aboutsummaryrefslogtreecommitdiffstats
path: root/src/gallium/drivers/r300/r300_render.c
blob: 175b83a1ace8eec0ae8668c1fa4b1d64f03c24b5 (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
1028
1029
1030
1031
1032
1033
1034
1035
1036
1037
1038
1039
1040
1041
1042
1043
1044
1045
1046
1047
1048
1049
1050
1051
1052
1053
1054
1055
1056
1057
1058
1059
1060
1061
1062
1063
1064
1065
1066
1067
1068
1069
1070
1071
1072
1073
1074
1075
1076
1077
1078
1079
1080
1081
1082
1083
1084
1085
1086
1087
1088
1089
1090
1091
1092
1093
1094
1095
1096
1097
1098
1099
1100
1101
1102
1103
1104
1105
1106
1107
1108
1109
1110
1111
1112
1113
1114
1115
1116
1117
1118
1119
1120
1121
1122
1123
1124
1125
1126
1127
1128
1129
1130
1131
1132
1133
1134
1135
1136
1137
1138
1139
1140
1141
1142
1143
1144
1145
1146
1147
1148
1149
1150
1151
1152
1153
1154
1155
1156
1157
1158
1159
1160
1161
1162
1163
1164
1165
1166
1167
1168
1169
1170
1171
1172
1173
1174
1175
1176
1177
1178
1179
1180
1181
1182
1183
1184
1185
1186
1187
1188
1189
1190
1191
1192
1193
1194
1195
1196
1197
1198
1199
1200
1201
1202
1203
1204
1205
1206
1207
1208
/*
 * Copyright 2009 Corbin Simpson <MostAwesomeDude@gmail.com>
 * Copyright 2010 Marek Olšák <maraeo@gmail.com>
 *
 * 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
 * on 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
 * THE AUTHOR(S) AND/OR THEIR 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. */

/* r300_render: Vertex and index buffer primitive emission. Contains both
 * HW TCL fastpath rendering, and SW TCL Draw-assisted rendering. */

#include "draw/draw_context.h"
#include "draw/draw_vbuf.h"

#include "util/u_inlines.h"

#include "util/u_format.h"
#include "util/u_memory.h"
#include "util/u_upload_mgr.h"
#include "util/u_prim.h"

#include "r300_cs.h"
#include "r300_context.h"
#include "r300_screen_buffer.h"
#include "r300_emit.h"
#include "r300_reg.h"

#include <limits.h>

#define IMMD_DWORDS 32

static uint32_t r300_translate_primitive(unsigned prim)
{
    static const int prim_conv[] = {
        R300_VAP_VF_CNTL__PRIM_POINTS,
        R300_VAP_VF_CNTL__PRIM_LINES,
        R300_VAP_VF_CNTL__PRIM_LINE_LOOP,
        R300_VAP_VF_CNTL__PRIM_LINE_STRIP,
        R300_VAP_VF_CNTL__PRIM_TRIANGLES,
        R300_VAP_VF_CNTL__PRIM_TRIANGLE_STRIP,
        R300_VAP_VF_CNTL__PRIM_TRIANGLE_FAN,
        R300_VAP_VF_CNTL__PRIM_QUADS,
        R300_VAP_VF_CNTL__PRIM_QUAD_STRIP,
        R300_VAP_VF_CNTL__PRIM_POLYGON,
        -1,
        -1,
        -1,
        -1
    };
    unsigned hwprim = prim_conv[prim];

    assert(hwprim != -1);
    return hwprim;
}

static uint32_t r300_provoking_vertex_fixes(struct r300_context *r300,
                                            unsigned mode)
{
    struct r300_rs_state* rs = (struct r300_rs_state*)r300->rs_state.state;
    uint32_t color_control = rs->color_control;

    /* By default (see r300_state.c:r300_create_rs_state) color_control is
     * initialized to provoking the first vertex.
     *
     * Triangle fans must be reduced to the second vertex, not the first, in
     * Gallium flatshade-first mode, as per the GL spec.
     * (http://www.opengl.org/registry/specs/ARB/provoking_vertex.txt)
     *
     * Quads never provoke correctly in flatshade-first mode. The first
     * vertex is never considered as provoking, so only the second, third,
     * and fourth vertices can be selected, and both "third" and "last" modes
     * select the fourth vertex. This is probably due to D3D lacking quads.
     *
     * Similarly, polygons reduce to the first, not the last, vertex, when in
     * "last" mode, and all other modes start from the second vertex.
     *
     * ~ C.
     */

    if (rs->rs.flatshade_first) {
        switch (mode) {
            case PIPE_PRIM_TRIANGLE_FAN:
                color_control |= R300_GA_COLOR_CONTROL_PROVOKING_VERTEX_SECOND;
                break;
            case PIPE_PRIM_QUADS:
            case PIPE_PRIM_QUAD_STRIP:
            case PIPE_PRIM_POLYGON:
                color_control |= R300_GA_COLOR_CONTROL_PROVOKING_VERTEX_LAST;
                break;
            default:
                color_control |= R300_GA_COLOR_CONTROL_PROVOKING_VERTEX_FIRST;
                break;
        }
    } else {
        color_control |= R300_GA_COLOR_CONTROL_PROVOKING_VERTEX_LAST;
    }

    return color_control;
}

void r500_emit_index_bias(struct r300_context *r300, int index_bias)
{
    CS_LOCALS(r300);

    BEGIN_CS(2);
    OUT_CS_REG(R500_VAP_INDEX_OFFSET,
               (index_bias & 0xFFFFFF) | (index_bias < 0 ? 1<<24 : 0));
    END_CS;
}

static void r300_emit_draw_init(struct r300_context *r300, unsigned mode,
                                unsigned max_index)
{
    CS_LOCALS(r300);

    assert(max_index < (1 << 24));

    BEGIN_CS(5);
    OUT_CS_REG(R300_GA_COLOR_CONTROL,
            r300_provoking_vertex_fixes(r300, mode));
    OUT_CS_REG_SEQ(R300_VAP_VF_MAX_VTX_INDX, 2);
    OUT_CS(max_index);
    OUT_CS(0);
    END_CS;
}

/* This function splits the index bias value into two parts:
 * - buffer_offset: the value that can be safely added to buffer offsets
 *   in r300_emit_vertex_arrays (it must yield a positive offset when added to
 *   a vertex buffer offset)
 * - index_offset: the value that must be manually subtracted from indices
 *   in an index buffer to achieve negative offsets. */
static void r300_split_index_bias(struct r300_context *r300, int index_bias,
                                  int *buffer_offset, int *index_offset)
{
    struct pipe_vertex_buffer *vb, *vbufs = r300->vertex_buffer;
    struct pipe_vertex_element *velem = r300->velems->velem;
    unsigned i, size;
    int max_neg_bias;

    if (index_bias < 0) {
        /* See how large index bias we may subtract. We must be careful
         * here because negative buffer offsets are not allowed
         * by the DRM API. */
        max_neg_bias = INT_MAX;
        for (i = 0; i < r300->velems->count; i++) {
            vb = &vbufs[velem[i].vertex_buffer_index];
            size = (vb->buffer_offset + velem[i].src_offset) / vb->stride;
            max_neg_bias = MIN2(max_neg_bias, size);
        }

        /* Now set the minimum allowed value. */
        *buffer_offset = MAX2(-max_neg_bias, index_bias);
    } else {
        /* A positive index bias is OK. */
        *buffer_offset = index_bias;
    }

    *index_offset = index_bias - *buffer_offset;
}

enum r300_prepare_flags {
    PREP_EMIT_STATES    = (1 << 0), /* call emit_dirty_state and friends? */
    PREP_VALIDATE_VBOS  = (1 << 1), /* validate VBOs? */
    PREP_EMIT_VARRAYS       = (1 << 2), /* call emit_vertex_arrays? */
    PREP_EMIT_VARRAYS_SWTCL = (1 << 3), /* call emit_vertex_arrays_swtcl? */
    PREP_INDEXED        = (1 << 4)  /* is this draw_elements? */
};

/**
 * Check if the requested number of dwords is available in the CS and
 * if not, flush.
 * \param r300          The context.
 * \param flags         See r300_prepare_flags.
 * \param cs_dwords     The number of dwords to reserve in CS.
 * \return TRUE if the CS was flushed
 */
static boolean r300_reserve_cs_dwords(struct r300_context *r300,
                                      enum r300_prepare_flags flags,
                                      unsigned cs_dwords)
{
    boolean flushed        = FALSE;
    boolean emit_states    = flags & PREP_EMIT_STATES;
    boolean emit_vertex_arrays       = flags & PREP_EMIT_VARRAYS;
    boolean emit_vertex_arrays_swtcl = flags & PREP_EMIT_VARRAYS_SWTCL;

    /* Add dirty state, index offset, and AOS. */
    if (emit_states)
        cs_dwords += r300_get_num_dirty_dwords(r300);

    if (r300->screen->caps.is_r500)
        cs_dwords += 2; /* emit_index_offset */

    if (emit_vertex_arrays)
        cs_dwords += 55; /* emit_vertex_arrays */

    if (emit_vertex_arrays_swtcl)
        cs_dwords += 7; /* emit_vertex_arrays_swtcl */

    cs_dwords += r300_get_num_cs_end_dwords(r300);

    /* Reserve requested CS space. */
    if (cs_dwords > (RADEON_MAX_CMDBUF_DWORDS - r300->cs->cdw)) {
        r300_flush(&r300->context, RADEON_FLUSH_ASYNC, NULL);
        flushed = TRUE;
    }

    return flushed;
}

/**
 * Validate buffers and emit dirty state.
 * \param r300          The context.
 * \param flags         See r300_prepare_flags.
 * \param index_buffer  The index buffer to validate. The parameter may be NULL.
 * \param buffer_offset The offset passed to emit_vertex_arrays.
 * \param index_bias    The index bias to emit.
 * \param instance_id   Index of instance to render
 * \return TRUE if rendering should be skipped
 */
static boolean r300_emit_states(struct r300_context *r300,
                                enum r300_prepare_flags flags,
                                struct pipe_resource *index_buffer,
                                int buffer_offset,
                                int index_bias, int instance_id)
{
    boolean emit_states    = flags & PREP_EMIT_STATES;
    boolean emit_vertex_arrays       = flags & PREP_EMIT_VARRAYS;
    boolean emit_vertex_arrays_swtcl = flags & PREP_EMIT_VARRAYS_SWTCL;
    boolean indexed        = flags & PREP_INDEXED;
    boolean validate_vbos  = flags & PREP_VALIDATE_VBOS;

    /* Validate buffers and emit dirty state if needed. */
    if (emit_states || (emit_vertex_arrays && validate_vbos)) {
        if (!r300_emit_buffer_validate(r300, validate_vbos,
                                       index_buffer)) {
           fprintf(stderr, "r300: CS space validation failed. "
                   "(not enough memory?) Skipping rendering.\n");
           return FALSE;
        }
    }

    if (emit_states)
        r300_emit_dirty_state(r300);

    if (r300->screen->caps.is_r500) {
        if (r300->screen->caps.has_tcl)
            r500_emit_index_bias(r300, index_bias);
        else
            r500_emit_index_bias(r300, 0);
    }

    if (emit_vertex_arrays &&
        (r300->vertex_arrays_dirty ||
         r300->vertex_arrays_indexed != indexed ||
         r300->vertex_arrays_offset != buffer_offset ||
         r300->vertex_arrays_instance_id != instance_id)) {
        r300_emit_vertex_arrays(r300, buffer_offset, indexed, instance_id);

        r300->vertex_arrays_dirty = FALSE;
        r300->vertex_arrays_indexed = indexed;
        r300->vertex_arrays_offset = buffer_offset;
        r300->vertex_arrays_instance_id = instance_id;
    }

    if (emit_vertex_arrays_swtcl)
        r300_emit_vertex_arrays_swtcl(r300, indexed);

    return TRUE;
}

/**
 * Check if the requested number of dwords is available in the CS and
 * if not, flush. Then validate buffers and emit dirty state.
 * \param r300          The context.
 * \param flags         See r300_prepare_flags.
 * \param index_buffer  The index buffer to validate. The parameter may be NULL.
 * \param cs_dwords     The number of dwords to reserve in CS.
 * \param buffer_offset The offset passed to emit_vertex_arrays.
 * \param index_bias    The index bias to emit.
 * \param instance_id The instance to render.
 * \return TRUE if rendering should be skipped
 */
static boolean r300_prepare_for_rendering(struct r300_context *r300,
                                          enum r300_prepare_flags flags,
                                          struct pipe_resource *index_buffer,
                                          unsigned cs_dwords,
                                          int buffer_offset,
                                          int index_bias,
                                          int instance_id)
{
    /* Make sure there is enough space in the command stream and emit states. */
    if (r300_reserve_cs_dwords(r300, flags, cs_dwords))
        flags |= PREP_EMIT_STATES;

    return r300_emit_states(r300, flags, index_buffer, buffer_offset,
                            index_bias, instance_id);
}

static boolean immd_is_good_idea(struct r300_context *r300,
                                 unsigned count)
{
    if (DBG_ON(r300, DBG_NO_IMMD)) {
        return FALSE;
    }

    if (count * r300->velems->vertex_size_dwords > IMMD_DWORDS) {
        return FALSE;
    }

    /* Buffers can only be used for read by r300 (except query buffers, but
     * those can't be bound by a state tracker as vertex buffers). */
    return TRUE;
}

/*****************************************************************************
 * The HWTCL draw functions.                                                 *
 ****************************************************************************/

static void r300_draw_arrays_immediate(struct r300_context *r300,
                                       const struct pipe_draw_info *info)
{
    struct pipe_vertex_element* velem;
    struct pipe_vertex_buffer* vbuf;
    unsigned vertex_element_count = r300->velems->count;
    unsigned i, v, vbi;

    /* Size of the vertex, in dwords. */
    unsigned vertex_size = r300->velems->vertex_size_dwords;

    /* The number of dwords for this draw operation. */
    unsigned dwords = 4 + info->count * vertex_size;

    /* Size of the vertex element, in dwords. */
    unsigned size[PIPE_MAX_ATTRIBS];

    /* Stride to the same attrib in the next vertex in the vertex buffer,
     * in dwords. */
    unsigned stride[PIPE_MAX_ATTRIBS];

    /* Mapped vertex buffers. */
    uint32_t* map[PIPE_MAX_ATTRIBS] = {0};
    uint32_t* mapelem[PIPE_MAX_ATTRIBS];

    CS_LOCALS(r300);

    if (!r300_prepare_for_rendering(r300, PREP_EMIT_STATES, NULL, dwords, 0, 0, -1))
        return;

    /* Calculate the vertex size, offsets, strides etc. and map the buffers. */
    for (i = 0; i < vertex_element_count; i++) {
        velem = &r300->velems->velem[i];
        size[i] = r300->velems->format_size[i] / 4;
        vbi = velem->vertex_buffer_index;
        vbuf = &r300->vertex_buffer[vbi];
        stride[i] = vbuf->stride / 4;

        /* Map the buffer. */
        if (!map[vbi]) {
            map[vbi] = (uint32_t*)r300->rws->buffer_map(
                r300_resource(vbuf->buffer)->cs_buf,
                r300->cs, PIPE_TRANSFER_READ | PIPE_TRANSFER_UNSYNCHRONIZED);
            map[vbi] += (vbuf->buffer_offset / 4) + stride[i] * info->start;
        }
        mapelem[i] = map[vbi] + (velem->src_offset / 4);
    }

    r300_emit_draw_init(r300, info->mode, info->count-1);

    BEGIN_CS(dwords);
    OUT_CS_REG(R300_VAP_VTX_SIZE, vertex_size);
    OUT_CS_PKT3(R300_PACKET3_3D_DRAW_IMMD_2, info->count * vertex_size);
    OUT_CS(R300_VAP_VF_CNTL__PRIM_WALK_VERTEX_EMBEDDED | (info->count << 16) |
            r300_translate_primitive(info->mode));

    /* Emit vertices. */
    for (v = 0; v < info->count; v++) {
        for (i = 0; i < vertex_element_count; i++) {
            OUT_CS_TABLE(&mapelem[i][stride[i] * v], size[i]);
        }
    }
    END_CS;
}

static void r300_emit_draw_arrays(struct r300_context *r300,
                                  unsigned mode,
                                  unsigned count)
{
    boolean alt_num_verts = count > 65535;
    CS_LOCALS(r300);

    if (count >= (1 << 24)) {
        fprintf(stderr, "r300: Got a huge number of vertices: %i, "
                "refusing to render.\n", count);
        return;
    }

    r300_emit_draw_init(r300, mode, count-1);

    BEGIN_CS(2 + (alt_num_verts ? 2 : 0));
    if (alt_num_verts) {
        OUT_CS_REG(R500_VAP_ALT_NUM_VERTICES, count);
    }
    OUT_CS_PKT3(R300_PACKET3_3D_DRAW_VBUF_2, 0);
    OUT_CS(R300_VAP_VF_CNTL__PRIM_WALK_VERTEX_LIST | (count << 16) |
           r300_translate_primitive(mode) |
           (alt_num_verts ? R500_VAP_VF_CNTL__USE_ALT_NUM_VERTS : 0));
    END_CS;
}

static void r300_emit_draw_elements(struct r300_context *r300,
                                    struct pipe_resource* indexBuffer,
                                    unsigned indexSize,
                                    unsigned max_index,
                                    unsigned mode,
                                    unsigned start,
                                    unsigned count,
                                    uint16_t *imm_indices3)
{
    uint32_t count_dwords, offset_dwords;
    boolean alt_num_verts = count > 65535;
    CS_LOCALS(r300);

    if (count >= (1 << 24)) {
        fprintf(stderr, "r300: Got a huge number of vertices: %i, "
                "refusing to render (max_index: %i).\n", count, max_index);
        return;
    }

    DBG(r300, DBG_DRAW, "r300: Indexbuf of %u indices, max %u\n",
        count, max_index);

    r300_emit_draw_init(r300, mode, max_index);

    /* If start is odd, render the first triangle with indices embedded
     * in the command stream. This will increase start by 3 and make it
     * even. We can then proceed without a fallback. */
    if (indexSize == 2 && (start & 1) &&
        mode == PIPE_PRIM_TRIANGLES) {
        BEGIN_CS(4);
        OUT_CS_PKT3(R300_PACKET3_3D_DRAW_INDX_2, 2);
        OUT_CS(R300_VAP_VF_CNTL__PRIM_WALK_INDICES | (3 << 16) |
               R300_VAP_VF_CNTL__PRIM_TRIANGLES);
        OUT_CS(imm_indices3[1] << 16 | imm_indices3[0]);
        OUT_CS(imm_indices3[2]);
        END_CS;

        start += 3;
        count -= 3;
        if (!count)
           return;
    }

    offset_dwords = indexSize * start / sizeof(uint32_t);

    BEGIN_CS(8 + (alt_num_verts ? 2 : 0));
    if (alt_num_verts) {
        OUT_CS_REG(R500_VAP_ALT_NUM_VERTICES, count);
    }
    OUT_CS_PKT3(R300_PACKET3_3D_DRAW_INDX_2, 0);
    if (indexSize == 4) {
        count_dwords = count;
        OUT_CS(R300_VAP_VF_CNTL__PRIM_WALK_INDICES | (count << 16) |
               R300_VAP_VF_CNTL__INDEX_SIZE_32bit |
               r300_translate_primitive(mode) |
               (alt_num_verts ? R500_VAP_VF_CNTL__USE_ALT_NUM_VERTS : 0));
    } else {
        count_dwords = (count + 1) / 2;
        OUT_CS(R300_VAP_VF_CNTL__PRIM_WALK_INDICES | (count << 16) |
               r300_translate_primitive(mode) |
               (alt_num_verts ? R500_VAP_VF_CNTL__USE_ALT_NUM_VERTS : 0));
    }

    OUT_CS_PKT3(R300_PACKET3_INDX_BUFFER, 2);
    OUT_CS(R300_INDX_BUFFER_ONE_REG_WR | (R300_VAP_PORT_IDX0 >> 2) |
           (0 << R300_INDX_BUFFER_SKIP_SHIFT));
    OUT_CS(offset_dwords << 2);
    OUT_CS(count_dwords);
    OUT_CS_RELOC(r300_resource(indexBuffer));
    END_CS;
}

static void r300_draw_elements_immediate(struct r300_context *r300,
                                         const struct pipe_draw_info *info)
{
    const uint8_t *ptr1;
    const uint16_t *ptr2;
    const uint32_t *ptr4;
    unsigned index_size = r300->index_buffer.index_size;
    unsigned i, count_dwords = index_size == 4 ? info->count :
                                                 (info->count + 1) / 2;
    CS_LOCALS(r300);

    /* 19 dwords for r300_draw_elements_immediate. Give up if the function fails. */
    if (!r300_prepare_for_rendering(r300,
            PREP_EMIT_STATES | PREP_VALIDATE_VBOS | PREP_EMIT_VARRAYS |
            PREP_INDEXED, NULL, 2+count_dwords, 0, info->index_bias, -1))
        return;

    r300_emit_draw_init(r300, info->mode, info->max_index);

    BEGIN_CS(2 + count_dwords);
    OUT_CS_PKT3(R300_PACKET3_3D_DRAW_INDX_2, count_dwords);

    switch (index_size) {
    case 1:
        ptr1 = (uint8_t*)r300->index_buffer.user_buffer;
        ptr1 += info->start;

        OUT_CS(R300_VAP_VF_CNTL__PRIM_WALK_INDICES | (info->count << 16) |
               r300_translate_primitive(info->mode));

        if (info->index_bias && !r300->screen->caps.is_r500) {
            for (i = 0; i < info->count-1; i += 2)
                OUT_CS(((ptr1[i+1] + info->index_bias) << 16) |
                        (ptr1[i]   + info->index_bias));

            if (info->count & 1)
                OUT_CS(ptr1[i] + info->index_bias);
        } else {
            for (i = 0; i < info->count-1; i += 2)
                OUT_CS(((ptr1[i+1]) << 16) |
                        (ptr1[i]  ));

            if (info->count & 1)
                OUT_CS(ptr1[i]);
        }
        break;

    case 2:
        ptr2 = (uint16_t*)r300->index_buffer.user_buffer;
        ptr2 += info->start;

        OUT_CS(R300_VAP_VF_CNTL__PRIM_WALK_INDICES | (info->count << 16) |
               r300_translate_primitive(info->mode));

        if (info->index_bias && !r300->screen->caps.is_r500) {
            for (i = 0; i < info->count-1; i += 2)
                OUT_CS(((ptr2[i+1] + info->index_bias) << 16) |
                        (ptr2[i]   + info->index_bias));

            if (info->count & 1)
                OUT_CS(ptr2[i] + info->index_bias);
        } else {
            OUT_CS_TABLE(ptr2, count_dwords);
        }
        break;

    case 4:
        ptr4 = (uint32_t*)r300->index_buffer.user_buffer;
        ptr4 += info->start;

        OUT_CS(R300_VAP_VF_CNTL__PRIM_WALK_INDICES | (info->count << 16) |
               R300_VAP_VF_CNTL__INDEX_SIZE_32bit |
               r300_translate_primitive(info->mode));

        if (info->index_bias && !r300->screen->caps.is_r500) {
            for (i = 0; i < info->count; i++)
                OUT_CS(ptr4[i] + info->index_bias);
        } else {
            OUT_CS_TABLE(ptr4, count_dwords);
        }
        break;
    }
    END_CS;
}

static void r300_draw_elements(struct r300_context *r300,
                               const struct pipe_draw_info *info,
                               int instance_id)
{
    struct pipe_resource *indexBuffer = r300->index_buffer.buffer;
    unsigned indexSize = r300->index_buffer.index_size;
    struct pipe_resource* orgIndexBuffer = indexBuffer;
    unsigned start = info->start;
    unsigned count = info->count;
    boolean alt_num_verts = r300->screen->caps.is_r500 &&
                            count > 65536;
    unsigned short_count;
    int buffer_offset = 0, index_offset = 0; /* for index bias emulation */
    uint16_t indices3[3];

    if (info->index_bias && !r300->screen->caps.is_r500) {
        r300_split_index_bias(r300, info->index_bias, &buffer_offset,
                              &index_offset);
    }

    r300_translate_index_buffer(r300, &r300->index_buffer, &indexBuffer,
                                &indexSize, index_offset, &start, count);

    /* Fallback for misaligned ushort indices. */
    if (indexSize == 2 && (start & 1) && indexBuffer) {
        /* If we got here, then orgIndexBuffer == indexBuffer. */
        uint16_t *ptr = r300->rws->buffer_map(r300_resource(orgIndexBuffer)->cs_buf,
                                              r300->cs,
                                              PIPE_TRANSFER_READ |
                                              PIPE_TRANSFER_UNSYNCHRONIZED);

        if (info->mode == PIPE_PRIM_TRIANGLES) {
           memcpy(indices3, ptr + start, 6);
        } else {
            /* Copy the mapped index buffer directly to the upload buffer.
             * The start index will be aligned simply from the fact that
             * every sub-buffer in the upload buffer is aligned. */
            r300_upload_index_buffer(r300, &indexBuffer, indexSize, &start,
                                     count, (uint8_t*)ptr);
        }
    } else {
        if (r300->index_buffer.user_buffer)
            r300_upload_index_buffer(r300, &indexBuffer, indexSize,
                                     &start, count,
                                     r300->index_buffer.user_buffer);
    }

    /* 19 dwords for emit_draw_elements. Give up if the function fails. */
    if (!r300_prepare_for_rendering(r300,
            PREP_EMIT_STATES | PREP_VALIDATE_VBOS | PREP_EMIT_VARRAYS |
            PREP_INDEXED, indexBuffer, 19, buffer_offset, info->index_bias,
            instance_id))
        goto done;

    if (alt_num_verts || count <= 65535) {
        r300_emit_draw_elements(r300, indexBuffer, indexSize,
                                info->max_index, info->mode, start, count,
                                indices3);
    } else {
        do {
            /* The maximum must be divisible by 4 and 3,
             * so that quad and triangle lists are split correctly.
             *
             * Strips, loops, and fans won't work. */
            short_count = MIN2(count, 65532);

            r300_emit_draw_elements(r300, indexBuffer, indexSize,
                                     info->max_index,
                                     info->mode, start, short_count, indices3);

            start += short_count;
            count -= short_count;

            /* 15 dwords for emit_draw_elements */
            if (count) {
                if (!r300_prepare_for_rendering(r300,
                        PREP_VALIDATE_VBOS | PREP_EMIT_VARRAYS | PREP_INDEXED,
                        indexBuffer, 19, buffer_offset, info->index_bias,
                        instance_id))
                    goto done;
            }
        } while (count);
    }

done:
    if (indexBuffer != orgIndexBuffer) {
        pipe_resource_reference( &indexBuffer, NULL );
    }
}

static void r300_draw_arrays(struct r300_context *r300,
                             const struct pipe_draw_info *info,
                             int instance_id)
{
    boolean alt_num_verts = r300->screen->caps.is_r500 &&
                            info->count > 65536;
    unsigned start = info->start;
    unsigned count = info->count;
    unsigned short_count;

    /* 9 spare dwords for emit_draw_arrays. Give up if the function fails. */
    if (!r300_prepare_for_rendering(r300,
                                    PREP_EMIT_STATES | PREP_VALIDATE_VBOS | PREP_EMIT_VARRAYS,
                                    NULL, 9, start, 0, instance_id))
        return;

    if (alt_num_verts || count <= 65535) {
        r300_emit_draw_arrays(r300, info->mode, count);
    } else {
        do {
            /* The maximum must be divisible by 4 and 3,
             * so that quad and triangle lists are split correctly.
             *
             * Strips, loops, and fans won't work. */
            short_count = MIN2(count, 65532);
            r300_emit_draw_arrays(r300, info->mode, short_count);

            start += short_count;
            count -= short_count;

            /* 9 spare dwords for emit_draw_arrays. Give up if the function fails. */
            if (count) {
                if (!r300_prepare_for_rendering(r300,
                                                PREP_VALIDATE_VBOS | PREP_EMIT_VARRAYS, NULL, 9,
                                                start, 0, instance_id))
                    return;
            }
        } while (count);
    }
}

static void r300_draw_arrays_instanced(struct r300_context *r300,
                                       const struct pipe_draw_info *info)
{
    int i;

    for (i = 0; i < info->instance_count; i++)
        r300_draw_arrays(r300, info, i);
}

static void r300_draw_elements_instanced(struct r300_context *r300,
                                         const struct pipe_draw_info *info)
{
    int i;

    for (i = 0; i < info->instance_count; i++)
        r300_draw_elements(r300, info, i);
}

static unsigned r300_max_vertex_count(struct r300_context *r300)
{
   unsigned i, nr = r300->velems->count;
   struct pipe_vertex_element *velems = r300->velems->velem;
   unsigned result = ~0;

   for (i = 0; i < nr; i++) {
      struct pipe_vertex_buffer *vb =
            &r300->vertex_buffer[velems[i].vertex_buffer_index];
      unsigned size, max_count, value;

      /* We're not interested in constant and per-instance attribs. */
      if (!vb->buffer ||
          !vb->stride ||
          velems[i].instance_divisor) {
         continue;
      }

      size = vb->buffer->width0;

      /* Subtract buffer_offset. */
      value = vb->buffer_offset;
      if (value >= size) {
         return 0;
      }
      size -= value;

      /* Subtract src_offset. */
      value = velems[i].src_offset;
      if (value >= size) {
         return 0;
      }
      size -= value;

      /* Subtract format_size. */
      value = r300->velems->format_size[i];
      if (value >= size) {
         return 0;
      }
      size -= value;

      /* Compute the max count. */
      max_count = 1 + size / vb->stride;
      result = MIN2(result, max_count);
   }
   return result;
}


static void r300_draw_vbo(struct pipe_context* pipe,
                          const struct pipe_draw_info *dinfo)
{
    struct r300_context* r300 = r300_context(pipe);
    struct pipe_draw_info info = *dinfo;

    info.indexed = info.indexed;

    if (r300->skip_rendering ||
        !u_trim_pipe_prim(info.mode, &info.count)) {
        return;
    }

    r300_update_derived_state(r300);

    /* Draw. */
    if (info.indexed) {
        unsigned max_count = r300_max_vertex_count(r300);

        if (!max_count) {
           fprintf(stderr, "r300: Skipping a draw command. There is a buffer "
                   " which is too small to be used for rendering.\n");
           return;
        }

        if (max_count == ~0) {
           /* There are no per-vertex vertex elements. Use the hardware maximum. */
           max_count = 0xffffff;
        }

        info.max_index = max_count - 1;
        info.start += r300->index_buffer.offset / r300->index_buffer.index_size;

        if (info.instance_count <= 1) {
            if (info.count <= 8 &&
                r300->index_buffer.user_buffer) {
                r300_draw_elements_immediate(r300, &info);
            } else {
                r300_draw_elements(r300, &info, -1);
            }
        } else {
            r300_draw_elements_instanced(r300, &info);
        }
    } else {
        if (info.instance_count <= 1) {
            if (immd_is_good_idea(r300, info.count)) {
                r300_draw_arrays_immediate(r300, &info);
            } else {
                r300_draw_arrays(r300, &info, -1);
            }
        } else {
            r300_draw_arrays_instanced(r300, &info);
        }
    }
}

/****************************************************************************
 * The rest of this file is for SW TCL rendering only. Please be polite and *
 * keep these functions separated so that they are easier to locate. ~C.    *
 ***************************************************************************/

/* SW TCL elements, using Draw. */
static void r300_swtcl_draw_vbo(struct pipe_context* pipe,
                                const struct pipe_draw_info *info)
{
    struct r300_context* r300 = r300_context(pipe);

    if (r300->skip_rendering) {
        return;
    }

    r300_update_derived_state(r300);

    draw_vbo(r300->draw, info);
    draw_flush(r300->draw);
}

/* Object for rendering using Draw. */
struct r300_render {
    /* Parent class */
    struct vbuf_render base;

    /* Pipe context */
    struct r300_context* r300;

    /* Vertex information */
    size_t vertex_size;
    unsigned prim;
    unsigned hwprim;

    /* VBO */
    size_t vbo_max_used;
    uint8_t *vbo_ptr;
};

static INLINE struct r300_render*
r300_render(struct vbuf_render* render)
{
    return (struct r300_render*)render;
}

static const struct vertex_info*
r300_render_get_vertex_info(struct vbuf_render* render)
{
    struct r300_render* r300render = r300_render(render);
    struct r300_context* r300 = r300render->r300;

    return &r300->vertex_info;
}

static boolean r300_render_allocate_vertices(struct vbuf_render* render,
                                             ushort vertex_size,
                                             ushort count)
{
    struct r300_render* r300render = r300_render(render);
    struct r300_context* r300 = r300render->r300;
    struct radeon_winsys *rws = r300->rws;
    size_t size = (size_t)vertex_size * (size_t)count;

    DBG(r300, DBG_DRAW, "r300: render_allocate_vertices (size: %d)\n", size);

    if (!r300->vbo || size + r300->draw_vbo_offset > r300->vbo->size) {
	pb_reference(&r300->vbo, NULL);
        r300->vbo_cs = NULL;
        r300render->vbo_ptr = NULL;

        r300->vbo = rws->buffer_create(rws,
                                       MAX2(R300_MAX_DRAW_VBO_SIZE, size),
                                       R300_BUFFER_ALIGNMENT, TRUE,
                                       RADEON_DOMAIN_GTT);
        if (!r300->vbo) {
            return FALSE;
        }
        r300->vbo_cs = rws->buffer_get_cs_handle(r300->vbo);
        r300->draw_vbo_offset = 0;
        r300render->vbo_ptr = rws->buffer_map(r300->vbo_cs, r300->cs,
                                              PIPE_TRANSFER_WRITE);
    }

    r300render->vertex_size = vertex_size;
    return TRUE;
}

static void* r300_render_map_vertices(struct vbuf_render* render)
{
    struct r300_render* r300render = r300_render(render);
    struct r300_context* r300 = r300render->r300;

    DBG(r300, DBG_DRAW, "r300: render_map_vertices\n");

    assert(r300render->vbo_ptr);
    return r300render->vbo_ptr + r300->draw_vbo_offset;
}

static void r300_render_unmap_vertices(struct vbuf_render* render,
                                             ushort min,
                                             ushort max)
{
    struct r300_render* r300render = r300_render(render);
    struct r300_context* r300 = r300render->r300;

    DBG(r300, DBG_DRAW, "r300: render_unmap_vertices\n");

    r300render->vbo_max_used = MAX2(r300render->vbo_max_used,
                                    r300render->vertex_size * (max + 1));
}

static void r300_render_release_vertices(struct vbuf_render* render)
{
    struct r300_render* r300render = r300_render(render);
    struct r300_context* r300 = r300render->r300;

    DBG(r300, DBG_DRAW, "r300: render_release_vertices\n");

    r300->draw_vbo_offset += r300render->vbo_max_used;
    r300render->vbo_max_used = 0;
}

static void r300_render_set_primitive(struct vbuf_render* render,
                                      unsigned prim)
{
    struct r300_render* r300render = r300_render(render);

    r300render->prim = prim;
    r300render->hwprim = r300_translate_primitive(prim);
}

static void r300_render_draw_arrays(struct vbuf_render* render,
                                    unsigned start,
                                    unsigned count)
{
    struct r300_render* r300render = r300_render(render);
    struct r300_context* r300 = r300render->r300;
    uint8_t* ptr;
    unsigned i;
    unsigned dwords = 6;

    CS_LOCALS(r300);
    (void) i; (void) ptr;

    assert(start == 0);
    assert(count < (1 << 16));

    DBG(r300, DBG_DRAW, "r300: render_draw_arrays (count: %d)\n", count);

    if (!r300_prepare_for_rendering(r300,
                                    PREP_EMIT_STATES | PREP_EMIT_VARRAYS_SWTCL,
                                    NULL, dwords, 0, 0, -1)) {
        return;
    }

    BEGIN_CS(dwords);
    OUT_CS_REG(R300_GA_COLOR_CONTROL,
            r300_provoking_vertex_fixes(r300, r300render->prim));
    OUT_CS_REG(R300_VAP_VF_MAX_VTX_INDX, count - 1);
    OUT_CS_PKT3(R300_PACKET3_3D_DRAW_VBUF_2, 0);
    OUT_CS(R300_VAP_VF_CNTL__PRIM_WALK_VERTEX_LIST | (count << 16) |
           r300render->hwprim);
    END_CS;
}

static void r300_render_draw_elements(struct vbuf_render* render,
                                      const ushort* indices,
                                      uint count)
{
    struct r300_render* r300render = r300_render(render);
    struct r300_context* r300 = r300render->r300;
    unsigned max_index = (r300->vbo->size - r300->draw_vbo_offset) /
                         (r300render->r300->vertex_info.size * 4) - 1;
    struct pipe_resource *index_buffer = NULL;
    unsigned index_buffer_offset;

    CS_LOCALS(r300);
    DBG(r300, DBG_DRAW, "r300: render_draw_elements (count: %d)\n", count);

    u_upload_data(r300->uploader, 0, count * 2, indices,
                  &index_buffer_offset, &index_buffer);
    if (!index_buffer) {
        return;
    }

    if (!r300_prepare_for_rendering(r300,
                                    PREP_EMIT_STATES |
                                    PREP_EMIT_VARRAYS_SWTCL | PREP_INDEXED,
                                    index_buffer, 12, 0, 0, -1)) {
        pipe_resource_reference(&index_buffer, NULL);
        return;
    }

    BEGIN_CS(12);
    OUT_CS_REG(R300_GA_COLOR_CONTROL,
               r300_provoking_vertex_fixes(r300, r300render->prim));
    OUT_CS_REG(R300_VAP_VF_MAX_VTX_INDX, max_index);

    OUT_CS_PKT3(R300_PACKET3_3D_DRAW_INDX_2, 0);
    OUT_CS(R300_VAP_VF_CNTL__PRIM_WALK_INDICES | (count << 16) |
           r300render->hwprim);

    OUT_CS_PKT3(R300_PACKET3_INDX_BUFFER, 2);
    OUT_CS(R300_INDX_BUFFER_ONE_REG_WR | (R300_VAP_PORT_IDX0 >> 2));
    OUT_CS(index_buffer_offset);
    OUT_CS((count + 1) / 2);
    OUT_CS_RELOC(r300_resource(index_buffer));
    END_CS;

    pipe_resource_reference(&index_buffer, NULL);
}

static void r300_render_destroy(struct vbuf_render* render)
{
    FREE(render);
}

static struct vbuf_render* r300_render_create(struct r300_context* r300)
{
    struct r300_render* r300render = CALLOC_STRUCT(r300_render);

    r300render->r300 = r300;

    r300render->base.max_vertex_buffer_bytes = R300_MAX_DRAW_VBO_SIZE;
    r300render->base.max_indices = 16 * 1024;

    r300render->base.get_vertex_info = r300_render_get_vertex_info;
    r300render->base.allocate_vertices = r300_render_allocate_vertices;
    r300render->base.map_vertices = r300_render_map_vertices;
    r300render->base.unmap_vertices = r300_render_unmap_vertices;
    r300render->base.set_primitive = r300_render_set_primitive;
    r300render->base.draw_elements = r300_render_draw_elements;
    r300render->base.draw_arrays = r300_render_draw_arrays;
    r300render->base.release_vertices = r300_render_release_vertices;
    r300render->base.destroy = r300_render_destroy;

    return &r300render->base;
}

struct draw_stage* r300_draw_stage(struct r300_context* r300)
{
    struct vbuf_render* render;
    struct draw_stage* stage;

    render = r300_render_create(r300);

    if (!render) {
        return NULL;
    }

    stage = draw_vbuf_stage(r300->draw, render);

    if (!stage) {
        render->destroy(render);
        return NULL;
    }

    draw_set_render(r300->draw, render);

    return stage;
}

/****************************************************************************
 *                         End of SW TCL functions                          *
 ***************************************************************************/

/* This functions is used to draw a rectangle for the blitter module.
 *
 * If we rendered a quad, the pixels on the main diagonal
 * would be computed and stored twice, which makes the clear/copy codepaths
 * somewhat inefficient. Instead we use a rectangular point sprite. */
void r300_blitter_draw_rectangle(struct blitter_context *blitter,
                                 int x1, int y1, int x2, int y2,
                                 float depth,
                                 enum blitter_attrib_type type,
                                 const union pipe_color_union *attrib)
{
    struct r300_context *r300 = r300_context(util_blitter_get_pipe(blitter));
    unsigned last_sprite_coord_enable = r300->sprite_coord_enable;
    unsigned width = x2 - x1;
    unsigned height = y2 - y1;
    unsigned vertex_size =
            type == UTIL_BLITTER_ATTRIB_COLOR || !r300->draw ? 8 : 4;
    unsigned dwords = 13 + vertex_size +
                      (type == UTIL_BLITTER_ATTRIB_TEXCOORD ? 7 : 0);
    static const union pipe_color_union zeros;
    CS_LOCALS(r300);

    /* XXX workaround for a lockup in MSAA resolve on SWTCL chipsets, this
     * function most probably doesn't handle type=NONE correctly */
    if (!r300->screen->caps.has_tcl && type == UTIL_BLITTER_ATTRIB_NONE) {
        util_blitter_draw_rectangle(blitter, x1, y1, x2, y2, depth, type, attrib);
        return;
    }

    if (r300->skip_rendering)
        return;

    if (type == UTIL_BLITTER_ATTRIB_TEXCOORD)
        r300->sprite_coord_enable = 1;

    r300_update_derived_state(r300);

    /* Mark some states we don't care about as non-dirty. */
    r300->viewport_state.dirty = FALSE;

    if (!r300_prepare_for_rendering(r300, PREP_EMIT_STATES, NULL, dwords, 0, 0, -1))
        goto done;

    DBG(r300, DBG_DRAW, "r300: draw_rectangle\n");

    BEGIN_CS(dwords);
    /* Set up GA. */
    OUT_CS_REG(R300_GA_POINT_SIZE, (height * 6) | ((width * 6) << 16));

    if (type == UTIL_BLITTER_ATTRIB_TEXCOORD) {
        /* Set up the GA to generate texcoords. */
        OUT_CS_REG(R300_GB_ENABLE, R300_GB_POINT_STUFF_ENABLE |
                   (R300_GB_TEX_STR << R300_GB_TEX0_SOURCE_SHIFT));
        OUT_CS_REG_SEQ(R300_GA_POINT_S0, 4);
        OUT_CS_32F(attrib->f[0]);
        OUT_CS_32F(attrib->f[3]);
        OUT_CS_32F(attrib->f[2]);
        OUT_CS_32F(attrib->f[1]);
    }

    /* Set up VAP controls. */
    OUT_CS_REG(R300_VAP_CLIP_CNTL, R300_CLIP_DISABLE);
    OUT_CS_REG(R300_VAP_VTE_CNTL, R300_VTX_XY_FMT | R300_VTX_Z_FMT);
    OUT_CS_REG(R300_VAP_VTX_SIZE, vertex_size);
    OUT_CS_REG_SEQ(R300_VAP_VF_MAX_VTX_INDX, 2);
    OUT_CS(1);
    OUT_CS(0);

    /* Draw. */
    OUT_CS_PKT3(R300_PACKET3_3D_DRAW_IMMD_2, vertex_size);
    OUT_CS(R300_VAP_VF_CNTL__PRIM_WALK_VERTEX_EMBEDDED | (1 << 16) |
           R300_VAP_VF_CNTL__PRIM_POINTS);

    OUT_CS_32F(x1 + width * 0.5f);
    OUT_CS_32F(y1 + height * 0.5f);
    OUT_CS_32F(depth);
    OUT_CS_32F(1);

    if (vertex_size == 8) {
        if (!attrib)
            attrib = &zeros;
        OUT_CS_TABLE(attrib->f, 4);
    }
    END_CS;

done:
    /* Restore the state. */
    r300_mark_atom_dirty(r300, &r300->rs_state);
    r300_mark_atom_dirty(r300, &r300->viewport_state);

    r300->sprite_coord_enable = last_sprite_coord_enable;
}

void r300_init_render_functions(struct r300_context *r300)
{
    /* Set draw functions based on presence of HW TCL. */
    if (r300->screen->caps.has_tcl) {
        r300->context.draw_vbo = r300_draw_vbo;
    } else {
        r300->context.draw_vbo = r300_swtcl_draw_vbo;
    }

    /* Plug in the two-sided stencil reference value fallback if needed. */
    if (!r300->screen->caps.is_r500)
        r300_plug_in_stencil_ref_fallback(r300);
}