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>
42 #include <sys/lockmgr.h>
43 #include <sys/tslog.h>
44 #include <sys/_mutex.h>
49 * NOTE: When changing statfs structure, mount structure, MNT_* flags or
50 * MNTK_* flags also update DDB show mount command in vfs_subr.c.
53 typedef struct fsid { int32_t val[2]; } fsid_t; /* filesystem id type */
55 #define fsidcmp(a, b) memcmp((a), (b), sizeof(fsid_t))
59 * These are unique per filesystem on a single machine.
64 u_short fid_len; /* length of data in bytes */
65 u_short fid_data0; /* force longword alignment */
66 char fid_data[MAXFIDSZ]; /* data (variable length) */
70 * filesystem statistics
72 #define MFSNAMELEN 16 /* length of type name including null */
73 #define MNAMELEN 1024 /* size of on/from name bufs */
74 #define STATFS_VERSION 0x20140518 /* current version number */
76 uint32_t f_version; /* structure version number */
77 uint32_t f_type; /* type of filesystem */
78 uint64_t f_flags; /* copy of mount exported flags */
79 uint64_t f_bsize; /* filesystem fragment size */
80 uint64_t f_iosize; /* optimal transfer block size */
81 uint64_t f_blocks; /* total data blocks in filesystem */
82 uint64_t f_bfree; /* free blocks in filesystem */
83 int64_t f_bavail; /* free blocks avail to non-superuser */
84 uint64_t f_files; /* total file nodes in filesystem */
85 int64_t f_ffree; /* free nodes avail to non-superuser */
86 uint64_t f_syncwrites; /* count of sync writes since mount */
87 uint64_t f_asyncwrites; /* count of async writes since mount */
88 uint64_t f_syncreads; /* count of sync reads since mount */
89 uint64_t f_asyncreads; /* count of async reads since mount */
90 uint64_t f_spare[10]; /* unused spare */
91 uint32_t f_namemax; /* maximum filename length */
92 uid_t f_owner; /* user that mounted the filesystem */
93 fsid_t f_fsid; /* filesystem id */
94 char f_charspare[80]; /* spare string space */
95 char f_fstypename[MFSNAMELEN]; /* filesystem type name */
96 char f_mntfromname[MNAMELEN]; /* mounted filesystem */
97 char f_mntonname[MNAMELEN]; /* directory on which mounted */
100 #if defined(_WANT_FREEBSD11_STATFS) || defined(_KERNEL)
101 #define FREEBSD11_STATFS_VERSION 0x20030518 /* current version number */
102 struct freebsd11_statfs {
103 uint32_t f_version; /* structure version number */
104 uint32_t f_type; /* type of filesystem */
105 uint64_t f_flags; /* copy of mount exported flags */
106 uint64_t f_bsize; /* filesystem fragment size */
107 uint64_t f_iosize; /* optimal transfer block size */
108 uint64_t f_blocks; /* total data blocks in filesystem */
109 uint64_t f_bfree; /* free blocks in filesystem */
110 int64_t f_bavail; /* free blocks avail to non-superuser */
111 uint64_t f_files; /* total file nodes in filesystem */
112 int64_t f_ffree; /* free nodes avail to non-superuser */
113 uint64_t f_syncwrites; /* count of sync writes since mount */
114 uint64_t f_asyncwrites; /* count of async writes since mount */
115 uint64_t f_syncreads; /* count of sync reads since mount */
116 uint64_t f_asyncreads; /* count of async reads since mount */
117 uint64_t f_spare[10]; /* unused spare */
118 uint32_t f_namemax; /* maximum filename length */
119 uid_t f_owner; /* user that mounted the filesystem */
120 fsid_t f_fsid; /* filesystem id */
121 char f_charspare[80]; /* spare string space */
122 char f_fstypename[16]; /* filesystem type name */
123 char f_mntfromname[88]; /* mounted filesystem */
124 char f_mntonname[88]; /* directory on which mounted */
126 #endif /* _WANT_FREEBSD11_STATFS || _KERNEL */
129 #define OMFSNAMELEN 16 /* length of fs type name, including null */
130 #define OMNAMELEN (88 - 2 * sizeof(long)) /* size of on/from name bufs */
132 /* XXX getfsstat.2 is out of date with write and read counter changes here. */
133 /* XXX statfs.2 is out of date with read counter changes here. */
135 long f_spare2; /* placeholder */
136 long f_bsize; /* fundamental filesystem block size */
137 long f_iosize; /* optimal transfer block size */
138 long f_blocks; /* total data blocks in filesystem */
139 long f_bfree; /* free blocks in fs */
140 long f_bavail; /* free blocks avail to non-superuser */
141 long f_files; /* total file nodes in filesystem */
142 long f_ffree; /* free file nodes in fs */
143 fsid_t f_fsid; /* filesystem id */
144 uid_t f_owner; /* user that mounted the filesystem */
145 int f_type; /* type of filesystem */
146 int f_flags; /* copy of mount exported flags */
147 long f_syncwrites; /* count of sync writes since mount */
148 long f_asyncwrites; /* count of async writes since mount */
149 char f_fstypename[OMFSNAMELEN]; /* fs type name */
150 char f_mntonname[OMNAMELEN]; /* directory on which mounted */
151 long f_syncreads; /* count of sync reads since mount */
152 long f_asyncreads; /* count of async reads since mount */
153 short f_spares1; /* unused spare */
154 char f_mntfromname[OMNAMELEN];/* mounted filesystem */
155 short f_spares2; /* unused spare */
157 * XXX on machines where longs are aligned to 8-byte boundaries, there
158 * is an unnamed int32_t here. This spare was after the apparent end
159 * of the struct until we bit off the read counters from f_mntonname.
161 long f_spare[2]; /* unused spare */
164 TAILQ_HEAD(vnodelst, vnode);
166 /* Mount options list */
167 TAILQ_HEAD(vfsoptlist, vfsopt);
169 TAILQ_ENTRY(vfsopt) link;
178 * Structure per mounted filesystem. Each mounted filesystem has an
179 * array of operations and an instance record. The filesystems are
180 * put on a doubly linked list.
186 * v - vnode freelist mutex
188 * Unmarked fields are considered stable as long as a ref is held.
192 struct mtx mnt_mtx; /* mount structure interlock */
193 int mnt_gen; /* struct mount generation */
194 #define mnt_startzero mnt_list
195 TAILQ_ENTRY(mount) mnt_list; /* (m) mount list */
196 struct vfsops *mnt_op; /* operations on fs */
197 struct vfsconf *mnt_vfc; /* configuration info */
198 struct vnode *mnt_vnodecovered; /* vnode we mounted on */
199 struct vnode *mnt_syncer; /* syncer vnode */
200 int mnt_ref; /* (i) Reference count */
201 struct vnodelst mnt_nvnodelist; /* (i) list of vnodes */
202 int mnt_nvnodelistsize; /* (i) # of vnodes */
203 int mnt_writeopcount; /* (i) write syscalls pending */
204 int mnt_kern_flag; /* (i) kernel only flags */
205 uint64_t mnt_flag; /* (i) flags shared with user */
206 struct vfsoptlist *mnt_opt; /* current mount options */
207 struct vfsoptlist *mnt_optnew; /* new options passed to fs */
208 int mnt_maxsymlinklen; /* max size of short symlink */
209 struct statfs mnt_stat; /* cache of filesystem stats */
210 struct ucred *mnt_cred; /* credentials of mounter */
211 void * mnt_data; /* private data */
212 time_t mnt_time; /* last time written*/
213 int mnt_iosize_max; /* max size for clusters, etc */
214 struct netexport *mnt_export; /* export list */
215 struct label *mnt_label; /* MAC label for the fs */
216 u_int mnt_hashseed; /* Random seed for vfs_hash */
217 int mnt_lockref; /* (i) Lock reference count */
218 int mnt_secondary_writes; /* (i) # of secondary writes */
219 int mnt_secondary_accwrites;/* (i) secondary wr. starts */
220 struct thread *mnt_susp_owner; /* (i) thread owning suspension */
221 #define mnt_endzero mnt_gjprovider
222 char *mnt_gjprovider; /* gjournal provider name */
223 struct mtx mnt_listmtx;
224 struct vnodelst mnt_lazyvnodelist; /* (l) list of lazy vnodes */
225 int mnt_lazyvnodelistsize; /* (l) # of lazy vnodes */
226 struct lock mnt_explock; /* vfs_export walkers lock */
227 TAILQ_ENTRY(mount) mnt_upper_link; /* (m) we in the all uppers */
228 TAILQ_HEAD(, mount) mnt_uppers; /* (m) upper mounts over us*/
229 int __aligned(CACHE_LINE_SIZE) mnt_vfs_ops;/* (i) pending vfs ops */
230 int *mnt_thread_in_ops_pcpu;
232 int *mnt_lockref_pcpu;
233 int *mnt_writeopcount_pcpu;
234 struct vnode *mnt_rootvnode;
238 * Definitions for MNT_VNODE_FOREACH_ALL.
240 struct vnode *__mnt_vnode_next_all(struct vnode **mvp, struct mount *mp);
241 struct vnode *__mnt_vnode_first_all(struct vnode **mvp, struct mount *mp);
242 void __mnt_vnode_markerfree_all(struct vnode **mvp, struct mount *mp);
244 #define MNT_VNODE_FOREACH_ALL(vp, mp, mvp) \
245 for (vp = __mnt_vnode_first_all(&(mvp), (mp)); \
246 (vp) != NULL; vp = __mnt_vnode_next_all(&(mvp), (mp)))
248 #define MNT_VNODE_FOREACH_ALL_ABORT(mp, mvp) \
251 __mnt_vnode_markerfree_all(&(mvp), (mp)); \
252 /* MNT_IUNLOCK(mp); -- done in above function */ \
253 mtx_assert(MNT_MTX(mp), MA_NOTOWNED); \
257 * Definitions for MNT_VNODE_FOREACH_LAZY.
259 typedef int mnt_lazy_cb_t(struct vnode *, void *);
260 struct vnode *__mnt_vnode_next_lazy(struct vnode **mvp, struct mount *mp,
261 mnt_lazy_cb_t *cb, void *cbarg);
262 struct vnode *__mnt_vnode_first_lazy(struct vnode **mvp, struct mount *mp,
263 mnt_lazy_cb_t *cb, void *cbarg);
264 void __mnt_vnode_markerfree_lazy(struct vnode **mvp, struct mount *mp);
266 #define MNT_VNODE_FOREACH_LAZY(vp, mp, mvp, cb, cbarg) \
267 for (vp = __mnt_vnode_first_lazy(&(mvp), (mp), (cb), (cbarg)); \
269 vp = __mnt_vnode_next_lazy(&(mvp), (mp), (cb), (cbarg)))
271 #define MNT_VNODE_FOREACH_LAZY_ABORT(mp, mvp) \
272 __mnt_vnode_markerfree_lazy(&(mvp), (mp))
274 #define MNT_ILOCK(mp) mtx_lock(&(mp)->mnt_mtx)
275 #define MNT_ITRYLOCK(mp) mtx_trylock(&(mp)->mnt_mtx)
276 #define MNT_IUNLOCK(mp) mtx_unlock(&(mp)->mnt_mtx)
277 #define MNT_MTX(mp) (&(mp)->mnt_mtx)
279 #define MNT_REF(mp) do { \
280 mtx_assert(MNT_MTX(mp), MA_OWNED); \
283 #define MNT_REL(mp) do { \
284 mtx_assert(MNT_MTX(mp), MA_OWNED); \
286 if ((mp)->mnt_vfs_ops && (mp)->mnt_ref < 0) \
287 vfs_dump_mount_counters(mp); \
288 if ((mp)->mnt_ref == 0 && (mp)->mnt_vfs_ops) \
295 * User specifiable flags, stored in mnt_flag.
297 #define MNT_RDONLY 0x0000000000000001ULL /* read only filesystem */
298 #define MNT_SYNCHRONOUS 0x0000000000000002ULL /* fs written synchronously */
299 #define MNT_NOEXEC 0x0000000000000004ULL /* can't exec from filesystem */
300 #define MNT_NOSUID 0x0000000000000008ULL /* don't honor setuid fs bits */
301 #define MNT_NFS4ACLS 0x0000000000000010ULL /* enable NFS version 4 ACLs */
302 #define MNT_UNION 0x0000000000000020ULL /* union with underlying fs */
303 #define MNT_ASYNC 0x0000000000000040ULL /* fs written asynchronously */
304 #define MNT_SUIDDIR 0x0000000000100000ULL /* special SUID dir handling */
305 #define MNT_SOFTDEP 0x0000000000200000ULL /* using soft updates */
306 #define MNT_NOSYMFOLLOW 0x0000000000400000ULL /* do not follow symlinks */
307 #define MNT_GJOURNAL 0x0000000002000000ULL /* GEOM journal support enabled */
308 #define MNT_MULTILABEL 0x0000000004000000ULL /* MAC support for objects */
309 #define MNT_ACLS 0x0000000008000000ULL /* ACL support enabled */
310 #define MNT_NOATIME 0x0000000010000000ULL /* dont update file access time */
311 #define MNT_NOCLUSTERR 0x0000000040000000ULL /* disable cluster read */
312 #define MNT_NOCLUSTERW 0x0000000080000000ULL /* disable cluster write */
313 #define MNT_SUJ 0x0000000100000000ULL /* using journaled soft updates */
314 #define MNT_AUTOMOUNTED 0x0000000200000000ULL /* mounted by automountd(8) */
315 #define MNT_UNTRUSTED 0x0000000800000000ULL /* filesys metadata untrusted */
318 * NFS export related mount flags.
320 #define MNT_EXRDONLY 0x0000000000000080ULL /* exported read only */
321 #define MNT_EXPORTED 0x0000000000000100ULL /* filesystem is exported */
322 #define MNT_DEFEXPORTED 0x0000000000000200ULL /* exported to the world */
323 #define MNT_EXPORTANON 0x0000000000000400ULL /* anon uid mapping for all */
324 #define MNT_EXKERB 0x0000000000000800ULL /* exported with Kerberos */
325 #define MNT_EXPUBLIC 0x0000000020000000ULL /* public export (WebNFS) */
328 * Flags set by internal operations,
329 * but visible to the user.
330 * XXX some of these are not quite right.. (I've never seen the root flag set)
332 #define MNT_LOCAL 0x0000000000001000ULL /* filesystem is stored locally */
333 #define MNT_QUOTA 0x0000000000002000ULL /* quotas are enabled on fs */
334 #define MNT_ROOTFS 0x0000000000004000ULL /* identifies the root fs */
335 #define MNT_USER 0x0000000000008000ULL /* mounted by a user */
336 #define MNT_IGNORE 0x0000000000800000ULL /* do not show entry in df */
337 #define MNT_VERIFIED 0x0000000400000000ULL /* filesystem is verified */
340 * Mask of flags that are visible to statfs().
341 * XXX I think that this could now become (~(MNT_CMDFLAGS))
342 * but the 'mount' program may need changing to handle this.
344 #define MNT_VISFLAGMASK (MNT_RDONLY | MNT_SYNCHRONOUS | MNT_NOEXEC | \
345 MNT_NOSUID | MNT_UNION | MNT_SUJ | \
346 MNT_ASYNC | MNT_EXRDONLY | MNT_EXPORTED | \
347 MNT_DEFEXPORTED | MNT_EXPORTANON| MNT_EXKERB | \
348 MNT_LOCAL | MNT_USER | MNT_QUOTA | \
349 MNT_ROOTFS | MNT_NOATIME | MNT_NOCLUSTERR| \
350 MNT_NOCLUSTERW | MNT_SUIDDIR | MNT_SOFTDEP | \
351 MNT_IGNORE | MNT_EXPUBLIC | MNT_NOSYMFOLLOW | \
352 MNT_GJOURNAL | MNT_MULTILABEL | MNT_ACLS | \
353 MNT_NFS4ACLS | MNT_AUTOMOUNTED | MNT_VERIFIED | \
356 /* Mask of flags that can be updated. */
357 #define MNT_UPDATEMASK (MNT_NOSUID | MNT_NOEXEC | \
358 MNT_SYNCHRONOUS | MNT_UNION | MNT_ASYNC | \
360 MNT_NOSYMFOLLOW | MNT_IGNORE | \
361 MNT_NOCLUSTERR | MNT_NOCLUSTERW | MNT_SUIDDIR | \
362 MNT_ACLS | MNT_USER | MNT_NFS4ACLS | \
363 MNT_AUTOMOUNTED | MNT_UNTRUSTED)
366 * External filesystem command modifier flags.
367 * Unmount can use the MNT_FORCE flag.
368 * XXX: These are not STATES and really should be somewhere else.
369 * XXX: MNT_BYFSID and MNT_NONBUSY collide with MNT_ACLS and MNT_MULTILABEL,
370 * but because MNT_ACLS and MNT_MULTILABEL are only used for mount(2),
371 * and MNT_BYFSID and MNT_NONBUSY are only used for unmount(2),
374 #define MNT_UPDATE 0x0000000000010000ULL /* not real mount, just update */
375 #define MNT_DELEXPORT 0x0000000000020000ULL /* delete export host lists */
376 #define MNT_RELOAD 0x0000000000040000ULL /* reload filesystem data */
377 #define MNT_FORCE 0x0000000000080000ULL /* force unmount or readonly */
378 #define MNT_SNAPSHOT 0x0000000001000000ULL /* snapshot the filesystem */
379 #define MNT_NONBUSY 0x0000000004000000ULL /* check vnode use counts. */
380 #define MNT_BYFSID 0x0000000008000000ULL /* specify filesystem by ID. */
381 #define MNT_NOCOVER 0x0000001000000000ULL /* Do not cover a mount point */
382 #define MNT_EMPTYDIR 0x0000002000000000ULL /* Only mount on empty dir */
383 #define MNT_CMDFLAGS (MNT_UPDATE | MNT_DELEXPORT | MNT_RELOAD | \
384 MNT_FORCE | MNT_SNAPSHOT | MNT_NONBUSY | \
385 MNT_BYFSID | MNT_NOCOVER | MNT_EMPTYDIR)
387 * Internal filesystem control flags stored in mnt_kern_flag.
389 * MNTK_UNMOUNT locks the mount entry so that name lookup cannot
390 * proceed past the mount point. This keeps the subtree stable during
391 * mounts and unmounts. When non-forced unmount flushes all vnodes
392 * from the mp queue, the MNTK_UNMOUNT flag prevents insmntque() from
393 * queueing new vnodes.
395 * MNTK_UNMOUNTF permits filesystems to detect a forced unmount while
396 * dounmount() is still waiting to lock the mountpoint. This allows
397 * the filesystem to cancel operations that might otherwise deadlock
398 * with the unmount attempt (used by NFS).
400 #define MNTK_UNMOUNTF 0x00000001 /* forced unmount in progress */
401 #define MNTK_ASYNC 0x00000002 /* filtered async flag */
402 #define MNTK_SOFTDEP 0x00000004 /* async disabled by softdep */
403 #define MNTK_NOMSYNC 0x00000008 /* don't do msync */
404 #define MNTK_DRAINING 0x00000010 /* lock draining is happening */
405 #define MNTK_REFEXPIRE 0x00000020 /* refcount expiring is happening */
406 #define MNTK_EXTENDED_SHARED 0x00000040 /* Allow shared locking for more ops */
407 #define MNTK_SHARED_WRITES 0x00000080 /* Allow shared locking for writes */
408 #define MNTK_NO_IOPF 0x00000100 /* Disallow page faults during reads
409 and writes. Filesystem shall properly
410 handle i/o state on EFAULT. */
411 #define MNTK_VGONE_UPPER 0x00000200
412 #define MNTK_VGONE_WAITER 0x00000400
413 #define MNTK_LOOKUP_EXCL_DOTDOT 0x00000800
414 #define MNTK_MARKER 0x00001000
415 #define MNTK_UNMAPPED_BUFS 0x00002000
416 #define MNTK_USES_BCACHE 0x00004000 /* FS uses the buffer cache. */
417 #define MNTK_TEXT_REFS 0x00008000 /* Keep use ref for text */
418 #define MNTK_VMSETSIZE_BUG 0x00010000
419 #define MNTK_UNIONFS 0x00020000 /* A hack for F_ISUNIONSTACK */
420 #define MNTK_NOASYNC 0x00800000 /* disable async */
421 #define MNTK_UNMOUNT 0x01000000 /* unmount in progress */
422 #define MNTK_MWAIT 0x02000000 /* waiting for unmount to finish */
423 #define MNTK_SUSPEND 0x08000000 /* request write suspension */
424 #define MNTK_SUSPEND2 0x04000000 /* block secondary writes */
425 #define MNTK_SUSPENDED 0x10000000 /* write operations are suspended */
426 #define MNTK_NULL_NOCACHE 0x20000000 /* auto disable cache for nullfs
427 mounts over this fs */
428 #define MNTK_LOOKUP_SHARED 0x40000000 /* FS supports shared lock lookups */
429 #define MNTK_NOKNOTE 0x80000000 /* Don't send KNOTEs from VOP hooks */
433 MNT_SHARED_WRITES(struct mount *mp)
436 return (mp != NULL && (mp->mnt_kern_flag & MNTK_SHARED_WRITES) != 0);
440 MNT_EXTENDED_SHARED(struct mount *mp)
443 return (mp != NULL && (mp->mnt_kern_flag & MNTK_EXTENDED_SHARED) != 0);
448 * Sysctl CTL_VFS definitions.
450 * Second level identifier specifies which filesystem. Second level
451 * identifier VFS_VFSCONF returns information about all filesystems.
452 * Second level identifier VFS_GENERIC is non-terminal.
454 #define VFS_VFSCONF 0 /* get configured filesystems */
455 #define VFS_GENERIC 0 /* generic filesystem information */
457 * Third level identifiers for VFS_GENERIC are given below; third
458 * level identifiers for specific filesystems are given in their
459 * mount specific header files.
461 #define VFS_MAXTYPENUM 1 /* int: highest defined filesystem type */
462 #define VFS_CONF 2 /* struct: vfsconf for filesystem given
466 * Flags for various system call interfaces.
468 * waitfor flags to vfs_sync() and getfsstat()
470 #define MNT_WAIT 1 /* synchronously wait for I/O to complete */
471 #define MNT_NOWAIT 2 /* start all I/O, but do not wait for it */
472 #define MNT_LAZY 3 /* push data not written by filesystem syncer */
473 #define MNT_SUSPEND 4 /* Suspend file system after sync */
476 * Generic file handle
479 fsid_t fh_fsid; /* Filesystem id of mount point */
480 struct fid fh_fid; /* Filesys specific id */
482 typedef struct fhandle fhandle_t;
485 * Old export arguments without security flavor list
487 struct oexport_args {
488 int ex_flags; /* export related flags */
489 uid_t ex_root; /* mapping for root uid */
490 struct xucred ex_anon; /* mapping for anonymous user */
491 struct sockaddr *ex_addr; /* net address to which exported */
492 u_char ex_addrlen; /* and the net address length */
493 struct sockaddr *ex_mask; /* mask of valid bits in saddr */
494 u_char ex_masklen; /* and the smask length */
495 char *ex_indexfile; /* index file for WebNFS URLs */
499 * Export arguments for local filesystem mount calls.
501 #define MAXSECFLAVORS 5
503 int ex_flags; /* export related flags */
504 uid_t ex_root; /* mapping for root uid */
505 struct xucred ex_anon; /* mapping for anonymous user */
506 struct sockaddr *ex_addr; /* net address to which exported */
507 u_char ex_addrlen; /* and the net address length */
508 struct sockaddr *ex_mask; /* mask of valid bits in saddr */
509 u_char ex_masklen; /* and the smask length */
510 char *ex_indexfile; /* index file for WebNFS URLs */
511 int ex_numsecflavors; /* security flavor count */
512 int ex_secflavors[MAXSECFLAVORS]; /* list of security flavors */
516 * Structure holding information for a publicly exported filesystem
517 * (WebNFS). Currently the specs allow just for one such filesystem.
520 int np_valid; /* Do we hold valid information */
521 fhandle_t np_handle; /* Filehandle for pub fs (internal) */
522 struct mount *np_mount; /* Mountpoint of exported fs */
523 char *np_index; /* Index file */
527 * Filesystem configuration information. One of these exists for each
528 * type of filesystem supported by the kernel. These are searched at
529 * mount time to identify the requested filesystem.
531 * XXX: Never change the first two arguments!
534 u_int vfc_version; /* ABI version number */
535 char vfc_name[MFSNAMELEN]; /* filesystem type name */
536 struct vfsops *vfc_vfsops; /* filesystem operations vector */
537 struct vfsops *vfc_vfsops_sd; /* ... signal-deferred */
538 int vfc_typenum; /* historic filesystem type number */
539 int vfc_refcount; /* number mounted of this type */
540 int vfc_flags; /* permanent flags */
541 int vfc_prison_flag; /* prison allow.mount.* flag */
542 struct vfsoptdecl *vfc_opts; /* mount options */
543 TAILQ_ENTRY(vfsconf) vfc_list; /* list of vfscons */
546 /* Userland version of the struct vfsconf. */
548 struct vfsops *vfc_vfsops; /* filesystem operations vector */
549 char vfc_name[MFSNAMELEN]; /* filesystem type name */
550 int vfc_typenum; /* historic filesystem type number */
551 int vfc_refcount; /* number mounted of this type */
552 int vfc_flags; /* permanent flags */
553 struct vfsconf *vfc_next; /* next in list */
567 * NB: these flags refer to IMPLEMENTATION properties, not properties of
568 * any actual mounts; i.e., it does not make sense to change the flags.
570 #define VFCF_STATIC 0x00010000 /* statically compiled into kernel */
571 #define VFCF_NETWORK 0x00020000 /* may get data over the network */
572 #define VFCF_READONLY 0x00040000 /* writes are not implemented */
573 #define VFCF_SYNTHETIC 0x00080000 /* data does not represent real files */
574 #define VFCF_LOOPBACK 0x00100000 /* aliases some other mounted FS */
575 #define VFCF_UNICODE 0x00200000 /* stores file names as Unicode */
576 #define VFCF_JAIL 0x00400000 /* can be mounted from within a jail */
577 #define VFCF_DELEGADMIN 0x00800000 /* supports delegated administration */
578 #define VFCF_SBDRY 0x01000000 /* Stop at Boundary: defer stop requests
579 to kernel->user (AST) transition */
581 typedef uint32_t fsctlop_t;
584 int vc_vers; /* should be VFSIDCTL_VERS1 (below) */
585 fsid_t vc_fsid; /* fsid to operate on */
586 char vc_fstypename[MFSNAMELEN];
587 /* type of fs 'nfs' or '*' */
588 fsctlop_t vc_op; /* operation VFS_CTL_* (below) */
589 void *vc_ptr; /* pointer to data structure */
590 size_t vc_len; /* sizeof said structure */
591 u_int32_t vc_spare[12]; /* spare (must be zero) */
594 /* vfsidctl API version. */
595 #define VFS_CTL_VERS1 0x01
598 * New style VFS sysctls, do not reuse/conflict with the namespace for
600 * All "global" sysctl ops have the 33rd bit set:
602 * Private sysctl ops should have the 33rd bit unset.
604 #define VFS_CTL_QUERY 0x00010001 /* anything wrong? (vfsquery) */
605 #define VFS_CTL_TIMEO 0x00010002 /* set timeout for vfs notification */
606 #define VFS_CTL_NOLOCKS 0x00010003 /* disable file locking */
610 u_int32_t vq_spare[31];
614 #define VQ_NOTRESP 0x0001 /* server down */
615 #define VQ_NEEDAUTH 0x0002 /* server bad auth */
616 #define VQ_LOWDISK 0x0004 /* we're low on space */
617 #define VQ_MOUNT 0x0008 /* new filesystem arrived */
618 #define VQ_UNMOUNT 0x0010 /* filesystem has left */
619 #define VQ_DEAD 0x0020 /* filesystem is dead, needs force unmount */
620 #define VQ_ASSIST 0x0040 /* filesystem needs assistance from external
622 #define VQ_NOTRESPLOCK 0x0080 /* server lockd down */
623 #define VQ_FLAG0100 0x0100 /* placeholder */
624 #define VQ_FLAG0200 0x0200 /* placeholder */
625 #define VQ_FLAG0400 0x0400 /* placeholder */
626 #define VQ_FLAG0800 0x0800 /* placeholder */
627 #define VQ_FLAG1000 0x1000 /* placeholder */
628 #define VQ_FLAG2000 0x2000 /* placeholder */
629 #define VQ_FLAG4000 0x4000 /* placeholder */
630 #define VQ_FLAG8000 0x8000 /* placeholder */
633 /* Point a sysctl request at a vfsidctl's data. */
634 #define VCTLTOREQ(vc, req) \
636 (req)->newptr = (vc)->vc_ptr; \
637 (req)->newlen = (vc)->vc_len; \
648 * vfs_busy specific flags and mask.
650 #define MBF_NOWAIT 0x01
651 #define MBF_MNTLSTLOCK 0x02
652 #define MBF_MASK (MBF_NOWAIT | MBF_MNTLSTLOCK)
654 #ifdef MALLOC_DECLARE
655 MALLOC_DECLARE(M_MOUNT);
656 MALLOC_DECLARE(M_STATFS);
658 extern int maxvfsconf; /* highest defined filesystem type */
660 TAILQ_HEAD(vfsconfhead, vfsconf);
661 extern struct vfsconfhead vfsconf;
664 * Operations supported on mounted filesystem.
672 * N.B., vfs_cmount is the ancient vfsop invoked by the old mount(2) syscall.
673 * The new way is vfs_mount.
675 * vfs_cmount implementations typically translate arguments from their
676 * respective old per-FS structures into the key-value list supported by
677 * nmount(2), then use kernel_mount(9) to mimic nmount(2) from kernelspace.
679 * Filesystems with mounters that use nmount(2) do not need to and should not
680 * implement vfs_cmount. Hopefully a future cleanup can remove vfs_cmount and
683 typedef int vfs_cmount_t(struct mntarg *ma, void *data, uint64_t flags);
684 typedef int vfs_unmount_t(struct mount *mp, int mntflags);
685 typedef int vfs_root_t(struct mount *mp, int flags, struct vnode **vpp);
686 typedef int vfs_quotactl_t(struct mount *mp, int cmds, uid_t uid, void *arg);
687 typedef int vfs_statfs_t(struct mount *mp, struct statfs *sbp);
688 typedef int vfs_sync_t(struct mount *mp, int waitfor);
689 typedef int vfs_vget_t(struct mount *mp, ino_t ino, int flags,
691 typedef int vfs_fhtovp_t(struct mount *mp, struct fid *fhp,
692 int flags, struct vnode **vpp);
693 typedef int vfs_checkexp_t(struct mount *mp, struct sockaddr *nam,
694 int *extflagsp, struct ucred **credanonp,
695 int *numsecflavors, int **secflavors);
696 typedef int vfs_init_t(struct vfsconf *);
697 typedef int vfs_uninit_t(struct vfsconf *);
698 typedef int vfs_extattrctl_t(struct mount *mp, int cmd,
699 struct vnode *filename_vp, int attrnamespace,
700 const char *attrname);
701 typedef int vfs_mount_t(struct mount *mp);
702 typedef int vfs_sysctl_t(struct mount *mp, fsctlop_t op,
703 struct sysctl_req *req);
704 typedef void vfs_susp_clean_t(struct mount *mp);
705 typedef void vfs_notify_lowervp_t(struct mount *mp, struct vnode *lowervp);
706 typedef void vfs_purge_t(struct mount *mp);
709 vfs_mount_t *vfs_mount;
710 vfs_cmount_t *vfs_cmount;
711 vfs_unmount_t *vfs_unmount;
712 vfs_root_t *vfs_root;
713 vfs_root_t *vfs_cachedroot;
714 vfs_quotactl_t *vfs_quotactl;
715 vfs_statfs_t *vfs_statfs;
716 vfs_sync_t *vfs_sync;
717 vfs_vget_t *vfs_vget;
718 vfs_fhtovp_t *vfs_fhtovp;
719 vfs_checkexp_t *vfs_checkexp;
720 vfs_init_t *vfs_init;
721 vfs_uninit_t *vfs_uninit;
722 vfs_extattrctl_t *vfs_extattrctl;
723 vfs_sysctl_t *vfs_sysctl;
724 vfs_susp_clean_t *vfs_susp_clean;
725 vfs_notify_lowervp_t *vfs_reclaim_lowervp;
726 vfs_notify_lowervp_t *vfs_unlink_lowervp;
727 vfs_purge_t *vfs_purge;
728 vfs_mount_t *vfs_spare[6]; /* spares for ABI compat */
731 vfs_statfs_t __vfs_statfs;
733 #define VFS_MOUNT(MP) ({ \
736 TSRAW(curthread, TS_ENTER, "VFS_MOUNT", (MP)->mnt_vfc->vfc_name);\
737 _rc = (*(MP)->mnt_op->vfs_mount)(MP); \
738 TSRAW(curthread, TS_EXIT, "VFS_MOUNT", (MP)->mnt_vfc->vfc_name);\
741 #define VFS_UNMOUNT(MP, FORCE) ({ \
744 _rc = (*(MP)->mnt_op->vfs_unmount)(MP, FORCE); \
747 #define VFS_ROOT(MP, FLAGS, VPP) ({ \
750 _rc = (*(MP)->mnt_op->vfs_root)(MP, FLAGS, VPP); \
753 #define VFS_CACHEDROOT(MP, FLAGS, VPP) ({ \
756 _rc = (*(MP)->mnt_op->vfs_cachedroot)(MP, FLAGS, VPP); \
759 #define VFS_QUOTACTL(MP, C, U, A) ({ \
762 _rc = (*(MP)->mnt_op->vfs_quotactl)(MP, C, U, A); \
765 #define VFS_STATFS(MP, SBP) ({ \
768 _rc = __vfs_statfs((MP), (SBP)); \
771 #define VFS_SYNC(MP, WAIT) ({ \
774 _rc = (*(MP)->mnt_op->vfs_sync)(MP, WAIT); \
777 #define VFS_VGET(MP, INO, FLAGS, VPP) ({ \
780 _rc = (*(MP)->mnt_op->vfs_vget)(MP, INO, FLAGS, VPP); \
783 #define VFS_FHTOVP(MP, FIDP, FLAGS, VPP) ({ \
786 _rc = (*(MP)->mnt_op->vfs_fhtovp)(MP, FIDP, FLAGS, VPP); \
789 #define VFS_CHECKEXP(MP, NAM, EXFLG, CRED, NUMSEC, SEC) ({ \
792 _rc = (*(MP)->mnt_op->vfs_checkexp)(MP, NAM, EXFLG, CRED, NUMSEC,\
796 #define VFS_EXTATTRCTL(MP, C, FN, NS, N) ({ \
799 _rc = (*(MP)->mnt_op->vfs_extattrctl)(MP, C, FN, NS, N); \
802 #define VFS_SYSCTL(MP, OP, REQ) ({ \
805 _rc = (*(MP)->mnt_op->vfs_sysctl)(MP, OP, REQ); \
808 #define VFS_SUSP_CLEAN(MP) do { \
809 if (*(MP)->mnt_op->vfs_susp_clean != NULL) { \
810 (*(MP)->mnt_op->vfs_susp_clean)(MP); \
814 #define VFS_RECLAIM_LOWERVP(MP, VP) do { \
815 if (*(MP)->mnt_op->vfs_reclaim_lowervp != NULL) { \
816 (*(MP)->mnt_op->vfs_reclaim_lowervp)((MP), (VP)); \
820 #define VFS_UNLINK_LOWERVP(MP, VP) do { \
821 if (*(MP)->mnt_op->vfs_unlink_lowervp != NULL) { \
822 (*(MP)->mnt_op->vfs_unlink_lowervp)((MP), (VP)); \
826 #define VFS_PURGE(MP) do { \
827 if (*(MP)->mnt_op->vfs_purge != NULL) { \
828 (*(MP)->mnt_op->vfs_purge)(MP); \
832 #define VFS_KNOTE_LOCKED(vp, hint) do \
834 if (((vp)->v_vflag & VV_NOKNOTE) == 0) \
835 VN_KNOTE((vp), (hint), KNF_LISTLOCKED); \
838 #define VFS_KNOTE_UNLOCKED(vp, hint) do \
840 if (((vp)->v_vflag & VV_NOKNOTE) == 0) \
841 VN_KNOTE((vp), (hint), 0); \
844 #define VFS_NOTIFY_UPPER_RECLAIM 1
845 #define VFS_NOTIFY_UPPER_UNLINK 2
847 #include <sys/module.h>
852 #define VFS_VERSION_00 0x19660120
853 #define VFS_VERSION_01 0x20121030
854 #define VFS_VERSION_02 0x20180504
855 #define VFS_VERSION VFS_VERSION_02
857 #define VFS_SET(vfsops, fsname, flags) \
858 static struct vfsconf fsname ## _vfsconf = { \
859 .vfc_version = VFS_VERSION, \
860 .vfc_name = #fsname, \
861 .vfc_vfsops = &vfsops, \
863 .vfc_flags = flags, \
865 static moduledata_t fsname ## _mod = { \
868 & fsname ## _vfsconf \
870 DECLARE_MODULE(fsname, fsname ## _mod, SI_SUB_VFS, SI_ORDER_MIDDLE)
873 * exported vnode operations
876 int dounmount(struct mount *, int, struct thread *);
878 int kernel_mount(struct mntarg *ma, uint64_t flags);
879 int kernel_vmount(int flags, ...);
880 struct mntarg *mount_arg(struct mntarg *ma, const char *name, const void *val, int len);
881 struct mntarg *mount_argb(struct mntarg *ma, int flag, const char *name);
882 struct mntarg *mount_argf(struct mntarg *ma, const char *name, const char *fmt, ...);
883 struct mntarg *mount_argsu(struct mntarg *ma, const char *name, const void *val, int len);
884 void statfs_scale_blocks(struct statfs *sf, long max_size);
885 struct vfsconf *vfs_byname(const char *);
886 struct vfsconf *vfs_byname_kld(const char *, struct thread *td, int *);
887 void vfs_mount_destroy(struct mount *);
888 void vfs_event_signal(fsid_t *, u_int32_t, intptr_t);
889 void vfs_freeopts(struct vfsoptlist *opts);
890 void vfs_deleteopt(struct vfsoptlist *opts, const char *name);
891 int vfs_buildopts(struct uio *auio, struct vfsoptlist **options);
892 int vfs_flagopt(struct vfsoptlist *opts, const char *name, uint64_t *w,
894 int vfs_getopt(struct vfsoptlist *, const char *, void **, int *);
895 int vfs_getopt_pos(struct vfsoptlist *opts, const char *name);
896 int vfs_getopt_size(struct vfsoptlist *opts, const char *name,
898 char *vfs_getopts(struct vfsoptlist *, const char *, int *error);
899 int vfs_copyopt(struct vfsoptlist *, const char *, void *, int);
900 int vfs_filteropt(struct vfsoptlist *, const char **legal);
901 void vfs_opterror(struct vfsoptlist *opts, const char *fmt, ...);
902 int vfs_scanopt(struct vfsoptlist *opts, const char *name, const char *fmt, ...);
903 int vfs_setopt(struct vfsoptlist *opts, const char *name, void *value,
905 int vfs_setopt_part(struct vfsoptlist *opts, const char *name, void *value,
907 int vfs_setopts(struct vfsoptlist *opts, const char *name,
909 int vfs_setpublicfs /* set publicly exported fs */
910 (struct mount *, struct netexport *, struct export_args *);
911 void vfs_periodic(struct mount *, int);
912 int vfs_busy(struct mount *, int);
913 int vfs_export /* process mount export info */
914 (struct mount *, struct export_args *);
915 void vfs_allocate_syncvnode(struct mount *);
916 void vfs_deallocate_syncvnode(struct mount *);
917 int vfs_donmount(struct thread *td, uint64_t fsflags,
918 struct uio *fsoptions);
919 void vfs_getnewfsid(struct mount *);
920 struct cdev *vfs_getrootfsid(struct mount *);
921 struct mount *vfs_getvfs(fsid_t *); /* return vfs given fsid */
922 struct mount *vfs_busyfs(fsid_t *);
923 int vfs_modevent(module_t, int, void *);
924 void vfs_mount_error(struct mount *, const char *, ...);
925 void vfs_mountroot(void); /* mount our root filesystem */
926 void vfs_mountedfrom(struct mount *, const char *from);
927 void vfs_notify_upper(struct vnode *, int);
928 void vfs_oexport_conv(const struct oexport_args *oexp,
929 struct export_args *exp);
930 void vfs_ref(struct mount *);
931 void vfs_rel(struct mount *);
932 struct mount *vfs_mount_alloc(struct vnode *, struct vfsconf *, const char *,
934 int vfs_suser(struct mount *, struct thread *);
935 void vfs_unbusy(struct mount *);
936 void vfs_unmountall(void);
937 extern TAILQ_HEAD(mntlist, mount) mountlist; /* mounted filesystem list */
938 extern struct mtx mountlist_mtx;
939 extern struct nfs_public nfs_pub;
940 extern struct sx vfsconf_sx;
941 #define vfsconf_lock() sx_xlock(&vfsconf_sx)
942 #define vfsconf_unlock() sx_xunlock(&vfsconf_sx)
943 #define vfsconf_slock() sx_slock(&vfsconf_sx)
944 #define vfsconf_sunlock() sx_sunlock(&vfsconf_sx)
945 struct vnode *mntfs_allocvp(struct mount *, struct vnode *);
946 void mntfs_freevp(struct vnode *);
949 * Declarations for these vfs default operations are located in
950 * kern/vfs_default.c. They will be automatically used to replace
951 * null entries in VFS ops tables when registering a new filesystem
952 * type in the global table.
954 vfs_root_t vfs_stdroot;
955 vfs_quotactl_t vfs_stdquotactl;
956 vfs_statfs_t vfs_stdstatfs;
957 vfs_sync_t vfs_stdsync;
958 vfs_sync_t vfs_stdnosync;
959 vfs_vget_t vfs_stdvget;
960 vfs_fhtovp_t vfs_stdfhtovp;
961 vfs_checkexp_t vfs_stdcheckexp;
962 vfs_init_t vfs_stdinit;
963 vfs_uninit_t vfs_stduninit;
964 vfs_extattrctl_t vfs_stdextattrctl;
965 vfs_sysctl_t vfs_stdsysctl;
967 void syncer_suspend(void);
968 void syncer_resume(void);
970 struct vnode *vfs_cache_root_clear(struct mount *);
971 void vfs_cache_root_set(struct mount *, struct vnode *);
973 void vfs_op_barrier_wait(struct mount *);
974 void vfs_op_enter(struct mount *);
975 void vfs_op_exit_locked(struct mount *);
976 void vfs_op_exit(struct mount *);
979 void vfs_assert_mount_counters(struct mount *);
980 void vfs_dump_mount_counters(struct mount *);
982 #define vfs_assert_mount_counters(mp) do { } while (0)
983 #define vfs_dump_mount_counters(mp) do { } while (0)
986 enum mount_counter { MNT_COUNT_REF, MNT_COUNT_LOCKREF, MNT_COUNT_WRITEOPCOUNT };
987 int vfs_mount_fetch_counter(struct mount *, enum mount_counter);
990 * Code transitioning mnt_vfs_ops to > 0 issues IPIs until it observes
991 * all CPUs not executing code enclosed by mnt_thread_in_ops_pcpu.
993 * This provides an invariant that by the time the last CPU is observed not
994 * executing, everyone else entering will see the counter > 0 and exit.
996 * Note there is no barrier between vfs_ops and the rest of the code in the
997 * section. It is not necessary as the writer has to wait for everyone to drain
998 * before making any changes or only make changes safe while the section is
1001 #define vfs_op_thread_entered(mp) ({ \
1002 MPASS(curthread->td_critnest > 0); \
1003 *zpcpu_get(mp->mnt_thread_in_ops_pcpu) == 1; \
1006 #define vfs_op_thread_enter(mp) ({ \
1007 bool _retval = true; \
1009 MPASS(!vfs_op_thread_entered(mp)); \
1010 zpcpu_set_protected(mp->mnt_thread_in_ops_pcpu, 1); \
1011 __compiler_membar(); \
1012 if (__predict_false(mp->mnt_vfs_ops > 0)) { \
1013 vfs_op_thread_exit(mp); \
1019 #define vfs_op_thread_exit(mp) do { \
1020 MPASS(vfs_op_thread_entered(mp)); \
1021 __compiler_membar(); \
1022 zpcpu_set_protected(mp->mnt_thread_in_ops_pcpu, 0); \
1026 #define vfs_mp_count_add_pcpu(mp, count, val) do { \
1027 MPASS(vfs_op_thread_entered(mp)); \
1028 zpcpu_add_protected(mp->mnt_##count##_pcpu, val); \
1031 #define vfs_mp_count_sub_pcpu(mp, count, val) do { \
1032 MPASS(vfs_op_thread_entered(mp)); \
1033 zpcpu_sub_protected(mp->mnt_##count##_pcpu, val); \
1036 #else /* !_KERNEL */
1038 #include <sys/cdefs.h>
1043 int fhlink(struct fhandle *, const char *);
1044 int fhlinkat(struct fhandle *, int, const char *);
1045 int fhopen(const struct fhandle *, int);
1046 int fhreadlink(struct fhandle *, char *, size_t);
1047 int fhstat(const struct fhandle *, struct stat *);
1048 int fhstatfs(const struct fhandle *, struct statfs *);
1049 int fstatfs(int, struct statfs *);
1050 int getfh(const char *, fhandle_t *);
1051 int getfhat(int, char *, struct fhandle *, int);
1052 int getfsstat(struct statfs *, long, int);
1053 int getmntinfo(struct statfs **, int);
1054 int lgetfh(const char *, fhandle_t *);
1055 int mount(const char *, const char *, int, void *);
1056 int nmount(struct iovec *, unsigned int, int);
1057 int statfs(const char *, struct statfs *);
1058 int unmount(const char *, int);
1060 /* C library stuff */
1061 int getvfsbyname(const char *, struct xvfsconf *);
1064 #endif /* _KERNEL */
1066 #endif /* !_SYS_MOUNT_H_ */