~ubuntu-branches/ubuntu/saucy/drizzle/saucy-proposed

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

Copyright (c) 1995, 2009, Innobase Oy. All Rights Reserved.
Copyright (c) 2008, Google Inc.

Portions of this file contain modifications contributed and copyrighted by
Google, Inc. Those modifications are gratefully acknowledged and are described
briefly in the InnoDB documentation. The contributions by Google are
incorporated with their permission, and subject to the conditions contained in
the file COPYING.Google.

This program is free software; you can redistribute it and/or modify it under
the terms of the GNU General Public License as published by the Free Software
Foundation; version 2 of the License.

This program is distributed in the hope that it will be useful, but WITHOUT
ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS
FOR A PARTICULAR PURPOSE. See the GNU General Public License for more details.

You should have received a copy of the GNU General Public License along with
this program; if not, write to the Free Software Foundation, Inc., 59 Temple
Place, Suite 330, Boston, MA 02111-1307 USA

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

/**************************************************//**
@file sync/sync0arr.c
The wait array used in synchronization primitives

Created 9/5/1995 Heikki Tuuri
*******************************************************/

#include "sync0arr.h"
#ifdef UNIV_NONINL
#include "sync0arr.ic"
#endif

#include "sync0sync.h"
#include "sync0rw.h"
#include "os0sync.h"
#include "os0file.h"
#include "srv0srv.h"

/*
			WAIT ARRAY
			==========

The wait array consists of cells each of which has an
an operating system event object created for it. The threads
waiting for a mutex, for example, can reserve a cell
in the array and suspend themselves to wait for the event
to become signaled. When using the wait array, remember to make
sure that some thread holding the synchronization object
will eventually know that there is a waiter in the array and
signal the object, to prevent infinite wait.
Why we chose to implement a wait array? First, to make
mutexes fast, we had to code our own implementation of them,
which only in usually uncommon cases resorts to using
slow operating system primitives. Then we had the choice of
assigning a unique OS event for each mutex, which would
be simpler, or using a global wait array. In some operating systems,
the global wait array solution is more efficient and flexible,
because we can do with a very small number of OS events,
say 200. In NT 3.51, allocating events seems to be a quadratic
algorithm, because 10 000 events are created fast, but
100 000 events takes a couple of minutes to create.

As of 5.0.30 the above mentioned design is changed. Since now
OS can handle millions of wait events efficiently, we no longer
have this concept of each cell of wait array having one event.
Instead, now the event that a thread wants to wait on is embedded
in the wait object (mutex or rw_lock). We still keep the global
wait array for the sake of diagnostics and also to avoid infinite
wait The error_monitor thread scans the global wait array to signal
any waiting threads who have missed the signal. */

/** A cell where an individual thread may wait suspended
until a resource is released. The suspending is implemented
using an operating system event semaphore. */
struct sync_cell_struct {
	void*		wait_object;	/*!< pointer to the object the
					thread is waiting for; if NULL
					the cell is free for use */
	mutex_t*	old_wait_mutex;	/*!< the latest wait mutex in cell */
	rw_lock_t*	old_wait_rw_lock;
					/*!< the latest wait rw-lock
					in cell */
	ulint		request_type;	/*!< lock type requested on the
					object */
	const char*	file;		/*!< in debug version file where
					requested */
	ulint		line;		/*!< in debug version line where
					requested */
	os_thread_id_t	thread;		/*!< thread id of this waiting
					thread */
	ibool		waiting;	/*!< TRUE if the thread has already
					called sync_array_event_wait
					on this cell */
	ib_int64_t	signal_count;	/*!< We capture the signal_count
					of the wait_object when we
					reset the event. This value is
					then passed on to os_event_wait
					and we wait only if the event
					has not been signalled in the
					period between the reset and
					wait call. */
	time_t		reservation_time;/*!< time when the thread reserved
					the wait cell */
};

/* NOTE: It is allowed for a thread to wait
for an event allocated for the array without owning the
protecting mutex (depending on the case: OS or database mutex), but
all changes (set or reset) to the state of the event must be made
while owning the mutex. */

/** Synchronization array */
struct sync_array_struct {
	ulint		n_reserved;	/*!< number of currently reserved
					cells in the wait array */
	ulint		n_cells;	/*!< number of cells in the
					wait array */
	sync_cell_t*	array;		/*!< pointer to wait array */
	ulint		protection;	/*!< this flag tells which
					mutex protects the data */
	mutex_t		mutex;		/*!< possible database mutex
					protecting this data structure */
	os_mutex_t	os_mutex;	/*!< Possible operating system mutex
					protecting the data structure.
					As this data structure is used in
					constructing the database mutex,
					to prevent infinite recursion
					in implementation, we fall back to
					an OS mutex. */
	ulint		sg_count;	/*!< count of how many times an
					object has been signalled */
	ulint		res_count;	/*!< count of cell reservations
					since creation of the array */
};

#ifdef UNIV_SYNC_DEBUG
/******************************************************************//**
This function is called only in the debug version. Detects a deadlock
of one or more threads because of waits of semaphores.
@return	TRUE if deadlock detected */
static
ibool
sync_array_detect_deadlock(
/*=======================*/
	sync_array_t*	arr,	/*!< in: wait array; NOTE! the caller must
				own the mutex to array */
	sync_cell_t*	start,	/*!< in: cell where recursive search started */
	sync_cell_t*	cell,	/*!< in: cell to search */
	ulint		depth);	/*!< in: recursion depth */
#endif /* UNIV_SYNC_DEBUG */

/*****************************************************************//**
Gets the nth cell in array.
@return	cell */
static
sync_cell_t*
sync_array_get_nth_cell(
/*====================*/
	sync_array_t*	arr,	/*!< in: sync array */
	ulint		n)	/*!< in: index */
{
	ut_a(arr);
	ut_a(n < arr->n_cells);

	return(arr->array + n);
}

/******************************************************************//**
Reserves the mutex semaphore protecting a sync array. */
static
void
sync_array_enter(
/*=============*/
	sync_array_t*	arr)	/*!< in: sync wait array */
{
	ulint	protection;

	protection = arr->protection;

	if (protection == SYNC_ARRAY_OS_MUTEX) {
		os_mutex_enter(arr->os_mutex);
	} else if (protection == SYNC_ARRAY_MUTEX) {
		mutex_enter(&(arr->mutex));
	} else {
		ut_error;
	}
}

/******************************************************************//**
Releases the mutex semaphore protecting a sync array. */
static
void
sync_array_exit(
/*============*/
	sync_array_t*	arr)	/*!< in: sync wait array */
{
	ulint	protection;

	protection = arr->protection;

	if (protection == SYNC_ARRAY_OS_MUTEX) {
		os_mutex_exit(arr->os_mutex);
	} else if (protection == SYNC_ARRAY_MUTEX) {
		mutex_exit(&(arr->mutex));
	} else {
		ut_error;
	}
}

/*******************************************************************//**
Creates a synchronization wait array. It is protected by a mutex
which is automatically reserved when the functions operating on it
are called.
@return	own: created wait array */
UNIV_INTERN
sync_array_t*
sync_array_create(
/*==============*/
	ulint	n_cells,	/*!< in: number of cells in the array
				to create */
	ulint	protection)	/*!< in: either SYNC_ARRAY_OS_MUTEX or
				SYNC_ARRAY_MUTEX: determines the type
				of mutex protecting the data structure */
{
	sync_array_t*	arr;
	sync_cell_t*	cell_array;
	sync_cell_t*	cell;
	ulint		i;

	ut_a(n_cells > 0);

	/* Allocate memory for the data structures */
	arr = ut_malloc(sizeof(sync_array_t));

	cell_array = ut_malloc(sizeof(sync_cell_t) * n_cells);

	arr->n_cells = n_cells;
	arr->n_reserved = 0;
	arr->array = cell_array;
	arr->protection = protection;
	arr->sg_count = 0;
	arr->res_count = 0;

	/* Then create the mutex to protect the wait array complex */
	if (protection == SYNC_ARRAY_OS_MUTEX) {
		arr->os_mutex = os_mutex_create(NULL);
	} else if (protection == SYNC_ARRAY_MUTEX) {
		mutex_create(&arr->mutex, SYNC_NO_ORDER_CHECK);
	} else {
		ut_error;
	}

	for (i = 0; i < n_cells; i++) {
		cell = sync_array_get_nth_cell(arr, i);
	cell->wait_object = NULL;
		cell->waiting = FALSE;
		cell->signal_count = 0;
	}

	return(arr);
}

/******************************************************************//**
Frees the resources in a wait array. */
UNIV_INTERN
void
sync_array_free(
/*============*/
	sync_array_t*	arr)	/*!< in, own: sync wait array */
{
	ulint		protection;

	ut_a(arr->n_reserved == 0);

	sync_array_validate(arr);

	protection = arr->protection;

	/* Release the mutex protecting the wait array complex */

	if (protection == SYNC_ARRAY_OS_MUTEX) {
		os_mutex_free(arr->os_mutex);
	} else if (protection == SYNC_ARRAY_MUTEX) {
		mutex_free(&(arr->mutex));
	} else {
		ut_error;
	}

	ut_free(arr->array);
	ut_free(arr);
}

/********************************************************************//**
Validates the integrity of the wait array. Checks
that the number of reserved cells equals the count variable. */
UNIV_INTERN
void
sync_array_validate(
/*================*/
	sync_array_t*	arr)	/*!< in: sync wait array */
{
	ulint		i;
	sync_cell_t*	cell;
	ulint		count		= 0;

	sync_array_enter(arr);

	for (i = 0; i < arr->n_cells; i++) {
		cell = sync_array_get_nth_cell(arr, i);
		if (cell->wait_object != NULL) {
			count++;
		}
	}

	ut_a(count == arr->n_reserved);

	sync_array_exit(arr);
}

/*******************************************************************//**
Returns the event that the thread owning the cell waits for. */
static
os_event_t
sync_cell_get_event(
/*================*/
	sync_cell_t*	cell) /*!< in: non-empty sync array cell */
{
	ulint type = cell->request_type;

	if (type == SYNC_MUTEX) {
		return(((mutex_t *) cell->wait_object)->event);
	} else if (type == RW_LOCK_WAIT_EX) {
		return(((rw_lock_t *) cell->wait_object)->wait_ex_event);
	} else { /* RW_LOCK_SHARED and RW_LOCK_EX wait on the same event */
		return(((rw_lock_t *) cell->wait_object)->event);
	}
}

/******************************************************************//**
Reserves a wait array cell for waiting for an object.
The event of the cell is reset to nonsignalled state. */
UNIV_INTERN
void
sync_array_reserve_cell(
/*====================*/
	sync_array_t*	arr,	/*!< in: wait array */
	void*		object, /*!< in: pointer to the object to wait for */
	ulint		type,	/*!< in: lock request type */
	const char*	file,	/*!< in: file where requested */
	ulint		line,	/*!< in: line where requested */
	ulint*		index)	/*!< out: index of the reserved cell */
{
	sync_cell_t*	cell;
	os_event_t      event;
	ulint		i;

	ut_a(object);
	ut_a(index);

	sync_array_enter(arr);

	arr->res_count++;

	/* Reserve a new cell. */
	for (i = 0; i < arr->n_cells; i++) {
		cell = sync_array_get_nth_cell(arr, i);

		if (cell->wait_object == NULL) {

			cell->waiting = FALSE;
			cell->wait_object = object;

			if (type == SYNC_MUTEX) {
				cell->old_wait_mutex = object;
			} else {
				cell->old_wait_rw_lock = object;
			}

			cell->request_type = type;

			cell->file = file;
			cell->line = line;

			arr->n_reserved++;

			*index = i;

			sync_array_exit(arr);

			/* Make sure the event is reset and also store
			the value of signal_count at which the event
			was reset. */
                        event = sync_cell_get_event(cell);
			cell->signal_count = os_event_reset(event);

			cell->reservation_time = time(NULL);

			cell->thread = os_thread_get_curr_id();

			return;
		}
	}

	ut_error; /* No free cell found */

	return;
}

/******************************************************************//**
This function should be called when a thread starts to wait on
a wait array cell. In the debug version this function checks
if the wait for a semaphore will result in a deadlock, in which
case prints info and asserts. */
UNIV_INTERN
void
sync_array_wait_event(
/*==================*/
	sync_array_t*	arr,	/*!< in: wait array */
	ulint		index)	/*!< in: index of the reserved cell */
{
	sync_cell_t*	cell;
	os_event_t	event;

	ut_a(arr);

	sync_array_enter(arr);

	cell = sync_array_get_nth_cell(arr, index);

	ut_a(cell->wait_object);
	ut_a(!cell->waiting);
	ut_ad(os_thread_get_curr_id() == cell->thread);

	event = sync_cell_get_event(cell);
		cell->waiting = TRUE;

#ifdef UNIV_SYNC_DEBUG

	/* We use simple enter to the mutex below, because if
	we cannot acquire it at once, mutex_enter would call
	recursively sync_array routines, leading to trouble.
	rw_lock_debug_mutex freezes the debug lists. */

	rw_lock_debug_mutex_enter();

	if (TRUE == sync_array_detect_deadlock(arr, cell, cell, 0)) {

		fputs("########################################\n", stderr);
		ut_error;
	}

	rw_lock_debug_mutex_exit();
#endif
	sync_array_exit(arr);

	os_event_wait_low(event, cell->signal_count);

	sync_array_free_cell(arr, index);
}

/******************************************************************//**
Reports info of a wait array cell. */
static
void
sync_array_cell_print(
/*==================*/
	FILE*		file,	/*!< in: file where to print */
	sync_cell_t*	cell)	/*!< in: sync cell */
{
	mutex_t*	mutex;
	rw_lock_t*	rwlock;
	ulint		type;
	ulint		writer;

	type = cell->request_type;

	fprintf(file,
		"--Thread %lu has waited at %s line %lu"
		" for %.2f seconds the semaphore:\n",
		(ulong) os_thread_pf(cell->thread), cell->file,
		(ulong) cell->line,
		difftime(time(NULL), cell->reservation_time));

	if (type == SYNC_MUTEX) {
		/* We use old_wait_mutex in case the cell has already
		been freed meanwhile */
		mutex = cell->old_wait_mutex;

		fprintf(file,
			"Mutex at %p created file %s line %lu, lock var %lu\n"
#ifdef UNIV_SYNC_DEBUG
			"Last time reserved in file %s line %lu, "
#endif /* UNIV_SYNC_DEBUG */
			"waiters flag %lu\n",
			(void*) mutex, mutex->cfile_name, (ulong) mutex->cline,
			(ulong) mutex->lock_word,
#ifdef UNIV_SYNC_DEBUG
			mutex->file_name, (ulong) mutex->line,
#endif /* UNIV_SYNC_DEBUG */
			(ulong) mutex->waiters);

	} else if (type == RW_LOCK_EX
		   || type == RW_LOCK_WAIT_EX
		   || type == RW_LOCK_SHARED) {

		fputs(type == RW_LOCK_EX ? "X-lock on" : "S-lock on", file);

		rwlock = cell->old_wait_rw_lock;

		fprintf(file,
			" RW-latch at %p created in file %s line %lu\n",
			(void*) rwlock, rwlock->cfile_name,
			(ulong) rwlock->cline);
		writer = rw_lock_get_writer(rwlock);
		if (writer != RW_LOCK_NOT_LOCKED) {
			fprintf(file,
				"a writer (thread id %lu) has"
				" reserved it in mode %s",
				(ulong) os_thread_pf(rwlock->writer_thread),
				writer == RW_LOCK_EX
				? " exclusive\n"
				: " wait exclusive\n");
		}

		fprintf(file,
			"number of readers %lu, waiters flag %lu, "
                        "lock_word: %lx\n"
			"Last time read locked in file %s line %lu\n"
			"Last time write locked in file %s line %lu\n",
			(ulong) rw_lock_get_reader_count(rwlock),
			(ulong) rwlock->waiters,
			rwlock->lock_word,
			rwlock->last_s_file_name,
			(ulong) rwlock->last_s_line,
			rwlock->last_x_file_name,
			(ulong) rwlock->last_x_line);
	} else {
		ut_error;
	}

	if (!cell->waiting) {
		fputs("wait has ended\n", file);
	}
}

#ifdef UNIV_SYNC_DEBUG
/******************************************************************//**
Looks for a cell with the given thread id.
@return	pointer to cell or NULL if not found */
static
sync_cell_t*
sync_array_find_thread(
/*===================*/
	sync_array_t*	arr,	/*!< in: wait array */
	os_thread_id_t	thread)	/*!< in: thread id */
{
	ulint		i;
	sync_cell_t*	cell;

	for (i = 0; i < arr->n_cells; i++) {

		cell = sync_array_get_nth_cell(arr, i);

		if (cell->wait_object != NULL
		    && os_thread_eq(cell->thread, thread)) {

			return(cell);	/* Found */
		}
	}

	return(NULL);	/* Not found */
}

/******************************************************************//**
Recursion step for deadlock detection.
@return	TRUE if deadlock detected */
static
ibool
sync_array_deadlock_step(
/*=====================*/
	sync_array_t*	arr,	/*!< in: wait array; NOTE! the caller must
				own the mutex to array */
	sync_cell_t*	start,	/*!< in: cell where recursive search
				started */
	os_thread_id_t	thread,	/*!< in: thread to look at */
	ulint		pass,	/*!< in: pass value */
	ulint		depth)	/*!< in: recursion depth */
{
	sync_cell_t*	new;
	ibool		ret;

	depth++;

	if (pass != 0) {
		/* If pass != 0, then we do not know which threads are
		responsible of releasing the lock, and no deadlock can
		be detected. */

		return(FALSE);
	}

	new = sync_array_find_thread(arr, thread);

	if (new == start) {
		/* Stop running of other threads */

		ut_dbg_stop_threads = TRUE;

		/* Deadlock */
		fputs("########################################\n"
		      "DEADLOCK of threads detected!\n", stderr);

		return(TRUE);

	} else if (new) {
		ret = sync_array_detect_deadlock(arr, start, new, depth);

		if (ret) {
			return(TRUE);
		}
	}
	return(FALSE);
}

/******************************************************************//**
This function is called only in the debug version. Detects a deadlock
of one or more threads because of waits of semaphores.
@return	TRUE if deadlock detected */
static
ibool
sync_array_detect_deadlock(
/*=======================*/
	sync_array_t*	arr,	/*!< in: wait array; NOTE! the caller must
				own the mutex to array */
	sync_cell_t*	start,	/*!< in: cell where recursive search started */
	sync_cell_t*	cell,	/*!< in: cell to search */
	ulint		depth)	/*!< in: recursion depth */
{
	mutex_t*	mutex;
	rw_lock_t*	lock;
	os_thread_id_t	thread;
	ibool		ret;
	rw_lock_debug_t*debug;

	ut_a(arr);
	ut_a(start);
	ut_a(cell);
	ut_ad(cell->wait_object);
	ut_ad(os_thread_get_curr_id() == start->thread);
	ut_ad(depth < 100);

	depth++;

	if (!cell->waiting) {

		return(FALSE); /* No deadlock here */
	}

	if (cell->request_type == SYNC_MUTEX) {

		mutex = cell->wait_object;

		if (mutex_get_lock_word(mutex) != 0) {

			thread = mutex->thread_id;

			/* Note that mutex->thread_id above may be
			also OS_THREAD_ID_UNDEFINED, because the
			thread which held the mutex maybe has not
			yet updated the value, or it has already
			released the mutex: in this case no deadlock
			can occur, as the wait array cannot contain
			a thread with ID_UNDEFINED value. */

			ret = sync_array_deadlock_step(arr, start, thread, 0,
						       depth);
			if (ret) {
				fprintf(stderr,
			"Mutex %p owned by thread %lu file %s line %lu\n",
					mutex, (ulong) os_thread_pf(mutex->thread_id),
					mutex->file_name, (ulong) mutex->line);
				sync_array_cell_print(stderr, cell);

				return(TRUE);
			}
		}

		return(FALSE); /* No deadlock */

	} else if (cell->request_type == RW_LOCK_EX
		   || cell->request_type == RW_LOCK_WAIT_EX) {

		lock = cell->wait_object;

		debug = UT_LIST_GET_FIRST(lock->debug_list);

		while (debug != NULL) {

			thread = debug->thread_id;

			if (((debug->lock_type == RW_LOCK_EX)
			     && !os_thread_eq(thread, cell->thread))
			    || ((debug->lock_type == RW_LOCK_WAIT_EX)
				&& !os_thread_eq(thread, cell->thread))
			    || (debug->lock_type == RW_LOCK_SHARED)) {

				/* The (wait) x-lock request can block
				infinitely only if someone (can be also cell
				thread) is holding s-lock, or someone
				(cannot be cell thread) (wait) x-lock, and
				he is blocked by start thread */

				ret = sync_array_deadlock_step(
					arr, start, thread, debug->pass,
					depth);
				if (ret) {
print:
					fprintf(stderr, "rw-lock %p ",
						(void*) lock);
					sync_array_cell_print(stderr, cell);
					rw_lock_debug_print(debug);
					return(TRUE);
				}
			}

			debug = UT_LIST_GET_NEXT(list, debug);
		}

		return(FALSE);

	} else if (cell->request_type == RW_LOCK_SHARED) {

		lock = cell->wait_object;
		debug = UT_LIST_GET_FIRST(lock->debug_list);

		while (debug != NULL) {

			thread = debug->thread_id;

			if ((debug->lock_type == RW_LOCK_EX)
			    || (debug->lock_type == RW_LOCK_WAIT_EX)) {

				/* The s-lock request can block infinitely
				only if someone (can also be cell thread) is
				holding (wait) x-lock, and he is blocked by
				start thread */

				ret = sync_array_deadlock_step(
					arr, start, thread, debug->pass,
					depth);
				if (ret) {
					goto print;
				}
			}

			debug = UT_LIST_GET_NEXT(list, debug);
		}

		return(FALSE);

	} else {
		ut_error;
	}

	return(TRUE);	/* Execution never reaches this line: for compiler
			fooling only */
}
#endif /* UNIV_SYNC_DEBUG */

/******************************************************************//**
Determines if we can wake up the thread waiting for a sempahore. */
static
ibool
sync_arr_cell_can_wake_up(
/*======================*/
	sync_cell_t*	cell)	/*!< in: cell to search */
{
	mutex_t*	mutex;
	rw_lock_t*	lock;

	if (cell->request_type == SYNC_MUTEX) {

		mutex = cell->wait_object;

		if (mutex_get_lock_word(mutex) == 0) {

			return(TRUE);
		}

	} else if (cell->request_type == RW_LOCK_EX) {

		lock = cell->wait_object;

		if (lock->lock_word > 0) {
		/* Either unlocked or only read locked. */

			return(TRUE);
		}

        } else if (cell->request_type == RW_LOCK_WAIT_EX) {

		lock = cell->wait_object;

                /* lock_word == 0 means all readers have left */
		if (lock->lock_word == 0) {

			return(TRUE);
		}
	} else if (cell->request_type == RW_LOCK_SHARED) {
		lock = cell->wait_object;

                /* lock_word > 0 means no writer or reserved writer */
		if (lock->lock_word > 0) {

			return(TRUE);
		}
	}

	return(FALSE);
}

/******************************************************************//**
Frees the cell. NOTE! sync_array_wait_event frees the cell
automatically! */
UNIV_INTERN
void
sync_array_free_cell(
/*=================*/
	sync_array_t*	arr,	/*!< in: wait array */
	ulint		index)  /*!< in: index of the cell in array */
{
	sync_cell_t*	cell;

	sync_array_enter(arr);

	cell = sync_array_get_nth_cell(arr, index);

	ut_a(cell->wait_object != NULL);

	cell->waiting = FALSE;
	cell->wait_object =  NULL;
	cell->signal_count = 0;

	ut_a(arr->n_reserved > 0);
	arr->n_reserved--;

	sync_array_exit(arr);
}

/**********************************************************************//**
Increments the signalled count. */
UNIV_INTERN
void
sync_array_object_signalled(
/*========================*/
	sync_array_t*	arr)	/*!< in: wait array */
{
#ifdef HAVE_ATOMIC_BUILTINS
	(void) os_atomic_increment_ulint(&arr->sg_count, 1);
#else
	sync_array_enter(arr);

	arr->sg_count++;

	sync_array_exit(arr);
#endif
}

/**********************************************************************//**
If the wakeup algorithm does not work perfectly at semaphore relases,
this function will do the waking (see the comment in mutex_exit). This
function should be called about every 1 second in the server.

Note that there's a race condition between this thread and mutex_exit
changing the lock_word and calling signal_object, so sometimes this finds
threads to wake up even when nothing has gone wrong. */
UNIV_INTERN
void
sync_arr_wake_threads_if_sema_free(void)
/*====================================*/
{
	sync_array_t*	arr	= sync_primary_wait_array;
	sync_cell_t*	cell;
	ulint		count;
	ulint		i;
	os_event_t      event;

	sync_array_enter(arr);

	i = 0;
	count = 0;

	while (count < arr->n_reserved) {

		cell = sync_array_get_nth_cell(arr, i);
		i++;

		if (cell->wait_object == NULL) {
			continue;
		}
			count++;

			if (sync_arr_cell_can_wake_up(cell)) {

			event = sync_cell_get_event(cell);

			os_event_set(event);
		}

	}

	sync_array_exit(arr);
}

/**********************************************************************//**
Prints warnings of long semaphore waits to stderr.
@return	TRUE if fatal semaphore wait threshold was exceeded */
UNIV_INTERN
ibool
sync_array_print_long_waits(void)
/*=============================*/
{
	sync_cell_t*	cell;
	ibool		old_val;
	ibool		noticed = FALSE;
	ulint		i;
	ulint		fatal_timeout = srv_fatal_semaphore_wait_threshold;
	ibool		fatal = FALSE;

	for (i = 0; i < sync_primary_wait_array->n_cells; i++) {

		cell = sync_array_get_nth_cell(sync_primary_wait_array, i);

		if (cell->wait_object != NULL && cell->waiting
		    && difftime(time(NULL), cell->reservation_time) > 240) {
			fputs("InnoDB: Warning: a long semaphore wait:\n",
			      stderr);
			sync_array_cell_print(stderr, cell);
			noticed = TRUE;
		}

		if (cell->wait_object != NULL && cell->waiting
		    && difftime(time(NULL), cell->reservation_time)
		    > fatal_timeout) {
			fatal = TRUE;
		}
	}

	if (noticed) {
		fprintf(stderr,
			"InnoDB: ###### Starts InnoDB Monitor"
			" for 30 secs to print diagnostic info:\n");
		old_val = srv_print_innodb_monitor;

		/* If some crucial semaphore is reserved, then also the InnoDB
		Monitor can hang, and we do not get diagnostics. Since in
		many cases an InnoDB hang is caused by a pwrite() or a pread()
		call hanging inside the operating system, let us print right
		now the values of pending calls of these. */

		fprintf(stderr,
			"InnoDB: Pending preads %lu, pwrites %lu\n",
			(ulong)os_file_n_pending_preads,
			(ulong)os_file_n_pending_pwrites);

		srv_print_innodb_monitor = TRUE;
		os_event_set(srv_lock_timeout_thread_event);

		os_thread_sleep(30000000);

		srv_print_innodb_monitor = old_val;
		fprintf(stderr,
			"InnoDB: ###### Diagnostic info printed"
			" to the standard error stream\n");
	}

	return(fatal);
}

/**********************************************************************//**
Prints info of the wait array. */
static
void
sync_array_output_info(
/*===================*/
	FILE*		file,	/*!< in: file where to print */
	sync_array_t*	arr)	/*!< in: wait array; NOTE! caller must own the
				mutex */
{
	sync_cell_t*	cell;
	ulint		count;
	ulint		i;

	fprintf(file,
		"OS WAIT ARRAY INFO: reservation count %ld, signal count %ld\n",
						(long) arr->res_count, (long) arr->sg_count);
	i = 0;
	count = 0;

	while (count < arr->n_reserved) {

		cell = sync_array_get_nth_cell(arr, i);

	if (cell->wait_object != NULL) {
		count++;
			sync_array_cell_print(file, cell);
		}

		i++;
	}
}

/**********************************************************************//**
Prints info of the wait array. */
UNIV_INTERN
void
sync_array_print_info(
/*==================*/
	FILE*		file,	/*!< in: file where to print */
	sync_array_t*	arr)	/*!< in: wait array */
{
	sync_array_enter(arr);

	sync_array_output_info(file, arr);

	sync_array_exit(arr);
}