aboutsummaryrefslogtreecommitdiffstats
path: root/src/gallium/drivers/ilo/ilo_3d.c
blob: 30686d6075f1029666ae010e20a9c758bb176945 (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
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
704
705
706
707
708
709
710
711
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
734
735
736
737
738
739
740
741
742
743
744
745
746
747
748
749
750
751
752
753
754
755
756
757
758
759
760
761
762
763
764
765
766
767
768
769
770
771
772
773
774
775
776
777
778
779
780
781
782
783
784
785
786
787
788
789
790
791
792
793
794
795
796
797
798
799
800
801
802
803
804
805
806
807
808
809
810
811
812
813
814
815
816
817
818
819
820
821
822
823
824
825
826
827
828
829
830
831
832
833
834
835
836
837
838
839
840
841
842
843
844
845
846
847
848
849
850
851
852
853
854
855
856
857
858
859
860
861
862
863
864
865
866
867
868
869
870
871
872
873
874
875
876
877
878
879
880
881
882
883
884
885
/*
 * Mesa 3-D graphics library
 *
 * Copyright (C) 2012-2013 LunarG, Inc.
 *
 * 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:
 *    Chia-I Wu <olv@lunarg.com>
 */

#include "util/u_prim.h"
#include "intel_winsys.h"

#include "ilo_3d_pipeline.h"
#include "ilo_blit.h"
#include "ilo_context.h"
#include "ilo_cp.h"
#include "ilo_query.h"
#include "ilo_shader.h"
#include "ilo_state.h"
#include "ilo_3d.h"

static void
process_query_for_occlusion_counter(struct ilo_3d *hw3d,
                                    struct ilo_query *q)
{
   uint64_t *vals, depth_count = 0;
   int i;

   /* in pairs */
   assert(q->reg_read % 2 == 0);

   vals = intel_bo_map(q->bo, false);
   for (i = 1; i < q->reg_read; i += 2)
      depth_count += vals[i] - vals[i - 1];
   intel_bo_unmap(q->bo);

   /* accumulate so that the query can be resumed if wanted */
   q->data.u64 += depth_count;
   q->reg_read = 0;
}

static uint64_t
timestamp_to_ns(uint64_t timestamp)
{
   /* see ilo_get_timestamp() */
   return (timestamp & 0xffffffff) * 80;
}

static void
process_query_for_timestamp(struct ilo_3d *hw3d, struct ilo_query *q)
{
   uint64_t *vals, timestamp;

   assert(q->reg_read == 1);

   vals = intel_bo_map(q->bo, false);
   timestamp = vals[0];
   intel_bo_unmap(q->bo);

   q->data.u64 = timestamp_to_ns(timestamp);
   q->reg_read = 0;
}

static void
process_query_for_time_elapsed(struct ilo_3d *hw3d, struct ilo_query *q)
{
   uint64_t *vals, elapsed = 0;
   int i;

   /* in pairs */
   assert(q->reg_read % 2 == 0);

   vals = intel_bo_map(q->bo, false);

   for (i = 1; i < q->reg_read; i += 2)
      elapsed += vals[i] - vals[i - 1];

   intel_bo_unmap(q->bo);

   /* accumulate so that the query can be resumed if wanted */
   q->data.u64 += timestamp_to_ns(elapsed);
   q->reg_read = 0;
}

static void
process_query_for_pipeline_statistics(struct ilo_3d *hw3d,
                                      struct ilo_query *q)
{
   const uint64_t *vals;
   int i;

   assert(q->reg_read % 22 == 0);

   vals = intel_bo_map(q->bo, false);

   for (i = 0; i < q->reg_read; i += 22) {
      struct pipe_query_data_pipeline_statistics *stats =
         &q->data.pipeline_statistics;
      const uint64_t *begin = vals + i;
      const uint64_t *end = begin + 11;

      stats->ia_vertices    += end[0] - begin[0];
      stats->ia_primitives  += end[1] - begin[1];
      stats->vs_invocations += end[2] - begin[2];
      stats->gs_invocations += end[3] - begin[3];
      stats->gs_primitives  += end[4] - begin[4];
      stats->c_invocations  += end[5] - begin[5];
      stats->c_primitives   += end[6] - begin[6];
      stats->ps_invocations += end[7] - begin[7];
      stats->hs_invocations += end[8] - begin[8];
      stats->ds_invocations += end[9] - begin[9];
      stats->cs_invocations += end[10] - begin[10];
   }

   intel_bo_unmap(q->bo);

   q->reg_read = 0;
}

static void
ilo_3d_resume_queries(struct ilo_3d *hw3d)
{
   struct ilo_query *q;

   /* resume occlusion queries */
   LIST_FOR_EACH_ENTRY(q, &hw3d->occlusion_queries, list) {
      /* accumulate the result if the bo is alreay full */
      if (q->reg_read >= q->reg_total)
         process_query_for_occlusion_counter(hw3d, q);

      ilo_3d_pipeline_emit_write_depth_count(hw3d->pipeline,
            q->bo, q->reg_read++);
   }

   /* resume timer queries */
   LIST_FOR_EACH_ENTRY(q, &hw3d->time_elapsed_queries, list) {
      /* accumulate the result if the bo is alreay full */
      if (q->reg_read >= q->reg_total)
         process_query_for_time_elapsed(hw3d, q);

      ilo_3d_pipeline_emit_write_timestamp(hw3d->pipeline,
            q->bo, q->reg_read++);
   }

   /* resume pipeline statistics queries */
   LIST_FOR_EACH_ENTRY(q, &hw3d->pipeline_statistics_queries, list) {
      /* accumulate the result if the bo is alreay full */
      if (q->reg_read >= q->reg_total)
         process_query_for_pipeline_statistics(hw3d, q);

      ilo_3d_pipeline_emit_write_statistics(hw3d->pipeline,
            q->bo, q->reg_read);
      q->reg_read += 11;
   }
}

static void
ilo_3d_pause_queries(struct ilo_3d *hw3d)
{
   struct ilo_query *q;

   /* pause occlusion queries */
   LIST_FOR_EACH_ENTRY(q, &hw3d->occlusion_queries, list) {
      assert(q->reg_read < q->reg_total);
      ilo_3d_pipeline_emit_write_depth_count(hw3d->pipeline,
            q->bo, q->reg_read++);
   }

   /* pause timer queries */
   LIST_FOR_EACH_ENTRY(q, &hw3d->time_elapsed_queries, list) {
      assert(q->reg_read < q->reg_total);
      ilo_3d_pipeline_emit_write_timestamp(hw3d->pipeline,
            q->bo, q->reg_read++);
   }

   /* pause pipeline statistics queries */
   LIST_FOR_EACH_ENTRY(q, &hw3d->pipeline_statistics_queries, list) {
      assert(q->reg_read < q->reg_total);
      ilo_3d_pipeline_emit_write_statistics(hw3d->pipeline,
            q->bo, q->reg_read);
      q->reg_read += 11;
   }
}

static void
ilo_3d_own_render_ring(struct ilo_3d *hw3d)
{
   ilo_cp_set_owner(hw3d->cp, INTEL_RING_RENDER, &hw3d->owner);
}

static void
ilo_3d_reserve_for_query(struct ilo_3d *hw3d, struct ilo_query *q,
                         enum ilo_3d_pipeline_action act)
{
   q->reg_cmd_size = ilo_3d_pipeline_estimate_size(hw3d->pipeline, act, NULL);

   /* XXX we should check the aperture size */
   if (ilo_cp_space(hw3d->cp) < q->reg_cmd_size * 2) {
      ilo_cp_submit(hw3d->cp, "out of space");
      assert(ilo_cp_space(hw3d->cp) >= q->reg_cmd_size * 2);
   }

   /* reserve space for pausing the query */
   hw3d->owner.reserve += q->reg_cmd_size;
}

/**
 * Begin a query.
 */
void
ilo_3d_begin_query(struct ilo_context *ilo, struct ilo_query *q)
{
   struct ilo_3d *hw3d = ilo->hw3d;

   ilo_3d_own_render_ring(hw3d);

   switch (q->type) {
   case PIPE_QUERY_OCCLUSION_COUNTER:
      ilo_3d_reserve_for_query(hw3d, q, ILO_3D_PIPELINE_WRITE_DEPTH_COUNT);
      q->data.u64 = 0;

      if (ilo_query_alloc_bo(q, 2, -1, hw3d->cp->winsys)) {
         ilo_3d_pipeline_emit_write_depth_count(hw3d->pipeline,
               q->bo, q->reg_read++);

         list_add(&q->list, &hw3d->occlusion_queries);
      }
      break;
   case PIPE_QUERY_TIMESTAMP:
      /* nop */
      break;
   case PIPE_QUERY_TIME_ELAPSED:
      ilo_3d_reserve_for_query(hw3d, q, ILO_3D_PIPELINE_WRITE_TIMESTAMP);
      q->data.u64 = 0;

      if (ilo_query_alloc_bo(q, 2, -1, hw3d->cp->winsys)) {
         ilo_3d_pipeline_emit_write_timestamp(hw3d->pipeline,
               q->bo, q->reg_read++);

         list_add(&q->list, &hw3d->time_elapsed_queries);
      }
      break;
   case PIPE_QUERY_PRIMITIVES_GENERATED:
      q->data.u64 = 0;
      list_add(&q->list, &hw3d->prim_generated_queries);
      break;
   case PIPE_QUERY_PRIMITIVES_EMITTED:
      q->data.u64 = 0;
      list_add(&q->list, &hw3d->prim_emitted_queries);
      break;
   case PIPE_QUERY_PIPELINE_STATISTICS:
      ilo_3d_reserve_for_query(hw3d, q, ILO_3D_PIPELINE_WRITE_STATISTICS);
      memset(&q->data.pipeline_statistics, 0,
            sizeof(q->data.pipeline_statistics));

      if (ilo_query_alloc_bo(q, 11 * 2, -1, hw3d->cp->winsys)) {
         ilo_3d_pipeline_emit_write_statistics(hw3d->pipeline,
               q->bo, q->reg_read);
         q->reg_read += 11;

         list_add(&q->list, &hw3d->pipeline_statistics_queries);
      }
      break;
   default:
      assert(!"unknown query type");
      break;
   }
}

/**
 * End a query.
 */
void
ilo_3d_end_query(struct ilo_context *ilo, struct ilo_query *q)
{
   struct ilo_3d *hw3d = ilo->hw3d;

   ilo_3d_own_render_ring(hw3d);

   switch (q->type) {
   case PIPE_QUERY_OCCLUSION_COUNTER:
      list_del(&q->list);

      assert(q->reg_read < q->reg_total);
      assert(hw3d->owner.reserve >= q->reg_cmd_size);
      hw3d->owner.reserve -= q->reg_cmd_size;

      ilo_3d_pipeline_emit_write_depth_count(hw3d->pipeline,
            q->bo, q->reg_read++);
      break;
   case PIPE_QUERY_TIMESTAMP:
      q->data.u64 = 0;

      if (ilo_query_alloc_bo(q, 1, 1, hw3d->cp->winsys)) {
         ilo_3d_pipeline_emit_write_timestamp(hw3d->pipeline,
               q->bo, q->reg_read++);
      }
      break;
   case PIPE_QUERY_TIME_ELAPSED:
      list_del(&q->list);

      assert(q->reg_read < q->reg_total);
      assert(hw3d->owner.reserve >= q->reg_cmd_size);
      hw3d->owner.reserve -= q->reg_cmd_size;

      ilo_3d_pipeline_emit_write_timestamp(hw3d->pipeline,
            q->bo, q->reg_read++);
      break;
   case PIPE_QUERY_PRIMITIVES_GENERATED:
   case PIPE_QUERY_PRIMITIVES_EMITTED:
      list_del(&q->list);
      break;
   case PIPE_QUERY_PIPELINE_STATISTICS:
      list_del(&q->list);

      assert(q->reg_read + 11 <= q->reg_total);
      assert(hw3d->owner.reserve >= q->reg_cmd_size);
      hw3d->owner.reserve -= q->reg_cmd_size;

      ilo_3d_pipeline_emit_write_statistics(hw3d->pipeline,
            q->bo, q->reg_read);
      q->reg_read += 11;
      break;
   default:
      assert(!"unknown query type");
      break;
   }
}

/**
 * Process the raw query data.
 */
void
ilo_3d_process_query(struct ilo_context *ilo, struct ilo_query *q)
{
   struct ilo_3d *hw3d = ilo->hw3d;

   switch (q->type) {
   case PIPE_QUERY_OCCLUSION_COUNTER:
      if (q->bo)
         process_query_for_occlusion_counter(hw3d, q);
      break;
   case PIPE_QUERY_TIMESTAMP:
      if (q->bo)
         process_query_for_timestamp(hw3d, q);
      break;
   case PIPE_QUERY_TIME_ELAPSED:
      if (q->bo)
         process_query_for_time_elapsed(hw3d, q);
      break;
   case PIPE_QUERY_PRIMITIVES_GENERATED:
   case PIPE_QUERY_PRIMITIVES_EMITTED:
      break;
   case PIPE_QUERY_PIPELINE_STATISTICS:
      if (q->bo)
         process_query_for_pipeline_statistics(hw3d, q);
      break;
   default:
      assert(!"unknown query type");
      break;
   }
}

/**
 * Hook for CP new-batch.
 */
void
ilo_3d_cp_submitted(struct ilo_3d *hw3d)
{
   /* invalidate the pipeline */
   ilo_3d_pipeline_invalidate(hw3d->pipeline,
         ILO_3D_PIPELINE_INVALIDATE_BATCH_BO |
         ILO_3D_PIPELINE_INVALIDATE_STATE_BO |
         ILO_3D_PIPELINE_INVALIDATE_KERNEL_BO);

   hw3d->new_batch = true;
}

static void
ilo_3d_own_cp(struct ilo_cp *cp, void *data)
{
   struct ilo_3d *hw3d = data;

   /* multiply by 2 for both resuming and pausing */
   if (ilo_cp_space(hw3d->cp) < hw3d->owner.reserve * 2) {
      ilo_cp_submit(hw3d->cp, "out of space");
      assert(ilo_cp_space(hw3d->cp) >= hw3d->owner.reserve * 2);
   }

   ilo_3d_resume_queries(hw3d);

   assert(ilo_cp_space(hw3d->cp) >= hw3d->owner.reserve);
}

static void
ilo_3d_release_cp(struct ilo_cp *cp, void *data)
{
   struct ilo_3d *hw3d = data;

   ilo_3d_pause_queries(hw3d);
}

/**
 * Create a 3D context.
 */
struct ilo_3d *
ilo_3d_create(struct ilo_cp *cp, const struct ilo_dev_info *dev)
{
   struct ilo_3d *hw3d;

   hw3d = CALLOC_STRUCT(ilo_3d);
   if (!hw3d)
      return NULL;

   hw3d->cp = cp;
   hw3d->owner.own = ilo_3d_own_cp;
   hw3d->owner.release = ilo_3d_release_cp;
   hw3d->owner.data = hw3d;
   hw3d->owner.reserve = 0;

   hw3d->new_batch = true;

   list_inithead(&hw3d->occlusion_queries);
   list_inithead(&hw3d->time_elapsed_queries);
   list_inithead(&hw3d->prim_generated_queries);
   list_inithead(&hw3d->prim_emitted_queries);
   list_inithead(&hw3d->pipeline_statistics_queries);

   hw3d->pipeline = ilo_3d_pipeline_create(cp, dev);
   if (!hw3d->pipeline) {
      FREE(hw3d);
      return NULL;
   }

   return hw3d;
}

/**
 * Destroy a 3D context.
 */
void
ilo_3d_destroy(struct ilo_3d *hw3d)
{
   ilo_3d_pipeline_destroy(hw3d->pipeline);
   FREE(hw3d);
}

static bool
draw_vbo(struct ilo_3d *hw3d, const struct ilo_state_vector *vec,
         int *prim_generated, int *prim_emitted)
{
   bool need_flush = false;
   int max_len;

   ilo_3d_own_render_ring(hw3d);

   if (!hw3d->new_batch) {
      /*
       * Without a better tracking mechanism, when the framebuffer changes, we
       * have to assume that the old framebuffer may be sampled from.  If that
       * happens in the middle of a batch buffer, we need to insert manual
       * flushes.
       */
      need_flush = (vec->dirty & ILO_DIRTY_FB);

      /* same to SO target changes */
      need_flush |= (vec->dirty & ILO_DIRTY_SO);
   }

   /* make sure there is enough room first */
   max_len = ilo_3d_pipeline_estimate_size(hw3d->pipeline,
         ILO_3D_PIPELINE_DRAW, vec);
   if (need_flush) {
      max_len += ilo_3d_pipeline_estimate_size(hw3d->pipeline,
            ILO_3D_PIPELINE_FLUSH, NULL);
   }

   if (max_len > ilo_cp_space(hw3d->cp)) {
      ilo_cp_submit(hw3d->cp, "out of space");
      need_flush = false;
      assert(max_len <= ilo_cp_space(hw3d->cp));
   }

   if (need_flush)
      ilo_3d_pipeline_emit_flush(hw3d->pipeline);

   return ilo_3d_pipeline_emit_draw(hw3d->pipeline, vec,
         prim_generated, prim_emitted);
}

static void
update_prim_count(struct ilo_3d *hw3d, int generated, int emitted)
{
   struct ilo_query *q;

   LIST_FOR_EACH_ENTRY(q, &hw3d->prim_generated_queries, list)
      q->data.u64 += generated;

   LIST_FOR_EACH_ENTRY(q, &hw3d->prim_emitted_queries, list)
      q->data.u64 += emitted;
}

bool
ilo_3d_pass_render_condition(struct ilo_context *ilo)
{
   struct ilo_3d *hw3d = ilo->hw3d;
   uint64_t result;
   bool wait;

   if (!hw3d->render_condition.query)
      return true;

   switch (hw3d->render_condition.mode) {
   case PIPE_RENDER_COND_WAIT:
   case PIPE_RENDER_COND_BY_REGION_WAIT:
      wait = true;
      break;
   case PIPE_RENDER_COND_NO_WAIT:
   case PIPE_RENDER_COND_BY_REGION_NO_WAIT:
   default:
      wait = false;
      break;
   }

   if (ilo->base.get_query_result(&ilo->base, hw3d->render_condition.query,
            wait, (union pipe_query_result *) &result))
      return (!result == hw3d->render_condition.cond);
   else
      return true;
}

void
ilo_3d_draw_rectlist(struct ilo_3d *hw3d, const struct ilo_blitter *blitter)
{
   ilo_3d_own_render_ring(hw3d);

   /*
    * From the Sandy Bridge PRM, volume 2 part 1, page 313:
    *
    *     "If other rendering operations have preceded this clear, a
    *      PIPE_CONTROL with write cache flush enabled and Z-inhibit
    *      disabled must be issued before the rectangle primitive used for
    *      the depth buffer clear operation."
    *
    * From the Sandy Bridge PRM, volume 2 part 1, page 314:
    *
    *     "Depth buffer clear pass must be followed by a PIPE_CONTROL
    *      command with DEPTH_STALL bit set and Then followed by Depth
    *      FLUSH"
    *
    * But the pipeline has to be flushed both before and after not only
    * because of these workarounds.  We need them for reasons such as
    *
    *  - we may sample from a texture that was rendered to
    *  - we may sample from the fb shortly after
    *
    * Skip checking blitter->op and do the flushes.
    *
    * XXX need space check
    */
   if (!hw3d->new_batch)
      ilo_3d_pipeline_emit_flush(hw3d->pipeline);

   ilo_3d_pipeline_emit_rectlist(hw3d->pipeline, blitter);

   ilo_3d_pipeline_emit_flush(hw3d->pipeline);

   hw3d->new_batch = false;
}

#define UPDATE_MIN2(a, b) (a) = MIN2((a), (b))
#define UPDATE_MAX2(a, b) (a) = MAX2((a), (b))

/**
 * \see find_sub_primitives() from core mesa
 */
static int
ilo_find_sub_primitives(const void *elements, unsigned element_size,
                    const struct pipe_draw_info *orig_info,
                    struct pipe_draw_info *info)
{
   const unsigned max_prims = orig_info->count - orig_info->start;
   unsigned i, cur_start, cur_count;
   int scan_index;
   unsigned scan_num;

   cur_start = orig_info->start;
   cur_count = 0;
   scan_num = 0;

#define IB_INDEX_READ(TYPE, INDEX) (((const TYPE *) elements)[INDEX])

#define SCAN_ELEMENTS(TYPE) \
   info[scan_num] = *orig_info; \
   info[scan_num].primitive_restart = false; \
   for (i = orig_info->start; i < orig_info->count; i++) { \
      scan_index = IB_INDEX_READ(TYPE, i); \
      if (scan_index == orig_info->restart_index) { \
         if (cur_count > 0) { \
            assert(scan_num < max_prims); \
            info[scan_num].start = cur_start; \
            info[scan_num].count = cur_count; \
            scan_num++; \
            info[scan_num] = *orig_info; \
            info[scan_num].primitive_restart = false; \
         } \
         cur_start = i + 1; \
         cur_count = 0; \
      } \
      else { \
         UPDATE_MIN2(info[scan_num].min_index, scan_index); \
         UPDATE_MAX2(info[scan_num].max_index, scan_index); \
         cur_count++; \
      } \
   } \
   if (cur_count > 0) { \
      assert(scan_num < max_prims); \
      info[scan_num].start = cur_start; \
      info[scan_num].count = cur_count; \
      scan_num++; \
   }

   switch (element_size) {
   case 1:
      SCAN_ELEMENTS(uint8_t);
      break;
   case 2:
      SCAN_ELEMENTS(uint16_t);
      break;
   case 4:
      SCAN_ELEMENTS(uint32_t);
      break;
   default:
      assert(0 && "bad index_size in find_sub_primitives()");
   }

#undef SCAN_ELEMENTS

   return scan_num;
}

static inline bool
ilo_check_restart_index(const struct ilo_context *ilo, unsigned restart_index)
{
   /*
    * Haswell (GEN(7.5)) supports an arbitrary cut index, check everything
    * older.
    */
   if (ilo_dev_gen(ilo->dev) >= ILO_GEN(7.5))
      return true;

   /* Note: indices must be unsigned byte, unsigned short or unsigned int */
   switch (ilo->state_vector.ib.index_size) {
   case 1:
      return ((restart_index & 0xff) == 0xff);
      break;
   case 2:
      return ((restart_index & 0xffff) == 0xffff);
      break;
   case 4:
      return (restart_index == 0xffffffff);
      break;
   }
   return false;
}

static inline bool
ilo_check_restart_prim_type(const struct ilo_context *ilo, unsigned prim)
{
   switch (prim) {
   case PIPE_PRIM_POINTS:
   case PIPE_PRIM_LINES:
   case PIPE_PRIM_LINE_STRIP:
   case PIPE_PRIM_TRIANGLES:
   case PIPE_PRIM_TRIANGLE_STRIP:
      /* All 965 GEN graphics support a cut index for these primitive types */
      return true;
      break;

   case PIPE_PRIM_LINE_LOOP:
   case PIPE_PRIM_POLYGON:
   case PIPE_PRIM_QUAD_STRIP:
   case PIPE_PRIM_QUADS:
   case PIPE_PRIM_TRIANGLE_FAN:
      if (ilo_dev_gen(ilo->dev) >= ILO_GEN(7.5)) {
         /* Haswell and newer parts can handle these prim types. */
         return true;
      }
      break;
   }

   return false;
}

/*
 * Handle VBOs using primitive restart.
 * Verify that restart index and primitive type can be handled by the HW.
 * Return true if this routine did the rendering
 * Return false if this routine did NOT render because restart can be handled
 * in HW.
 */
static void
ilo_draw_vbo_with_sw_restart(struct pipe_context *pipe,
                             const struct pipe_draw_info *info)
{
   struct ilo_state_vector *vec = &ilo_context(pipe)->state_vector;
   struct pipe_draw_info *restart_info = NULL;
   int sub_prim_count = 1;

   /*
    * We have to break up the primitive into chunks manually
    * Worst case, every other index could be a restart index so
    * need to have space for that many primitives
    */
   restart_info = MALLOC(((info->count + 1) / 2) * sizeof(*info));
   if (NULL == restart_info) {
      /* If we can't get memory for this, bail out */
      ilo_err("%s:%d - Out of memory", __FILE__, __LINE__);
      return;
   }

   if (vec->ib.buffer) {
      struct pipe_transfer *transfer;
      const void *map;

      map = pipe_buffer_map(pipe, vec->ib.buffer,
            PIPE_TRANSFER_READ, &transfer);

      sub_prim_count = ilo_find_sub_primitives(map + vec->ib.offset,
            vec->ib.index_size, info, restart_info);

      pipe_buffer_unmap(pipe, transfer);
   }
   else {
      sub_prim_count =
         ilo_find_sub_primitives(vec->ib.user_buffer,
               vec->ib.index_size, info, restart_info);
   }

   info = restart_info;

   while (sub_prim_count > 0) {
      pipe->draw_vbo(pipe, info);

      sub_prim_count--;
      info++;
   }

   FREE(restart_info);
}

static void
ilo_draw_vbo(struct pipe_context *pipe, const struct pipe_draw_info *info)
{
   struct ilo_context *ilo = ilo_context(pipe);
   struct ilo_3d *hw3d = ilo->hw3d;
   int prim_generated, prim_emitted;

   if (ilo_debug & ILO_DEBUG_DRAW) {
      if (info->indexed) {
         ilo_printf("indexed draw %s: "
               "index start %d, count %d, vertex range [%d, %d]\n",
               u_prim_name(info->mode), info->start, info->count,
               info->min_index, info->max_index);
      }
      else {
         ilo_printf("draw %s: vertex start %d, count %d\n",
               u_prim_name(info->mode), info->start, info->count);
      }

      ilo_state_vector_dump_dirty(&ilo->state_vector);
   }

   if (!ilo_3d_pass_render_condition(ilo))
      return;

   if (info->primitive_restart && info->indexed) {
      /*
       * Want to draw an indexed primitive using primitive restart
       * Check that HW can handle the request and fall to SW if not.
       */
      if (!ilo_check_restart_index(ilo, info->restart_index) ||
          !ilo_check_restart_prim_type(ilo, info->mode)) {
         ilo_draw_vbo_with_sw_restart(pipe, info);
         return;
      }
   }

   ilo_finalize_3d_states(ilo, info);

   ilo_shader_cache_upload(ilo->shader_cache, &hw3d->cp->builder);

   ilo_blit_resolve_framebuffer(ilo);

   /* If draw_vbo ever fails, return immediately. */
   if (!draw_vbo(hw3d, &ilo->state_vector, &prim_generated, &prim_emitted))
      return;

   /* clear dirty status */
   ilo->state_vector.dirty = 0x0;
   hw3d->new_batch = false;

   /* avoid dangling pointer reference */
   ilo->state_vector.draw = NULL;

   update_prim_count(hw3d, prim_generated, prim_emitted);

   if (ilo_debug & ILO_DEBUG_NOCACHE)
      ilo_3d_pipeline_emit_flush(hw3d->pipeline);
}

static void
ilo_render_condition(struct pipe_context *pipe,
                     struct pipe_query *query,
                     boolean condition,
                     uint mode)
{
   struct ilo_context *ilo = ilo_context(pipe);
   struct ilo_3d *hw3d = ilo->hw3d;

   /* reference count? */
   hw3d->render_condition.query = query;
   hw3d->render_condition.mode = mode;
   hw3d->render_condition.cond = condition;
}

static void
ilo_texture_barrier(struct pipe_context *pipe)
{
   struct ilo_context *ilo = ilo_context(pipe);
   struct ilo_3d *hw3d = ilo->hw3d;

   if (ilo->cp->ring != INTEL_RING_RENDER)
      return;

   ilo_3d_pipeline_emit_flush(hw3d->pipeline);

   /* don't know why */
   if (ilo_dev_gen(ilo->dev) >= ILO_GEN(7))
      ilo_cp_submit(hw3d->cp, "texture barrier");
}

static void
ilo_get_sample_position(struct pipe_context *pipe,
                        unsigned sample_count,
                        unsigned sample_index,
                        float *out_value)
{
   struct ilo_context *ilo = ilo_context(pipe);
   struct ilo_3d *hw3d = ilo->hw3d;

   ilo_3d_pipeline_get_sample_position(hw3d->pipeline,
         sample_count, sample_index,
         &out_value[0], &out_value[1]);
}

/**
 * Initialize 3D-related functions.
 */
void
ilo_init_3d_functions(struct ilo_context *ilo)
{
   ilo->base.draw_vbo = ilo_draw_vbo;
   ilo->base.render_condition = ilo_render_condition;
   ilo->base.texture_barrier = ilo_texture_barrier;
   ilo->base.get_sample_position = ilo_get_sample_position;
}