r300g: move vborender context function to seperate file
[mesa.git] / src / gallium / drivers / r300 / r300_render.c
1 /*
2 * Copyright 2009 Corbin Simpson <MostAwesomeDude@gmail.com>
3 *
4 * Permission is hereby granted, free of charge, to any person obtaining a
5 * copy of this software and associated documentation files (the "Software"),
6 * to deal in the Software without restriction, including without limitation
7 * on the rights to use, copy, modify, merge, publish, distribute, sub
8 * license, and/or sell copies of the Software, and to permit persons to whom
9 * the Software is furnished to do so, subject to the following conditions:
10 *
11 * The above copyright notice and this permission notice (including the next
12 * paragraph) shall be included in all copies or substantial portions of the
13 * Software.
14 *
15 * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
16 * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
17 * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL
18 * THE AUTHOR(S) AND/OR THEIR SUPPLIERS BE LIABLE FOR ANY CLAIM,
19 * DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR
20 * OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE
21 * USE OR OTHER DEALINGS IN THE SOFTWARE. */
22
23 /* r300_render: Vertex and index buffer primitive emission. Contains both
24 * HW TCL fastpath rendering, and SW TCL Draw-assisted rendering. */
25
26 #include "draw/draw_context.h"
27 #include "draw/draw_vbuf.h"
28
29 #include "pipe/p_inlines.h"
30
31 #include "util/u_memory.h"
32 #include "util/u_prim.h"
33
34 #include "r300_cs.h"
35 #include "r300_context.h"
36 #include "r300_emit.h"
37 #include "r300_reg.h"
38 #include "r300_render.h"
39 #include "r300_state_derived.h"
40
41 /* r300_render: Vertex and index buffer primitive emission. */
42 #define R300_MAX_VBO_SIZE (1024 * 1024)
43
44 uint32_t r300_translate_primitive(unsigned prim)
45 {
46 switch (prim) {
47 case PIPE_PRIM_POINTS:
48 return R300_VAP_VF_CNTL__PRIM_POINTS;
49 case PIPE_PRIM_LINES:
50 return R300_VAP_VF_CNTL__PRIM_LINES;
51 case PIPE_PRIM_LINE_LOOP:
52 return R300_VAP_VF_CNTL__PRIM_LINE_LOOP;
53 case PIPE_PRIM_LINE_STRIP:
54 return R300_VAP_VF_CNTL__PRIM_LINE_STRIP;
55 case PIPE_PRIM_TRIANGLES:
56 return R300_VAP_VF_CNTL__PRIM_TRIANGLES;
57 case PIPE_PRIM_TRIANGLE_STRIP:
58 return R300_VAP_VF_CNTL__PRIM_TRIANGLE_STRIP;
59 case PIPE_PRIM_TRIANGLE_FAN:
60 return R300_VAP_VF_CNTL__PRIM_TRIANGLE_FAN;
61 case PIPE_PRIM_QUADS:
62 return R300_VAP_VF_CNTL__PRIM_QUADS;
63 case PIPE_PRIM_QUAD_STRIP:
64 return R300_VAP_VF_CNTL__PRIM_QUAD_STRIP;
65 case PIPE_PRIM_POLYGON:
66 return R300_VAP_VF_CNTL__PRIM_POLYGON;
67 default:
68 return 0;
69 }
70 }
71
72 /* This is the fast-path drawing & emission for HW TCL. */
73 boolean r300_draw_range_elements(struct pipe_context* pipe,
74 struct pipe_buffer* indexBuffer,
75 unsigned indexSize,
76 unsigned minIndex,
77 unsigned maxIndex,
78 unsigned mode,
79 unsigned start,
80 unsigned count)
81 {
82 struct r300_context* r300 = r300_context(pipe);
83 uint32_t prim = r300_translate_primitive(mode);
84 struct pipe_vertex_buffer* aos = r300->vertex_buffers;
85 unsigned aos_count = r300->vertex_buffer_count;
86 short* indices;
87 unsigned packet_size;
88 unsigned i;
89 bool invalid = FALSE;
90
91 CS_LOCALS(r300);
92
93 if (!u_trim_pipe_prim(mode, &count)) {
94 return FALSE;
95 }
96
97 validate:
98 for (i = 0; i < aos_count; i++) {
99 if (!r300->winsys->add_buffer(r300->winsys, aos[i].buffer,
100 RADEON_GEM_DOMAIN_GTT, 0)) {
101 pipe->flush(pipe, 0, NULL);
102 goto validate;
103 }
104 }
105 if (!r300->winsys->validate(r300->winsys)) {
106 pipe->flush(pipe, 0, NULL);
107 if (invalid) {
108 /* Well, hell. */
109 debug_printf("r300: Stuck in validation loop, gonna quit now.");
110 exit(1);
111 }
112 invalid = TRUE;
113 goto validate;
114 }
115
116 r300_emit_dirty_state(r300);
117
118 packet_size = (aos_count >> 1) * 3 + (aos_count & 1) * 2;
119
120 BEGIN_CS(3 + packet_size + (aos_count * 2));
121 OUT_CS_PKT3(R300_PACKET3_3D_LOAD_VBPNTR, packet_size);
122 OUT_CS(aos_count);
123 for (i = 0; i < aos_count - 1; i += 2) {
124 OUT_CS(aos[i].stride |
125 (aos[i].stride << 8) |
126 (aos[i + 1].stride << 16) |
127 (aos[i + 1].stride << 24));
128 OUT_CS(aos[i].buffer_offset + start * 4 * aos[i].stride);
129 OUT_CS(aos[i + 1].buffer_offset + start * 4 * aos[i + 1].stride);
130 }
131 if (aos_count & 1) {
132 OUT_CS(aos[i].stride | (aos[i].stride << 8));
133 OUT_CS(aos[i].buffer_offset + start * 4 * aos[i].stride);
134 }
135 for (i = 0; i < aos_count; i++) {
136 OUT_CS_RELOC(aos[i].buffer, 0, RADEON_GEM_DOMAIN_GTT, 0, 0);
137 }
138 END_CS;
139
140 if (indexBuffer) {
141 indices = (short*)pipe_buffer_map(pipe->screen, indexBuffer,
142 PIPE_BUFFER_USAGE_CPU_READ);
143
144 /* Set the starting point. */
145 indices += start;
146
147 BEGIN_CS(2 + (count+1)/2);
148 OUT_CS_PKT3(R300_PACKET3_3D_DRAW_INDX_2, (count + 1)/2);
149 OUT_CS(R300_VAP_VF_CNTL__PRIM_WALK_INDICES | (count << 16) | prim);
150 for (i = 0; i < count - 1; i += 2) {
151 OUT_CS(indices[i + 1] << 16 | indices[i]);
152 }
153 if (count % 2) {
154 OUT_CS(indices[count - 1]);
155 }
156 END_CS;
157 } else {
158 BEGIN_CS(2);
159 OUT_CS_PKT3(R300_PACKET3_3D_DRAW_VBUF_2, 0);
160 OUT_CS(R300_VAP_VF_CNTL__PRIM_WALK_VERTEX_LIST | (count << 16) |
161 prim);
162 END_CS;
163 }
164
165 return TRUE;
166 }
167
168 /* Simple helpers for context setup. Should probably be moved to util. */
169 boolean r300_draw_elements(struct pipe_context* pipe,
170 struct pipe_buffer* indexBuffer,
171 unsigned indexSize, unsigned mode,
172 unsigned start, unsigned count)
173 {
174 return pipe->draw_range_elements(pipe, indexBuffer, indexSize, 0, ~0,
175 mode, start, count);
176 }
177
178 boolean r300_draw_arrays(struct pipe_context* pipe, unsigned mode,
179 unsigned start, unsigned count)
180 {
181 return pipe->draw_elements(pipe, NULL, 0, mode, start, count);
182 }
183
184 /****************************************************************************
185 * The rest of this file is for SW TCL rendering only. Please be polite and *
186 * keep these functions separated so that they are easier to locate. ~C. *
187 ***************************************************************************/
188
189 /* Draw-based drawing for SW TCL chipsets. */
190 boolean r300_swtcl_draw_range_elements(struct pipe_context* pipe,
191 struct pipe_buffer* indexBuffer,
192 unsigned indexSize,
193 unsigned minIndex,
194 unsigned maxIndex,
195 unsigned mode,
196 unsigned start,
197 unsigned count)
198 {
199 struct r300_context* r300 = r300_context(pipe);
200 int i;
201
202 if (!u_trim_pipe_prim(mode, &count)) {
203 return FALSE;
204 }
205
206 for (i = 0; i < r300->vertex_buffer_count; i++) {
207 void* buf = pipe_buffer_map(pipe->screen,
208 r300->vertex_buffers[i].buffer,
209 PIPE_BUFFER_USAGE_CPU_READ);
210 draw_set_mapped_vertex_buffer(r300->draw, i, buf);
211 }
212
213 if (indexBuffer) {
214 void* indices = pipe_buffer_map(pipe->screen, indexBuffer,
215 PIPE_BUFFER_USAGE_CPU_READ);
216 draw_set_mapped_element_buffer_range(r300->draw, indexSize,
217 minIndex, maxIndex, indices);
218 } else {
219 draw_set_mapped_element_buffer(r300->draw, 0, NULL);
220 }
221
222 draw_set_mapped_constant_buffer(r300->draw,
223 r300->shader_constants[PIPE_SHADER_VERTEX].constants,
224 r300->shader_constants[PIPE_SHADER_VERTEX].count *
225 (sizeof(float) * 4));
226
227 draw_arrays(r300->draw, mode, start, count);
228
229 for (i = 0; i < r300->vertex_buffer_count; i++) {
230 pipe_buffer_unmap(pipe->screen, r300->vertex_buffers[i].buffer);
231 draw_set_mapped_vertex_buffer(r300->draw, i, NULL);
232 }
233
234 if (indexBuffer) {
235 pipe_buffer_unmap(pipe->screen, indexBuffer);
236 draw_set_mapped_element_buffer_range(r300->draw, 0, start,
237 start + count - 1, NULL);
238 }
239
240 return TRUE;
241 }
242
243 /* Object for rendering using Draw. */
244 struct r300_render {
245 /* Parent class */
246 struct vbuf_render base;
247
248 /* Pipe context */
249 struct r300_context* r300;
250
251 /* Vertex information */
252 size_t vertex_size;
253 unsigned prim;
254 unsigned hwprim;
255
256 /* VBO */
257 struct pipe_buffer* vbo;
258 size_t vbo_size;
259 size_t vbo_offset;
260 size_t vbo_max_used;
261 void * vbo_ptr;
262 };
263
264 static INLINE struct r300_render*
265 r300_render(struct vbuf_render* render)
266 {
267 return (struct r300_render*)render;
268 }
269
270 static const struct vertex_info*
271 r300_render_get_vertex_info(struct vbuf_render* render)
272 {
273 struct r300_render* r300render = r300_render(render);
274 struct r300_context* r300 = r300render->r300;
275
276 r300_update_derived_state(r300);
277
278 return &r300->vertex_info->vinfo;
279 }
280
281 static boolean r300_render_allocate_vertices(struct vbuf_render* render,
282 ushort vertex_size,
283 ushort count)
284 {
285 struct r300_render* r300render = r300_render(render);
286 struct r300_context* r300 = r300render->r300;
287 struct pipe_screen* screen = r300->context.screen;
288 size_t size = (size_t)vertex_size * (size_t)count;
289
290 if (size + r300render->vbo_offset > r300render->vbo_size)
291 {
292 pipe_buffer_reference(&r300->vbo, NULL);
293 r300render->vbo = pipe_buffer_create(screen,
294 64,
295 PIPE_BUFFER_USAGE_VERTEX,
296 R300_MAX_VBO_SIZE);
297 r300render->vbo_offset = 0;
298 r300render->vbo_size = R300_MAX_VBO_SIZE;
299 }
300
301 r300render->vertex_size = vertex_size;
302 r300->vbo = r300render->vbo;
303 r300->vbo_offset = r300render->vbo_offset;
304
305 return (r300render->vbo) ? TRUE : FALSE;
306 }
307
308 static void* r300_render_map_vertices(struct vbuf_render* render)
309 {
310 struct r300_render* r300render = r300_render(render);
311 struct pipe_screen* screen = r300render->r300->context.screen;
312
313 r300render->vbo_ptr = pipe_buffer_map(screen, r300render->vbo,
314 PIPE_BUFFER_USAGE_CPU_WRITE);
315
316 return (r300render->vbo_ptr + r300render->vbo_offset);
317 }
318
319 static void r300_render_unmap_vertices(struct vbuf_render* render,
320 ushort min,
321 ushort max)
322 {
323 struct r300_render* r300render = r300_render(render);
324 struct pipe_screen* screen = r300render->r300->context.screen;
325 CS_LOCALS(r300render->r300);
326 BEGIN_CS(2);
327 OUT_CS_REG(R300_VAP_VF_MAX_VTX_INDX, max);
328 END_CS;
329
330 r300render->vbo_max_used = MAX2(r300render->vbo_max_used,
331 r300render->vertex_size * (max + 1));
332 pipe_buffer_unmap(screen, r300render->vbo);
333 }
334
335 static void r300_render_release_vertices(struct vbuf_render* render)
336 {
337 struct r300_render* r300render = r300_render(render);
338
339 r300render->vbo_offset += r300render->vbo_max_used;
340 r300render->vbo_max_used = 0;
341 }
342
343 static boolean r300_render_set_primitive(struct vbuf_render* render,
344 unsigned prim)
345 {
346 struct r300_render* r300render = r300_render(render);
347
348 r300render->prim = prim;
349 r300render->hwprim = r300_translate_primitive(prim);
350
351 return TRUE;
352 }
353
354 static void r300_render_draw_arrays(struct vbuf_render* render,
355 unsigned start,
356 unsigned count)
357 {
358 struct r300_render* r300render = r300_render(render);
359 struct r300_context* r300 = r300render->r300;
360
361 CS_LOCALS(r300);
362
363 r300_emit_dirty_state(r300);
364
365 DBG(r300, DBG_DRAW, "r300: Doing vbuf render, count %d\n", count);
366
367 BEGIN_CS(2);
368 OUT_CS_PKT3(R300_PACKET3_3D_DRAW_VBUF_2, 0);
369 OUT_CS(R300_VAP_VF_CNTL__PRIM_WALK_VERTEX_LIST | (count << 16) |
370 r300render->hwprim);
371 END_CS;
372 }
373
374 static void r300_render_draw(struct vbuf_render* render,
375 const ushort* indices,
376 uint count)
377 {
378 struct r300_render* r300render = r300_render(render);
379 struct r300_context* r300 = r300render->r300;
380 int i;
381
382 CS_LOCALS(r300);
383
384 r300_emit_dirty_state(r300);
385
386 BEGIN_CS(2 + (count+1)/2);
387 OUT_CS_PKT3(R300_PACKET3_3D_DRAW_INDX_2, (count+1)/2);
388 OUT_CS(R300_VAP_VF_CNTL__PRIM_WALK_INDICES | (count << 16) |
389 r300render->hwprim);
390 for (i = 0; i < count-1; i += 2) {
391 OUT_CS(indices[i+1] << 16 | indices[i]);
392 }
393 if (count % 2) {
394 OUT_CS(indices[count-1]);
395 }
396 END_CS;
397 }
398
399 static void r300_render_destroy(struct vbuf_render* render)
400 {
401 FREE(render);
402 }
403
404 static struct vbuf_render* r300_render_create(struct r300_context* r300)
405 {
406 struct r300_render* r300render = CALLOC_STRUCT(r300_render);
407
408 r300render->r300 = r300;
409
410 /* XXX find real numbers plz */
411 r300render->base.max_vertex_buffer_bytes = 128 * 1024;
412 r300render->base.max_indices = 16 * 1024;
413
414 r300render->base.get_vertex_info = r300_render_get_vertex_info;
415 r300render->base.allocate_vertices = r300_render_allocate_vertices;
416 r300render->base.map_vertices = r300_render_map_vertices;
417 r300render->base.unmap_vertices = r300_render_unmap_vertices;
418 r300render->base.set_primitive = r300_render_set_primitive;
419 r300render->base.draw = r300_render_draw;
420 r300render->base.draw_arrays = r300_render_draw_arrays;
421 r300render->base.release_vertices = r300_render_release_vertices;
422 r300render->base.destroy = r300_render_destroy;
423
424 r300render->vbo = NULL;
425 r300render->vbo_size = 0;
426 r300render->vbo_offset = 0;
427
428 return &r300render->base;
429 }
430
431 struct draw_stage* r300_draw_stage(struct r300_context* r300)
432 {
433 struct vbuf_render* render;
434 struct draw_stage* stage;
435
436 render = r300_render_create(r300);
437
438 if (!render) {
439 return NULL;
440 }
441
442 stage = draw_vbuf_stage(r300->draw, render);
443
444 if (!stage) {
445 render->destroy(render);
446 return NULL;
447 }
448
449 draw_set_render(r300->draw, render);
450
451 return stage;
452 }