summaryrefslogtreecommitdiff
path: root/sys/kern/vfs_vnode.c
blob: e4729dc72d3acabfae69762b21a5d6df018b0acb (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
/*	$NetBSD: vfs_vnode.c,v 1.32 2014/02/27 16:51:38 hannken Exp $	*/

/*-
 * Copyright (c) 1997-2011 The NetBSD Foundation, Inc.
 * All rights reserved.
 *
 * This code is derived from software contributed to The NetBSD Foundation
 * by Jason R. Thorpe of the Numerical Aerospace Simulation Facility,
 * NASA Ames Research Center, by Charles M. Hannum, and by Andrew Doran.
 *
 * 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.
 *
 * THIS SOFTWARE IS PROVIDED BY THE NETBSD FOUNDATION, INC. AND CONTRIBUTORS
 * ``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 FOUNDATION OR CONTRIBUTORS
 * 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.
 */

/*
 * Copyright (c) 1989, 1993
 *	The Regents of the University of California.  All rights reserved.
 * (c) UNIX System Laboratories, Inc.
 * All or some portions of this file are derived from material licensed
 * to the University of California by American Telephone and Telegraph
 * Co. or Unix System Laboratories, Inc. and are reproduced herein with
 * the permission of UNIX System Laboratories, Inc.
 *
 * 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. Neither the name of the University nor the names of its contributors
 *    may be used to endorse or promote products derived from this software
 *    without specific prior written permission.
 *
 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``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 REGENTS OR CONTRIBUTORS 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.
 *
 *	@(#)vfs_subr.c	8.13 (Berkeley) 4/18/94
 */

/*
 * The vnode cache subsystem.
 *
 * Life-cycle
 *
 *	Normally, there are two points where new vnodes are created:
 *	VOP_CREATE(9) and VOP_LOOKUP(9).  The life-cycle of a vnode
 *	starts in one of the following ways:
 *
 *	- Allocation, via getnewvnode(9) and/or vnalloc(9).
 *	- Reclamation of inactive vnode, via vget(9).
 *
 *	Recycle from a free list, via getnewvnode(9) -> getcleanvnode(9)
 *	was another, traditional way.  Currently, only the draining thread
 *	recycles the vnodes.  This behaviour might be revisited.
 *
 *	The life-cycle ends when the last reference is dropped, usually
 *	in VOP_REMOVE(9).  In such case, VOP_INACTIVE(9) is called to inform
 *	the file system that vnode is inactive.  Via this call, file system
 *	indicates whether vnode can be recycled (usually, it checks its own
 *	references, e.g. count of links, whether the file was removed).
 *
 *	Depending on indication, vnode can be put into a free list (cache),
 *	or cleaned via vclean(9), which calls VOP_RECLAIM(9) to disassociate
 *	underlying file system from the vnode, and finally destroyed.
 *
 * Reference counting
 *
 *	Vnode is considered active, if reference count (vnode_t::v_usecount)
 *	is non-zero.  It is maintained using: vref(9) and vrele(9), as well
 *	as vput(9), routines.  Common points holding references are e.g.
 *	file openings, current working directory, mount points, etc.  
 *
 * Note on v_usecount and its locking
 *
 *	At nearly all points it is known that v_usecount could be zero,
 *	the vnode_t::v_interlock will be held.  To change v_usecount away
 *	from zero, the interlock must be held.  To change from a non-zero
 *	value to zero, again the interlock must be held.
 *
 *	Changing the usecount from a non-zero value to a non-zero value can
 *	safely be done using atomic operations, without the interlock held.
 *
 *	Note: if VI_CLEAN is set, vnode_t::v_interlock will be released while
 *	mntvnode_lock is still held.
 *
 *	See PR 41374.
 */

#include <sys/cdefs.h>
__KERNEL_RCSID(0, "$NetBSD: vfs_vnode.c,v 1.32 2014/02/27 16:51:38 hannken Exp $");

#define _VFS_VNODE_PRIVATE

#include <sys/param.h>
#include <sys/kernel.h>

#include <sys/atomic.h>
#include <sys/buf.h>
#include <sys/conf.h>
#include <sys/device.h>
#include <sys/kauth.h>
#include <sys/kmem.h>
#include <sys/kthread.h>
#include <sys/module.h>
#include <sys/mount.h>
#include <sys/namei.h>
#include <sys/syscallargs.h>
#include <sys/sysctl.h>
#include <sys/systm.h>
#include <sys/vnode.h>
#include <sys/wapbl.h>
#include <sys/fstrans.h>

#include <uvm/uvm.h>
#include <uvm/uvm_readahead.h>

/* Flags to vrelel. */
#define	VRELEL_ASYNC_RELE	0x0001	/* Always defer to vrele thread. */
#define	VRELEL_CHANGING_SET	0x0002	/* VI_CHANGING set by caller. */

u_int			numvnodes		__cacheline_aligned;

static pool_cache_t	vnode_cache		__read_mostly;
static struct mount	*dead_mount;

/*
 * There are two free lists: one is for vnodes which have no buffer/page
 * references and one for those which do (i.e. v_holdcnt is non-zero).
 * Vnode recycling mechanism first attempts to look into the former list.
 */
static kmutex_t		vnode_free_list_lock	__cacheline_aligned;
static vnodelst_t	vnode_free_list		__cacheline_aligned;
static vnodelst_t	vnode_hold_list		__cacheline_aligned;
static kcondvar_t	vdrain_cv		__cacheline_aligned;

static vnodelst_t	vrele_list		__cacheline_aligned;
static kmutex_t		vrele_lock		__cacheline_aligned;
static kcondvar_t	vrele_cv		__cacheline_aligned;
static lwp_t *		vrele_lwp		__cacheline_aligned;
static int		vrele_pending		__cacheline_aligned;
static int		vrele_gen		__cacheline_aligned;

static int		cleanvnode(void);
static void		vclean(vnode_t *);
static void		vrelel(vnode_t *, int);
static void		vdrain_thread(void *);
static void		vrele_thread(void *);
static void		vnpanic(vnode_t *, const char *, ...)
    __printflike(2, 3);

/* Routines having to do with the management of the vnode table. */
extern int		(**dead_vnodeop_p)(void *);
extern struct vfsops	dead_vfsops;

void
vfs_vnode_sysinit(void)
{
	int error __diagused;

	vnode_cache = pool_cache_init(sizeof(vnode_t), 0, 0, 0, "vnodepl",
	    NULL, IPL_NONE, NULL, NULL, NULL);
	KASSERT(vnode_cache != NULL);

	dead_mount = vfs_mountalloc(&dead_vfsops, NULL);
	KASSERT(dead_mount != NULL);
	dead_mount->mnt_iflag = IMNT_MPSAFE;

	mutex_init(&vnode_free_list_lock, MUTEX_DEFAULT, IPL_NONE);
	TAILQ_INIT(&vnode_free_list);
	TAILQ_INIT(&vnode_hold_list);
	TAILQ_INIT(&vrele_list);

	mutex_init(&vrele_lock, MUTEX_DEFAULT, IPL_NONE);
	cv_init(&vdrain_cv, "vdrain");
	cv_init(&vrele_cv, "vrele");
	error = kthread_create(PRI_VM, KTHREAD_MPSAFE, NULL, vdrain_thread,
	    NULL, NULL, "vdrain");
	KASSERT(error == 0);
	error = kthread_create(PRI_VM, KTHREAD_MPSAFE, NULL, vrele_thread,
	    NULL, &vrele_lwp, "vrele");
	KASSERT(error == 0);
}

/*
 * Allocate a new, uninitialized vnode.  If 'mp' is non-NULL, this is a
 * marker vnode.
 */
vnode_t *
vnalloc(struct mount *mp)
{
	vnode_t *vp;

	vp = pool_cache_get(vnode_cache, PR_WAITOK);
	KASSERT(vp != NULL);

	memset(vp, 0, sizeof(*vp));
	uvm_obj_init(&vp->v_uobj, &uvm_vnodeops, true, 0);
	cv_init(&vp->v_cv, "vnode");
	/*
	 * Done by memset() above.
	 *	LIST_INIT(&vp->v_nclist);
	 *	LIST_INIT(&vp->v_dnclist);
	 */

	if (mp != NULL) {
		vp->v_mount = mp;
		vp->v_type = VBAD;
		vp->v_iflag = VI_MARKER;
	} else {
		rw_init(&vp->v_lock);
	}

	return vp;
}

/*
 * Free an unused, unreferenced vnode.
 */
void
vnfree(vnode_t *vp)
{

	KASSERT(vp->v_usecount == 0);

	if ((vp->v_iflag & VI_MARKER) == 0) {
		rw_destroy(&vp->v_lock);
		mutex_enter(&vnode_free_list_lock);
		numvnodes--;
		mutex_exit(&vnode_free_list_lock);
	}

	/*
	 * Note: the vnode interlock will either be freed, of reference
	 * dropped (if VI_LOCKSHARE was in use).
	 */
	uvm_obj_destroy(&vp->v_uobj, true);
	cv_destroy(&vp->v_cv);
	pool_cache_put(vnode_cache, vp);
}

/*
 * cleanvnode: grab a vnode from freelist, clean and free it.
 *
 * => Releases vnode_free_list_lock.
 */
static int
cleanvnode(void)
{
	vnode_t *vp;
	vnodelst_t *listhd;
	struct mount *mp;

	KASSERT(mutex_owned(&vnode_free_list_lock));

	listhd = &vnode_free_list;
try_nextlist:
	TAILQ_FOREACH(vp, listhd, v_freelist) {
		/*
		 * It's safe to test v_usecount and v_iflag
		 * without holding the interlock here, since
		 * these vnodes should never appear on the
		 * lists.
		 */
		KASSERT(vp->v_usecount == 0);
		KASSERT((vp->v_iflag & VI_CLEAN) == 0);
		KASSERT(vp->v_freelisthd == listhd);

		if (!mutex_tryenter(vp->v_interlock))
			continue;
		if ((vp->v_iflag & VI_XLOCK) != 0) {
			mutex_exit(vp->v_interlock);
			continue;
		}
		mp = vp->v_mount;
		if (fstrans_start_nowait(mp, FSTRANS_SHARED) != 0) {
			mutex_exit(vp->v_interlock);
			continue;
		}
		break;
	}

	if (vp == NULL) {
		if (listhd == &vnode_free_list) {
			listhd = &vnode_hold_list;
			goto try_nextlist;
		}
		mutex_exit(&vnode_free_list_lock);
		return EBUSY;
	}

	/* Remove it from the freelist. */
	TAILQ_REMOVE(listhd, vp, v_freelist);
	vp->v_freelisthd = NULL;
	mutex_exit(&vnode_free_list_lock);

	KASSERT(vp->v_usecount == 0);

	/*
	 * The vnode is still associated with a file system, so we must
	 * clean it out before freeing it.  We need to add a reference
	 * before doing this.
	 */
	vp->v_usecount = 1;
	KASSERT((vp->v_iflag & VI_CHANGING) == 0);
	vp->v_iflag |= VI_CHANGING;
	vclean(vp);
	vrelel(vp, VRELEL_CHANGING_SET);
	fstrans_done(mp);

	return 0;
}

/*
 * getnewvnode: return a fresh vnode.
 *
 * => Returns referenced vnode, moved into the mount queue.
 * => Shares the interlock specified by 'slock', if it is not NULL.
 */
int
getnewvnode(enum vtagtype tag, struct mount *mp, int (**vops)(void *),
    kmutex_t *slock, vnode_t **vpp)
{
	struct uvm_object *uobj __diagused;
	vnode_t *vp;
	int error = 0;

	if (mp != NULL) {
		/*
		 * Mark filesystem busy while we are creating a vnode.
		 * If unmount is in progress, this will fail.
		 */
		error = vfs_busy(mp, NULL);
		if (error)
			return error;
	}

	vp = NULL;

	/* Allocate a new vnode. */
	mutex_enter(&vnode_free_list_lock);
	numvnodes++;
	if (numvnodes > desiredvnodes + desiredvnodes / 10)
		cv_signal(&vdrain_cv);
	mutex_exit(&vnode_free_list_lock);
	vp = vnalloc(NULL);

	KASSERT(vp->v_freelisthd == NULL);
	KASSERT(LIST_EMPTY(&vp->v_nclist));
	KASSERT(LIST_EMPTY(&vp->v_dnclist));

	/* Initialize vnode. */
	vp->v_usecount = 1;
	vp->v_type = VNON;
	vp->v_tag = tag;
	vp->v_op = vops;
	vp->v_data = NULL;

	uobj = &vp->v_uobj;
	KASSERT(uobj->pgops == &uvm_vnodeops);
	KASSERT(uobj->uo_npages == 0);
	KASSERT(TAILQ_FIRST(&uobj->memq) == NULL);
	vp->v_size = vp->v_writesize = VSIZENOTSET;

	/* Share the vnode_t::v_interlock, if requested. */
	if (slock) {
		/* Set the interlock and mark that it is shared. */
		KASSERT(vp->v_mount == NULL);
		mutex_obj_hold(slock);
		uvm_obj_setlock(&vp->v_uobj, slock);
		KASSERT(vp->v_interlock == slock);
		vp->v_iflag |= VI_LOCKSHARE;
	}

	/* Finally, move vnode into the mount queue. */
	vfs_insmntque(vp, mp);

	if (mp != NULL) {
		if ((mp->mnt_iflag & IMNT_MPSAFE) != 0)
			vp->v_vflag |= VV_MPSAFE;
		vfs_unbusy(mp, true, NULL);
	}

	*vpp = vp;
	return 0;
}

/*
 * This is really just the reverse of getnewvnode(). Needed for
 * VFS_VGET functions who may need to push back a vnode in case
 * of a locking race.
 */
void
ungetnewvnode(vnode_t *vp)
{

	KASSERT(vp->v_usecount == 1);
	KASSERT(vp->v_data == NULL);
	KASSERT(vp->v_freelisthd == NULL);

	mutex_enter(vp->v_interlock);
	vp->v_iflag |= VI_CLEAN;
	vrelel(vp, 0);
}

/*
 * Helper thread to keep the number of vnodes below desiredvnodes.
 */
static void
vdrain_thread(void *cookie)
{
	int error;

	mutex_enter(&vnode_free_list_lock);

	for (;;) {
		cv_timedwait(&vdrain_cv, &vnode_free_list_lock, hz);
		while (numvnodes > desiredvnodes) {
			error = cleanvnode();
			if (error)
				kpause("vndsbusy", false, hz, NULL);
			mutex_enter(&vnode_free_list_lock);
			if (error)
				break;
		}
	}
}

/*
 * Remove a vnode from its freelist.
 */
void
vremfree(vnode_t *vp)
{

	KASSERT(mutex_owned(vp->v_interlock));
	KASSERT(vp->v_usecount == 0);

	/*
	 * Note that the reference count must not change until
	 * the vnode is removed.
	 */
	mutex_enter(&vnode_free_list_lock);
	if (vp->v_holdcnt > 0) {
		KASSERT(vp->v_freelisthd == &vnode_hold_list);
	} else {
		KASSERT(vp->v_freelisthd == &vnode_free_list);
	}
	TAILQ_REMOVE(vp->v_freelisthd, vp, v_freelist);
	vp->v_freelisthd = NULL;
	mutex_exit(&vnode_free_list_lock);
}

/*
 * vget: get a particular vnode from the free list, increment its reference
 * count and lock it.
 *
 * => Should be called with v_interlock held.
 *
 * If VI_CHANGING is set, the vnode may be eliminated in vgone()/vclean().
 * In that case, we cannot grab the vnode, so the process is awakened when
 * the transition is completed, and an error returned to indicate that the
 * vnode is no longer usable.
 */
int
vget(vnode_t *vp, int flags)
{
	int error = 0;

	KASSERT((vp->v_iflag & VI_MARKER) == 0);
	KASSERT(mutex_owned(vp->v_interlock));
	KASSERT((flags & ~(LK_SHARED|LK_EXCLUSIVE|LK_NOWAIT)) == 0);

	/*
	 * Before adding a reference, we must remove the vnode
	 * from its freelist.
	 */
	if (vp->v_usecount == 0) {
		vremfree(vp);
		vp->v_usecount = 1;
	} else {
		atomic_inc_uint(&vp->v_usecount);
	}

	/*
	 * If the vnode is in the process of changing state we wait
	 * for the change to complete and take care not to return
	 * a clean vnode.
	 */
	if ((vp->v_iflag & VI_CHANGING) != 0) {
		if ((flags & LK_NOWAIT) != 0) {
			vrelel(vp, 0);
			return EBUSY;
		}
		vwait(vp, VI_CHANGING);
		if ((vp->v_iflag & VI_CLEAN) != 0) {
			vrelel(vp, 0);
			return ENOENT;
		}
	}

	/*
	 * Ok, we got it in good shape.  Just locking left.
	 */
	KASSERT((vp->v_iflag & VI_CLEAN) == 0);
	mutex_exit(vp->v_interlock);
	if (flags & (LK_EXCLUSIVE | LK_SHARED)) {
		error = vn_lock(vp, flags);
		if (error != 0) {
			vrele(vp);
		}
	}
	return error;
}

/*
 * vput: unlock and release the reference.
 */
void
vput(vnode_t *vp)
{

	KASSERT((vp->v_iflag & VI_MARKER) == 0);

	VOP_UNLOCK(vp);
	vrele(vp);
}

/*
 * Try to drop reference on a vnode.  Abort if we are releasing the
 * last reference.  Note: this _must_ succeed if not the last reference.
 */
static inline bool
vtryrele(vnode_t *vp)
{
	u_int use, next;

	for (use = vp->v_usecount;; use = next) {
		if (use == 1) {
			return false;
		}
		KASSERT(use > 1);
		next = atomic_cas_uint(&vp->v_usecount, use, use - 1);
		if (__predict_true(next == use)) {
			return true;
		}
	}
}

/*
 * Vnode release.  If reference count drops to zero, call inactive
 * routine and either return to freelist or free to the pool.
 */
static void
vrelel(vnode_t *vp, int flags)
{
	bool recycle, defer;
	int error;

	KASSERT(mutex_owned(vp->v_interlock));
	KASSERT((vp->v_iflag & VI_MARKER) == 0);
	KASSERT(vp->v_freelisthd == NULL);

	if (__predict_false(vp->v_op == dead_vnodeop_p &&
	    (vp->v_iflag & (VI_CLEAN|VI_XLOCK)) == 0)) {
		vnpanic(vp, "dead but not clean");
	}

	/*
	 * If not the last reference, just drop the reference count
	 * and unlock.
	 */
	if (vtryrele(vp)) {
		if ((flags & VRELEL_CHANGING_SET) != 0) {
			KASSERT((vp->v_iflag & VI_CHANGING) != 0);
			vp->v_iflag &= ~VI_CHANGING;
			cv_broadcast(&vp->v_cv);
		}
		mutex_exit(vp->v_interlock);
		return;
	}
	if (vp->v_usecount <= 0 || vp->v_writecount != 0) {
		vnpanic(vp, "%s: bad ref count", __func__);
	}

	KASSERT((vp->v_iflag & VI_XLOCK) == 0);

#ifdef DIAGNOSTIC
	if ((vp->v_type == VBLK || vp->v_type == VCHR) &&
	    vp->v_specnode != NULL && vp->v_specnode->sn_opencnt != 0) {
		vprint("vrelel: missing VOP_CLOSE()", vp);
	}
#endif

	/*
	 * If not clean, deactivate the vnode, but preserve
	 * our reference across the call to VOP_INACTIVE().
	 */
	if ((vp->v_iflag & VI_CLEAN) == 0) {
		recycle = false;

		/*
		 * XXX This ugly block can be largely eliminated if
		 * locking is pushed down into the file systems.
		 *
		 * Defer vnode release to vrele_thread if caller
		 * requests it explicitly or is the pagedaemon.
		 */
		if ((curlwp == uvm.pagedaemon_lwp) ||
		    (flags & VRELEL_ASYNC_RELE) != 0) {
			defer = true;
		} else if (curlwp == vrele_lwp) {
			/*
			 * We have to try harder.
			 */
			mutex_exit(vp->v_interlock);
			error = vn_lock(vp, LK_EXCLUSIVE | LK_RETRY);
			KASSERT(error == 0);
			mutex_enter(vp->v_interlock);
			defer = false;
		} else {
			/* If we can't acquire the lock, then defer. */
			mutex_exit(vp->v_interlock);
			error = vn_lock(vp,
			    LK_EXCLUSIVE | LK_RETRY | LK_NOWAIT);
			defer = (error != 0);
			mutex_enter(vp->v_interlock);
		}

		KASSERT(mutex_owned(vp->v_interlock));
		KASSERT(! (curlwp == vrele_lwp && defer));

		if (defer) {
			/*
			 * Defer reclaim to the kthread; it's not safe to
			 * clean it here.  We donate it our last reference.
			 */
			if ((flags & VRELEL_CHANGING_SET) != 0) {
				KASSERT((vp->v_iflag & VI_CHANGING) != 0);
				vp->v_iflag &= ~VI_CHANGING;
				cv_broadcast(&vp->v_cv);
			}
			mutex_enter(&vrele_lock);
			TAILQ_INSERT_TAIL(&vrele_list, vp, v_freelist);
			if (++vrele_pending > (desiredvnodes >> 8))
				cv_signal(&vrele_cv); 
			mutex_exit(&vrele_lock);
			mutex_exit(vp->v_interlock);
			return;
		}

		/*
		 * If the node got another reference while we
		 * released the interlock, don't try to inactivate it yet.
		 */
		if (__predict_false(vtryrele(vp))) {
			VOP_UNLOCK(vp);
			if ((flags & VRELEL_CHANGING_SET) != 0) {
				KASSERT((vp->v_iflag & VI_CHANGING) != 0);
				vp->v_iflag &= ~VI_CHANGING;
				cv_broadcast(&vp->v_cv);
			}
			mutex_exit(vp->v_interlock);
			return;
		}

		if ((flags & VRELEL_CHANGING_SET) == 0) {
			KASSERT((vp->v_iflag & VI_CHANGING) == 0);
			vp->v_iflag |= VI_CHANGING;
		}
		mutex_exit(vp->v_interlock);

		/*
		 * The vnode can gain another reference while being
		 * deactivated.  If VOP_INACTIVE() indicates that
		 * the described file has been deleted, then recycle
		 * the vnode irrespective of additional references.
		 * Another thread may be waiting to re-use the on-disk
		 * inode.
		 *
		 * Note that VOP_INACTIVE() will drop the vnode lock.
		 */
		VOP_INACTIVE(vp, &recycle);
		mutex_enter(vp->v_interlock);
		if (!recycle) {
			if (vtryrele(vp)) {
				KASSERT((vp->v_iflag & VI_CHANGING) != 0);
				vp->v_iflag &= ~VI_CHANGING;
				cv_broadcast(&vp->v_cv);
				mutex_exit(vp->v_interlock);
				return;
			}
		}

		/* Take care of space accounting. */
		if (vp->v_iflag & VI_EXECMAP) {
			atomic_add_int(&uvmexp.execpages,
			    -vp->v_uobj.uo_npages);
			atomic_add_int(&uvmexp.filepages,
			    vp->v_uobj.uo_npages);
		}
		vp->v_iflag &= ~(VI_TEXT|VI_EXECMAP|VI_WRMAP);
		vp->v_vflag &= ~VV_MAPPED;

		/*
		 * Recycle the vnode if the file is now unused (unlinked),
		 * otherwise just free it.
		 */
		if (recycle) {
			vclean(vp);
		}
		KASSERT(vp->v_usecount > 0);
	} else { /* vnode was already clean */
		if ((flags & VRELEL_CHANGING_SET) == 0) {
			KASSERT((vp->v_iflag & VI_CHANGING) == 0);
			vp->v_iflag |= VI_CHANGING;
		}
	}

	if (atomic_dec_uint_nv(&vp->v_usecount) != 0) {
		/* Gained another reference while being reclaimed. */
		KASSERT((vp->v_iflag & VI_CHANGING) != 0);
		vp->v_iflag &= ~VI_CHANGING;
		cv_broadcast(&vp->v_cv);
		mutex_exit(vp->v_interlock);
		return;
	}

	if ((vp->v_iflag & VI_CLEAN) != 0) {
		/*
		 * It's clean so destroy it.  It isn't referenced
		 * anywhere since it has been reclaimed.
		 */
		KASSERT(vp->v_holdcnt == 0);
		KASSERT(vp->v_writecount == 0);
		mutex_exit(vp->v_interlock);
		vfs_insmntque(vp, NULL);
		if (vp->v_type == VBLK || vp->v_type == VCHR) {
			spec_node_destroy(vp);
		}
		vnfree(vp);
	} else {
		/*
		 * Otherwise, put it back onto the freelist.  It
		 * can't be destroyed while still associated with
		 * a file system.
		 */
		mutex_enter(&vnode_free_list_lock);
		if (vp->v_holdcnt > 0) {
			vp->v_freelisthd = &vnode_hold_list;
		} else {
			vp->v_freelisthd = &vnode_free_list;
		}
		TAILQ_INSERT_TAIL(vp->v_freelisthd, vp, v_freelist);
		mutex_exit(&vnode_free_list_lock);
		KASSERT((vp->v_iflag & VI_CHANGING) != 0);
		vp->v_iflag &= ~VI_CHANGING;
		cv_broadcast(&vp->v_cv);
		mutex_exit(vp->v_interlock);
	}
}

void
vrele(vnode_t *vp)
{

	KASSERT((vp->v_iflag & VI_MARKER) == 0);

	if (vtryrele(vp)) {
		return;
	}
	mutex_enter(vp->v_interlock);
	vrelel(vp, 0);
}

/*
 * Asynchronous vnode release, vnode is released in different context.
 */
void
vrele_async(vnode_t *vp)
{

	KASSERT((vp->v_iflag & VI_MARKER) == 0);

	if (vtryrele(vp)) {
		return;
	}
	mutex_enter(vp->v_interlock);
	vrelel(vp, VRELEL_ASYNC_RELE);
}

static void
vrele_thread(void *cookie)
{
	vnode_t *vp;

	for (;;) {
		mutex_enter(&vrele_lock);
		while (TAILQ_EMPTY(&vrele_list)) {
			vrele_gen++;
			cv_broadcast(&vrele_cv);
			cv_timedwait(&vrele_cv, &vrele_lock, hz);
		}
		vp = TAILQ_FIRST(&vrele_list);
		TAILQ_REMOVE(&vrele_list, vp, v_freelist);
		vrele_pending--;
		mutex_exit(&vrele_lock);

		/*
		 * If not the last reference, then ignore the vnode
		 * and look for more work.
		 */
		mutex_enter(vp->v_interlock);
		vrelel(vp, 0);
	}
}

void
vrele_flush(void)
{
	int gen;

	mutex_enter(&vrele_lock);
	gen = vrele_gen;
	while (vrele_pending && gen == vrele_gen) {
		cv_broadcast(&vrele_cv);
		cv_wait(&vrele_cv, &vrele_lock);
	}
	mutex_exit(&vrele_lock);
}

/*
 * Vnode reference, where a reference is already held by some other
 * object (for example, a file structure).
 */
void
vref(vnode_t *vp)
{

	KASSERT((vp->v_iflag & VI_MARKER) == 0);
	KASSERT(vp->v_usecount != 0);

	atomic_inc_uint(&vp->v_usecount);
}

/*
 * Page or buffer structure gets a reference.
 * Called with v_interlock held.
 */
void
vholdl(vnode_t *vp)
{

	KASSERT(mutex_owned(vp->v_interlock));
	KASSERT((vp->v_iflag & VI_MARKER) == 0);

	if (vp->v_holdcnt++ == 0 && vp->v_usecount == 0) {
		mutex_enter(&vnode_free_list_lock);
		KASSERT(vp->v_freelisthd == &vnode_free_list);
		TAILQ_REMOVE(vp->v_freelisthd, vp, v_freelist);
		vp->v_freelisthd = &vnode_hold_list;
		TAILQ_INSERT_TAIL(vp->v_freelisthd, vp, v_freelist);
		mutex_exit(&vnode_free_list_lock);
	}
}

/*
 * Page or buffer structure frees a reference.
 * Called with v_interlock held.
 */
void
holdrelel(vnode_t *vp)
{

	KASSERT(mutex_owned(vp->v_interlock));
	KASSERT((vp->v_iflag & VI_MARKER) == 0);

	if (vp->v_holdcnt <= 0) {
		vnpanic(vp, "%s: holdcnt vp %p", __func__, vp);
	}

	vp->v_holdcnt--;
	if (vp->v_holdcnt == 0 && vp->v_usecount == 0) {
		mutex_enter(&vnode_free_list_lock);
		KASSERT(vp->v_freelisthd == &vnode_hold_list);
		TAILQ_REMOVE(vp->v_freelisthd, vp, v_freelist);
		vp->v_freelisthd = &vnode_free_list;
		TAILQ_INSERT_TAIL(vp->v_freelisthd, vp, v_freelist);
		mutex_exit(&vnode_free_list_lock);
	}
}

/*
 * Disassociate the underlying file system from a vnode.
 *
 * Must be called with the interlock held, and will return with it held.
 */
static void
vclean(vnode_t *vp)
{
	lwp_t *l = curlwp;
	bool recycle, active, doclose;
	int error;

	KASSERT(mutex_owned(vp->v_interlock));
	KASSERT((vp->v_iflag & VI_MARKER) == 0);
	KASSERT(vp->v_usecount != 0);

	/* If already clean, nothing to do. */
	if ((vp->v_iflag & VI_CLEAN) != 0) {
		return;
	}

	active = (vp->v_usecount > 1);
	doclose = ! (active && vp->v_type == VBLK &&
	    spec_node_getmountedfs(vp) != NULL);
	mutex_exit(vp->v_interlock);

	vn_lock(vp, LK_EXCLUSIVE | LK_RETRY);

	/*
	 * Prevent the vnode from being recycled or brought into use
	 * while we clean it out.
	 */
	mutex_enter(vp->v_interlock);
	KASSERT((vp->v_iflag & (VI_XLOCK | VI_CLEAN)) == 0);
	vp->v_iflag |= VI_XLOCK;
	if (vp->v_iflag & VI_EXECMAP) {
		atomic_add_int(&uvmexp.execpages, -vp->v_uobj.uo_npages);
		atomic_add_int(&uvmexp.filepages, vp->v_uobj.uo_npages);
	}
	vp->v_iflag &= ~(VI_TEXT|VI_EXECMAP);
	mutex_exit(vp->v_interlock);

	/*
	 * Clean out any cached data associated with the vnode.
	 * If purging an active vnode, it must be closed and
	 * deactivated before being reclaimed. Note that the
	 * VOP_INACTIVE will unlock the vnode.
	 */
	if (doclose) {
		error = vinvalbuf(vp, V_SAVE, NOCRED, l, 0, 0);
		if (error != 0) {
			if (wapbl_vphaswapbl(vp))
				WAPBL_DISCARD(wapbl_vptomp(vp));
			error = vinvalbuf(vp, 0, NOCRED, l, 0, 0);
		}
		KASSERT(error == 0);
		KASSERT((vp->v_iflag & VI_ONWORKLST) == 0);
		if (active && (vp->v_type == VBLK || vp->v_type == VCHR)) {
			 spec_node_revoke(vp);
		}
	}
	if (active) {
		VOP_INACTIVE(vp, &recycle);
	} else {
		/*
		 * Any other processes trying to obtain this lock must first
		 * wait for VI_XLOCK to clear, then call the new lock operation.
		 */
		VOP_UNLOCK(vp);
	}

	/* Disassociate the underlying file system from the vnode. */
	if (VOP_RECLAIM(vp)) {
		vnpanic(vp, "%s: cannot reclaim", __func__);
	}

	KASSERT(vp->v_data == NULL);
	KASSERT(vp->v_uobj.uo_npages == 0);

	if (vp->v_type == VREG && vp->v_ractx != NULL) {
		uvm_ra_freectx(vp->v_ractx);
		vp->v_ractx = NULL;
	}

	/* Purge name cache. */
	cache_purge(vp);

	/* Move to dead mount. */
	vp->v_vflag &= ~VV_ROOT;
	atomic_inc_uint(&dead_mount->mnt_refcnt);
	vfs_insmntque(vp, dead_mount);

	/* Done with purge, notify sleepers of the grim news. */
	mutex_enter(vp->v_interlock);
	if (doclose) {
		vp->v_op = dead_vnodeop_p;
		vp->v_vflag |= VV_LOCKSWORK;
		vp->v_iflag |= VI_CLEAN;
	} else {
		vp->v_op = spec_vnodeop_p;
		vp->v_vflag &= ~VV_LOCKSWORK;
	}
	vp->v_tag = VT_NON;
	KNOTE(&vp->v_klist, NOTE_REVOKE);
	vp->v_iflag &= ~VI_XLOCK;
	cv_broadcast(&vp->v_cv);

	KASSERT((vp->v_iflag & VI_ONWORKLST) == 0);
}

/*
 * Recycle an unused vnode to the front of the free list.
 * Release the passed interlock if the vnode will be recycled.
 */
int
vrecycle(vnode_t *vp, kmutex_t *inter_lkp)
{

	KASSERT((vp->v_iflag & VI_MARKER) == 0);

	mutex_enter(vp->v_interlock);
	if (vp->v_usecount != 0 || (vp->v_iflag & (VI_CLEAN|VI_XLOCK)) != 0) {
		mutex_exit(vp->v_interlock);
		return 0;
	}
	if (inter_lkp) {
		mutex_exit(inter_lkp);
	}
	vremfree(vp);
	vp->v_usecount = 1;
	KASSERT((vp->v_iflag & VI_CHANGING) == 0);
	vp->v_iflag |= VI_CHANGING;
	vclean(vp);
	vrelel(vp, VRELEL_CHANGING_SET);
	return 1;
}

/*
 * Eliminate all activity associated with the requested vnode
 * and with all vnodes aliased to the requested vnode.
 */
void
vrevoke(vnode_t *vp)
{
	vnode_t *vq;
	enum vtype type;
	dev_t dev;

	KASSERT(vp->v_usecount > 0);

	mutex_enter(vp->v_interlock);
	if ((vp->v_iflag & VI_CLEAN) != 0) {
		mutex_exit(vp->v_interlock);
		return;
	} else if (vp->v_type != VBLK && vp->v_type != VCHR) {
		atomic_inc_uint(&vp->v_usecount);
		mutex_exit(vp->v_interlock);
		vgone(vp);
		return;
	} else {
		dev = vp->v_rdev;
		type = vp->v_type;
		mutex_exit(vp->v_interlock);
	}

	while (spec_node_lookup_by_dev(type, dev, &vq) == 0) {
		vgone(vq);
	}
}

/*
 * Eliminate all activity associated with a vnode in preparation for
 * reuse.  Drops a reference from the vnode.
 */
void
vgone(vnode_t *vp)
{

	mutex_enter(vp->v_interlock);
	if ((vp->v_iflag & VI_CHANGING) != 0)
		vwait(vp, VI_CHANGING);
	vp->v_iflag |= VI_CHANGING;
	vclean(vp);
	vrelel(vp, VRELEL_CHANGING_SET);
}

/*
 * Update outstanding I/O count and do wakeup if requested.
 */
void
vwakeup(struct buf *bp)
{
	vnode_t *vp;

	if ((vp = bp->b_vp) == NULL)
		return;

	KASSERT(bp->b_objlock == vp->v_interlock);
	KASSERT(mutex_owned(bp->b_objlock));

	if (--vp->v_numoutput < 0)
		vnpanic(vp, "%s: neg numoutput, vp %p", __func__, vp);
	if (vp->v_numoutput == 0)
		cv_broadcast(&vp->v_cv);
}

/*
 * Wait for a vnode (typically with VI_XLOCK set) to be cleaned or
 * recycled.
 */
void
vwait(vnode_t *vp, int flags)
{

	KASSERT(mutex_owned(vp->v_interlock));
	KASSERT(vp->v_usecount != 0);

	while ((vp->v_iflag & flags) != 0)
		cv_wait(&vp->v_cv, vp->v_interlock);
}

int
vfs_drainvnodes(long target)
{
	int error;

	mutex_enter(&vnode_free_list_lock);

	while (numvnodes > target) {
		error = cleanvnode();
		if (error != 0)
			return error;
		mutex_enter(&vnode_free_list_lock);
	}

	mutex_exit(&vnode_free_list_lock);

	return 0;
}

void
vnpanic(vnode_t *vp, const char *fmt, ...)
{
	va_list ap;

#ifdef DIAGNOSTIC
	vprint(NULL, vp);
#endif
	va_start(ap, fmt);
	vpanic(fmt, ap);
	va_end(ap);
}