summaryrefslogtreecommitdiffstats
path: root/drivers/gpu/drm/vc4/vc4_validate.c
blob: 2feba55bcef7bb7d171ad13a10c28ee792a119aa (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
/*
 * Copyright © 2014 Broadcom
 *
 * Permission is hereby granted, free of charge, to any person obtaining a
 * copy of this software and associated documentation files (the "Software"),
 * to deal in the Software without restriction, including without limitation
 * the rights to use, copy, modify, merge, publish, distribute, sublicense,
 * and/or sell copies of the Software, and to permit persons to whom the
 * Software is furnished to do so, subject to the following conditions:
 *
 * The above copyright notice and this permission notice (including the next
 * paragraph) shall be included in all copies or substantial portions of the
 * Software.
 *
 * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
 * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
 * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
 * THE AUTHORS OR COPYRIGHT HOLDERS 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.
 */

/**
 * DOC: Command list validator for VC4.
 *
 * Since the VC4 has no IOMMU between it and system memory, a user
 * with access to execute command lists could escalate privilege by
 * overwriting system memory (drawing to it as a framebuffer) or
 * reading system memory it shouldn't (reading it as a vertex buffer
 * or index buffer)
 *
 * We validate binner command lists to ensure that all accesses are
 * within the bounds of the GEM objects referenced by the submitted
 * job.  It explicitly whitelists packets, and looks at the offsets in
 * any address fields to make sure they're contained within the BOs
 * they reference.
 *
 * Note that because CL validation is already reading the
 * user-submitted CL and writing the validated copy out to the memory
 * that the GPU will actually read, this is also where GEM relocation
 * processing (turning BO references into actual addresses for the GPU
 * to use) happens.
 */

#include "uapi/drm/vc4_drm.h"
#include "vc4_drv.h"
#include "vc4_packet.h"

#define VALIDATE_ARGS \
	struct vc4_exec_info *exec,			\
	void *validated,				\
	void *untrusted

/** Return the width in pixels of a 64-byte microtile. */
static uint32_t
utile_width(int cpp)
{
	switch (cpp) {
	case 1:
	case 2:
		return 8;
	case 4:
		return 4;
	case 8:
		return 2;
	default:
		DRM_ERROR("unknown cpp: %d\n", cpp);
		return 1;
	}
}

/** Return the height in pixels of a 64-byte microtile. */
static uint32_t
utile_height(int cpp)
{
	switch (cpp) {
	case 1:
		return 8;
	case 2:
	case 4:
	case 8:
		return 4;
	default:
		DRM_ERROR("unknown cpp: %d\n", cpp);
		return 1;
	}
}

/**
 * size_is_lt() - Returns whether a miplevel of the given size will
 * use the lineartile (LT) tiling layout rather than the normal T
 * tiling layout.
 * @width: Width in pixels of the miplevel
 * @height: Height in pixels of the miplevel
 * @cpp: Bytes per pixel of the pixel format
 */
static bool
size_is_lt(uint32_t width, uint32_t height, int cpp)
{
	return (width <= 4 * utile_width(cpp) ||
		height <= 4 * utile_height(cpp));
}

struct drm_gem_cma_object *
vc4_use_bo(struct vc4_exec_info *exec, uint32_t hindex)
{
	struct vc4_dev *vc4 = exec->dev;
	struct drm_gem_cma_object *obj;
	struct vc4_bo *bo;

	if (WARN_ON_ONCE(vc4->is_vc5))
		return NULL;

	if (hindex >= exec->bo_count) {
		DRM_DEBUG("BO index %d greater than BO count %d\n",
			  hindex, exec->bo_count);
		return NULL;
	}
	obj = exec->bo[hindex];
	bo = to_vc4_bo(&obj->base);

	if (bo->validated_shader) {
		DRM_DEBUG("Trying to use shader BO as something other than "
			  "a shader\n");
		return NULL;
	}

	return obj;
}

static struct drm_gem_cma_object *
vc4_use_handle(struct vc4_exec_info *exec, uint32_t gem_handles_packet_index)
{
	return vc4_use_bo(exec, exec->bo_index[gem_handles_packet_index]);
}

static bool
validate_bin_pos(struct vc4_exec_info *exec, void *untrusted, uint32_t pos)
{
	/* Note that the untrusted pointer passed to these functions is
	 * incremented past the packet byte.
	 */
	return (untrusted - 1 == exec->bin_u + pos);
}

static uint32_t
gl_shader_rec_size(uint32_t pointer_bits)
{
	uint32_t attribute_count = pointer_bits & 7;
	bool extended = pointer_bits & 8;

	if (attribute_count == 0)
		attribute_count = 8;

	if (extended)
		return 100 + attribute_count * 4;
	else
		return 36 + attribute_count * 8;
}

bool
vc4_check_tex_size(struct vc4_exec_info *exec, struct drm_gem_cma_object *fbo,
		   uint32_t offset, uint8_t tiling_format,
		   uint32_t width, uint32_t height, uint8_t cpp)
{
	struct vc4_dev *vc4 = exec->dev;
	uint32_t aligned_width, aligned_height, stride, size;
	uint32_t utile_w = utile_width(cpp);
	uint32_t utile_h = utile_height(cpp);

	if (WARN_ON_ONCE(vc4->is_vc5))
		return false;

	/* The shaded vertex format stores signed 12.4 fixed point
	 * (-2048,2047) offsets from the viewport center, so we should
	 * never have a render target larger than 4096.  The texture
	 * unit can only sample from 2048x2048, so it's even more
	 * restricted.  This lets us avoid worrying about overflow in
	 * our math.
	 */
	if (width > 4096 || height > 4096) {
		DRM_DEBUG("Surface dimensions (%d,%d) too large",
			  width, height);
		return false;
	}

	switch (tiling_format) {
	case VC4_TILING_FORMAT_LINEAR:
		aligned_width = round_up(width, utile_w);
		aligned_height = height;
		break;
	case VC4_TILING_FORMAT_T:
		aligned_width = round_up(width, utile_w * 8);
		aligned_height = round_up(height, utile_h * 8);
		break;
	case VC4_TILING_FORMAT_LT:
		aligned_width = round_up(width, utile_w);
		aligned_height = round_up(height, utile_h);
		break;
	default:
		DRM_DEBUG("buffer tiling %d unsupported\n", tiling_format);
		return false;
	}

	stride = aligned_width * cpp;
	size = stride * aligned_height;

	if (size + offset < size ||
	    size + offset > fbo->base.size) {
		DRM_DEBUG("Overflow in %dx%d (%dx%d) fbo size (%d + %d > %zd)\n",
			  width, height,
			  aligned_width, aligned_height,
			  size, offset, fbo->base.size);
		return false;
	}

	return true;
}

static int
validate_flush(VALIDATE_ARGS)
{
	if (!validate_bin_pos(exec, untrusted, exec->args->bin_cl_size - 1)) {
		DRM_DEBUG("Bin CL must end with VC4_PACKET_FLUSH\n");
		return -EINVAL;
	}
	exec->found_flush = true;

	return 0;
}

static int
validate_start_tile_binning(VALIDATE_ARGS)
{
	if (exec->found_start_tile_binning_packet) {
		DRM_DEBUG("Duplicate VC4_PACKET_START_TILE_BINNING\n");
		return -EINVAL;
	}
	exec->found_start_tile_binning_packet = true;

	if (!exec->found_tile_binning_mode_config_packet) {
		DRM_DEBUG("missing VC4_PACKET_TILE_BINNING_MODE_CONFIG\n");
		return -EINVAL;
	}

	return 0;
}

static int
validate_increment_semaphore(VALIDATE_ARGS)
{
	if (!validate_bin_pos(exec, untrusted, exec->args->bin_cl_size - 2)) {
		DRM_DEBUG("Bin CL must end with "
			  "VC4_PACKET_INCREMENT_SEMAPHORE\n");
		return -EINVAL;
	}
	exec->found_increment_semaphore_packet = true;

	return 0;
}

static int
validate_indexed_prim_list(VALIDATE_ARGS)
{
	struct drm_gem_cma_object *ib;
	uint32_t length = *(uint32_t *)(untrusted + 1);
	uint32_t offset = *(uint32_t *)(untrusted + 5);
	uint32_t max_index = *(uint32_t *)(untrusted + 9);
	uint32_t index_size = (*(uint8_t *)(untrusted + 0) >> 4) ? 2 : 1;
	struct vc4_shader_state *shader_state;

	/* Check overflow condition */
	if (exec->shader_state_count == 0) {
		DRM_DEBUG("shader state must precede primitives\n");
		return -EINVAL;
	}
	shader_state = &exec->shader_state[exec->shader_state_count - 1];

	if (max_index > shader_state->max_index)
		shader_state->max_index = max_index;

	ib = vc4_use_handle(exec, 0);
	if (!ib)
		return -EINVAL;

	exec->bin_dep_seqno = max(exec->bin_dep_seqno,
				  to_vc4_bo(&ib->base)->write_seqno);

	if (offset > ib->base.size ||
	    (ib->base.size - offset) / index_size < length) {
		DRM_DEBUG("IB access overflow (%d + %d*%d > %zd)\n",
			  offset, length, index_size, ib->base.size);
		return -EINVAL;
	}

	*(uint32_t *)(validated + 5) = ib->paddr + offset;

	return 0;
}

static int
validate_gl_array_primitive(VALIDATE_ARGS)
{
	uint32_t length = *(uint32_t *)(untrusted + 1);
	uint32_t base_index = *(uint32_t *)(untrusted + 5);
	uint32_t max_index;
	struct vc4_shader_state *shader_state;

	/* Check overflow condition */
	if (exec->shader_state_count == 0) {
		DRM_DEBUG("shader state must precede primitives\n");
		return -EINVAL;
	}
	shader_state = &exec->shader_state[exec->shader_state_count - 1];

	if (length + base_index < length) {
		DRM_DEBUG("primitive vertex count overflow\n");
		return -EINVAL;
	}
	max_index = length + base_index - 1;

	if (max_index > shader_state->max_index)
		shader_state->max_index = max_index;

	return 0;
}

static int
validate_gl_shader_state(VALIDATE_ARGS)
{
	uint32_t i = exec->shader_state_count++;

	if (i >= exec->shader_state_size) {
		DRM_DEBUG("More requests for shader states than declared\n");
		return -EINVAL;
	}

	exec->shader_state[i].addr = *(uint32_t *)untrusted;
	exec->shader_state[i].max_index = 0;

	if (exec->shader_state[i].addr & ~0xf) {
		DRM_DEBUG("high bits set in GL shader rec reference\n");
		return -EINVAL;
	}

	*(uint32_t *)validated = (exec->shader_rec_p +
				  exec->shader_state[i].addr);

	exec->shader_rec_p +=
		roundup(gl_shader_rec_size(exec->shader_state[i].addr), 16);

	return 0;
}

static int
validate_tile_binning_config(VALIDATE_ARGS)
{
	struct drm_device *dev = exec->exec_bo->base.dev;
	struct vc4_dev *vc4 = to_vc4_dev(dev);
	uint8_t flags;
	uint32_t tile_state_size;
	uint32_t tile_count, bin_addr;
	int bin_slot;

	if (exec->found_tile_binning_mode_config_packet) {
		DRM_DEBUG("Duplicate VC4_PACKET_TILE_BINNING_MODE_CONFIG\n");
		return -EINVAL;
	}
	exec->found_tile_binning_mode_config_packet = true;

	exec->bin_tiles_x = *(uint8_t *)(untrusted + 12);
	exec->bin_tiles_y = *(uint8_t *)(untrusted + 13);
	tile_count = exec->bin_tiles_x * exec->bin_tiles_y;
	flags = *(uint8_t *)(untrusted + 14);

	if (exec->bin_tiles_x == 0 ||
	    exec->bin_tiles_y == 0) {
		DRM_DEBUG("Tile binning config of %dx%d too small\n",
			  exec->bin_tiles_x, exec->bin_tiles_y);
		return -EINVAL;
	}

	if (flags & (VC4_BIN_CONFIG_DB_NON_MS |
		     VC4_BIN_CONFIG_TILE_BUFFER_64BIT)) {
		DRM_DEBUG("unsupported binning config flags 0x%02x\n", flags);
		return -EINVAL;
	}

	bin_slot = vc4_v3d_get_bin_slot(vc4);
	if (bin_slot < 0) {
		if (bin_slot != -EINTR && bin_slot != -ERESTARTSYS) {
			DRM_ERROR("Failed to allocate binner memory: %d\n",
				  bin_slot);
		}
		return bin_slot;
	}

	/* The slot we allocated will only be used by this job, and is
	 * free when the job completes rendering.
	 */
	exec->bin_slots |= BIT(bin_slot);
	bin_addr = vc4->bin_bo->base.paddr + bin_slot * vc4->bin_alloc_size;

	/* The tile state data array is 48 bytes per tile, and we put it at
	 * the start of a BO containing both it and the tile alloc.
	 */
	tile_state_size = 48 * tile_count;

	/* Since the tile alloc array will follow us, align. */
	exec->tile_alloc_offset = bin_addr + roundup(tile_state_size, 4096);

	*(uint8_t *)(validated + 14) =
		((flags & ~(VC4_BIN_CONFIG_ALLOC_INIT_BLOCK_SIZE_MASK |
			    VC4_BIN_CONFIG_ALLOC_BLOCK_SIZE_MASK)) |
		 VC4_BIN_CONFIG_AUTO_INIT_TSDA |
		 VC4_SET_FIELD(VC4_BIN_CONFIG_ALLOC_INIT_BLOCK_SIZE_32,
			       VC4_BIN_CONFIG_ALLOC_INIT_BLOCK_SIZE) |
		 VC4_SET_FIELD(VC4_BIN_CONFIG_ALLOC_BLOCK_SIZE_128,
			       VC4_BIN_CONFIG_ALLOC_BLOCK_SIZE));

	/* tile alloc address. */
	*(uint32_t *)(validated + 0) = exec->tile_alloc_offset;
	/* tile alloc size. */
	*(uint32_t *)(validated + 4) = (bin_addr + vc4->bin_alloc_size -
					exec->tile_alloc_offset);
	/* tile state address. */
	*(uint32_t *)(validated + 8) = bin_addr;

	return 0;
}

static int
validate_gem_handles(VALIDATE_ARGS)
{
	memcpy(exec->bo_index, untrusted, sizeof(exec->bo_index));
	return 0;
}

#define VC4_DEFINE_PACKET(packet, func) \
	[packet] = { packet ## _SIZE, #packet, func }

static const struct cmd_info {
	uint16_t len;
	const char *name;
	int (*func)(struct vc4_exec_info *exec, void *validated,
		    void *untrusted);
} cmd_info[] = {
	VC4_DEFINE_PACKET(VC4_PACKET_HALT, NULL),
	VC4_DEFINE_PACKET(VC4_PACKET_NOP, NULL),
	VC4_DEFINE_PACKET(VC4_PACKET_FLUSH, validate_flush),
	VC4_DEFINE_PACKET(VC4_PACKET_FLUSH_ALL, NULL),
	VC4_DEFINE_PACKET(VC4_PACKET_START_TILE_BINNING,
			  validate_start_tile_binning),
	VC4_DEFINE_PACKET(VC4_PACKET_INCREMENT_SEMAPHORE,
			  validate_increment_semaphore),

	VC4_DEFINE_PACKET(VC4_PACKET_GL_INDEXED_PRIMITIVE,
			  validate_indexed_prim_list),
	VC4_DEFINE_PACKET(VC4_PACKET_GL_ARRAY_PRIMITIVE,
			  validate_gl_array_primitive),

	VC4_DEFINE_PACKET(VC4_PACKET_PRIMITIVE_LIST_FORMAT, NULL),

	VC4_DEFINE_PACKET(VC4_PACKET_GL_SHADER_STATE, validate_gl_shader_state),

	VC4_DEFINE_PACKET(VC4_PACKET_CONFIGURATION_BITS, NULL),
	VC4_DEFINE_PACKET(VC4_PACKET_FLAT_SHADE_FLAGS, NULL),
	VC4_DEFINE_PACKET(VC4_PACKET_POINT_SIZE, NULL),
	VC4_DEFINE_PACKET(VC4_PACKET_LINE_WIDTH, NULL),
	VC4_DEFINE_PACKET(VC4_PACKET_RHT_X_BOUNDARY, NULL),
	VC4_DEFINE_PACKET(VC4_PACKET_DEPTH_OFFSET, NULL),
	VC4_DEFINE_PACKET(VC4_PACKET_CLIP_WINDOW, NULL),
	VC4_DEFINE_PACKET(VC4_PACKET_VIEWPORT_OFFSET, NULL),
	VC4_DEFINE_PACKET(VC4_PACKET_CLIPPER_XY_SCALING, NULL),
	/* Note: The docs say this was also 105, but it was 106 in the
	 * initial userland code drop.
	 */
	VC4_DEFINE_PACKET(VC4_PACKET_CLIPPER_Z_SCALING, NULL),

	VC4_DEFINE_PACKET(VC4_PACKET_TILE_BINNING_MODE_CONFIG,
			  validate_tile_binning_config),

	VC4_DEFINE_PACKET(VC4_PACKET_GEM_HANDLES, validate_gem_handles),
};

int
vc4_validate_bin_cl(struct drm_device *dev,
		    void *validated,
		    void *unvalidated,
		    struct vc4_exec_info *exec)
{
	struct vc4_dev *vc4 = to_vc4_dev(dev);
	uint32_t len = exec->args->bin_cl_size;
	uint32_t dst_offset = 0;
	uint32_t src_offset = 0;

	if (WARN_ON_ONCE(vc4->is_vc5))
		return -ENODEV;

	while (src_offset < len) {
		void *dst_pkt = validated + dst_offset;
		void *src_pkt = unvalidated + src_offset;
		u8 cmd = *(uint8_t *)src_pkt;
		const struct cmd_info *info;

		if (cmd >= ARRAY_SIZE(cmd_info)) {
			DRM_DEBUG("0x%08x: packet %d out of bounds\n",
				  src_offset, cmd);
			return -EINVAL;
		}

		info = &cmd_info[cmd];
		if (!info->name) {
			DRM_DEBUG("0x%08x: packet %d invalid\n",
				  src_offset, cmd);
			return -EINVAL;
		}

		if (src_offset + info->len > len) {
			DRM_DEBUG("0x%08x: packet %d (%s) length 0x%08x "
				  "exceeds bounds (0x%08x)\n",
				  src_offset, cmd, info->name, info->len,
				  src_offset + len);
			return -EINVAL;
		}

		if (cmd != VC4_PACKET_GEM_HANDLES)
			memcpy(dst_pkt, src_pkt, info->len);

		if (info->func && info->func(exec,
					     dst_pkt + 1,
					     src_pkt + 1)) {
			DRM_DEBUG("0x%08x: packet %d (%s) failed to validate\n",
				  src_offset, cmd, info->name);
			return -EINVAL;
		}

		src_offset += info->len;
		/* GEM handle loading doesn't produce HW packets. */
		if (cmd != VC4_PACKET_GEM_HANDLES)
			dst_offset += info->len;

		/* When the CL hits halt, it'll stop reading anything else. */
		if (cmd == VC4_PACKET_HALT)
			break;
	}

	exec->ct0ea = exec->ct0ca + dst_offset;

	if (!exec->found_start_tile_binning_packet) {
		DRM_DEBUG("Bin CL missing VC4_PACKET_START_TILE_BINNING\n");
		return -EINVAL;
	}

	/* The bin CL must be ended with INCREMENT_SEMAPHORE and FLUSH.  The
	 * semaphore is used to trigger the render CL to start up, and the
	 * FLUSH is what caps the bin lists with
	 * VC4_PACKET_RETURN_FROM_SUB_LIST (so they jump back to the main
	 * render CL when they get called to) and actually triggers the queued
	 * semaphore increment.
	 */
	if (!exec->found_increment_semaphore_packet || !exec->found_flush) {
		DRM_DEBUG("Bin CL missing VC4_PACKET_INCREMENT_SEMAPHORE + "
			  "VC4_PACKET_FLUSH\n");
		return -EINVAL;
	}

	return 0;
}

static bool
reloc_tex(struct vc4_exec_info *exec,
	  void *uniform_data_u,
	  struct vc4_texture_sample_info *sample,
	  uint32_t texture_handle_index, bool is_cs)
{
	struct drm_gem_cma_object *tex;
	uint32_t p0 = *(uint32_t *)(uniform_data_u + sample->p_offset[0]);
	uint32_t p1 = *(uint32_t *)(uniform_data_u + sample->p_offset[1]);
	uint32_t p2 = (sample->p_offset[2] != ~0 ?
		       *(uint32_t *)(uniform_data_u + sample->p_offset[2]) : 0);
	uint32_t p3 = (sample->p_offset[3] != ~0 ?
		       *(uint32_t *)(uniform_data_u + sample->p_offset[3]) : 0);
	uint32_t *validated_p0 = exec->uniforms_v + sample->p_offset[0];
	uint32_t offset = p0 & VC4_TEX_P0_OFFSET_MASK;
	uint32_t miplevels = VC4_GET_FIELD(p0, VC4_TEX_P0_MIPLVLS);
	uint32_t width = VC4_GET_FIELD(p1, VC4_TEX_P1_WIDTH);
	uint32_t height = VC4_GET_FIELD(p1, VC4_TEX_P1_HEIGHT);
	uint32_t cpp, tiling_format, utile_w, utile_h;
	uint32_t i;
	uint32_t cube_map_stride = 0;
	enum vc4_texture_data_type type;

	tex = vc4_use_bo(exec, texture_handle_index);
	if (!tex)
		return false;

	if (sample->is_direct) {
		uint32_t remaining_size = tex->base.size - p0;

		if (p0 > tex->base.size - 4) {
			DRM_DEBUG("UBO offset greater than UBO size\n");
			goto fail;
		}
		if (p1 > remaining_size - 4) {
			DRM_DEBUG("UBO clamp would allow reads "
				  "outside of UBO\n");
			goto fail;
		}
		*validated_p0 = tex->paddr + p0;
		return true;
	}

	if (width == 0)
		width = 2048;
	if (height == 0)
		height = 2048;

	if (p0 & VC4_TEX_P0_CMMODE_MASK) {
		if (VC4_GET_FIELD(p2, VC4_TEX_P2_PTYPE) ==
		    VC4_TEX_P2_PTYPE_CUBE_MAP_STRIDE)
			cube_map_stride = p2 & VC4_TEX_P2_CMST_MASK;
		if (VC4_GET_FIELD(p3, VC4_TEX_P2_PTYPE) ==
		    VC4_TEX_P2_PTYPE_CUBE_MAP_STRIDE) {
			if (cube_map_stride) {
				DRM_DEBUG("Cube map stride set twice\n");
				goto fail;
			}

			cube_map_stride = p3 & VC4_TEX_P2_CMST_MASK;
		}
		if (!cube_map_stride) {
			DRM_DEBUG("Cube map stride not set\n");
			goto fail;
		}
	}

	type = (VC4_GET_FIELD(p0, VC4_TEX_P0_TYPE) |
		(VC4_GET_FIELD(p1, VC4_TEX_P1_TYPE4) << 4));

	switch (type) {
	case VC4_TEXTURE_TYPE_RGBA8888:
	case VC4_TEXTURE_TYPE_RGBX8888:
	case VC4_TEXTURE_TYPE_RGBA32R:
		cpp = 4;
		break;
	case VC4_TEXTURE_TYPE_RGBA4444:
	case VC4_TEXTURE_TYPE_RGBA5551:
	case VC4_TEXTURE_TYPE_RGB565:
	case VC4_TEXTURE_TYPE_LUMALPHA:
	case VC4_TEXTURE_TYPE_S16F:
	case VC4_TEXTURE_TYPE_S16:
		cpp = 2;
		break;
	case VC4_TEXTURE_TYPE_LUMINANCE:
	case VC4_TEXTURE_TYPE_ALPHA:
	case VC4_TEXTURE_TYPE_S8:
		cpp = 1;
		break;
	case VC4_TEXTURE_TYPE_ETC1:
		/* ETC1 is arranged as 64-bit blocks, where each block is 4x4
		 * pixels.
		 */
		cpp = 8;
		width = (width + 3) >> 2;
		height = (height + 3) >> 2;
		break;
	case VC4_TEXTURE_TYPE_BW1:
	case VC4_TEXTURE_TYPE_A4:
	case VC4_TEXTURE_TYPE_A1:
	case VC4_TEXTURE_TYPE_RGBA64:
	case VC4_TEXTURE_TYPE_YUV422R:
	default:
		DRM_DEBUG("Texture format %d unsupported\n", type);
		goto fail;
	}
	utile_w = utile_width(cpp);
	utile_h = utile_height(cpp);

	if (type == VC4_TEXTURE_TYPE_RGBA32R) {
		tiling_format = VC4_TILING_FORMAT_LINEAR;
	} else {
		if (size_is_lt(width, height, cpp))
			tiling_format = VC4_TILING_FORMAT_LT;
		else
			tiling_format = VC4_TILING_FORMAT_T;
	}

	if (!vc4_check_tex_size(exec, tex, offset + cube_map_stride * 5,
				tiling_format, width, height, cpp)) {
		goto fail;
	}

	/* The mipmap levels are stored before the base of the texture.  Make
	 * sure there is actually space in the BO.
	 */
	for (i = 1; i <= miplevels; i++) {
		uint32_t level_width = max(width >> i, 1u);
		uint32_t level_height = max(height >> i, 1u);
		uint32_t aligned_width, aligned_height;
		uint32_t level_size;

		/* Once the levels get small enough, they drop from T to LT. */
		if (tiling_format == VC4_TILING_FORMAT_T &&
		    size_is_lt(level_width, level_height, cpp)) {
			tiling_format = VC4_TILING_FORMAT_LT;
		}

		switch (tiling_format) {
		case VC4_TILING_FORMAT_T:
			aligned_width = round_up(level_width, utile_w * 8);
			aligned_height = round_up(level_height, utile_h * 8);
			break;
		case VC4_TILING_FORMAT_LT:
			aligned_width = round_up(level_width, utile_w);
			aligned_height = round_up(level_height, utile_h);
			break;
		default:
			aligned_width = round_up(level_width, utile_w);
			aligned_height = level_height;
			break;
		}

		level_size = aligned_width * cpp * aligned_height;

		if (offset < level_size) {
			DRM_DEBUG("Level %d (%dx%d -> %dx%d) size %db "
				  "overflowed buffer bounds (offset %d)\n",
				  i, level_width, level_height,
				  aligned_width, aligned_height,
				  level_size, offset);
			goto fail;
		}

		offset -= level_size;
	}

	*validated_p0 = tex->paddr + p0;

	if (is_cs) {
		exec->bin_dep_seqno = max(exec->bin_dep_seqno,
					  to_vc4_bo(&tex->base)->write_seqno);
	}

	return true;
 fail:
	DRM_INFO("Texture p0 at %d: 0x%08x\n", sample->p_offset[0], p0);
	DRM_INFO("Texture p1 at %d: 0x%08x\n", sample->p_offset[1], p1);
	DRM_INFO("Texture p2 at %d: 0x%08x\n", sample->p_offset[2], p2);
	DRM_INFO("Texture p3 at %d: 0x%08x\n", sample->p_offset[3], p3);
	return false;
}

static int
validate_gl_shader_rec(struct drm_device *dev,
		       struct vc4_exec_info *exec,
		       struct vc4_shader_state *state)
{
	uint32_t *src_handles;
	void *pkt_u, *pkt_v;
	static const uint32_t shader_reloc_offsets[] = {
		4, /* fs */
		16, /* vs */
		28, /* cs */
	};
	uint32_t shader_reloc_count = ARRAY_SIZE(shader_reloc_offsets);
	struct drm_gem_cma_object *bo[ARRAY_SIZE(shader_reloc_offsets) + 8];
	uint32_t nr_attributes, nr_relocs, packet_size;
	int i;

	nr_attributes = state->addr & 0x7;
	if (nr_attributes == 0)
		nr_attributes = 8;
	packet_size = gl_shader_rec_size(state->addr);

	nr_relocs = ARRAY_SIZE(shader_reloc_offsets) + nr_attributes;
	if (nr_relocs * 4 > exec->shader_rec_size) {
		DRM_DEBUG("overflowed shader recs reading %d handles "
			  "from %d bytes left\n",
			  nr_relocs, exec->shader_rec_size);
		return -EINVAL;
	}
	src_handles = exec->shader_rec_u;
	exec->shader_rec_u += nr_relocs * 4;
	exec->shader_rec_size -= nr_relocs * 4;

	if (packet_size > exec->shader_rec_size) {
		DRM_DEBUG("overflowed shader recs copying %db packet "
			  "from %d bytes left\n",
			  packet_size, exec->shader_rec_size);
		return -EINVAL;
	}
	pkt_u = exec->shader_rec_u;
	pkt_v = exec->shader_rec_v;
	memcpy(pkt_v, pkt_u, packet_size);
	exec->shader_rec_u += packet_size;
	/* Shader recs have to be aligned to 16 bytes (due to the attribute
	 * flags being in the low bytes), so round the next validated shader
	 * rec address up.  This should be safe, since we've got so many
	 * relocations in a shader rec packet.
	 */
	BUG_ON(roundup(packet_size, 16) - packet_size > nr_relocs * 4);
	exec->shader_rec_v += roundup(packet_size, 16);
	exec->shader_rec_size -= packet_size;

	for (i = 0; i < shader_reloc_count; i++) {
		if (src_handles[i] > exec->bo_count) {
			DRM_DEBUG("Shader handle %d too big\n", src_handles[i]);
			return -EINVAL;
		}

		bo[i] = exec->bo[src_handles[i]];
		if (!bo[i])
			return -EINVAL;
	}
	for (i = shader_reloc_count; i < nr_relocs; i++) {
		bo[i] = vc4_use_bo(exec, src_handles[i]);
		if (!bo[i])
			return -EINVAL;
	}

	if (((*(uint16_t *)pkt_u & VC4_SHADER_FLAG_FS_SINGLE_THREAD) == 0) !=
	    to_vc4_bo(&bo[0]->base)->validated_shader->is_threaded) {
		DRM_DEBUG("Thread mode of CL and FS do not match\n");
		return -EINVAL;
	}

	if (to_vc4_bo(&bo[1]->base)->validated_shader->is_threaded ||
	    to_vc4_bo(&bo[2]->base)->validated_shader->is_threaded) {
		DRM_DEBUG("cs and vs cannot be threaded\n");
		return -EINVAL;
	}

	for (i = 0; i < shader_reloc_count; i++) {
		struct vc4_validated_shader_info *validated_shader;
		uint32_t o = shader_reloc_offsets[i];
		uint32_t src_offset = *(uint32_t *)(pkt_u + o);
		uint32_t *texture_handles_u;
		void *uniform_data_u;
		uint32_t tex, uni;

		*(uint32_t *)(pkt_v + o) = bo[i]->paddr + src_offset;

		if (src_offset != 0) {
			DRM_DEBUG("Shaders must be at offset 0 of "
				  "the BO.\n");
			return -EINVAL;
		}

		validated_shader = to_vc4_bo(&bo[i]->base)->validated_shader;
		if (!validated_shader)
			return -EINVAL;

		if (validated_shader->uniforms_src_size >
		    exec->uniforms_size) {
			DRM_DEBUG("Uniforms src buffer overflow\n");
			return -EINVAL;
		}

		texture_handles_u = exec->uniforms_u;
		uniform_data_u = (texture_handles_u +
				  validated_shader->num_texture_samples);

		memcpy(exec->uniforms_v, uniform_data_u,
		       validated_shader->uniforms_size);

		for (tex = 0;
		     tex < validated_shader->num_texture_samples;
		     tex++) {
			if (!reloc_tex(exec,
				       uniform_data_u,
				       &validated_shader->texture_samples[tex],
				       texture_handles_u[tex],
				       i == 2)) {
				return -EINVAL;
			}
		}

		/* Fill in the uniform slots that need this shader's
		 * start-of-uniforms address (used for resetting the uniform
		 * stream in the presence of control flow).
		 */
		for (uni = 0;
		     uni < validated_shader->num_uniform_addr_offsets;
		     uni++) {
			uint32_t o = validated_shader->uniform_addr_offsets[uni];
			((uint32_t *)exec->uniforms_v)[o] = exec->uniforms_p;
		}

		*(uint32_t *)(pkt_v + o + 4) = exec->uniforms_p;

		exec->uniforms_u += validated_shader->uniforms_src_size;
		exec->uniforms_v += validated_shader->uniforms_size;
		exec->uniforms_p += validated_shader->uniforms_size;
	}

	for (i = 0; i < nr_attributes; i++) {
		struct drm_gem_cma_object *vbo =
			bo[ARRAY_SIZE(shader_reloc_offsets) + i];
		uint32_t o = 36 + i * 8;
		uint32_t offset = *(uint32_t *)(pkt_u + o + 0);
		uint32_t attr_size = *(uint8_t *)(pkt_u + o + 4) + 1;
		uint32_t stride = *(uint8_t *)(pkt_u + o + 5);
		uint32_t max_index;

		exec->bin_dep_seqno = max(exec->bin_dep_seqno,
					  to_vc4_bo(&vbo->base)->write_seqno);

		if (state->addr & 0x8)
			stride |= (*(uint32_t *)(pkt_u + 100 + i * 4)) & ~0xff;

		if (vbo->base.size < offset ||
		    vbo->base.size - offset < attr_size) {
			DRM_DEBUG("BO offset overflow (%d + %d > %zu)\n",
				  offset, attr_size, vbo->base.size);
			return -EINVAL;
		}

		if (stride != 0) {
			max_index = ((vbo->base.size - offset - attr_size) /
				     stride);
			if (state->max_index > max_index) {
				DRM_DEBUG("primitives use index %d out of "
					  "supplied %d\n",
					  state->max_index, max_index);
				return -EINVAL;
			}
		}

		*(uint32_t *)(pkt_v + o) = vbo->paddr + offset;
	}

	return 0;
}

int
vc4_validate_shader_recs(struct drm_device *dev,
			 struct vc4_exec_info *exec)
{
	struct vc4_dev *vc4 = to_vc4_dev(dev);
	uint32_t i;
	int ret = 0;

	if (WARN_ON_ONCE(vc4->is_vc5))
		return -ENODEV;

	for (i = 0; i < exec->shader_state_count; i++) {
		ret = validate_gl_shader_rec(dev, exec, &exec->shader_state[i]);
		if (ret)
			return ret;
	}

	return ret;
}