summaryrefslogtreecommitdiff
path: root/src/gallium/drivers/radeon/r600_texture.c
blob: 9a46c53f3f7e3558e21baccde4598104fcefd28d (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
1209
1210
1211
1212
1213
1214
1215
1216
1217
1218
1219
1220
1221
1222
1223
1224
1225
1226
1227
1228
1229
1230
1231
1232
1233
1234
1235
1236
1237
1238
1239
1240
1241
1242
1243
1244
1245
1246
1247
1248
1249
1250
1251
1252
1253
1254
1255
1256
1257
1258
1259
1260
1261
1262
1263
1264
1265
1266
1267
1268
1269
1270
1271
1272
1273
1274
1275
1276
1277
1278
1279
1280
1281
1282
1283
1284
1285
1286
1287
1288
1289
1290
1291
1292
1293
1294
1295
1296
1297
1298
1299
1300
1301
1302
1303
1304
1305
1306
1307
1308
1309
1310
1311
/*
 * Copyright 2010 Jerome Glisse <glisse@freedesktop.org>
 *
 * 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.
 *
 * Authors:
 *      Jerome Glisse
 *      Corbin Simpson
 */
#include "r600_pipe_common.h"
#include "r600_cs.h"
#include "util/u_format.h"
#include "util/u_memory.h"
#include "util/u_pack_color.h"
#include <errno.h>
#include <inttypes.h>

/* Same as resource_copy_region, except that both upsampling and downsampling are allowed. */
static void r600_copy_region_with_blit(struct pipe_context *pipe,
				       struct pipe_resource *dst,
                                       unsigned dst_level,
                                       unsigned dstx, unsigned dsty, unsigned dstz,
                                       struct pipe_resource *src,
                                       unsigned src_level,
                                       const struct pipe_box *src_box)
{
	struct pipe_blit_info blit;

	memset(&blit, 0, sizeof(blit));
	blit.src.resource = src;
	blit.src.format = src->format;
	blit.src.level = src_level;
	blit.src.box = *src_box;
	blit.dst.resource = dst;
	blit.dst.format = dst->format;
	blit.dst.level = dst_level;
	blit.dst.box.x = dstx;
	blit.dst.box.y = dsty;
	blit.dst.box.z = dstz;
	blit.dst.box.width = src_box->width;
	blit.dst.box.height = src_box->height;
	blit.dst.box.depth = src_box->depth;
	blit.mask = util_format_get_mask(src->format) &
		    util_format_get_mask(dst->format);
	blit.filter = PIPE_TEX_FILTER_NEAREST;

	if (blit.mask) {
		pipe->blit(pipe, &blit);
	}
}

/* Copy from a full GPU texture to a transfer's staging one. */
static void r600_copy_to_staging_texture(struct pipe_context *ctx, struct r600_transfer *rtransfer)
{
	struct r600_common_context *rctx = (struct r600_common_context*)ctx;
	struct pipe_transfer *transfer = (struct pipe_transfer*)rtransfer;
	struct pipe_resource *dst = &rtransfer->staging->b.b;
	struct pipe_resource *src = transfer->resource;

	if (src->nr_samples > 1) {
		r600_copy_region_with_blit(ctx, dst, 0, 0, 0, 0,
					   src, transfer->level, &transfer->box);
		return;
	}

	rctx->dma_copy(ctx, dst, 0, 0, 0, 0, src, transfer->level,
		       &transfer->box);
}

/* Copy from a transfer's staging texture to a full GPU one. */
static void r600_copy_from_staging_texture(struct pipe_context *ctx, struct r600_transfer *rtransfer)
{
	struct r600_common_context *rctx = (struct r600_common_context*)ctx;
	struct pipe_transfer *transfer = (struct pipe_transfer*)rtransfer;
	struct pipe_resource *dst = transfer->resource;
	struct pipe_resource *src = &rtransfer->staging->b.b;
	struct pipe_box sbox;

	u_box_3d(0, 0, 0, transfer->box.width, transfer->box.height, transfer->box.depth, &sbox);

	if (dst->nr_samples > 1) {
		r600_copy_region_with_blit(ctx, dst, transfer->level,
					   transfer->box.x, transfer->box.y, transfer->box.z,
					   src, 0, &sbox);
		return;
	}

	rctx->dma_copy(ctx, dst, transfer->level,
		       transfer->box.x, transfer->box.y, transfer->box.z,
		       src, 0, &sbox);
}

static unsigned r600_texture_get_offset(struct r600_texture *rtex, unsigned level,
					const struct pipe_box *box)
{
	enum pipe_format format = rtex->resource.b.b.format;

	return rtex->surface.level[level].offset +
	       box->z * rtex->surface.level[level].slice_size +
	       box->y / util_format_get_blockheight(format) * rtex->surface.level[level].pitch_bytes +
	       box->x / util_format_get_blockwidth(format) * util_format_get_blocksize(format);
}

static int r600_init_surface(struct r600_common_screen *rscreen,
			     struct radeon_surface *surface,
			     const struct pipe_resource *ptex,
			     unsigned array_mode,
			     bool is_flushed_depth)
{
	const struct util_format_description *desc =
		util_format_description(ptex->format);
	bool is_depth, is_stencil;

	is_depth = util_format_has_depth(desc);
	is_stencil = util_format_has_stencil(desc);

	surface->npix_x = ptex->width0;
	surface->npix_y = ptex->height0;
	surface->npix_z = ptex->depth0;
	surface->blk_w = util_format_get_blockwidth(ptex->format);
	surface->blk_h = util_format_get_blockheight(ptex->format);
	surface->blk_d = 1;
	surface->array_size = 1;
	surface->last_level = ptex->last_level;

	if (rscreen->chip_class >= EVERGREEN && !is_flushed_depth &&
	    ptex->format == PIPE_FORMAT_Z32_FLOAT_S8X24_UINT) {
		surface->bpe = 4; /* stencil is allocated separately on evergreen */
	} else {
		surface->bpe = util_format_get_blocksize(ptex->format);
		/* align byte per element on dword */
		if (surface->bpe == 3) {
			surface->bpe = 4;
		}
	}

	surface->nsamples = ptex->nr_samples ? ptex->nr_samples : 1;
	surface->flags = RADEON_SURF_SET(array_mode, MODE);

	switch (ptex->target) {
	case PIPE_TEXTURE_1D:
		surface->flags |= RADEON_SURF_SET(RADEON_SURF_TYPE_1D, TYPE);
		break;
	case PIPE_TEXTURE_RECT:
	case PIPE_TEXTURE_2D:
		surface->flags |= RADEON_SURF_SET(RADEON_SURF_TYPE_2D, TYPE);
		break;
	case PIPE_TEXTURE_3D:
		surface->flags |= RADEON_SURF_SET(RADEON_SURF_TYPE_3D, TYPE);
		break;
	case PIPE_TEXTURE_1D_ARRAY:
		surface->flags |= RADEON_SURF_SET(RADEON_SURF_TYPE_1D_ARRAY, TYPE);
		surface->array_size = ptex->array_size;
		break;
	case PIPE_TEXTURE_2D_ARRAY:
	case PIPE_TEXTURE_CUBE_ARRAY: /* cube array layout like 2d array */
		surface->flags |= RADEON_SURF_SET(RADEON_SURF_TYPE_2D_ARRAY, TYPE);
		surface->array_size = ptex->array_size;
		break;
	case PIPE_TEXTURE_CUBE:
		surface->flags |= RADEON_SURF_SET(RADEON_SURF_TYPE_CUBEMAP, TYPE);
		break;
	case PIPE_BUFFER:
	default:
		return -EINVAL;
	}
	if (ptex->bind & PIPE_BIND_SCANOUT) {
		surface->flags |= RADEON_SURF_SCANOUT;
	}

	if (!is_flushed_depth && is_depth) {
		surface->flags |= RADEON_SURF_ZBUFFER;

		if (is_stencil) {
			surface->flags |= RADEON_SURF_SBUFFER |
					  RADEON_SURF_HAS_SBUFFER_MIPTREE;
		}
	}
	if (rscreen->chip_class >= SI) {
		surface->flags |= RADEON_SURF_HAS_TILE_MODE_INDEX;
	}
	return 0;
}

static int r600_setup_surface(struct pipe_screen *screen,
			      struct r600_texture *rtex,
			      unsigned pitch_in_bytes_override)
{
	struct r600_common_screen *rscreen = (struct r600_common_screen*)screen;
	int r;

	r = rscreen->ws->surface_init(rscreen->ws, &rtex->surface);
	if (r) {
		return r;
	}

	rtex->size = rtex->surface.bo_size;

	if (pitch_in_bytes_override && pitch_in_bytes_override != rtex->surface.level[0].pitch_bytes) {
		/* old ddx on evergreen over estimate alignment for 1d, only 1 level
		 * for those
		 */
		rtex->surface.level[0].nblk_x = pitch_in_bytes_override / rtex->surface.bpe;
		rtex->surface.level[0].pitch_bytes = pitch_in_bytes_override;
		rtex->surface.level[0].slice_size = pitch_in_bytes_override * rtex->surface.level[0].nblk_y;
		if (rtex->surface.flags & RADEON_SURF_SBUFFER) {
			rtex->surface.stencil_offset =
			rtex->surface.stencil_level[0].offset = rtex->surface.level[0].slice_size;
		}
	}
	return 0;
}

static boolean r600_texture_get_handle(struct pipe_screen* screen,
				       struct pipe_resource *ptex,
				       struct winsys_handle *whandle)
{
	struct r600_texture *rtex = (struct r600_texture*)ptex;
	struct r600_resource *resource = &rtex->resource;
	struct radeon_surface *surface = &rtex->surface;
	struct r600_common_screen *rscreen = (struct r600_common_screen*)screen;

	rscreen->ws->buffer_set_tiling(resource->buf,
				       NULL,
				       surface->level[0].mode >= RADEON_SURF_MODE_1D ?
				       RADEON_LAYOUT_TILED : RADEON_LAYOUT_LINEAR,
				       surface->level[0].mode >= RADEON_SURF_MODE_2D ?
				       RADEON_LAYOUT_TILED : RADEON_LAYOUT_LINEAR,
				       surface->bankw, surface->bankh,
				       surface->tile_split,
				       surface->stencil_tile_split,
				       surface->mtilea,
				       surface->level[0].pitch_bytes,
				       (surface->flags & RADEON_SURF_SCANOUT) != 0);

	return rscreen->ws->buffer_get_handle(resource->buf,
						surface->level[0].pitch_bytes, whandle);
}

static void r600_texture_destroy(struct pipe_screen *screen,
				 struct pipe_resource *ptex)
{
	struct r600_texture *rtex = (struct r600_texture*)ptex;
	struct r600_resource *resource = &rtex->resource;

	if (rtex->flushed_depth_texture)
		pipe_resource_reference((struct pipe_resource **)&rtex->flushed_depth_texture, NULL);

	pipe_resource_reference((struct pipe_resource**)&rtex->htile_buffer, NULL);
	if (rtex->cmask_buffer != &rtex->resource) {
	    pipe_resource_reference((struct pipe_resource**)&rtex->cmask_buffer, NULL);
	}
	pb_reference(&resource->buf, NULL);
	FREE(rtex);
}

static const struct u_resource_vtbl r600_texture_vtbl;

/* The number of samples can be specified independently of the texture. */
void r600_texture_get_fmask_info(struct r600_common_screen *rscreen,
				 struct r600_texture *rtex,
				 unsigned nr_samples,
				 struct r600_fmask_info *out)
{
	/* FMASK is allocated like an ordinary texture. */
	struct radeon_surface fmask = rtex->surface;

	memset(out, 0, sizeof(*out));

	fmask.bo_alignment = 0;
	fmask.bo_size = 0;
	fmask.nsamples = 1;
	fmask.flags |= RADEON_SURF_FMASK;

	/* Force 2D tiling if it wasn't set. This may occur when creating
	 * FMASK for MSAA resolve on R6xx. On R6xx, the single-sample
	 * destination buffer must have an FMASK too. */
	fmask.flags = RADEON_SURF_CLR(fmask.flags, MODE);
	fmask.flags |= RADEON_SURF_SET(RADEON_SURF_MODE_2D, MODE);

	if (rscreen->chip_class >= SI) {
		fmask.flags |= RADEON_SURF_HAS_TILE_MODE_INDEX;
	}

	switch (nr_samples) {
	case 2:
	case 4:
		fmask.bpe = 1;
		if (rscreen->chip_class <= CAYMAN) {
			fmask.bankh = 4;
		}
		break;
	case 8:
		fmask.bpe = 4;
		break;
	default:
		R600_ERR("Invalid sample count for FMASK allocation.\n");
		return;
	}

	/* Overallocate FMASK on R600-R700 to fix colorbuffer corruption.
	 * This can be fixed by writing a separate FMASK allocator specifically
	 * for R600-R700 asics. */
	if (rscreen->chip_class <= R700) {
		fmask.bpe *= 2;
	}

	if (rscreen->ws->surface_init(rscreen->ws, &fmask)) {
		R600_ERR("Got error in surface_init while allocating FMASK.\n");
		return;
	}

	assert(fmask.level[0].mode == RADEON_SURF_MODE_2D);

	out->slice_tile_max = (fmask.level[0].nblk_x * fmask.level[0].nblk_y) / 64;
	if (out->slice_tile_max)
		out->slice_tile_max -= 1;

	out->tile_mode_index = fmask.tiling_index[0];
	out->pitch = fmask.level[0].nblk_x;
	out->bank_height = fmask.bankh;
	out->alignment = MAX2(256, fmask.bo_alignment);
	out->size = fmask.bo_size;
}

static void r600_texture_allocate_fmask(struct r600_common_screen *rscreen,
					struct r600_texture *rtex)
{
	r600_texture_get_fmask_info(rscreen, rtex,
				    rtex->resource.b.b.nr_samples, &rtex->fmask);

	rtex->fmask.offset = align(rtex->size, rtex->fmask.alignment);
	rtex->size = rtex->fmask.offset + rtex->fmask.size;
}

void r600_texture_get_cmask_info(struct r600_common_screen *rscreen,
				 struct r600_texture *rtex,
				 struct r600_cmask_info *out)
{
	unsigned cmask_tile_width = 8;
	unsigned cmask_tile_height = 8;
	unsigned cmask_tile_elements = cmask_tile_width * cmask_tile_height;
	unsigned element_bits = 4;
	unsigned cmask_cache_bits = 1024;
	unsigned num_pipes = rscreen->tiling_info.num_channels;
	unsigned pipe_interleave_bytes = rscreen->tiling_info.group_bytes;

	unsigned elements_per_macro_tile = (cmask_cache_bits / element_bits) * num_pipes;
	unsigned pixels_per_macro_tile = elements_per_macro_tile * cmask_tile_elements;
	unsigned sqrt_pixels_per_macro_tile = sqrt(pixels_per_macro_tile);
	unsigned macro_tile_width = util_next_power_of_two(sqrt_pixels_per_macro_tile);
	unsigned macro_tile_height = pixels_per_macro_tile / macro_tile_width;

	unsigned pitch_elements = align(rtex->surface.npix_x, macro_tile_width);
	unsigned height = align(rtex->surface.npix_y, macro_tile_height);

	unsigned base_align = num_pipes * pipe_interleave_bytes;
	unsigned slice_bytes =
		((pitch_elements * height * element_bits + 7) / 8) / cmask_tile_elements;

	assert(macro_tile_width % 128 == 0);
	assert(macro_tile_height % 128 == 0);

	out->slice_tile_max = ((pitch_elements * height) / (128*128)) - 1;
	out->alignment = MAX2(256, base_align);
	out->size = (util_max_layer(&rtex->resource.b.b, 0) + 1) *
		    align(slice_bytes, base_align);
}

static void si_texture_get_cmask_info(struct r600_common_screen *rscreen,
				      struct r600_texture *rtex,
				      struct r600_cmask_info *out)
{
	unsigned pipe_interleave_bytes = rscreen->tiling_info.group_bytes;
	unsigned num_pipes = rscreen->tiling_info.num_channels;
	unsigned cl_width, cl_height;

	switch (num_pipes) {
	case 2:
		cl_width = 32;
		cl_height = 16;
		break;
	case 4:
		cl_width = 32;
		cl_height = 32;
		break;
	case 8:
		cl_width = 64;
		cl_height = 32;
		break;
	case 16: /* Hawaii */
		cl_width = 64;
		cl_height = 64;
		break;
	default:
		assert(0);
		return;
	}

	unsigned base_align = num_pipes * pipe_interleave_bytes;

	unsigned width = align(rtex->surface.npix_x, cl_width*8);
	unsigned height = align(rtex->surface.npix_y, cl_height*8);
	unsigned slice_elements = (width * height) / (8*8);

	/* Each element of CMASK is a nibble. */
	unsigned slice_bytes = slice_elements / 2;

	out->slice_tile_max = (width * height) / (128*128);
	if (out->slice_tile_max)
		out->slice_tile_max -= 1;

	out->alignment = MAX2(256, base_align);
	out->size = (util_max_layer(&rtex->resource.b.b, 0) + 1) *
		    align(slice_bytes, base_align);
}

static void r600_texture_allocate_cmask(struct r600_common_screen *rscreen,
					struct r600_texture *rtex)
{
	if (rscreen->chip_class >= SI) {
		si_texture_get_cmask_info(rscreen, rtex, &rtex->cmask);
	} else {
		r600_texture_get_cmask_info(rscreen, rtex, &rtex->cmask);
	}

	rtex->cmask.offset = align(rtex->size, rtex->cmask.alignment);
	rtex->size = rtex->cmask.offset + rtex->cmask.size;

	if (rscreen->chip_class >= SI)
		rtex->cb_color_info |= SI_S_028C70_FAST_CLEAR(1);
	else
		rtex->cb_color_info |= EG_S_028C70_FAST_CLEAR(1);
}

static void r600_texture_alloc_cmask_separate(struct r600_common_screen *rscreen,
					      struct r600_texture *rtex)
{
	if (rtex->cmask_buffer)
                return;

	assert(rtex->cmask.size == 0);

	if (rscreen->chip_class >= SI) {
		si_texture_get_cmask_info(rscreen, rtex, &rtex->cmask);
	} else {
		r600_texture_get_cmask_info(rscreen, rtex, &rtex->cmask);
	}

	rtex->cmask_buffer = (struct r600_resource *)
		pipe_buffer_create(&rscreen->b, PIPE_BIND_CUSTOM,
				   PIPE_USAGE_DEFAULT, rtex->cmask.size);
	if (rtex->cmask_buffer == NULL) {
		rtex->cmask.size = 0;
		return;
	}

	/* update colorbuffer state bits */
	rtex->cmask.base_address_reg =
		r600_resource_va(&rscreen->b, &rtex->cmask_buffer->b.b) >> 8;

	if (rscreen->chip_class >= SI)
		rtex->cb_color_info |= SI_S_028C70_FAST_CLEAR(1);
	else
		rtex->cb_color_info |= EG_S_028C70_FAST_CLEAR(1);
}

static unsigned si_texture_htile_alloc_size(struct r600_common_screen *rscreen,
					    struct r600_texture *rtex)
{
	unsigned cl_width, cl_height, width, height;
	unsigned slice_elements, slice_bytes, pipe_interleave_bytes, base_align;
	unsigned num_pipes = rscreen->tiling_info.num_channels;

	/* HTILE is broken with 1D tiling on old kernels and CIK. */
	if (rtex->surface.level[0].mode == RADEON_SURF_MODE_1D &&
	    rscreen->chip_class >= CIK && rscreen->info.drm_minor < 38)
		return 0;

	switch (num_pipes) {
	case 2:
		cl_width = 32;
		cl_height = 32;
		break;
	case 4:
		cl_width = 64;
		cl_height = 32;
		break;
	case 8:
		cl_width = 64;
		cl_height = 64;
		break;
	case 16:
		cl_width = 128;
		cl_height = 64;
		break;
	default:
		assert(0);
		return 0;
	}

	width = align(rtex->surface.npix_x, cl_width * 8);
	height = align(rtex->surface.npix_y, cl_height * 8);

	slice_elements = (width * height) / (8 * 8);
	slice_bytes = slice_elements * 4;

	pipe_interleave_bytes = rscreen->tiling_info.group_bytes;
	base_align = num_pipes * pipe_interleave_bytes;

	return (util_max_layer(&rtex->resource.b.b, 0) + 1) *
		align(slice_bytes, base_align);
}

static unsigned r600_texture_htile_alloc_size(struct r600_common_screen *rscreen,
					      struct r600_texture *rtex)
{
	unsigned sw = rtex->surface.level[0].nblk_x * rtex->surface.blk_w;
	unsigned sh = rtex->surface.level[0].nblk_y * rtex->surface.blk_h;
	unsigned npipes = rscreen->info.r600_num_tile_pipes;
	unsigned htile_size;

	/* XXX also use it for other texture targets */
	if (rscreen->info.drm_minor < 26 ||
	    rtex->resource.b.b.target != PIPE_TEXTURE_2D ||
	    rtex->surface.level[0].nblk_x < 32 ||
	    rtex->surface.level[0].nblk_y < 32) {
		return 0;
	}

	/* HW bug on R6xx. */
	if (rscreen->chip_class == R600 &&
	    (rtex->surface.level[0].npix_x > 7680 ||
	     rtex->surface.level[0].npix_y > 7680))
		return 0;

	/* this alignment and htile size only apply to linear htile buffer */
	sw = align(sw, 16 << 3);
	sh = align(sh, npipes << 3);
	htile_size = (sw >> 3) * (sh >> 3) * 4;
	/* must be aligned with 2K * npipes */
	htile_size = align(htile_size, (2 << 10) * npipes);
	return htile_size;
}

static void r600_texture_allocate_htile(struct r600_common_screen *rscreen,
					struct r600_texture *rtex)
{
	unsigned htile_size;
	if (rscreen->chip_class >= SI) {
		htile_size = si_texture_htile_alloc_size(rscreen, rtex);
	} else {
		htile_size = r600_texture_htile_alloc_size(rscreen, rtex);
	}

	if (!htile_size)
		return;

	/* XXX don't allocate it separately */
	rtex->htile_buffer = (struct r600_resource*)
			     pipe_buffer_create(&rscreen->b, PIPE_BIND_CUSTOM,
						PIPE_USAGE_DEFAULT, htile_size);
	if (rtex->htile_buffer == NULL) {
		/* this is not a fatal error as we can still keep rendering
		 * without htile buffer */
		R600_ERR("Failed to create buffer object for htile buffer.\n");
	} else {
		r600_screen_clear_buffer(rscreen, &rtex->htile_buffer->b.b, 0, htile_size, 0);
	}
}

/* Common processing for r600_texture_create and r600_texture_from_handle */
static struct r600_texture *
r600_texture_create_object(struct pipe_screen *screen,
			   const struct pipe_resource *base,
			   unsigned pitch_in_bytes_override,
			   struct pb_buffer *buf,
			   struct radeon_surface *surface)
{
	struct r600_texture *rtex;
	struct r600_resource *resource;
	struct r600_common_screen *rscreen = (struct r600_common_screen*)screen;
	uint64_t va;

	rtex = CALLOC_STRUCT(r600_texture);
	if (rtex == NULL)
		return NULL;

	resource = &rtex->resource;
	resource->b.b = *base;
	resource->b.vtbl = &r600_texture_vtbl;
	pipe_reference_init(&resource->b.b.reference, 1);
	resource->b.b.screen = screen;
	rtex->pitch_override = pitch_in_bytes_override;

	/* don't include stencil-only formats which we don't support for rendering */
	rtex->is_depth = util_format_has_depth(util_format_description(rtex->resource.b.b.format));

	rtex->surface = *surface;
	if (r600_setup_surface(screen, rtex, pitch_in_bytes_override)) {
		FREE(rtex);
		return NULL;
	}

	/* Tiled depth textures utilize the non-displayable tile order.
	 * This must be done after r600_setup_surface.
	 * Applies to R600-Cayman. */
	rtex->non_disp_tiling = rtex->is_depth && rtex->surface.level[0].mode >= RADEON_SURF_MODE_1D;

	if (rtex->is_depth) {
		if (!(base->flags & (R600_RESOURCE_FLAG_TRANSFER |
				     R600_RESOURCE_FLAG_FLUSHED_DEPTH)) &&
		    (rscreen->debug_flags & DBG_HYPERZ)) {

			r600_texture_allocate_htile(rscreen, rtex);
		}
	} else {
		if (base->nr_samples > 1) {
			if (!buf) {
				r600_texture_allocate_fmask(rscreen, rtex);
				r600_texture_allocate_cmask(rscreen, rtex);
				rtex->cmask_buffer = &rtex->resource;
			}
			if (!rtex->fmask.size || !rtex->cmask.size) {
				FREE(rtex);
				return NULL;
			}
		}
	}

	/* Now create the backing buffer. */
	if (!buf) {
		if (!r600_init_resource(rscreen, resource, rtex->size,
					rtex->surface.bo_alignment, TRUE)) {
			FREE(rtex);
			return NULL;
		}
	} else {
		resource->buf = buf;
		resource->cs_buf = rscreen->ws->buffer_get_cs_handle(buf);
		resource->domains = rscreen->ws->buffer_get_initial_domain(resource->cs_buf);
	}

	if (rtex->cmask.size) {
		/* Initialize the cmask to 0xCC (= compressed state). */
		r600_screen_clear_buffer(rscreen, &rtex->cmask_buffer->b.b,
					 rtex->cmask.offset, rtex->cmask.size, 0xCCCCCCCC);
	}

	/* Initialize the CMASK base register value. */
	va = r600_resource_va(&rscreen->b, &rtex->resource.b.b);
	rtex->cmask.base_address_reg = (va + rtex->cmask.offset) >> 8;

	if (rscreen->debug_flags & DBG_VM) {
		fprintf(stderr, "VM start=0x%"PRIu64"  end=0x%"PRIu64" | Texture %ix%ix%i, %i levels, %i samples, %s\n",
			r600_resource_va(screen, &rtex->resource.b.b),
			r600_resource_va(screen, &rtex->resource.b.b) + rtex->resource.buf->size,
			base->width0, base->height0, util_max_layer(base, 0)+1, base->last_level+1,
			base->nr_samples ? base->nr_samples : 1, util_format_short_name(base->format));
	}

	if (rscreen->debug_flags & DBG_TEX ||
	    (rtex->resource.b.b.last_level > 0 && rscreen->debug_flags & DBG_TEXMIP)) {
		printf("Texture: npix_x=%u, npix_y=%u, npix_z=%u, blk_w=%u, "
		       "blk_h=%u, blk_d=%u, array_size=%u, last_level=%u, "
		       "bpe=%u, nsamples=%u, flags=0x%x, %s\n",
		       rtex->surface.npix_x, rtex->surface.npix_y,
		       rtex->surface.npix_z, rtex->surface.blk_w,
		       rtex->surface.blk_h, rtex->surface.blk_d,
		       rtex->surface.array_size, rtex->surface.last_level,
		       rtex->surface.bpe, rtex->surface.nsamples,
		       rtex->surface.flags, util_format_short_name(base->format));
		for (int i = 0; i <= rtex->surface.last_level; i++) {
			printf("  L %i: offset=%"PRIu64", slice_size=%"PRIu64", npix_x=%u, "
			       "npix_y=%u, npix_z=%u, nblk_x=%u, nblk_y=%u, "
			       "nblk_z=%u, pitch_bytes=%u, mode=%u\n",
			       i, rtex->surface.level[i].offset,
			       rtex->surface.level[i].slice_size,
			       u_minify(rtex->resource.b.b.width0, i),
			       u_minify(rtex->resource.b.b.height0, i),
			       u_minify(rtex->resource.b.b.depth0, i),
			       rtex->surface.level[i].nblk_x,
			       rtex->surface.level[i].nblk_y,
			       rtex->surface.level[i].nblk_z,
			       rtex->surface.level[i].pitch_bytes,
			       rtex->surface.level[i].mode);
		}
		if (rtex->surface.flags & RADEON_SURF_SBUFFER) {
			for (int i = 0; i <= rtex->surface.last_level; i++) {
				printf("  S %i: offset=%"PRIu64", slice_size=%"PRIu64", npix_x=%u, "
				       "npix_y=%u, npix_z=%u, nblk_x=%u, nblk_y=%u, "
				       "nblk_z=%u, pitch_bytes=%u, mode=%u\n",
				       i, rtex->surface.stencil_level[i].offset,
				       rtex->surface.stencil_level[i].slice_size,
				       u_minify(rtex->resource.b.b.width0, i),
				       u_minify(rtex->resource.b.b.height0, i),
				       u_minify(rtex->resource.b.b.depth0, i),
				       rtex->surface.stencil_level[i].nblk_x,
				       rtex->surface.stencil_level[i].nblk_y,
				       rtex->surface.stencil_level[i].nblk_z,
				       rtex->surface.stencil_level[i].pitch_bytes,
				       rtex->surface.stencil_level[i].mode);
			}
		}
	}
	return rtex;
}

static unsigned r600_choose_tiling(struct r600_common_screen *rscreen,
				   const struct pipe_resource *templ)
{
	const struct util_format_description *desc = util_format_description(templ->format);

	/* MSAA resources must be 2D tiled. */
	if (templ->nr_samples > 1)
		return RADEON_SURF_MODE_2D;

	/* Transfer resources should be linear. */
	if (templ->flags & R600_RESOURCE_FLAG_TRANSFER)
		return RADEON_SURF_MODE_LINEAR_ALIGNED;

	/* Handle common candidates for the linear mode.
	 * Compressed textures must always be tiled. */
	if (!(templ->flags & R600_RESOURCE_FLAG_FORCE_TILING) &&
	    !util_format_is_compressed(templ->format)) {
		/* Tiling doesn't work with the 422 (SUBSAMPLED) formats on R600-Cayman. */
		if (rscreen->chip_class <= CAYMAN &&
		    desc->layout == UTIL_FORMAT_LAYOUT_SUBSAMPLED)
			return RADEON_SURF_MODE_LINEAR_ALIGNED;

		/* Cursors are linear on SI.
		 * (XXX double-check, maybe also use RADEON_SURF_SCANOUT) */
		if (rscreen->chip_class >= SI &&
		    (templ->bind & PIPE_BIND_CURSOR))
			return RADEON_SURF_MODE_LINEAR_ALIGNED;

		if (templ->bind & PIPE_BIND_LINEAR)
			return RADEON_SURF_MODE_LINEAR_ALIGNED;

		/* Textures with a very small height are recommended to be linear. */
		if (templ->target == PIPE_TEXTURE_1D ||
		    templ->target == PIPE_TEXTURE_1D_ARRAY ||
		    templ->height0 <= 4)
			return RADEON_SURF_MODE_LINEAR_ALIGNED;

		/* Textures likely to be mapped often. */
		if (templ->usage == PIPE_USAGE_STAGING ||
		    templ->usage == PIPE_USAGE_STREAM)
			return RADEON_SURF_MODE_LINEAR_ALIGNED;
	}

	/* Make small textures 1D tiled. */
	if (templ->width0 <= 16 || templ->height0 <= 16)
		return RADEON_SURF_MODE_1D;

	/* The allocator will switch to 1D if needed. */
	return RADEON_SURF_MODE_2D;
}

struct pipe_resource *r600_texture_create(struct pipe_screen *screen,
					  const struct pipe_resource *templ)
{
	struct r600_common_screen *rscreen = (struct r600_common_screen*)screen;
	struct radeon_surface surface = {0};
	int r;

	r = r600_init_surface(rscreen, &surface, templ,
			      r600_choose_tiling(rscreen, templ),
			      templ->flags & R600_RESOURCE_FLAG_FLUSHED_DEPTH);
	if (r) {
		return NULL;
	}
	r = rscreen->ws->surface_best(rscreen->ws, &surface);
	if (r) {
		return NULL;
	}
	return (struct pipe_resource *)r600_texture_create_object(screen, templ,
								  0, NULL, &surface);
}

static struct pipe_resource *r600_texture_from_handle(struct pipe_screen *screen,
						      const struct pipe_resource *templ,
						      struct winsys_handle *whandle)
{
	struct r600_common_screen *rscreen = (struct r600_common_screen*)screen;
	struct pb_buffer *buf = NULL;
	unsigned stride = 0;
	unsigned array_mode;
	enum radeon_bo_layout micro, macro;
	struct radeon_surface surface;
	bool scanout;
	int r;

	/* Support only 2D textures without mipmaps */
	if ((templ->target != PIPE_TEXTURE_2D && templ->target != PIPE_TEXTURE_RECT) ||
	      templ->depth0 != 1 || templ->last_level != 0)
		return NULL;

	buf = rscreen->ws->buffer_from_handle(rscreen->ws, whandle, &stride);
	if (!buf)
		return NULL;

	rscreen->ws->buffer_get_tiling(buf, &micro, &macro,
				       &surface.bankw, &surface.bankh,
				       &surface.tile_split,
				       &surface.stencil_tile_split,
				       &surface.mtilea, &scanout);

	if (macro == RADEON_LAYOUT_TILED)
		array_mode = RADEON_SURF_MODE_2D;
	else if (micro == RADEON_LAYOUT_TILED)
		array_mode = RADEON_SURF_MODE_1D;
	else
		array_mode = RADEON_SURF_MODE_LINEAR_ALIGNED;

	r = r600_init_surface(rscreen, &surface, templ, array_mode, false);
	if (r) {
		return NULL;
	}

	if (scanout)
		surface.flags |= RADEON_SURF_SCANOUT;

	return (struct pipe_resource *)r600_texture_create_object(screen, templ,
								  stride, buf, &surface);
}

bool r600_init_flushed_depth_texture(struct pipe_context *ctx,
				     struct pipe_resource *texture,
				     struct r600_texture **staging)
{
	struct r600_texture *rtex = (struct r600_texture*)texture;
	struct pipe_resource resource;
	struct r600_texture **flushed_depth_texture = staging ?
			staging : &rtex->flushed_depth_texture;

	if (!staging && rtex->flushed_depth_texture)
		return true; /* it's ready */

	resource.target = texture->target;
	resource.format = texture->format;
	resource.width0 = texture->width0;
	resource.height0 = texture->height0;
	resource.depth0 = texture->depth0;
	resource.array_size = texture->array_size;
	resource.last_level = texture->last_level;
	resource.nr_samples = texture->nr_samples;
	resource.usage = staging ? PIPE_USAGE_STAGING : PIPE_USAGE_DEFAULT;
	resource.bind = texture->bind & ~PIPE_BIND_DEPTH_STENCIL;
	resource.flags = texture->flags | R600_RESOURCE_FLAG_FLUSHED_DEPTH;

	if (staging)
		resource.flags |= R600_RESOURCE_FLAG_TRANSFER;

	*flushed_depth_texture = (struct r600_texture *)ctx->screen->resource_create(ctx->screen, &resource);
	if (*flushed_depth_texture == NULL) {
		R600_ERR("failed to create temporary texture to hold flushed depth\n");
		return false;
	}

	(*flushed_depth_texture)->is_flushing_texture = TRUE;
	(*flushed_depth_texture)->non_disp_tiling = false;
	return true;
}

/**
 * Initialize the pipe_resource descriptor to be of the same size as the box,
 * which is supposed to hold a subregion of the texture "orig" at the given
 * mipmap level.
 */
static void r600_init_temp_resource_from_box(struct pipe_resource *res,
					     struct pipe_resource *orig,
					     const struct pipe_box *box,
					     unsigned level, unsigned flags)
{
	memset(res, 0, sizeof(*res));
	res->format = orig->format;
	res->width0 = box->width;
	res->height0 = box->height;
	res->depth0 = 1;
	res->array_size = 1;
	res->usage = flags & R600_RESOURCE_FLAG_TRANSFER ? PIPE_USAGE_STAGING : PIPE_USAGE_DEFAULT;
	res->flags = flags;

	/* We must set the correct texture target and dimensions for a 3D box. */
	if (box->depth > 1 && util_max_layer(orig, level) > 0)
		res->target = orig->target;
	else
		res->target = PIPE_TEXTURE_2D;

	switch (res->target) {
	case PIPE_TEXTURE_1D_ARRAY:
	case PIPE_TEXTURE_2D_ARRAY:
	case PIPE_TEXTURE_CUBE_ARRAY:
		res->array_size = box->depth;
		break;
	case PIPE_TEXTURE_3D:
		res->depth0 = box->depth;
		break;
	default:;
	}
}

static void *r600_texture_transfer_map(struct pipe_context *ctx,
				       struct pipe_resource *texture,
				       unsigned level,
				       unsigned usage,
				       const struct pipe_box *box,
				       struct pipe_transfer **ptransfer)
{
	struct r600_common_context *rctx = (struct r600_common_context*)ctx;
	struct r600_texture *rtex = (struct r600_texture*)texture;
	struct r600_transfer *trans;
	boolean use_staging_texture = FALSE;
	struct r600_resource *buf;
	unsigned offset = 0;
	char *map;

	/* We cannot map a tiled texture directly because the data is
	 * in a different order, therefore we do detiling using a blit.
	 *
	 * Also, use a temporary in GTT memory for read transfers, as
	 * the CPU is much happier reading out of cached system memory
	 * than uncached VRAM.
	 */
	if (rtex->surface.level[level].mode >= RADEON_SURF_MODE_1D)
		use_staging_texture = TRUE;

	/* Untiled buffers in VRAM, which is slow for CPU reads */
	if ((usage & PIPE_TRANSFER_READ) && !(usage & PIPE_TRANSFER_MAP_DIRECTLY) &&
	    (rtex->resource.domains == RADEON_DOMAIN_VRAM)) {
		use_staging_texture = TRUE;
	}

	/* Use a staging texture for uploads if the underlying BO is busy. */
	if (!(usage & PIPE_TRANSFER_READ) &&
	    (r600_rings_is_buffer_referenced(rctx, rtex->resource.cs_buf, RADEON_USAGE_READWRITE) ||
	     rctx->ws->buffer_is_busy(rtex->resource.buf, RADEON_USAGE_READWRITE))) {
		use_staging_texture = TRUE;
	}

	if (texture->flags & R600_RESOURCE_FLAG_TRANSFER) {
		use_staging_texture = FALSE;
	}

	if (use_staging_texture && (usage & PIPE_TRANSFER_MAP_DIRECTLY)) {
		return NULL;
	}

	trans = CALLOC_STRUCT(r600_transfer);
	if (trans == NULL)
		return NULL;
	trans->transfer.resource = texture;
	trans->transfer.level = level;
	trans->transfer.usage = usage;
	trans->transfer.box = *box;

	if (rtex->is_depth) {
		struct r600_texture *staging_depth;

		if (rtex->resource.b.b.nr_samples > 1) {
			/* MSAA depth buffers need to be converted to single sample buffers.
			 *
			 * Mapping MSAA depth buffers can occur if ReadPixels is called
			 * with a multisample GLX visual.
			 *
			 * First downsample the depth buffer to a temporary texture,
			 * then decompress the temporary one to staging.
			 *
			 * Only the region being mapped is transfered.
			 */
			struct pipe_resource resource;

			r600_init_temp_resource_from_box(&resource, texture, box, level, 0);

			if (!r600_init_flushed_depth_texture(ctx, &resource, &staging_depth)) {
				R600_ERR("failed to create temporary texture to hold untiled copy\n");
				FREE(trans);
				return NULL;
			}

			if (usage & PIPE_TRANSFER_READ) {
				struct pipe_resource *temp = ctx->screen->resource_create(ctx->screen, &resource);

				r600_copy_region_with_blit(ctx, temp, 0, 0, 0, 0, texture, level, box);
				rctx->blit_decompress_depth(ctx, (struct r600_texture*)temp, staging_depth,
							    0, 0, 0, box->depth, 0, 0);
				pipe_resource_reference((struct pipe_resource**)&temp, NULL);
			}
		}
		else {
			/* XXX: only readback the rectangle which is being mapped? */
			/* XXX: when discard is true, no need to read back from depth texture */
			if (!r600_init_flushed_depth_texture(ctx, texture, &staging_depth)) {
				R600_ERR("failed to create temporary texture to hold untiled copy\n");
				FREE(trans);
				return NULL;
			}

			rctx->blit_decompress_depth(ctx, rtex, staging_depth,
						    level, level,
						    box->z, box->z + box->depth - 1,
						    0, 0);

			offset = r600_texture_get_offset(staging_depth, level, box);
		}

		trans->transfer.stride = staging_depth->surface.level[level].pitch_bytes;
		trans->transfer.layer_stride = staging_depth->surface.level[level].slice_size;
		trans->staging = (struct r600_resource*)staging_depth;
	} else if (use_staging_texture) {
		struct pipe_resource resource;
		struct r600_texture *staging;

		r600_init_temp_resource_from_box(&resource, texture, box, level,
						 R600_RESOURCE_FLAG_TRANSFER);

		/* Create the temporary texture. */
		staging = (struct r600_texture*)ctx->screen->resource_create(ctx->screen, &resource);
		if (staging == NULL) {
			R600_ERR("failed to create temporary texture to hold untiled copy\n");
			FREE(trans);
			return NULL;
		}
		trans->staging = &staging->resource;
		trans->transfer.stride = staging->surface.level[0].pitch_bytes;
		trans->transfer.layer_stride = staging->surface.level[0].slice_size;
		if (usage & PIPE_TRANSFER_READ) {
			r600_copy_to_staging_texture(ctx, trans);
		}
	} else {
		/* the resource is mapped directly */
		trans->transfer.stride = rtex->surface.level[level].pitch_bytes;
		trans->transfer.layer_stride = rtex->surface.level[level].slice_size;
		offset = r600_texture_get_offset(rtex, level, box);
	}

	if (trans->staging) {
		buf = trans->staging;
		if (!rtex->is_depth && !(usage & PIPE_TRANSFER_READ))
			usage |= PIPE_TRANSFER_UNSYNCHRONIZED;
	} else {
		buf = &rtex->resource;
	}

	if (!(map = r600_buffer_map_sync_with_rings(rctx, buf, usage))) {
		pipe_resource_reference((struct pipe_resource**)&trans->staging, NULL);
		FREE(trans);
		return NULL;
	}

	*ptransfer = &trans->transfer;
	return map + offset;
}

static void r600_texture_transfer_unmap(struct pipe_context *ctx,
					struct pipe_transfer* transfer)
{
	struct r600_transfer *rtransfer = (struct r600_transfer*)transfer;
	struct r600_common_context *rctx = (struct r600_common_context*)ctx;
	struct radeon_winsys_cs_handle *buf;
	struct pipe_resource *texture = transfer->resource;
	struct r600_texture *rtex = (struct r600_texture*)texture;

	if (rtransfer->staging) {
		buf = rtransfer->staging->cs_buf;
	} else {
		buf = r600_resource(transfer->resource)->cs_buf;
	}
	rctx->ws->buffer_unmap(buf);

	if ((transfer->usage & PIPE_TRANSFER_WRITE) && rtransfer->staging) {
		if (rtex->is_depth && rtex->resource.b.b.nr_samples <= 1) {
			ctx->resource_copy_region(ctx, texture, transfer->level,
						  transfer->box.x, transfer->box.y, transfer->box.z,
						  &rtransfer->staging->b.b, transfer->level,
						  &transfer->box);
		} else {
			r600_copy_from_staging_texture(ctx, rtransfer);
		}
	}

	if (rtransfer->staging)
		pipe_resource_reference((struct pipe_resource**)&rtransfer->staging, NULL);

	FREE(transfer);
}

static const struct u_resource_vtbl r600_texture_vtbl =
{
	NULL,				/* get_handle */
	r600_texture_destroy,		/* resource_destroy */
	r600_texture_transfer_map,	/* transfer_map */
	NULL,				/* transfer_flush_region */
	r600_texture_transfer_unmap,	/* transfer_unmap */
	NULL				/* transfer_inline_write */
};

struct pipe_surface *r600_create_surface_custom(struct pipe_context *pipe,
						struct pipe_resource *texture,
						const struct pipe_surface *templ,
						unsigned width, unsigned height)
{
	struct r600_surface *surface = CALLOC_STRUCT(r600_surface);

	if (surface == NULL)
		return NULL;

	assert(templ->u.tex.first_layer <= util_max_layer(texture, templ->u.tex.level));
	assert(templ->u.tex.last_layer <= util_max_layer(texture, templ->u.tex.level));

	pipe_reference_init(&surface->base.reference, 1);
	pipe_resource_reference(&surface->base.texture, texture);
	surface->base.context = pipe;
	surface->base.format = templ->format;
	surface->base.width = width;
	surface->base.height = height;
	surface->base.u = templ->u;
	return &surface->base;
}

static struct pipe_surface *r600_create_surface(struct pipe_context *pipe,
						struct pipe_resource *tex,
						const struct pipe_surface *templ)
{
	unsigned level = templ->u.tex.level;

	return r600_create_surface_custom(pipe, tex, templ,
					  u_minify(tex->width0, level),
					  u_minify(tex->height0, level));
}

static void r600_surface_destroy(struct pipe_context *pipe,
				 struct pipe_surface *surface)
{
	struct r600_surface *surf = (struct r600_surface*)surface;
	pipe_resource_reference((struct pipe_resource**)&surf->cb_buffer_fmask, NULL);
	pipe_resource_reference((struct pipe_resource**)&surf->cb_buffer_cmask, NULL);
	pipe_resource_reference(&surface->texture, NULL);
	FREE(surface);
}

unsigned r600_translate_colorswap(enum pipe_format format)
{
	const struct util_format_description *desc = util_format_description(format);

#define HAS_SWIZZLE(chan,swz) (desc->swizzle[chan] == UTIL_FORMAT_SWIZZLE_##swz)

	if (format == PIPE_FORMAT_R11G11B10_FLOAT) /* isn't plain */
		return V_0280A0_SWAP_STD;

	if (desc->layout != UTIL_FORMAT_LAYOUT_PLAIN)
		return ~0U;

	switch (desc->nr_channels) {
	case 1:
		if (HAS_SWIZZLE(0,X))
			return V_0280A0_SWAP_STD; /* X___ */
		else if (HAS_SWIZZLE(3,X))
			return V_0280A0_SWAP_ALT_REV; /* ___X */
		break;
	case 2:
		if ((HAS_SWIZZLE(0,X) && HAS_SWIZZLE(1,Y)) ||
		    (HAS_SWIZZLE(0,X) && HAS_SWIZZLE(1,NONE)) ||
		    (HAS_SWIZZLE(0,NONE) && HAS_SWIZZLE(1,Y)))
			return V_0280A0_SWAP_STD; /* XY__ */
		else if ((HAS_SWIZZLE(0,Y) && HAS_SWIZZLE(1,X)) ||
			 (HAS_SWIZZLE(0,Y) && HAS_SWIZZLE(1,NONE)) ||
		         (HAS_SWIZZLE(0,NONE) && HAS_SWIZZLE(1,X)))
			return V_0280A0_SWAP_STD_REV; /* YX__ */
		else if (HAS_SWIZZLE(0,X) && HAS_SWIZZLE(3,Y))
			return V_0280A0_SWAP_ALT; /* X__Y */
		else if (HAS_SWIZZLE(0,Y) && HAS_SWIZZLE(3,X))
			return V_0280A0_SWAP_ALT_REV; /* Y__X */
		break;
	case 3:
		if (HAS_SWIZZLE(0,X))
			return V_0280A0_SWAP_STD; /* XYZ */
		else if (HAS_SWIZZLE(0,Z))
			return V_0280A0_SWAP_STD_REV; /* ZYX */
		break;
	case 4:
		/* check the middle channels, the 1st and 4th channel can be NONE */
		if (HAS_SWIZZLE(1,Y) && HAS_SWIZZLE(2,Z))
			return V_0280A0_SWAP_STD; /* XYZW */
		else if (HAS_SWIZZLE(1,Z) && HAS_SWIZZLE(2,Y))
			return V_0280A0_SWAP_STD_REV; /* WZYX */
		else if (HAS_SWIZZLE(1,Y) && HAS_SWIZZLE(2,X))
			return V_0280A0_SWAP_ALT; /* ZYXW */
		else if (HAS_SWIZZLE(1,X) && HAS_SWIZZLE(2,Y))
			return V_0280A0_SWAP_ALT_REV; /* WXYZ */
		break;
	}
	return ~0U;
}

static void evergreen_set_clear_color(struct r600_texture *rtex,
				      enum pipe_format surface_format,
				      const union pipe_color_union *color)
{
	union util_color uc;

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

	if (util_format_is_pure_uint(surface_format)) {
		util_format_write_4ui(surface_format, color->ui, 0, &uc, 0, 0, 0, 1, 1);
	} else if (util_format_is_pure_sint(surface_format)) {
		util_format_write_4i(surface_format, color->i, 0, &uc, 0, 0, 0, 1, 1);
	} else {
		util_pack_color(color->f, surface_format, &uc);
	}

	memcpy(rtex->color_clear_value, &uc, 2 * sizeof(uint32_t));
}

void evergreen_do_fast_color_clear(struct r600_common_context *rctx,
				   struct pipe_framebuffer_state *fb,
				   struct r600_atom *fb_state,
				   unsigned *buffers,
				   const union pipe_color_union *color)
{
	int i;

	if (rctx->current_render_cond)
		return;

	for (i = 0; i < fb->nr_cbufs; i++) {
		struct r600_texture *tex;
		unsigned clear_bit = PIPE_CLEAR_COLOR0 << i;

		if (!fb->cbufs[i])
			continue;

		/* if this colorbuffer is not being cleared */
		if (!(*buffers & clear_bit))
			continue;

		tex = (struct r600_texture *)fb->cbufs[i]->texture;

		/* 128-bit formats are unusupported */
		if (util_format_get_blocksizebits(fb->cbufs[i]->format) > 64) {
			continue;
		}

		/* the clear is allowed if all layers are bound */
		if (fb->cbufs[i]->u.tex.first_layer != 0 ||
		    fb->cbufs[i]->u.tex.last_layer != util_max_layer(&tex->resource.b.b, 0)) {
			continue;
		}

		/* cannot clear mipmapped textures */
		if (fb->cbufs[i]->texture->last_level != 0) {
			continue;
		}

		/* only supported on tiled surfaces */
		if (tex->surface.level[0].mode < RADEON_SURF_MODE_1D) {
			continue;
		}

		/* fast color clear with 1D tiling doesn't work on old kernels and CIK */
		if (tex->surface.level[0].mode == RADEON_SURF_MODE_1D &&
		    rctx->chip_class >= CIK && rctx->screen->info.drm_minor < 38) {
			continue;
		}

		/* ensure CMASK is enabled */
		r600_texture_alloc_cmask_separate(rctx->screen, tex);
		if (tex->cmask.size == 0) {
			continue;
		}

		/* Do the fast clear. */
		evergreen_set_clear_color(tex, fb->cbufs[i]->format, color);
		rctx->clear_buffer(&rctx->b, &tex->cmask_buffer->b.b,
				   tex->cmask.offset, tex->cmask.size, 0);

		tex->dirty_level_mask |= 1 << fb->cbufs[i]->u.tex.level;
		fb_state->dirty = true;
		*buffers &= ~clear_bit;
	}
}

void r600_init_screen_texture_functions(struct r600_common_screen *rscreen)
{
	rscreen->b.resource_from_handle = r600_texture_from_handle;
	rscreen->b.resource_get_handle = r600_texture_get_handle;
}

void r600_init_context_texture_functions(struct r600_common_context *rctx)
{
	rctx->b.create_surface = r600_create_surface;
	rctx->b.surface_destroy = r600_surface_destroy;
}