draw: s/Elements/ARRAY_SIZE/
[mesa.git] / src / gallium / auxiliary / draw / draw_vs_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 * Authors:
30 * Keith Whitwell <keithw@vmware.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 debug_assert(!draw->llvm);
67 /* Specify the vertex program to interpret/execute.
68 * Avoid rebinding when possible.
69 */
70 if (evs->machine->Tokens != shader->state.tokens) {
71 tgsi_exec_machine_bind_shader(evs->machine,
72 shader->state.tokens,
73 draw->vs.tgsi.sampler,
74 draw->vs.tgsi.image,
75 draw->vs.tgsi.buffer);
76 }
77 }
78
79
80
81
82 /* Simplified vertex shader interface for the pt paths. Given the
83 * complexity of code-generating all the above operations together,
84 * it's time to try doing all the other stuff separately.
85 */
86 static void
87 vs_exec_run_linear( struct draw_vertex_shader *shader,
88 const float (*input)[4],
89 float (*output)[4],
90 const void *constants[PIPE_MAX_CONSTANT_BUFFERS],
91 const unsigned const_size[PIPE_MAX_CONSTANT_BUFFERS],
92 unsigned count,
93 unsigned input_stride,
94 unsigned output_stride )
95 {
96 struct exec_vertex_shader *evs = exec_vertex_shader(shader);
97 struct tgsi_exec_machine *machine = evs->machine;
98 unsigned int i, j;
99 unsigned slot;
100 boolean clamp_vertex_color = shader->draw->rasterizer->clamp_vertex_color;
101
102 debug_assert(!shader->draw->llvm);
103 tgsi_exec_set_constant_buffers(machine, PIPE_MAX_CONSTANT_BUFFERS,
104 constants, const_size);
105
106 if (shader->info.uses_instanceid) {
107 unsigned i = machine->SysSemanticToIndex[TGSI_SEMANTIC_INSTANCEID];
108 assert(i < ARRAY_SIZE(machine->SystemValue));
109 for (j = 0; j < TGSI_QUAD_SIZE; j++)
110 machine->SystemValue[i].xyzw[0].i[j] = shader->draw->instance_id;
111 }
112
113 for (i = 0; i < count; i += MAX_TGSI_VERTICES) {
114 unsigned int max_vertices = MIN2(MAX_TGSI_VERTICES, count - i);
115
116 /* Swizzle inputs.
117 */
118 for (j = 0; j < max_vertices; j++) {
119 #if 0
120 debug_printf("%d) Input vert:\n", i + j);
121 for (slot = 0; slot < shader->info.num_inputs; slot++) {
122 debug_printf("\t%d: %f %f %f %f\n", slot,
123 input[slot][0],
124 input[slot][1],
125 input[slot][2],
126 input[slot][3]);
127 }
128 #endif
129
130 if (shader->info.uses_vertexid) {
131 unsigned vid = machine->SysSemanticToIndex[TGSI_SEMANTIC_VERTEXID];
132 assert(vid < ARRAY_SIZE(machine->SystemValue));
133 machine->SystemValue[vid].xyzw[0].i[j] = i + j;
134 /* XXX this should include base vertex. Where to get it??? */
135 }
136 if (shader->info.uses_basevertex) {
137 unsigned vid = machine->SysSemanticToIndex[TGSI_SEMANTIC_BASEVERTEX];
138 assert(vid < ARRAY_SIZE(machine->SystemValue));
139 machine->SystemValue[vid].xyzw[0].i[j] = 0;
140 /* XXX Where to get it??? */
141 }
142 if (shader->info.uses_vertexid_nobase) {
143 unsigned vid = machine->SysSemanticToIndex[TGSI_SEMANTIC_VERTEXID_NOBASE];
144 assert(vid < ARRAY_SIZE(machine->SystemValue));
145 machine->SystemValue[vid].xyzw[0].i[j] = i + j;
146 }
147
148 for (slot = 0; slot < shader->info.num_inputs; slot++) {
149 #if 0
150 assert(!util_is_inf_or_nan(input[slot][0]));
151 assert(!util_is_inf_or_nan(input[slot][1]));
152 assert(!util_is_inf_or_nan(input[slot][2]));
153 assert(!util_is_inf_or_nan(input[slot][3]));
154 #endif
155 machine->Inputs[slot].xyzw[0].f[j] = input[slot][0];
156 machine->Inputs[slot].xyzw[1].f[j] = input[slot][1];
157 machine->Inputs[slot].xyzw[2].f[j] = input[slot][2];
158 machine->Inputs[slot].xyzw[3].f[j] = input[slot][3];
159 }
160
161 input = (const float (*)[4])((const char *)input + input_stride);
162 }
163
164 machine->NonHelperMask = (1 << max_vertices) - 1;
165 /* run interpreter */
166 tgsi_exec_machine_run( machine, 0 );
167
168 /* Unswizzle all output results.
169 */
170 for (j = 0; j < max_vertices; j++) {
171 for (slot = 0; slot < shader->info.num_outputs; slot++) {
172 unsigned name = shader->info.output_semantic_name[slot];
173 if(clamp_vertex_color &&
174 (name == TGSI_SEMANTIC_COLOR || name == TGSI_SEMANTIC_BCOLOR))
175 {
176 output[slot][0] = CLAMP(machine->Outputs[slot].xyzw[0].f[j], 0.0f, 1.0f);
177 output[slot][1] = CLAMP(machine->Outputs[slot].xyzw[1].f[j], 0.0f, 1.0f);
178 output[slot][2] = CLAMP(machine->Outputs[slot].xyzw[2].f[j], 0.0f, 1.0f);
179 output[slot][3] = CLAMP(machine->Outputs[slot].xyzw[3].f[j], 0.0f, 1.0f);
180 }
181 else
182 {
183 output[slot][0] = machine->Outputs[slot].xyzw[0].f[j];
184 output[slot][1] = machine->Outputs[slot].xyzw[1].f[j];
185 output[slot][2] = machine->Outputs[slot].xyzw[2].f[j];
186 output[slot][3] = machine->Outputs[slot].xyzw[3].f[j];
187 }
188 }
189
190 #if 0
191 debug_printf("%d) Post xform vert:\n", i + j);
192 for (slot = 0; slot < shader->info.num_outputs; slot++) {
193 debug_printf("\t%d: %f %f %f %f\n", slot,
194 output[slot][0],
195 output[slot][1],
196 output[slot][2],
197 output[slot][3]);
198 assert(!util_is_inf_or_nan(output[slot][0]));
199 }
200 #endif
201
202 output = (float (*)[4])((char *)output + output_stride);
203 }
204
205 }
206 }
207
208
209
210
211 static void
212 vs_exec_delete( struct draw_vertex_shader *dvs )
213 {
214 FREE((void*) dvs->state.tokens);
215 FREE( dvs );
216 }
217
218
219 struct draw_vertex_shader *
220 draw_create_vs_exec(struct draw_context *draw,
221 const struct pipe_shader_state *state)
222 {
223 struct exec_vertex_shader *vs = CALLOC_STRUCT( exec_vertex_shader );
224
225 if (!vs)
226 return NULL;
227
228 /* we make a private copy of the tokens */
229 vs->base.state.tokens = tgsi_dup_tokens(state->tokens);
230 if (!vs->base.state.tokens) {
231 FREE(vs);
232 return NULL;
233 }
234
235 tgsi_scan_shader(state->tokens, &vs->base.info);
236
237 vs->base.state.stream_output = state->stream_output;
238 vs->base.draw = draw;
239 vs->base.prepare = vs_exec_prepare;
240 vs->base.run_linear = vs_exec_run_linear;
241 vs->base.delete = vs_exec_delete;
242 vs->base.create_variant = draw_vs_create_variant_generic;
243 vs->machine = draw->vs.tgsi.machine;
244
245 return &vs->base;
246 }