Merge branch 'mesa_7_5_branch'
[mesa.git] / src / gallium / auxiliary / draw / draw_vs_exec.c
1 /**************************************************************************
2 *
3 * Copyright 2007 Tungsten Graphics, Inc., Cedar Park, Texas.
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 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.
25 *
26 **************************************************************************/
27
28 /*
29 * Authors:
30 * Keith Whitwell <keith@tungstengraphics.com>
31 * Brian Paul
32 */
33
34 #include "util/u_math.h"
35 #include "util/u_memory.h"
36 #include "pipe/p_shader_tokens.h"
37
38 #include "draw_private.h"
39 #include "draw_context.h"
40 #include "draw_vs.h"
41
42 #include "tgsi/tgsi_parse.h"
43 #include "tgsi/tgsi_scan.h"
44 #include "tgsi/tgsi_exec.h"
45
46
47 struct exec_vertex_shader {
48 struct draw_vertex_shader base;
49 struct tgsi_exec_machine *machine;
50 };
51
52 static struct exec_vertex_shader *exec_vertex_shader( struct draw_vertex_shader *vs )
53 {
54 return (struct exec_vertex_shader *)vs;
55 }
56
57
58 /* Not required for run_linear.
59 */
60 static void
61 vs_exec_prepare( struct draw_vertex_shader *shader,
62 struct draw_context *draw )
63 {
64 struct exec_vertex_shader *evs = exec_vertex_shader(shader);
65
66 /* Specify the vertex program to interpret/execute.
67 * Avoid rebinding when possible.
68 */
69 if (evs->machine->Tokens != shader->state.tokens) {
70 tgsi_exec_machine_bind_shader(evs->machine,
71 shader->state.tokens,
72 draw->vs.num_samplers,
73 draw->vs.samplers);
74 }
75 }
76
77
78
79
80 /* Simplified vertex shader interface for the pt paths. Given the
81 * complexity of code-generating all the above operations together,
82 * it's time to try doing all the other stuff separately.
83 */
84 static void
85 vs_exec_run_linear( struct draw_vertex_shader *shader,
86 const float (*input)[4],
87 float (*output)[4],
88 const float (*constants)[4],
89 unsigned count,
90 unsigned input_stride,
91 unsigned output_stride )
92 {
93 struct exec_vertex_shader *evs = exec_vertex_shader(shader);
94 struct tgsi_exec_machine *machine = evs->machine;
95 unsigned int i, j;
96 unsigned slot;
97
98 machine->Consts = constants;
99
100 for (i = 0; i < count; i += MAX_TGSI_VERTICES) {
101 unsigned int max_vertices = MIN2(MAX_TGSI_VERTICES, count - i);
102
103 /* Swizzle inputs.
104 */
105 for (j = 0; j < max_vertices; j++) {
106 #if 0
107 debug_printf("%d) Input vert:\n", i + j);
108 for (slot = 0; slot < shader->info.num_inputs; slot++) {
109 debug_printf("\t%d: %f %f %f %f\n", slot,
110 input[slot][0],
111 input[slot][1],
112 input[slot][2],
113 input[slot][3]);
114 }
115 #endif
116
117 for (slot = 0; slot < shader->info.num_inputs; slot++) {
118 #if 0
119 assert(!util_is_inf_or_nan(input[slot][0]));
120 assert(!util_is_inf_or_nan(input[slot][1]));
121 assert(!util_is_inf_or_nan(input[slot][2]));
122 assert(!util_is_inf_or_nan(input[slot][3]));
123 #endif
124 machine->Inputs[slot].xyzw[0].f[j] = input[slot][0];
125 machine->Inputs[slot].xyzw[1].f[j] = input[slot][1];
126 machine->Inputs[slot].xyzw[2].f[j] = input[slot][2];
127 machine->Inputs[slot].xyzw[3].f[j] = input[slot][3];
128 }
129
130 input = (const float (*)[4])((const char *)input + input_stride);
131 }
132
133 tgsi_set_exec_mask(machine,
134 1,
135 max_vertices > 1,
136 max_vertices > 2,
137 max_vertices > 3);
138
139 /* run interpreter */
140 tgsi_exec_machine_run( machine );
141
142 /* Unswizzle all output results.
143 */
144 for (j = 0; j < max_vertices; j++) {
145 for (slot = 0; slot < shader->info.num_outputs; slot++) {
146 output[slot][0] = machine->Outputs[slot].xyzw[0].f[j];
147 output[slot][1] = machine->Outputs[slot].xyzw[1].f[j];
148 output[slot][2] = machine->Outputs[slot].xyzw[2].f[j];
149 output[slot][3] = machine->Outputs[slot].xyzw[3].f[j];
150
151 }
152
153 #if 0
154 debug_printf("%d) Post xform vert:\n", i + j);
155 for (slot = 0; slot < shader->info.num_outputs; slot++) {
156 debug_printf("\t%d: %f %f %f %f\n", slot,
157 output[slot][0],
158 output[slot][1],
159 output[slot][2],
160 output[slot][3]);
161 assert(!util_is_inf_or_nan(output[slot][0]));
162 }
163 #endif
164
165 output = (float (*)[4])((char *)output + output_stride);
166 }
167
168 }
169 }
170
171
172
173
174 static void
175 vs_exec_delete( struct draw_vertex_shader *dvs )
176 {
177 FREE((void*) dvs->state.tokens);
178 FREE( dvs );
179 }
180
181
182 struct draw_vertex_shader *
183 draw_create_vs_exec(struct draw_context *draw,
184 const struct pipe_shader_state *state)
185 {
186 struct exec_vertex_shader *vs = CALLOC_STRUCT( exec_vertex_shader );
187
188 if (vs == NULL)
189 return NULL;
190
191 /* we make a private copy of the tokens */
192 vs->base.state.tokens = tgsi_dup_tokens(state->tokens);
193 if (!vs->base.state.tokens) {
194 FREE(vs);
195 return NULL;
196 }
197
198 tgsi_scan_shader(state->tokens, &vs->base.info);
199
200 vs->base.draw = draw;
201 vs->base.prepare = vs_exec_prepare;
202 vs->base.run_linear = vs_exec_run_linear;
203 vs->base.delete = vs_exec_delete;
204 vs->base.create_varient = draw_vs_varient_generic;
205 vs->machine = draw->vs.machine;
206
207 return &vs->base;
208 }