summaryrefslogtreecommitdiff
path: root/src/mesa/drivers/dri/i965/brw_draw_upload.c
blob: 2952027ebd736f2004cd606d0213ee9cae9dd42b (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
/**************************************************************************
 * 
 * Copyright 2003 Tungsten Graphics, Inc., Cedar Park, Texas.
 * 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, sub license, 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 NON-INFRINGEMENT.
 * IN NO EVENT SHALL TUNGSTEN GRAPHICS 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.
 * 
 **************************************************************************/

#include "main/glheader.h"
#include "main/bufferobj.h"
#include "main/context.h"
#include "main/enums.h"
#include "main/macros.h"
#include "main/glformats.h"

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

#include "intel_batchbuffer.h"
#include "intel_buffer_objects.h"

static GLuint double_types[5] = {
   0,
   BRW_SURFACEFORMAT_R64_FLOAT,
   BRW_SURFACEFORMAT_R64G64_FLOAT,
   BRW_SURFACEFORMAT_R64G64B64_FLOAT,
   BRW_SURFACEFORMAT_R64G64B64A64_FLOAT
};

static GLuint float_types[5] = {
   0,
   BRW_SURFACEFORMAT_R32_FLOAT,
   BRW_SURFACEFORMAT_R32G32_FLOAT,
   BRW_SURFACEFORMAT_R32G32B32_FLOAT,
   BRW_SURFACEFORMAT_R32G32B32A32_FLOAT
};

static GLuint half_float_types[5] = {
   0,
   BRW_SURFACEFORMAT_R16_FLOAT,
   BRW_SURFACEFORMAT_R16G16_FLOAT,
   BRW_SURFACEFORMAT_R16G16B16A16_FLOAT,
   BRW_SURFACEFORMAT_R16G16B16A16_FLOAT
};

static GLuint fixed_point_types[5] = {
   0,
   BRW_SURFACEFORMAT_R32_SFIXED,
   BRW_SURFACEFORMAT_R32G32_SFIXED,
   BRW_SURFACEFORMAT_R32G32B32_SFIXED,
   BRW_SURFACEFORMAT_R32G32B32A32_SFIXED,
};

static GLuint uint_types_direct[5] = {
   0,
   BRW_SURFACEFORMAT_R32_UINT,
   BRW_SURFACEFORMAT_R32G32_UINT,
   BRW_SURFACEFORMAT_R32G32B32_UINT,
   BRW_SURFACEFORMAT_R32G32B32A32_UINT
};

static GLuint uint_types_norm[5] = {
   0,
   BRW_SURFACEFORMAT_R32_UNORM,
   BRW_SURFACEFORMAT_R32G32_UNORM,
   BRW_SURFACEFORMAT_R32G32B32_UNORM,
   BRW_SURFACEFORMAT_R32G32B32A32_UNORM
};

static GLuint uint_types_scale[5] = {
   0,
   BRW_SURFACEFORMAT_R32_USCALED,
   BRW_SURFACEFORMAT_R32G32_USCALED,
   BRW_SURFACEFORMAT_R32G32B32_USCALED,
   BRW_SURFACEFORMAT_R32G32B32A32_USCALED
};

static GLuint int_types_direct[5] = {
   0,
   BRW_SURFACEFORMAT_R32_SINT,
   BRW_SURFACEFORMAT_R32G32_SINT,
   BRW_SURFACEFORMAT_R32G32B32_SINT,
   BRW_SURFACEFORMAT_R32G32B32A32_SINT
};

static GLuint int_types_norm[5] = {
   0,
   BRW_SURFACEFORMAT_R32_SNORM,
   BRW_SURFACEFORMAT_R32G32_SNORM,
   BRW_SURFACEFORMAT_R32G32B32_SNORM,
   BRW_SURFACEFORMAT_R32G32B32A32_SNORM
};

static GLuint int_types_scale[5] = {
   0,
   BRW_SURFACEFORMAT_R32_SSCALED,
   BRW_SURFACEFORMAT_R32G32_SSCALED,
   BRW_SURFACEFORMAT_R32G32B32_SSCALED,
   BRW_SURFACEFORMAT_R32G32B32A32_SSCALED
};

static GLuint ushort_types_direct[5] = {
   0,
   BRW_SURFACEFORMAT_R16_UINT,
   BRW_SURFACEFORMAT_R16G16_UINT,
   BRW_SURFACEFORMAT_R16G16B16A16_UINT,
   BRW_SURFACEFORMAT_R16G16B16A16_UINT
};

static GLuint ushort_types_norm[5] = {
   0,
   BRW_SURFACEFORMAT_R16_UNORM,
   BRW_SURFACEFORMAT_R16G16_UNORM,
   BRW_SURFACEFORMAT_R16G16B16_UNORM,
   BRW_SURFACEFORMAT_R16G16B16A16_UNORM
};

static GLuint ushort_types_scale[5] = {
   0,
   BRW_SURFACEFORMAT_R16_USCALED,
   BRW_SURFACEFORMAT_R16G16_USCALED,
   BRW_SURFACEFORMAT_R16G16B16_USCALED,
   BRW_SURFACEFORMAT_R16G16B16A16_USCALED
};

static GLuint short_types_direct[5] = {
   0,
   BRW_SURFACEFORMAT_R16_SINT,
   BRW_SURFACEFORMAT_R16G16_SINT,
   BRW_SURFACEFORMAT_R16G16B16A16_SINT,
   BRW_SURFACEFORMAT_R16G16B16A16_SINT
};

static GLuint short_types_norm[5] = {
   0,
   BRW_SURFACEFORMAT_R16_SNORM,
   BRW_SURFACEFORMAT_R16G16_SNORM,
   BRW_SURFACEFORMAT_R16G16B16_SNORM,
   BRW_SURFACEFORMAT_R16G16B16A16_SNORM
};

static GLuint short_types_scale[5] = {
   0,
   BRW_SURFACEFORMAT_R16_SSCALED,
   BRW_SURFACEFORMAT_R16G16_SSCALED,
   BRW_SURFACEFORMAT_R16G16B16_SSCALED,
   BRW_SURFACEFORMAT_R16G16B16A16_SSCALED
};

static GLuint ubyte_types_direct[5] = {
   0,
   BRW_SURFACEFORMAT_R8_UINT,
   BRW_SURFACEFORMAT_R8G8_UINT,
   BRW_SURFACEFORMAT_R8G8B8A8_UINT,
   BRW_SURFACEFORMAT_R8G8B8A8_UINT
};

static GLuint ubyte_types_norm[5] = {
   0,
   BRW_SURFACEFORMAT_R8_UNORM,
   BRW_SURFACEFORMAT_R8G8_UNORM,
   BRW_SURFACEFORMAT_R8G8B8_UNORM,
   BRW_SURFACEFORMAT_R8G8B8A8_UNORM
};

static GLuint ubyte_types_scale[5] = {
   0,
   BRW_SURFACEFORMAT_R8_USCALED,
   BRW_SURFACEFORMAT_R8G8_USCALED,
   BRW_SURFACEFORMAT_R8G8B8_USCALED,
   BRW_SURFACEFORMAT_R8G8B8A8_USCALED
};

static GLuint byte_types_direct[5] = {
   0,
   BRW_SURFACEFORMAT_R8_SINT,
   BRW_SURFACEFORMAT_R8G8_SINT,
   BRW_SURFACEFORMAT_R8G8B8A8_SINT,
   BRW_SURFACEFORMAT_R8G8B8A8_SINT
};

static GLuint byte_types_norm[5] = {
   0,
   BRW_SURFACEFORMAT_R8_SNORM,
   BRW_SURFACEFORMAT_R8G8_SNORM,
   BRW_SURFACEFORMAT_R8G8B8_SNORM,
   BRW_SURFACEFORMAT_R8G8B8A8_SNORM
};

static GLuint byte_types_scale[5] = {
   0,
   BRW_SURFACEFORMAT_R8_SSCALED,
   BRW_SURFACEFORMAT_R8G8_SSCALED,
   BRW_SURFACEFORMAT_R8G8B8_SSCALED,
   BRW_SURFACEFORMAT_R8G8B8A8_SSCALED
};


/**
 * Given vertex array type/size/format/normalized info, return
 * the appopriate hardware surface type.
 * Format will be GL_RGBA or possibly GL_BGRA for GLubyte[4] color arrays.
 */
static unsigned
get_surface_type(struct brw_context *brw,
                 const struct gl_client_array *glarray)
{
   int size = glarray->Size;

   if (unlikely(INTEL_DEBUG & DEBUG_VERTS))
      printf("type %s size %d normalized %d\n", 
             _mesa_lookup_enum_by_nr(glarray->Type),
             glarray->Size, glarray->Normalized);

   if (glarray->Integer) {
      assert(glarray->Format == GL_RGBA); /* sanity check */
      switch (glarray->Type) {
      case GL_INT: return int_types_direct[size];
      case GL_SHORT: return short_types_direct[size];
      case GL_BYTE: return byte_types_direct[size];
      case GL_UNSIGNED_INT: return uint_types_direct[size];
      case GL_UNSIGNED_SHORT: return ushort_types_direct[size];
      case GL_UNSIGNED_BYTE: return ubyte_types_direct[size];
      default: assert(0); return 0;
      }
   } else if (glarray->Normalized) {
      switch (glarray->Type) {
      case GL_DOUBLE: return double_types[size];
      case GL_FLOAT: return float_types[size];
      case GL_HALF_FLOAT: return half_float_types[size];
      case GL_INT: return int_types_norm[size];
      case GL_SHORT: return short_types_norm[size];
      case GL_BYTE: return byte_types_norm[size];
      case GL_UNSIGNED_INT: return uint_types_norm[size];
      case GL_UNSIGNED_SHORT: return ushort_types_norm[size];
      case GL_UNSIGNED_BYTE:
         if (glarray->Format == GL_BGRA) {
            /* See GL_EXT_vertex_array_bgra */
            assert(size == 4);
            return BRW_SURFACEFORMAT_B8G8R8A8_UNORM;
         }
         else {
            return ubyte_types_norm[size];
         }
      case GL_FIXED:
         if (brw->gen >= 8 || brw->is_haswell)
            return fixed_point_types[size];

         /* This produces GL_FIXED inputs as values between INT32_MIN and
          * INT32_MAX, which will be scaled down by 1/65536 by the VS.
          */
         return int_types_scale[size];
      /* See GL_ARB_vertex_type_2_10_10_10_rev.
       * W/A: Pre-Haswell, the hardware doesn't really support the formats we'd
       * like to use here, so upload everything as UINT and fix
       * it in the shader
       */
      case GL_INT_2_10_10_10_REV:
         assert(size == 4);
         if (brw->gen >= 8 || brw->is_haswell) {
            return glarray->Format == GL_BGRA
               ? BRW_SURFACEFORMAT_B10G10R10A2_SNORM
               : BRW_SURFACEFORMAT_R10G10B10A2_SNORM;
         }
         return BRW_SURFACEFORMAT_R10G10B10A2_UINT;
      case GL_UNSIGNED_INT_2_10_10_10_REV:
         assert(size == 4);
         if (brw->gen >= 8 || brw->is_haswell) {
            return glarray->Format == GL_BGRA
               ? BRW_SURFACEFORMAT_B10G10R10A2_UNORM
               : BRW_SURFACEFORMAT_R10G10B10A2_UNORM;
         }
         return BRW_SURFACEFORMAT_R10G10B10A2_UINT;
      default: assert(0); return 0;
      }
   }
   else {
      /* See GL_ARB_vertex_type_2_10_10_10_rev.
       * W/A: the hardware doesn't really support the formats we'd
       * like to use here, so upload everything as UINT and fix
       * it in the shader
       */
      if (glarray->Type == GL_INT_2_10_10_10_REV) {
         assert(size == 4);
         if (brw->gen >= 8 || brw->is_haswell) {
            return glarray->Format == GL_BGRA
               ? BRW_SURFACEFORMAT_B10G10R10A2_SSCALED
               : BRW_SURFACEFORMAT_R10G10B10A2_SSCALED;
         }
         return BRW_SURFACEFORMAT_R10G10B10A2_UINT;
      } else if (glarray->Type == GL_UNSIGNED_INT_2_10_10_10_REV) {
         assert(size == 4);
         if (brw->gen >= 8 || brw->is_haswell) {
            return glarray->Format == GL_BGRA
               ? BRW_SURFACEFORMAT_B10G10R10A2_USCALED
               : BRW_SURFACEFORMAT_R10G10B10A2_USCALED;
         }
         return BRW_SURFACEFORMAT_R10G10B10A2_UINT;
      }
      assert(glarray->Format == GL_RGBA); /* sanity check */
      switch (glarray->Type) {
      case GL_DOUBLE: return double_types[size];
      case GL_FLOAT: return float_types[size];
      case GL_HALF_FLOAT: return half_float_types[size];
      case GL_INT: return int_types_scale[size];
      case GL_SHORT: return short_types_scale[size];
      case GL_BYTE: return byte_types_scale[size];
      case GL_UNSIGNED_INT: return uint_types_scale[size];
      case GL_UNSIGNED_SHORT: return ushort_types_scale[size];
      case GL_UNSIGNED_BYTE: return ubyte_types_scale[size];
      case GL_FIXED:
         if (brw->gen >= 8 || brw->is_haswell)
            return fixed_point_types[size];

         /* This produces GL_FIXED inputs as values between INT32_MIN and
          * INT32_MAX, which will be scaled down by 1/65536 by the VS.
          */
         return int_types_scale[size];
      default: assert(0); return 0;
      }
   }
}

static GLuint get_index_type(GLenum type)
{
   switch (type) {
   case GL_UNSIGNED_BYTE:  return BRW_INDEX_BYTE;
   case GL_UNSIGNED_SHORT: return BRW_INDEX_WORD;
   case GL_UNSIGNED_INT:   return BRW_INDEX_DWORD;
   default: assert(0); return 0;
   }
}

static void
copy_array_to_vbo_array(struct brw_context *brw,
			struct brw_vertex_element *element,
			int min, int max,
			struct brw_vertex_buffer *buffer,
			GLuint dst_stride)
{
   const int src_stride = element->glarray->StrideB;

   /* If the source stride is zero, we just want to upload the current
    * attribute once and set the buffer's stride to 0.  There's no need
    * to replicate it out.
    */
   if (src_stride == 0) {
      intel_upload_data(brw, element->glarray->Ptr,
                        element->glarray->_ElementSize,
                        element->glarray->_ElementSize,
			&buffer->bo, &buffer->offset);

      buffer->stride = 0;
      return;
   }

   const unsigned char *src = element->glarray->Ptr + min * src_stride;
   int count = max - min + 1;
   GLuint size = count * dst_stride;

   if (dst_stride == src_stride) {
      intel_upload_data(brw, src, size, dst_stride,
			&buffer->bo, &buffer->offset);
   } else {
      char * const map = intel_upload_map(brw, size, dst_stride);
      char *dst = map;

      while (count--) {
	 memcpy(dst, src, dst_stride);
	 src += src_stride;
	 dst += dst_stride;
      }
      intel_upload_unmap(brw, map, size, dst_stride,
			 &buffer->bo, &buffer->offset);
   }
   buffer->stride = dst_stride;
}

static void brw_prepare_vertices(struct brw_context *brw)
{
   struct gl_context *ctx = &brw->ctx;
   /* CACHE_NEW_VS_PROG */
   GLbitfield64 vs_inputs = brw->vs.prog_data->inputs_read;
   const unsigned char *ptr = NULL;
   GLuint interleaved = 0;
   unsigned int min_index = brw->vb.min_index + brw->basevertex;
   unsigned int max_index = brw->vb.max_index + brw->basevertex;
   int delta, i, j;

   struct brw_vertex_element *upload[VERT_ATTRIB_MAX];
   GLuint nr_uploads = 0;

   /* _NEW_POLYGON
    *
    * On gen6+, edge flags don't end up in the VUE (either in or out of the
    * VS).  Instead, they're uploaded as the last vertex element, and the data
    * is passed sideband through the fixed function units.  So, we need to
    * prepare the vertex buffer for it, but it's not present in inputs_read.
    */
   if (brw->gen >= 6 && (ctx->Polygon.FrontMode != GL_FILL ||
                           ctx->Polygon.BackMode != GL_FILL)) {
      vs_inputs |= VERT_BIT_EDGEFLAG;
   }

   if (0)
      printf("%s %d..%d\n", __FUNCTION__, min_index, max_index);

   /* Accumulate the list of enabled arrays. */
   brw->vb.nr_enabled = 0;
   while (vs_inputs) {
      GLuint i = ffsll(vs_inputs) - 1;
      struct brw_vertex_element *input = &brw->vb.inputs[i];

      vs_inputs &= ~BITFIELD64_BIT(i);
      brw->vb.enabled[brw->vb.nr_enabled++] = input;
   }

   if (brw->vb.nr_enabled == 0)
      return;

   if (brw->vb.nr_buffers)
      return;

   for (i = j = 0; i < brw->vb.nr_enabled; i++) {
      struct brw_vertex_element *input = brw->vb.enabled[i];
      const struct gl_client_array *glarray = input->glarray;

      if (_mesa_is_bufferobj(glarray->BufferObj)) {
	 struct intel_buffer_object *intel_buffer =
	    intel_buffer_object(glarray->BufferObj);
	 int k;

	 /* If we have a VB set to be uploaded for this buffer object
	  * already, reuse that VB state so that we emit fewer
	  * relocations.
	  */
	 for (k = 0; k < i; k++) {
	    const struct gl_client_array *other = brw->vb.enabled[k]->glarray;
	    if (glarray->BufferObj == other->BufferObj &&
		glarray->StrideB == other->StrideB &&
		glarray->InstanceDivisor == other->InstanceDivisor &&
		(uintptr_t)(glarray->Ptr - other->Ptr) < glarray->StrideB)
	    {
	       input->buffer = brw->vb.enabled[k]->buffer;
	       input->offset = glarray->Ptr - other->Ptr;
	       break;
	    }
	 }
	 if (k == i) {
	    struct brw_vertex_buffer *buffer = &brw->vb.buffers[j];

	    /* Named buffer object: Just reference its contents directly. */
            buffer->bo = intel_bufferobj_source(brw,
                                                intel_buffer, 1,
						&buffer->offset);
	    drm_intel_bo_reference(buffer->bo);
	    buffer->offset += (uintptr_t)glarray->Ptr;
	    buffer->stride = glarray->StrideB;
	    buffer->step_rate = glarray->InstanceDivisor;

	    input->buffer = j++;
	    input->offset = 0;
	 }

	 /* This is a common place to reach if the user mistakenly supplies
	  * a pointer in place of a VBO offset.  If we just let it go through,
	  * we may end up dereferencing a pointer beyond the bounds of the
	  * GTT.  We would hope that the VBO's max_index would save us, but
	  * Mesa appears to hand us min/max values not clipped to the
	  * array object's _MaxElement, and _MaxElement frequently appears
	  * to be wrong anyway.
	  *
	  * The VBO spec allows application termination in this case, and it's
	  * probably a service to the poor programmer to do so rather than
	  * trying to just not render.
	  */
	 assert(input->offset < brw->vb.buffers[input->buffer].bo->size);
      } else {
	 /* Queue the buffer object up to be uploaded in the next pass,
	  * when we've decided if we're doing interleaved or not.
	  */
	 if (nr_uploads == 0) {
	    interleaved = glarray->StrideB;
	    ptr = glarray->Ptr;
	 }
	 else if (interleaved != glarray->StrideB ||
                  glarray->Ptr < ptr ||
                  (uintptr_t)(glarray->Ptr - ptr) + glarray->_ElementSize > interleaved)
	 {
            /* If our stride is different from the first attribute's stride,
             * or if the first attribute's stride didn't cover our element,
             * disable the interleaved upload optimization.  The second case
             * can most commonly occur in cases where there is a single vertex
             * and, for example, the data is stored on the application's
             * stack.
             *
             * NOTE: This will also disable the optimization in cases where
             * the data is in a different order than the array indices.
             * Something like:
             *
             *     float data[...];
             *     glVertexAttribPointer(0, 4, GL_FLOAT, 32, &data[4]);
             *     glVertexAttribPointer(1, 4, GL_FLOAT, 32, &data[0]);
             */
	    interleaved = 0;
	 }

	 upload[nr_uploads++] = input;
      }
   }

   /* If we need to upload all the arrays, then we can trim those arrays to
    * only the used elements [min_index, max_index] so long as we adjust all
    * the values used in the 3DPRIMITIVE i.e. by setting the vertex bias.
    */
   brw->vb.start_vertex_bias = 0;
   delta = min_index;
   if (nr_uploads == brw->vb.nr_enabled) {
      brw->vb.start_vertex_bias = -delta;
      delta = 0;
   }

   /* Handle any arrays to be uploaded. */
   if (nr_uploads > 1) {
      if (interleaved) {
	 struct brw_vertex_buffer *buffer = &brw->vb.buffers[j];
	 /* All uploads are interleaved, so upload the arrays together as
	  * interleaved.  First, upload the contents and set up upload[0].
	  */
	 copy_array_to_vbo_array(brw, upload[0], min_index, max_index,
				 buffer, interleaved);
	 buffer->offset -= delta * interleaved;

	 for (i = 0; i < nr_uploads; i++) {
	    /* Then, just point upload[i] at upload[0]'s buffer. */
	    upload[i]->offset =
	       ((const unsigned char *)upload[i]->glarray->Ptr - ptr);
	    upload[i]->buffer = j;
	 }
	 j++;

	 nr_uploads = 0;
      }
   }
   /* Upload non-interleaved arrays */
   for (i = 0; i < nr_uploads; i++) {
      struct brw_vertex_buffer *buffer = &brw->vb.buffers[j];
      if (upload[i]->glarray->InstanceDivisor == 0) {
         copy_array_to_vbo_array(brw, upload[i], min_index, max_index,
                                 buffer, upload[i]->glarray->_ElementSize);
      } else {
         /* This is an instanced attribute, since its InstanceDivisor
          * is not zero. Therefore, its data will be stepped after the
          * instanced draw has been run InstanceDivisor times.
          */
         uint32_t instanced_attr_max_index =
            (brw->num_instances - 1) / upload[i]->glarray->InstanceDivisor;
         copy_array_to_vbo_array(brw, upload[i], 0, instanced_attr_max_index,
                                 buffer, upload[i]->glarray->_ElementSize);
      }
      buffer->offset -= delta * buffer->stride;
      buffer->step_rate = upload[i]->glarray->InstanceDivisor;
      upload[i]->buffer = j++;
      upload[i]->offset = 0;
   }

   brw->vb.nr_buffers = j;
}

static void brw_emit_vertices(struct brw_context *brw)
{
   GLuint i, nr_elements;

   brw_prepare_vertices(brw);

   brw_emit_query_begin(brw);

   nr_elements = brw->vb.nr_enabled + brw->vs.prog_data->uses_vertexid;

   /* If the VS doesn't read any inputs (calculating vertex position from
    * a state variable for some reason, for example), emit a single pad
    * VERTEX_ELEMENT struct and bail.
    *
    * The stale VB state stays in place, but they don't do anything unless
    * a VE loads from them.
    */
   if (nr_elements == 0) {
      BEGIN_BATCH(3);
      OUT_BATCH((_3DSTATE_VERTEX_ELEMENTS << 16) | 1);
      if (brw->gen >= 6) {
	 OUT_BATCH((0 << GEN6_VE0_INDEX_SHIFT) |
		   GEN6_VE0_VALID |
		   (BRW_SURFACEFORMAT_R32G32B32A32_FLOAT << BRW_VE0_FORMAT_SHIFT) |
		   (0 << BRW_VE0_SRC_OFFSET_SHIFT));
      } else {
	 OUT_BATCH((0 << BRW_VE0_INDEX_SHIFT) |
		   BRW_VE0_VALID |
		   (BRW_SURFACEFORMAT_R32G32B32A32_FLOAT << BRW_VE0_FORMAT_SHIFT) |
		   (0 << BRW_VE0_SRC_OFFSET_SHIFT));
      }
      OUT_BATCH((BRW_VE1_COMPONENT_STORE_0 << BRW_VE1_COMPONENT_0_SHIFT) |
		(BRW_VE1_COMPONENT_STORE_0 << BRW_VE1_COMPONENT_1_SHIFT) |
		(BRW_VE1_COMPONENT_STORE_0 << BRW_VE1_COMPONENT_2_SHIFT) |
		(BRW_VE1_COMPONENT_STORE_1_FLT << BRW_VE1_COMPONENT_3_SHIFT));
      CACHED_BATCH();
      return;
   }

   /* Now emit VB and VEP state packets.
    */

   if (brw->vb.nr_buffers) {
      if (brw->gen >= 6) {
	 assert(brw->vb.nr_buffers <= 33);
      } else {
	 assert(brw->vb.nr_buffers <= 17);
      }

      BEGIN_BATCH(1 + 4*brw->vb.nr_buffers);
      OUT_BATCH((_3DSTATE_VERTEX_BUFFERS << 16) | (4*brw->vb.nr_buffers - 1));
      for (i = 0; i < brw->vb.nr_buffers; i++) {
	 struct brw_vertex_buffer *buffer = &brw->vb.buffers[i];
	 uint32_t dw0;

	 if (brw->gen >= 6) {
	    dw0 = buffer->step_rate
	             ? GEN6_VB0_ACCESS_INSTANCEDATA
	             : GEN6_VB0_ACCESS_VERTEXDATA;
	    dw0 |= i << GEN6_VB0_INDEX_SHIFT;
	 } else {
	    dw0 = buffer->step_rate
	             ? BRW_VB0_ACCESS_INSTANCEDATA
	             : BRW_VB0_ACCESS_VERTEXDATA;
	    dw0 |= i << BRW_VB0_INDEX_SHIFT;
	 }

	 if (brw->gen >= 7)
	    dw0 |= GEN7_VB0_ADDRESS_MODIFYENABLE;

	 OUT_BATCH(dw0 | (buffer->stride << BRW_VB0_PITCH_SHIFT));
	 OUT_RELOC(buffer->bo, I915_GEM_DOMAIN_VERTEX, 0, buffer->offset);
	 if (brw->gen >= 5) {
	    OUT_RELOC(buffer->bo, I915_GEM_DOMAIN_VERTEX, 0, buffer->bo->size - 1);
	 } else
	    OUT_BATCH(0);
	 OUT_BATCH(buffer->step_rate);
      }
      ADVANCE_BATCH();
   }

   /* The hardware allows one more VERTEX_ELEMENTS than VERTEX_BUFFERS, presumably
    * for VertexID/InstanceID.
    */
   if (brw->gen >= 6) {
      assert(nr_elements <= 34);
   } else {
      assert(nr_elements <= 18);
   }

   struct brw_vertex_element *gen6_edgeflag_input = NULL;

   BEGIN_BATCH(1 + nr_elements * 2);
   OUT_BATCH((_3DSTATE_VERTEX_ELEMENTS << 16) | (2 * nr_elements - 1));
   for (i = 0; i < brw->vb.nr_enabled; i++) {
      struct brw_vertex_element *input = brw->vb.enabled[i];
      uint32_t format = get_surface_type(brw, input->glarray);
      uint32_t comp0 = BRW_VE1_COMPONENT_STORE_SRC;
      uint32_t comp1 = BRW_VE1_COMPONENT_STORE_SRC;
      uint32_t comp2 = BRW_VE1_COMPONENT_STORE_SRC;
      uint32_t comp3 = BRW_VE1_COMPONENT_STORE_SRC;

      /* The gen4 driver expects edgeflag to come in as a float, and passes
       * that float on to the tests in the clipper.  Mesa's current vertex
       * attribute value for EdgeFlag is stored as a float, which works out.
       * glEdgeFlagPointer, on the other hand, gives us an unnormalized
       * integer ubyte.  Just rewrite that to convert to a float.
       */
      if (input->attrib == VERT_ATTRIB_EDGEFLAG) {
         /* Gen6+ passes edgeflag as sideband along with the vertex, instead
          * of in the VUE.  We have to upload it sideband as the last vertex
          * element according to the B-Spec.
          */
         if (brw->gen >= 6) {
            gen6_edgeflag_input = input;
            continue;
         }

         if (format == BRW_SURFACEFORMAT_R8_UINT)
            format = BRW_SURFACEFORMAT_R8_SSCALED;
      }

      switch (input->glarray->Size) {
      case 0: comp0 = BRW_VE1_COMPONENT_STORE_0;
      case 1: comp1 = BRW_VE1_COMPONENT_STORE_0;
      case 2: comp2 = BRW_VE1_COMPONENT_STORE_0;
      case 3: comp3 = input->glarray->Integer ? BRW_VE1_COMPONENT_STORE_1_INT
                                              : BRW_VE1_COMPONENT_STORE_1_FLT;
	 break;
      }

      if (brw->gen >= 6) {
	 OUT_BATCH((input->buffer << GEN6_VE0_INDEX_SHIFT) |
		   GEN6_VE0_VALID |
		   (format << BRW_VE0_FORMAT_SHIFT) |
		   (input->offset << BRW_VE0_SRC_OFFSET_SHIFT));
      } else {
	 OUT_BATCH((input->buffer << BRW_VE0_INDEX_SHIFT) |
		   BRW_VE0_VALID |
		   (format << BRW_VE0_FORMAT_SHIFT) |
		   (input->offset << BRW_VE0_SRC_OFFSET_SHIFT));
      }

      if (brw->gen >= 5)
          OUT_BATCH((comp0 << BRW_VE1_COMPONENT_0_SHIFT) |
                    (comp1 << BRW_VE1_COMPONENT_1_SHIFT) |
                    (comp2 << BRW_VE1_COMPONENT_2_SHIFT) |
                    (comp3 << BRW_VE1_COMPONENT_3_SHIFT));
      else
          OUT_BATCH((comp0 << BRW_VE1_COMPONENT_0_SHIFT) |
                    (comp1 << BRW_VE1_COMPONENT_1_SHIFT) |
                    (comp2 << BRW_VE1_COMPONENT_2_SHIFT) |
                    (comp3 << BRW_VE1_COMPONENT_3_SHIFT) |
                    ((i * 4) << BRW_VE1_DST_OFFSET_SHIFT));
   }

   if (brw->gen >= 6 && gen6_edgeflag_input) {
      uint32_t format = get_surface_type(brw, gen6_edgeflag_input->glarray);

      OUT_BATCH((gen6_edgeflag_input->buffer << GEN6_VE0_INDEX_SHIFT) |
                GEN6_VE0_VALID |
                GEN6_VE0_EDGE_FLAG_ENABLE |
                (format << BRW_VE0_FORMAT_SHIFT) |
                (gen6_edgeflag_input->offset << BRW_VE0_SRC_OFFSET_SHIFT));
      OUT_BATCH((BRW_VE1_COMPONENT_STORE_SRC << BRW_VE1_COMPONENT_0_SHIFT) |
                (BRW_VE1_COMPONENT_STORE_0 << BRW_VE1_COMPONENT_1_SHIFT) |
                (BRW_VE1_COMPONENT_STORE_0 << BRW_VE1_COMPONENT_2_SHIFT) |
                (BRW_VE1_COMPONENT_STORE_0 << BRW_VE1_COMPONENT_3_SHIFT));
   }

   if (brw->vs.prog_data->uses_vertexid) {
      uint32_t dw0 = 0, dw1 = 0;

      dw1 = ((BRW_VE1_COMPONENT_STORE_VID << BRW_VE1_COMPONENT_0_SHIFT) |
	     (BRW_VE1_COMPONENT_STORE_IID << BRW_VE1_COMPONENT_1_SHIFT) |
	     (BRW_VE1_COMPONENT_STORE_0 << BRW_VE1_COMPONENT_2_SHIFT) |
	     (BRW_VE1_COMPONENT_STORE_0 << BRW_VE1_COMPONENT_3_SHIFT));

      if (brw->gen >= 6) {
	 dw0 |= GEN6_VE0_VALID;
      } else {
	 dw0 |= BRW_VE0_VALID;
	 dw1 |= (i * 4) << BRW_VE1_DST_OFFSET_SHIFT;
      }

      /* Note that for gl_VertexID, gl_InstanceID, and gl_PrimitiveID values,
       * the format is ignored and the value is always int.
       */

      OUT_BATCH(dw0);
      OUT_BATCH(dw1);
   }

   CACHED_BATCH();
}

const struct brw_tracked_state brw_vertices = {
   .dirty = {
      .mesa = _NEW_POLYGON,
      .brw = BRW_NEW_BATCH | BRW_NEW_VERTICES,
      .cache = CACHE_NEW_VS_PROG,
   },
   .emit = brw_emit_vertices,
};

static void brw_upload_indices(struct brw_context *brw)
{
   struct gl_context *ctx = &brw->ctx;
   const struct _mesa_index_buffer *index_buffer = brw->ib.ib;
   GLuint ib_size;
   drm_intel_bo *bo = NULL;
   struct gl_buffer_object *bufferobj;
   GLuint offset;
   GLuint ib_type_size;

   if (index_buffer == NULL)
      return;

   ib_type_size = _mesa_sizeof_type(index_buffer->type);
   ib_size = ib_type_size * index_buffer->count;
   bufferobj = index_buffer->obj;

   /* Turn into a proper VBO:
    */
   if (!_mesa_is_bufferobj(bufferobj)) {

      /* Get new bufferobj, offset:
       */
      intel_upload_data(brw, index_buffer->ptr, ib_size, ib_type_size,
			&bo, &offset);
      brw->ib.start_vertex_offset = offset / ib_type_size;
   } else {
      offset = (GLuint) (unsigned long) index_buffer->ptr;

      /* If the index buffer isn't aligned to its element size, we have to
       * rebase it into a temporary.
       */
       if ((ib_type_size - 1) & offset) {
          perf_debug("copying index buffer to a temporary to work around "
                     "misaligned offset %d\n", offset);

          GLubyte *map = ctx->Driver.MapBufferRange(ctx,
                                                    offset,
                                                    ib_size,
                                                    GL_MAP_READ_BIT,
                                                    bufferobj);

          intel_upload_data(brw, map, ib_size, ib_type_size, &bo, &offset);
          brw->ib.start_vertex_offset = offset / ib_type_size;

          ctx->Driver.UnmapBuffer(ctx, bufferobj);
       } else {
	  /* Use CMD_3D_PRIM's start_vertex_offset to avoid re-uploading
	   * the index buffer state when we're just moving the start index
	   * of our drawing.
	   */
	  brw->ib.start_vertex_offset = offset / ib_type_size;

	  bo = intel_bufferobj_source(brw,
				      intel_buffer_object(bufferobj),
				      ib_type_size,
				      &offset);
	  drm_intel_bo_reference(bo);

	  brw->ib.start_vertex_offset += offset / ib_type_size;
       }
   }

   if (brw->ib.bo != bo) {
      drm_intel_bo_unreference(brw->ib.bo);
      brw->ib.bo = bo;

      brw->state.dirty.brw |= BRW_NEW_INDEX_BUFFER;
   } else {
      drm_intel_bo_unreference(bo);
   }

   if (index_buffer->type != brw->ib.type) {
      brw->ib.type = index_buffer->type;
      brw->state.dirty.brw |= BRW_NEW_INDEX_BUFFER;
   }
}

const struct brw_tracked_state brw_indices = {
   .dirty = {
      .mesa = 0,
      .brw = BRW_NEW_INDICES,
      .cache = 0,
   },
   .emit = brw_upload_indices,
};

static void brw_emit_index_buffer(struct brw_context *brw)
{
   const struct _mesa_index_buffer *index_buffer = brw->ib.ib;
   GLuint cut_index_setting;

   if (index_buffer == NULL)
      return;

   if (brw->prim_restart.enable_cut_index && !brw->is_haswell) {
      cut_index_setting = BRW_CUT_INDEX_ENABLE;
   } else {
      cut_index_setting = 0;
   }

   BEGIN_BATCH(3);
   OUT_BATCH(CMD_INDEX_BUFFER << 16 |
             cut_index_setting |
             get_index_type(index_buffer->type) << 8 |
             1);
   OUT_RELOC(brw->ib.bo,
             I915_GEM_DOMAIN_VERTEX, 0,
             0);
   OUT_RELOC(brw->ib.bo,
             I915_GEM_DOMAIN_VERTEX, 0,
	     brw->ib.bo->size - 1);
   ADVANCE_BATCH();
}

const struct brw_tracked_state brw_index_buffer = {
   .dirty = {
      .mesa = 0,
      .brw = BRW_NEW_BATCH | BRW_NEW_INDEX_BUFFER,
      .cache = 0,
   },
   .emit = brw_emit_index_buffer,
};