aboutsummaryrefslogtreecommitdiff
path: root/sys/kern/subr_witness.c
blob: 3019a7a4016dfccb382959168169876805731ed5 (plain) (blame)
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
1518
1519
1520
1521
1522
1523
1524
1525
1526
1527
1528
1529
1530
1531
1532
1533
1534
1535
1536
1537
1538
1539
1540
1541
1542
1543
1544
1545
1546
1547
1548
1549
1550
1551
1552
1553
1554
1555
1556
1557
1558
1559
1560
1561
1562
1563
1564
1565
1566
1567
1568
1569
1570
1571
1572
1573
1574
1575
1576
1577
1578
1579
1580
1581
1582
1583
1584
1585
1586
1587
1588
1589
1590
1591
1592
1593
1594
1595
1596
1597
1598
1599
1600
1601
1602
1603
1604
1605
1606
1607
1608
1609
1610
1611
1612
1613
1614
1615
1616
1617
1618
1619
1620
1621
1622
1623
1624
1625
1626
1627
1628
1629
1630
1631
1632
1633
1634
1635
1636
1637
1638
1639
1640
1641
1642
1643
1644
1645
1646
1647
1648
1649
1650
1651
1652
1653
1654
1655
1656
1657
1658
1659
1660
1661
1662
1663
1664
1665
1666
1667
1668
1669
1670
1671
1672
1673
1674
1675
1676
1677
1678
1679
1680
1681
1682
1683
1684
1685
1686
1687
1688
1689
1690
1691
1692
1693
1694
1695
1696
1697
1698
1699
1700
1701
1702
1703
1704
1705
1706
1707
1708
1709
1710
1711
1712
1713
1714
1715
1716
1717
1718
1719
1720
1721
1722
1723
1724
1725
1726
1727
1728
1729
1730
1731
1732
1733
1734
1735
1736
1737
1738
1739
1740
1741
1742
1743
1744
1745
1746
1747
1748
1749
1750
1751
1752
1753
1754
1755
1756
1757
1758
1759
1760
1761
1762
1763
1764
1765
1766
1767
1768
1769
1770
1771
1772
1773
1774
1775
1776
1777
1778
1779
1780
1781
1782
1783
1784
1785
1786
1787
1788
1789
1790
1791
1792
1793
1794
1795
1796
1797
1798
1799
1800
1801
1802
1803
1804
1805
1806
1807
1808
1809
1810
1811
1812
1813
1814
1815
1816
1817
1818
1819
1820
1821
1822
1823
1824
1825
1826
1827
1828
1829
1830
1831
1832
1833
1834
1835
1836
1837
1838
1839
1840
1841
1842
1843
1844
1845
1846
1847
1848
1849
1850
1851
1852
1853
1854
1855
1856
1857
1858
1859
1860
1861
1862
1863
1864
1865
1866
1867
1868
1869
1870
1871
1872
1873
1874
1875
1876
1877
1878
1879
1880
1881
1882
1883
1884
1885
1886
1887
1888
1889
1890
1891
1892
1893
1894
1895
1896
1897
1898
1899
1900
1901
1902
1903
1904
1905
1906
1907
1908
1909
1910
1911
1912
1913
1914
1915
1916
1917
1918
1919
1920
1921
1922
1923
1924
1925
1926
1927
1928
1929
1930
1931
1932
1933
1934
1935
1936
1937
1938
1939
1940
1941
1942
1943
1944
1945
1946
1947
1948
1949
1950
1951
1952
1953
1954
1955
1956
1957
1958
1959
1960
1961
1962
1963
1964
1965
1966
1967
1968
1969
1970
1971
1972
1973
1974
1975
1976
1977
1978
1979
1980
1981
1982
1983
1984
1985
1986
1987
1988
1989
1990
1991
1992
1993
1994
1995
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
2011
2012
2013
2014
2015
2016
2017
2018
2019
2020
2021
2022
2023
2024
2025
2026
2027
2028
2029
2030
2031
2032
/*-
 * Copyright (c) 1998 Berkeley Software Design, Inc. 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, this list of conditions and the following disclaimer.
 * 2. Redistributions in binary form must reproduce the above copyright
 *    notice, this list of conditions and the following disclaimer in the
 *    documentation and/or other materials provided with the distribution.
 * 3. Berkeley Software Design Inc's name may not be used to endorse or
 *    promote products derived from this software without specific prior
 *    written permission.
 *
 * THIS SOFTWARE IS PROVIDED BY BERKELEY SOFTWARE DESIGN INC ``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 BERKELEY SOFTWARE DESIGN INC 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.
 *
 *	from BSDI $Id: mutex_witness.c,v 1.1.2.20 2000/04/27 03:10:27 cp Exp $
 *	and BSDI $Id: synch_machdep.c,v 2.3.2.39 2000/04/27 03:10:25 cp Exp $
 */

/*
 * Implementation of the `witness' lock verifier.  Originally implemented for
 * mutexes in BSD/OS.  Extended to handle generic lock objects and lock
 * classes in FreeBSD.
 */

/*
 *	Main Entry: witness
 *	Pronunciation: 'wit-n&s
 *	Function: noun
 *	Etymology: Middle English witnesse, from Old English witnes knowledge,
 *	    testimony, witness, from 2wit
 *	Date: before 12th century
 *	1 : attestation of a fact or event : TESTIMONY
 *	2 : one that gives evidence; specifically : one who testifies in
 *	    a cause or before a judicial tribunal
 *	3 : one asked to be present at a transaction so as to be able to
 *	    testify to its having taken place
 *	4 : one who has personal knowledge of something
 *	5 a : something serving as evidence or proof : SIGN
 *	  b : public affirmation by word or example of usually
 *	      religious faith or conviction <the heroic witness to divine
 *	      life -- Pilot>
 *	6 capitalized : a member of the Jehovah's Witnesses 
 */

/*
 * Special rules concerning Giant and lock orders:
 *
 * 1) Giant must be acquired before any other mutexes.  Stated another way,
 *    no other mutex may be held when Giant is acquired.
 *
 * 2) Giant must be released when blocking on a sleepable lock.
 *
 * This rule is less obvious, but is a result of Giant providing the same
 * semantics as spl().  Basically, when a thread sleeps, it must release
 * Giant.  When a thread blocks on a sleepable lock, it sleeps.  Hence rule
 * 2).
 *
 * 3) Giant may be acquired before or after sleepable locks.
 *
 * This rule is also not quite as obvious.  Giant may be acquired after
 * a sleepable lock because it is a non-sleepable lock and non-sleepable
 * locks may always be acquired while holding a sleepable lock.  The second
 * case, Giant before a sleepable lock, follows from rule 2) above.  Suppose
 * you have two threads T1 and T2 and a sleepable lock X.  Suppose that T1
 * acquires X and blocks on Giant.  Then suppose that T2 acquires Giant and
 * blocks on X.  When T2 blocks on X, T2 will release Giant allowing T1 to
 * execute.  Thus, acquiring Giant both before and after a sleepable lock
 * will not result in a lock order reversal.
 */

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

#include "opt_ddb.h"
#include "opt_hwpmc_hooks.h"
#include "opt_witness.h"

#include <sys/param.h>
#include <sys/bus.h>
#include <sys/kdb.h>
#include <sys/kernel.h>
#include <sys/ktr.h>
#include <sys/lock.h>
#include <sys/malloc.h>
#include <sys/mutex.h>
#include <sys/priv.h>
#include <sys/proc.h>
#include <sys/sysctl.h>
#include <sys/systm.h>

#include <ddb/ddb.h>

#include <machine/stdarg.h>

/* Note that these traces do not work with KTR_ALQ. */
#if 0
#define	KTR_WITNESS	KTR_SUBSYS
#else
#define	KTR_WITNESS	0
#endif

/* Easier to stay with the old names. */
#define	lo_list		lo_witness_data.lod_list
#define	lo_witness	lo_witness_data.lod_witness

/* Define this to check for blessed mutexes */
#undef BLESSING

#define WITNESS_COUNT 1024
#define WITNESS_CHILDCOUNT (WITNESS_COUNT * 4)
/*
 * XXX: This is somewhat bogus, as we assume here that at most 1024 threads
 * will hold LOCK_NCHILDREN * 2 locks.  We handle failure ok, and we should
 * probably be safe for the most part, but it's still a SWAG.
 */
#define LOCK_CHILDCOUNT (MAXCPU + 1024) * 2

#define	WITNESS_NCHILDREN 6

struct witness_child_list_entry;

struct witness {
	const	char *w_name;
	struct	lock_class *w_class;
	STAILQ_ENTRY(witness) w_list;		/* List of all witnesses. */
	STAILQ_ENTRY(witness) w_typelist;	/* Witnesses of a type. */
	struct	witness_child_list_entry *w_children;	/* Great evilness... */
	const	char *w_file;
	int	w_line;
	u_int	w_level;
	u_int	w_refcount;
	u_char	w_Giant_squawked:1;
	u_char	w_other_squawked:1;
	u_char	w_same_squawked:1;
	u_char	w_displayed:1;
};

struct witness_child_list_entry {
	struct	witness_child_list_entry *wcl_next;
	struct	witness *wcl_children[WITNESS_NCHILDREN];
	u_int	wcl_count;
};

STAILQ_HEAD(witness_list, witness);

#ifdef BLESSING
struct witness_blessed {
	const	char *b_lock1;
	const	char *b_lock2;
};
#endif

struct witness_order_list_entry {
	const	char *w_name;
	struct	lock_class *w_class;
};

#ifdef BLESSING
static int	blessed(struct witness *, struct witness *);
#endif
static int	depart(struct witness *w);
static struct	witness *enroll(const char *description,
				struct lock_class *lock_class);
static int	insertchild(struct witness *parent, struct witness *child);
static int	isitmychild(struct witness *parent, struct witness *child);
static int	isitmydescendant(struct witness *parent, struct witness *child);
static int	itismychild(struct witness *parent, struct witness *child);
static void	removechild(struct witness *parent, struct witness *child);
static int	sysctl_debug_witness_watch(SYSCTL_HANDLER_ARGS);
static const char *fixup_filename(const char *file);
static struct	witness *witness_get(void);
static void	witness_free(struct witness *m);
static struct	witness_child_list_entry *witness_child_get(void);
static void	witness_child_free(struct witness_child_list_entry *wcl);
static struct	lock_list_entry *witness_lock_list_get(void);
static void	witness_lock_list_free(struct lock_list_entry *lle);
static struct	lock_instance *find_instance(struct lock_list_entry *lock_list,
					     struct lock_object *lock);
static void	witness_list_lock(struct lock_instance *instance);
#ifdef DDB
static void	witness_leveldescendents(struct witness *parent, int level);
static void	witness_levelall(void);
static void	witness_displaydescendants(void(*)(const char *fmt, ...),
					   struct witness *, int indent);
static void	witness_display_list(void(*prnt)(const char *fmt, ...),
				     struct witness_list *list);
static void	witness_display(void(*)(const char *fmt, ...));
static void	witness_list(struct thread *td);
#endif

SYSCTL_NODE(_debug, OID_AUTO, witness, CTLFLAG_RW, 0, "Witness Locking");

/*
 * If set to 0, witness is disabled.  If set to a non-zero value, witness
 * performs full lock order checking for all locks.  At runtime, this
 * value may be set to 0 to turn off witness.  witness is not allowed be
 * turned on once it is turned off, however.
 */
static int witness_watch = 1;
TUNABLE_INT("debug.witness.watch", &witness_watch);
SYSCTL_PROC(_debug_witness, OID_AUTO, watch, CTLFLAG_RW | CTLTYPE_INT, NULL, 0,
    sysctl_debug_witness_watch, "I", "witness is watching lock operations");

#ifdef KDB
/*
 * When KDB is enabled and witness_kdb is set to 1, it will cause the system
 * to drop into kdebug() when:
 *	- a lock hierarchy violation occurs
 *	- locks are held when going to sleep.
 */
#ifdef WITNESS_KDB
int	witness_kdb = 1;
#else
int	witness_kdb = 0;
#endif
TUNABLE_INT("debug.witness.kdb", &witness_kdb);
SYSCTL_INT(_debug_witness, OID_AUTO, kdb, CTLFLAG_RW, &witness_kdb, 0, "");

/*
 * When KDB is enabled and witness_trace is set to 1, it will cause the system
 * to print a stack trace:
 *	- a lock hierarchy violation occurs
 *	- locks are held when going to sleep.
 */
int	witness_trace = 1;
TUNABLE_INT("debug.witness.trace", &witness_trace);
SYSCTL_INT(_debug_witness, OID_AUTO, trace, CTLFLAG_RW, &witness_trace, 0, "");
#endif /* KDB */

#ifdef WITNESS_SKIPSPIN
int	witness_skipspin = 1;
#else
int	witness_skipspin = 0;
#endif
TUNABLE_INT("debug.witness.skipspin", &witness_skipspin);
SYSCTL_INT(_debug_witness, OID_AUTO, skipspin, CTLFLAG_RDTUN,
    &witness_skipspin, 0, "");

static struct mtx w_mtx;
static struct witness_list w_free = STAILQ_HEAD_INITIALIZER(w_free);
static struct witness_list w_all = STAILQ_HEAD_INITIALIZER(w_all);
static struct witness_list w_spin = STAILQ_HEAD_INITIALIZER(w_spin);
static struct witness_list w_sleep = STAILQ_HEAD_INITIALIZER(w_sleep);
static struct witness_child_list_entry *w_child_free = NULL;
static struct lock_list_entry *w_lock_list_free = NULL;

static int w_free_cnt, w_spin_cnt, w_sleep_cnt, w_child_free_cnt, w_child_cnt;
SYSCTL_INT(_debug_witness, OID_AUTO, free_cnt, CTLFLAG_RD, &w_free_cnt, 0, "");
SYSCTL_INT(_debug_witness, OID_AUTO, spin_cnt, CTLFLAG_RD, &w_spin_cnt, 0, "");
SYSCTL_INT(_debug_witness, OID_AUTO, sleep_cnt, CTLFLAG_RD, &w_sleep_cnt, 0,
    "");
SYSCTL_INT(_debug_witness, OID_AUTO, child_free_cnt, CTLFLAG_RD,
    &w_child_free_cnt, 0, "");
SYSCTL_INT(_debug_witness, OID_AUTO, child_cnt, CTLFLAG_RD, &w_child_cnt, 0,
    "");

static struct witness w_data[WITNESS_COUNT];
static struct witness_child_list_entry w_childdata[WITNESS_CHILDCOUNT];
static struct lock_list_entry w_locklistdata[LOCK_CHILDCOUNT];

static struct witness_order_list_entry order_lists[] = {
	/*
	 * sx locks
	 */
	{ "proctree", &lock_class_sx },
	{ "allproc", &lock_class_sx },
	{ "allprison", &lock_class_sx },
	{ NULL, NULL },
	/*
	 * Various mutexes
	 */
	{ "Giant", &lock_class_mtx_sleep },
	{ "pipe mutex", &lock_class_mtx_sleep },
	{ "sigio lock", &lock_class_mtx_sleep },
	{ "process group", &lock_class_mtx_sleep },
	{ "process lock", &lock_class_mtx_sleep },
	{ "session", &lock_class_mtx_sleep },
	{ "uidinfo hash", &lock_class_mtx_sleep },
	{ "uidinfo struct", &lock_class_mtx_sleep },
#ifdef	HWPMC_HOOKS
	{ "pmc-sleep", &lock_class_mtx_sleep },
#endif
	{ NULL, NULL },
	/*
	 * Sockets
	 */
	{ "accept", &lock_class_mtx_sleep },
	{ "so_snd", &lock_class_mtx_sleep },
	{ "so_rcv", &lock_class_mtx_sleep },
	{ "sellck", &lock_class_mtx_sleep },
	{ NULL, NULL },
	/*
	 * Routing
	 */
	{ "so_rcv", &lock_class_mtx_sleep },
	{ "radix node head", &lock_class_mtx_sleep },
	{ "rtentry", &lock_class_mtx_sleep },
	{ "ifaddr", &lock_class_mtx_sleep },
	{ NULL, NULL },
	/*
	 * Multicast - protocol locks before interface locks, after UDP locks.
	 */
	{ "udpinp", &lock_class_mtx_sleep },
	{ "in_multi_mtx", &lock_class_mtx_sleep },
	{ "igmp_mtx", &lock_class_mtx_sleep },
	{ "if_addr_mtx", &lock_class_mtx_sleep },
	{ NULL, NULL },
	/*
	 * UNIX Domain Sockets
	 */
	{ "unp", &lock_class_mtx_sleep },
	{ "so_snd", &lock_class_mtx_sleep },
	{ NULL, NULL },
	/*
	 * UDP/IP
	 */
	{ "udp", &lock_class_mtx_sleep },
	{ "udpinp", &lock_class_mtx_sleep },
	{ "so_snd", &lock_class_mtx_sleep },
	{ NULL, NULL },
	/*
	 * TCP/IP
	 */
	{ "tcp", &lock_class_mtx_sleep },
	{ "tcpinp", &lock_class_mtx_sleep },
	{ "so_snd", &lock_class_mtx_sleep },
	{ NULL, NULL },
	/*
	 * SLIP
	 */
	{ "slip_mtx", &lock_class_mtx_sleep },
	{ "slip sc_mtx", &lock_class_mtx_sleep },
	{ NULL, NULL },
	/*
	 * netatalk
	 */
	{ "ddp_list_mtx", &lock_class_mtx_sleep },
	{ "ddp_mtx", &lock_class_mtx_sleep },
	{ NULL, NULL },
	/*
	 * BPF
	 */
	{ "bpf global lock", &lock_class_mtx_sleep },
	{ "bpf interface lock", &lock_class_mtx_sleep },
	{ "bpf cdev lock", &lock_class_mtx_sleep },
	{ NULL, NULL },
	/*
	 * NFS server
	 */
	{ "nfsd_mtx", &lock_class_mtx_sleep },
	{ "so_snd", &lock_class_mtx_sleep },
	{ NULL, NULL },

	/*
	 * IEEE 802.11
	 */
	{ "802.11 com lock", &lock_class_mtx_sleep},
	{ NULL, NULL },
	/*
	 * Network drivers
	 */
	{ "network driver", &lock_class_mtx_sleep},
	{ NULL, NULL },

	/*
	 * Netgraph
	 */
	{ "ng_node", &lock_class_mtx_sleep },
	{ "ng_worklist", &lock_class_mtx_sleep },
	{ NULL, NULL },
	/*
	 * CDEV
	 */
	{ "system map", &lock_class_mtx_sleep },
	{ "vm page queue mutex", &lock_class_mtx_sleep },
	{ "vnode interlock", &lock_class_mtx_sleep },
	{ "cdev", &lock_class_mtx_sleep },
	{ NULL, NULL },
	/*
	 * kqueue/VFS interaction
	 */
	{ "kqueue", &lock_class_mtx_sleep },
	{ "struct mount mtx", &lock_class_mtx_sleep },
	{ "vnode interlock", &lock_class_mtx_sleep },
	{ NULL, NULL },
	/*
	 * spin locks
	 */
#ifdef SMP
	{ "ap boot", &lock_class_mtx_spin },
#endif
	{ "rm.mutex_mtx", &lock_class_mtx_spin },
	{ "sio", &lock_class_mtx_spin },
	{ "scrlock", &lock_class_mtx_spin },
#ifdef __i386__
	{ "cy", &lock_class_mtx_spin },
#endif
#ifdef __sparc64__
	{ "pcib_mtx", &lock_class_mtx_spin },
	{ "rtc_mtx", &lock_class_mtx_spin },
#endif
	{ "scc_hwmtx", &lock_class_mtx_spin },
	{ "uart_hwmtx", &lock_class_mtx_spin },
	{ "fast_taskqueue", &lock_class_mtx_spin },
	{ "intr table", &lock_class_mtx_spin },
#ifdef	HWPMC_HOOKS
	{ "pmc-per-proc", &lock_class_mtx_spin },
#endif
	{ "process slock", &lock_class_mtx_spin },
	{ "sleepq chain", &lock_class_mtx_spin },
	{ "umtx lock", &lock_class_mtx_spin },
	{ "rm_spinlock", &lock_class_mtx_spin },
	{ "turnstile chain", &lock_class_mtx_spin },
	{ "turnstile lock", &lock_class_mtx_spin },
	{ "sched lock", &lock_class_mtx_spin },
	{ "td_contested", &lock_class_mtx_spin },
	{ "callout", &lock_class_mtx_spin },
	{ "entropy harvest mutex", &lock_class_mtx_spin },
	{ "syscons video lock", &lock_class_mtx_spin },
	{ "time lock", &lock_class_mtx_spin },
#ifdef SMP
	{ "smp rendezvous", &lock_class_mtx_spin },
#endif
	/*
	 * leaf locks
	 */
	{ "icu", &lock_class_mtx_spin },
#if defined(SMP) && defined(__sparc64__)
	{ "ipi", &lock_class_mtx_spin },
#endif
#ifdef __i386__
	{ "allpmaps", &lock_class_mtx_spin },
	{ "descriptor tables", &lock_class_mtx_spin },
#endif
	{ "clk", &lock_class_mtx_spin },
	{ "mprof lock", &lock_class_mtx_spin },
	{ "kse lock", &lock_class_mtx_spin },
	{ "zombie lock", &lock_class_mtx_spin },
	{ "ALD Queue", &lock_class_mtx_spin },
#ifdef __ia64__
	{ "MCA spin lock", &lock_class_mtx_spin },
#endif
#if defined(__i386__) || defined(__amd64__)
	{ "pcicfg", &lock_class_mtx_spin },
	{ "NDIS thread lock", &lock_class_mtx_spin },
#endif
	{ "tw_osl_io_lock", &lock_class_mtx_spin },
	{ "tw_osl_q_lock", &lock_class_mtx_spin },
	{ "tw_cl_io_lock", &lock_class_mtx_spin },
	{ "tw_cl_intr_lock", &lock_class_mtx_spin },
	{ "tw_cl_gen_lock", &lock_class_mtx_spin },
#ifdef	HWPMC_HOOKS
	{ "pmc-leaf", &lock_class_mtx_spin },
#endif
	{ "blocked lock", &lock_class_mtx_spin },
	{ NULL, NULL },
	{ NULL, NULL }
};

#ifdef BLESSING
/*
 * Pairs of locks which have been blessed
 * Don't complain about order problems with blessed locks
 */
static struct witness_blessed blessed_list[] = {
};
static int blessed_count =
	sizeof(blessed_list) / sizeof(struct witness_blessed);
#endif

/*
 * List of locks initialized prior to witness being initialized whose
 * enrollment is currently deferred.
 */
STAILQ_HEAD(, lock_object) pending_locks =
    STAILQ_HEAD_INITIALIZER(pending_locks);

/*
 * This global is set to 0 once it becomes safe to use the witness code.
 */
static int witness_cold = 1;

/*
 * This global is set to 1 once the static lock orders have been enrolled
 * so that a warning can be issued for any spin locks enrolled later.
 */
static int witness_spin_warn = 0;

/*
 * The WITNESS-enabled diagnostic code.  Note that the witness code does
 * assume that the early boot is single-threaded at least until after this
 * routine is completed.
 */
static void
witness_initialize(void *dummy __unused)
{
	struct lock_object *lock;
	struct witness_order_list_entry *order;
	struct witness *w, *w1;
	int i;

	/*
	 * We have to release Giant before initializing its witness
	 * structure so that WITNESS doesn't get confused.
	 */
	mtx_unlock(&Giant);
	mtx_assert(&Giant, MA_NOTOWNED);

	CTR1(KTR_WITNESS, "%s: initializing witness", __func__);
	mtx_init(&w_mtx, "witness lock", NULL, MTX_SPIN | MTX_QUIET |
	    MTX_NOWITNESS | MTX_NOPROFILE);
	for (i = 0; i < WITNESS_COUNT; i++)
		witness_free(&w_data[i]);
	for (i = 0; i < WITNESS_CHILDCOUNT; i++)
		witness_child_free(&w_childdata[i]);
	for (i = 0; i < LOCK_CHILDCOUNT; i++)
		witness_lock_list_free(&w_locklistdata[i]);

	/* First add in all the specified order lists. */
	for (order = order_lists; order->w_name != NULL; order++) {
		w = enroll(order->w_name, order->w_class);
		if (w == NULL)
			continue;
		w->w_file = "order list";
		for (order++; order->w_name != NULL; order++) {
			w1 = enroll(order->w_name, order->w_class);
			if (w1 == NULL)
				continue;
			w1->w_file = "order list";
			if (!itismychild(w, w1))
				panic("Not enough memory for static orders!");
			w = w1;
		}
	}
	witness_spin_warn = 1;

	/* Iterate through all locks and add them to witness. */
	while (!STAILQ_EMPTY(&pending_locks)) {
		lock = STAILQ_FIRST(&pending_locks);
		STAILQ_REMOVE_HEAD(&pending_locks, lo_list);
		KASSERT(lock->lo_flags & LO_WITNESS,
		    ("%s: lock %s is on pending list but not LO_WITNESS",
		    __func__, lock->lo_name));
		lock->lo_witness = enroll(lock->lo_type, LOCK_CLASS(lock));
	}

	/* Mark the witness code as being ready for use. */
	witness_cold = 0;

	mtx_lock(&Giant);
}
SYSINIT(witness_init, SI_SUB_WITNESS, SI_ORDER_FIRST, witness_initialize, NULL)

static int
sysctl_debug_witness_watch(SYSCTL_HANDLER_ARGS)
{
	int error, value;

	value = witness_watch;
	error = sysctl_handle_int(oidp, &value, 0, req);
	if (error != 0 || req->newptr == NULL)
		return (error);
	if (value == witness_watch)
		return (0);
	if (value != 0)
		return (EINVAL);
	witness_watch = 0;
	return (0);
}

void
witness_init(struct lock_object *lock)
{
	struct lock_class *class;

	/* Various sanity checks. */
	class = LOCK_CLASS(lock);
	if ((lock->lo_flags & LO_RECURSABLE) != 0 &&
	    (class->lc_flags & LC_RECURSABLE) == 0)
		panic("%s: lock (%s) %s can not be recursable", __func__,
		    class->lc_name, lock->lo_name);
	if ((lock->lo_flags & LO_SLEEPABLE) != 0 &&
	    (class->lc_flags & LC_SLEEPABLE) == 0)
		panic("%s: lock (%s) %s can not be sleepable", __func__,
		    class->lc_name, lock->lo_name);
	if ((lock->lo_flags & LO_UPGRADABLE) != 0 &&
	    (class->lc_flags & LC_UPGRADABLE) == 0)
		panic("%s: lock (%s) %s can not be upgradable", __func__,
		    class->lc_name, lock->lo_name);

	/*
	 * If we shouldn't watch this lock, then just clear lo_witness.
	 * Otherwise, if witness_cold is set, then it is too early to
	 * enroll this lock, so defer it to witness_initialize() by adding
	 * it to the pending_locks list.  If it is not too early, then enroll
	 * the lock now.
	 */
	if (witness_watch == 0 || panicstr != NULL ||
	    (lock->lo_flags & LO_WITNESS) == 0)
		lock->lo_witness = NULL;
	else if (witness_cold) {
		STAILQ_INSERT_TAIL(&pending_locks, lock, lo_list);
		lock->lo_flags |= LO_ENROLLPEND;
	} else
		lock->lo_witness = enroll(lock->lo_type, class);
}

void
witness_destroy(struct lock_object *lock)
{
	struct lock_class *class;
	struct witness *w;

	class = LOCK_CLASS(lock);
	if (witness_cold)
		panic("lock (%s) %s destroyed while witness_cold",
		    class->lc_name, lock->lo_name);

	/* XXX: need to verify that no one holds the lock */
	if ((lock->lo_flags & (LO_WITNESS | LO_ENROLLPEND)) == LO_WITNESS &&
	    lock->lo_witness != NULL) {
		w = lock->lo_witness;
		mtx_lock_spin(&w_mtx);
		MPASS(w->w_refcount > 0);
		w->w_refcount--;

		/*
		 * Lock is already released if we have an allocation failure
		 * and depart() fails.
		 */
		if (w->w_refcount != 0 || depart(w))
			mtx_unlock_spin(&w_mtx);
	}

	/*
	 * If this lock is destroyed before witness is up and running,
	 * remove it from the pending list.
	 */
	if (lock->lo_flags & LO_ENROLLPEND) {
		STAILQ_REMOVE(&pending_locks, lock, lock_object, lo_list);
		lock->lo_flags &= ~LO_ENROLLPEND;
	}
}

#ifdef DDB
static void
witness_levelall (void)
{
	struct witness_list *list;
	struct witness *w, *w1;

	/*
	 * First clear all levels.
	 */
	STAILQ_FOREACH(w, &w_all, w_list) {
		w->w_level = 0;
	}

	/*
	 * Look for locks with no parent and level all their descendants.
	 */
	STAILQ_FOREACH(w, &w_all, w_list) {
		/*
		 * This is just an optimization, technically we could get
		 * away just walking the all list each time.
		 */
		if (w->w_class->lc_flags & LC_SLEEPLOCK)
			list = &w_sleep;
		else
			list = &w_spin;
		STAILQ_FOREACH(w1, list, w_typelist) {
			if (isitmychild(w1, w))
				goto skip;
		}
		witness_leveldescendents(w, 0);
	skip:
		;	/* silence GCC 3.x */
	}
}

static void
witness_leveldescendents(struct witness *parent, int level)
{
	struct witness_child_list_entry *wcl;
	int i;

	if (parent->w_level < level)
		parent->w_level = level;
	level++;
	for (wcl = parent->w_children; wcl != NULL; wcl = wcl->wcl_next)
		for (i = 0; i < wcl->wcl_count; i++)
			witness_leveldescendents(wcl->wcl_children[i], level);
}

static void
witness_displaydescendants(void(*prnt)(const char *fmt, ...),
			   struct witness *parent, int indent)
{
	struct witness_child_list_entry *wcl;
	int i, level;

	level = parent->w_level;
	prnt("%-2d", level);
	for (i = 0; i < indent; i++)
		prnt(" ");
	if (parent->w_refcount > 0)
		prnt("%s", parent->w_name);
	else
		prnt("(dead)");
	if (parent->w_displayed) {
		prnt(" -- (already displayed)\n");
		return;
	}
	parent->w_displayed = 1;
	if (parent->w_refcount > 0) {
		if (parent->w_file != NULL)
			prnt(" -- last acquired @ %s:%d", parent->w_file,
			    parent->w_line);
	}
	prnt("\n");
	for (wcl = parent->w_children; wcl != NULL; wcl = wcl->wcl_next)
		for (i = 0; i < wcl->wcl_count; i++)
			    witness_displaydescendants(prnt,
				wcl->wcl_children[i], indent + 1);
}

static void
witness_display_list(void(*prnt)(const char *fmt, ...),
		     struct witness_list *list)
{
	struct witness *w;

	STAILQ_FOREACH(w, list, w_typelist) {
		if (w->w_file == NULL || w->w_level > 0)
			continue;
		/*
		 * This lock has no anscestors, display its descendants. 
		 */
		witness_displaydescendants(prnt, w, 0);
	}
}
	
static void
witness_display(void(*prnt)(const char *fmt, ...))
{
	struct witness *w;

	KASSERT(!witness_cold, ("%s: witness_cold", __func__));
	witness_levelall();

	/* Clear all the displayed flags. */
	STAILQ_FOREACH(w, &w_all, w_list) {
		w->w_displayed = 0;
	}

	/*
	 * First, handle sleep locks which have been acquired at least
	 * once.
	 */
	prnt("Sleep locks:\n");
	witness_display_list(prnt, &w_sleep);
	
	/*
	 * Now do spin locks which have been acquired at least once.
	 */
	prnt("\nSpin locks:\n");
	witness_display_list(prnt, &w_spin);
	
	/*
	 * Finally, any locks which have not been acquired yet.
	 */
	prnt("\nLocks which were never acquired:\n");
	STAILQ_FOREACH(w, &w_all, w_list) {
		if (w->w_file != NULL || w->w_refcount == 0)
			continue;
		prnt("%s\n", w->w_name);
	}
}
#endif /* DDB */

/* Trim useless garbage from filenames. */
static const char *
fixup_filename(const char *file)
{

	if (file == NULL)
		return (NULL);
	while (strncmp(file, "../", 3) == 0)
		file += 3;
	return (file);
}

int
witness_defineorder(struct lock_object *lock1, struct lock_object *lock2)
{

	if (witness_watch == 0 || panicstr != NULL)
		return (0);

	/* Require locks that witness knows about. */
	if (lock1 == NULL || lock1->lo_witness == NULL || lock2 == NULL ||
	    lock2->lo_witness == NULL)
		return (EINVAL);

	MPASS(!mtx_owned(&w_mtx));
	mtx_lock_spin(&w_mtx);

	/*
	 * If we already have either an explicit or implied lock order that
	 * is the other way around, then return an error.
	 */
	if (isitmydescendant(lock2->lo_witness, lock1->lo_witness)) {
		mtx_unlock_spin(&w_mtx);
		return (EDOOFUS);
	}
	
	/* Try to add the new order. */
	CTR3(KTR_WITNESS, "%s: adding %s as a child of %s", __func__,
	    lock2->lo_type, lock1->lo_type);
	if (!itismychild(lock1->lo_witness, lock2->lo_witness))
		return (ENOMEM);
	mtx_unlock_spin(&w_mtx);
	return (0);
}

void
witness_checkorder(struct lock_object *lock, int flags, const char *file,
    int line)
{
	struct lock_list_entry **lock_list, *lle;
	struct lock_instance *lock1, *lock2;
	struct lock_class *class;
	struct witness *w, *w1;
	struct thread *td;
	int i, j;

	if (witness_cold || witness_watch == 0 || lock->lo_witness == NULL ||
	    panicstr != NULL)
		return;

	/*
	 * Try locks do not block if they fail to acquire the lock, thus
	 * there is no danger of deadlocks or of switching while holding a
	 * spin lock if we acquire a lock via a try operation.  This
	 * function shouldn't even be called for try locks, so panic if
	 * that happens.
	 */
	if (flags & LOP_TRYLOCK)
		panic("%s should not be called for try lock operations",
		    __func__);

	w = lock->lo_witness;
	class = LOCK_CLASS(lock);
	td = curthread;
	file = fixup_filename(file);

	if (class->lc_flags & LC_SLEEPLOCK) {
		/*
		 * Since spin locks include a critical section, this check
		 * implicitly enforces a lock order of all sleep locks before
		 * all spin locks.
		 */
		if (td->td_critnest != 0 && !kdb_active)
			panic("blockable sleep lock (%s) %s @ %s:%d",
			    class->lc_name, lock->lo_name, file, line);

		/*
		 * If this is the first lock acquired then just return as
		 * no order checking is needed.
		 */
		if (td->td_sleeplocks == NULL)
			return;
		lock_list = &td->td_sleeplocks;
	} else {
		/*
		 * If this is the first lock, just return as no order
		 * checking is needed.  We check this in both if clauses
		 * here as unifying the check would require us to use a
		 * critical section to ensure we don't migrate while doing
		 * the check.  Note that if this is not the first lock, we
		 * are already in a critical section and are safe for the
		 * rest of the check.
		 */
		if (PCPU_GET(spinlocks) == NULL)
			return;
		lock_list = PCPU_PTR(spinlocks);
	}

	/*
	 * Check to see if we are recursing on a lock we already own.  If
	 * so, make sure that we don't mismatch exclusive and shared lock
	 * acquires.
	 */
	lock1 = find_instance(*lock_list, lock);
	if (lock1 != NULL) {
		if ((lock1->li_flags & LI_EXCLUSIVE) != 0 &&
		    (flags & LOP_EXCLUSIVE) == 0) {
			printf("shared lock of (%s) %s @ %s:%d\n",
			    class->lc_name, lock->lo_name, file, line);
			printf("while exclusively locked from %s:%d\n",
			    lock1->li_file, lock1->li_line);
			panic("share->excl");
		}
		if ((lock1->li_flags & LI_EXCLUSIVE) == 0 &&
		    (flags & LOP_EXCLUSIVE) != 0) {
			printf("exclusive lock of (%s) %s @ %s:%d\n",
			    class->lc_name, lock->lo_name, file, line);
			printf("while share locked from %s:%d\n",
			    lock1->li_file, lock1->li_line);
			panic("excl->share");
		}
		return;
	}

	/*
	 * Try locks do not block if they fail to acquire the lock, thus
	 * there is no danger of deadlocks or of switching while holding a
	 * spin lock if we acquire a lock via a try operation.
	 */
	if (flags & LOP_TRYLOCK)
		return;

	/*
	 * Check for duplicate locks of the same type.  Note that we only
	 * have to check for this on the last lock we just acquired.  Any
	 * other cases will be caught as lock order violations.
	 */
	lock1 = &(*lock_list)->ll_children[(*lock_list)->ll_count - 1];
	w1 = lock1->li_lock->lo_witness;
	if (w1 == w) {
		if (w->w_same_squawked || (lock->lo_flags & LO_DUPOK) ||
		    (flags & LOP_DUPOK))
			return;
		w->w_same_squawked = 1;
		printf("acquiring duplicate lock of same type: \"%s\"\n", 
			lock->lo_type);
		printf(" 1st %s @ %s:%d\n", lock1->li_lock->lo_name,
		    lock1->li_file, lock1->li_line);
		printf(" 2nd %s @ %s:%d\n", lock->lo_name, file, line);
#ifdef KDB
		goto debugger;
#else
		return;
#endif
	}
	MPASS(!mtx_owned(&w_mtx));
	mtx_lock_spin(&w_mtx);
	/*
	 * If we know that the the lock we are acquiring comes after
	 * the lock we most recently acquired in the lock order tree,
	 * then there is no need for any further checks.
	 */
	if (isitmychild(w1, w)) {
		mtx_unlock_spin(&w_mtx);
		return;
	}
	for (j = 0, lle = *lock_list; lle != NULL; lle = lle->ll_next) {
		for (i = lle->ll_count - 1; i >= 0; i--, j++) {

			MPASS(j < WITNESS_COUNT);
			lock1 = &lle->ll_children[i];
			w1 = lock1->li_lock->lo_witness;

			/*
			 * If this lock doesn't undergo witness checking,
			 * then skip it.
			 */
			if (w1 == NULL) {
				KASSERT((lock1->li_lock->lo_flags & LO_WITNESS) == 0,
				    ("lock missing witness structure"));
				continue;
			}
			/*
			 * If we are locking Giant and this is a sleepable
			 * lock, then skip it.
			 */
			if ((lock1->li_lock->lo_flags & LO_SLEEPABLE) != 0 &&
			    lock == &Giant.lock_object)
				continue;
			/*
			 * If we are locking a sleepable lock and this lock
			 * is Giant, then skip it.
			 */
			if ((lock->lo_flags & LO_SLEEPABLE) != 0 &&
			    lock1->li_lock == &Giant.lock_object)
				continue;
			/*
			 * If we are locking a sleepable lock and this lock
			 * isn't sleepable, we want to treat it as a lock
			 * order violation to enfore a general lock order of
			 * sleepable locks before non-sleepable locks.
			 */
			if (((lock->lo_flags & LO_SLEEPABLE) != 0 &&
			    (lock1->li_lock->lo_flags & LO_SLEEPABLE) == 0))
				goto reversal;
			/*
			 * If we are locking Giant and this is a non-sleepable
			 * lock, then treat it as a reversal.
			 */
			if ((lock1->li_lock->lo_flags & LO_SLEEPABLE) == 0 &&
			    lock == &Giant.lock_object)
				goto reversal;
			/*
			 * Check the lock order hierarchy for a reveresal.
			 */
			if (!isitmydescendant(w, w1))
				continue;
		reversal:
			/*
			 * We have a lock order violation, check to see if it
			 * is allowed or has already been yelled about.
			 */
			mtx_unlock_spin(&w_mtx);
#ifdef BLESSING
			/*
			 * If the lock order is blessed, just bail.  We don't
			 * look for other lock order violations though, which
			 * may be a bug.
			 */
			if (blessed(w, w1))
				return;
#endif
			if (lock1->li_lock == &Giant.lock_object) {
				if (w1->w_Giant_squawked)
					return;
				else
					w1->w_Giant_squawked = 1;
			} else {
				if (w1->w_other_squawked)
					return;
				else
					w1->w_other_squawked = 1;
			}
			/*
			 * Ok, yell about it.
			 */
			if (((lock->lo_flags & LO_SLEEPABLE) != 0 &&
			    (lock1->li_lock->lo_flags & LO_SLEEPABLE) == 0))
				printf(
		"lock order reversal: (sleepable after non-sleepable)\n");
			else if ((lock1->li_lock->lo_flags & LO_SLEEPABLE) == 0
			    && lock == &Giant.lock_object)
				printf(
		"lock order reversal: (Giant after non-sleepable)\n");
			else
				printf("lock order reversal:\n");
			/*
			 * Try to locate an earlier lock with
			 * witness w in our list.
			 */
			do {
				lock2 = &lle->ll_children[i];
				MPASS(lock2->li_lock != NULL);
				if (lock2->li_lock->lo_witness == w)
					break;
				if (i == 0 && lle->ll_next != NULL) {
					lle = lle->ll_next;
					i = lle->ll_count - 1;
					MPASS(i >= 0 && i < LOCK_NCHILDREN);
				} else
					i--;
			} while (i >= 0);
			if (i < 0) {
				printf(" 1st %p %s (%s) @ %s:%d\n",
				    lock1->li_lock, lock1->li_lock->lo_name,
				    lock1->li_lock->lo_type, lock1->li_file,
				    lock1->li_line);
				printf(" 2nd %p %s (%s) @ %s:%d\n", lock,
				    lock->lo_name, lock->lo_type, file, line);
			} else {
				printf(" 1st %p %s (%s) @ %s:%d\n",
				    lock2->li_lock, lock2->li_lock->lo_name,
				    lock2->li_lock->lo_type, lock2->li_file,
				    lock2->li_line);
				printf(" 2nd %p %s (%s) @ %s:%d\n",
				    lock1->li_lock, lock1->li_lock->lo_name,
				    lock1->li_lock->lo_type, lock1->li_file,
				    lock1->li_line);
				printf(" 3rd %p %s (%s) @ %s:%d\n", lock,
				    lock->lo_name, lock->lo_type, file, line);
			}
#ifdef KDB
			goto debugger;
#else
			return;
#endif
		}
	}
	lock1 = &(*lock_list)->ll_children[(*lock_list)->ll_count - 1];
	/*
	 * If requested, build a new lock order.  However, don't build a new
	 * relationship between a sleepable lock and Giant if it is in the
	 * wrong direction.  The correct lock order is that sleepable locks
	 * always come before Giant.
	 */
	if (flags & LOP_NEWORDER &&
	    !(lock1->li_lock == &Giant.lock_object &&
	    (lock->lo_flags & LO_SLEEPABLE) != 0)) {
		CTR3(KTR_WITNESS, "%s: adding %s as a child of %s", __func__,
		    lock->lo_type, lock1->li_lock->lo_type);
		if (!itismychild(lock1->li_lock->lo_witness, w))
			/* Witness is dead. */
			return;
	} 
	mtx_unlock_spin(&w_mtx);
	return;

#ifdef KDB
debugger:
	if (witness_trace)
		kdb_backtrace();
	if (witness_kdb)
		kdb_enter(__func__);
#endif
}

void
witness_lock(struct lock_object *lock, int flags, const char *file, int line)
{
	struct lock_list_entry **lock_list, *lle;
	struct lock_instance *instance;
	struct witness *w;
	struct thread *td;

	if (witness_cold || witness_watch == 0 || lock->lo_witness == NULL ||
	    panicstr != NULL)
		return;
	w = lock->lo_witness;
	td = curthread;
	file = fixup_filename(file);

	/* Determine lock list for this lock. */
	if (LOCK_CLASS(lock)->lc_flags & LC_SLEEPLOCK)
		lock_list = &td->td_sleeplocks;
	else
		lock_list = PCPU_PTR(spinlocks);

	/* Check to see if we are recursing on a lock we already own. */
	instance = find_instance(*lock_list, lock);
	if (instance != NULL) {
		instance->li_flags++;
		CTR4(KTR_WITNESS, "%s: pid %d recursed on %s r=%d", __func__,
		    td->td_proc->p_pid, lock->lo_name,
		    instance->li_flags & LI_RECURSEMASK);
		instance->li_file = file;
		instance->li_line = line;
		return;
	}

	/* Update per-witness last file and line acquire. */
	w->w_file = file;
	w->w_line = line;

	/* Find the next open lock instance in the list and fill it. */
	lle = *lock_list;
	if (lle == NULL || lle->ll_count == LOCK_NCHILDREN) {
		lle = witness_lock_list_get();
		if (lle == NULL)
			return;
		lle->ll_next = *lock_list;
		CTR3(KTR_WITNESS, "%s: pid %d added lle %p", __func__,
		    td->td_proc->p_pid, lle);
		*lock_list = lle;
	}
	instance = &lle->ll_children[lle->ll_count++];
	instance->li_lock = lock;
	instance->li_line = line;
	instance->li_file = file;
	if ((flags & LOP_EXCLUSIVE) != 0)
		instance->li_flags = LI_EXCLUSIVE;
	else
		instance->li_flags = 0;
	CTR4(KTR_WITNESS, "%s: pid %d added %s as lle[%d]", __func__,
	    td->td_proc->p_pid, lock->lo_name, lle->ll_count - 1);
}

void
witness_upgrade(struct lock_object *lock, int flags, const char *file, int line)
{
	struct lock_instance *instance;
	struct lock_class *class;

	KASSERT(!witness_cold, ("%s: witness_cold", __func__));
	if (lock->lo_witness == NULL || witness_watch == 0 || panicstr != NULL)
		return;
	class = LOCK_CLASS(lock);
	file = fixup_filename(file);
	if ((lock->lo_flags & LO_UPGRADABLE) == 0)
		panic("upgrade of non-upgradable lock (%s) %s @ %s:%d",
		    class->lc_name, lock->lo_name, file, line);
	if ((flags & LOP_TRYLOCK) == 0)
		panic("non-try upgrade of lock (%s) %s @ %s:%d", class->lc_name,
		    lock->lo_name, file, line);
	if ((class->lc_flags & LC_SLEEPLOCK) == 0)
		panic("upgrade of non-sleep lock (%s) %s @ %s:%d",
		    class->lc_name, lock->lo_name, file, line);
	instance = find_instance(curthread->td_sleeplocks, lock);
	if (instance == NULL)
		panic("upgrade of unlocked lock (%s) %s @ %s:%d",
		    class->lc_name, lock->lo_name, file, line);
	if ((instance->li_flags & LI_EXCLUSIVE) != 0)
		panic("upgrade of exclusive lock (%s) %s @ %s:%d",
		    class->lc_name, lock->lo_name, file, line);
	if ((instance->li_flags & LI_RECURSEMASK) != 0)
		panic("upgrade of recursed lock (%s) %s r=%d @ %s:%d",
		    class->lc_name, lock->lo_name,
		    instance->li_flags & LI_RECURSEMASK, file, line);
	instance->li_flags |= LI_EXCLUSIVE;
}

void
witness_downgrade(struct lock_object *lock, int flags, const char *file,
    int line)
{
	struct lock_instance *instance;
	struct lock_class *class;

	KASSERT(!witness_cold, ("%s: witness_cold", __func__));
	if (lock->lo_witness == NULL || witness_watch == 0 || panicstr != NULL)
		return;
	class = LOCK_CLASS(lock);
	file = fixup_filename(file);
	if ((lock->lo_flags & LO_UPGRADABLE) == 0)
		panic("downgrade of non-upgradable lock (%s) %s @ %s:%d",
		    class->lc_name, lock->lo_name, file, line);
	if ((class->lc_flags & LC_SLEEPLOCK) == 0)
		panic("downgrade of non-sleep lock (%s) %s @ %s:%d",
		    class->lc_name, lock->lo_name, file, line);
	instance = find_instance(curthread->td_sleeplocks, lock);
	if (instance == NULL)
		panic("downgrade of unlocked lock (%s) %s @ %s:%d",
		    class->lc_name, lock->lo_name, file, line);
	if ((instance->li_flags & LI_EXCLUSIVE) == 0)
		panic("downgrade of shared lock (%s) %s @ %s:%d",
		    class->lc_name, lock->lo_name, file, line);
	if ((instance->li_flags & LI_RECURSEMASK) != 0)
		panic("downgrade of recursed lock (%s) %s r=%d @ %s:%d",
		    class->lc_name, lock->lo_name,
		    instance->li_flags & LI_RECURSEMASK, file, line);
	instance->li_flags &= ~LI_EXCLUSIVE;
}

void
witness_unlock(struct lock_object *lock, int flags, const char *file, int line)
{
	struct lock_list_entry **lock_list, *lle;
	struct lock_instance *instance;
	struct lock_class *class;
	struct thread *td;
	register_t s;
	int i, j;

	if (witness_cold || witness_watch == 0 || lock->lo_witness == NULL ||
	    panicstr != NULL)
		return;
	td = curthread;
	class = LOCK_CLASS(lock);
	file = fixup_filename(file);

	/* Find lock instance associated with this lock. */
	if (class->lc_flags & LC_SLEEPLOCK)
		lock_list = &td->td_sleeplocks;
	else
		lock_list = PCPU_PTR(spinlocks);
	for (; *lock_list != NULL; lock_list = &(*lock_list)->ll_next)
		for (i = 0; i < (*lock_list)->ll_count; i++) {
			instance = &(*lock_list)->ll_children[i];
			if (instance->li_lock == lock)
				goto found;
		}
	panic("lock (%s) %s not locked @ %s:%d", class->lc_name, lock->lo_name,
	    file, line);
found:

	/* First, check for shared/exclusive mismatches. */
	if ((instance->li_flags & LI_EXCLUSIVE) != 0 &&
	    (flags & LOP_EXCLUSIVE) == 0) {
		printf("shared unlock of (%s) %s @ %s:%d\n", class->lc_name,
		    lock->lo_name, file, line);
		printf("while exclusively locked from %s:%d\n",
		    instance->li_file, instance->li_line);
		panic("excl->ushare");
	}
	if ((instance->li_flags & LI_EXCLUSIVE) == 0 &&
	    (flags & LOP_EXCLUSIVE) != 0) {
		printf("exclusive unlock of (%s) %s @ %s:%d\n", class->lc_name,
		    lock->lo_name, file, line);
		printf("while share locked from %s:%d\n", instance->li_file,
		    instance->li_line);
		panic("share->uexcl");
	}

	/* If we are recursed, unrecurse. */
	if ((instance->li_flags & LI_RECURSEMASK) > 0) {
		CTR4(KTR_WITNESS, "%s: pid %d unrecursed on %s r=%d", __func__,
		    td->td_proc->p_pid, instance->li_lock->lo_name,
		    instance->li_flags);
		instance->li_flags--;
		return;
	}

	/* Otherwise, remove this item from the list. */
	s = intr_disable();
	CTR4(KTR_WITNESS, "%s: pid %d removed %s from lle[%d]", __func__,
	    td->td_proc->p_pid, instance->li_lock->lo_name,
	    (*lock_list)->ll_count - 1);
	for (j = i; j < (*lock_list)->ll_count - 1; j++)
		(*lock_list)->ll_children[j] =
		    (*lock_list)->ll_children[j + 1];
	(*lock_list)->ll_count--;
	intr_restore(s);

	/* If this lock list entry is now empty, free it. */
	if ((*lock_list)->ll_count == 0) {
		lle = *lock_list;
		*lock_list = lle->ll_next;
		CTR3(KTR_WITNESS, "%s: pid %d removed lle %p", __func__,
		    td->td_proc->p_pid, lle);
		witness_lock_list_free(lle);
	}
}

/*
 * Warn if any locks other than 'lock' are held.  Flags can be passed in to
 * exempt Giant and sleepable locks from the checks as well.  If any
 * non-exempt locks are held, then a supplied message is printed to the
 * console along with a list of the offending locks.  If indicated in the
 * flags then a failure results in a panic as well.
 */
int
witness_warn(int flags, struct lock_object *lock, const char *fmt, ...)
{
	struct lock_list_entry *lle;
	struct lock_instance *lock1;
	struct thread *td;
	va_list ap;
	int i, n;

	if (witness_cold || witness_watch == 0 || panicstr != NULL)
		return (0);
	n = 0;
	td = curthread;
	for (lle = td->td_sleeplocks; lle != NULL; lle = lle->ll_next)
		for (i = lle->ll_count - 1; i >= 0; i--) {
			lock1 = &lle->ll_children[i];
			if (lock1->li_lock == lock)
				continue;
			if (flags & WARN_GIANTOK &&
			    lock1->li_lock == &Giant.lock_object)
				continue;
			if (flags & WARN_SLEEPOK &&
			    (lock1->li_lock->lo_flags & LO_SLEEPABLE) != 0)
				continue;
			if (n == 0) {
				va_start(ap, fmt);
				vprintf(fmt, ap);
				va_end(ap);
				printf(" with the following");
				if (flags & WARN_SLEEPOK)
					printf(" non-sleepable");
				printf(" locks held:\n");
			}
			n++;
			witness_list_lock(lock1);
		}
	if (PCPU_GET(spinlocks) != NULL) {
		/*
		 * Since we already hold a spinlock preemption is
		 * already blocked.
		 */
		if (n == 0) {
			va_start(ap, fmt);
			vprintf(fmt, ap);
			va_end(ap);
			printf(" with the following");
			if (flags & WARN_SLEEPOK)
				printf(" non-sleepable");
			printf(" locks held:\n");
		}
		n += witness_list_locks(PCPU_PTR(spinlocks));
	}
	if (flags & WARN_PANIC && n)
		panic("witness_warn");
#ifdef KDB
	else if (witness_kdb && n)
		kdb_enter(__func__);
	else if (witness_trace && n)
		kdb_backtrace();
#endif
	return (n);
}

const char *
witness_file(struct lock_object *lock)
{
	struct witness *w;

	if (witness_cold || witness_watch == 0 || lock->lo_witness == NULL)
		return ("?");
	w = lock->lo_witness;
	return (w->w_file);
}

int
witness_line(struct lock_object *lock)
{
	struct witness *w;

	if (witness_cold || witness_watch == 0 || lock->lo_witness == NULL)
		return (0);
	w = lock->lo_witness;
	return (w->w_line);
}

static struct witness *
enroll(const char *description, struct lock_class *lock_class)
{
	struct witness *w;

	if (witness_watch == 0 || panicstr != NULL)
		return (NULL);
	if ((lock_class->lc_flags & LC_SPINLOCK) && witness_skipspin)
		return (NULL);
	mtx_lock_spin(&w_mtx);
	STAILQ_FOREACH(w, &w_all, w_list) {
		if (w->w_name == description || (w->w_refcount > 0 &&
		    strcmp(description, w->w_name) == 0)) {
			w->w_refcount++;
			mtx_unlock_spin(&w_mtx);
			if (lock_class != w->w_class)
				panic(
				"lock (%s) %s does not match earlier (%s) lock",
				    description, lock_class->lc_name,
				    w->w_class->lc_name);
			return (w);
		}
	}
	if ((w = witness_get()) == NULL)
		goto out;
	w->w_name = description;
	w->w_class = lock_class;
	w->w_refcount = 1;
	STAILQ_INSERT_HEAD(&w_all, w, w_list);
	if (lock_class->lc_flags & LC_SPINLOCK) {
		STAILQ_INSERT_HEAD(&w_spin, w, w_typelist);
		w_spin_cnt++;
	} else if (lock_class->lc_flags & LC_SLEEPLOCK) {
		STAILQ_INSERT_HEAD(&w_sleep, w, w_typelist);
		w_sleep_cnt++;
	} else {
		mtx_unlock_spin(&w_mtx);
		panic("lock class %s is not sleep or spin",
		    lock_class->lc_name);
	}
	mtx_unlock_spin(&w_mtx);
out:
	/*
	 * We issue a warning for any spin locks not defined in the static
	 * order list as a way to discourage their use (folks should really
	 * be using non-spin mutexes most of the time).  However, several
	 * 3rd part device drivers use spin locks because that is all they
	 * have available on Windows and Linux and they think that normal
	 * mutexes are insufficient.
	 */
	if ((lock_class->lc_flags & LC_SPINLOCK) && witness_spin_warn)
		printf("WITNESS: spin lock %s not in order list\n",
		    description);
	return (w);
}

/* Don't let the door bang you on the way out... */
static int
depart(struct witness *w)
{
	struct witness_child_list_entry *wcl, *nwcl;
	struct witness_list *list;
	struct witness *parent;

	MPASS(w->w_refcount == 0);
	if (w->w_class->lc_flags & LC_SLEEPLOCK) {
		list = &w_sleep;
		w_sleep_cnt--;
	} else {
		list = &w_spin;
		w_spin_cnt--;
	}
	/*
	 * First, we run through the entire tree looking for any
	 * witnesses that the outgoing witness is a child of.  For
	 * each parent that we find, we reparent all the direct
	 * children of the outgoing witness to its parent.
	 */
	STAILQ_FOREACH(parent, list, w_typelist) {
		if (!isitmychild(parent, w))
			continue;
		removechild(parent, w);
	}

	/*
	 * Now we go through and free up the child list of the
	 * outgoing witness.
	 */
	for (wcl = w->w_children; wcl != NULL; wcl = nwcl) {
		nwcl = wcl->wcl_next;
        	w_child_cnt--;
		witness_child_free(wcl);
	}

	/*
	 * Detach from various lists and free.
	 */
	STAILQ_REMOVE(list, w, witness, w_typelist);
	STAILQ_REMOVE(&w_all, w, witness, w_list);
	witness_free(w);

	return (1);
}

/*
 * Add "child" as a direct child of "parent".  Returns false if
 * we fail due to out of memory.
 */
static int
insertchild(struct witness *parent, struct witness *child)
{
	struct witness_child_list_entry **wcl;

	MPASS(child != NULL && parent != NULL);

	/*
	 * Insert "child" after "parent"
	 */
	wcl = &parent->w_children;
	while (*wcl != NULL && (*wcl)->wcl_count == WITNESS_NCHILDREN)
		wcl = &(*wcl)->wcl_next;
	if (*wcl == NULL) {
		*wcl = witness_child_get();
		if (*wcl == NULL)
			return (0);
        	w_child_cnt++;
	}
	(*wcl)->wcl_children[(*wcl)->wcl_count++] = child;

	return (1);
}


static int
itismychild(struct witness *parent, struct witness *child)
{
	struct witness_list *list;

	MPASS(child != NULL && parent != NULL);
	if ((parent->w_class->lc_flags & (LC_SLEEPLOCK | LC_SPINLOCK)) !=
	    (child->w_class->lc_flags & (LC_SLEEPLOCK | LC_SPINLOCK)))
		panic(
		"%s: parent (%s) and child (%s) are not the same lock type",
		    __func__, parent->w_class->lc_name,
		    child->w_class->lc_name);

	if (!insertchild(parent, child))
		return (0);

	if (parent->w_class->lc_flags & LC_SLEEPLOCK)
		list = &w_sleep;
	else
		list = &w_spin;
	return (1);
}

static void
removechild(struct witness *parent, struct witness *child)
{
	struct witness_child_list_entry **wcl, *wcl1;
	int i;

	for (wcl = &parent->w_children; *wcl != NULL; wcl = &(*wcl)->wcl_next)
		for (i = 0; i < (*wcl)->wcl_count; i++)
			if ((*wcl)->wcl_children[i] == child)
				goto found;
	return;
found:
	(*wcl)->wcl_count--;
	if ((*wcl)->wcl_count > i)
		(*wcl)->wcl_children[i] =
		    (*wcl)->wcl_children[(*wcl)->wcl_count];
	MPASS((*wcl)->wcl_children[i] != NULL);
	if ((*wcl)->wcl_count != 0)
		return;
	wcl1 = *wcl;
	*wcl = wcl1->wcl_next;
	w_child_cnt--;
	witness_child_free(wcl1);
}

static int
isitmychild(struct witness *parent, struct witness *child)
{
	struct witness_child_list_entry *wcl;
	int i;

	for (wcl = parent->w_children; wcl != NULL; wcl = wcl->wcl_next) {
		for (i = 0; i < wcl->wcl_count; i++) {
			if (wcl->wcl_children[i] == child)
				return (1);
		}
	}
	return (0);
}

static int
isitmydescendant(struct witness *parent, struct witness *child)
{
	struct witness_child_list_entry *wcl;
	int i, j;

	if (isitmychild(parent, child))
		return (1);
	j = 0;
	for (wcl = parent->w_children; wcl != NULL; wcl = wcl->wcl_next) {
		MPASS(j < 1000);
		for (i = 0; i < wcl->wcl_count; i++) {
			if (isitmydescendant(wcl->wcl_children[i], child))
				return (1);
		}
		j++;
	}
	return (0);
}

#ifdef BLESSING
static int
blessed(struct witness *w1, struct witness *w2)
{
	int i;
	struct witness_blessed *b;

	for (i = 0; i < blessed_count; i++) {
		b = &blessed_list[i];
		if (strcmp(w1->w_name, b->b_lock1) == 0) {
			if (strcmp(w2->w_name, b->b_lock2) == 0)
				return (1);
			continue;
		}
		if (strcmp(w1->w_name, b->b_lock2) == 0)
			if (strcmp(w2->w_name, b->b_lock1) == 0)
				return (1);
	}
	return (0);
}
#endif

static struct witness *
witness_get(void)
{
	struct witness *w;

	if (witness_watch == 0) {
		mtx_unlock_spin(&w_mtx);
		return (NULL);
	}
	if (STAILQ_EMPTY(&w_free)) {
		witness_watch = 0;
		mtx_unlock_spin(&w_mtx);
		printf("%s: witness exhausted\n", __func__);
		return (NULL);
	}
	w = STAILQ_FIRST(&w_free);
	STAILQ_REMOVE_HEAD(&w_free, w_list);
	w_free_cnt--;
	bzero(w, sizeof(*w));
	return (w);
}

static void
witness_free(struct witness *w)
{

	STAILQ_INSERT_HEAD(&w_free, w, w_list);
	w_free_cnt++;
}

static struct witness_child_list_entry *
witness_child_get(void)
{
	struct witness_child_list_entry *wcl;

	if (witness_watch == 0) {
		mtx_unlock_spin(&w_mtx);
		return (NULL);
	}
	wcl = w_child_free;
	if (wcl == NULL) {
		witness_watch = 0;
		mtx_unlock_spin(&w_mtx);
		printf("%s: witness exhausted\n", __func__);
		return (NULL);
	}
	w_child_free = wcl->wcl_next;
	w_child_free_cnt--;
	bzero(wcl, sizeof(*wcl));
	return (wcl);
}

static void
witness_child_free(struct witness_child_list_entry *wcl)
{

	wcl->wcl_next = w_child_free;
	w_child_free = wcl;
	w_child_free_cnt++;
}

static struct lock_list_entry *
witness_lock_list_get(void)
{
	struct lock_list_entry *lle;

	if (witness_watch == 0)
		return (NULL);
	mtx_lock_spin(&w_mtx);
	lle = w_lock_list_free;
	if (lle == NULL) {
		witness_watch = 0;
		mtx_unlock_spin(&w_mtx);
		printf("%s: witness exhausted\n", __func__);
		return (NULL);
	}
	w_lock_list_free = lle->ll_next;
	mtx_unlock_spin(&w_mtx);
	bzero(lle, sizeof(*lle));
	return (lle);
}
		
static void
witness_lock_list_free(struct lock_list_entry *lle)
{

	mtx_lock_spin(&w_mtx);
	lle->ll_next = w_lock_list_free;
	w_lock_list_free = lle;
	mtx_unlock_spin(&w_mtx);
}

static struct lock_instance *
find_instance(struct lock_list_entry *lock_list, struct lock_object *lock)
{
	struct lock_list_entry *lle;
	struct lock_instance *instance;
	int i;

	for (lle = lock_list; lle != NULL; lle = lle->ll_next)
		for (i = lle->ll_count - 1; i >= 0; i--) {
			instance = &lle->ll_children[i];
			if (instance->li_lock == lock)
				return (instance);
		}
	return (NULL);
}

static void
witness_list_lock(struct lock_instance *instance)
{
	struct lock_object *lock;

	lock = instance->li_lock;
	printf("%s %s %s", (instance->li_flags & LI_EXCLUSIVE) != 0 ?
	    "exclusive" : "shared", LOCK_CLASS(lock)->lc_name, lock->lo_name);
	if (lock->lo_type != lock->lo_name)
		printf(" (%s)", lock->lo_type);
	printf(" r = %d (%p) locked @ %s:%d\n",
	    instance->li_flags & LI_RECURSEMASK, lock, instance->li_file,
	    instance->li_line);
}

#ifdef DDB
static int
witness_thread_has_locks(struct thread *td)
{

	return (td->td_sleeplocks != NULL);
}

static int
witness_proc_has_locks(struct proc *p)
{
	struct thread *td;

	FOREACH_THREAD_IN_PROC(p, td) {
		if (witness_thread_has_locks(td))
			return (1);
	}
	return (0);
}
#endif

int
witness_list_locks(struct lock_list_entry **lock_list)
{
	struct lock_list_entry *lle;
	int i, nheld;

	nheld = 0;
	for (lle = *lock_list; lle != NULL; lle = lle->ll_next)
		for (i = lle->ll_count - 1; i >= 0; i--) {
			witness_list_lock(&lle->ll_children[i]);
			nheld++;
		}
	return (nheld);
}

/*
 * This is a bit risky at best.  We call this function when we have timed
 * out acquiring a spin lock, and we assume that the other CPU is stuck
 * with this lock held.  So, we go groveling around in the other CPU's
 * per-cpu data to try to find the lock instance for this spin lock to
 * see when it was last acquired.
 */
void
witness_display_spinlock(struct lock_object *lock, struct thread *owner)
{
	struct lock_instance *instance;
	struct pcpu *pc;

	if (owner->td_critnest == 0 || owner->td_oncpu == NOCPU)
		return;
	pc = pcpu_find(owner->td_oncpu);
	instance = find_instance(pc->pc_spinlocks, lock);
	if (instance != NULL)
		witness_list_lock(instance);
}

void
witness_save(struct lock_object *lock, const char **filep, int *linep)
{
	struct lock_list_entry *lock_list;
	struct lock_instance *instance;
	struct lock_class *class;

	KASSERT(!witness_cold, ("%s: witness_cold", __func__));
	if (lock->lo_witness == NULL || witness_watch == 0 || panicstr != NULL)
		return;
	class = LOCK_CLASS(lock);
	if (class->lc_flags & LC_SLEEPLOCK)
		lock_list = curthread->td_sleeplocks;
	else {
		if (witness_skipspin)
			return;
		lock_list = PCPU_GET(spinlocks);
	}
	instance = find_instance(lock_list, lock);
	if (instance == NULL)
		panic("%s: lock (%s) %s not locked", __func__,
		    class->lc_name, lock->lo_name);
	*filep = instance->li_file;
	*linep = instance->li_line;
}

void
witness_restore(struct lock_object *lock, const char *file, int line)
{
	struct lock_list_entry *lock_list;
	struct lock_instance *instance;
	struct lock_class *class;

	KASSERT(!witness_cold, ("%s: witness_cold", __func__));
	if (lock->lo_witness == NULL || witness_watch == 0 || panicstr != NULL)
		return;
	class = LOCK_CLASS(lock);
	if (class->lc_flags & LC_SLEEPLOCK)
		lock_list = curthread->td_sleeplocks;
	else {
		if (witness_skipspin)
			return;
		lock_list = PCPU_GET(spinlocks);
	}
	instance = find_instance(lock_list, lock);
	if (instance == NULL)
		panic("%s: lock (%s) %s not locked", __func__,
		    class->lc_name, lock->lo_name);
	lock->lo_witness->w_file = file;
	lock->lo_witness->w_line = line;
	instance->li_file = file;
	instance->li_line = line;
}

void
witness_assert(struct lock_object *lock, int flags, const char *file, int line)
{
#ifdef INVARIANT_SUPPORT
	struct lock_instance *instance;
	struct lock_class *class;

	if (lock->lo_witness == NULL || witness_watch == 0 || panicstr != NULL)
		return;
	class = LOCK_CLASS(lock);
	if ((class->lc_flags & LC_SLEEPLOCK) != 0)
		instance = find_instance(curthread->td_sleeplocks, lock);
	else if ((class->lc_flags & LC_SPINLOCK) != 0)
		instance = find_instance(PCPU_GET(spinlocks), lock);
	else {
		panic("Lock (%s) %s is not sleep or spin!",
		    class->lc_name, lock->lo_name);
	}
	file = fixup_filename(file);
	switch (flags) {
	case LA_UNLOCKED:
		if (instance != NULL)
			panic("Lock (%s) %s locked @ %s:%d.",
			    class->lc_name, lock->lo_name, file, line);
		break;
	case LA_LOCKED:
	case LA_LOCKED | LA_RECURSED:
	case LA_LOCKED | LA_NOTRECURSED:
	case LA_SLOCKED:
	case LA_SLOCKED | LA_RECURSED:
	case LA_SLOCKED | LA_NOTRECURSED:
	case LA_XLOCKED:
	case LA_XLOCKED | LA_RECURSED:
	case LA_XLOCKED | LA_NOTRECURSED:
		if (instance == NULL) {
			panic("Lock (%s) %s not locked @ %s:%d.",
			    class->lc_name, lock->lo_name, file, line);
			break;
		}
		if ((flags & LA_XLOCKED) != 0 &&
		    (instance->li_flags & LI_EXCLUSIVE) == 0)
			panic("Lock (%s) %s not exclusively locked @ %s:%d.",
			    class->lc_name, lock->lo_name, file, line);
		if ((flags & LA_SLOCKED) != 0 &&
		    (instance->li_flags & LI_EXCLUSIVE) != 0)
			panic("Lock (%s) %s exclusively locked @ %s:%d.",
			    class->lc_name, lock->lo_name, file, line);
		if ((flags & LA_RECURSED) != 0 &&
		    (instance->li_flags & LI_RECURSEMASK) == 0)
			panic("Lock (%s) %s not recursed @ %s:%d.",
			    class->lc_name, lock->lo_name, file, line);
		if ((flags & LA_NOTRECURSED) != 0 &&
		    (instance->li_flags & LI_RECURSEMASK) != 0)
			panic("Lock (%s) %s recursed @ %s:%d.",
			    class->lc_name, lock->lo_name, file, line);
		break;
	default:
		panic("Invalid lock assertion at %s:%d.", file, line);

	}
#endif	/* INVARIANT_SUPPORT */
}

#ifdef DDB
static void
witness_list(struct thread *td)
{

	KASSERT(!witness_cold, ("%s: witness_cold", __func__));
	KASSERT(kdb_active, ("%s: not in the debugger", __func__));

	if (witness_watch == 0)
		return;

	witness_list_locks(&td->td_sleeplocks);

	/*
	 * We only handle spinlocks if td == curthread.  This is somewhat broken
	 * if td is currently executing on some other CPU and holds spin locks
	 * as we won't display those locks.  If we had a MI way of getting
	 * the per-cpu data for a given cpu then we could use
	 * td->td_oncpu to get the list of spinlocks for this thread
	 * and "fix" this.
	 *
	 * That still wouldn't really fix this unless we locked the scheduler
	 * lock or stopped the other CPU to make sure it wasn't changing the
	 * list out from under us.  It is probably best to just not try to
	 * handle threads on other CPU's for now.
	 */
	if (td == curthread && PCPU_GET(spinlocks) != NULL)
		witness_list_locks(PCPU_PTR(spinlocks));
}

DB_SHOW_COMMAND(locks, db_witness_list)
{
	struct thread *td;

	if (have_addr)
		td = db_lookup_thread(addr, TRUE);
	else
		td = kdb_thread;
	witness_list(td);
}

DB_SHOW_COMMAND(alllocks, db_witness_list_all)
{
	struct thread *td;
	struct proc *p;

	/*
	 * It would be nice to list only threads and processes that actually
	 * held sleep locks, but that information is currently not exported
	 * by WITNESS.
	 */
	FOREACH_PROC_IN_SYSTEM(p) {
		if (!witness_proc_has_locks(p))
			continue;
		FOREACH_THREAD_IN_PROC(p, td) {
			if (!witness_thread_has_locks(td))
				continue;
			db_printf("Process %d (%s) thread %p (%d)\n", p->p_pid,
			    td->td_name, td, td->td_tid);
			witness_list(td);
		}
	}
}

DB_SHOW_COMMAND(witness, db_witness_display)
{

	witness_display(db_printf);
}
#endif