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 static void
exec_prepare(const struct sp_fragment_shader_variant * var,struct tgsi_exec_machine * machine,struct tgsi_sampler * sampler,struct tgsi_image * image,struct tgsi_buffer * buffer)55 exec_prepare( const struct sp_fragment_shader_variant *var,
56 struct tgsi_exec_machine *machine,
57 struct tgsi_sampler *sampler,
58 struct tgsi_image *image,
59 struct tgsi_buffer *buffer )
60 {
61 /*
62 * Bind tokens/shader to the interpreter's machine state.
63 */
64 tgsi_exec_machine_bind_shader(machine,
65 var->tokens,
66 sampler, image, buffer);
67 }
68
69
70
71 /**
72 * Compute quad X,Y,Z,W for the four fragments in a quad.
73 *
74 * This should really be part of the compiled shader.
75 */
76 static void
setup_pos_vector(const struct tgsi_interp_coef * coef,float x,float y,struct tgsi_exec_vector * quadpos)77 setup_pos_vector(const struct tgsi_interp_coef *coef,
78 float x, float y,
79 struct tgsi_exec_vector *quadpos)
80 {
81 uint chan;
82 /* do X */
83 quadpos->xyzw[0].f[0] = x;
84 quadpos->xyzw[0].f[1] = x + 1;
85 quadpos->xyzw[0].f[2] = x;
86 quadpos->xyzw[0].f[3] = x + 1;
87
88 /* do Y */
89 quadpos->xyzw[1].f[0] = y;
90 quadpos->xyzw[1].f[1] = y;
91 quadpos->xyzw[1].f[2] = y + 1;
92 quadpos->xyzw[1].f[3] = y + 1;
93
94 /* do Z and W for all fragments in the quad */
95 for (chan = 2; chan < 4; chan++) {
96 const float dadx = coef->dadx[chan];
97 const float dady = coef->dady[chan];
98 const float a0 = coef->a0[chan] + dadx * x + dady * y;
99 quadpos->xyzw[chan].f[0] = a0;
100 quadpos->xyzw[chan].f[1] = a0 + dadx;
101 quadpos->xyzw[chan].f[2] = a0 + dady;
102 quadpos->xyzw[chan].f[3] = a0 + dadx + dady;
103 }
104 }
105
106
107 /* TODO: hide the machine struct in here somewhere, remove from this
108 * interface:
109 */
110 static unsigned
exec_run(const struct sp_fragment_shader_variant * var,struct tgsi_exec_machine * machine,struct quad_header * quad,bool early_depth_test)111 exec_run( const struct sp_fragment_shader_variant *var,
112 struct tgsi_exec_machine *machine,
113 struct quad_header *quad,
114 bool early_depth_test )
115 {
116 /* Compute X, Y, Z, W vals for this quad */
117 setup_pos_vector(quad->posCoef,
118 (float)quad->input.x0, (float)quad->input.y0,
119 &machine->QuadPos);
120
121 /* convert 0 to 1.0 and 1 to -1.0 */
122 machine->Face = (float) (quad->input.facing * -2 + 1);
123
124 machine->NonHelperMask = quad->inout.mask;
125 quad->inout.mask &= tgsi_exec_machine_run( machine, 0 );
126 if (quad->inout.mask == 0)
127 return FALSE;
128
129 /* store outputs */
130 {
131 const ubyte *sem_name = var->info.output_semantic_name;
132 const ubyte *sem_index = var->info.output_semantic_index;
133 const uint n = var->info.num_outputs;
134 uint i;
135 for (i = 0; i < n; i++) {
136 switch (sem_name[i]) {
137 case TGSI_SEMANTIC_COLOR:
138 {
139 uint cbuf = sem_index[i];
140
141 assert(sizeof(quad->output.color[cbuf]) ==
142 sizeof(machine->Outputs[i]));
143
144 /* copy float[4][4] result */
145 memcpy(quad->output.color[cbuf],
146 &machine->Outputs[i],
147 sizeof(quad->output.color[0]) );
148 }
149 break;
150 case TGSI_SEMANTIC_POSITION:
151 {
152 uint j;
153
154 if (!early_depth_test) {
155 for (j = 0; j < 4; j++)
156 quad->output.depth[j] = machine->Outputs[i].xyzw[2].f[j];
157 }
158 }
159 break;
160 case TGSI_SEMANTIC_STENCIL:
161 {
162 uint j;
163 if (!early_depth_test) {
164 for (j = 0; j < 4; j++)
165 quad->output.stencil[j] = (unsigned)machine->Outputs[i].xyzw[1].u[j];
166 }
167 }
168 break;
169 }
170 }
171 }
172
173 return TRUE;
174 }
175
176
177 static void
exec_delete(struct sp_fragment_shader_variant * var,struct tgsi_exec_machine * machine)178 exec_delete(struct sp_fragment_shader_variant *var,
179 struct tgsi_exec_machine *machine)
180 {
181 if (machine->Tokens == var->tokens) {
182 tgsi_exec_machine_bind_shader(machine, NULL, NULL, NULL, NULL);
183 }
184
185 FREE( (void *) var->tokens );
186 FREE(var);
187 }
188
189
190 struct sp_fragment_shader_variant *
softpipe_create_fs_variant_exec(struct softpipe_context * softpipe)191 softpipe_create_fs_variant_exec(struct softpipe_context *softpipe)
192 {
193 struct sp_exec_fragment_shader *shader;
194
195 shader = CALLOC_STRUCT(sp_exec_fragment_shader);
196 if (!shader)
197 return NULL;
198
199 shader->base.prepare = exec_prepare;
200 shader->base.run = exec_run;
201 shader->base.delete = exec_delete;
202
203 return &shader->base;
204 }
205