summaryrefslogtreecommitdiff
path: root/src/mesa/tnl/t_split_copy.c
blob: 87c0020e311f43144e9ddd672900ce95ed2eeaf0 (plain)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637

/*
 * Mesa 3-D graphics library
 *
 * Copyright (C) 1999-2006  Brian Paul   All Rights Reserved.
 *
 * Permission is hereby granted, free of charge, to any person obtaining a
 * copy of this software and associated documentation files (the "Software"),
 * to deal in the Software without restriction, including without limitation
 * the rights to use, copy, modify, merge, publish, distribute, sublicense,
 * and/or sell copies of the Software, and to permit persons to whom the
 * Software is furnished to do so, subject to the following conditions:
 *
 * The above copyright notice and this permission notice shall be included
 * in all copies or substantial portions of the Software.
 *
 * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
 * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
 * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
 * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR
 * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
 * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
 * OTHER DEALINGS IN THE SOFTWARE.
 *
 * Authors:
 *    Keith Whitwell <keithw@vmware.com>
 */

/* Split indexed primitives with per-vertex copying.
 */

#include <stdio.h>

#include "main/glheader.h"
#include "main/bufferobj.h"
#include "util/imports.h"
#include "main/glformats.h"
#include "main/macros.h"
#include "main/mtypes.h"
#include "main/varray.h"
#include "vbo/vbo.h"

#include "t_split.h"
#include "tnl.h"


#define ELT_TABLE_SIZE 16

/**
 * Used for vertex-level splitting of indexed buffers.  Note that
 * non-indexed primitives may be converted to indexed in some cases
 * (eg loops, fans) in order to use this splitting path.
 */
struct copy_context {
   struct gl_context *ctx;
   const struct tnl_vertex_array *array;
   const struct _mesa_prim *prim;
   GLuint nr_prims;
   const struct _mesa_index_buffer *ib;
   tnl_draw_func draw;

   const struct split_limits *limits;

   struct {
      GLuint attr;
      GLuint size;
      const struct tnl_vertex_array *array;
      const GLubyte *src_ptr;

      struct gl_vertex_buffer_binding dstbinding;
      struct gl_array_attributes dstattribs;

   } varying[VERT_ATTRIB_MAX];
   GLuint nr_varying;

   struct tnl_vertex_array dstarray[VERT_ATTRIB_MAX];
   struct _mesa_index_buffer dstib;

   GLuint *translated_elt_buf;
   const GLuint *srcelt;

   /** A baby hash table to avoid re-emitting (some) duplicate
    * vertices when splitting indexed primitives.
    */
   struct {
      GLuint in;
      GLuint out;
   } vert_cache[ELT_TABLE_SIZE];

   GLuint vertex_size;
   GLubyte *dstbuf;
   GLubyte *dstptr;     /**< dstptr == dstbuf + dstelt_max * vertsize */
   GLuint dstbuf_size;  /**< in vertices */
   GLuint dstbuf_nr;    /**< count of emitted vertices, also the largest value
                         * in dstelt.  Our MaxIndex.
                         */

   GLuint *dstelt;
   GLuint dstelt_nr;
   GLuint dstelt_size;

#define MAX_PRIM 32
   struct _mesa_prim dstprim[MAX_PRIM];
   GLuint dstprim_nr;
};


/**
 * Shallow copy one vertex array to another.
 */
static inline void
copy_vertex_array(struct tnl_vertex_array *dst,
                  const struct tnl_vertex_array *src)
{
   dst->VertexAttrib = src->VertexAttrib;
   dst->BufferBinding = src->BufferBinding;
}


/**
 * Starts returning true slightly before the buffer fills, to ensure
 * that there is sufficient room for any remaining vertices to finish
 * off the prim:
 */
static GLboolean
check_flush(struct copy_context *copy)
{
   GLenum mode = copy->dstprim[copy->dstprim_nr].mode;

   if (GL_TRIANGLE_STRIP == mode &&
       copy->dstelt_nr & 1) { /* see bug9962 */
       return GL_FALSE;
   }

   if (copy->dstbuf_nr + 4 > copy->dstbuf_size)
      return GL_TRUE;

   if (copy->dstelt_nr + 4 > copy->dstelt_size)
      return GL_TRUE;

   return GL_FALSE;
}


/**
 * Dump the parameters/info for a vbo->draw() call.
 */
static void
dump_draw_info(const struct tnl_vertex_array *arrays,
               const struct _mesa_prim *prims,
               GLuint nr_prims,
               const struct _mesa_index_buffer *ib)
{
   GLuint i, j;

   printf("VBO Draw:\n");
   for (i = 0; i < nr_prims; i++) {
      printf("Prim %u of %u\n", i, nr_prims);
      printf("  Prim mode 0x%x\n", prims[i].mode);
      printf("  IB: %p\n", (void*) ib);
      for (j = 0; j < VERT_ATTRIB_MAX; j++) {
         const struct tnl_vertex_array *array = &arrays[j];
         const struct gl_vertex_buffer_binding *binding
            = array->BufferBinding;
         const struct gl_array_attributes *attrib = array->VertexAttrib;
         const GLubyte *ptr = _mesa_vertex_attrib_address(attrib, binding);
         printf("    array %d at %p:\n", j, (void*) &arrays[j]);
         printf("      ptr %p, size %d, type 0x%x, stride %d\n",
                ptr, attrib->Format.Size, attrib->Format.Type, binding->Stride);
         if (0) {
            GLint k = prims[i].start + prims[i].count - 1;
            GLfloat *last = (GLfloat *) (ptr + binding->Stride * k);
            printf("        last: %f %f %f\n",
                   last[0], last[1], last[2]);
         }
      }
   }
}


static void
flush(struct copy_context *copy)
{
   struct gl_context *ctx = copy->ctx;
   GLuint i;

   /* Set some counters:
    */
   copy->dstib.count = copy->dstelt_nr;

#if 0
   dump_draw_info(copy->dstarray,
                  copy->dstprim,
                  copy->dstprim_nr,
                  &copy->dstib);
#else
   (void) dump_draw_info;
#endif

   copy->draw(ctx,
              copy->dstarray,
              copy->dstprim,
              copy->dstprim_nr,
              &copy->dstib,
              GL_TRUE,
              0,
              copy->dstbuf_nr - 1,
              1,
              0);

   /* Reset all pointers:
    */
   copy->dstprim_nr = 0;
   copy->dstelt_nr = 0;
   copy->dstbuf_nr = 0;
   copy->dstptr = copy->dstbuf;

   /* Clear the vertex cache:
    */
   for (i = 0; i < ELT_TABLE_SIZE; i++)
      copy->vert_cache[i].in = ~0;
}


/**
 * Called at begin of each primitive during replay.
 */
static void
begin(struct copy_context *copy, GLenum mode, GLboolean begin_flag)
{
   struct _mesa_prim *prim = &copy->dstprim[copy->dstprim_nr];

   prim->mode = mode;
   prim->begin = begin_flag;
}


/**
 * Use a hashtable to attempt to identify recently-emitted vertices
 * and avoid re-emitting them.
 */
static GLuint
elt(struct copy_context *copy, GLuint elt_idx)
{
   GLuint elt = copy->srcelt[elt_idx] + copy->prim->basevertex;
   GLuint slot = elt & (ELT_TABLE_SIZE-1);

   /* Look up the incoming element in the vertex cache.  Re-emit if
    * necessary.
    */
   if (copy->vert_cache[slot].in != elt) {
      GLubyte *csr = copy->dstptr;
      GLuint i;

      for (i = 0; i < copy->nr_varying; i++) {
         const struct tnl_vertex_array *srcarray = copy->varying[i].array;
         const struct gl_vertex_buffer_binding* srcbinding
            = srcarray->BufferBinding;
         const GLubyte *srcptr
            = copy->varying[i].src_ptr + elt * srcbinding->Stride;

         memcpy(csr, srcptr, copy->varying[i].size);
         csr += copy->varying[i].size;

#ifdef NAN_CHECK
         if (srcarray->Format.Type == GL_FLOAT) {
            GLuint k;
            GLfloat *f = (GLfloat *) srcptr;
            for (k = 0; k < srcarray->Size; k++) {
               assert(!IS_INF_OR_NAN(f[k]));
               assert(f[k] <= 1.0e20 && f[k] >= -1.0e20);
            }
         }
#endif

         if (0) {
            const GLuint *f = (const GLuint *)srcptr;
            GLuint j;
            printf("  varying %d: ", i);
            for (j = 0; j < copy->varying[i].size / 4; j++)
               printf("%x ", f[j]);
            printf("\n");
         }
      }

      copy->vert_cache[slot].in = elt;
      copy->vert_cache[slot].out = copy->dstbuf_nr++;
      copy->dstptr += copy->vertex_size;

      assert(csr == copy->dstptr);
      assert(copy->dstptr == (copy->dstbuf +
                              copy->dstbuf_nr * copy->vertex_size));
   }

   copy->dstelt[copy->dstelt_nr++] = copy->vert_cache[slot].out;
   return check_flush(copy);
}


/**
 * Called at end of each primitive during replay.
 */
static void
end(struct copy_context *copy, GLboolean end_flag)
{
   struct _mesa_prim *prim = &copy->dstprim[copy->dstprim_nr];

   prim->end = end_flag;
   prim->count = copy->dstelt_nr - prim->start;

   if (++copy->dstprim_nr == MAX_PRIM || check_flush(copy)) {
      flush(copy);
   }
}


static void
replay_elts(struct copy_context *copy)
{
   GLuint i, j, k;
   GLboolean split;

   for (i = 0; i < copy->nr_prims; i++) {
      const struct _mesa_prim *prim = &copy->prim[i];
      const GLuint start = prim->start;
      GLuint first, incr;

      switch (prim->mode) {
      case GL_LINE_LOOP:
         /* Convert to linestrip and emit the final vertex explicitly,
          * but only in the resultant strip that requires it.
          */
         j = 0;
         while (j != prim->count) {
            begin(copy, GL_LINE_STRIP, prim->begin && j == 0);

            for (split = GL_FALSE; j != prim->count && !split; j++)
               split = elt(copy, start + j);

            if (j == prim->count) {
               /* Done, emit final line.  Split doesn't matter as
                * it is always raised a bit early so we can emit
                * the last verts if necessary!
                */
               if (prim->end)
                  (void)elt(copy, start + 0);

               end(copy, prim->end);
            }
            else {
               /* Wrap
                */
               assert(split);
               end(copy, 0);
               j--;
            }
         }
         break;

      case GL_TRIANGLE_FAN:
      case GL_POLYGON:
         j = 2;
         while (j != prim->count) {
            begin(copy, prim->mode, prim->begin && j == 0);

            split = elt(copy, start+0);
            assert(!split);

            split = elt(copy, start+j-1);
            assert(!split);

            for (; j != prim->count && !split; j++)
               split = elt(copy, start+j);

            end(copy, prim->end && j == prim->count);

            if (j != prim->count) {
               /* Wrapped the primitive, need to repeat some vertices:
                */
               j -= 1;
            }
         }
         break;

      default:
         (void)_tnl_split_prim_inplace(prim->mode, &first, &incr);

         j = 0;
         while (j != prim->count) {

            begin(copy, prim->mode, prim->begin && j == 0);

            split = 0;
            for (k = 0; k < first; k++, j++)
               split |= elt(copy, start+j);

            assert(!split);

            for (; j != prim->count && !split;)
               for (k = 0; k < incr; k++, j++)
                  split |= elt(copy, start+j);

            end(copy, prim->end && j == prim->count);

            if (j != prim->count) {
               /* Wrapped the primitive, need to repeat some vertices:
                */
               assert(j > first - incr);
               j -= (first - incr);
            }
         }
         break;
      }
   }

   if (copy->dstprim_nr)
      flush(copy);
}


static void
replay_init(struct copy_context *copy)
{
   struct gl_context *ctx = copy->ctx;
   GLuint i;
   GLuint offset;
   const GLvoid *srcptr;

   /* Make a list of varying attributes and their vbo's.  Also
    * calculate vertex size.
    */
   copy->vertex_size = 0;
   for (i = 0; i < VERT_ATTRIB_MAX; i++) {
      const struct tnl_vertex_array *array = &copy->array[i];
      const struct gl_vertex_buffer_binding *binding = array->BufferBinding;

      if (binding->Stride == 0) {
         copy_vertex_array(&copy->dstarray[i], array);
      }
      else {
         const struct gl_array_attributes *attrib = array->VertexAttrib;
         struct gl_buffer_object *vbo = binding->BufferObj;
         const GLubyte *ptr = _mesa_vertex_attrib_address(attrib, binding);
         GLuint j = copy->nr_varying++;

         copy->varying[j].attr = i;
         copy->varying[j].array = &copy->array[i];
         copy->varying[j].size = attrib->Format._ElementSize;
         copy->vertex_size += attrib->Format._ElementSize;

         if (vbo) {
            if (!_mesa_bufferobj_mapped(vbo, MAP_INTERNAL)) {
               ctx->Driver.MapBufferRange(ctx, 0, vbo->Size, GL_MAP_READ_BIT, vbo,
                                          MAP_INTERNAL);
            }

            copy->varying[j].src_ptr =
                  ADD_POINTERS(vbo->Mappings[MAP_INTERNAL].Pointer, ptr);
         } else {
            copy->varying[j].src_ptr = ptr;
         }

         copy->dstarray[i].VertexAttrib = &copy->varying[j].dstattribs;
         copy->dstarray[i].BufferBinding = &copy->varying[j].dstbinding;
      }
   }

   /* There must always be an index buffer.  Currently require the
    * caller convert non-indexed prims to indexed.  Could alternately
    * do it internally.
    */
   if (copy->ib->obj) {
      if (!_mesa_bufferobj_mapped(copy->ib->obj, MAP_INTERNAL))
         ctx->Driver.MapBufferRange(ctx, 0, copy->ib->obj->Size, GL_MAP_READ_BIT,
                                    copy->ib->obj, MAP_INTERNAL);

      srcptr = (const GLubyte *)
         ADD_POINTERS(copy->ib->obj->Mappings[MAP_INTERNAL].Pointer,
                      copy->ib->ptr);
   } else
      srcptr = copy->ib->ptr;

   switch (copy->ib->index_size_shift) {
   case 0:
      copy->translated_elt_buf = malloc(sizeof(GLuint) * copy->ib->count);
      copy->srcelt = copy->translated_elt_buf;

      for (i = 0; i < copy->ib->count; i++)
         copy->translated_elt_buf[i] = ((const GLubyte *)srcptr)[i];
      break;

   case 1:
      copy->translated_elt_buf = malloc(sizeof(GLuint) * copy->ib->count);
      copy->srcelt = copy->translated_elt_buf;

      for (i = 0; i < copy->ib->count; i++)
         copy->translated_elt_buf[i] = ((const GLushort *)srcptr)[i];
      break;

   case 2:
      copy->translated_elt_buf = NULL;
      copy->srcelt = (const GLuint *)srcptr;
      break;
   }

   /* Figure out the maximum allowed vertex buffer size:
    */
   if (copy->vertex_size * copy->limits->max_verts <= copy->limits->max_vb_size) {
      copy->dstbuf_size = copy->limits->max_verts;
   }
   else {
      copy->dstbuf_size = copy->limits->max_vb_size / copy->vertex_size;
   }

   /* Allocate an output vertex buffer:
    *
    * XXX:  This should be a VBO!
    */
   copy->dstbuf = malloc(copy->dstbuf_size * copy->vertex_size);
   copy->dstptr = copy->dstbuf;

   /* Setup new vertex arrays to point into the output buffer:
    */
   for (offset = 0, i = 0; i < copy->nr_varying; i++) {
      const struct tnl_vertex_array *src = copy->varying[i].array;
      const struct gl_array_attributes *srcattr = src->VertexAttrib;
      struct tnl_vertex_array *dst = &copy->dstarray[copy->varying[i].attr];
      struct gl_vertex_buffer_binding *dstbind = &copy->varying[i].dstbinding;
      struct gl_array_attributes *dstattr = &copy->varying[i].dstattribs;

      dstattr->Format = srcattr->Format;
      dstattr->Ptr = copy->dstbuf + offset;
      dstbind->Stride = copy->vertex_size;
      dstbind->BufferObj = NULL;
      dst->BufferBinding = dstbind;
      dst->VertexAttrib = dstattr;

      offset += copy->varying[i].size;
   }

   /* Allocate an output element list:
    */
   copy->dstelt_size = MIN2(65536, copy->ib->count * 2 + 3);
   copy->dstelt_size = MIN2(copy->dstelt_size, copy->limits->max_indices);
   copy->dstelt = malloc(sizeof(GLuint) * copy->dstelt_size);
   copy->dstelt_nr = 0;

   /* Setup the new index buffer to point to the allocated element
    * list:
    */
   copy->dstib.count = 0;        /* duplicates dstelt_nr */
   copy->dstib.index_size_shift = 2;
   copy->dstib.obj = NULL;
   copy->dstib.ptr = copy->dstelt;
}


/**
 * Free up everything allocated during split/replay.
 */
static void
replay_finish(struct copy_context *copy)
{
   struct gl_context *ctx = copy->ctx;
   GLuint i;

   /* Free our vertex and index buffers */
   free(copy->translated_elt_buf);
   free(copy->dstbuf);
   free(copy->dstelt);

   /* Unmap VBO's */
   for (i = 0; i < copy->nr_varying; i++) {
      struct gl_buffer_object *vbo =
         copy->varying[i].array->BufferBinding->BufferObj;
      if (vbo && _mesa_bufferobj_mapped(vbo, MAP_INTERNAL))
         ctx->Driver.UnmapBuffer(ctx, vbo, MAP_INTERNAL);
   }

   /* Unmap index buffer */
   if (copy->ib->obj &&
       _mesa_bufferobj_mapped(copy->ib->obj, MAP_INTERNAL)) {
      ctx->Driver.UnmapBuffer(ctx, copy->ib->obj, MAP_INTERNAL);
   }
}


/**
 * Split VBO into smaller pieces, draw the pieces.
 */
void
_tnl_split_copy(struct gl_context *ctx,
                const struct tnl_vertex_array *arrays,
                const struct _mesa_prim *prim,
                GLuint nr_prims,
                const struct _mesa_index_buffer *ib,
                tnl_draw_func draw,
                const struct split_limits *limits)
{
   struct copy_context copy;
   GLuint i, this_nr_prims;

   for (i = 0; i < nr_prims;) {
      /* Our SW TNL pipeline doesn't handle basevertex yet, so bind_indices
       * will rebase the elements to the basevertex, and we'll only
       * emit strings of prims with the same basevertex in one draw call.
       */
      for (this_nr_prims = 1; i + this_nr_prims < nr_prims;
           this_nr_prims++) {
         if (prim[i].basevertex != prim[i + this_nr_prims].basevertex)
            break;
      }

      memset(&copy, 0, sizeof(copy));

      /* Require indexed primitives:
       */
      assert(ib);

      copy.ctx = ctx;
      copy.array = arrays;
      copy.prim = &prim[i];
      copy.nr_prims = this_nr_prims;
      copy.ib = ib;
      copy.draw = draw;
      copy.limits = limits;

      /* Clear the vertex cache:
       */
      for (i = 0; i < ELT_TABLE_SIZE; i++)
         copy.vert_cache[i].in = ~0;

      replay_init(&copy);
      replay_elts(&copy);
      replay_finish(&copy);
   }
}