aboutsummaryrefslogtreecommitdiffstats
path: root/src/mesa/drivers/dri/i965/brw_misc_state.c
blob: 78a46cb050dde3fb81644ee3e7e7feb08f60dbfb (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
886
887
888
889
890
891
892
893
894
895
896
897
898
899
900
901
902
903
904
905
906
907
908
909
910
911
912
913
914
915
916
917
918
919
920
921
922
923
924
925
926
927
928
929
930
931
932
933
934
935
936
937
938
939
940
941
942
943
944
945
946
947
948
949
950
951
952
953
954
955
956
957
958
959
960
961
962
963
964
965
966
967
968
969
970
971
972
973
974
975
976
977
978
979
980
981
982
983
984
985
986
987
988
989
990
991
992
993
994
995
996
997
998
999
1000
1001
1002
1003
1004
1005
1006
1007
1008
1009
1010
1011
1012
1013
1014
1015
1016
1017
1018
1019
1020
1021
1022
1023
1024
1025
1026
1027
/*
 Copyright (C) Intel Corp.  2006.  All Rights Reserved.
 Intel funded Tungsten Graphics to
 develop this 3D driver.

 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 (including the
 next paragraph) 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 COPYRIGHT OWNER(S) AND/OR ITS SUPPLIERS 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>
  */



#include "intel_batchbuffer.h"
#include "intel_fbo.h"
#include "intel_mipmap_tree.h"

#include "brw_context.h"
#include "brw_state.h"
#include "brw_defines.h"

#include "main/fbobject.h"
#include "main/glformats.h"

/* Constant single cliprect for framebuffer object or DRI2 drawing */
static void upload_drawing_rect(struct brw_context *brw)
{
   struct gl_context *ctx = &brw->ctx;

   BEGIN_BATCH(4);
   OUT_BATCH(_3DSTATE_DRAWING_RECTANGLE << 16 | (4 - 2));
   OUT_BATCH(0); /* xmin, ymin */
   OUT_BATCH(((ctx->DrawBuffer->Width - 1) & 0xffff) |
	    ((ctx->DrawBuffer->Height - 1) << 16));
   OUT_BATCH(0);
   ADVANCE_BATCH();
}

const struct brw_tracked_state brw_drawing_rect = {
   .dirty = {
      .mesa = _NEW_BUFFERS,
      .brw = BRW_NEW_CONTEXT,
   },
   .emit = upload_drawing_rect
};

/**
 * Upload pointers to the per-stage state.
 *
 * The state pointers in this packet are all relative to the general state
 * base address set by CMD_STATE_BASE_ADDRESS, which is 0.
 */
static void upload_pipelined_state_pointers(struct brw_context *brw )
{
   if (brw->gen == 5) {
      /* Need to flush before changing clip max threads for errata. */
      BEGIN_BATCH(1);
      OUT_BATCH(MI_FLUSH);
      ADVANCE_BATCH();
   }

   BEGIN_BATCH(7);
   OUT_BATCH(_3DSTATE_PIPELINED_POINTERS << 16 | (7 - 2));
   OUT_RELOC(brw->batch.bo, I915_GEM_DOMAIN_INSTRUCTION, 0,
	     brw->vs.base.state_offset);
   if (brw->ff_gs.prog_active)
      OUT_RELOC(brw->batch.bo, I915_GEM_DOMAIN_INSTRUCTION, 0,
		brw->ff_gs.state_offset | 1);
   else
      OUT_BATCH(0);
   OUT_RELOC(brw->batch.bo, I915_GEM_DOMAIN_INSTRUCTION, 0,
	     brw->clip.state_offset | 1);
   OUT_RELOC(brw->batch.bo, I915_GEM_DOMAIN_INSTRUCTION, 0,
	     brw->sf.state_offset);
   OUT_RELOC(brw->batch.bo, I915_GEM_DOMAIN_INSTRUCTION, 0,
	     brw->wm.base.state_offset);
   OUT_RELOC(brw->batch.bo, I915_GEM_DOMAIN_INSTRUCTION, 0,
	     brw->cc.state_offset);
   ADVANCE_BATCH();

   brw->ctx.NewDriverState |= BRW_NEW_PSP;
}

static void upload_psp_urb_cbs(struct brw_context *brw )
{
   upload_pipelined_state_pointers(brw);
   brw_upload_urb_fence(brw);
   brw_upload_cs_urb_state(brw);
}

const struct brw_tracked_state brw_psp_urb_cbs = {
   .dirty = {
      .mesa = 0,
      .brw = BRW_NEW_BATCH |
             BRW_NEW_FF_GS_PROG_DATA |
             BRW_NEW_GEN4_UNIT_STATE |
             BRW_NEW_STATE_BASE_ADDRESS |
             BRW_NEW_URB_FENCE,
   },
   .emit = upload_psp_urb_cbs,
};

uint32_t
brw_depthbuffer_format(struct brw_context *brw)
{
   struct gl_context *ctx = &brw->ctx;
   struct gl_framebuffer *fb = ctx->DrawBuffer;
   struct intel_renderbuffer *drb = intel_get_renderbuffer(fb, BUFFER_DEPTH);
   struct intel_renderbuffer *srb;

   if (!drb &&
       (srb = intel_get_renderbuffer(fb, BUFFER_STENCIL)) &&
       !srb->mt->stencil_mt &&
       (intel_rb_format(srb) == MESA_FORMAT_Z24_UNORM_S8_UINT ||
	intel_rb_format(srb) == MESA_FORMAT_Z32_FLOAT_S8X24_UINT)) {
      drb = srb;
   }

   if (!drb)
      return BRW_DEPTHFORMAT_D32_FLOAT;

   return brw_depth_format(brw, drb->mt->format);
}

/**
 * Returns the mask of how many bits of x and y must be handled through the
 * depthbuffer's draw offset x and y fields.
 *
 * The draw offset x/y field of the depthbuffer packet is unfortunately shared
 * between the depth, hiz, and stencil buffers.  Because it can be hard to get
 * all 3 to agree on this value, we want to do as much drawing offset
 * adjustment as possible by moving the base offset of the 3 buffers, which is
 * restricted to tile boundaries.
 *
 * For each buffer, the remainder must be applied through the x/y draw offset.
 * This returns the worst-case mask of the low bits that have to go into the
 * packet.  If the 3 buffers don't agree on the drawing offset ANDed with this
 * mask, then we're in trouble.
 */
void
brw_get_depthstencil_tile_masks(struct intel_mipmap_tree *depth_mt,
                                uint32_t depth_level,
                                uint32_t depth_layer,
                                struct intel_mipmap_tree *stencil_mt,
                                uint32_t *out_tile_mask_x,
                                uint32_t *out_tile_mask_y)
{
   uint32_t tile_mask_x = 0, tile_mask_y = 0;

   if (depth_mt) {
      intel_miptree_get_tile_masks(depth_mt, &tile_mask_x, &tile_mask_y, false);

      if (intel_miptree_level_has_hiz(depth_mt, depth_level)) {
         uint32_t hiz_tile_mask_x, hiz_tile_mask_y;
         intel_miptree_get_tile_masks(depth_mt->hiz_buf->mt,
                                      &hiz_tile_mask_x, &hiz_tile_mask_y,
                                      false);

         /* Each HiZ row represents 2 rows of pixels */
         hiz_tile_mask_y = hiz_tile_mask_y << 1 | 1;

         tile_mask_x |= hiz_tile_mask_x;
         tile_mask_y |= hiz_tile_mask_y;
      }
   }

   if (stencil_mt) {
      if (stencil_mt->stencil_mt)
	 stencil_mt = stencil_mt->stencil_mt;

      if (stencil_mt->format == MESA_FORMAT_S_UINT8) {
         /* Separate stencil buffer uses 64x64 tiles. */
         tile_mask_x |= 63;
         tile_mask_y |= 63;
      } else {
         uint32_t stencil_tile_mask_x, stencil_tile_mask_y;
         intel_miptree_get_tile_masks(stencil_mt,
                                      &stencil_tile_mask_x,
                                      &stencil_tile_mask_y, false);

         tile_mask_x |= stencil_tile_mask_x;
         tile_mask_y |= stencil_tile_mask_y;
      }
   }

   *out_tile_mask_x = tile_mask_x;
   *out_tile_mask_y = tile_mask_y;
}

static struct intel_mipmap_tree *
get_stencil_miptree(struct intel_renderbuffer *irb)
{
   if (!irb)
      return NULL;
   if (irb->mt->stencil_mt)
      return irb->mt->stencil_mt;
   return irb->mt;
}

void
brw_workaround_depthstencil_alignment(struct brw_context *brw,
                                      GLbitfield clear_mask)
{
   struct gl_context *ctx = &brw->ctx;
   struct gl_framebuffer *fb = ctx->DrawBuffer;
   bool rebase_depth = false;
   bool rebase_stencil = false;
   struct intel_renderbuffer *depth_irb = intel_get_renderbuffer(fb, BUFFER_DEPTH);
   struct intel_renderbuffer *stencil_irb = intel_get_renderbuffer(fb, BUFFER_STENCIL);
   struct intel_mipmap_tree *depth_mt = NULL;
   struct intel_mipmap_tree *stencil_mt = get_stencil_miptree(stencil_irb);
   uint32_t tile_x = 0, tile_y = 0, stencil_tile_x = 0, stencil_tile_y = 0;
   uint32_t stencil_draw_x = 0, stencil_draw_y = 0;
   bool invalidate_depth = clear_mask & BUFFER_BIT_DEPTH;
   bool invalidate_stencil = clear_mask & BUFFER_BIT_STENCIL;

   if (depth_irb)
      depth_mt = depth_irb->mt;

   /* Initialize brw->depthstencil to 'nop' workaround state.
    */
   brw->depthstencil.tile_x = 0;
   brw->depthstencil.tile_y = 0;
   brw->depthstencil.depth_offset = 0;
   brw->depthstencil.stencil_offset = 0;
   brw->depthstencil.hiz_offset = 0;
   brw->depthstencil.depth_mt = NULL;
   brw->depthstencil.stencil_mt = NULL;
   if (depth_irb)
      brw->depthstencil.depth_mt = depth_mt;
   if (stencil_irb)
      brw->depthstencil.stencil_mt = get_stencil_miptree(stencil_irb);

   /* Gen6+ doesn't require the workarounds, since we always program the
    * surface state at the start of the whole surface.
    */
   if (brw->gen >= 6)
      return;

   /* Check if depth buffer is in depth/stencil format.  If so, then it's only
    * safe to invalidate it if we're also clearing stencil, and both depth_irb
    * and stencil_irb point to the same miptree.
    *
    * Note: it's not sufficient to check for the case where
    * _mesa_get_format_base_format(depth_mt->format) == GL_DEPTH_STENCIL,
    * because this fails to catch depth/stencil buffers on hardware that uses
    * separate stencil.  To catch that case, we check whether
    * depth_mt->stencil_mt is non-NULL.
    */
   if (depth_irb && invalidate_depth &&
       (_mesa_get_format_base_format(depth_mt->format) == GL_DEPTH_STENCIL ||
        depth_mt->stencil_mt)) {
      invalidate_depth = invalidate_stencil && depth_irb && stencil_irb
         && depth_irb->mt == stencil_irb->mt;
   }

   uint32_t tile_mask_x, tile_mask_y;
   brw_get_depthstencil_tile_masks(depth_mt,
                                   depth_mt ? depth_irb->mt_level : 0,
                                   depth_mt ? depth_irb->mt_layer : 0,
                                   stencil_mt,
                                   &tile_mask_x, &tile_mask_y);

   if (depth_irb) {
      tile_x = depth_irb->draw_x & tile_mask_x;
      tile_y = depth_irb->draw_y & tile_mask_y;

      /* According to the Sandy Bridge PRM, volume 2 part 1, pp326-327
       * (3DSTATE_DEPTH_BUFFER dw5), in the documentation for "Depth
       * Coordinate Offset X/Y":
       *
       *   "The 3 LSBs of both offsets must be zero to ensure correct
       *   alignment"
       */
      if (tile_x & 7 || tile_y & 7)
         rebase_depth = true;

      /* We didn't even have intra-tile offsets before g45. */
      if (!brw->has_surface_tile_offset) {
         if (tile_x || tile_y)
            rebase_depth = true;
      }

      if (rebase_depth) {
         perf_debug("HW workaround: blitting depth level %d to a temporary "
                    "to fix alignment (depth tile offset %d,%d)\n",
                    depth_irb->mt_level, tile_x, tile_y);
         intel_renderbuffer_move_to_temp(brw, depth_irb, invalidate_depth);
         /* In the case of stencil_irb being the same packed depth/stencil
          * texture but not the same rb, make it point at our rebased mt, too.
          */
         if (stencil_irb &&
             stencil_irb != depth_irb &&
             stencil_irb->mt == depth_mt) {
            intel_miptree_reference(&stencil_irb->mt, depth_irb->mt);
            intel_renderbuffer_set_draw_offset(stencil_irb);
         }

         stencil_mt = get_stencil_miptree(stencil_irb);

         tile_x = depth_irb->draw_x & tile_mask_x;
         tile_y = depth_irb->draw_y & tile_mask_y;
      }

      if (stencil_irb) {
         stencil_mt = get_stencil_miptree(stencil_irb);
         intel_miptree_get_image_offset(stencil_mt,
                                        stencil_irb->mt_level,
                                        stencil_irb->mt_layer,
                                        &stencil_draw_x, &stencil_draw_y);
         int stencil_tile_x = stencil_draw_x & tile_mask_x;
         int stencil_tile_y = stencil_draw_y & tile_mask_y;

         /* If stencil doesn't match depth, then we'll need to rebase stencil
          * as well.  (if we hadn't decided to rebase stencil before, the
          * post-stencil depth test will also rebase depth to try to match it
          * up).
          */
         if (tile_x != stencil_tile_x ||
             tile_y != stencil_tile_y) {
            rebase_stencil = true;
         }
      }
   }

   /* If we have (just) stencil, check it for ignored low bits as well */
   if (stencil_irb) {
      intel_miptree_get_image_offset(stencil_mt,
                                     stencil_irb->mt_level,
                                     stencil_irb->mt_layer,
                                     &stencil_draw_x, &stencil_draw_y);
      stencil_tile_x = stencil_draw_x & tile_mask_x;
      stencil_tile_y = stencil_draw_y & tile_mask_y;

      if (stencil_tile_x & 7 || stencil_tile_y & 7)
         rebase_stencil = true;

      if (!brw->has_surface_tile_offset) {
         if (stencil_tile_x || stencil_tile_y)
            rebase_stencil = true;
      }
   }

   if (rebase_stencil) {
      perf_debug("HW workaround: blitting stencil level %d to a temporary "
                 "to fix alignment (stencil tile offset %d,%d)\n",
                 stencil_irb->mt_level, stencil_tile_x, stencil_tile_y);

      intel_renderbuffer_move_to_temp(brw, stencil_irb, invalidate_stencil);
      stencil_mt = get_stencil_miptree(stencil_irb);

      intel_miptree_get_image_offset(stencil_mt,
                                     stencil_irb->mt_level,
                                     stencil_irb->mt_layer,
                                     &stencil_draw_x, &stencil_draw_y);
      stencil_tile_x = stencil_draw_x & tile_mask_x;
      stencil_tile_y = stencil_draw_y & tile_mask_y;

      if (depth_irb && depth_irb->mt == stencil_irb->mt) {
         intel_miptree_reference(&depth_irb->mt, stencil_irb->mt);
         intel_renderbuffer_set_draw_offset(depth_irb);
      } else if (depth_irb && !rebase_depth) {
         if (tile_x != stencil_tile_x ||
             tile_y != stencil_tile_y) {
            perf_debug("HW workaround: blitting depth level %d to a temporary "
                       "to match stencil level %d alignment (depth tile offset "
                       "%d,%d, stencil offset %d,%d)\n",
                       depth_irb->mt_level,
                       stencil_irb->mt_level,
                       tile_x, tile_y,
                       stencil_tile_x, stencil_tile_y);

            intel_renderbuffer_move_to_temp(brw, depth_irb, invalidate_depth);

            tile_x = depth_irb->draw_x & tile_mask_x;
            tile_y = depth_irb->draw_y & tile_mask_y;

            if (stencil_irb && stencil_irb->mt == depth_mt) {
               intel_miptree_reference(&stencil_irb->mt, depth_irb->mt);
               intel_renderbuffer_set_draw_offset(stencil_irb);
            }

            WARN_ONCE(stencil_tile_x != tile_x ||
                      stencil_tile_y != tile_y,
                      "Rebased stencil tile offset (%d,%d) doesn't match depth "
                      "tile offset (%d,%d).\n",
                      stencil_tile_x, stencil_tile_y,
                      tile_x, tile_y);
         }
      }
   }

   if (!depth_irb) {
      tile_x = stencil_tile_x;
      tile_y = stencil_tile_y;
   }

   /* While we just tried to get everything aligned, we may have failed to do
    * so in the case of rendering to array or 3D textures, where nonzero faces
    * will still have an offset post-rebase.  At least give an informative
    * warning.
    */
   WARN_ONCE((tile_x & 7) || (tile_y & 7),
             "Depth/stencil buffer needs alignment to 8-pixel boundaries.\n"
             "Truncating offset, bad rendering may occur.\n");
   tile_x &= ~7;
   tile_y &= ~7;

   /* Now, after rebasing, save off the new dephtstencil state so the hardware
    * packets can just dereference that without re-calculating tile offsets.
    */
   brw->depthstencil.tile_x = tile_x;
   brw->depthstencil.tile_y = tile_y;
   if (depth_irb) {
      depth_mt = depth_irb->mt;
      brw->depthstencil.depth_mt = depth_mt;
      brw->depthstencil.depth_offset =
         intel_miptree_get_aligned_offset(depth_mt,
                                          depth_irb->draw_x & ~tile_mask_x,
                                          depth_irb->draw_y & ~tile_mask_y,
                                          false);
      if (intel_renderbuffer_has_hiz(depth_irb)) {
         brw->depthstencil.hiz_offset =
            intel_miptree_get_aligned_offset(depth_mt,
                                             depth_irb->draw_x & ~tile_mask_x,
                                             (depth_irb->draw_y & ~tile_mask_y) / 2,
                                             false);
      }
   }
   if (stencil_irb) {
      stencil_mt = get_stencil_miptree(stencil_irb);

      brw->depthstencil.stencil_mt = stencil_mt;
      if (stencil_mt->format == MESA_FORMAT_S_UINT8) {
         /* Note: we can't compute the stencil offset using
          * intel_region_get_aligned_offset(), because stencil_region claims
          * that the region is untiled even though it's W tiled.
          */
         brw->depthstencil.stencil_offset =
            (stencil_draw_y & ~tile_mask_y) * stencil_mt->pitch +
            (stencil_draw_x & ~tile_mask_x) * 64;
      }
   }
}

void
brw_emit_depthbuffer(struct brw_context *brw)
{
   struct gl_context *ctx = &brw->ctx;
   struct gl_framebuffer *fb = ctx->DrawBuffer;
   /* _NEW_BUFFERS */
   struct intel_renderbuffer *depth_irb = intel_get_renderbuffer(fb, BUFFER_DEPTH);
   struct intel_renderbuffer *stencil_irb = intel_get_renderbuffer(fb, BUFFER_STENCIL);
   struct intel_mipmap_tree *depth_mt = brw->depthstencil.depth_mt;
   struct intel_mipmap_tree *stencil_mt = brw->depthstencil.stencil_mt;
   uint32_t tile_x = brw->depthstencil.tile_x;
   uint32_t tile_y = brw->depthstencil.tile_y;
   bool hiz = depth_irb && intel_renderbuffer_has_hiz(depth_irb);
   bool separate_stencil = false;
   uint32_t depth_surface_type = BRW_SURFACE_NULL;
   uint32_t depthbuffer_format = BRW_DEPTHFORMAT_D32_FLOAT;
   uint32_t depth_offset = 0;
   uint32_t width = 1, height = 1;

   if (stencil_mt) {
      separate_stencil = stencil_mt->format == MESA_FORMAT_S_UINT8;

      /* Gen7 supports only separate stencil */
      assert(separate_stencil || brw->gen < 7);
   }

   /* If there's a packed depth/stencil bound to stencil only, we need to
    * emit the packed depth/stencil buffer packet.
    */
   if (!depth_irb && stencil_irb && !separate_stencil) {
      depth_irb = stencil_irb;
      depth_mt = stencil_mt;
   }

   if (depth_irb && depth_mt) {
      /* When 3DSTATE_DEPTH_BUFFER.Separate_Stencil_Enable is set, then
       * 3DSTATE_DEPTH_BUFFER.Surface_Format is not permitted to be a packed
       * depthstencil format.
       *
       * Gens prior to 7 require that HiZ_Enable and Separate_Stencil_Enable be
       * set to the same value. Gens after 7 implicitly always set
       * Separate_Stencil_Enable; software cannot disable it.
       */
      if ((brw->gen < 7 && hiz) || brw->gen >= 7) {
         assert(!_mesa_is_format_packed_depth_stencil(depth_mt->format));
      }

      /* Prior to Gen7, if using separate stencil, hiz must be enabled. */
      assert(brw->gen >= 7 || !separate_stencil || hiz);

      assert(brw->gen < 6 || depth_mt->tiling == I915_TILING_Y);
      assert(!hiz || depth_mt->tiling == I915_TILING_Y);

      depthbuffer_format = brw_depthbuffer_format(brw);
      depth_surface_type = BRW_SURFACE_2D;
      depth_offset = brw->depthstencil.depth_offset;
      width = depth_irb->Base.Base.Width;
      height = depth_irb->Base.Base.Height;
   } else if (separate_stencil) {
      /*
       * There exists a separate stencil buffer but no depth buffer.
       *
       * The stencil buffer inherits most of its fields from
       * 3DSTATE_DEPTH_BUFFER: namely the tile walk, surface type, width, and
       * height.
       *
       * The tiled bit must be set. From the Sandybridge PRM, Volume 2, Part 1,
       * Section 7.5.5.1.1 3DSTATE_DEPTH_BUFFER, Bit 1.27 Tiled Surface:
       *     [DevGT+]: This field must be set to TRUE.
       */
      assert(brw->has_separate_stencil);

      depth_surface_type = BRW_SURFACE_2D;
      width = stencil_irb->Base.Base.Width;
      height = stencil_irb->Base.Base.Height;
   }

   if (depth_mt)
      brw_render_cache_set_check_flush(brw, depth_mt->bo);
   if (stencil_mt)
      brw_render_cache_set_check_flush(brw, stencil_mt->bo);

   brw->vtbl.emit_depth_stencil_hiz(brw, depth_mt, depth_offset,
                                    depthbuffer_format, depth_surface_type,
                                    stencil_mt, hiz, separate_stencil,
                                    width, height, tile_x, tile_y);
}

void
brw_emit_depth_stencil_hiz(struct brw_context *brw,
                           struct intel_mipmap_tree *depth_mt,
                           uint32_t depth_offset, uint32_t depthbuffer_format,
                           uint32_t depth_surface_type,
                           struct intel_mipmap_tree *stencil_mt,
                           bool hiz, bool separate_stencil,
                           uint32_t width, uint32_t height,
                           uint32_t tile_x, uint32_t tile_y)
{
   /* Enable the hiz bit if we're doing separate stencil, because it and the
    * separate stencil bit must have the same value. From Section 2.11.5.6.1.1
    * 3DSTATE_DEPTH_BUFFER, Bit 1.21 "Separate Stencil Enable":
    *     [DevIL]: If this field is enabled, Hierarchical Depth Buffer
    *     Enable must also be enabled.
    *
    *     [DevGT]: This field must be set to the same value (enabled or
    *     disabled) as Hierarchical Depth Buffer Enable
    */
   bool enable_hiz_ss = hiz || separate_stencil;


   /* 3DSTATE_DEPTH_BUFFER, 3DSTATE_STENCIL_BUFFER are both
    * non-pipelined state that will need the PIPE_CONTROL workaround.
    */
   if (brw->gen == 6) {
      intel_emit_depth_stall_flushes(brw);
   }

   unsigned int len;
   if (brw->gen >= 6)
      len = 7;
   else if (brw->is_g4x || brw->gen == 5)
      len = 6;
   else
      len = 5;

   BEGIN_BATCH(len);
   OUT_BATCH(_3DSTATE_DEPTH_BUFFER << 16 | (len - 2));
   OUT_BATCH((depth_mt ? depth_mt->pitch - 1 : 0) |
             (depthbuffer_format << 18) |
             ((enable_hiz_ss ? 1 : 0) << 21) | /* separate stencil enable */
             ((enable_hiz_ss ? 1 : 0) << 22) | /* hiz enable */
             (BRW_TILEWALK_YMAJOR << 26) |
             ((depth_mt ? depth_mt->tiling != I915_TILING_NONE : 1)
              << 27) |
             (depth_surface_type << 29));

   if (depth_mt) {
      OUT_RELOC(depth_mt->bo,
		I915_GEM_DOMAIN_RENDER, I915_GEM_DOMAIN_RENDER,
		depth_offset);
   } else {
      OUT_BATCH(0);
   }

   OUT_BATCH(((width + tile_x - 1) << 6) |
             ((height + tile_y - 1) << 19));
   OUT_BATCH(0);

   if (brw->is_g4x || brw->gen >= 5)
      OUT_BATCH(tile_x | (tile_y << 16));
   else
      assert(tile_x == 0 && tile_y == 0);

   if (brw->gen >= 6)
      OUT_BATCH(0);

   ADVANCE_BATCH();

   if (hiz || separate_stencil) {
      /*
       * In the 3DSTATE_DEPTH_BUFFER batch emitted above, the 'separate
       * stencil enable' and 'hiz enable' bits were set. Therefore we must
       * emit 3DSTATE_HIER_DEPTH_BUFFER and 3DSTATE_STENCIL_BUFFER. Even if
       * there is no stencil buffer, 3DSTATE_STENCIL_BUFFER must be emitted;
       * failure to do so causes hangs on gen5 and a stall on gen6.
       */

      /* Emit hiz buffer. */
      if (hiz) {
         struct intel_mipmap_tree *hiz_mt = depth_mt->hiz_buf->mt;
	 BEGIN_BATCH(3);
	 OUT_BATCH((_3DSTATE_HIER_DEPTH_BUFFER << 16) | (3 - 2));
	 OUT_BATCH(hiz_mt->pitch - 1);
	 OUT_RELOC(hiz_mt->bo,
		   I915_GEM_DOMAIN_RENDER, I915_GEM_DOMAIN_RENDER,
		   brw->depthstencil.hiz_offset);
	 ADVANCE_BATCH();
      } else {
	 BEGIN_BATCH(3);
	 OUT_BATCH((_3DSTATE_HIER_DEPTH_BUFFER << 16) | (3 - 2));
	 OUT_BATCH(0);
	 OUT_BATCH(0);
	 ADVANCE_BATCH();
      }

      /* Emit stencil buffer. */
      if (separate_stencil) {
	 BEGIN_BATCH(3);
	 OUT_BATCH((_3DSTATE_STENCIL_BUFFER << 16) | (3 - 2));
         /* The stencil buffer has quirky pitch requirements.  From Vol 2a,
          * 11.5.6.2.1 3DSTATE_STENCIL_BUFFER, field "Surface Pitch":
          *    The pitch must be set to 2x the value computed based on width, as
          *    the stencil buffer is stored with two rows interleaved.
          */
	 OUT_BATCH(2 * stencil_mt->pitch - 1);
	 OUT_RELOC(stencil_mt->bo,
		   I915_GEM_DOMAIN_RENDER, I915_GEM_DOMAIN_RENDER,
		   brw->depthstencil.stencil_offset);
	 ADVANCE_BATCH();
      } else {
	 BEGIN_BATCH(3);
	 OUT_BATCH((_3DSTATE_STENCIL_BUFFER << 16) | (3 - 2));
	 OUT_BATCH(0);
	 OUT_BATCH(0);
	 ADVANCE_BATCH();
      }
   }

   /*
    * On Gen >= 6, emit clear params for safety. If using hiz, then clear
    * params must be emitted.
    *
    * From Section 2.11.5.6.4.1 3DSTATE_CLEAR_PARAMS:
    *     3DSTATE_CLEAR_PARAMS packet must follow the DEPTH_BUFFER_STATE packet
    *     when HiZ is enabled and the DEPTH_BUFFER_STATE changes.
    */
   if (brw->gen >= 6 || hiz) {
      BEGIN_BATCH(2);
      OUT_BATCH(_3DSTATE_CLEAR_PARAMS << 16 |
		GEN5_DEPTH_CLEAR_VALID |
		(2 - 2));
      OUT_BATCH(depth_mt ? depth_mt->depth_clear_value : 0);
      ADVANCE_BATCH();
   }
}

const struct brw_tracked_state brw_depthbuffer = {
   .dirty = {
      .mesa = _NEW_BUFFERS,
      .brw = BRW_NEW_BATCH,
   },
   .emit = brw_emit_depthbuffer,
};



/***********************************************************************
 * Polygon stipple packet
 */

static void upload_polygon_stipple(struct brw_context *brw)
{
   struct gl_context *ctx = &brw->ctx;
   GLuint i;

   /* _NEW_POLYGON */
   if (!ctx->Polygon.StippleFlag)
      return;

   BEGIN_BATCH(33);
   OUT_BATCH(_3DSTATE_POLY_STIPPLE_PATTERN << 16 | (33 - 2));

   /* Polygon stipple is provided in OpenGL order, i.e. bottom
    * row first.  If we're rendering to a window (i.e. the
    * default frame buffer object, 0), then we need to invert
    * it to match our pixel layout.  But if we're rendering
    * to a FBO (i.e. any named frame buffer object), we *don't*
    * need to invert - we already match the layout.
    */
   if (_mesa_is_winsys_fbo(ctx->DrawBuffer)) {
      for (i = 0; i < 32; i++)
	  OUT_BATCH(ctx->PolygonStipple[31 - i]); /* invert */
   }
   else {
      for (i = 0; i < 32; i++)
	 OUT_BATCH(ctx->PolygonStipple[i]);
   }
   ADVANCE_BATCH();
}

const struct brw_tracked_state brw_polygon_stipple = {
   .dirty = {
      .mesa = _NEW_POLYGON |
              _NEW_POLYGONSTIPPLE,
      .brw = BRW_NEW_CONTEXT,
   },
   .emit = upload_polygon_stipple
};


/***********************************************************************
 * Polygon stipple offset packet
 */

static void upload_polygon_stipple_offset(struct brw_context *brw)
{
   struct gl_context *ctx = &brw->ctx;

   /* _NEW_POLYGON */
   if (!ctx->Polygon.StippleFlag)
      return;

   BEGIN_BATCH(2);
   OUT_BATCH(_3DSTATE_POLY_STIPPLE_OFFSET << 16 | (2-2));

   /* _NEW_BUFFERS
    *
    * If we're drawing to a system window we have to invert the Y axis
    * in order to match the OpenGL pixel coordinate system, and our
    * offset must be matched to the window position.  If we're drawing
    * to a user-created FBO then our native pixel coordinate system
    * works just fine, and there's no window system to worry about.
    */
   if (_mesa_is_winsys_fbo(ctx->DrawBuffer))
      OUT_BATCH((32 - (ctx->DrawBuffer->Height & 31)) & 31);
   else
      OUT_BATCH(0);
   ADVANCE_BATCH();
}

const struct brw_tracked_state brw_polygon_stipple_offset = {
   .dirty = {
      .mesa = _NEW_BUFFERS |
              _NEW_POLYGON,
      .brw = BRW_NEW_CONTEXT,
   },
   .emit = upload_polygon_stipple_offset
};

/**********************************************************************
 * AA Line parameters
 */
static void upload_aa_line_parameters(struct brw_context *brw)
{
   struct gl_context *ctx = &brw->ctx;

   if (!ctx->Line.SmoothFlag)
      return;

   /* Original Gen4 doesn't have 3DSTATE_AA_LINE_PARAMETERS. */
   if (brw->gen == 4 && !brw->is_g4x)
      return;

   BEGIN_BATCH(3);
   OUT_BATCH(_3DSTATE_AA_LINE_PARAMETERS << 16 | (3 - 2));
   /* use legacy aa line coverage computation */
   OUT_BATCH(0);
   OUT_BATCH(0);
   ADVANCE_BATCH();
}

const struct brw_tracked_state brw_aa_line_parameters = {
   .dirty = {
      .mesa = _NEW_LINE,
      .brw = BRW_NEW_CONTEXT,
   },
   .emit = upload_aa_line_parameters
};

/***********************************************************************
 * Line stipple packet
 */

static void upload_line_stipple(struct brw_context *brw)
{
   struct gl_context *ctx = &brw->ctx;
   GLfloat tmp;
   GLint tmpi;

   if (!ctx->Line.StippleFlag)
      return;

   BEGIN_BATCH(3);
   OUT_BATCH(_3DSTATE_LINE_STIPPLE_PATTERN << 16 | (3 - 2));
   OUT_BATCH(ctx->Line.StipplePattern);

   if (brw->gen >= 7) {
      /* in U1.16 */
      tmp = 1.0 / (GLfloat) ctx->Line.StippleFactor;
      tmpi = tmp * (1<<16);
      OUT_BATCH(tmpi << 15 | ctx->Line.StippleFactor);
   }
   else {
      /* in U1.13 */
      tmp = 1.0 / (GLfloat) ctx->Line.StippleFactor;
      tmpi = tmp * (1<<13);
      OUT_BATCH(tmpi << 16 | ctx->Line.StippleFactor);
   }

   ADVANCE_BATCH();
}

const struct brw_tracked_state brw_line_stipple = {
   .dirty = {
      .mesa = _NEW_LINE,
      .brw = BRW_NEW_CONTEXT,
   },
   .emit = upload_line_stipple
};


/***********************************************************************
 * Misc invariant state packets
 */

void
brw_upload_invariant_state(struct brw_context *brw)
{
   const bool is_965 = brw->gen == 4 && !brw->is_g4x;

   /* Select the 3D pipeline (as opposed to media) */
   const uint32_t _3DSTATE_PIPELINE_SELECT =
      is_965 ? CMD_PIPELINE_SELECT_965 : CMD_PIPELINE_SELECT_GM45;
   BEGIN_BATCH(1);
   OUT_BATCH(_3DSTATE_PIPELINE_SELECT << 16 | (brw->gen >= 9 ? (3 << 8) : 0));
   ADVANCE_BATCH();

   if (brw->gen < 6) {
      /* Disable depth offset clamping. */
      BEGIN_BATCH(2);
      OUT_BATCH(_3DSTATE_GLOBAL_DEPTH_OFFSET_CLAMP << 16 | (2 - 2));
      OUT_BATCH_F(0.0);
      ADVANCE_BATCH();
   }

   if (brw->gen >= 8) {
      BEGIN_BATCH(3);
      OUT_BATCH(CMD_STATE_SIP << 16 | (3 - 2));
      OUT_BATCH(0);
      OUT_BATCH(0);
      ADVANCE_BATCH();
   } else {
      BEGIN_BATCH(2);
      OUT_BATCH(CMD_STATE_SIP << 16 | (2 - 2));
      OUT_BATCH(0);
      ADVANCE_BATCH();
   }

   const uint32_t _3DSTATE_VF_STATISTICS =
      is_965 ? GEN4_3DSTATE_VF_STATISTICS : GM45_3DSTATE_VF_STATISTICS;
   BEGIN_BATCH(1);
   OUT_BATCH(_3DSTATE_VF_STATISTICS << 16 | 1);
   ADVANCE_BATCH();
}

const struct brw_tracked_state brw_invariant_state = {
   .dirty = {
      .mesa = 0,
      .brw = BRW_NEW_CONTEXT,
   },
   .emit = brw_upload_invariant_state
};

/**
 * Define the base addresses which some state is referenced from.
 *
 * This allows us to avoid having to emit relocations for the objects,
 * and is actually required for binding table pointers on gen6.
 *
 * Surface state base address covers binding table pointers and
 * surface state objects, but not the surfaces that the surface state
 * objects point to.
 */
static void upload_state_base_address( struct brw_context *brw )
{
   /* FINISHME: According to section 3.6.1 "STATE_BASE_ADDRESS" of
    * vol1a of the G45 PRM, MI_FLUSH with the ISC invalidate should be
    * programmed prior to STATE_BASE_ADDRESS.
    *
    * However, given that the instruction SBA (general state base
    * address) on this chipset is always set to 0 across X and GL,
    * maybe this isn't required for us in particular.
    */

   if (brw->gen >= 6) {
      uint8_t mocs = brw->gen == 7 ? GEN7_MOCS_L3 : 0;

       BEGIN_BATCH(10);
       OUT_BATCH(CMD_STATE_BASE_ADDRESS << 16 | (10 - 2));
       OUT_BATCH(mocs << 8 | /* General State Memory Object Control State */
                 mocs << 4 | /* Stateless Data Port Access Memory Object Control State */
                 1); /* General State Base Address Modify Enable */
       /* Surface state base address:
	* BINDING_TABLE_STATE
	* SURFACE_STATE
	*/
       OUT_RELOC(brw->batch.bo, I915_GEM_DOMAIN_SAMPLER, 0, 1);
        /* Dynamic state base address:
	 * SAMPLER_STATE
	 * SAMPLER_BORDER_COLOR_STATE
	 * CLIP, SF, WM/CC viewport state
	 * COLOR_CALC_STATE
	 * DEPTH_STENCIL_STATE
	 * BLEND_STATE
	 * Push constants (when INSTPM: CONSTANT_BUFFER Address Offset
	 * Disable is clear, which we rely on)
	 */
       OUT_RELOC(brw->batch.bo, (I915_GEM_DOMAIN_RENDER |
				   I915_GEM_DOMAIN_INSTRUCTION), 0, 1);

       OUT_BATCH(1); /* Indirect object base address: MEDIA_OBJECT data */
       OUT_RELOC(brw->cache.bo, I915_GEM_DOMAIN_INSTRUCTION, 0,
		 1); /* Instruction base address: shader kernels (incl. SIP) */

       OUT_BATCH(1); /* General state upper bound */
       /* Dynamic state upper bound.  Although the documentation says that
	* programming it to zero will cause it to be ignored, that is a lie.
	* If this isn't programmed to a real bound, the sampler border color
	* pointer is rejected, causing border color to mysteriously fail.
	*/
       OUT_BATCH(0xfffff001);
       OUT_BATCH(1); /* Indirect object upper bound */
       OUT_BATCH(1); /* Instruction access upper bound */
       ADVANCE_BATCH();
   } else if (brw->gen == 5) {
       BEGIN_BATCH(8);
       OUT_BATCH(CMD_STATE_BASE_ADDRESS << 16 | (8 - 2));
       OUT_BATCH(1); /* General state base address */
       OUT_RELOC(brw->batch.bo, I915_GEM_DOMAIN_SAMPLER, 0,
		 1); /* Surface state base address */
       OUT_BATCH(1); /* Indirect object base address */
       OUT_RELOC(brw->cache.bo, I915_GEM_DOMAIN_INSTRUCTION, 0,
		 1); /* Instruction base address */
       OUT_BATCH(0xfffff001); /* General state upper bound */
       OUT_BATCH(1); /* Indirect object upper bound */
       OUT_BATCH(1); /* Instruction access upper bound */
       ADVANCE_BATCH();
   } else {
       BEGIN_BATCH(6);
       OUT_BATCH(CMD_STATE_BASE_ADDRESS << 16 | (6 - 2));
       OUT_BATCH(1); /* General state base address */
       OUT_RELOC(brw->batch.bo, I915_GEM_DOMAIN_SAMPLER, 0,
		 1); /* Surface state base address */
       OUT_BATCH(1); /* Indirect object base address */
       OUT_BATCH(1); /* General state upper bound */
       OUT_BATCH(1); /* Indirect object upper bound */
       ADVANCE_BATCH();
   }

   /* According to section 3.6.1 of VOL1 of the 965 PRM,
    * STATE_BASE_ADDRESS updates require a reissue of:
    *
    * 3DSTATE_PIPELINE_POINTERS
    * 3DSTATE_BINDING_TABLE_POINTERS
    * MEDIA_STATE_POINTERS
    *
    * and this continues through Ironlake.  The Sandy Bridge PRM, vol
    * 1 part 1 says that the folowing packets must be reissued:
    *
    * 3DSTATE_CC_POINTERS
    * 3DSTATE_BINDING_TABLE_POINTERS
    * 3DSTATE_SAMPLER_STATE_POINTERS
    * 3DSTATE_VIEWPORT_STATE_POINTERS
    * MEDIA_STATE_POINTERS
    *
    * Those are always reissued following SBA updates anyway (new
    * batch time), except in the case of the program cache BO
    * changing.  Having a separate state flag makes the sequence more
    * obvious.
    */

   brw->ctx.NewDriverState |= BRW_NEW_STATE_BASE_ADDRESS;
}

const struct brw_tracked_state brw_state_base_address = {
   .dirty = {
      .mesa = 0,
      .brw = BRW_NEW_BATCH |
             BRW_NEW_PROGRAM_CACHE,
   },
   .emit = upload_state_base_address
};