summaryrefslogtreecommitdiffstats
path: root/sys/kern/kern_lock.c
blob: 5b6910656746e19dfd3a5a77cd73eb8dcc11157b (plain)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
704
705
706
707
708
709
710
711
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
734
735
736
737
738
739
740
741
742
743
744
745
746
747
748
749
750
751
752
753
754
755
756
757
758
759
760
761
762
763
764
765
766
767
768
769
770
771
772
773
774
775
776
777
778
779
780
781
782
783
784
785
786
787
788
789
790
791
792
793
794
795
796
797
798
799
800
801
802
803
804
805
806
807
808
809
810
811
812
813
814
815
816
817
818
819
820
821
822
823
824
825
826
827
828
829
830
831
832
833
834
835
836
837
838
839
840
841
842
843
844
845
846
847
848
849
850
851
852
853
854
855
856
857
858
859
860
861
862
863
864
865
866
867
868
869
870
871
872
873
874
875
876
877
878
879
880
881
882
883
884
885
886
887
888
889
890
891
892
893
894
895
896
897
898
899
900
901
902
903
904
905
906
907
908
909
910
911
912
913
914
915
916
917
918
919
920
921
922
923
924
925
926
927
928
929
930
931
932
933
934
935
936
937
938
939
940
941
942
943
944
945
946
947
948
949
950
951
952
953
954
955
956
957
958
959
960
961
962
963
964
965
966
967
968
969
970
971
972
973
974
975
976
977
978
979
980
981
982
983
984
985
986
987
988
989
990
991
992
993
994
995
996
997
998
999
1000
1001
1002
1003
1004
1005
1006
1007
1008
1009
1010
1011
1012
1013
1014
1015
1016
1017
1018
1019
1020
1021
1022
1023
1024
1025
1026
1027
1028
1029
1030
1031
1032
1033
1034
1035
1036
1037
1038
1039
1040
1041
1042
1043
1044
1045
1046
1047
1048
1049
1050
1051
1052
1053
1054
1055
1056
1057
1058
1059
1060
1061
1062
1063
1064
1065
1066
1067
1068
1069
1070
1071
1072
1073
1074
1075
1076
1077
1078
1079
1080
1081
1082
1083
1084
1085
1086
1087
1088
1089
1090
1091
1092
1093
1094
1095
1096
1097
1098
1099
1100
1101
1102
1103
1104
1105
1106
1107
1108
1109
1110
1111
1112
1113
1114
1115
1116
1117
1118
1119
1120
1121
1122
1123
1124
1125
1126
1127
1128
1129
1130
1131
1132
1133
1134
1135
1136
1137
1138
1139
1140
1141
1142
1143
1144
1145
1146
1147
1148
1149
1150
1151
1152
1153
1154
1155
1156
1157
1158
1159
1160
1161
1162
1163
1164
1165
1166
1167
1168
1169
1170
1171
1172
1173
1174
1175
1176
1177
1178
1179
1180
1181
1182
1183
1184
1185
1186
1187
1188
1189
1190
1191
1192
1193
1194
1195
1196
1197
1198
1199
1200
1201
1202
1203
1204
1205
1206
1207
1208
1209
1210
1211
1212
1213
1214
1215
1216
1217
1218
1219
1220
1221
1222
1223
1224
1225
1226
1227
1228
1229
1230
1231
1232
1233
1234
1235
1236
1237
1238
1239
1240
1241
1242
1243
1244
1245
1246
1247
1248
1249
1250
1251
1252
1253
1254
1255
1256
1257
1258
1259
1260
1261
1262
1263
1264
1265
1266
1267
1268
1269
1270
1271
1272
1273
1274
1275
1276
1277
1278
1279
1280
1281
1282
1283
1284
1285
1286
1287
1288
1289
1290
1291
1292
1293
1294
1295
1296
1297
1298
1299
1300
1301
1302
1303
1304
1305
1306
1307
1308
1309
1310
1311
1312
1313
1314
1315
1316
1317
1318
1319
1320
1321
1322
1323
1324
1325
1326
1327
1328
1329
1330
1331
1332
1333
1334
1335
1336
1337
1338
1339
1340
1341
1342
1343
1344
1345
1346
1347
1348
1349
1350
1351
1352
1353
1354
1355
1356
1357
1358
1359
1360
1361
1362
1363
1364
1365
1366
1367
1368
1369
1370
1371
1372
1373
1374
1375
1376
1377
1378
1379
1380
1381
1382
1383
1384
1385
1386
1387
1388
1389
1390
1391
1392
1393
1394
1395
1396
1397
1398
1399
1400
1401
1402
1403
1404
1405
1406
1407
1408
1409
1410
1411
1412
1413
1414
1415
1416
1417
1418
1419
1420
1421
1422
1423
1424
1425
1426
1427
1428
1429
1430
1431
1432
1433
1434
1435
1436
1437
1438
1439
1440
1441
1442
1443
1444
1445
1446
1447
1448
1449
1450
1451
1452
1453
1454
1455
1456
1457
1458
1459
1460
1461
1462
1463
1464
1465
1466
1467
1468
1469
1470
1471
1472
1473
1474
1475
1476
1477
1478
1479
1480
1481
1482
1483
1484
1485
1486
1487
1488
1489
1490
1491
1492
1493
1494
1495
1496
1497
1498
1499
1500
1501
1502
1503
1504
1505
1506
1507
1508
1509
1510
1511
1512
1513
1514
1515
1516
1517
/*-
 * Copyright (c) 2008 Attilio Rao <attilio@FreeBSD.org>
 * All rights reserved.
 *
 * Redistribution and use in source and binary forms, with or without
 * modification, are permitted provided that the following conditions
 * are met:
 * 1. Redistributions of source code must retain the above copyright
 *    notice(s), this list of conditions and the following disclaimer as
 *    the first lines of this file unmodified other than the possible
 *    addition of one or more copyright notices.
 * 2. Redistributions in binary form must reproduce the above copyright
 *    notice(s), this list of conditions and the following disclaimer in the
 *    documentation and/or other materials provided with the distribution.
 *
 * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY
 * EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED
 * WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
 * DISCLAIMED.  IN NO EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY
 * DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES
 * (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR
 * SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER
 * CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH
 * DAMAGE.
 */

#include "opt_adaptive_lockmgrs.h"
#include "opt_ddb.h"
#include "opt_hwpmc_hooks.h"

#include <sys/cdefs.h>
__FBSDID("$FreeBSD$");

#include <sys/param.h>
#include <sys/kdb.h>
#include <sys/ktr.h>
#include <sys/lock.h>
#include <sys/lock_profile.h>
#include <sys/lockmgr.h>
#include <sys/mutex.h>
#include <sys/proc.h>
#include <sys/sleepqueue.h>
#ifdef DEBUG_LOCKS
#include <sys/stack.h>
#endif
#include <sys/sysctl.h>
#include <sys/systm.h>

#include <machine/cpu.h>

#ifdef DDB
#include <ddb/ddb.h>
#endif

#ifdef HWPMC_HOOKS
#include <sys/pmckern.h>
PMC_SOFT_DECLARE( , , lock, failed);
#endif

CTASSERT(((LK_ADAPTIVE | LK_NOSHARE) & LO_CLASSFLAGS) ==
    (LK_ADAPTIVE | LK_NOSHARE));
CTASSERT(LK_UNLOCKED == (LK_UNLOCKED &
    ~(LK_ALL_WAITERS | LK_EXCLUSIVE_SPINNERS)));

#define	SQ_EXCLUSIVE_QUEUE	0
#define	SQ_SHARED_QUEUE		1

#ifndef INVARIANTS
#define	_lockmgr_assert(lk, what, file, line)
#define	TD_LOCKS_INC(td)
#define	TD_LOCKS_DEC(td)
#else
#define	TD_LOCKS_INC(td)	((td)->td_locks++)
#define	TD_LOCKS_DEC(td)	((td)->td_locks--)
#endif
#define	TD_SLOCKS_INC(td)	((td)->td_lk_slocks++)
#define	TD_SLOCKS_DEC(td)	((td)->td_lk_slocks--)

#ifndef DEBUG_LOCKS
#define	STACK_PRINT(lk)
#define	STACK_SAVE(lk)
#define	STACK_ZERO(lk)
#else
#define	STACK_PRINT(lk)	stack_print_ddb(&(lk)->lk_stack)
#define	STACK_SAVE(lk)	stack_save(&(lk)->lk_stack)
#define	STACK_ZERO(lk)	stack_zero(&(lk)->lk_stack)
#endif

#define	LOCK_LOG2(lk, string, arg1, arg2)				\
	if (LOCK_LOG_TEST(&(lk)->lock_object, 0))			\
		CTR2(KTR_LOCK, (string), (arg1), (arg2))
#define	LOCK_LOG3(lk, string, arg1, arg2, arg3)				\
	if (LOCK_LOG_TEST(&(lk)->lock_object, 0))			\
		CTR3(KTR_LOCK, (string), (arg1), (arg2), (arg3))

#define	GIANT_DECLARE							\
	int _i = 0;							\
	WITNESS_SAVE_DECL(Giant)
#define	GIANT_RESTORE() do {						\
	if (_i > 0) {							\
		while (_i--)						\
			mtx_lock(&Giant);				\
		WITNESS_RESTORE(&Giant.lock_object, Giant);		\
	}								\
} while (0)
#define	GIANT_SAVE() do {						\
	if (mtx_owned(&Giant)) {					\
		WITNESS_SAVE(&Giant.lock_object, Giant);		\
		while (mtx_owned(&Giant)) {				\
			_i++;						\
			mtx_unlock(&Giant);				\
		}							\
	}								\
} while (0)

#define	LK_CAN_SHARE(x)							\
	(((x) & LK_SHARE) && (((x) & LK_EXCLUSIVE_WAITERS) == 0 ||	\
	((x) & LK_EXCLUSIVE_SPINNERS) == 0 ||				\
	curthread->td_lk_slocks || (curthread->td_pflags & TDP_DEADLKTREAT)))
#define	LK_TRYOP(x)							\
	((x) & LK_NOWAIT)

#define	LK_CAN_WITNESS(x)						\
	(((x) & LK_NOWITNESS) == 0 && !LK_TRYOP(x))
#define	LK_TRYWIT(x)							\
	(LK_TRYOP(x) ? LOP_TRYLOCK : 0)

#define	LK_CAN_ADAPT(lk, f)						\
	(((lk)->lock_object.lo_flags & LK_ADAPTIVE) != 0 &&		\
	((f) & LK_SLEEPFAIL) == 0)

#define	lockmgr_disowned(lk)						\
	(((lk)->lk_lock & ~(LK_FLAGMASK & ~LK_SHARE)) == LK_KERNPROC)

#define	lockmgr_xlocked(lk)						\
	(((lk)->lk_lock & ~(LK_FLAGMASK & ~LK_SHARE)) == (uintptr_t)curthread)

static void	assert_lockmgr(const struct lock_object *lock, int how);
#ifdef DDB
static void	db_show_lockmgr(const struct lock_object *lock);
#endif
static void	lock_lockmgr(struct lock_object *lock, uintptr_t how);
#ifdef KDTRACE_HOOKS
static int	owner_lockmgr(const struct lock_object *lock,
		    struct thread **owner);
#endif
static uintptr_t unlock_lockmgr(struct lock_object *lock);

struct lock_class lock_class_lockmgr = {
	.lc_name = "lockmgr",
	.lc_flags = LC_RECURSABLE | LC_SLEEPABLE | LC_SLEEPLOCK | LC_UPGRADABLE,
	.lc_assert = assert_lockmgr,
#ifdef DDB
	.lc_ddb_show = db_show_lockmgr,
#endif
	.lc_lock = lock_lockmgr,
	.lc_unlock = unlock_lockmgr,
#ifdef KDTRACE_HOOKS
	.lc_owner = owner_lockmgr,
#endif
};

#ifdef ADAPTIVE_LOCKMGRS
static u_int alk_retries = 10;
static u_int alk_loops = 10000;
static SYSCTL_NODE(_debug, OID_AUTO, lockmgr, CTLFLAG_RD, NULL,
    "lockmgr debugging");
SYSCTL_UINT(_debug_lockmgr, OID_AUTO, retries, CTLFLAG_RW, &alk_retries, 0, "");
SYSCTL_UINT(_debug_lockmgr, OID_AUTO, loops, CTLFLAG_RW, &alk_loops, 0, "");
#endif

static __inline struct thread *
lockmgr_xholder(const struct lock *lk)
{
	uintptr_t x;

	x = lk->lk_lock;
	return ((x & LK_SHARE) ? NULL : (struct thread *)LK_HOLDER(x));
}

/*
 * It assumes sleepq_lock held and returns with this one unheld.
 * It also assumes the generic interlock is sane and previously checked.
 * If LK_INTERLOCK is specified the interlock is not reacquired after the
 * sleep.
 */
static __inline int
sleeplk(struct lock *lk, u_int flags, struct lock_object *ilk,
    const char *wmesg, int pri, int timo, int queue)
{
	GIANT_DECLARE;
	struct lock_class *class;
	int catch, error;

	class = (flags & LK_INTERLOCK) ? LOCK_CLASS(ilk) : NULL;
	catch = pri & PCATCH;
	pri &= PRIMASK;
	error = 0;

	LOCK_LOG3(lk, "%s: %p blocking on the %s sleepqueue", __func__, lk,
	    (queue == SQ_EXCLUSIVE_QUEUE) ? "exclusive" : "shared");

	if (flags & LK_INTERLOCK)
		class->lc_unlock(ilk);
	if (queue == SQ_EXCLUSIVE_QUEUE && (flags & LK_SLEEPFAIL) != 0)
		lk->lk_exslpfail++;
	GIANT_SAVE();
	sleepq_add(&lk->lock_object, NULL, wmesg, SLEEPQ_LK | (catch ?
	    SLEEPQ_INTERRUPTIBLE : 0), queue);
	if ((flags & LK_TIMELOCK) && timo)
		sleepq_set_timeout(&lk->lock_object, timo);

	/*
	 * Decisional switch for real sleeping.
	 */
	if ((flags & LK_TIMELOCK) && timo && catch)
		error = sleepq_timedwait_sig(&lk->lock_object, pri);
	else if ((flags & LK_TIMELOCK) && timo)
		error = sleepq_timedwait(&lk->lock_object, pri);
	else if (catch)
		error = sleepq_wait_sig(&lk->lock_object, pri);
	else
		sleepq_wait(&lk->lock_object, pri);
	GIANT_RESTORE();
	if ((flags & LK_SLEEPFAIL) && error == 0)
		error = ENOLCK;

	return (error);
}

static __inline int
wakeupshlk(struct lock *lk, const char *file, int line)
{
	uintptr_t v, x;
	u_int realexslp;
	int queue, wakeup_swapper;

	WITNESS_UNLOCK(&lk->lock_object, 0, file, line);
	LOCK_LOG_LOCK("SUNLOCK", &lk->lock_object, 0, 0, file, line);

	wakeup_swapper = 0;
	for (;;) {
		x = lk->lk_lock;

		/*
		 * If there is more than one shared lock held, just drop one
		 * and return.
		 */
		if (LK_SHARERS(x) > 1) {
			if (atomic_cmpset_rel_ptr(&lk->lk_lock, x,
			    x - LK_ONE_SHARER))
				break;
			continue;
		}

		/*
		 * If there are not waiters on the exclusive queue, drop the
		 * lock quickly.
		 */
		if ((x & LK_ALL_WAITERS) == 0) {
			MPASS((x & ~LK_EXCLUSIVE_SPINNERS) ==
			    LK_SHARERS_LOCK(1));
			if (atomic_cmpset_rel_ptr(&lk->lk_lock, x, LK_UNLOCKED))
				break;
			continue;
		}

		/*
		 * We should have a sharer with waiters, so enter the hard
		 * path in order to handle wakeups correctly.
		 */
		sleepq_lock(&lk->lock_object);
		x = lk->lk_lock & (LK_ALL_WAITERS | LK_EXCLUSIVE_SPINNERS);
		v = LK_UNLOCKED;

		/*
		 * If the lock has exclusive waiters, give them preference in
		 * order to avoid deadlock with shared runners up.
		 * If interruptible sleeps left the exclusive queue empty
		 * avoid a starvation for the threads sleeping on the shared
		 * queue by giving them precedence and cleaning up the
		 * exclusive waiters bit anyway.
		 * Please note that lk_exslpfail count may be lying about
		 * the real number of waiters with the LK_SLEEPFAIL flag on
		 * because they may be used in conjuction with interruptible
		 * sleeps so lk_exslpfail might be considered an 'upper limit'
		 * bound, including the edge cases.
		 */
		realexslp = sleepq_sleepcnt(&lk->lock_object,
		    SQ_EXCLUSIVE_QUEUE);
		if ((x & LK_EXCLUSIVE_WAITERS) != 0 && realexslp != 0) {
			if (lk->lk_exslpfail < realexslp) {
				lk->lk_exslpfail = 0;
				queue = SQ_EXCLUSIVE_QUEUE;
				v |= (x & LK_SHARED_WAITERS);
			} else {
				lk->lk_exslpfail = 0;
				LOCK_LOG2(lk,
				    "%s: %p has only LK_SLEEPFAIL sleepers",
				    __func__, lk);
				LOCK_LOG2(lk,
			    "%s: %p waking up threads on the exclusive queue",
				    __func__, lk);
				wakeup_swapper =
				    sleepq_broadcast(&lk->lock_object,
				    SLEEPQ_LK, 0, SQ_EXCLUSIVE_QUEUE);
				queue = SQ_SHARED_QUEUE;
			}
				
		} else {

			/*
			 * Exclusive waiters sleeping with LK_SLEEPFAIL on
			 * and using interruptible sleeps/timeout may have
			 * left spourious lk_exslpfail counts on, so clean
			 * it up anyway.
			 */
			lk->lk_exslpfail = 0;
			queue = SQ_SHARED_QUEUE;
		}

		if (!atomic_cmpset_rel_ptr(&lk->lk_lock, LK_SHARERS_LOCK(1) | x,
		    v)) {
			sleepq_release(&lk->lock_object);
			continue;
		}
		LOCK_LOG3(lk, "%s: %p waking up threads on the %s queue",
		    __func__, lk, queue == SQ_SHARED_QUEUE ? "shared" :
		    "exclusive");
		wakeup_swapper |= sleepq_broadcast(&lk->lock_object, SLEEPQ_LK,
		    0, queue);
		sleepq_release(&lk->lock_object);
		break;
	}

	lock_profile_release_lock(&lk->lock_object);
	TD_LOCKS_DEC(curthread);
	TD_SLOCKS_DEC(curthread);
	return (wakeup_swapper);
}

static void
assert_lockmgr(const struct lock_object *lock, int what)
{

	panic("lockmgr locks do not support assertions");
}

static void
lock_lockmgr(struct lock_object *lock, uintptr_t how)
{

	panic("lockmgr locks do not support sleep interlocking");
}

static uintptr_t
unlock_lockmgr(struct lock_object *lock)
{

	panic("lockmgr locks do not support sleep interlocking");
}

#ifdef KDTRACE_HOOKS
static int
owner_lockmgr(const struct lock_object *lock, struct thread **owner)
{

	panic("lockmgr locks do not support owner inquiring");
}
#endif

void
lockinit(struct lock *lk, int pri, const char *wmesg, int timo, int flags)
{
	int iflags;

	MPASS((flags & ~LK_INIT_MASK) == 0);
	ASSERT_ATOMIC_LOAD_PTR(lk->lk_lock,
            ("%s: lockmgr not aligned for %s: %p", __func__, wmesg,
            &lk->lk_lock));

	iflags = LO_SLEEPABLE | LO_UPGRADABLE;
	if (flags & LK_CANRECURSE)
		iflags |= LO_RECURSABLE;
	if ((flags & LK_NODUP) == 0)
		iflags |= LO_DUPOK;
	if (flags & LK_NOPROFILE)
		iflags |= LO_NOPROFILE;
	if ((flags & LK_NOWITNESS) == 0)
		iflags |= LO_WITNESS;
	if (flags & LK_QUIET)
		iflags |= LO_QUIET;
	if (flags & LK_IS_VNODE)
		iflags |= LO_IS_VNODE;
	iflags |= flags & (LK_ADAPTIVE | LK_NOSHARE);

	lock_init(&lk->lock_object, &lock_class_lockmgr, wmesg, NULL, iflags);
	lk->lk_lock = LK_UNLOCKED;
	lk->lk_recurse = 0;
	lk->lk_exslpfail = 0;
	lk->lk_timo = timo;
	lk->lk_pri = pri;
	STACK_ZERO(lk);
}

/*
 * XXX: Gross hacks to manipulate external lock flags after
 * initialization.  Used for certain vnode and buf locks.
 */
void
lockallowshare(struct lock *lk)
{

	lockmgr_assert(lk, KA_XLOCKED);
	lk->lock_object.lo_flags &= ~LK_NOSHARE;
}

void
lockallowrecurse(struct lock *lk)
{

	lockmgr_assert(lk, KA_XLOCKED);
	lk->lock_object.lo_flags |= LO_RECURSABLE;
}

void
lockdisablerecurse(struct lock *lk)
{

	lockmgr_assert(lk, KA_XLOCKED);
	lk->lock_object.lo_flags &= ~LO_RECURSABLE;
}

void
lockdestroy(struct lock *lk)
{

	KASSERT(lk->lk_lock == LK_UNLOCKED, ("lockmgr still held"));
	KASSERT(lk->lk_recurse == 0, ("lockmgr still recursed"));
	KASSERT(lk->lk_exslpfail == 0, ("lockmgr still exclusive waiters"));
	lock_destroy(&lk->lock_object);
}

int
__lockmgr_args(struct lock *lk, u_int flags, struct lock_object *ilk,
    const char *wmesg, int pri, int timo, const char *file, int line)
{
	GIANT_DECLARE;
	struct lock_class *class;
	const char *iwmesg;
	uintptr_t tid, v, x;
	u_int op, realexslp;
	int error, ipri, itimo, queue, wakeup_swapper;
#ifdef LOCK_PROFILING
	uint64_t waittime = 0;
	int contested = 0;
#endif
#ifdef ADAPTIVE_LOCKMGRS
	volatile struct thread *owner;
	u_int i, spintries = 0;
#endif

	error = 0;
	tid = (uintptr_t)curthread;
	op = (flags & LK_TYPE_MASK);
	iwmesg = (wmesg == LK_WMESG_DEFAULT) ? lk->lock_object.lo_name : wmesg;
	ipri = (pri == LK_PRIO_DEFAULT) ? lk->lk_pri : pri;
	itimo = (timo == LK_TIMO_DEFAULT) ? lk->lk_timo : timo;

	MPASS((flags & ~LK_TOTAL_MASK) == 0);
	KASSERT((op & (op - 1)) == 0,
	    ("%s: Invalid requested operation @ %s:%d", __func__, file, line));
	KASSERT((flags & (LK_NOWAIT | LK_SLEEPFAIL)) == 0 ||
	    (op != LK_DOWNGRADE && op != LK_RELEASE),
	    ("%s: Invalid flags in regard of the operation desired @ %s:%d",
	    __func__, file, line));
	KASSERT((flags & LK_INTERLOCK) == 0 || ilk != NULL,
	    ("%s: LK_INTERLOCK passed without valid interlock @ %s:%d",
	    __func__, file, line));
	KASSERT(kdb_active != 0 || !TD_IS_IDLETHREAD(curthread),
	    ("%s: idle thread %p on lockmgr %s @ %s:%d", __func__, curthread,
	    lk->lock_object.lo_name, file, line));

	class = (flags & LK_INTERLOCK) ? LOCK_CLASS(ilk) : NULL;
	if (panicstr != NULL) {
		if (flags & LK_INTERLOCK)
			class->lc_unlock(ilk);
		return (0);
	}

	if (lk->lock_object.lo_flags & LK_NOSHARE) {
		switch (op) {
		case LK_SHARED:
			op = LK_EXCLUSIVE;
			break;
		case LK_UPGRADE:
		case LK_TRYUPGRADE:
		case LK_DOWNGRADE:
			_lockmgr_assert(lk, KA_XLOCKED | KA_NOTRECURSED,
			    file, line);
			if (flags & LK_INTERLOCK)
				class->lc_unlock(ilk);
			return (0);
		}
	}

	wakeup_swapper = 0;
	switch (op) {
	case LK_SHARED:
		if (LK_CAN_WITNESS(flags))
			WITNESS_CHECKORDER(&lk->lock_object, LOP_NEWORDER,
			    file, line, flags & LK_INTERLOCK ? ilk : NULL);
		for (;;) {
			x = lk->lk_lock;

			/*
			 * If no other thread has an exclusive lock, or
			 * no exclusive waiter is present, bump the count of
			 * sharers.  Since we have to preserve the state of
			 * waiters, if we fail to acquire the shared lock
			 * loop back and retry.
			 */
			if (LK_CAN_SHARE(x)) {
				if (atomic_cmpset_acq_ptr(&lk->lk_lock, x,
				    x + LK_ONE_SHARER))
					break;
				continue;
			}
#ifdef HWPMC_HOOKS
			PMC_SOFT_CALL( , , lock, failed);
#endif
			lock_profile_obtain_lock_failed(&lk->lock_object,
			    &contested, &waittime);

			/*
			 * If the lock is already held by curthread in
			 * exclusive way avoid a deadlock.
			 */
			if (LK_HOLDER(x) == tid) {
				LOCK_LOG2(lk,
				    "%s: %p already held in exclusive mode",
				    __func__, lk);
				error = EDEADLK;
				break;
			}

			/*
			 * If the lock is expected to not sleep just give up
			 * and return.
			 */
			if (LK_TRYOP(flags)) {
				LOCK_LOG2(lk, "%s: %p fails the try operation",
				    __func__, lk);
				error = EBUSY;
				break;
			}

#ifdef ADAPTIVE_LOCKMGRS
			/*
			 * If the owner is running on another CPU, spin until
			 * the owner stops running or the state of the lock
			 * changes.  We need a double-state handle here
			 * because for a failed acquisition the lock can be
			 * either held in exclusive mode or shared mode
			 * (for the writer starvation avoidance technique).
			 */
			if (LK_CAN_ADAPT(lk, flags) && (x & LK_SHARE) == 0 &&
			    LK_HOLDER(x) != LK_KERNPROC) {
				owner = (struct thread *)LK_HOLDER(x);
				if (LOCK_LOG_TEST(&lk->lock_object, 0))
					CTR3(KTR_LOCK,
					    "%s: spinning on %p held by %p",
					    __func__, lk, owner);

				/*
				 * If we are holding also an interlock drop it
				 * in order to avoid a deadlock if the lockmgr
				 * owner is adaptively spinning on the
				 * interlock itself.
				 */
				if (flags & LK_INTERLOCK) {
					class->lc_unlock(ilk);
					flags &= ~LK_INTERLOCK;
				}
				GIANT_SAVE();
				while (LK_HOLDER(lk->lk_lock) ==
				    (uintptr_t)owner && TD_IS_RUNNING(owner))
					cpu_spinwait();
				GIANT_RESTORE();
				continue;
			} else if (LK_CAN_ADAPT(lk, flags) &&
			    (x & LK_SHARE) != 0 && LK_SHARERS(x) &&
			    spintries < alk_retries) {
				if (flags & LK_INTERLOCK) {
					class->lc_unlock(ilk);
					flags &= ~LK_INTERLOCK;
				}
				GIANT_SAVE();
				spintries++;
				for (i = 0; i < alk_loops; i++) {
					if (LOCK_LOG_TEST(&lk->lock_object, 0))
						CTR4(KTR_LOCK,
				    "%s: shared spinning on %p with %u and %u",
						    __func__, lk, spintries, i);
					x = lk->lk_lock;
					if ((x & LK_SHARE) == 0 ||
					    LK_CAN_SHARE(x) != 0)
						break;
					cpu_spinwait();
				}
				GIANT_RESTORE();
				if (i != alk_loops)
					continue;
			}
#endif

			/*
			 * Acquire the sleepqueue chain lock because we
			 * probabilly will need to manipulate waiters flags.
			 */
			sleepq_lock(&lk->lock_object);
			x = lk->lk_lock;

			/*
			 * if the lock can be acquired in shared mode, try
			 * again.
			 */
			if (LK_CAN_SHARE(x)) {
				sleepq_release(&lk->lock_object);
				continue;
			}

#ifdef ADAPTIVE_LOCKMGRS
			/*
			 * The current lock owner might have started executing
			 * on another CPU (or the lock could have changed
			 * owner) while we were waiting on the turnstile
			 * chain lock.  If so, drop the turnstile lock and try
			 * again.
			 */
			if (LK_CAN_ADAPT(lk, flags) && (x & LK_SHARE) == 0 &&
			    LK_HOLDER(x) != LK_KERNPROC) {
				owner = (struct thread *)LK_HOLDER(x);
				if (TD_IS_RUNNING(owner)) {
					sleepq_release(&lk->lock_object);
					continue;
				}
			}
#endif

			/*
			 * Try to set the LK_SHARED_WAITERS flag.  If we fail,
			 * loop back and retry.
			 */
			if ((x & LK_SHARED_WAITERS) == 0) {
				if (!atomic_cmpset_acq_ptr(&lk->lk_lock, x,
				    x | LK_SHARED_WAITERS)) {
					sleepq_release(&lk->lock_object);
					continue;
				}
				LOCK_LOG2(lk, "%s: %p set shared waiters flag",
				    __func__, lk);
			}

			/*
			 * As far as we have been unable to acquire the
			 * shared lock and the shared waiters flag is set,
			 * we will sleep.
			 */
			error = sleeplk(lk, flags, ilk, iwmesg, ipri, itimo,
			    SQ_SHARED_QUEUE);
			flags &= ~LK_INTERLOCK;
			if (error) {
				LOCK_LOG3(lk,
				    "%s: interrupted sleep for %p with %d",
				    __func__, lk, error);
				break;
			}
			LOCK_LOG2(lk, "%s: %p resuming from the sleep queue",
			    __func__, lk);
		}
		if (error == 0) {
			lock_profile_obtain_lock_success(&lk->lock_object,
			    contested, waittime, file, line);
			LOCK_LOG_LOCK("SLOCK", &lk->lock_object, 0, 0, file,
			    line);
			WITNESS_LOCK(&lk->lock_object, LK_TRYWIT(flags), file,
			    line);
			TD_LOCKS_INC(curthread);
			TD_SLOCKS_INC(curthread);
			STACK_SAVE(lk);
		}
		break;
	case LK_UPGRADE:
	case LK_TRYUPGRADE:
		_lockmgr_assert(lk, KA_SLOCKED, file, line);
		v = lk->lk_lock;
		x = v & LK_ALL_WAITERS;
		v &= LK_EXCLUSIVE_SPINNERS;

		/*
		 * Try to switch from one shared lock to an exclusive one.
		 * We need to preserve waiters flags during the operation.
		 */
		if (atomic_cmpset_ptr(&lk->lk_lock, LK_SHARERS_LOCK(1) | x | v,
		    tid | x)) {
			LOCK_LOG_LOCK("XUPGRADE", &lk->lock_object, 0, 0, file,
			    line);
			WITNESS_UPGRADE(&lk->lock_object, LOP_EXCLUSIVE |
			    LK_TRYWIT(flags), file, line);
			TD_SLOCKS_DEC(curthread);
			break;
		}

		/*
		 * In LK_TRYUPGRADE mode, do not drop the lock,
		 * returning EBUSY instead.
		 */
		if (op == LK_TRYUPGRADE) {
			LOCK_LOG2(lk, "%s: %p failed the nowait upgrade",
			    __func__, lk);
			error = EBUSY;
			break;
		}

		/*
		 * We have been unable to succeed in upgrading, so just
		 * give up the shared lock.
		 */
		wakeup_swapper |= wakeupshlk(lk, file, line);

		/* FALLTHROUGH */
	case LK_EXCLUSIVE:
		if (LK_CAN_WITNESS(flags))
			WITNESS_CHECKORDER(&lk->lock_object, LOP_NEWORDER |
			    LOP_EXCLUSIVE, file, line, flags & LK_INTERLOCK ?
			    ilk : NULL);

		/*
		 * If curthread already holds the lock and this one is
		 * allowed to recurse, simply recurse on it.
		 */
		if (lockmgr_xlocked(lk)) {
			if ((flags & LK_CANRECURSE) == 0 &&
			    (lk->lock_object.lo_flags & LO_RECURSABLE) == 0) {

				/*
				 * If the lock is expected to not panic just
				 * give up and return.
				 */
				if (LK_TRYOP(flags)) {
					LOCK_LOG2(lk,
					    "%s: %p fails the try operation",
					    __func__, lk);
					error = EBUSY;
					break;
				}
				if (flags & LK_INTERLOCK)
					class->lc_unlock(ilk);
		panic("%s: recursing on non recursive lockmgr %s @ %s:%d\n",
				    __func__, iwmesg, file, line);
			}
			lk->lk_recurse++;
			LOCK_LOG2(lk, "%s: %p recursing", __func__, lk);
			LOCK_LOG_LOCK("XLOCK", &lk->lock_object, 0,
			    lk->lk_recurse, file, line);
			WITNESS_LOCK(&lk->lock_object, LOP_EXCLUSIVE |
			    LK_TRYWIT(flags), file, line);
			TD_LOCKS_INC(curthread);
			break;
		}

		while (!atomic_cmpset_acq_ptr(&lk->lk_lock, LK_UNLOCKED,
		    tid)) {
#ifdef HWPMC_HOOKS
			PMC_SOFT_CALL( , , lock, failed);
#endif
			lock_profile_obtain_lock_failed(&lk->lock_object,
			    &contested, &waittime);

			/*
			 * If the lock is expected to not sleep just give up
			 * and return.
			 */
			if (LK_TRYOP(flags)) {
				LOCK_LOG2(lk, "%s: %p fails the try operation",
				    __func__, lk);
				error = EBUSY;
				break;
			}

#ifdef ADAPTIVE_LOCKMGRS
			/*
			 * If the owner is running on another CPU, spin until
			 * the owner stops running or the state of the lock
			 * changes.
			 */
			x = lk->lk_lock;
			if (LK_CAN_ADAPT(lk, flags) && (x & LK_SHARE) == 0 &&
			    LK_HOLDER(x) != LK_KERNPROC) {
				owner = (struct thread *)LK_HOLDER(x);
				if (LOCK_LOG_TEST(&lk->lock_object, 0))
					CTR3(KTR_LOCK,
					    "%s: spinning on %p held by %p",
					    __func__, lk, owner);

				/*
				 * If we are holding also an interlock drop it
				 * in order to avoid a deadlock if the lockmgr
				 * owner is adaptively spinning on the
				 * interlock itself.
				 */
				if (flags & LK_INTERLOCK) {
					class->lc_unlock(ilk);
					flags &= ~LK_INTERLOCK;
				}
				GIANT_SAVE();
				while (LK_HOLDER(lk->lk_lock) ==
				    (uintptr_t)owner && TD_IS_RUNNING(owner))
					cpu_spinwait();
				GIANT_RESTORE();
				continue;
			} else if (LK_CAN_ADAPT(lk, flags) &&
			    (x & LK_SHARE) != 0 && LK_SHARERS(x) &&
			    spintries < alk_retries) {
				if ((x & LK_EXCLUSIVE_SPINNERS) == 0 &&
				    !atomic_cmpset_ptr(&lk->lk_lock, x,
				    x | LK_EXCLUSIVE_SPINNERS))
					continue;
				if (flags & LK_INTERLOCK) {
					class->lc_unlock(ilk);
					flags &= ~LK_INTERLOCK;
				}
				GIANT_SAVE();
				spintries++;
				for (i = 0; i < alk_loops; i++) {
					if (LOCK_LOG_TEST(&lk->lock_object, 0))
						CTR4(KTR_LOCK,
				    "%s: shared spinning on %p with %u and %u",
						    __func__, lk, spintries, i);
					if ((lk->lk_lock &
					    LK_EXCLUSIVE_SPINNERS) == 0)
						break;
					cpu_spinwait();
				}
				GIANT_RESTORE();
				if (i != alk_loops)
					continue;
			}
#endif

			/*
			 * Acquire the sleepqueue chain lock because we
			 * probabilly will need to manipulate waiters flags.
			 */
			sleepq_lock(&lk->lock_object);
			x = lk->lk_lock;

			/*
			 * if the lock has been released while we spun on
			 * the sleepqueue chain lock just try again.
			 */
			if (x == LK_UNLOCKED) {
				sleepq_release(&lk->lock_object);
				continue;
			}

#ifdef ADAPTIVE_LOCKMGRS
			/*
			 * The current lock owner might have started executing
			 * on another CPU (or the lock could have changed
			 * owner) while we were waiting on the turnstile
			 * chain lock.  If so, drop the turnstile lock and try
			 * again.
			 */
			if (LK_CAN_ADAPT(lk, flags) && (x & LK_SHARE) == 0 &&
			    LK_HOLDER(x) != LK_KERNPROC) {
				owner = (struct thread *)LK_HOLDER(x);
				if (TD_IS_RUNNING(owner)) {
					sleepq_release(&lk->lock_object);
					continue;
				}
			}
#endif

			/*
			 * The lock can be in the state where there is a
			 * pending queue of waiters, but still no owner.
			 * This happens when the lock is contested and an
			 * owner is going to claim the lock.
			 * If curthread is the one successfully acquiring it
			 * claim lock ownership and return, preserving waiters
			 * flags.
			 */
			v = x & (LK_ALL_WAITERS | LK_EXCLUSIVE_SPINNERS);
			if ((x & ~v) == LK_UNLOCKED) {
				v &= ~LK_EXCLUSIVE_SPINNERS;
				if (atomic_cmpset_acq_ptr(&lk->lk_lock, x,
				    tid | v)) {
					sleepq_release(&lk->lock_object);
					LOCK_LOG2(lk,
					    "%s: %p claimed by a new writer",
					    __func__, lk);
					break;
				}
				sleepq_release(&lk->lock_object);
				continue;
			}

			/*
			 * Try to set the LK_EXCLUSIVE_WAITERS flag.  If we
			 * fail, loop back and retry.
			 */
			if ((x & LK_EXCLUSIVE_WAITERS) == 0) {
				if (!atomic_cmpset_ptr(&lk->lk_lock, x,
				    x | LK_EXCLUSIVE_WAITERS)) {
					sleepq_release(&lk->lock_object);
					continue;
				}
				LOCK_LOG2(lk, "%s: %p set excl waiters flag",
				    __func__, lk);
			}

			/*
			 * As far as we have been unable to acquire the
			 * exclusive lock and the exclusive waiters flag
			 * is set, we will sleep.
			 */
			error = sleeplk(lk, flags, ilk, iwmesg, ipri, itimo,
			    SQ_EXCLUSIVE_QUEUE);
			flags &= ~LK_INTERLOCK;
			if (error) {
				LOCK_LOG3(lk,
				    "%s: interrupted sleep for %p with %d",
				    __func__, lk, error);
				break;
			}
			LOCK_LOG2(lk, "%s: %p resuming from the sleep queue",
			    __func__, lk);
		}
		if (error == 0) {
			lock_profile_obtain_lock_success(&lk->lock_object,
			    contested, waittime, file, line);
			LOCK_LOG_LOCK("XLOCK", &lk->lock_object, 0,
			    lk->lk_recurse, file, line);
			WITNESS_LOCK(&lk->lock_object, LOP_EXCLUSIVE |
			    LK_TRYWIT(flags), file, line);
			TD_LOCKS_INC(curthread);
			STACK_SAVE(lk);
		}
		break;
	case LK_DOWNGRADE:
		_lockmgr_assert(lk, KA_XLOCKED, file, line);
		LOCK_LOG_LOCK("XDOWNGRADE", &lk->lock_object, 0, 0, file, line);
		WITNESS_DOWNGRADE(&lk->lock_object, 0, file, line);

		/*
		 * Panic if the lock is recursed.
		 */
		if (lockmgr_xlocked(lk) && lockmgr_recursed(lk)) {
			if (flags & LK_INTERLOCK)
				class->lc_unlock(ilk);
			panic("%s: downgrade a recursed lockmgr %s @ %s:%d\n",
			    __func__, iwmesg, file, line);
		}
		TD_SLOCKS_INC(curthread);

		/*
		 * In order to preserve waiters flags, just spin.
		 */
		for (;;) {
			x = lk->lk_lock;
			MPASS((x & LK_EXCLUSIVE_SPINNERS) == 0);
			x &= LK_ALL_WAITERS;
			if (atomic_cmpset_rel_ptr(&lk->lk_lock, tid | x,
			    LK_SHARERS_LOCK(1) | x))
				break;
			cpu_spinwait();
		}
		break;
	case LK_RELEASE:
		_lockmgr_assert(lk, KA_LOCKED, file, line);
		x = lk->lk_lock;

		if ((x & LK_SHARE) == 0) {

			/*
			 * As first option, treact the lock as if it has not
			 * any waiter.
			 * Fix-up the tid var if the lock has been disowned.
			 */
			if (LK_HOLDER(x) == LK_KERNPROC)
				tid = LK_KERNPROC;
			else {
				WITNESS_UNLOCK(&lk->lock_object, LOP_EXCLUSIVE,
				    file, line);
				TD_LOCKS_DEC(curthread);
			}
			LOCK_LOG_LOCK("XUNLOCK", &lk->lock_object, 0,
			    lk->lk_recurse, file, line);

			/*
			 * The lock is held in exclusive mode.
			 * If the lock is recursed also, then unrecurse it.
			 */
			if (lockmgr_xlocked(lk) && lockmgr_recursed(lk)) {
				LOCK_LOG2(lk, "%s: %p unrecursing", __func__,
				    lk);
				lk->lk_recurse--;
				break;
			}
			if (tid != LK_KERNPROC)
				lock_profile_release_lock(&lk->lock_object);

			if (atomic_cmpset_rel_ptr(&lk->lk_lock, tid,
			    LK_UNLOCKED))
				break;

			sleepq_lock(&lk->lock_object);
			x = lk->lk_lock;
			v = LK_UNLOCKED;

			/*
		 	 * If the lock has exclusive waiters, give them
			 * preference in order to avoid deadlock with
			 * shared runners up.
			 * If interruptible sleeps left the exclusive queue
			 * empty avoid a starvation for the threads sleeping
			 * on the shared queue by giving them precedence
			 * and cleaning up the exclusive waiters bit anyway.
			 * Please note that lk_exslpfail count may be lying
			 * about the real number of waiters with the
			 * LK_SLEEPFAIL flag on because they may be used in
			 * conjuction with interruptible sleeps so
			 * lk_exslpfail might be considered an 'upper limit'
			 * bound, including the edge cases.
			 */
			MPASS((x & LK_EXCLUSIVE_SPINNERS) == 0);
			realexslp = sleepq_sleepcnt(&lk->lock_object,
			    SQ_EXCLUSIVE_QUEUE);
			if ((x & LK_EXCLUSIVE_WAITERS) != 0 && realexslp != 0) {
				if (lk->lk_exslpfail < realexslp) {
					lk->lk_exslpfail = 0;
					queue = SQ_EXCLUSIVE_QUEUE;
					v |= (x & LK_SHARED_WAITERS);
				} else {
					lk->lk_exslpfail = 0;
					LOCK_LOG2(lk,
					"%s: %p has only LK_SLEEPFAIL sleepers",
					    __func__, lk);
					LOCK_LOG2(lk,
			"%s: %p waking up threads on the exclusive queue",
					    __func__, lk);
					wakeup_swapper =
					    sleepq_broadcast(&lk->lock_object,
					    SLEEPQ_LK, 0, SQ_EXCLUSIVE_QUEUE);
					queue = SQ_SHARED_QUEUE;
				}
			} else {

				/*
				 * Exclusive waiters sleeping with LK_SLEEPFAIL
				 * on and using interruptible sleeps/timeout
				 * may have left spourious lk_exslpfail counts
				 * on, so clean it up anyway. 
				 */
				lk->lk_exslpfail = 0;
				queue = SQ_SHARED_QUEUE;
			}

			LOCK_LOG3(lk,
			    "%s: %p waking up threads on the %s queue",
			    __func__, lk, queue == SQ_SHARED_QUEUE ? "shared" :
			    "exclusive");
			atomic_store_rel_ptr(&lk->lk_lock, v);
			wakeup_swapper |= sleepq_broadcast(&lk->lock_object,
			    SLEEPQ_LK, 0, queue);
			sleepq_release(&lk->lock_object);
			break;
		} else
			wakeup_swapper = wakeupshlk(lk, file, line);
		break;
	case LK_DRAIN:
		if (LK_CAN_WITNESS(flags))
			WITNESS_CHECKORDER(&lk->lock_object, LOP_NEWORDER |
			    LOP_EXCLUSIVE, file, line, flags & LK_INTERLOCK ?
			    ilk : NULL);

		/*
		 * Trying to drain a lock we already own will result in a
		 * deadlock.
		 */
		if (lockmgr_xlocked(lk)) {
			if (flags & LK_INTERLOCK)
				class->lc_unlock(ilk);
			panic("%s: draining %s with the lock held @ %s:%d\n",
			    __func__, iwmesg, file, line);
		}

		while (!atomic_cmpset_acq_ptr(&lk->lk_lock, LK_UNLOCKED, tid)) {
#ifdef HWPMC_HOOKS
			PMC_SOFT_CALL( , , lock, failed);
#endif
			lock_profile_obtain_lock_failed(&lk->lock_object,
			    &contested, &waittime);

			/*
			 * If the lock is expected to not sleep just give up
			 * and return.
			 */
			if (LK_TRYOP(flags)) {
				LOCK_LOG2(lk, "%s: %p fails the try operation",
				    __func__, lk);
				error = EBUSY;
				break;
			}

			/*
			 * Acquire the sleepqueue chain lock because we
			 * probabilly will need to manipulate waiters flags.
			 */
			sleepq_lock(&lk->lock_object);
			x = lk->lk_lock;

			/*
			 * if the lock has been released while we spun on
			 * the sleepqueue chain lock just try again.
			 */
			if (x == LK_UNLOCKED) {
				sleepq_release(&lk->lock_object);
				continue;
			}

			v = x & (LK_ALL_WAITERS | LK_EXCLUSIVE_SPINNERS);
			if ((x & ~v) == LK_UNLOCKED) {
				v = (x & ~LK_EXCLUSIVE_SPINNERS);

				/*
				 * If interruptible sleeps left the exclusive
				 * queue empty avoid a starvation for the
				 * threads sleeping on the shared queue by
				 * giving them precedence and cleaning up the
				 * exclusive waiters bit anyway.
				 * Please note that lk_exslpfail count may be
				 * lying about the real number of waiters with
				 * the LK_SLEEPFAIL flag on because they may
				 * be used in conjuction with interruptible
				 * sleeps so lk_exslpfail might be considered
				 * an 'upper limit' bound, including the edge
				 * cases.
				 */
				if (v & LK_EXCLUSIVE_WAITERS) {
					queue = SQ_EXCLUSIVE_QUEUE;
					v &= ~LK_EXCLUSIVE_WAITERS;
				} else {

					/*
					 * Exclusive waiters sleeping with
					 * LK_SLEEPFAIL on and using
					 * interruptible sleeps/timeout may
					 * have left spourious lk_exslpfail
					 * counts on, so clean it up anyway.
					 */
					MPASS(v & LK_SHARED_WAITERS);
					lk->lk_exslpfail = 0;
					queue = SQ_SHARED_QUEUE;
					v &= ~LK_SHARED_WAITERS;
				}
				if (queue == SQ_EXCLUSIVE_QUEUE) {
					realexslp =
					    sleepq_sleepcnt(&lk->lock_object,
					    SQ_EXCLUSIVE_QUEUE);
					if (lk->lk_exslpfail >= realexslp) {
						lk->lk_exslpfail = 0;
						queue = SQ_SHARED_QUEUE;
						v &= ~LK_SHARED_WAITERS;
						if (realexslp != 0) {
							LOCK_LOG2(lk,
					"%s: %p has only LK_SLEEPFAIL sleepers",
							    __func__, lk);
							LOCK_LOG2(lk,
			"%s: %p waking up threads on the exclusive queue",
							    __func__, lk);
							wakeup_swapper =
							    sleepq_broadcast(
							    &lk->lock_object,
							    SLEEPQ_LK, 0,
							    SQ_EXCLUSIVE_QUEUE);
						}
					} else
						lk->lk_exslpfail = 0;
				}
				if (!atomic_cmpset_ptr(&lk->lk_lock, x, v)) {
					sleepq_release(&lk->lock_object);
					continue;
				}
				LOCK_LOG3(lk,
				"%s: %p waking up all threads on the %s queue",
				    __func__, lk, queue == SQ_SHARED_QUEUE ?
				    "shared" : "exclusive");
				wakeup_swapper |= sleepq_broadcast(
				    &lk->lock_object, SLEEPQ_LK, 0, queue);

				/*
				 * If shared waiters have been woken up we need
				 * to wait for one of them to acquire the lock
				 * before to set the exclusive waiters in
				 * order to avoid a deadlock.
				 */
				if (queue == SQ_SHARED_QUEUE) {
					for (v = lk->lk_lock;
					    (v & LK_SHARE) && !LK_SHARERS(v);
					    v = lk->lk_lock)
						cpu_spinwait();
				}
			}

			/*
			 * Try to set the LK_EXCLUSIVE_WAITERS flag.  If we
			 * fail, loop back and retry.
			 */
			if ((x & LK_EXCLUSIVE_WAITERS) == 0) {
				if (!atomic_cmpset_ptr(&lk->lk_lock, x,
				    x | LK_EXCLUSIVE_WAITERS)) {
					sleepq_release(&lk->lock_object);
					continue;
				}
				LOCK_LOG2(lk, "%s: %p set drain waiters flag",
				    __func__, lk);
			}

			/*
			 * As far as we have been unable to acquire the
			 * exclusive lock and the exclusive waiters flag
			 * is set, we will sleep.
			 */
			if (flags & LK_INTERLOCK) {
				class->lc_unlock(ilk);
				flags &= ~LK_INTERLOCK;
			}
			GIANT_SAVE();
			sleepq_add(&lk->lock_object, NULL, iwmesg, SLEEPQ_LK,
			    SQ_EXCLUSIVE_QUEUE);
			sleepq_wait(&lk->lock_object, ipri & PRIMASK);
			GIANT_RESTORE();
			LOCK_LOG2(lk, "%s: %p resuming from the sleep queue",
			    __func__, lk);
		}

		if (error == 0) {
			lock_profile_obtain_lock_success(&lk->lock_object,
			    contested, waittime, file, line);
			LOCK_LOG_LOCK("DRAIN", &lk->lock_object, 0,
			    lk->lk_recurse, file, line);
			WITNESS_LOCK(&lk->lock_object, LOP_EXCLUSIVE |
			    LK_TRYWIT(flags), file, line);
			TD_LOCKS_INC(curthread);
			STACK_SAVE(lk);
		}
		break;
	default:
		if (flags & LK_INTERLOCK)
			class->lc_unlock(ilk);
		panic("%s: unknown lockmgr request 0x%x\n", __func__, op);
	}

	if (flags & LK_INTERLOCK)
		class->lc_unlock(ilk);
	if (wakeup_swapper)
		kick_proc0();

	return (error);
}

void
_lockmgr_disown(struct lock *lk, const char *file, int line)
{
	uintptr_t tid, x;

	if (SCHEDULER_STOPPED())
		return;

	tid = (uintptr_t)curthread;
	_lockmgr_assert(lk, KA_XLOCKED, file, line);

	/*
	 * Panic if the lock is recursed.
	 */
	if (lockmgr_xlocked(lk) && lockmgr_recursed(lk))
		panic("%s: disown a recursed lockmgr @ %s:%d\n",
		    __func__,  file, line);

	/*
	 * If the owner is already LK_KERNPROC just skip the whole operation.
	 */
	if (LK_HOLDER(lk->lk_lock) != tid)
		return;
	lock_profile_release_lock(&lk->lock_object);
	LOCK_LOG_LOCK("XDISOWN", &lk->lock_object, 0, 0, file, line);
	WITNESS_UNLOCK(&lk->lock_object, LOP_EXCLUSIVE, file, line);
	TD_LOCKS_DEC(curthread);
	STACK_SAVE(lk);

	/*
	 * In order to preserve waiters flags, just spin.
	 */
	for (;;) {
		x = lk->lk_lock;
		MPASS((x & LK_EXCLUSIVE_SPINNERS) == 0);
		x &= LK_ALL_WAITERS;
		if (atomic_cmpset_rel_ptr(&lk->lk_lock, tid | x,
		    LK_KERNPROC | x))
			return;
		cpu_spinwait();
	}
}

void
lockmgr_printinfo(const struct lock *lk)
{
	struct thread *td;
	uintptr_t x;

	if (lk->lk_lock == LK_UNLOCKED)
		printf("lock type %s: UNLOCKED\n", lk->lock_object.lo_name);
	else if (lk->lk_lock & LK_SHARE)
		printf("lock type %s: SHARED (count %ju)\n",
		    lk->lock_object.lo_name,
		    (uintmax_t)LK_SHARERS(lk->lk_lock));
	else {
		td = lockmgr_xholder(lk);
		printf("lock type %s: EXCL by thread %p "
		    "(pid %d, %s, tid %d)\n", lk->lock_object.lo_name, td,
		    td->td_proc->p_pid, td->td_proc->p_comm, td->td_tid);
	}

	x = lk->lk_lock;
	if (x & LK_EXCLUSIVE_WAITERS)
		printf(" with exclusive waiters pending\n");
	if (x & LK_SHARED_WAITERS)
		printf(" with shared waiters pending\n");
	if (x & LK_EXCLUSIVE_SPINNERS)
		printf(" with exclusive spinners pending\n");

	STACK_PRINT(lk);
}

int
lockstatus(const struct lock *lk)
{
	uintptr_t v, x;
	int ret;

	ret = LK_SHARED;
	x = lk->lk_lock;
	v = LK_HOLDER(x);

	if ((x & LK_SHARE) == 0) {
		if (v == (uintptr_t)curthread || v == LK_KERNPROC)
			ret = LK_EXCLUSIVE;
		else
			ret = LK_EXCLOTHER;
	} else if (x == LK_UNLOCKED)
		ret = 0;

	return (ret);
}

#ifdef INVARIANT_SUPPORT

FEATURE(invariant_support,
    "Support for modules compiled with INVARIANTS option");

#ifndef INVARIANTS
#undef	_lockmgr_assert
#endif

void
_lockmgr_assert(const struct lock *lk, int what, const char *file, int line)
{
	int slocked = 0;

	if (panicstr != NULL)
		return;
	switch (what) {
	case KA_SLOCKED:
	case KA_SLOCKED | KA_NOTRECURSED:
	case KA_SLOCKED | KA_RECURSED:
		slocked = 1;
	case KA_LOCKED:
	case KA_LOCKED | KA_NOTRECURSED:
	case KA_LOCKED | KA_RECURSED:
#ifdef WITNESS

		/*
		 * We cannot trust WITNESS if the lock is held in exclusive
		 * mode and a call to lockmgr_disown() happened.
		 * Workaround this skipping the check if the lock is held in
		 * exclusive mode even for the KA_LOCKED case.
		 */
		if (slocked || (lk->lk_lock & LK_SHARE)) {
			witness_assert(&lk->lock_object, what, file, line);
			break;
		}
#endif
		if (lk->lk_lock == LK_UNLOCKED ||
		    ((lk->lk_lock & LK_SHARE) == 0 && (slocked ||
		    (!lockmgr_xlocked(lk) && !lockmgr_disowned(lk)))))
			panic("Lock %s not %slocked @ %s:%d\n",
			    lk->lock_object.lo_name, slocked ? "share" : "",
			    file, line);

		if ((lk->lk_lock & LK_SHARE) == 0) {
			if (lockmgr_recursed(lk)) {
				if (what & KA_NOTRECURSED)
					panic("Lock %s recursed @ %s:%d\n",
					    lk->lock_object.lo_name, file,
					    line);
			} else if (what & KA_RECURSED)
				panic("Lock %s not recursed @ %s:%d\n",
				    lk->lock_object.lo_name, file, line);
		}
		break;
	case KA_XLOCKED:
	case KA_XLOCKED | KA_NOTRECURSED:
	case KA_XLOCKED | KA_RECURSED:
		if (!lockmgr_xlocked(lk) && !lockmgr_disowned(lk))
			panic("Lock %s not exclusively locked @ %s:%d\n",
			    lk->lock_object.lo_name, file, line);
		if (lockmgr_recursed(lk)) {
			if (what & KA_NOTRECURSED)
				panic("Lock %s recursed @ %s:%d\n",
				    lk->lock_object.lo_name, file, line);
		} else if (what & KA_RECURSED)
			panic("Lock %s not recursed @ %s:%d\n",
			    lk->lock_object.lo_name, file, line);
		break;
	case KA_UNLOCKED:
		if (lockmgr_xlocked(lk) || lockmgr_disowned(lk))
			panic("Lock %s exclusively locked @ %s:%d\n",
			    lk->lock_object.lo_name, file, line);
		break;
	default:
		panic("Unknown lockmgr assertion: %d @ %s:%d\n", what, file,
		    line);
	}
}
#endif

#ifdef DDB
int
lockmgr_chain(struct thread *td, struct thread **ownerp)
{
	struct lock *lk;

	lk = td->td_wchan;

	if (LOCK_CLASS(&lk->lock_object) != &lock_class_lockmgr)
		return (0);
	db_printf("blocked on lockmgr %s", lk->lock_object.lo_name);
	if (lk->lk_lock & LK_SHARE)
		db_printf("SHARED (count %ju)\n",
		    (uintmax_t)LK_SHARERS(lk->lk_lock));
	else
		db_printf("EXCL\n");
	*ownerp = lockmgr_xholder(lk);

	return (1);
}

static void
db_show_lockmgr(const struct lock_object *lock)
{
	struct thread *td;
	const struct lock *lk;

	lk = (const struct lock *)lock;

	db_printf(" state: ");
	if (lk->lk_lock == LK_UNLOCKED)
		db_printf("UNLOCKED\n");
	else if (lk->lk_lock & LK_SHARE)
		db_printf("SLOCK: %ju\n", (uintmax_t)LK_SHARERS(lk->lk_lock));
	else {
		td = lockmgr_xholder(lk);
		if (td == (struct thread *)LK_KERNPROC)
			db_printf("XLOCK: LK_KERNPROC\n");
		else
			db_printf("XLOCK: %p (tid %d, pid %d, \"%s\")\n", td,
			    td->td_tid, td->td_proc->p_pid,
			    td->td_proc->p_comm);
		if (lockmgr_recursed(lk))
			db_printf(" recursed: %d\n", lk->lk_recurse);
	}
	db_printf(" waiters: ");
	switch (lk->lk_lock & LK_ALL_WAITERS) {
	case LK_SHARED_WAITERS:
		db_printf("shared\n");
		break;
	case LK_EXCLUSIVE_WAITERS:
		db_printf("exclusive\n");
		break;
	case LK_ALL_WAITERS:
		db_printf("shared and exclusive\n");
		break;
	default:
		db_printf("none\n");
	}
	db_printf(" spinners: ");
	if (lk->lk_lock & LK_EXCLUSIVE_SPINNERS)
		db_printf("exclusive\n");
	else
		db_printf("none\n");
}
#endif
OpenPOWER on IntegriCloud