-
Notifications
You must be signed in to change notification settings - Fork 117
/
vfs_mount.c
1355 lines (1238 loc) · 36.9 KB
/
vfs_mount.c
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
/*
* Copyright (c) 2004 The DragonFly Project. All rights reserved.
*
* This code is derived from software contributed to The DragonFly Project
* by Matthew Dillon <dillon@backplane.com>
*
* 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 DragonFly Project 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 COPYRIGHT HOLDERS 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
* COPYRIGHT HOLDERS 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. All advertising materials mentioning features or use of this software
* must display the following acknowledgement:
* This product includes software developed by the University of
* California, Berkeley and its contributors.
* 4. 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.
*/
/*
* External virtual filesystem routines
*/
#include <sys/param.h>
#include <sys/systm.h>
#include <sys/kernel.h>
#include <sys/malloc.h>
#include <sys/mount.h>
#include <sys/proc.h>
#include <sys/vnode.h>
#include <sys/buf.h>
#include <sys/eventhandler.h>
#include <sys/kthread.h>
#include <sys/sysctl.h>
#include <machine/limits.h>
#include <sys/buf2.h>
#include <sys/thread2.h>
#include <sys/sysref2.h>
#include <vm/vm.h>
#include <vm/vm_object.h>
struct mountscan_info {
TAILQ_ENTRY(mountscan_info) msi_entry;
int msi_how;
struct mount *msi_node;
};
struct vmntvnodescan_info {
TAILQ_ENTRY(vmntvnodescan_info) entry;
struct vnode *vp;
};
struct vnlru_info {
int pass;
};
static int vnlru_nowhere = 0;
SYSCTL_INT(_debug, OID_AUTO, vnlru_nowhere, CTLFLAG_RD,
&vnlru_nowhere, 0,
"Number of times the vnlru process ran without success");
static struct lwkt_token mntid_token;
static struct mount dummymount;
/* note: mountlist exported to pstat */
struct mntlist mountlist = TAILQ_HEAD_INITIALIZER(mountlist);
static TAILQ_HEAD(,mountscan_info) mountscan_list;
static struct lwkt_token mountlist_token;
static TAILQ_HEAD(,vmntvnodescan_info) mntvnodescan_list;
struct lwkt_token mntvnode_token;
static TAILQ_HEAD(,bio_ops) bio_ops_list = TAILQ_HEAD_INITIALIZER(bio_ops_list);
/*
* Called from vfsinit()
*/
void
vfs_mount_init(void)
{
lwkt_token_init(&mountlist_token, "mntlist");
lwkt_token_init(&mntvnode_token, "mntvnode");
lwkt_token_init(&mntid_token, "mntid");
TAILQ_INIT(&mountscan_list);
TAILQ_INIT(&mntvnodescan_list);
mount_init(&dummymount);
dummymount.mnt_flag |= MNT_RDONLY;
dummymount.mnt_kern_flag |= MNTK_ALL_MPSAFE;
}
/*
* Support function called with mntvnode_token held to remove a vnode
* from the mountlist. We must update any list scans which are in progress.
*/
static void
vremovevnodemnt(struct vnode *vp)
{
struct vmntvnodescan_info *info;
TAILQ_FOREACH(info, &mntvnodescan_list, entry) {
if (info->vp == vp)
info->vp = TAILQ_NEXT(vp, v_nmntvnodes);
}
TAILQ_REMOVE(&vp->v_mount->mnt_nvnodelist, vp, v_nmntvnodes);
}
/*
* Allocate a new vnode and associate it with a tag, mount point, and
* operations vector.
*
* A VX locked and refd vnode is returned. The caller should setup the
* remaining fields and vx_put() or, if he wishes to leave a vref,
* vx_unlock() the vnode.
*/
int
getnewvnode(enum vtagtype tag, struct mount *mp,
struct vnode **vpp, int lktimeout, int lkflags)
{
struct vnode *vp;
KKASSERT(mp != NULL);
vp = allocvnode(lktimeout, lkflags);
vp->v_tag = tag;
vp->v_data = NULL;
/*
* By default the vnode is assigned the mount point's normal
* operations vector.
*/
vp->v_ops = &mp->mnt_vn_use_ops;
/*
* Placing the vnode on the mount point's queue makes it visible.
* VNON prevents it from being messed with, however.
*/
insmntque(vp, mp);
/*
* A VX locked & refd vnode is returned.
*/
*vpp = vp;
return (0);
}
/*
* This function creates vnodes with special operations vectors. The
* mount point is optional.
*
* This routine is being phased out but is still used by vfs_conf to
* create vnodes for devices prior to the root mount (with mp == NULL).
*/
int
getspecialvnode(enum vtagtype tag, struct mount *mp,
struct vop_ops **ops,
struct vnode **vpp, int lktimeout, int lkflags)
{
struct vnode *vp;
vp = allocvnode(lktimeout, lkflags);
vp->v_tag = tag;
vp->v_data = NULL;
vp->v_ops = ops;
if (mp == NULL)
mp = &dummymount;
/*
* Placing the vnode on the mount point's queue makes it visible.
* VNON prevents it from being messed with, however.
*/
insmntque(vp, mp);
/*
* A VX locked & refd vnode is returned.
*/
*vpp = vp;
return (0);
}
/*
* Interlock against an unmount, return 0 on success, non-zero on failure.
*
* The passed flag may be 0 or LK_NOWAIT and is only used if an unmount
* is in-progress.
*
* If no unmount is in-progress LK_NOWAIT is ignored. No other flag bits
* are used. A shared locked will be obtained and the filesystem will not
* be unmountable until the lock is released.
*/
int
vfs_busy(struct mount *mp, int flags)
{
int lkflags;
atomic_add_int(&mp->mnt_refs, 1);
if (mp->mnt_kern_flag & MNTK_UNMOUNT) {
if (flags & LK_NOWAIT) {
atomic_add_int(&mp->mnt_refs, -1);
return (ENOENT);
}
/* XXX not MP safe */
mp->mnt_kern_flag |= MNTK_MWAIT;
/*
* Since all busy locks are shared except the exclusive
* lock granted when unmounting, the only place that a
* wakeup needs to be done is at the release of the
* exclusive lock at the end of dounmount.
*/
tsleep((caddr_t)mp, 0, "vfs_busy", 0);
atomic_add_int(&mp->mnt_refs, -1);
return (ENOENT);
}
lkflags = LK_SHARED;
if (lockmgr(&mp->mnt_lock, lkflags))
panic("vfs_busy: unexpected lock failure");
return (0);
}
/*
* Free a busy filesystem.
*
* Decrement refs before releasing the lock so e.g. a pending umount
* doesn't give us an unexpected busy error.
*/
void
vfs_unbusy(struct mount *mp)
{
atomic_add_int(&mp->mnt_refs, -1);
lockmgr(&mp->mnt_lock, LK_RELEASE);
}
/*
* Lookup a filesystem type, and if found allocate and initialize
* a mount structure for it.
*
* Devname is usually updated by mount(8) after booting.
*/
int
vfs_rootmountalloc(char *fstypename, char *devname, struct mount **mpp)
{
struct vfsconf *vfsp;
struct mount *mp;
if (fstypename == NULL)
return (ENODEV);
vfsp = vfsconf_find_by_name(fstypename);
if (vfsp == NULL)
return (ENODEV);
mp = kmalloc(sizeof(struct mount), M_MOUNT, M_WAITOK | M_ZERO);
mount_init(mp);
lockinit(&mp->mnt_lock, "vfslock", VLKTIMEOUT, 0);
vfs_busy(mp, 0);
mp->mnt_vfc = vfsp;
mp->mnt_op = vfsp->vfc_vfsops;
vfsp->vfc_refcount++;
mp->mnt_stat.f_type = vfsp->vfc_typenum;
mp->mnt_flag |= MNT_RDONLY;
mp->mnt_flag |= vfsp->vfc_flags & MNT_VISFLAGMASK;
strncpy(mp->mnt_stat.f_fstypename, vfsp->vfc_name, MFSNAMELEN);
copystr(devname, mp->mnt_stat.f_mntfromname, MNAMELEN - 1, 0);
*mpp = mp;
return (0);
}
/*
* Basic mount structure initialization
*/
void
mount_init(struct mount *mp)
{
lockinit(&mp->mnt_lock, "vfslock", 0, 0);
lwkt_token_init(&mp->mnt_token, "permnt");
TAILQ_INIT(&mp->mnt_nvnodelist);
TAILQ_INIT(&mp->mnt_reservedvnlist);
TAILQ_INIT(&mp->mnt_jlist);
mp->mnt_nvnodelistsize = 0;
mp->mnt_flag = 0;
mp->mnt_iosize_max = MAXPHYS;
}
/*
* Lookup a mount point by filesystem identifier.
*/
struct mount *
vfs_getvfs(fsid_t *fsid)
{
struct mount *mp;
lwkt_gettoken(&mountlist_token);
TAILQ_FOREACH(mp, &mountlist, mnt_list) {
if (mp->mnt_stat.f_fsid.val[0] == fsid->val[0] &&
mp->mnt_stat.f_fsid.val[1] == fsid->val[1]) {
break;
}
}
lwkt_reltoken(&mountlist_token);
return (mp);
}
/*
* Get a new unique fsid. Try to make its val[0] unique, since this value
* will be used to create fake device numbers for stat(). Also try (but
* not so hard) make its val[0] unique mod 2^16, since some emulators only
* support 16-bit device numbers. We end up with unique val[0]'s for the
* first 2^16 calls and unique val[0]'s mod 2^16 for the first 2^8 calls.
*
* Keep in mind that several mounts may be running in parallel. Starting
* the search one past where the previous search terminated is both a
* micro-optimization and a defense against returning the same fsid to
* different mounts.
*/
void
vfs_getnewfsid(struct mount *mp)
{
static u_int16_t mntid_base;
fsid_t tfsid;
int mtype;
lwkt_gettoken(&mntid_token);
mtype = mp->mnt_vfc->vfc_typenum;
tfsid.val[1] = mtype;
mtype = (mtype & 0xFF) << 24;
for (;;) {
tfsid.val[0] = makeudev(255,
mtype | ((mntid_base & 0xFF00) << 8) | (mntid_base & 0xFF));
mntid_base++;
if (vfs_getvfs(&tfsid) == NULL)
break;
}
mp->mnt_stat.f_fsid.val[0] = tfsid.val[0];
mp->mnt_stat.f_fsid.val[1] = tfsid.val[1];
lwkt_reltoken(&mntid_token);
}
/*
* Set the FSID for a new mount point to the template. Adjust
* the FSID to avoid collisions.
*/
int
vfs_setfsid(struct mount *mp, fsid_t *template)
{
int didmunge = 0;
bzero(&mp->mnt_stat.f_fsid, sizeof(mp->mnt_stat.f_fsid));
for (;;) {
if (vfs_getvfs(template) == NULL)
break;
didmunge = 1;
++template->val[1];
}
mp->mnt_stat.f_fsid = *template;
return(didmunge);
}
/*
* This routine is called when we have too many vnodes. It attempts
* to free <count> vnodes and will potentially free vnodes that still
* have VM backing store (VM backing store is typically the cause
* of a vnode blowout so we want to do this). Therefore, this operation
* is not considered cheap.
*
* A number of conditions may prevent a vnode from being reclaimed.
* the buffer cache may have references on the vnode, a directory
* vnode may still have references due to the namei cache representing
* underlying files, or the vnode may be in active use. It is not
* desireable to reuse such vnodes. These conditions may cause the
* number of vnodes to reach some minimum value regardless of what
* you set kern.maxvnodes to. Do not set kern.maxvnodes too low.
*/
/*
* This is a quick non-blocking check to determine if the vnode is a good
* candidate for being (eventually) vgone()'d. Returns 0 if the vnode is
* not a good candidate, 1 if it is.
*/
static __inline int
vmightfree(struct vnode *vp, int page_count, int pass)
{
if (vp->v_flag & VRECLAIMED)
return (0);
#if 0
if ((vp->v_flag & VFREE) && TAILQ_EMPTY(&vp->v_namecache))
return (0);
#endif
if (sysref_isactive(&vp->v_sysref))
return (0);
if (vp->v_object && vp->v_object->resident_page_count >= page_count)
return (0);
/*
* XXX horrible hack. Up to four passes will be taken. Each pass
* makes a larger set of vnodes eligible. For now what this really
* means is that we try to recycle files opened only once before
* recycling files opened multiple times.
*/
switch(vp->v_flag & (VAGE0 | VAGE1)) {
case 0:
if (pass < 3)
return(0);
break;
case VAGE0:
if (pass < 2)
return(0);
break;
case VAGE1:
if (pass < 1)
return(0);
break;
case VAGE0 | VAGE1:
break;
}
return (1);
}
/*
* The vnode was found to be possibly vgone()able and the caller has locked it
* (thus the usecount should be 1 now). Determine if the vnode is actually
* vgone()able, doing some cleanups in the process. Returns 1 if the vnode
* can be vgone()'d, 0 otherwise.
*
* Note that v_auxrefs may be non-zero because (A) this vnode is not a leaf
* in the namecache topology and (B) this vnode has buffer cache bufs.
* We cannot remove vnodes with non-leaf namecache associations. We do a
* tentitive leaf check prior to attempting to flush out any buffers but the
* 'real' test when all is said in done is that v_auxrefs must become 0 for
* the vnode to be freeable.
*
* We could theoretically just unconditionally flush when v_auxrefs != 0,
* but flushing data associated with non-leaf nodes (which are always
* directories), just throws it away for no benefit. It is the buffer
* cache's responsibility to choose buffers to recycle from the cached
* data point of view.
*/
static int
visleaf(struct vnode *vp)
{
struct namecache *ncp;
spin_lock(&vp->v_spin);
TAILQ_FOREACH(ncp, &vp->v_namecache, nc_vnode) {
if (!TAILQ_EMPTY(&ncp->nc_list)) {
spin_unlock(&vp->v_spin);
return(0);
}
}
spin_unlock(&vp->v_spin);
return(1);
}
/*
* Try to clean up the vnode to the point where it can be vgone()'d, returning
* 0 if it cannot be vgone()'d (or already has been), 1 if it can. Unlike
* vmightfree() this routine may flush the vnode and block. Vnodes marked
* VFREE are still candidates for vgone()ing because they may hold namecache
* resources and could be blocking the namecache directory hierarchy (and
* related vnodes) from being freed.
*/
static int
vtrytomakegoneable(struct vnode *vp, int page_count)
{
if (vp->v_flag & VRECLAIMED)
return (0);
if (vp->v_sysref.refcnt > 1)
return (0);
if (vp->v_object && vp->v_object->resident_page_count >= page_count)
return (0);
if (vp->v_auxrefs && visleaf(vp)) {
vinvalbuf(vp, V_SAVE, 0, 0);
#if 0 /* DEBUG */
kprintf((vp->v_auxrefs ? "vrecycle: vp %p failed: %s\n" :
"vrecycle: vp %p succeeded: %s\n"), vp,
(TAILQ_FIRST(&vp->v_namecache) ?
TAILQ_FIRST(&vp->v_namecache)->nc_name : "?"));
#endif
}
/*
* This sequence may seem a little strange, but we need to optimize
* the critical path a bit. We can't recycle vnodes with other
* references and because we are trying to recycle an otherwise
* perfectly fine vnode we have to invalidate the namecache in a
* way that avoids possible deadlocks (since the vnode lock is being
* held here). Finally, we have to check for other references one
* last time in case something snuck in during the inval.
*/
if (vp->v_sysref.refcnt > 1 || vp->v_auxrefs != 0)
return (0);
if (cache_inval_vp_nonblock(vp))
return (0);
return (vp->v_sysref.refcnt <= 1 && vp->v_auxrefs == 0);
}
/*
* Reclaim up to 1/10 of the vnodes associated with a mount point. Try
* to avoid vnodes which have lots of resident pages (we are trying to free
* vnodes, not memory).
*
* This routine is a callback from the mountlist scan. The mount point
* in question will be busied.
*
* NOTE: The 1/10 reclamation also ensures that the inactive data set
* (the vnodes being recycled by the one-time use) does not degenerate
* into too-small a set. This is important because once a vnode is
* marked as not being one-time-use (VAGE0/VAGE1 both 0) that vnode
* will not be destroyed EXCEPT by this mechanism. VM pages can still
* be cleaned/freed by the pageout daemon.
*/
static int
vlrureclaim(struct mount *mp, void *data)
{
struct vnlru_info *info = data;
struct vnode *vp;
int done;
int trigger;
int usevnodes;
int count;
int trigger_mult = vnlru_nowhere;
/*
* Calculate the trigger point for the resident pages check. The
* minimum trigger value is approximately the number of pages in
* the system divded by the number of vnodes. However, due to
* various other system memory overheads unrelated to data caching
* it is a good idea to double the trigger (at least).
*
* trigger_mult starts at 0. If the recycler is having problems
* finding enough freeable vnodes it will increase trigger_mult.
* This should not happen in normal operation, even on machines with
* low amounts of memory, but extraordinary memory use by the system
* verses the amount of cached data can trigger it.
*/
usevnodes = desiredvnodes;
if (usevnodes <= 0)
usevnodes = 1;
trigger = vmstats.v_page_count * (trigger_mult + 2) / usevnodes;
done = 0;
lwkt_gettoken(&mntvnode_token);
count = mp->mnt_nvnodelistsize / 10 + 1;
while (count && mp->mnt_syncer) {
/*
* Next vnode. Use the special syncer vnode to placemark
* the LRU. This way the LRU code does not interfere with
* vmntvnodescan().
*/
vp = TAILQ_NEXT(mp->mnt_syncer, v_nmntvnodes);
TAILQ_REMOVE(&mp->mnt_nvnodelist, mp->mnt_syncer, v_nmntvnodes);
if (vp) {
TAILQ_INSERT_AFTER(&mp->mnt_nvnodelist, vp,
mp->mnt_syncer, v_nmntvnodes);
} else {
TAILQ_INSERT_HEAD(&mp->mnt_nvnodelist, mp->mnt_syncer,
v_nmntvnodes);
vp = TAILQ_NEXT(mp->mnt_syncer, v_nmntvnodes);
if (vp == NULL)
break;
}
/*
* __VNODESCAN__
*
* The VP will stick around while we hold mntvnode_token,
* at least until we block, so we can safely do an initial
* check, and then must check again after we lock the vnode.
*/
if (vp->v_type == VNON || /* syncer or indeterminant */
!vmightfree(vp, trigger, info->pass) /* critical path opt */
) {
--count;
continue;
}
/*
* VX get the candidate vnode. If the VX get fails the
* vnode might still be on the mountlist. Our loop depends
* on us at least cycling the vnode to the end of the
* mountlist.
*/
if (vx_get_nonblock(vp) != 0) {
--count;
continue;
}
/*
* Since we blocked locking the vp, make sure it is still
* a candidate for reclamation. That is, it has not already
* been reclaimed and only has our VX reference associated
* with it.
*/
if (vp->v_type == VNON || /* syncer or indeterminant */
(vp->v_flag & VRECLAIMED) ||
vp->v_mount != mp ||
!vtrytomakegoneable(vp, trigger) /* critical path opt */
) {
--count;
vx_put(vp);
continue;
}
/*
* All right, we are good, move the vp to the end of the
* mountlist and clean it out. The vget will have returned
* an error if the vnode was destroyed (VRECLAIMED set), so we
* do not have to check again. The vput() will move the
* vnode to the free list if the vgone() was successful.
*/
KKASSERT(vp->v_mount == mp);
vgone_vxlocked(vp);
vx_put(vp);
++done;
--count;
}
lwkt_reltoken(&mntvnode_token);
return (done);
}
/*
* Attempt to recycle vnodes in a context that is always safe to block.
* Calling vlrurecycle() from the bowels of file system code has some
* interesting deadlock problems.
*/
static struct thread *vnlruthread;
static int vnlruproc_sig;
void
vnlru_proc_wait(void)
{
tsleep_interlock(&vnlruproc_sig, 0);
if (vnlruproc_sig == 0) {
vnlruproc_sig = 1; /* avoid unnecessary wakeups */
wakeup(vnlruthread);
}
tsleep(&vnlruproc_sig, PINTERLOCKED, "vlruwk", hz);
}
static void
vnlru_proc(void)
{
struct thread *td = curthread;
struct vnlru_info info;
int done;
EVENTHANDLER_REGISTER(shutdown_pre_sync, shutdown_kproc, td,
SHUTDOWN_PRI_FIRST);
for (;;) {
kproc_suspend_loop();
/*
* Try to free some vnodes if we have too many
*/
if (numvnodes > desiredvnodes &&
freevnodes > desiredvnodes * 2 / 10) {
int count = numvnodes - desiredvnodes;
if (count > freevnodes / 100)
count = freevnodes / 100;
if (count < 5)
count = 5;
freesomevnodes(count);
}
/*
* Nothing to do if most of our vnodes are already on
* the free list.
*/
if (numvnodes - freevnodes <= desiredvnodes * 9 / 10) {
vnlruproc_sig = 0;
wakeup(&vnlruproc_sig);
tsleep(vnlruthread, 0, "vlruwt", hz);
continue;
}
cache_hysteresis();
/*
* The pass iterates through the four combinations of
* VAGE0/VAGE1. We want to get rid of aged small files
* first.
*/
info.pass = 0;
done = 0;
while (done == 0 && info.pass < 4) {
done = mountlist_scan(vlrureclaim, &info,
MNTSCAN_FORWARD);
++info.pass;
}
/*
* The vlrureclaim() call only processes 1/10 of the vnodes
* on each mount. If we couldn't find any repeat the loop
* at least enough times to cover all available vnodes before
* we start sleeping. Complain if the failure extends past
* 30 second, every 30 seconds.
*/
if (done == 0) {
++vnlru_nowhere;
if (vnlru_nowhere % 10 == 0)
tsleep(vnlruthread, 0, "vlrup", hz * 3);
if (vnlru_nowhere % 100 == 0)
kprintf("vnlru_proc: vnode recycler stopped working!\n");
if (vnlru_nowhere == 1000)
vnlru_nowhere = 900;
} else {
vnlru_nowhere = 0;
}
}
}
/*
* MOUNTLIST FUNCTIONS
*/
/*
* mountlist_insert (MP SAFE)
*
* Add a new mount point to the mount list.
*/
void
mountlist_insert(struct mount *mp, int how)
{
lwkt_gettoken(&mountlist_token);
if (how == MNTINS_FIRST)
TAILQ_INSERT_HEAD(&mountlist, mp, mnt_list);
else
TAILQ_INSERT_TAIL(&mountlist, mp, mnt_list);
lwkt_reltoken(&mountlist_token);
}
/*
* mountlist_interlock (MP SAFE)
*
* Execute the specified interlock function with the mountlist token
* held. The function will be called in a serialized fashion verses
* other functions called through this mechanism.
*/
int
mountlist_interlock(int (*callback)(struct mount *), struct mount *mp)
{
int error;
lwkt_gettoken(&mountlist_token);
error = callback(mp);
lwkt_reltoken(&mountlist_token);
return (error);
}
/*
* mountlist_boot_getfirst (DURING BOOT ONLY)
*
* This function returns the first mount on the mountlist, which is
* expected to be the root mount. Since no interlocks are obtained
* this function is only safe to use during booting.
*/
struct mount *
mountlist_boot_getfirst(void)
{
return(TAILQ_FIRST(&mountlist));
}
/*
* mountlist_remove (MP SAFE)
*
* Remove a node from the mountlist. If this node is the next scan node
* for any active mountlist scans, the active mountlist scan will be
* adjusted to skip the node, thus allowing removals during mountlist
* scans.
*/
void
mountlist_remove(struct mount *mp)
{
struct mountscan_info *msi;
lwkt_gettoken(&mountlist_token);
TAILQ_FOREACH(msi, &mountscan_list, msi_entry) {
if (msi->msi_node == mp) {
if (msi->msi_how & MNTSCAN_FORWARD)
msi->msi_node = TAILQ_NEXT(mp, mnt_list);
else
msi->msi_node = TAILQ_PREV(mp, mntlist, mnt_list);
}
}
TAILQ_REMOVE(&mountlist, mp, mnt_list);
lwkt_reltoken(&mountlist_token);
}
/*
* mountlist_exists (MP SAFE)
*
* Checks if a node exists in the mountlist.
* This function is mainly used by VFS quota code to check if a
* cached nullfs struct mount pointer is still valid at use time
*
* FIXME: there is no warranty the mp passed to that function
* will be the same one used by VFS_ACCOUNT() later
*/
int
mountlist_exists(struct mount *mp)
{
int node_exists = 0;
struct mount* lmp;
lwkt_gettoken(&mountlist_token);
TAILQ_FOREACH(lmp, &mountlist, mnt_list) {
if (lmp == mp) {
node_exists = 1;
break;
}
}
lwkt_reltoken(&mountlist_token);
return(node_exists);
}
/*
* mountlist_scan (MP SAFE)
*
* Safely scan the mount points on the mount list. Unless otherwise
* specified each mount point will be busied prior to the callback and
* unbusied afterwords. The callback may safely remove any mount point
* without interfering with the scan. If the current callback
* mount is removed the scanner will not attempt to unbusy it.
*
* If a mount node cannot be busied it is silently skipped.
*
* The callback return value is aggregated and a total is returned. A return
* value of < 0 is not aggregated and will terminate the scan.
*
* MNTSCAN_FORWARD - the mountlist is scanned in the forward direction
* MNTSCAN_REVERSE - the mountlist is scanned in reverse
* MNTSCAN_NOBUSY - the scanner will make the callback without busying
* the mount node.
*/
int
mountlist_scan(int (*callback)(struct mount *, void *), void *data, int how)
{
struct mountscan_info info;
struct mount *mp;
int count;
int res;
lwkt_gettoken(&mountlist_token);
info.msi_how = how;
info.msi_node = NULL; /* paranoia */
TAILQ_INSERT_TAIL(&mountscan_list, &info, msi_entry);
res = 0;
if (how & MNTSCAN_FORWARD) {
info.msi_node = TAILQ_FIRST(&mountlist);
while ((mp = info.msi_node) != NULL) {
if (how & MNTSCAN_NOBUSY) {
count = callback(mp, data);
} else if (vfs_busy(mp, LK_NOWAIT) == 0) {
count = callback(mp, data);
if (mp == info.msi_node)
vfs_unbusy(mp);
} else {
count = 0;
}
if (count < 0)
break;
res += count;
if (mp == info.msi_node)
info.msi_node = TAILQ_NEXT(mp, mnt_list);
}
} else if (how & MNTSCAN_REVERSE) {
info.msi_node = TAILQ_LAST(&mountlist, mntlist);
while ((mp = info.msi_node) != NULL) {
if (how & MNTSCAN_NOBUSY) {
count = callback(mp, data);
} else if (vfs_busy(mp, LK_NOWAIT) == 0) {
count = callback(mp, data);
if (mp == info.msi_node)
vfs_unbusy(mp);
} else {
count = 0;
}
if (count < 0)
break;
res += count;
if (mp == info.msi_node)
info.msi_node = TAILQ_PREV(mp, mntlist, mnt_list);
}
}
TAILQ_REMOVE(&mountscan_list, &info, msi_entry);
lwkt_reltoken(&mountlist_token);
return(res);
}
/*
* MOUNT RELATED VNODE FUNCTIONS
*/
static struct kproc_desc vnlru_kp = {
"vnlru",
vnlru_proc,
&vnlruthread
};
SYSINIT(vnlru, SI_SUB_KTHREAD_UPDATE, SI_ORDER_FIRST, kproc_start, &vnlru_kp)
/*
* Move a vnode from one mount queue to another.
*
* MPSAFE
*/
void
insmntque(struct vnode *vp, struct mount *mp)
{
lwkt_gettoken(&mntvnode_token);
/*
* Delete from old mount point vnode list, if on one.
*/
if (vp->v_mount != NULL) {
KASSERT(vp->v_mount->mnt_nvnodelistsize > 0,
("bad mount point vnode list size"));
vremovevnodemnt(vp);
vp->v_mount->mnt_nvnodelistsize--;
}
/*
* Insert into list of vnodes for the new mount point, if available.
* The 'end' of the LRU list is the vnode prior to mp->mnt_syncer.
*/
if ((vp->v_mount = mp) == NULL) {
lwkt_reltoken(&mntvnode_token);
return;
}
if (mp->mnt_syncer) {
TAILQ_INSERT_BEFORE(mp->mnt_syncer, vp, v_nmntvnodes);
} else {
TAILQ_INSERT_TAIL(&mp->mnt_nvnodelist, vp, v_nmntvnodes);
}
mp->mnt_nvnodelistsize++;
lwkt_reltoken(&mntvnode_token);
}
/*
* Scan the vnodes under a mount point and issue appropriate callbacks.
*
* The fastfunc() callback is called with just the mountlist token held