summaryrefslogtreecommitdiff
path: root/storage/innobase/btr/btr0scrub.cc
blob: 22e997b60aaba9aa9e7ad2b9cb62d3838418f8c8 (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
// Copyright (c) 2014, Google Inc.
// Copyright (c) 2017, MariaDB Corporation.

/**************************************************//**
@file btr/btr0scrub.cc
Scrubbing of btree pages

*******************************************************/

#include "btr0btr.h"
#include "btr0cur.h"
#include "btr0scrub.h"
#include "ibuf0ibuf.h"
#include "fsp0fsp.h"
#include "dict0dict.h"
#include "mtr0mtr.h"

/* used when trying to acquire dict-lock */
UNIV_INTERN bool fil_crypt_is_closing(ulint space);

/**
* scrub data at delete time (e.g purge thread)
*/
my_bool srv_immediate_scrub_data_uncompressed = false;

/**
* background scrub uncompressed data
*
* if srv_immediate_scrub_data_uncompressed is enabled
* this is only needed to handle "old" data
*/
my_bool srv_background_scrub_data_uncompressed = false;

/**
* backgrounds scrub compressed data
*
* reorganize compressed page for scrubbing
* (only way to scrub compressed data)
*/
my_bool srv_background_scrub_data_compressed = false;

/* check spaces once per hour */
UNIV_INTERN uint srv_background_scrub_data_check_interval = (60 * 60);

/* default to scrub spaces that hasn't been scrubbed in a week */
UNIV_INTERN uint srv_background_scrub_data_interval = (7 * 24 * 60 * 60);

/**
* statistics for scrubbing by background threads
*/
static btr_scrub_stat_t scrub_stat;
static ib_mutex_t scrub_stat_mutex;
#ifdef UNIV_PFS_MUTEX
UNIV_INTERN mysql_pfs_key_t scrub_stat_mutex_key;
#endif

#ifdef UNIV_DEBUG
/**
* srv_scrub_force_testing
*
* - force scrubbing using background threads even for uncompressed tables
* - force pessimistic scrubbing (page split) even if not needed
*   (see test_pessimistic_scrub_pct)
*/
my_bool srv_scrub_force_testing = true;

/**
* Force pessimistic scrubbing in 50% of the cases (UNIV_DEBUG only)
*/
static int test_pessimistic_scrub_pct = 50;

#endif
static uint scrub_compression_level = page_zip_level;

/**************************************************************//**
Log a scrubbing failure */
static
void
log_scrub_failure(
/*===============*/
	dict_index_t* index,     /*!< in: index */
	btr_scrub_t* scrub_data, /*!< in: data to store statistics on */
	buf_block_t* block,	 /*!< in: block */
	dberr_t err)             /*!< in: error */
{
	const char* reason = "unknown";
	switch(err) {
	case DB_UNDERFLOW:
		reason = "too few records on page";
		scrub_data->scrub_stat.page_split_failures_underflow++;
		break;
	case DB_INDEX_CORRUPT:
		reason = "unable to find index!";
		scrub_data->scrub_stat.page_split_failures_missing_index++;
		break;
	case DB_OUT_OF_FILE_SPACE:
		reason = "out of filespace";
		scrub_data->scrub_stat.page_split_failures_out_of_filespace++;
		break;
	default:
		ut_ad(0);
		reason = "unknown";
		scrub_data->scrub_stat.page_split_failures_unknown++;
	}

	ib::warn() << "Failed to scrub index " << index->name
		   << " of table " << index->table->name
		   << " page " << block->page.id << ": " << reason;
}

/****************************************************************
Lock dict mutexes */
static
bool
btr_scrub_lock_dict_func(ulint space_id, bool lock_to_close_table,
			 const char * file, uint line)
{
	time_t start = time(0);
	time_t last = start;

	/* FIXME: this is not the proper way of doing things. The
	dict_sys->mutex should not be held by any thread for longer
	than a few microseconds. It must not be held during I/O,
	for example. So, what is the purpose for this busy-waiting?
	This function should be rewritten as part of MDEV-8139:
	Fix scrubbing tests. */

	while (mutex_enter_nowait(&(dict_sys->mutex))) {
		/* if we lock to close a table, we wait forever
		* if we don't lock to close a table, we check if space
		* is closing, and then instead give up
		*/
		if (lock_to_close_table) {
		} else if (fil_space_t* space = fil_space_acquire(space_id)) {
			bool stopping = space->is_stopping();
			space->release();
			if (stopping) {
				return false;
			}
		} else {
			return false;
		}

		os_thread_sleep(250000);

		time_t now = time(0);

		if (now >= last + 30) {
			fprintf(stderr,
				"WARNING: %s:%u waited %ld seconds for"
				" dict_sys lock, space: " ULINTPF
				" lock_to_close_table: %d\n",
				file, line, long(now - start), space_id,
				lock_to_close_table);

			last = now;
		}
	}

	ut_ad(mutex_own(&dict_sys->mutex));
	return true;
}

#define btr_scrub_lock_dict(space, lock_to_close_table)			\
	btr_scrub_lock_dict_func(space, lock_to_close_table, __FILE__, __LINE__)

/****************************************************************
Unlock dict mutexes */
static
void
btr_scrub_unlock_dict()
{
	dict_mutex_exit_for_mysql();
}

/****************************************************************
Release reference to table
*/
static
void
btr_scrub_table_close(
/*==================*/
	dict_table_t* table)  /*!< in: table */
{
	bool dict_locked = true;
	bool try_drop = false;
	table->stats_bg_flag &= ~BG_SCRUB_IN_PROGRESS;
	dict_table_close(table, dict_locked, try_drop);
}

/****************************************************************
Release reference to table
*/
static
void
btr_scrub_table_close_for_thread(
	btr_scrub_t *scrub_data)
{
	if (scrub_data->current_table == NULL) {
		return;
	}

	if (fil_space_t* space = fil_space_acquire(scrub_data->space)) {
		/* If tablespace is not marked as stopping perform
		the actual close. */
		if (!space->is_stopping()) {
			mutex_enter(&dict_sys->mutex);
			/* perform the actual closing */
			btr_scrub_table_close(scrub_data->current_table);
			mutex_exit(&dict_sys->mutex);
		}
		space->release();
	}

	scrub_data->current_table = NULL;
	scrub_data->current_index = NULL;
}

/**************************************************************//**
Check if scrubbing is turned ON or OFF */
static
bool
check_scrub_setting(
/*=====================*/
	btr_scrub_t*	scrub_data) /*!< in: scrub data  */
{
	if (scrub_data->compressed)
		return srv_background_scrub_data_compressed;
	else
		return srv_background_scrub_data_uncompressed;
}

#define IBUF_INDEX_ID (DICT_IBUF_ID_MIN + IBUF_SPACE_ID)

/**************************************************************//**
Check if a page needs scrubbing */
UNIV_INTERN
int
btr_page_needs_scrubbing(
/*=====================*/
	btr_scrub_t*	scrub_data, /*!< in: scrub data  */
	buf_block_t*	block,	    /*!< in: block to check, latched */
	btr_scrub_page_allocation_status_t allocated)  /*!< in: is block known
						       to be allocated */
{
	/**
	* Check if scrubbing has been turned OFF.
	*
	* at start of space, we check if scrubbing is ON or OFF
	* here we only check if scrubbing is turned OFF.
	*
	* Motivation is that it's only valueable to have a full table (space)
	* scrubbed.
	*/
	if (!check_scrub_setting(scrub_data)) {
		bool before_value = scrub_data->scrubbing;
		scrub_data->scrubbing = false;

		if (before_value == true) {
			/* we toggle scrubbing from on to off */
			return BTR_SCRUB_TURNED_OFF;
		}
	}

	if (scrub_data->scrubbing == false) {
		return BTR_SCRUB_SKIP_PAGE_AND_CLOSE_TABLE;
	}

	const page_t*	page = buf_block_get_frame(block);

	if (allocated == BTR_SCRUB_PAGE_ALLOCATED) {
		if (fil_page_get_type(page) != FIL_PAGE_INDEX) {
			/* this function is called from fil-crypt-threads.
			* these threads iterate all pages of all tablespaces
			* and don't know about fil_page_type.
			* But scrubbing is only needed for index-pages. */

			/**
			* NOTE: scrubbing is also needed for UNDO pages,
			* but they are scrubbed at purge-time, since they are
			* uncompressed
			*/

			/* if encountering page type not needing scrubbing
			release reference to table object */
			return BTR_SCRUB_SKIP_PAGE_AND_CLOSE_TABLE;
		}

		if (!page_has_garbage(page)) {
			/* no garbage (from deleted/shrunken records) */
			return BTR_SCRUB_SKIP_PAGE_AND_CLOSE_TABLE;
		}

	} else if (allocated == BTR_SCRUB_PAGE_FREE ||
		   allocated == BTR_SCRUB_PAGE_ALLOCATION_UNKNOWN) {

		switch (fil_page_get_type(page)) {
		case FIL_PAGE_INDEX:
		case FIL_PAGE_TYPE_ZBLOB:
		case FIL_PAGE_TYPE_ZBLOB2:
			break;
		default:
			/**
			* If this is a dropped page, we also need to scrub
			* BLOB pages
			*/

			/* if encountering page type not needing scrubbing
			release reference to table object */
			return BTR_SCRUB_SKIP_PAGE_AND_CLOSE_TABLE;
		}
	}

	if (block->page.id.space() == TRX_SYS_SPACE
	    && btr_page_get_index_id(page) == IBUF_INDEX_ID) {
		/* skip ibuf */
		return BTR_SCRUB_SKIP_PAGE_AND_CLOSE_TABLE;
	}

	return BTR_SCRUB_PAGE;
}

/****************************************************************
Handle a skipped page
*/
UNIV_INTERN
void
btr_scrub_skip_page(
/*==================*/
	btr_scrub_t* scrub_data, /*!< in: data with scrub state */
	int needs_scrubbing)     /*!< in: return code from
				 btr_page_needs_scrubbing */
{
	switch(needs_scrubbing) {
	case BTR_SCRUB_SKIP_PAGE:
		/* nothing todo */
		return;
	case BTR_SCRUB_SKIP_PAGE_AND_CLOSE_TABLE:
		btr_scrub_table_close_for_thread(scrub_data);
		return;
	case BTR_SCRUB_TURNED_OFF:
	case BTR_SCRUB_SKIP_PAGE_AND_COMPLETE_SPACE:
		btr_scrub_complete_space(scrub_data);
		return;
	}

	/* unknown value. should not happen */
	ut_a(0);
}

/****************************************************************
Try to scrub a page using btr_page_reorganize_low
return DB_SUCCESS on success or DB_OVERFLOW on failure */
static
dberr_t
btr_optimistic_scrub(
/*==================*/
	btr_scrub_t* scrub_data, /*!< in: data with scrub state */
	buf_block_t* block,      /*!< in: block to scrub */
	dict_index_t* index,     /*!< in: index */
	mtr_t* mtr)              /*!< in: mtr */
{
#ifdef UNIV_DEBUG
	if (srv_scrub_force_testing &&
	    page_get_n_recs(buf_block_get_frame(block)) > 2 &&
	    (rand() % 100) < test_pessimistic_scrub_pct) {

		log_scrub_failure(index, scrub_data, block, DB_OVERFLOW);
		return DB_OVERFLOW;
	}
#endif

	page_cur_t cur;
	page_cur_set_before_first(block, &cur);
	bool recovery = false;
	if (!btr_page_reorganize_low(recovery, scrub_compression_level,
				     &cur, index, mtr)) {
		return DB_OVERFLOW;
	}

	/* We play safe and reset the free bits */
	if (!dict_index_is_clust(index) &&
	    block != NULL) {
		buf_frame_t* frame = buf_block_get_frame(block);
		if (frame &&
		    page_is_leaf(frame)) {

			ibuf_reset_free_bits(block);
		}
	}

	scrub_data->scrub_stat.page_reorganizations++;

	return DB_SUCCESS;
}

/****************************************************************
Try to scrub a page by splitting it
return DB_SUCCESS on success
DB_UNDERFLOW if page has too few records
DB_OUT_OF_FILE_SPACE if we can't find space for split */
static
dberr_t
btr_pessimistic_scrub(
/*==================*/
	btr_scrub_t* scrub_data, /*!< in: data with scrub state */
	buf_block_t* block,      /*!< in: block to scrub */
	dict_index_t* index,     /*!< in: index */
	mtr_t* mtr)              /*!< in: mtr */
{
	page_t*	page = buf_block_get_frame(block);

	if (page_get_n_recs(page) < 2) {
		/**
		* There is no way we can split a page with < 2 records
		*/
		log_scrub_failure(index, scrub_data, block, DB_UNDERFLOW);
		return DB_UNDERFLOW;
	}

	/**
	* Splitting page needs new space, allocate it here
	* so that splitting won't fail due to this */
	ulint n_extents = 3;
	ulint n_reserved = 0;
	if (!fsp_reserve_free_extents(&n_reserved, index->table->space,
				      n_extents, FSP_NORMAL, mtr)) {
		log_scrub_failure(index, scrub_data, block,
				  DB_OUT_OF_FILE_SPACE);
		return DB_OUT_OF_FILE_SPACE;
	}

	/* read block variables */
	const ulint page_no =  mach_read_from_4(page + FIL_PAGE_OFFSET);
	const ulint left_page_no = mach_read_from_4(page + FIL_PAGE_PREV);
	const ulint right_page_no = mach_read_from_4(page + FIL_PAGE_NEXT);
	const page_size_t page_size(index->table->space->flags);

	/**
	* When splitting page, we need X-latches on left/right brothers
	* see e.g btr_cur_latch_leaves
	*/

	if (left_page_no != FIL_NULL) {
		/**
		* pages needs to be locked left-to-right, release block
		* and re-lock. We still have x-lock on index
		* so this should be safe
		*/
		mtr->release_block_at_savepoint(scrub_data->savepoint, block);

		buf_block_t* get_block __attribute__((unused)) = btr_block_get(
			page_id_t(index->table->space->id, left_page_no),
			page_size, RW_X_LATCH, index, mtr);

		/**
		* Refetch block and re-initialize page
		*/
		block = btr_block_get(
			page_id_t(index->table->space->id, page_no),
			page_size, RW_X_LATCH, index, mtr);

		page = buf_block_get_frame(block);

		/**
		* structure should be unchanged
		*/
		ut_a(left_page_no == btr_page_get_prev(page, mtr));
		ut_a(right_page_no == btr_page_get_next(page, mtr));
	}

	if (right_page_no != FIL_NULL) {
		buf_block_t* get_block __attribute__((unused))= btr_block_get(
			page_id_t(index->table->space->id, right_page_no),
			page_size, RW_X_LATCH, index, mtr);
	}

	/* arguments to btr_page_split_and_insert */
	mem_heap_t* heap = NULL;
	dtuple_t* entry = NULL;
	ulint* offsets = NULL;
	ulint n_ext = 0;
	ulint flags = BTR_MODIFY_TREE;

	/**
	* position a cursor on first record on page
	*/
	rec_t* rec = page_rec_get_next(page_get_infimum_rec(page));
	btr_cur_t cursor;
	btr_cur_position(index, rec, block, &cursor);

	/**
	* call split page with NULL as argument for entry to insert
	*/
	if (dict_index_get_page(index) == page_no) {
		/* The page is the root page
		* NOTE: ibuf_reset_free_bits is called inside
		* btr_root_raise_and_insert */
		rec = btr_root_raise_and_insert(
			flags, &cursor, &offsets, &heap, entry, n_ext, mtr);
	} else {
		/* We play safe and reset the free bits
		* NOTE: need to call this prior to btr_page_split_and_insert */
		if (!dict_index_is_clust(index) &&
		    block != NULL) {
			buf_frame_t* frame = buf_block_get_frame(block);
			if (frame &&
			    page_is_leaf(frame)) {

				ibuf_reset_free_bits(block);
			}
		}

		rec = btr_page_split_and_insert(
			flags, &cursor, &offsets, &heap, entry, n_ext, mtr);
	}

	if (heap) {
		mem_heap_free(heap);
	}

	index->table->space->release_free_extents(n_reserved);
	scrub_data->scrub_stat.page_splits++;
	return DB_SUCCESS;
}

/****************************************************************
Location index by id for a table
return index or NULL */
static
dict_index_t*
find_index(
/*========*/
	dict_table_t* table, /*!< in: table */
	index_id_t index_id) /*!< in: index id */
{
	if (table != NULL) {
		dict_index_t* index = dict_table_get_first_index(table);
		while (index != NULL) {
			if (index->id == index_id)
				return index;
			index = dict_table_get_next_index(index);
		}
	}

	return NULL;
}

/****************************************************************
Check if table should be scrubbed
*/
static
bool
btr_scrub_table_needs_scrubbing(
/*============================*/
	dict_table_t* table) /*!< in: table */
{
	if (table == NULL)
		return false;

	if (table->stats_bg_flag & BG_STAT_SHOULD_QUIT) {
		return false;
	}

	if (table->to_be_dropped) {
		return false;
	}

	if (!table->is_readable()) {
		return false;
	}

	return true;
}

/****************************************************************
Check if index should be scrubbed
*/
static
bool
btr_scrub_index_needs_scrubbing(
/*============================*/
	dict_index_t* index) /*!< in: index */
{
	if (index == NULL)
		return false;

	if (dict_index_is_ibuf(index)) {
		return false;
	}

	if (dict_index_is_online_ddl(index)) {
		return false;
	}

	return true;
}

/****************************************************************
Get table and index and store it on scrub_data
*/
static
void
btr_scrub_get_table_and_index(
/*=========================*/
	btr_scrub_t* scrub_data, /*!< in/out: scrub data */
	index_id_t index_id)     /*!< in: index id */
{
	/* first check if it's an index to current table */
	scrub_data->current_index = find_index(scrub_data->current_table,
					       index_id);

	if (scrub_data->current_index != NULL) {
		/* yes it was */
		return;
	}

	if (!btr_scrub_lock_dict(scrub_data->space, false)) {
		btr_scrub_complete_space(scrub_data);
		return;
	}

	/* close current table (if any) */
	if (scrub_data->current_table != NULL) {
		btr_scrub_table_close(scrub_data->current_table);
		scrub_data->current_table = NULL;
	}

	/* argument to dict_table_open_on_index_id */
	bool dict_locked = true;

	/* open table based on index_id */
	dict_table_t* table = dict_table_open_on_index_id(
		index_id,
		dict_locked);

	if (table != NULL) {
		/* mark table as being scrubbed */
		table->stats_bg_flag |= BG_SCRUB_IN_PROGRESS;

		if (!btr_scrub_table_needs_scrubbing(table)) {
			btr_scrub_table_close(table);
			btr_scrub_unlock_dict();
			return;
		}
	}

	btr_scrub_unlock_dict();
	scrub_data->current_table = table;
	scrub_data->current_index = find_index(table, index_id);
}

/****************************************************************
Handle free page */
UNIV_INTERN
int
btr_scrub_free_page(
/*====================*/
	btr_scrub_t* scrub_data,  /*!< in/out: scrub data */
	buf_block_t* block,       /*!< in: block to scrub */
	mtr_t* mtr)               /*!< in: mtr */
{
	// TODO(jonaso): scrub only what is actually needed

	{
		/* note: perform both the memset and setting of FIL_PAGE_TYPE
		* wo/ logging. so that if we crash before page is flushed
		* it will be found by scrubbing thread again
		*/
		memset(buf_block_get_frame(block) + PAGE_HEADER, 0,
		       srv_page_size - PAGE_HEADER);

		mach_write_to_2(buf_block_get_frame(block) + FIL_PAGE_TYPE,
				FIL_PAGE_TYPE_ALLOCATED);
	}

	page_create(block, mtr,
		    dict_table_is_comp(scrub_data->current_table),
		    dict_index_is_spatial(scrub_data->current_index));

	mtr_commit(mtr);

	/* page doesn't need further processing => SKIP
	* and close table/index so that we don't keep references too long */
	return BTR_SCRUB_SKIP_PAGE_AND_CLOSE_TABLE;
}

/****************************************************************
Recheck if a page needs scrubbing, and if it does load appropriate
table and index */
UNIV_INTERN
int
btr_scrub_recheck_page(
/*====================*/
	btr_scrub_t* scrub_data,  /*!< inut: scrub data */
	buf_block_t* block,       /*!< in: block */
	btr_scrub_page_allocation_status_t allocated, /*!< in: is block
						      allocated or free */
	mtr_t* mtr)               /*!< in: mtr */
{
	/* recheck if page needs scrubbing (knowing allocation status) */
	int needs_scrubbing = btr_page_needs_scrubbing(
		scrub_data, block, allocated);

	if (needs_scrubbing != BTR_SCRUB_PAGE) {
		mtr_commit(mtr);
		return needs_scrubbing;
	}

	if (allocated == BTR_SCRUB_PAGE_FREE) {
		/** we don't need to load table/index for free pages
		* so scrub directly here */
		/* mtr is committed inside btr_scrub_page_free */
		return btr_scrub_free_page(scrub_data,
					   block,
					   mtr);
	}

	page_t*	page = buf_block_get_frame(block);
	index_id_t index_id = btr_page_get_index_id(page);

	if (scrub_data->current_index == NULL ||
	    scrub_data->current_index->id != index_id) {

		/**
		* commit mtr (i.e release locks on block)
		* and try to get table&index potentially loading it
		* from disk
		*/
		mtr_commit(mtr);
		btr_scrub_get_table_and_index(scrub_data, index_id);
	} else {
		/* we already have correct index
		* commit mtr so that we can lock index before fetching page
		*/
		mtr_commit(mtr);
	}

	/* check if table is about to be dropped */
	if (!btr_scrub_table_needs_scrubbing(scrub_data->current_table)) {
		return BTR_SCRUB_SKIP_PAGE_AND_CLOSE_TABLE;
	}

	/* check if index is scrubbable */
	if (!btr_scrub_index_needs_scrubbing(scrub_data->current_index)) {
		return BTR_SCRUB_SKIP_PAGE_AND_CLOSE_TABLE;
	}

	mtr_start(mtr);
	mtr_x_lock(dict_index_get_lock(scrub_data->current_index), mtr);
	/** set savepoint for X-latch of block */
	scrub_data->savepoint = mtr_set_savepoint(mtr);
	return BTR_SCRUB_PAGE;
}

/****************************************************************
Perform actual scrubbing of page */
UNIV_INTERN
int
btr_scrub_page(
/*============*/
	btr_scrub_t* scrub_data,  /*!< in/out: scrub data */
	buf_block_t* block,       /*!< in: block */
	btr_scrub_page_allocation_status_t allocated, /*!< in: is block
						      allocated or free */
	mtr_t* mtr)               /*!< in: mtr */
{
	/* recheck if page needs scrubbing (knowing allocation status) */
	int needs_scrubbing = BTR_SCRUB_SKIP_PAGE_AND_CLOSE_TABLE;

	if (block) {
		btr_page_needs_scrubbing(scrub_data, block, allocated);
	}

	if (!block || needs_scrubbing != BTR_SCRUB_PAGE) {
		mtr_commit(mtr);
		return needs_scrubbing;
	}

	if (allocated == BTR_SCRUB_PAGE_FREE) {
		/* mtr is committed inside btr_scrub_page_free */
		return btr_scrub_free_page(scrub_data,
					   block,
					   mtr);
	}

	/* check that table/index still match now that they are loaded */

	if (!scrub_data->current_table->space
	    || scrub_data->current_table->space->id != scrub_data->space) {
		/* this is truncate table */
		mtr_commit(mtr);
		return BTR_SCRUB_SKIP_PAGE_AND_CLOSE_TABLE;
	}

	if (scrub_data->current_index->table != scrub_data->current_table) {
		/* this is truncate table */
		mtr_commit(mtr);
		return BTR_SCRUB_SKIP_PAGE_AND_CLOSE_TABLE;
	}

	if (scrub_data->current_index->page == FIL_NULL) {
		/* this is truncate table */
		mtr_commit(mtr);
		return BTR_SCRUB_SKIP_PAGE_AND_CLOSE_TABLE;
	}

	buf_frame_t* frame = buf_block_get_frame(block);

	if (!frame || btr_page_get_index_id(frame) !=
	    scrub_data->current_index->id) {
		/* page has been reallocated to new index */
		mtr_commit(mtr);
		return BTR_SCRUB_SKIP_PAGE_AND_CLOSE_TABLE;
	}

	/* check if I can scrub (reorganize) page wo/ overflow */
	if (btr_optimistic_scrub(scrub_data,
				 block,
				 scrub_data->current_index,
				 mtr) != DB_SUCCESS) {

		/**
		* Can't reorganize page...need to split it
		*/
		btr_pessimistic_scrub(scrub_data,
				      block,
				      scrub_data->current_index,
				      mtr);
	}
	mtr_commit(mtr);

	return BTR_SCRUB_SKIP_PAGE; // no further action needed
}

/**************************************************************//**
Start iterating a space */
UNIV_INTERN
bool
btr_scrub_start_space(
/*===================*/
	ulint space,             /*!< in: space */
	btr_scrub_t* scrub_data) /*!< in/out: scrub data */
{
	bool found;
	scrub_data->space = space;
	scrub_data->current_table = NULL;
	scrub_data->current_index = NULL;
	const page_size_t page_size = fil_space_get_page_size(space, &found);

	scrub_data->compressed = page_size.is_compressed();
	scrub_data->scrubbing = check_scrub_setting(scrub_data);
	return scrub_data->scrubbing;
}

/***********************************************************************
Update global statistics with thread statistics */
static
void
btr_scrub_update_total_stat(btr_scrub_t *scrub_data)
{
	mutex_enter(&scrub_stat_mutex);
	scrub_stat.page_reorganizations +=
		scrub_data->scrub_stat.page_reorganizations;
	scrub_stat.page_splits +=
		scrub_data->scrub_stat.page_splits;
	scrub_stat.page_split_failures_underflow +=
		scrub_data->scrub_stat.page_split_failures_underflow;
	scrub_stat.page_split_failures_out_of_filespace +=
		scrub_data->scrub_stat.page_split_failures_out_of_filespace;
	scrub_stat.page_split_failures_missing_index +=
		scrub_data->scrub_stat.page_split_failures_missing_index;
	scrub_stat.page_split_failures_unknown +=
		scrub_data->scrub_stat.page_split_failures_unknown;
	mutex_exit(&scrub_stat_mutex);

	// clear stat
	memset(&scrub_data->scrub_stat, 0, sizeof(scrub_data->scrub_stat));
}

/** Complete iterating a space.
@param[in,out]	scrub_data	 scrub data */
UNIV_INTERN
void
btr_scrub_complete_space(btr_scrub_t* scrub_data)
{
	ut_ad(scrub_data->scrubbing);
	btr_scrub_table_close_for_thread(scrub_data);
	btr_scrub_update_total_stat(scrub_data);
}

/*********************************************************************
Return scrub statistics */
void
btr_scrub_total_stat(btr_scrub_stat_t *stat)
{
	mutex_enter(&scrub_stat_mutex);
	*stat = scrub_stat;
	mutex_exit(&scrub_stat_mutex);
}

/*********************************************************************
Init global variables */
UNIV_INTERN
void
btr_scrub_init()
{
	mutex_create(LATCH_ID_SCRUB_STAT_MUTEX, &scrub_stat_mutex);

	memset(&scrub_stat, 0, sizeof(scrub_stat));
}

/*********************************************************************
Cleanup globals */
UNIV_INTERN
void
btr_scrub_cleanup()
{
	mutex_free(&scrub_stat_mutex);
}