summaryrefslogtreecommitdiff
path: root/src/mesa/drivers/dri/i965/brw_draw_upload.c
blob: 0fefbd9d8142a58075c9fd5a51378f48cfa1a651 (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
/**************************************************************************
 * 
 * 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/state.h"
#include "main/api_validate.h"
#include "main/enums.h"

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

#include "intel_batchbuffer.h"
#include "intel_buffer_objects.h"
#include "intel_tex.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 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_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_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_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_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_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 GLuint get_surface_type( GLenum type, GLuint size,
                                GLenum format, GLboolean normalized )
{
   if (INTEL_DEBUG & DEBUG_VERTS)
      _mesa_printf("type %s size %d normalized %d\n", 
		   _mesa_lookup_enum_by_nr(type), size, normalized);

   if (normalized) {
      switch (type) {
      case GL_DOUBLE: return double_types[size];
      case GL_FLOAT: return 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 (format == GL_BGRA) {
            /* See GL_EXT_vertex_array_bgra */
            assert(size == 4);
            return BRW_SURFACEFORMAT_B8G8R8A8_UNORM;
         }
         else {
            return ubyte_types_norm[size];
         }
      default: assert(0); return 0;
      }      
   }
   else {
      assert(format == GL_RGBA); /* sanity check */
      switch (type) {
      case GL_DOUBLE: return double_types[size];
      case GL_FLOAT: return 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];
      default: assert(0); return 0;
      }      
   }
}


static GLuint get_size( GLenum type )
{
   switch (type) {
   case GL_DOUBLE: return sizeof(GLdouble);
   case GL_FLOAT: return sizeof(GLfloat);
   case GL_INT: return sizeof(GLint);
   case GL_SHORT: return sizeof(GLshort);
   case GL_BYTE: return sizeof(GLbyte);
   case GL_UNSIGNED_INT: return sizeof(GLuint);
   case GL_UNSIGNED_SHORT: return sizeof(GLushort);
   case GL_UNSIGNED_BYTE: return sizeof(GLubyte);
   default: 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 wrap_buffers( struct brw_context *brw,
			  GLuint size )
{
   if (size < BRW_UPLOAD_INIT_SIZE)
      size = BRW_UPLOAD_INIT_SIZE;

   brw->vb.upload.offset = 0;

   if (brw->vb.upload.bo != NULL)
      dri_bo_unreference(brw->vb.upload.bo);
   brw->vb.upload.bo = dri_bo_alloc(brw->intel.bufmgr, "temporary VBO",
				    size, 1);

   /* Set the internal VBO\ to no-backing-store.  We only use them as a
    * temporary within a brw_try_draw_prims while the lock is held.
    */
   /* DON'T DO THIS AS IF WE HAVE TO RE-ORG MEMORY WE NEED SOMEWHERE WITH
      FAKE TO PUSH THIS STUFF */
//   if (!brw->intel.ttm)
//      dri_bo_fake_disable_backing_store(brw->vb.upload.bo, NULL, NULL);
}

static void get_space( struct brw_context *brw,
		       GLuint size,
		       dri_bo **bo_return,
		       GLuint *offset_return )
{
   size = ALIGN(size, 64);

   if (brw->vb.upload.bo == NULL ||
       brw->vb.upload.offset + size > brw->vb.upload.bo->size) {
      wrap_buffers(brw, size);
   }

   assert(*bo_return == NULL);
   dri_bo_reference(brw->vb.upload.bo);
   *bo_return = brw->vb.upload.bo;
   *offset_return = brw->vb.upload.offset;
   brw->vb.upload.offset += size;
}

static void
copy_array_to_vbo_array( struct brw_context *brw,
			 struct brw_vertex_element *element,
			 GLuint dst_stride)
{
   struct intel_context *intel = &brw->intel;
   GLuint size = element->count * dst_stride;

   get_space(brw, size, &element->bo, &element->offset);

   if (element->glarray->StrideB == 0) {
      assert(element->count == 1);
      element->stride = 0;
   } else {
      element->stride = dst_stride;
   }

   if (dst_stride == element->glarray->StrideB) {
      if (intel->intelScreen->kernel_exec_fencing) {
	 drm_intel_gem_bo_map_gtt(element->bo);
	 memcpy((char *)element->bo->virtual + element->offset,
		element->glarray->Ptr, size);
	 drm_intel_gem_bo_unmap_gtt(element->bo);
      } else {
	 dri_bo_subdata(element->bo,
			element->offset,
			size,
			element->glarray->Ptr);
      }
   } else {
      char *dest;
      const unsigned char *src = element->glarray->Ptr;
      int i;

      if (intel->intelScreen->kernel_exec_fencing) {
	 drm_intel_gem_bo_map_gtt(element->bo);
	 dest = element->bo->virtual;
	 dest += element->offset;

	 for (i = 0; i < element->count; i++) {
	    memcpy(dest, src, dst_stride);
	    src += element->glarray->StrideB;
	    dest += dst_stride;
	 }

	 drm_intel_gem_bo_unmap_gtt(element->bo);
      } else {
	 void *data;

	 data = _mesa_malloc(dst_stride * element->count);
	 dest = data;
	 for (i = 0; i < element->count; i++) {
	    memcpy(dest, src, dst_stride);
	    src += element->glarray->StrideB;
	    dest += dst_stride;
	 }

	 dri_bo_subdata(element->bo,
			element->offset,
			size,
			data);

	 _mesa_free(data);
      }
   }
}

static void brw_prepare_vertices(struct brw_context *brw)
{
   GLcontext *ctx = &brw->intel.ctx;
   struct intel_context *intel = intel_context(ctx);
   GLbitfield vs_inputs = brw->vs.prog_data->inputs_read; 
   GLuint i;
   const unsigned char *ptr = NULL;
   GLuint interleave = 0;
   unsigned int min_index = brw->vb.min_index;
   unsigned int max_index = brw->vb.max_index;

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

   /* First build an array of pointers to ve's in vb.inputs_read
    */
   if (0)
      _mesa_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 = _mesa_ffsll(vs_inputs) - 1;
      struct brw_vertex_element *input = &brw->vb.inputs[i];

      vs_inputs &= ~(1 << i);
      brw->vb.enabled[brw->vb.nr_enabled++] = input;
   }

   /* XXX: In the rare cases where this happens we fallback all
    * the way to software rasterization, although a tnl fallback
    * would be sufficient.  I don't know of *any* real world
    * cases with > 17 vertex attributes enabled, so it probably
    * isn't an issue at this point.
    */
   if (brw->vb.nr_enabled >= BRW_VEP_MAX) {
      intel->Fallback = GL_TRUE; /* boolean, not bitfield */
      return;
   }

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

      input->element_size = get_size(input->glarray->Type) * input->glarray->Size;

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

	 /* Named buffer object: Just reference its contents directly. */
	 dri_bo_unreference(input->bo);
	 input->bo = intel_bufferobj_buffer(intel, intel_buffer,
					    INTEL_READ);
	 dri_bo_reference(input->bo);
	 input->offset = (unsigned long)input->glarray->Ptr;
	 input->stride = input->glarray->StrideB;
	 input->count = input->glarray->_MaxElement;

	 /* 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 < input->bo->size);
      } else {
	 input->count = input->glarray->StrideB ? max_index + 1 - min_index : 1;
	 if (input->bo != NULL) {
	    /* Already-uploaded vertex data is present from a previous
	     * prepare_vertices, but we had to re-validate state due to
	     * check_aperture failing and a new batch being produced.
	     */
	    continue;
	 }

	 /* Queue the buffer object up to be uploaded in the next pass,
	  * when we've decided if we're doing interleaved or not.
	  */
	 if (input->attrib == VERT_ATTRIB_POS) {
	    /* Position array not properly enabled:
	     */
            if (input->glarray->StrideB == 0) {
               intel->Fallback = GL_TRUE; /* boolean, not bitfield */
               return;
            }

	    interleave = input->glarray->StrideB;
	    ptr = input->glarray->Ptr;
	 }
	 else if (interleave != input->glarray->StrideB ||
		  (const unsigned char *)input->glarray->Ptr - ptr < 0 ||
		  (const unsigned char *)input->glarray->Ptr - ptr > interleave)
	 {
	    interleave = 0;
	 }

	 upload[nr_uploads++] = input;
	 
	 /* We rebase drawing to start at element zero only when
	  * varyings are not in vbos, which means we can end up
	  * uploading non-varying arrays (stride != 0) when min_index
	  * is zero.  This doesn't matter as the amount to upload is
	  * the same for these arrays whether the draw call is rebased
	  * or not - we just have to upload the one element.
	  */
	 assert(min_index == 0 || input->glarray->StrideB == 0);
      }
   }

   /* Handle any arrays to be uploaded. */
   if (nr_uploads > 1 && interleave && interleave <= 256) {
      /* 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], interleave);

      for (i = 1; i < nr_uploads; i++) {
	 /* Then, just point upload[i] at upload[0]'s buffer. */
	 upload[i]->stride = interleave;
	 upload[i]->offset = upload[0]->offset +
	    ((const unsigned char *)upload[i]->glarray->Ptr - ptr);
	 upload[i]->bo = upload[0]->bo;
	 dri_bo_reference(upload[i]->bo);
      }
   }
   else {
      /* Upload non-interleaved arrays */
      for (i = 0; i < nr_uploads; i++) {
          copy_array_to_vbo_array(brw, upload[i], upload[i]->element_size);
      }
   }

   brw_prepare_query_begin(brw);

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

      brw_add_validated_bo(brw, input->bo);
   }
}

static void brw_emit_vertices(struct brw_context *brw)
{
   GLcontext *ctx = &brw->intel.ctx;
   struct intel_context *intel = intel_context(ctx);
   GLuint i;

   brw_emit_query_begin(brw);

   /* 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 (brw->vb.nr_enabled == 0) {
      BEGIN_BATCH(3, IGNORE_CLIPRECTS);
      OUT_BATCH((CMD_VERTEX_ELEMENT << 16) | 1);
      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));
      ADVANCE_BATCH();
      return;
   }

   /* Now emit VB and VEP state packets.
    *
    * This still defines a hardware VB for each input, even if they
    * are interleaved or from the same VBO.  TBD if this makes a
    * performance difference.
    */
   BEGIN_BATCH(1 + brw->vb.nr_enabled * 4, IGNORE_CLIPRECTS);
   OUT_BATCH((CMD_VERTEX_BUFFER << 16) |
	     ((1 + brw->vb.nr_enabled * 4) - 2));

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

      OUT_BATCH((i << BRW_VB0_INDEX_SHIFT) |
		BRW_VB0_ACCESS_VERTEXDATA |
		(input->stride << BRW_VB0_PITCH_SHIFT));
      OUT_RELOC(input->bo,
		I915_GEM_DOMAIN_VERTEX, 0,
		input->offset);
      if (BRW_IS_IGDNG(brw)) {
          if (input->stride) {
              OUT_RELOC(input->bo,
                        I915_GEM_DOMAIN_VERTEX, 0,
                        input->offset + input->stride * input->count - 1);
          } else {
              assert(input->count == 1);
              OUT_RELOC(input->bo,
                        I915_GEM_DOMAIN_VERTEX, 0,
                        input->offset + input->element_size - 1);
          }
      } else
          OUT_BATCH(input->stride ? input->count : 0);
      OUT_BATCH(0); /* Instance data step rate */
   }
   ADVANCE_BATCH();

   BEGIN_BATCH(1 + brw->vb.nr_enabled * 2, IGNORE_CLIPRECTS);
   OUT_BATCH((CMD_VERTEX_ELEMENT << 16) | ((1 + brw->vb.nr_enabled * 2) - 2));
   for (i = 0; i < brw->vb.nr_enabled; i++) {
      struct brw_vertex_element *input = brw->vb.enabled[i];
      uint32_t format = get_surface_type(input->glarray->Type,
					 input->glarray->Size,
					 input->glarray->Format,
					 input->glarray->Normalized);
      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;

      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 = BRW_VE1_COMPONENT_STORE_1_FLT;
	 break;
      }

      OUT_BATCH((i << BRW_VE0_INDEX_SHIFT) |
		BRW_VE0_VALID |
		(format << BRW_VE0_FORMAT_SHIFT) |
		(0 << BRW_VE0_SRC_OFFSET_SHIFT));

      if (BRW_IS_IGDNG(brw))
          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));
   }
   ADVANCE_BATCH();
}

const struct brw_tracked_state brw_vertices = {
   .dirty = {
      .mesa = 0,
      .brw = BRW_NEW_BATCH | BRW_NEW_VERTICES,
      .cache = 0,
   },
   .prepare = brw_prepare_vertices,
   .emit = brw_emit_vertices,
};

static void brw_prepare_indices(struct brw_context *brw)
{
   GLcontext *ctx = &brw->intel.ctx;
   struct intel_context *intel = &brw->intel;
   const struct _mesa_index_buffer *index_buffer = brw->ib.ib;
   GLuint ib_size;
   dri_bo *bo = NULL;
   struct gl_buffer_object *bufferobj;
   GLuint offset;
   GLuint ib_type_size;

   if (index_buffer == NULL)
      return;

   ib_type_size = get_size(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)) {
      brw->ib.start_vertex_offset = 0;

      /* Get new bufferobj, offset:
       */
      get_space(brw, ib_size, &bo, &offset);

      /* Straight upload
       */
      if (intel->intelScreen->kernel_exec_fencing) {
	 drm_intel_gem_bo_map_gtt(bo);
	 memcpy((char *)bo->virtual + offset, index_buffer->ptr, ib_size);
	 drm_intel_gem_bo_unmap_gtt(bo);
      } else {
	 dri_bo_subdata(bo, offset, ib_size, index_buffer->ptr);
      }
   } else {
      offset = (GLuint) (unsigned long) index_buffer->ptr;
      brw->ib.start_vertex_offset = 0;

      /* If the index buffer isn't aligned to its element size, we have to
       * rebase it into a temporary.
       */
       if ((get_size(index_buffer->type) - 1) & offset) {
           GLubyte *map = ctx->Driver.MapBuffer(ctx,
                                                GL_ELEMENT_ARRAY_BUFFER_ARB,
                                                GL_DYNAMIC_DRAW_ARB,
                                                bufferobj);
           map += offset;

	   get_space(brw, ib_size, &bo, &offset);

	   dri_bo_subdata(bo, offset, ib_size, map);

           ctx->Driver.UnmapBuffer(ctx, GL_ELEMENT_ARRAY_BUFFER_ARB, bufferobj);
       } else {
	  bo = intel_bufferobj_buffer(intel, intel_buffer_object(bufferobj),
				      INTEL_READ);
	  dri_bo_reference(bo);

	  /* 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;
	  offset = 0;
	  ib_size = bo->size;
       }
   }

   if (brw->ib.bo != bo ||
       brw->ib.offset != offset ||
       brw->ib.size != ib_size)
   {
      drm_intel_bo_unreference(brw->ib.bo);
      brw->ib.bo = bo;
      brw->ib.offset = offset;
      brw->ib.size = ib_size;

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

   brw_add_validated_bo(brw, brw->ib.bo);
}

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

static void brw_emit_index_buffer(struct brw_context *brw)
{
   struct intel_context *intel = &brw->intel;
   const struct _mesa_index_buffer *index_buffer = brw->ib.ib;

   if (index_buffer == NULL)
      return;

   /* Emit the indexbuffer packet:
    */
   {
      struct brw_indexbuffer ib;

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

      ib.header.bits.opcode = CMD_INDEX_BUFFER;
      ib.header.bits.length = sizeof(ib)/4 - 2;
      ib.header.bits.index_format = get_index_type(index_buffer->type);
      ib.header.bits.cut_index_enable = 0;

      BEGIN_BATCH(4, IGNORE_CLIPRECTS);
      OUT_BATCH( ib.header.dword );
      OUT_RELOC(brw->ib.bo,
		I915_GEM_DOMAIN_VERTEX, 0,
		brw->ib.offset);
      OUT_RELOC(brw->ib.bo,
		I915_GEM_DOMAIN_VERTEX, 0,
		brw->ib.offset + brw->ib.size - 1);
      OUT_BATCH( 0 );
      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,
};