summaryrefslogtreecommitdiff
path: root/src/btree/bt_vrfy_dsk.c
blob: d6c300d200a11fb799dcd1559856b739637c5c6c (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
/*-
 * Copyright (c) 2014-2015 MongoDB, Inc.
 * Copyright (c) 2008-2014 WiredTiger, Inc.
 *	All rights reserved.
 *
 * See the file LICENSE for redistribution information.
 */

#include "wt_internal.h"

static int __err_cell_corrupted(WT_SESSION_IMPL *, uint32_t, const char *);
static int __err_cell_type(
	WT_SESSION_IMPL *, uint32_t, const char *, uint8_t, uint8_t);
static int __err_eof(WT_SESSION_IMPL *, uint32_t, const char *);
static int __verify_dsk_chunk(
	WT_SESSION_IMPL *, const char *, const WT_PAGE_HEADER *, uint32_t);
static int __verify_dsk_col_fix(
	WT_SESSION_IMPL *, const char *, const WT_PAGE_HEADER *);
static int __verify_dsk_col_int(
	WT_SESSION_IMPL *, const char *, const WT_PAGE_HEADER *);
static int __verify_dsk_col_var(
	WT_SESSION_IMPL *, const char *, const WT_PAGE_HEADER *);
static int __verify_dsk_memsize(
	WT_SESSION_IMPL *, const char *, const WT_PAGE_HEADER *, WT_CELL *);
static int __verify_dsk_row(
	WT_SESSION_IMPL *, const char *, const WT_PAGE_HEADER *);

#define	WT_ERR_VRFY(session, ...) do {					\
	if (!(F_ISSET(session, WT_SESSION_SALVAGE_CORRUPT_OK)))		\
		__wt_errx(session, __VA_ARGS__);			\
	goto err;							\
} while (0)

#define	WT_RET_VRFY(session, ...) do {					\
	if (!(F_ISSET(session, WT_SESSION_SALVAGE_CORRUPT_OK)))		\
		__wt_errx(session, __VA_ARGS__);			\
	return (WT_ERROR);						\
} while (0)

/*
 * __wt_verify_dsk_image --
 *	Verify a single block as read from disk.
 */
int
__wt_verify_dsk_image(WT_SESSION_IMPL *session,
    const char *addr, const WT_PAGE_HEADER *dsk, size_t size, int empty_page_ok)
{
	const uint8_t *p, *end;
	u_int i;
	uint8_t flags;

	/* Check the page type. */
	switch (dsk->type) {
	case WT_PAGE_BLOCK_MANAGER:
	case WT_PAGE_COL_FIX:
	case WT_PAGE_COL_INT:
	case WT_PAGE_COL_VAR:
	case WT_PAGE_OVFL:
	case WT_PAGE_ROW_INT:
	case WT_PAGE_ROW_LEAF:
		break;
	case WT_PAGE_INVALID:
	default:
		WT_RET_VRFY(session,
		    "page at %s has an invalid type of %" PRIu32,
		    addr, dsk->type);
	}

	/* Check the page record number. */
	switch (dsk->type) {
	case WT_PAGE_COL_FIX:
	case WT_PAGE_COL_INT:
	case WT_PAGE_COL_VAR:
		if (dsk->recno != 0)
			break;
		WT_RET_VRFY(session,
		    "%s page at %s has a record number of zero",
		    __wt_page_type_string(dsk->type), addr);
	case WT_PAGE_BLOCK_MANAGER:
	case WT_PAGE_OVFL:
	case WT_PAGE_ROW_INT:
	case WT_PAGE_ROW_LEAF:
		if (dsk->recno == 0)
			break;
		WT_RET_VRFY(session,
		    "%s page at %s has a non-zero record number",
		    __wt_page_type_string(dsk->type), addr);
	}

	/* Check the page flags. */
	flags = dsk->flags;
	if (LF_ISSET(WT_PAGE_COMPRESSED))
		LF_CLR(WT_PAGE_COMPRESSED);
	if (dsk->type == WT_PAGE_ROW_LEAF) {
		if (LF_ISSET(WT_PAGE_EMPTY_V_ALL) &&
		    LF_ISSET(WT_PAGE_EMPTY_V_NONE))
			WT_RET_VRFY(session,
			    "page at %s has invalid flags combination: 0x%"
			    PRIx8,
			    addr, dsk->flags);
		if (LF_ISSET(WT_PAGE_EMPTY_V_ALL))
			LF_CLR(WT_PAGE_EMPTY_V_ALL);
		if (LF_ISSET(WT_PAGE_EMPTY_V_NONE))
			LF_CLR(WT_PAGE_EMPTY_V_NONE);
	}
	if (flags != 0)
		WT_RET_VRFY(session,
		    "page at %s has invalid flags set: 0x%" PRIx8,
		    addr, flags);

	/* Unused bytes */
	for (p = dsk->unused, i = sizeof(dsk->unused); i > 0; --i)
		if (*p != '\0')
			WT_RET_VRFY(session,
			    "page at %s has non-zero unused page header bytes",
			    addr);

	/*
	 * Any bytes after the data chunk should be nul bytes; ignore if the
	 * size is 0, that allows easy checking of disk images where we don't
	 * have the size.
	 */
	if (size != 0) {
		p = (uint8_t *)dsk + dsk->mem_size;
		end = (uint8_t *)dsk + size;
		for (; p < end; ++p)
			if (*p != '\0')
				WT_RET_VRFY(session,
				    "%s page at %s has non-zero trailing bytes",
				    __wt_page_type_string(dsk->type), addr);
	}

	/* Check for empty pages, then verify the items on the page. */
	switch (dsk->type) {
	case WT_PAGE_COL_INT:
	case WT_PAGE_COL_FIX:
	case WT_PAGE_COL_VAR:
	case WT_PAGE_ROW_INT:
	case WT_PAGE_ROW_LEAF:
		if (!empty_page_ok && dsk->u.entries == 0)
			WT_RET_VRFY(session, "%s page at %s has no entries",
			    __wt_page_type_string(dsk->type), addr);
		break;
	case WT_PAGE_BLOCK_MANAGER:
	case WT_PAGE_OVFL:
		if (dsk->u.datalen == 0)
			WT_RET_VRFY(session, "%s page at %s has no data",
			    __wt_page_type_string(dsk->type), addr);
		break;
	}
	switch (dsk->type) {
	case WT_PAGE_COL_INT:
		return (__verify_dsk_col_int(session, addr, dsk));
	case WT_PAGE_COL_FIX:
		return (__verify_dsk_col_fix(session, addr, dsk));
	case WT_PAGE_COL_VAR:
		return (__verify_dsk_col_var(session, addr, dsk));
	case WT_PAGE_ROW_INT:
	case WT_PAGE_ROW_LEAF:
		return (__verify_dsk_row(session, addr, dsk));
	case WT_PAGE_BLOCK_MANAGER:
	case WT_PAGE_OVFL:
		return (__verify_dsk_chunk(session, addr, dsk, dsk->u.datalen));
	WT_ILLEGAL_VALUE(session);
	}
	/* NOTREACHED */
}

/*
 * __wt_verify_dsk --
 *	Verify a single Btree page as read from disk.
 */
int
__wt_verify_dsk(WT_SESSION_IMPL *session, const char *addr, WT_ITEM *buf)
{
	return (__wt_verify_dsk_image(session, addr, buf->data, buf->size, 0));
}

/*
 * __verify_dsk_row --
 *	Walk a WT_PAGE_ROW_INT or WT_PAGE_ROW_LEAF disk page and verify it.
 */
static int
__verify_dsk_row(
    WT_SESSION_IMPL *session, const char *addr, const WT_PAGE_HEADER *dsk)
{
	WT_BM *bm;
	WT_BTREE *btree;
	WT_CELL *cell;
	WT_CELL_UNPACK *unpack, _unpack;
	WT_DECL_ITEM(current);
	WT_DECL_ITEM(last_ovfl);
	WT_DECL_ITEM(last_pfx);
	WT_DECL_RET;
	WT_ITEM *last;
	enum { FIRST, WAS_KEY, WAS_VALUE } last_cell_type;
	void *huffman;
	uint32_t cell_num, cell_type, i, key_cnt, prefix;
	uint8_t *end;
	int cmp;

	btree = S2BT(session);
	bm = btree->bm;
	unpack = &_unpack;
	huffman = dsk->type == WT_PAGE_ROW_INT ? NULL : btree->huffman_key;

	WT_ERR(__wt_scr_alloc(session, 0, &current));
	WT_ERR(__wt_scr_alloc(session, 0, &last_pfx));
	WT_ERR(__wt_scr_alloc(session, 0, &last_ovfl));
	last = last_ovfl;

	end = (uint8_t *)dsk + dsk->mem_size;

	last_cell_type = FIRST;
	cell_num = 0;
	key_cnt = 0;
	WT_CELL_FOREACH(btree, dsk, cell, unpack, i) {
		++cell_num;

		/* Carefully unpack the cell. */
		if (__wt_cell_unpack_safe(cell, unpack, end) != 0) {
			ret = __err_cell_corrupted(session, cell_num, addr);
			goto err;
		}

		/* Check the raw and collapsed cell types. */
		WT_ERR(__err_cell_type(
		    session, cell_num, addr, unpack->raw, dsk->type));
		WT_ERR(__err_cell_type(
		    session, cell_num, addr, unpack->type, dsk->type));
		cell_type = unpack->type;

		/*
		 * Check ordering relationships between the WT_CELL entries.
		 * For row-store internal pages, check for:
		 *	two values in a row,
		 *	two keys in a row,
		 *	a value as the first cell on a page.
		 * For row-store leaf pages, check for:
		 *	two values in a row,
		 *	a value as the first cell on a page.
		 */
		switch (cell_type) {
		case WT_CELL_KEY:
		case WT_CELL_KEY_OVFL:
			++key_cnt;
			switch (last_cell_type) {
			case FIRST:
			case WAS_VALUE:
				break;
			case WAS_KEY:
				if (dsk->type == WT_PAGE_ROW_LEAF)
					break;
				WT_ERR_VRFY(session,
				    "cell %" PRIu32 " on page at %s is the "
				    "first of two adjacent keys",
				    cell_num - 1, addr);
			}
			last_cell_type = WAS_KEY;
			break;
		case WT_CELL_ADDR_DEL:
		case WT_CELL_ADDR_INT:
		case WT_CELL_ADDR_LEAF:
		case WT_CELL_ADDR_LEAF_NO:
		case WT_CELL_VALUE:
		case WT_CELL_VALUE_OVFL:
			switch (last_cell_type) {
			case FIRST:
				WT_ERR_VRFY(session,
				    "page at %s begins with a value", addr);
			case WAS_KEY:
				break;
			case WAS_VALUE:
				WT_ERR_VRFY(session,
				    "cell %" PRIu32 " on page at %s is the "
				    "first of two adjacent values",
				    cell_num - 1, addr);
			}
			last_cell_type = WAS_VALUE;
			break;
		}

		/* Check if any referenced item has a valid address. */
		switch (cell_type) {
		case WT_CELL_ADDR_DEL:
		case WT_CELL_ADDR_INT:
		case WT_CELL_ADDR_LEAF:
		case WT_CELL_ADDR_LEAF_NO:
		case WT_CELL_KEY_OVFL:
		case WT_CELL_VALUE_OVFL:
			if (!bm->addr_valid(bm,
			    session, unpack->data, unpack->size))
				goto eof;
			break;
		}

		/*
		 * Remaining checks are for key order and prefix compression.
		 * If this cell isn't a key, we're done, move to the next cell.
		 * If this cell is an overflow item, instantiate the key and
		 * compare it with the last key.   Otherwise, we have to deal
		 * with prefix compression.
		 */
		switch (cell_type) {
		case WT_CELL_KEY:
			break;
		case WT_CELL_KEY_OVFL:
			WT_ERR(__wt_dsk_cell_data_ref(
			    session, dsk->type, unpack, current));
			goto key_compare;
		default:
			/* Not a key -- continue with the next cell. */
			continue;
		}

		/*
		 * Prefix compression checks.
		 *
		 * Confirm the first non-overflow key on a page has a zero
		 * prefix compression count.
		 */
		prefix = unpack->prefix;
		if (last_pfx->size == 0 && prefix != 0)
			WT_ERR_VRFY(session,
			    "the %" PRIu32 " key on page at %s is the first "
			    "non-overflow key on the page and has a non-zero "
			    "prefix compression value",
			    cell_num, addr);

		/* Confirm the prefix compression count is possible. */
		if (cell_num > 1 && prefix > last->size)
			WT_ERR_VRFY(session,
			    "key %" PRIu32 " on page at %s has a prefix "
			    "compression count of %" PRIu32 ", larger than "
			    "the length of the previous key, %" WT_SIZET_FMT,
			    cell_num, addr, prefix, last->size);

		/*
		 * If Huffman decoding required, unpack the cell to build the
		 * key, then resolve the prefix.  Else, we can do it faster
		 * internally because we don't have to shuffle memory around as
		 * much.
		 */
		if (huffman != NULL) {
			WT_ERR(__wt_dsk_cell_data_ref(
			    session, dsk->type, unpack, current));

			/*
			 * If there's a prefix, make sure there's enough buffer
			 * space, then shift the decoded data past the prefix
			 * and copy the prefix into place.  Take care with the
			 * pointers: current->data may be pointing inside the
			 * buffer.
			 */
			if (prefix != 0) {
				WT_ERR(__wt_buf_grow(
				    session, current, prefix + current->size));
				memmove((uint8_t *)current->mem + prefix,
				    current->data, current->size);
				memcpy(current->mem, last->data, prefix);
				current->data = current->mem;
				current->size += prefix;
			}
		} else {
			/*
			 * Get the cell's data/length and make sure we have
			 * enough buffer space.
			 */
			WT_ERR(__wt_buf_init(
			    session, current, prefix + unpack->size));

			/* Copy the prefix then the data into place. */
			if (prefix != 0)
				memcpy(current->mem, last->data, prefix);
			memcpy((uint8_t *)current->mem + prefix, unpack->data,
			    unpack->size);
			current->size = prefix + unpack->size;
		}

key_compare:	/*
		 * Compare the current key against the last key.
		 *
		 * Be careful about the 0th key on internal pages: we only store
		 * the first byte and custom collators may not be able to handle
		 * truncated keys.
		 */
		if ((dsk->type == WT_PAGE_ROW_INT && cell_num > 3) ||
		    (dsk->type != WT_PAGE_ROW_INT && cell_num > 1)) {
			WT_ERR(__wt_compare(
			    session, btree->collator, last, current, &cmp));
			if (cmp >= 0)
				WT_ERR_VRFY(session,
				    "the %" PRIu32 " and %" PRIu32 " keys on "
				    "page at %s are incorrectly sorted",
				    cell_num - 2, cell_num, addr);
		}

		/*
		 * Swap the buffers: last always references the last key entry,
		 * last_pfx and last_ovfl reference the last prefix-compressed
		 * and last overflow key entries.  Current gets pointed to the
		 * buffer we're not using this time around, which is where the
		 * next key goes.
		 */
		last = current;
		if (cell_type == WT_CELL_KEY) {
			current = last_pfx;
			last_pfx = last;
		} else {
			current = last_ovfl;
			last_ovfl = last;
		}
		WT_ASSERT(session, last != current);
	}
	WT_ERR(__verify_dsk_memsize(session, addr, dsk, cell));

	/*
	 * On row-store internal pages, and on row-store leaf pages, where the
	 * "no empty values" flag is set, the key count should be equal to half
	 * the number of physical entries.  On row-store leaf pages where the
	 * "all empty values" flag is set, the key count should be equal to the
	 * number of physical entries.
	 */
	if (dsk->type == WT_PAGE_ROW_INT && key_cnt * 2 != dsk->u.entries)
		WT_ERR_VRFY(session,
		    "%s page at %s has a key count of %" PRIu32 " and a "
		    "physical entry count of %" PRIu32,
		    __wt_page_type_string(dsk->type),
		    addr, key_cnt, dsk->u.entries);
	if (dsk->type == WT_PAGE_ROW_LEAF &&
	    F_ISSET(dsk, WT_PAGE_EMPTY_V_ALL) &&
	    key_cnt != dsk->u.entries)
		WT_ERR_VRFY(session,
		    "%s page at %s with the 'all empty values' flag set has a "
		    "key count of %" PRIu32 " and a physical entry count of %"
		    PRIu32,
		    __wt_page_type_string(dsk->type),
		    addr, key_cnt, dsk->u.entries);
	if (dsk->type == WT_PAGE_ROW_LEAF &&
	    F_ISSET(dsk, WT_PAGE_EMPTY_V_NONE) &&
	    key_cnt * 2 != dsk->u.entries)
		WT_ERR_VRFY(session,
		    "%s page at %s with the 'no empty values' flag set has a "
		    "key count of %" PRIu32 " and a physical entry count of %"
		    PRIu32,
		    __wt_page_type_string(dsk->type),
		    addr, key_cnt, dsk->u.entries);

	if (0) {
eof:		ret = __err_eof(session, cell_num, addr);
	}

	if (0) {
err:		if (ret == 0)
			ret = WT_ERROR;
	}
	__wt_scr_free(session, &current);
	__wt_scr_free(session, &last_pfx);
	__wt_scr_free(session, &last_ovfl);
	return (ret);
}

/*
 * __verify_dsk_col_int --
 *	Walk a WT_PAGE_COL_INT disk page and verify it.
 */
static int
__verify_dsk_col_int(
    WT_SESSION_IMPL *session, const char *addr, const WT_PAGE_HEADER *dsk)
{
	WT_BM *bm;
	WT_BTREE *btree;
	WT_CELL *cell;
	WT_CELL_UNPACK *unpack, _unpack;
	uint32_t cell_num, i;
	uint8_t *end;

	btree = S2BT(session);
	bm = btree->bm;
	unpack = &_unpack;
	end = (uint8_t *)dsk + dsk->mem_size;

	cell_num = 0;
	WT_CELL_FOREACH(btree, dsk, cell, unpack, i) {
		++cell_num;

		/* Carefully unpack the cell. */
		if (__wt_cell_unpack_safe(cell, unpack, end) != 0)
			return (__err_cell_corrupted(session, cell_num, addr));

		/* Check the raw and collapsed cell types. */
		WT_RET(__err_cell_type(
		    session, cell_num, addr, unpack->raw, dsk->type));
		WT_RET(__err_cell_type(
		    session, cell_num, addr, unpack->type, dsk->type));

		/* Check if any referenced item is entirely in the file. */
		if (!bm->addr_valid(bm, session, unpack->data, unpack->size))
			return (__err_eof(session, cell_num, addr));
	}
	WT_RET(__verify_dsk_memsize(session, addr, dsk, cell));

	return (0);
}

/*
 * __verify_dsk_col_fix --
 *	Walk a WT_PAGE_COL_FIX disk page and verify it.
 */
static int
__verify_dsk_col_fix(
    WT_SESSION_IMPL *session, const char *addr, const WT_PAGE_HEADER *dsk)
{
	WT_BTREE *btree;
	uint32_t datalen;

	btree = S2BT(session);

	datalen = __bitstr_size(btree->bitcnt * dsk->u.entries);
	return (__verify_dsk_chunk(session, addr, dsk, datalen));
}

/*
 * __verify_dsk_col_var --
 *	Walk a WT_PAGE_COL_VAR disk page and verify it.
 */
static int
__verify_dsk_col_var(
    WT_SESSION_IMPL *session, const char *addr, const WT_PAGE_HEADER *dsk)
{
	WT_BM *bm;
	WT_BTREE *btree;
	WT_CELL *cell;
	WT_CELL_UNPACK *unpack, _unpack;
	size_t last_size;
	uint32_t cell_num, cell_type, i;
	int last_deleted;
	const uint8_t *last_data;
	uint8_t *end;

	btree = S2BT(session);
	bm = btree->bm;
	unpack = &_unpack;
	end = (uint8_t *)dsk + dsk->mem_size;

	last_data = NULL;
	last_size = 0;
	last_deleted = 0;

	cell_num = 0;
	WT_CELL_FOREACH(btree, dsk, cell, unpack, i) {
		++cell_num;

		/* Carefully unpack the cell. */
		if (__wt_cell_unpack_safe(cell, unpack, end) != 0)
			return (__err_cell_corrupted(session, cell_num, addr));

		/* Check the raw and collapsed cell types. */
		WT_RET(__err_cell_type(
		    session, cell_num, addr, unpack->raw, dsk->type));
		WT_RET(__err_cell_type(
		    session, cell_num, addr, unpack->type, dsk->type));
		cell_type = unpack->type;

		/* Check if any referenced item is entirely in the file. */
		if (cell_type == WT_CELL_VALUE_OVFL &&
		    !bm->addr_valid(bm, session, unpack->data, unpack->size))
			return (__err_eof(session, cell_num, addr));

		/*
		 * Compare the last two items and see if reconciliation missed
		 * a chance for RLE encoding.  We don't have to care about data
		 * encoding or anything else, a byte comparison is enough.
		 */
		if (last_deleted == 1) {
			if (cell_type == WT_CELL_DEL)
				goto match_err;
		} else
			if (cell_type == WT_CELL_VALUE &&
			    last_data != NULL &&
			    last_size == unpack->size &&
			    memcmp(last_data, unpack->data, last_size) == 0)
match_err:			WT_RET_VRFY(session,
				    "data entries %" PRIu32 " and %" PRIu32
				    " on page at %s are identical and should "
				    "have been run-length encoded",
				    cell_num - 1, cell_num, addr);

		switch (cell_type) {
		case WT_CELL_DEL:
			last_deleted = 1;
			last_data = NULL;
			break;
		case WT_CELL_VALUE_OVFL:
			last_deleted = 0;
			last_data = NULL;
			break;
		case WT_CELL_VALUE:
			last_deleted = 0;
			last_data = unpack->data;
			last_size = unpack->size;
			break;
		}
	}
	WT_RET(__verify_dsk_memsize(session, addr, dsk, cell));

	return (0);
}

/*
 * __verify_dsk_memsize --
 *	Verify the last cell on the page matches the page's memory size.
 */
static int
__verify_dsk_memsize(WT_SESSION_IMPL *session,
    const char *addr, const WT_PAGE_HEADER *dsk, WT_CELL *cell)
{
	size_t len;

	/*
	 * We use the fact that cells exactly fill a page to detect the case of
	 * a row-store leaf page where the last cell is a key (that is, there's
	 * no subsequent value cell).  Check for any page type containing cells.
	 */
	len = WT_PTRDIFF((uint8_t *)dsk + dsk->mem_size, cell);
	if (len == 0)
		return (0);
	WT_RET_VRFY(session,
	    "%s page at %s has %" WT_SIZET_FMT " unexpected bytes of data "
	    "after the last cell",
	    __wt_page_type_string(dsk->type), addr, len);
}

/*
 * __verify_dsk_chunk --
 *	Verify a Chunk O' Data on a Btree page.
 */
static int
__verify_dsk_chunk(WT_SESSION_IMPL *session,
    const char *addr, const WT_PAGE_HEADER *dsk, uint32_t datalen)
{
	WT_BTREE *btree;
	uint8_t *p, *end;

	btree = S2BT(session);
	end = (uint8_t *)dsk + dsk->mem_size;

	/*
	 * Fixed-length column-store and overflow pages are simple chunks of
	 * data. Verify the data doesn't overflow the end of the page.
	 */
	p = WT_PAGE_HEADER_BYTE(btree, dsk);
	if (p + datalen > end)
		WT_RET_VRFY(session,
		    "data on page at %s extends past the end of the page",
		    addr);

	/* Any bytes after the data chunk should be nul bytes. */
	for (p += datalen; p < end; ++p)
		if (*p != '\0')
			WT_RET_VRFY(session,
			    "%s page at %s has non-zero trailing bytes",
			    __wt_page_type_string(dsk->type), addr);

	return (0);
}

/*
 * __err_cell_corrupted --
 *	Generic corrupted cell, we couldn't read it.
 */
static int
__err_cell_corrupted(
    WT_SESSION_IMPL *session, uint32_t entry_num, const char *addr)
{
	WT_RET_VRFY(session,
	    "item %" PRIu32 " on page at %s is a corrupted cell",
	    entry_num, addr);
}

/*
 * __err_cell_type --
 *	Generic illegal cell type for a particular page type error.
 */
static int
__err_cell_type(WT_SESSION_IMPL *session,
    uint32_t entry_num, const char *addr, uint8_t cell_type, uint8_t dsk_type)
{
	switch (cell_type) {
	case WT_CELL_ADDR_DEL:
	case WT_CELL_ADDR_INT:
	case WT_CELL_ADDR_LEAF:
	case WT_CELL_ADDR_LEAF_NO:
		if (dsk_type == WT_PAGE_COL_INT ||
		    dsk_type == WT_PAGE_ROW_INT)
			return (0);
		break;
	case WT_CELL_DEL:
		if (dsk_type == WT_PAGE_COL_VAR)
			return (0);
		break;
	case WT_CELL_KEY:
	case WT_CELL_KEY_OVFL:
	case WT_CELL_KEY_SHORT:
		if (dsk_type == WT_PAGE_ROW_INT ||
		    dsk_type == WT_PAGE_ROW_LEAF)
			return (0);
		break;
	case WT_CELL_KEY_PFX:
	case WT_CELL_KEY_SHORT_PFX:
		if (dsk_type == WT_PAGE_ROW_LEAF)
			return (0);
		break;
	case WT_CELL_KEY_OVFL_RM:
	case WT_CELL_VALUE_OVFL_RM:
		/*
		 * Removed overflow cells are in-memory only, it's an error to
		 * ever see one on a disk page.
		 */
		break;
	case WT_CELL_VALUE:
	case WT_CELL_VALUE_COPY:
	case WT_CELL_VALUE_OVFL:
	case WT_CELL_VALUE_SHORT:
		if (dsk_type == WT_PAGE_COL_VAR ||
		    dsk_type == WT_PAGE_ROW_LEAF)
			return (0);
		break;
	default:
		break;
	}

	WT_RET_VRFY(session,
	    "illegal cell and page type combination: cell %" PRIu32
	    " on page at %s is a %s cell on a %s page",
	    entry_num, addr,
	    __wt_cell_type_string(cell_type), __wt_page_type_string(dsk_type));
}

/*
 * __err_eof --
 *	Generic item references non-existent file pages error.
 */
static int
__err_eof(WT_SESSION_IMPL *session, uint32_t entry_num, const char *addr)
{
	WT_RET_VRFY(session,
	    "off-page item %" PRIu32
	    " on page at %s references non-existent file pages",
	    entry_num, addr);
}