2 * SPDX-License-Identifier: BSD-3-Clause
4 * Copyright (c) 1989, 1991, 1993
5 * The Regents of the University of California. All rights reserved.
7 * Redistribution and use in source and binary forms, with or without
8 * modification, are permitted provided that the following conditions
10 * 1. Redistributions of source code must retain the above copyright
11 * notice, this list of conditions and the following disclaimer.
12 * 2. Redistributions in binary form must reproduce the above copyright
13 * notice, this list of conditions and the following disclaimer in the
14 * documentation and/or other materials provided with the distribution.
15 * 3. Neither the name of the University nor the names of its contributors
16 * may be used to endorse or promote products derived from this software
17 * without specific prior written permission.
19 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
20 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
21 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
22 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
23 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
24 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
25 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
26 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
27 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
28 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
31 * @(#)mount.h 8.21 (Berkeley) 5/20/95
38 #include <sys/ucred.h>
39 #include <sys/queue.h>
41 #include <sys/types.h>
43 #include <sys/lockmgr.h>
44 #include <sys/tslog.h>
45 #include <sys/_mutex.h>
50 * NOTE: When changing statfs structure, mount structure, MNT_* flags or
51 * MNTK_* flags also update DDB show mount command in vfs_subr.c.
54 typedef struct fsid { int32_t val[2]; } fsid_t; /* filesystem id type */
56 #define fsidcmp(a, b) memcmp((a), (b), sizeof(fsid_t))
60 * These are unique per filesystem on a single machine.
62 * Note that the offset of fid_data is 4 bytes, so care must be taken to avoid
63 * undefined behavior accessing unaligned fields within an embedded struct.
68 u_short fid_len; /* length of data in bytes */
69 u_short fid_data0; /* force longword alignment */
70 char fid_data[MAXFIDSZ]; /* data (variable length) */
74 * filesystem statistics
76 #define MFSNAMELEN 16 /* length of type name including null */
77 #define MNAMELEN 1024 /* size of on/from name bufs */
78 #define STATFS_VERSION 0x20140518 /* current version number */
80 uint32_t f_version; /* structure version number */
81 uint32_t f_type; /* type of filesystem */
82 uint64_t f_flags; /* copy of mount exported flags */
83 uint64_t f_bsize; /* filesystem fragment size */
84 uint64_t f_iosize; /* optimal transfer block size */
85 uint64_t f_blocks; /* total data blocks in filesystem */
86 uint64_t f_bfree; /* free blocks in filesystem */
87 int64_t f_bavail; /* free blocks avail to non-superuser */
88 uint64_t f_files; /* total file nodes in filesystem */
89 int64_t f_ffree; /* free nodes avail to non-superuser */
90 uint64_t f_syncwrites; /* count of sync writes since mount */
91 uint64_t f_asyncwrites; /* count of async writes since mount */
92 uint64_t f_syncreads; /* count of sync reads since mount */
93 uint64_t f_asyncreads; /* count of async reads since mount */
94 uint64_t f_spare[10]; /* unused spare */
95 uint32_t f_namemax; /* maximum filename length */
96 uid_t f_owner; /* user that mounted the filesystem */
97 fsid_t f_fsid; /* filesystem id */
98 char f_charspare[80]; /* spare string space */
99 char f_fstypename[MFSNAMELEN]; /* filesystem type name */
100 char f_mntfromname[MNAMELEN]; /* mounted filesystem */
101 char f_mntonname[MNAMELEN]; /* directory on which mounted */
104 #if defined(_WANT_FREEBSD11_STATFS) || defined(_KERNEL)
105 #define FREEBSD11_STATFS_VERSION 0x20030518 /* current version number */
106 struct freebsd11_statfs {
107 uint32_t f_version; /* structure version number */
108 uint32_t f_type; /* type of filesystem */
109 uint64_t f_flags; /* copy of mount exported flags */
110 uint64_t f_bsize; /* filesystem fragment size */
111 uint64_t f_iosize; /* optimal transfer block size */
112 uint64_t f_blocks; /* total data blocks in filesystem */
113 uint64_t f_bfree; /* free blocks in filesystem */
114 int64_t f_bavail; /* free blocks avail to non-superuser */
115 uint64_t f_files; /* total file nodes in filesystem */
116 int64_t f_ffree; /* free nodes avail to non-superuser */
117 uint64_t f_syncwrites; /* count of sync writes since mount */
118 uint64_t f_asyncwrites; /* count of async writes since mount */
119 uint64_t f_syncreads; /* count of sync reads since mount */
120 uint64_t f_asyncreads; /* count of async reads since mount */
121 uint64_t f_spare[10]; /* unused spare */
122 uint32_t f_namemax; /* maximum filename length */
123 uid_t f_owner; /* user that mounted the filesystem */
124 fsid_t f_fsid; /* filesystem id */
125 char f_charspare[80]; /* spare string space */
126 char f_fstypename[16]; /* filesystem type name */
127 char f_mntfromname[88]; /* mounted filesystem */
128 char f_mntonname[88]; /* directory on which mounted */
130 #endif /* _WANT_FREEBSD11_STATFS || _KERNEL */
133 #define OMFSNAMELEN 16 /* length of fs type name, including null */
134 #define OMNAMELEN (88 - 2 * sizeof(long)) /* size of on/from name bufs */
136 /* XXX getfsstat.2 is out of date with write and read counter changes here. */
137 /* XXX statfs.2 is out of date with read counter changes here. */
139 long f_spare2; /* placeholder */
140 long f_bsize; /* fundamental filesystem block size */
141 long f_iosize; /* optimal transfer block size */
142 long f_blocks; /* total data blocks in filesystem */
143 long f_bfree; /* free blocks in fs */
144 long f_bavail; /* free blocks avail to non-superuser */
145 long f_files; /* total file nodes in filesystem */
146 long f_ffree; /* free file nodes in fs */
147 fsid_t f_fsid; /* filesystem id */
148 uid_t f_owner; /* user that mounted the filesystem */
149 int f_type; /* type of filesystem */
150 int f_flags; /* copy of mount exported flags */
151 long f_syncwrites; /* count of sync writes since mount */
152 long f_asyncwrites; /* count of async writes since mount */
153 char f_fstypename[OMFSNAMELEN]; /* fs type name */
154 char f_mntonname[OMNAMELEN]; /* directory on which mounted */
155 long f_syncreads; /* count of sync reads since mount */
156 long f_asyncreads; /* count of async reads since mount */
157 short f_spares1; /* unused spare */
158 char f_mntfromname[OMNAMELEN];/* mounted filesystem */
159 short f_spares2; /* unused spare */
161 * XXX on machines where longs are aligned to 8-byte boundaries, there
162 * is an unnamed int32_t here. This spare was after the apparent end
163 * of the struct until we bit off the read counters from f_mntonname.
165 long f_spare[2]; /* unused spare */
169 #if defined(_WANT_MOUNT) || defined(_KERNEL)
170 TAILQ_HEAD(vnodelst, vnode);
172 /* Mount options list */
173 TAILQ_HEAD(vfsoptlist, vfsopt);
175 TAILQ_ENTRY(vfsopt) link;
184 int mntp_thread_in_ops;
187 int mntp_writeopcount;
190 _Static_assert(sizeof(struct mount_pcpu) == 16,
191 "the struct is allocated from pcpu 16 zone");
194 * Structure for tracking a stacked filesystem mounted above another
195 * filesystem. This is expected to be stored in the upper FS' per-mount data.
198 * i - lower mount interlock
199 * c - constant from node initialization
201 struct mount_upper_node {
202 struct mount *mp; /* (c) mount object for upper FS */
203 TAILQ_ENTRY(mount_upper_node) mnt_upper_link; /* (i) position in uppers list */
207 * Structure per mounted filesystem. Each mounted filesystem has an
208 * array of operations and an instance record. The filesystems are
209 * put on a doubly linked list.
215 * v - vnode freelist mutex
216 * d - deferred unmount list mutex
218 * Unmarked fields are considered stable as long as a ref is held.
222 int mnt_vfs_ops; /* (i) pending vfs ops */
223 int mnt_kern_flag; /* (i) kernel only flags */
224 uint64_t mnt_flag; /* (i) flags shared with user */
225 struct mount_pcpu *mnt_pcpu; /* per-CPU data */
226 struct vnode *mnt_rootvnode;
227 struct vnode *mnt_vnodecovered; /* vnode we mounted on */
228 struct vfsops *mnt_op; /* operations on fs */
229 struct vfsconf *mnt_vfc; /* configuration info */
230 struct mtx __aligned(CACHE_LINE_SIZE) mnt_mtx; /* mount structure interlock */
231 int mnt_gen; /* struct mount generation */
232 #define mnt_startzero mnt_list
233 TAILQ_ENTRY(mount) mnt_list; /* (m) mount list */
234 struct vnode *mnt_syncer; /* syncer vnode */
235 int mnt_ref; /* (i) Reference count */
236 struct vnodelst mnt_nvnodelist; /* (i) list of vnodes */
237 int mnt_nvnodelistsize; /* (i) # of vnodes */
238 int mnt_writeopcount; /* (i) write syscalls pending */
239 struct vfsoptlist *mnt_opt; /* current mount options */
240 struct vfsoptlist *mnt_optnew; /* new options passed to fs */
241 struct statfs mnt_stat; /* cache of filesystem stats */
242 struct ucred *mnt_cred; /* credentials of mounter */
243 void * mnt_data; /* private data */
244 time_t mnt_time; /* last time written*/
245 int mnt_iosize_max; /* max size for clusters, etc */
246 struct netexport *mnt_export; /* export list */
247 struct label *mnt_label; /* MAC label for the fs */
248 u_int mnt_hashseed; /* Random seed for vfs_hash */
249 int mnt_lockref; /* (i) Lock reference count */
250 int mnt_secondary_writes; /* (i) # of secondary writes */
251 int mnt_secondary_accwrites;/* (i) secondary wr. starts */
252 struct thread *mnt_susp_owner; /* (i) thread owning suspension */
253 #define mnt_endzero mnt_gjprovider
254 char *mnt_gjprovider; /* gjournal provider name */
255 struct mtx mnt_listmtx;
256 struct vnodelst mnt_lazyvnodelist; /* (l) list of lazy vnodes */
257 int mnt_lazyvnodelistsize; /* (l) # of lazy vnodes */
258 int mnt_upper_pending; /* (i) # of pending ops on mnt_uppers */
259 struct lock mnt_explock; /* vfs_export walkers lock */
260 TAILQ_HEAD(, mount_upper_node) mnt_uppers; /* (i) upper mounts over us */
261 TAILQ_HEAD(, mount_upper_node) mnt_notify; /* (i) upper mounts for notification */
262 STAILQ_ENTRY(mount) mnt_taskqueue_link; /* (d) our place in deferred unmount list */
263 uint64_t mnt_taskqueue_flags; /* (d) unmount flags passed from taskqueue */
264 unsigned int mnt_unmount_retries; /* (d) # of failed deferred unmount attempts */
266 #endif /* _WANT_MOUNT || _KERNEL */
270 * Definitions for MNT_VNODE_FOREACH_ALL.
272 struct vnode *__mnt_vnode_next_all(struct vnode **mvp, struct mount *mp);
273 struct vnode *__mnt_vnode_first_all(struct vnode **mvp, struct mount *mp);
274 void __mnt_vnode_markerfree_all(struct vnode **mvp, struct mount *mp);
276 #define MNT_VNODE_FOREACH_ALL(vp, mp, mvp) \
277 for (vp = __mnt_vnode_first_all(&(mvp), (mp)); \
278 (vp) != NULL; vp = __mnt_vnode_next_all(&(mvp), (mp)))
280 #define MNT_VNODE_FOREACH_ALL_ABORT(mp, mvp) \
283 __mnt_vnode_markerfree_all(&(mvp), (mp)); \
284 /* MNT_IUNLOCK(mp); -- done in above function */ \
285 mtx_assert(MNT_MTX(mp), MA_NOTOWNED); \
289 * Definitions for MNT_VNODE_FOREACH_LAZY.
291 typedef int mnt_lazy_cb_t(struct vnode *, void *);
292 struct vnode *__mnt_vnode_next_lazy(struct vnode **mvp, struct mount *mp,
293 mnt_lazy_cb_t *cb, void *cbarg);
294 struct vnode *__mnt_vnode_first_lazy(struct vnode **mvp, struct mount *mp,
295 mnt_lazy_cb_t *cb, void *cbarg);
296 void __mnt_vnode_markerfree_lazy(struct vnode **mvp, struct mount *mp);
298 #define MNT_VNODE_FOREACH_LAZY(vp, mp, mvp, cb, cbarg) \
299 for (vp = __mnt_vnode_first_lazy(&(mvp), (mp), (cb), (cbarg)); \
301 vp = __mnt_vnode_next_lazy(&(mvp), (mp), (cb), (cbarg)))
303 #define MNT_VNODE_FOREACH_LAZY_ABORT(mp, mvp) \
304 __mnt_vnode_markerfree_lazy(&(mvp), (mp))
306 #define MNT_ILOCK(mp) mtx_lock(&(mp)->mnt_mtx)
307 #define MNT_ITRYLOCK(mp) mtx_trylock(&(mp)->mnt_mtx)
308 #define MNT_IUNLOCK(mp) mtx_unlock(&(mp)->mnt_mtx)
309 #define MNT_MTX(mp) (&(mp)->mnt_mtx)
311 #define MNT_REF(mp) do { \
312 mtx_assert(MNT_MTX(mp), MA_OWNED); \
315 #define MNT_REL(mp) do { \
316 mtx_assert(MNT_MTX(mp), MA_OWNED); \
318 if ((mp)->mnt_vfs_ops && (mp)->mnt_ref < 0) \
319 vfs_dump_mount_counters(mp); \
320 if ((mp)->mnt_ref == 0 && (mp)->mnt_vfs_ops) \
326 #if defined(_WANT_MNTOPTNAMES) || defined(_KERNEL)
331 #define MNTOPT_NAMES \
332 { MNT_ASYNC, "asynchronous" }, \
333 { MNT_EXPORTED, "NFS exported" }, \
334 { MNT_LOCAL, "local" }, \
335 { MNT_NOATIME, "noatime" }, \
336 { MNT_NOEXEC, "noexec" }, \
337 { MNT_NOSUID, "nosuid" }, \
338 { MNT_NOSYMFOLLOW, "nosymfollow" }, \
339 { MNT_QUOTA, "with quotas" }, \
340 { MNT_RDONLY, "read-only" }, \
341 { MNT_SYNCHRONOUS, "synchronous" }, \
342 { MNT_UNION, "union" }, \
343 { MNT_NOCLUSTERR, "noclusterr" }, \
344 { MNT_NOCLUSTERW, "noclusterw" }, \
345 { MNT_SUIDDIR, "suiddir" }, \
346 { MNT_SOFTDEP, "soft-updates" }, \
347 { MNT_SUJ, "journaled soft-updates" }, \
348 { MNT_MULTILABEL, "multilabel" }, \
349 { MNT_ACLS, "acls" }, \
350 { MNT_NFS4ACLS, "nfsv4acls" }, \
351 { MNT_GJOURNAL, "gjournal" }, \
352 { MNT_AUTOMOUNTED, "automounted" }, \
353 { MNT_VERIFIED, "verified" }, \
354 { MNT_UNTRUSTED, "untrusted" }, \
355 { MNT_NOCOVER, "nocover" }, \
356 { MNT_EMPTYDIR, "emptydir" }, \
357 { MNT_UPDATE, "update" }, \
358 { MNT_DELEXPORT, "delexport" }, \
359 { MNT_RELOAD, "reload" }, \
360 { MNT_FORCE, "force" }, \
361 { MNT_SNAPSHOT, "snapshot" }, \
366 * User specifiable flags, stored in mnt_flag.
368 #define MNT_RDONLY 0x0000000000000001ULL /* read only filesystem */
369 #define MNT_SYNCHRONOUS 0x0000000000000002ULL /* fs written synchronously */
370 #define MNT_NOEXEC 0x0000000000000004ULL /* can't exec from filesystem */
371 #define MNT_NOSUID 0x0000000000000008ULL /* don't honor setuid fs bits */
372 #define MNT_NFS4ACLS 0x0000000000000010ULL /* enable NFS version 4 ACLs */
373 #define MNT_UNION 0x0000000000000020ULL /* union with underlying fs */
374 #define MNT_ASYNC 0x0000000000000040ULL /* fs written asynchronously */
375 #define MNT_SUIDDIR 0x0000000000100000ULL /* special SUID dir handling */
376 #define MNT_SOFTDEP 0x0000000000200000ULL /* using soft updates */
377 #define MNT_NOSYMFOLLOW 0x0000000000400000ULL /* do not follow symlinks */
378 #define MNT_GJOURNAL 0x0000000002000000ULL /* GEOM journal support enabled */
379 #define MNT_MULTILABEL 0x0000000004000000ULL /* MAC support for objects */
380 #define MNT_ACLS 0x0000000008000000ULL /* ACL support enabled */
381 #define MNT_NOATIME 0x0000000010000000ULL /* dont update file access time */
382 #define MNT_NOCLUSTERR 0x0000000040000000ULL /* disable cluster read */
383 #define MNT_NOCLUSTERW 0x0000000080000000ULL /* disable cluster write */
384 #define MNT_SUJ 0x0000000100000000ULL /* using journaled soft updates */
385 #define MNT_AUTOMOUNTED 0x0000000200000000ULL /* mounted by automountd(8) */
386 #define MNT_UNTRUSTED 0x0000000800000000ULL /* filesys metadata untrusted */
389 * NFS export related mount flags.
391 #define MNT_EXRDONLY 0x0000000000000080ULL /* exported read only */
392 #define MNT_EXPORTED 0x0000000000000100ULL /* filesystem is exported */
393 #define MNT_DEFEXPORTED 0x0000000000000200ULL /* exported to the world */
394 #define MNT_EXPORTANON 0x0000000000000400ULL /* anon uid mapping for all */
395 #define MNT_EXKERB 0x0000000000000800ULL /* exported with Kerberos */
396 #define MNT_EXPUBLIC 0x0000000020000000ULL /* public export (WebNFS) */
397 #define MNT_EXTLS 0x0000004000000000ULL /* require TLS */
398 #define MNT_EXTLSCERT 0x0000008000000000ULL /* require TLS with client cert */
399 #define MNT_EXTLSCERTUSER 0x0000010000000000ULL /* require TLS with user cert */
402 * Flags set by internal operations, but visible to the user.
404 #define MNT_LOCAL 0x0000000000001000ULL /* filesystem is stored locally */
405 #define MNT_QUOTA 0x0000000000002000ULL /* quotas are enabled on fs */
406 #define MNT_ROOTFS 0x0000000000004000ULL /* identifies the root fs */
407 #define MNT_USER 0x0000000000008000ULL /* mounted by a user */
408 #define MNT_IGNORE 0x0000000000800000ULL /* do not show entry in df */
409 #define MNT_VERIFIED 0x0000000400000000ULL /* filesystem is verified */
412 * Mask of flags that are visible to statfs().
413 * XXX I think that this could now become (~(MNT_CMDFLAGS))
414 * but the 'mount' program may need changing to handle this.
416 #define MNT_VISFLAGMASK (MNT_RDONLY | MNT_SYNCHRONOUS | MNT_NOEXEC | \
417 MNT_NOSUID | MNT_UNION | MNT_SUJ | \
418 MNT_ASYNC | MNT_EXRDONLY | MNT_EXPORTED | \
419 MNT_DEFEXPORTED | MNT_EXPORTANON| MNT_EXKERB | \
420 MNT_LOCAL | MNT_USER | MNT_QUOTA | \
421 MNT_ROOTFS | MNT_NOATIME | MNT_NOCLUSTERR| \
422 MNT_NOCLUSTERW | MNT_SUIDDIR | MNT_SOFTDEP | \
423 MNT_IGNORE | MNT_EXPUBLIC | MNT_NOSYMFOLLOW | \
424 MNT_GJOURNAL | MNT_MULTILABEL | MNT_ACLS | \
425 MNT_NFS4ACLS | MNT_AUTOMOUNTED | MNT_VERIFIED | \
428 /* Mask of flags that can be updated. */
429 #define MNT_UPDATEMASK (MNT_NOSUID | MNT_NOEXEC | \
430 MNT_SYNCHRONOUS | MNT_UNION | MNT_ASYNC | \
432 MNT_NOSYMFOLLOW | MNT_IGNORE | \
433 MNT_NOCLUSTERR | MNT_NOCLUSTERW | MNT_SUIDDIR | \
434 MNT_ACLS | MNT_USER | MNT_NFS4ACLS | \
435 MNT_AUTOMOUNTED | MNT_UNTRUSTED)
438 * External filesystem command modifier flags.
439 * Unmount can use the MNT_FORCE flag.
440 * XXX: These are not STATES and really should be somewhere else.
441 * XXX: MNT_BYFSID and MNT_NONBUSY collide with MNT_ACLS and MNT_MULTILABEL,
442 * but because MNT_ACLS and MNT_MULTILABEL are only used for mount(2),
443 * and MNT_BYFSID and MNT_NONBUSY are only used for unmount(2),
446 #define MNT_UPDATE 0x0000000000010000ULL /* not real mount, just update */
447 #define MNT_DELEXPORT 0x0000000000020000ULL /* delete export host lists */
448 #define MNT_RELOAD 0x0000000000040000ULL /* reload filesystem data */
449 #define MNT_FORCE 0x0000000000080000ULL /* force unmount or readonly */
450 #define MNT_SNAPSHOT 0x0000000001000000ULL /* snapshot the filesystem */
451 #define MNT_NONBUSY 0x0000000004000000ULL /* check vnode use counts. */
452 #define MNT_BYFSID 0x0000000008000000ULL /* specify filesystem by ID. */
453 #define MNT_NOCOVER 0x0000001000000000ULL /* Do not cover a mount point */
454 #define MNT_EMPTYDIR 0x0000002000000000ULL /* Only mount on empty dir */
455 #define MNT_RECURSE 0x0000100000000000ULL /* recursively unmount uppers */
456 #define MNT_DEFERRED 0x0000200000000000ULL /* unmount in async context */
457 #define MNT_CMDFLAGS (MNT_UPDATE | MNT_DELEXPORT | MNT_RELOAD | \
458 MNT_FORCE | MNT_SNAPSHOT | MNT_NONBUSY | \
459 MNT_BYFSID | MNT_NOCOVER | MNT_EMPTYDIR | \
460 MNT_RECURSE | MNT_DEFERRED)
463 * Internal filesystem control flags stored in mnt_kern_flag.
465 * MNTK_UNMOUNT locks the mount entry so that name lookup cannot
466 * proceed past the mount point. This keeps the subtree stable during
467 * mounts and unmounts. When non-forced unmount flushes all vnodes
468 * from the mp queue, the MNTK_UNMOUNT flag prevents insmntque() from
469 * queueing new vnodes.
471 * MNTK_UNMOUNTF permits filesystems to detect a forced unmount while
472 * dounmount() is still waiting to lock the mountpoint. This allows
473 * the filesystem to cancel operations that might otherwise deadlock
474 * with the unmount attempt (used by NFS).
476 #define MNTK_UNMOUNTF 0x00000001 /* forced unmount in progress */
477 #define MNTK_ASYNC 0x00000002 /* filtered async flag */
478 #define MNTK_SOFTDEP 0x00000004 /* async disabled by softdep */
479 #define MNTK_NOMSYNC 0x00000008 /* don't do msync */
480 #define MNTK_DRAINING 0x00000010 /* lock draining is happening */
481 #define MNTK_REFEXPIRE 0x00000020 /* refcount expiring is happening */
482 #define MNTK_EXTENDED_SHARED 0x00000040 /* Allow shared locking for more ops */
483 #define MNTK_SHARED_WRITES 0x00000080 /* Allow shared locking for writes */
484 #define MNTK_NO_IOPF 0x00000100 /* Disallow page faults during reads
485 and writes. Filesystem shall properly
486 handle i/o state on EFAULT. */
487 #define MNTK_RECURSE 0x00000200 /* pending recursive unmount */
488 #define MNTK_UPPER_WAITER 0x00000400 /* waiting to drain MNTK_UPPER_PENDING */
489 #define MNTK_LOOKUP_EXCL_DOTDOT 0x00000800
490 /* UNUSED 0x00001000 */
491 #define MNTK_UNMAPPED_BUFS 0x00002000
492 #define MNTK_USES_BCACHE 0x00004000 /* FS uses the buffer cache. */
493 /* UNUSED 0x00008000 */
494 #define MNTK_VMSETSIZE_BUG 0x00010000
495 #define MNTK_UNIONFS 0x00020000 /* A hack for F_ISUNIONSTACK */
496 #define MNTK_FPLOOKUP 0x00040000 /* fast path lookup is supported */
497 #define MNTK_SUSPEND_ALL 0x00080000 /* Suspended by all-fs suspension */
498 #define MNTK_TASKQUEUE_WAITER 0x00100000 /* Waiting on unmount taskqueue */
499 /* UNUSED 0x00200000 */
500 /* UNUSED 0x00400000 */
501 #define MNTK_NOASYNC 0x00800000 /* disable async */
502 #define MNTK_UNMOUNT 0x01000000 /* unmount in progress */
503 #define MNTK_MWAIT 0x02000000 /* waiting for unmount to finish */
504 #define MNTK_SUSPEND 0x08000000 /* request write suspension */
505 #define MNTK_SUSPEND2 0x04000000 /* block secondary writes */
506 #define MNTK_SUSPENDED 0x10000000 /* write operations are suspended */
507 #define MNTK_NULL_NOCACHE 0x20000000 /* auto disable cache for nullfs
508 mounts over this fs */
509 #define MNTK_LOOKUP_SHARED 0x40000000 /* FS supports shared lock lookups */
510 /* UNUSED 0x80000000 */
514 MNT_SHARED_WRITES(struct mount *mp)
517 return (mp != NULL && (mp->mnt_kern_flag & MNTK_SHARED_WRITES) != 0);
521 MNT_EXTENDED_SHARED(struct mount *mp)
524 return (mp != NULL && (mp->mnt_kern_flag & MNTK_EXTENDED_SHARED) != 0);
529 * Sysctl CTL_VFS definitions.
531 * Second level identifier specifies which filesystem. Second level
532 * identifier VFS_VFSCONF returns information about all filesystems.
533 * Second level identifier VFS_GENERIC is non-terminal.
535 #define VFS_VFSCONF 0 /* get configured filesystems */
536 #define VFS_GENERIC 0 /* generic filesystem information */
538 * Third level identifiers for VFS_GENERIC are given below; third
539 * level identifiers for specific filesystems are given in their
540 * mount specific header files.
542 #define VFS_MAXTYPENUM 1 /* int: highest defined filesystem type */
543 #define VFS_CONF 2 /* struct: vfsconf for filesystem given
547 * Flags for various system call interfaces.
549 * waitfor flags to vfs_sync() and getfsstat()
551 #define MNT_WAIT 1 /* synchronously wait for I/O to complete */
552 #define MNT_NOWAIT 2 /* start all I/O, but do not wait for it */
553 #define MNT_LAZY 3 /* push data not written by filesystem syncer */
554 #define MNT_SUSPEND 4 /* Suspend file system after sync */
557 * Generic file handle
560 fsid_t fh_fsid; /* Filesystem id of mount point */
561 struct fid fh_fid; /* Filesys specific id */
563 typedef struct fhandle fhandle_t;
566 * Old export arguments without security flavor list
568 struct oexport_args {
569 int ex_flags; /* export related flags */
570 uid_t ex_root; /* mapping for root uid */
571 struct xucred ex_anon; /* mapping for anonymous user */
572 struct sockaddr *ex_addr; /* net address to which exported */
573 u_char ex_addrlen; /* and the net address length */
574 struct sockaddr *ex_mask; /* mask of valid bits in saddr */
575 u_char ex_masklen; /* and the smask length */
576 char *ex_indexfile; /* index file for WebNFS URLs */
580 * Not quite so old export arguments with 32bit ex_flags and xucred ex_anon.
582 #define MAXSECFLAVORS 5
583 struct o2export_args {
584 int ex_flags; /* export related flags */
585 uid_t ex_root; /* mapping for root uid */
586 struct xucred ex_anon; /* mapping for anonymous user */
587 struct sockaddr *ex_addr; /* net address to which exported */
588 u_char ex_addrlen; /* and the net address length */
589 struct sockaddr *ex_mask; /* mask of valid bits in saddr */
590 u_char ex_masklen; /* and the smask length */
591 char *ex_indexfile; /* index file for WebNFS URLs */
592 int ex_numsecflavors; /* security flavor count */
593 int ex_secflavors[MAXSECFLAVORS]; /* list of security flavors */
597 * Export arguments for local filesystem mount calls.
600 uint64_t ex_flags; /* export related flags */
601 uid_t ex_root; /* mapping for root uid */
602 uid_t ex_uid; /* mapping for anonymous user */
605 struct sockaddr *ex_addr; /* net address to which exported */
606 u_char ex_addrlen; /* and the net address length */
607 struct sockaddr *ex_mask; /* mask of valid bits in saddr */
608 u_char ex_masklen; /* and the smask length */
609 char *ex_indexfile; /* index file for WebNFS URLs */
610 int ex_numsecflavors; /* security flavor count */
611 int ex_secflavors[MAXSECFLAVORS]; /* list of security flavors */
615 * Structure holding information for a publicly exported filesystem
616 * (WebNFS). Currently the specs allow just for one such filesystem.
619 int np_valid; /* Do we hold valid information */
620 fhandle_t np_handle; /* Filehandle for pub fs (internal) */
621 struct mount *np_mount; /* Mountpoint of exported fs */
622 char *np_index; /* Index file */
626 * Filesystem configuration information. One of these exists for each
627 * type of filesystem supported by the kernel. These are searched at
628 * mount time to identify the requested filesystem.
630 * XXX: Never change the first two arguments!
633 u_int vfc_version; /* ABI version number */
634 char vfc_name[MFSNAMELEN]; /* filesystem type name */
635 struct vfsops *vfc_vfsops; /* filesystem operations vector */
636 struct vfsops *vfc_vfsops_sd; /* ... signal-deferred */
637 int vfc_typenum; /* historic filesystem type number */
638 int vfc_refcount; /* number mounted of this type */
639 int vfc_flags; /* permanent flags */
640 int vfc_prison_flag; /* prison allow.mount.* flag */
641 struct vfsoptdecl *vfc_opts; /* mount options */
642 TAILQ_ENTRY(vfsconf) vfc_list; /* list of vfscons */
645 /* Userland version of the struct vfsconf. */
647 struct vfsops *vfc_vfsops; /* filesystem operations vector */
648 char vfc_name[MFSNAMELEN]; /* filesystem type name */
649 int vfc_typenum; /* historic filesystem type number */
650 int vfc_refcount; /* number mounted of this type */
651 int vfc_flags; /* permanent flags */
652 struct vfsconf *vfc_next; /* next in list */
666 * NB: these flags refer to IMPLEMENTATION properties, not properties of
667 * any actual mounts; i.e., it does not make sense to change the flags.
669 #define VFCF_STATIC 0x00010000 /* statically compiled into kernel */
670 #define VFCF_NETWORK 0x00020000 /* may get data over the network */
671 #define VFCF_READONLY 0x00040000 /* writes are not implemented */
672 #define VFCF_SYNTHETIC 0x00080000 /* data does not represent real files */
673 #define VFCF_LOOPBACK 0x00100000 /* aliases some other mounted FS */
674 #define VFCF_UNICODE 0x00200000 /* stores file names as Unicode */
675 #define VFCF_JAIL 0x00400000 /* can be mounted from within a jail */
676 #define VFCF_DELEGADMIN 0x00800000 /* supports delegated administration */
677 #define VFCF_SBDRY 0x01000000 /* Stop at Boundary: defer stop requests
678 to kernel->user (AST) transition */
680 typedef uint32_t fsctlop_t;
683 int vc_vers; /* should be VFSIDCTL_VERS1 (below) */
684 fsid_t vc_fsid; /* fsid to operate on */
685 char vc_fstypename[MFSNAMELEN];
686 /* type of fs 'nfs' or '*' */
687 fsctlop_t vc_op; /* operation VFS_CTL_* (below) */
688 void *vc_ptr; /* pointer to data structure */
689 size_t vc_len; /* sizeof said structure */
690 u_int32_t vc_spare[12]; /* spare (must be zero) */
693 /* vfsidctl API version. */
694 #define VFS_CTL_VERS1 0x01
697 * New style VFS sysctls, do not reuse/conflict with the namespace for
699 * All "global" sysctl ops have the 33rd bit set:
701 * Private sysctl ops should have the 33rd bit unset.
703 #define VFS_CTL_QUERY 0x00010001 /* anything wrong? (vfsquery) */
704 #define VFS_CTL_TIMEO 0x00010002 /* set timeout for vfs notification */
705 #define VFS_CTL_NOLOCKS 0x00010003 /* disable file locking */
709 u_int32_t vq_spare[31];
713 #define VQ_NOTRESP 0x0001 /* server down */
714 #define VQ_NEEDAUTH 0x0002 /* server bad auth */
715 #define VQ_LOWDISK 0x0004 /* we're low on space */
716 #define VQ_MOUNT 0x0008 /* new filesystem arrived */
717 #define VQ_UNMOUNT 0x0010 /* filesystem has left */
718 #define VQ_DEAD 0x0020 /* filesystem is dead, needs force unmount */
719 #define VQ_ASSIST 0x0040 /* filesystem needs assistance from external
721 #define VQ_NOTRESPLOCK 0x0080 /* server lockd down */
722 #define VQ_FLAG0100 0x0100 /* placeholder */
723 #define VQ_FLAG0200 0x0200 /* placeholder */
724 #define VQ_FLAG0400 0x0400 /* placeholder */
725 #define VQ_FLAG0800 0x0800 /* placeholder */
726 #define VQ_FLAG1000 0x1000 /* placeholder */
727 #define VQ_FLAG2000 0x2000 /* placeholder */
728 #define VQ_FLAG4000 0x4000 /* placeholder */
729 #define VQ_FLAG8000 0x8000 /* placeholder */
732 /* Point a sysctl request at a vfsidctl's data. */
733 #define VCTLTOREQ(vc, req) \
735 (req)->newptr = (vc)->vc_ptr; \
736 (req)->newlen = (vc)->vc_len; \
747 * vfs_busy specific flags and mask.
749 #define MBF_NOWAIT 0x01
750 #define MBF_MNTLSTLOCK 0x02
751 #define MBF_MASK (MBF_NOWAIT | MBF_MNTLSTLOCK)
753 #ifdef MALLOC_DECLARE
754 MALLOC_DECLARE(M_MOUNT);
755 MALLOC_DECLARE(M_STATFS);
757 extern int maxvfsconf; /* highest defined filesystem type */
759 TAILQ_HEAD(vfsconfhead, vfsconf);
760 extern struct vfsconfhead vfsconf;
763 * Operations supported on mounted filesystem.
771 * N.B., vfs_cmount is the ancient vfsop invoked by the old mount(2) syscall.
772 * The new way is vfs_mount.
774 * vfs_cmount implementations typically translate arguments from their
775 * respective old per-FS structures into the key-value list supported by
776 * nmount(2), then use kernel_mount(9) to mimic nmount(2) from kernelspace.
778 * Filesystems with mounters that use nmount(2) do not need to and should not
779 * implement vfs_cmount. Hopefully a future cleanup can remove vfs_cmount and
782 typedef int vfs_cmount_t(struct mntarg *ma, void *data, uint64_t flags);
783 typedef int vfs_unmount_t(struct mount *mp, int mntflags);
784 typedef int vfs_root_t(struct mount *mp, int flags, struct vnode **vpp);
785 typedef int vfs_quotactl_t(struct mount *mp, int cmds, uid_t uid, void *arg,
787 typedef int vfs_statfs_t(struct mount *mp, struct statfs *sbp);
788 typedef int vfs_sync_t(struct mount *mp, int waitfor);
789 typedef int vfs_vget_t(struct mount *mp, ino_t ino, int flags,
791 typedef int vfs_fhtovp_t(struct mount *mp, struct fid *fhp,
792 int flags, struct vnode **vpp);
793 typedef int vfs_checkexp_t(struct mount *mp, struct sockaddr *nam,
794 uint64_t *extflagsp, struct ucred **credanonp,
795 int *numsecflavors, int *secflavors);
796 typedef int vfs_init_t(struct vfsconf *);
797 typedef int vfs_uninit_t(struct vfsconf *);
798 typedef int vfs_extattrctl_t(struct mount *mp, int cmd,
799 struct vnode *filename_vp, int attrnamespace,
800 const char *attrname);
801 typedef int vfs_mount_t(struct mount *mp);
802 typedef int vfs_sysctl_t(struct mount *mp, fsctlop_t op,
803 struct sysctl_req *req);
804 typedef void vfs_susp_clean_t(struct mount *mp);
805 typedef void vfs_notify_lowervp_t(struct mount *mp, struct vnode *lowervp);
806 typedef void vfs_purge_t(struct mount *mp);
809 vfs_mount_t *vfs_mount;
810 vfs_cmount_t *vfs_cmount;
811 vfs_unmount_t *vfs_unmount;
812 vfs_root_t *vfs_root;
813 vfs_root_t *vfs_cachedroot;
814 vfs_quotactl_t *vfs_quotactl;
815 vfs_statfs_t *vfs_statfs;
816 vfs_sync_t *vfs_sync;
817 vfs_vget_t *vfs_vget;
818 vfs_fhtovp_t *vfs_fhtovp;
819 vfs_checkexp_t *vfs_checkexp;
820 vfs_init_t *vfs_init;
821 vfs_uninit_t *vfs_uninit;
822 vfs_extattrctl_t *vfs_extattrctl;
823 vfs_sysctl_t *vfs_sysctl;
824 vfs_susp_clean_t *vfs_susp_clean;
825 vfs_notify_lowervp_t *vfs_reclaim_lowervp;
826 vfs_notify_lowervp_t *vfs_unlink_lowervp;
827 vfs_purge_t *vfs_purge;
828 vfs_mount_t *vfs_spare[6]; /* spares for ABI compat */
831 vfs_statfs_t __vfs_statfs;
833 #define VFS_MOUNT(MP) ({ \
836 TSRAW(curthread, TS_ENTER, "VFS_MOUNT", (MP)->mnt_vfc->vfc_name);\
837 _rc = (*(MP)->mnt_op->vfs_mount)(MP); \
838 TSRAW(curthread, TS_EXIT, "VFS_MOUNT", (MP)->mnt_vfc->vfc_name);\
841 #define VFS_UNMOUNT(MP, FORCE) ({ \
844 _rc = (*(MP)->mnt_op->vfs_unmount)(MP, FORCE); \
847 #define VFS_ROOT(MP, FLAGS, VPP) ({ \
850 _rc = (*(MP)->mnt_op->vfs_root)(MP, FLAGS, VPP); \
853 #define VFS_CACHEDROOT(MP, FLAGS, VPP) ({ \
856 _rc = (*(MP)->mnt_op->vfs_cachedroot)(MP, FLAGS, VPP); \
859 #define VFS_QUOTACTL(MP, C, U, A, MP_BUSY) ({ \
862 _rc = (*(MP)->mnt_op->vfs_quotactl)(MP, C, U, A, MP_BUSY); \
865 #define VFS_STATFS(MP, SBP) ({ \
868 _rc = __vfs_statfs((MP), (SBP)); \
871 #define VFS_SYNC(MP, WAIT) ({ \
874 _rc = (*(MP)->mnt_op->vfs_sync)(MP, WAIT); \
877 #define VFS_VGET(MP, INO, FLAGS, VPP) ({ \
880 _rc = (*(MP)->mnt_op->vfs_vget)(MP, INO, FLAGS, VPP); \
883 #define VFS_FHTOVP(MP, FIDP, FLAGS, VPP) ({ \
886 _rc = (*(MP)->mnt_op->vfs_fhtovp)(MP, FIDP, FLAGS, VPP); \
889 #define VFS_CHECKEXP(MP, NAM, EXFLG, CRED, NUMSEC, SEC) ({ \
892 _rc = (*(MP)->mnt_op->vfs_checkexp)(MP, NAM, EXFLG, CRED, NUMSEC,\
896 #define VFS_EXTATTRCTL(MP, C, FN, NS, N) ({ \
899 _rc = (*(MP)->mnt_op->vfs_extattrctl)(MP, C, FN, NS, N); \
902 #define VFS_SYSCTL(MP, OP, REQ) ({ \
905 _rc = (*(MP)->mnt_op->vfs_sysctl)(MP, OP, REQ); \
908 #define VFS_SUSP_CLEAN(MP) do { \
909 if (*(MP)->mnt_op->vfs_susp_clean != NULL) { \
910 (*(MP)->mnt_op->vfs_susp_clean)(MP); \
914 #define VFS_RECLAIM_LOWERVP(MP, VP) do { \
915 if (*(MP)->mnt_op->vfs_reclaim_lowervp != NULL) { \
916 (*(MP)->mnt_op->vfs_reclaim_lowervp)((MP), (VP)); \
920 #define VFS_UNLINK_LOWERVP(MP, VP) do { \
921 if (*(MP)->mnt_op->vfs_unlink_lowervp != NULL) { \
922 (*(MP)->mnt_op->vfs_unlink_lowervp)((MP), (VP)); \
926 #define VFS_PURGE(MP) do { \
927 if (*(MP)->mnt_op->vfs_purge != NULL) { \
928 (*(MP)->mnt_op->vfs_purge)(MP); \
932 #define VFS_KNOTE_LOCKED(vp, hint) do \
934 VN_KNOTE((vp), (hint), KNF_LISTLOCKED); \
937 #define VFS_KNOTE_UNLOCKED(vp, hint) do \
939 VN_KNOTE((vp), (hint), 0); \
942 #define VFS_NOTIFY_UPPER_RECLAIM 1
943 #define VFS_NOTIFY_UPPER_UNLINK 2
945 #include <sys/module.h>
950 #define VFS_VERSION_00 0x19660120
951 #define VFS_VERSION_01 0x20121030
952 #define VFS_VERSION_02 0x20180504
953 #define VFS_VERSION VFS_VERSION_02
955 #define VFS_SET(vfsops, fsname, flags) \
956 static struct vfsconf fsname ## _vfsconf = { \
957 .vfc_version = VFS_VERSION, \
958 .vfc_name = #fsname, \
959 .vfc_vfsops = &vfsops, \
961 .vfc_flags = flags, \
963 static moduledata_t fsname ## _mod = { \
966 & fsname ## _vfsconf \
968 DECLARE_MODULE(fsname, fsname ## _mod, SI_SUB_VFS, SI_ORDER_MIDDLE)
971 * exported vnode operations
974 int dounmount(struct mount *, uint64_t, struct thread *);
976 int kernel_mount(struct mntarg *ma, uint64_t flags);
977 struct mntarg *mount_arg(struct mntarg *ma, const char *name, const void *val, int len);
978 struct mntarg *mount_argb(struct mntarg *ma, int flag, const char *name);
979 struct mntarg *mount_argf(struct mntarg *ma, const char *name, const char *fmt, ...);
980 struct mntarg *mount_argsu(struct mntarg *ma, const char *name, const void *val, int len);
981 void statfs_scale_blocks(struct statfs *sf, long max_size);
982 struct vfsconf *vfs_byname(const char *);
983 struct vfsconf *vfs_byname_kld(const char *, struct thread *td, int *);
984 void vfs_mount_destroy(struct mount *);
985 void vfs_event_signal(fsid_t *, u_int32_t, intptr_t);
986 void vfs_freeopts(struct vfsoptlist *opts);
987 void vfs_deleteopt(struct vfsoptlist *opts, const char *name);
988 int vfs_buildopts(struct uio *auio, struct vfsoptlist **options);
989 int vfs_flagopt(struct vfsoptlist *opts, const char *name, uint64_t *w,
991 int vfs_getopt(struct vfsoptlist *, const char *, void **, int *);
992 int vfs_getopt_pos(struct vfsoptlist *opts, const char *name);
993 int vfs_getopt_size(struct vfsoptlist *opts, const char *name,
995 char *vfs_getopts(struct vfsoptlist *, const char *, int *error);
996 int vfs_copyopt(struct vfsoptlist *, const char *, void *, int);
997 int vfs_filteropt(struct vfsoptlist *, const char **legal);
998 void vfs_opterror(struct vfsoptlist *opts, const char *fmt, ...);
999 int vfs_scanopt(struct vfsoptlist *opts, const char *name, const char *fmt, ...);
1000 int vfs_setopt(struct vfsoptlist *opts, const char *name, void *value,
1002 int vfs_setopt_part(struct vfsoptlist *opts, const char *name, void *value,
1004 int vfs_setopts(struct vfsoptlist *opts, const char *name,
1006 int vfs_setpublicfs /* set publicly exported fs */
1007 (struct mount *, struct netexport *, struct export_args *);
1008 void vfs_periodic(struct mount *, int);
1009 int vfs_busy(struct mount *, int);
1010 int vfs_export /* process mount export info */
1011 (struct mount *, struct export_args *);
1012 void vfs_allocate_syncvnode(struct mount *);
1013 void vfs_deallocate_syncvnode(struct mount *);
1014 int vfs_donmount(struct thread *td, uint64_t fsflags,
1015 struct uio *fsoptions);
1016 void vfs_getnewfsid(struct mount *);
1017 struct mount *vfs_getvfs(fsid_t *); /* return vfs given fsid */
1018 struct mount *vfs_busyfs(fsid_t *);
1019 int vfs_modevent(module_t, int, void *);
1020 void vfs_mount_error(struct mount *, const char *, ...);
1021 void vfs_mountroot(void); /* mount our root filesystem */
1022 void vfs_mountedfrom(struct mount *, const char *from);
1023 void vfs_notify_upper(struct vnode *, int);
1024 struct mount *vfs_ref_from_vp(struct vnode *);
1025 void vfs_ref(struct mount *);
1026 void vfs_rel(struct mount *);
1027 struct mount *vfs_mount_alloc(struct vnode *, struct vfsconf *, const char *,
1029 int vfs_suser(struct mount *, struct thread *);
1030 void vfs_unbusy(struct mount *);
1031 void vfs_unmountall(void);
1032 struct mount *vfs_register_upper_from_vp(struct vnode *,
1033 struct mount *ump, struct mount_upper_node *);
1034 void vfs_register_for_notification(struct mount *, struct mount *,
1035 struct mount_upper_node *);
1036 void vfs_unregister_for_notification(struct mount *,
1037 struct mount_upper_node *);
1038 void vfs_unregister_upper(struct mount *, struct mount_upper_node *);
1039 int vfs_remount_ro(struct mount *mp);
1041 extern TAILQ_HEAD(mntlist, mount) mountlist; /* mounted filesystem list */
1042 extern struct mtx_padalign mountlist_mtx;
1043 extern struct nfs_public nfs_pub;
1044 extern struct sx vfsconf_sx;
1045 #define vfsconf_lock() sx_xlock(&vfsconf_sx)
1046 #define vfsconf_unlock() sx_xunlock(&vfsconf_sx)
1047 #define vfsconf_slock() sx_slock(&vfsconf_sx)
1048 #define vfsconf_sunlock() sx_sunlock(&vfsconf_sx)
1049 struct vnode *mntfs_allocvp(struct mount *, struct vnode *);
1050 void mntfs_freevp(struct vnode *);
1053 * Declarations for these vfs default operations are located in
1054 * kern/vfs_default.c. They will be automatically used to replace
1055 * null entries in VFS ops tables when registering a new filesystem
1056 * type in the global table.
1058 vfs_root_t vfs_stdroot;
1059 vfs_quotactl_t vfs_stdquotactl;
1060 vfs_statfs_t vfs_stdstatfs;
1061 vfs_sync_t vfs_stdsync;
1062 vfs_sync_t vfs_stdnosync;
1063 vfs_vget_t vfs_stdvget;
1064 vfs_fhtovp_t vfs_stdfhtovp;
1065 vfs_checkexp_t vfs_stdcheckexp;
1066 vfs_init_t vfs_stdinit;
1067 vfs_uninit_t vfs_stduninit;
1068 vfs_extattrctl_t vfs_stdextattrctl;
1069 vfs_sysctl_t vfs_stdsysctl;
1071 void syncer_suspend(void);
1072 void syncer_resume(void);
1074 struct vnode *vfs_cache_root_clear(struct mount *);
1075 void vfs_cache_root_set(struct mount *, struct vnode *);
1077 void vfs_op_barrier_wait(struct mount *);
1078 void vfs_op_enter(struct mount *);
1079 void vfs_op_exit_locked(struct mount *);
1080 void vfs_op_exit(struct mount *);
1083 void vfs_assert_mount_counters(struct mount *);
1084 void vfs_dump_mount_counters(struct mount *);
1086 #define vfs_assert_mount_counters(mp) do { } while (0)
1087 #define vfs_dump_mount_counters(mp) do { } while (0)
1090 enum mount_counter { MNT_COUNT_REF, MNT_COUNT_LOCKREF, MNT_COUNT_WRITEOPCOUNT };
1091 int vfs_mount_fetch_counter(struct mount *, enum mount_counter);
1093 void suspend_all_fs(void);
1094 void resume_all_fs(void);
1097 * Code transitioning mnt_vfs_ops to > 0 issues IPIs until it observes
1098 * all CPUs not executing code enclosed by thread_in_ops_pcpu variable.
1100 * This provides an invariant that by the time the last CPU is observed not
1101 * executing, everyone else entering will see the counter > 0 and exit.
1103 * Note there is no barrier between vfs_ops and the rest of the code in the
1104 * section. It is not necessary as the writer has to wait for everyone to drain
1105 * before making any changes or only make changes safe while the section is
1108 #define vfs_mount_pcpu(mp) zpcpu_get(mp->mnt_pcpu)
1109 #define vfs_mount_pcpu_remote(mp, cpu) zpcpu_get_cpu(mp->mnt_pcpu, cpu)
1111 #define vfs_op_thread_entered(mp) ({ \
1112 MPASS(curthread->td_critnest > 0); \
1113 struct mount_pcpu *_mpcpu = vfs_mount_pcpu(mp); \
1114 _mpcpu->mntp_thread_in_ops == 1; \
1117 #define vfs_op_thread_enter_crit(mp, _mpcpu) ({ \
1118 bool _retval_crit = true; \
1119 MPASS(curthread->td_critnest > 0); \
1120 _mpcpu = vfs_mount_pcpu(mp); \
1121 MPASS(mpcpu->mntp_thread_in_ops == 0); \
1122 _mpcpu->mntp_thread_in_ops = 1; \
1123 atomic_interrupt_fence(); \
1124 if (__predict_false(mp->mnt_vfs_ops > 0)) { \
1125 vfs_op_thread_exit_crit(mp, _mpcpu); \
1126 _retval_crit = false; \
1131 #define vfs_op_thread_enter(mp, _mpcpu) ({ \
1134 _retval = vfs_op_thread_enter_crit(mp, _mpcpu); \
1135 if (__predict_false(!_retval)) \
1140 #define vfs_op_thread_exit_crit(mp, _mpcpu) do { \
1141 MPASS(_mpcpu == vfs_mount_pcpu(mp)); \
1142 MPASS(_mpcpu->mntp_thread_in_ops == 1); \
1143 atomic_interrupt_fence(); \
1144 _mpcpu->mntp_thread_in_ops = 0; \
1147 #define vfs_op_thread_exit(mp, _mpcpu) do { \
1148 vfs_op_thread_exit_crit(mp, _mpcpu); \
1152 #define vfs_mp_count_add_pcpu(_mpcpu, count, val) do { \
1153 MPASS(_mpcpu->mntp_thread_in_ops == 1); \
1154 _mpcpu->mntp_##count += val; \
1157 #define vfs_mp_count_sub_pcpu(_mpcpu, count, val) do { \
1158 MPASS(_mpcpu->mntp_thread_in_ops == 1); \
1159 _mpcpu->mntp_##count -= val; \
1162 #else /* !_KERNEL */
1164 #include <sys/cdefs.h>
1169 int fhlink(struct fhandle *, const char *);
1170 int fhlinkat(struct fhandle *, int, const char *);
1171 int fhopen(const struct fhandle *, int);
1172 int fhreadlink(struct fhandle *, char *, size_t);
1173 int fhstat(const struct fhandle *, struct stat *);
1174 int fhstatfs(const struct fhandle *, struct statfs *);
1175 int fstatfs(int, struct statfs *);
1176 int getfh(const char *, fhandle_t *);
1177 int getfhat(int, char *, struct fhandle *, int);
1178 int getfsstat(struct statfs *, long, int);
1179 int getmntinfo(struct statfs **, int);
1180 int lgetfh(const char *, fhandle_t *);
1181 int mount(const char *, const char *, int, void *);
1182 int nmount(struct iovec *, unsigned int, int);
1183 int statfs(const char *, struct statfs *);
1184 int unmount(const char *, int);
1186 /* C library stuff */
1187 int getvfsbyname(const char *, struct xvfsconf *);
1190 #endif /* _KERNEL */
1192 #endif /* !_SYS_MOUNT_H_ */