remove dead code
[mesa.git] / src / mesa / pipe / draw / draw_vertex_shader.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 "pipe/p_util.h"
35 #include "draw_private.h"
36 #include "draw_context.h"
37 #include "draw_vertex.h"
38
39 #include "x86/rtasm/x86sse.h"
40
41 #include "pipe/tgsi/exec/tgsi_core.h"
42
43 static INLINE unsigned
44 compute_clipmask(float cx, float cy, float cz, float cw)
45 {
46 unsigned mask = 0;
47
48 if (-cx + cw < 0) mask |= CLIP_RIGHT_BIT;
49 if ( cx + cw < 0) mask |= CLIP_LEFT_BIT;
50 if (-cy + cw < 0) mask |= CLIP_TOP_BIT;
51 if ( cy + cw < 0) mask |= CLIP_BOTTOM_BIT;
52 if (-cz + cw < 0) mask |= CLIP_FAR_BIT;
53 if ( cz + cw < 0) mask |= CLIP_NEAR_BIT;
54
55 return mask;
56 }
57
58
59
60
61 #if !defined(XSTDCALL)
62 #if defined(WIN32)
63 #define XSTDCALL __stdcall
64 #else
65 #define XSTDCALL
66 #endif
67 #endif
68
69 typedef void (XSTDCALL *codegen_function) (
70 const struct tgsi_exec_vector *input,
71 struct tgsi_exec_vector *output,
72 float (*constant)[4],
73 struct tgsi_exec_vector *temporary );
74
75
76 /**
77 * Transform vertices with the current vertex program/shader
78 * Up to four vertices can be shaded at a time.
79 * \param vbuffer the input vertex data
80 * \param elts indexes of four input vertices
81 * \param count number of vertices to shade [1..4]
82 * \param vOut array of pointers to four output vertices
83 */
84 static void
85 run_vertex_program(struct draw_context *draw,
86 unsigned elts[4], unsigned count,
87 struct vertex_header *vOut[])
88 {
89 struct tgsi_exec_machine *machine = &draw->machine;
90 unsigned int j;
91
92 ALIGN16_DECL(struct tgsi_exec_vector, inputs, PIPE_ATTRIB_MAX);
93 ALIGN16_DECL(struct tgsi_exec_vector, outputs, PIPE_ATTRIB_MAX);
94 const float *scale = draw->viewport.scale;
95 const float *trans = draw->viewport.translate;
96
97 assert(count <= 4);
98 assert(draw->vertex_shader->state->output_semantic_name[0]
99 == TGSI_SEMANTIC_POSITION);
100
101 /* Consts does not require 16 byte alignment. */
102 machine->Consts = (float (*)[4]) draw->mapped_constants;
103
104 machine->Inputs = ALIGN16_ASSIGN(inputs);
105 machine->Outputs = ALIGN16_ASSIGN(outputs);
106
107 draw_vertex_fetch( draw, machine, elts, count );
108
109 /* run shader */
110 if( draw->vertex_shader->state->executable != NULL ) {
111 /* SSE */
112 codegen_function func = (codegen_function) draw->vertex_shader->state->executable;
113 func(
114 machine->Inputs,
115 machine->Outputs,
116 machine->Consts,
117 machine->Temps );
118 }
119 else {
120 /* interpreter */
121 tgsi_exec_machine_run( machine );
122 }
123
124
125 /* store machine results */
126 for (j = 0; j < count; j++) {
127 unsigned slot;
128 float x, y, z, w;
129
130 /* Handle attr[0] (position) specially: */
131 x = vOut[j]->clip[0] = machine->Outputs[0].xyzw[0].f[j];
132 y = vOut[j]->clip[1] = machine->Outputs[0].xyzw[1].f[j];
133 z = vOut[j]->clip[2] = machine->Outputs[0].xyzw[2].f[j];
134 w = vOut[j]->clip[3] = machine->Outputs[0].xyzw[3].f[j];
135
136 vOut[j]->clipmask = compute_clipmask(x, y, z, w) | draw->user_clipmask;
137 vOut[j]->edgeflag = 1;
138
139 /* divide by w */
140 w = 1.0f / w;
141 x *= w;
142 y *= w;
143 z *= w;
144
145 /* Viewport mapping */
146 vOut[j]->data[0][0] = x * scale[0] + trans[0];
147 vOut[j]->data[0][1] = y * scale[1] + trans[1];
148 vOut[j]->data[0][2] = z * scale[2] + trans[2];
149 vOut[j]->data[0][3] = w;
150
151 /* Remaining attributes are packed into sequential post-transform
152 * vertex attrib slots.
153 * Skip 0 since we just did it above.
154 * Subtract two because of the VERTEX_HEADER, CLIP_POS attribs.
155 */
156 for (slot = 1; slot < draw->vertex_info.num_attribs - 2; slot++) {
157 vOut[j]->data[slot][0] = machine->Outputs[slot].xyzw[0].f[j];
158 vOut[j]->data[slot][1] = machine->Outputs[slot].xyzw[1].f[j];
159 vOut[j]->data[slot][2] = machine->Outputs[slot].xyzw[2].f[j];
160 vOut[j]->data[slot][3] = machine->Outputs[slot].xyzw[3].f[j];
161 /*
162 printf("output %d: %f %f %f %f\n", slot,
163 vOut[j]->data[slot][0],
164 vOut[j]->data[slot][1],
165 vOut[j]->data[slot][2],
166 vOut[j]->data[slot][3]);
167 */
168 }
169 } /* loop over vertices */
170 }
171
172
173 /**
174 * Called by the draw module when the vertx cache needs to be flushed.
175 * This involves running the vertex shader.
176 */
177 void draw_vertex_shader_queue_flush( struct draw_context *draw )
178 {
179 unsigned i, j;
180
181 // fprintf(stderr, " q(%d) ", draw->vs.queue_nr );
182
183 /* run vertex shader on vertex cache entries, four per invokation */
184 for (i = 0; i < draw->vs.queue_nr; i += 4) {
185 struct vertex_header *dests[4];
186 unsigned elts[4];
187 int n;
188
189 for (j = 0; j < 4; j++) {
190 elts[j] = draw->vs.queue[i + j].elt;
191 dests[j] = draw->vs.queue[i + j].dest;
192 }
193
194 n = MIN2(4, draw->vs.queue_nr - i);
195 assert(n > 0);
196 assert(n <= 4);
197
198 run_vertex_program(draw, elts, n, dests);
199 }
200
201 draw->vs.queue_nr = 0;
202 }
203
204
205 void *
206 draw_create_vertex_shader(struct draw_context *draw,
207 const struct pipe_shader_state *shader)
208 {
209 struct draw_vertex_shader *vs = calloc(1, sizeof(struct draw_vertex_shader));
210
211 vs->state = shader;
212 #if defined(__i386__) || defined(__386__)
213 x86_init_func(&vs->sse2_program);
214
215 if (draw->use_sse) {
216 tgsi_emit_sse2(shader->tokens, &vs->sse2_program);
217 ((struct pipe_shader_state*)(vs->state))->executable =
218 x86_get_func(&vs->sse2_program);
219 }
220 #endif
221
222 return vs;
223 }
224
225 void draw_bind_vertex_shader(struct draw_context *draw,
226 void *vcso)
227 {
228 draw_flush(draw);
229 draw->vertex_shader = (struct draw_vertex_shader*)(vcso);
230
231 /* specify the fragment program to interpret/execute */
232 tgsi_exec_machine_init(&draw->machine,
233 draw->vertex_shader->state->tokens,
234 PIPE_MAX_SAMPLERS,
235 NULL /*samplers*/ );
236 }
237
238 void draw_delete_vertex_shader(struct draw_context *draw,
239 void *vcso)
240 {
241 struct draw_vertex_shader *vs = (struct draw_vertex_shader*)(vcso);
242 #if defined(__i386__) || defined(__386__)
243 x86_release_func(&vs->sse2_program);
244 #endif
245 free(vcso);
246 }
247
248
249