161 quad->output.depth[j] = machine->Outputs[i].xyzw[2].f[j];
162 }
163 }
164 break;
165 case TGSI_SEMANTIC_STENCIL:
166 {
167 uint j;
168 if (!early_depth_test) {
169 for (j = 0; j < 4; j++)
170 quad->output.stencil[j] = (unsigned)machine->Outputs[i].xyzw[1].u[j];
171 }
172 }
173 break;
174 }
175 }
176 }
177
178 return TRUE;
179 }
180
181
182 static void
183 exec_delete(struct sp_fragment_shader_variant *var,
184 struct tgsi_exec_machine *machine)
185 {
186 if (machine->Tokens == var->tokens) {
187 tgsi_exec_machine_bind_shader(machine, NULL, NULL);
188 }
189
190 FREE( (void *) var->tokens );
191 FREE(var);
192 }
193
194
195 struct sp_fragment_shader_variant *
196 softpipe_create_fs_variant_exec(struct softpipe_context *softpipe)
197 {
198 struct sp_exec_fragment_shader *shader;
199
200 shader = CALLOC_STRUCT(sp_exec_fragment_shader);
201 if (!shader)
202 return NULL;
203
204 shader->base.prepare = exec_prepare;
205 shader->base.run = exec_run;
206 shader->base.delete = exec_delete;
207
208 return &shader->base;
209 }